South Asia Economic Focus | Fall 2020
Informality and COVID-19
Beaten or Broken? Informality and COVID-19
2
Beaten or Broken? Informality and COVID-19
© 2020 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved 1 2 3 4 23 22 21 20 This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions
This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creativecommons.org/licenses/ by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: World Bank. 2020. “COVID-19 and Informality” South Asia Economic Focus (October), World Bank, Washington, DC. Doi: 10.1596/978-1-4648-1640-6. License: Creative Commons Attribution CC BY 3.0 IGO Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third-party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. ISBN (electronic): 978-1-4648-1640-6 DOI: 10.1596/ 978-1-4648-1640-6
Cover photo: © Mahendra N Parikh / Shutterstock Cover design: Alejandro Espinosa / sonideas.com
PHOTO BY: IM_ROHITBHAKAR / SHUTTERSTOCK.COM
Beaten or Broken? Informality and COVID-19
3
4
Beaten or Broken? Informality and COVID-19 
Beaten or Broken? Informality and COVID-19
T
his report is a joint product of the Office of the Chief Economist for the South Asia Region (SARCE) and the Macroeconomics, Trade and Investment (MTI) Global Practice. Its preparation was led by Robert C. M. Beyer (Economist, SARCE), Valerie Anne Mercer Blackman (Senior Economist, SARCE), and Maurizio Bussolo (Lead Economist, SARCE) under the oversight of Hans Timmer (Chief Economist, South Asia Region), in close collaboration with Manuela Francisco (Practice Manager, MTI) and Zoubida Kherous Allaoua (EFI South Asia Regional Director). The core team responsible for Chapter 1 consisted of Robert C. M. Beyer, Milagros Chocce, Sebastián Franco-Bedoya, and Virgilio Galdo (all SARCE); boxes were contributed by Koen Martijn Geven and Diana Goldemberg (both Education Global Practice), Jiemin Ren, and Rucheta Singh (both SARCE). The core team responsible for Chapter 2 consisted of Valerie Anne Mercer Blackman (SARCE), Benoit Campagne (MTI) and Yi ‘Claire’ Li (SARCE); Sebastián Franco-Bedoya and Muthukumara Mani (SARCE) contributed boxes. The core team of Chapter 3 consisted of Maurizio Bussolo, Ananya Kotia, Jean Nahrae Lee, Nayantara Sarma, Siddharth Sharma, Anaise Williams and Yue Zhou (all SARCE). In Chapter 3, boxes were contributed by the Poverty Global Practice: Urmila Chatterjee on the relief policies, and Luz Carazo, Maria Eugenia Genoni and Nandini Krishnan on labor market impacts from real-time data collection efforts. Colleagues from MTI providing information for the country briefs in Chapter 4 include Sayed Murtaza Muzaffari, Tobias Akhtar Haque (Afghanistan); Melanie Simone Trost (Bhutan); Mona Prasad, Bernard Haven, Nazmus Sadat Khan (Bangladesh); Pui Shen Yoong (Maldives); Kene Ezemenari, Nayan Krishna Joshi, Florian Blum (Nepal); Aurélien Kruse, Rangeet Ghosh, Dhruv Sharma, Tanvir Malik, Rishabh Choudhary (India); Adnan Ashraf Ghumman, Muhammad Waheed, Derek Hung Chiat Chen, Zehra Aslam (Pakistan); Fernando Gabriel Im and Kishan Abeygunawardana (Sri Lanka) under supervision of Manuela Francisco (Practice Manager, MTI). Chapter 3 greatly benefitted from extensive discussions and suggestions from Ravi Kanbur (Cornell University) and Chris Woodruff (Oxford University), advisors to the SARCE informality research program. Valuable inputs were provided by Milagros Alejandra Chocce Falla (SARCE), Laura Liliana Moreno Herrera and Minh Cong Nguyen, Yeon Soo Kim (all Poverty GP), Joshua Wimpey (DEC). We thank these (in alphabetical order) for their helpful comments at various stages of the work: Syud Amer Ahmed (Social Protection and Jobs GP), Andras Bodor (Social Protection and Jobs GP), Andrew Dabalen (Poverty GP), Ayago Esmubancha Wambile (Poverty GP), Christoph Lakner (DEC), Pedro Olinto (Poverty GP), Stefano Paternostro (Social Protection and Jobs GP), Silvia Redaelli (Poverty GP), Sutirtha Sinha Roy (Poverty GP), Indhira Vanessa Santos (Social Protection and Jobs GP), Michael Weber (Social Protection and Jobs GP) Nishant Yonzan (DEC).
PHOTO BY: MANOEJ PAATEEL / SHUTTERSTOCK.COM
Useful comments and suggestions were also provided by Zoubida Allaoua (EFI South Asia Regional Director), Najy Benhassine (Country Director for Pakistan), Faris Hadad-Zervos (Country Director for Nepal, Sri Lanka and the Maldives), Mercy Miyang Tembon (Country Director for Bangladesh and Bhutan), Lei ‘Sandy’ Ye and other colleagues in the Prospects Group (EPGDR), by Sanjay Gupta (Consultant) and Cecile Fruman (Engagement) Krizia Anne Garay (Asian Development Bank), by numerous colleagues from the Office of the Chief Economist for the South Asia Region, the Macroeconomics, Trade and Investment Global Practice – including by Kishan Weditha Namanayaka Abeygunawardana, Zehra Aslam, Florian Blum, Derek Hung Chiat Chen, Bernard Haven, Kene Ezemenari, Nayan Krishna Joshi, Zayed Murtaza Muzzafari, Nazmus Sadat Khan, Aurélien Kruse, Nyda Mukthar, Melanie Simone Trost, and Pui Shen Yoong – as well as by participants of the 6th South Asia Economic Policy Network Conference on ‘COVID-19 and Informality in South Asia’ held virtually in September 2020. Alejandro Espinosa at Sonideas was responsible for the layout, design, typesetting, and an accompanying video, and William Shaw edited the chapters. Elena Karaban (Manager, SAR External Affairs), Yann Doignon (External Affairs Officer), Rana Damayo AlGazzaz, and Adnan Javaid Siddiqi (both Consultants) coordinated the dissemination, and Neelam Chowdhry provided valuable administrative support. South Asia as used in this report includes Afghanistan, Bangladesh, Bhutan, India, Maldives, Nepal, Pakistan and Sri Lanka. The cutoff date for this report was September 30, 2020.
South Asia Chief Economist Office Macroeconomics, Trade and Investment Global Practice
5
Beaten or Broken? Informality and COVID-19 
PHOTO BY: MANOEJ PAATEEL / SHUTTERSTOCK.COM
6
Beaten or Broken? Informality and COVID-19
Table of contents C H A PTER 1
Recent economic developments Summary From dire straits to gradual recovery COVID-19 hit South Asia late but hard The economic impact was sudden and steep Spatially heterogeneous COVID-19 impacts Strong policy measures preserved macro-financial stability … … but the situation is fragile, and policy tools have been exhausted Conclusion References
8 9 10 12 15 21 24 26 30 33
C H A PTER 2
With a new world, a changing outlook Summary Introduction Growth in the region downgraded Simulating the risks to the forecast outlook Longer-term effects: a preliminary assessment The role of government and policy recommendations References
36 37 38 38 42 50 55 58
C H A PTER 3
The impact of COVID-19 on the informal sector Summary Introduction COVID-19 is severely reducing employment and incomes, particularly in the informal sector Model simulations indicate that COVID-19 has particularly harmed informal sector workers
66 67 68 68 74
The informal sector cannot (especially in the short run) expect huge benefits from key aspects of the digital economy Understanding key characteristics of the informal sector is critical for policy effectiveness Policy implications Conclusions References
81 86 88 95 96
C H A PTER 4
South Asia country briefs
107
7
8
Beaten or Broken? Informality and COVID-19
1
Chapter
Recent economic developments
Beaten or Broken? Informality and COVID-19 Recent economic developments
Summary The COVID-19 pandemic is not yet under control in South Asia, despite early containment measures. In March, South Asian countries quickly imposed lockdowns and travel restrictions, but not all countries were able to contain the domestic spread of COVID-19. Due to low testing, social stigma, and a young population, the actual extent of COVID-19 infections is highly uncertain, but likely much higher than recorded numbers suggest. The crisis brought South Asia to a near standstill. Travel restrictions prevented travelers from reaching South Asia and lockdown measures triggered massive supply disruptions. Information from high-frequency variables, combined in activity indicators, show an unprecedented contraction. In April, activity dropped by 40 percent in Pakistan and by around two thirds in the other countries. Activity recovered subsequently across the region, but it remained below pre-COVID levels in August. High-frequency approximations of GDP suggest year-over-year contractions during the second quarter of this year in all countries and a subsequent gradual recovery. The collapse in activity was widespread. The economic disruption is even visible from space: South Asia has darkened since March. Between March and August, nighttime light intensity declined in more than three quarters of South Asia’s districts. In August, the average nighttime light intensity across districts was still 10 percent below its level a year earlier. Mobility declined strongly in nearly all districts, as a result both of national containment measures and local COVID-19 infections. Some of the observed heterogeneity across districts can be explained by voluntary reductions in mobility due to higher local prevalence of COVID-19. During the national lockdown in India, districts with more recorded COVID-19 infections per capita experienced larger declines in mobility and nighttime lights.
PHOTO BY: HAREN GOGOI / SHUTTERSTOCK.COM
South Asian governments proactively stabilized economic activity through monetary easing, fiscal stimulus, and supportive financial regulation. For now, macro-financial stability has been preserved. However, the situation is fragile amid weak buffers and exhausted policy tools in some countries. Regulatory adjustments to the COVID-19 pandemic have exacerbated financial sector vulnerabilities, and fiscal stimulus despite large revenue shortfalls have resulted in rising fiscal deficits. Public debt, already high in Sri Lanka and Maldives before the pandemic, has risen further.
9
Beaten or Broken? Informality and COVID-19 Recent economic developments
From dire straits to gradual recovery
reported official GDP figures for the first half of this year recorded a contraction. The impact was
The world is in an unprecedented crisis. The global
especially large during the second quarter, with
pandemic of coronavirus disease 2019 (COVID‑19)
economic activity contracting by double digits in
is still worsening, with around a quarter million
most countries (Figure 1.1.B), and with more severe
new cases registered every day (Figure 1.1.A). As of
impacts in those with higher infection rates and
September 30, over 34 million cases of COVID-19
stricter containment measures (see Box 1.1).
(of which 7.7 million were active cases) and over 1 million deaths have been reported across more
On average, countries contracted by more than 10
than 180 countries. On September 30, India alone
percent and the contraction was particularly deep in
recorded over 80,000 new cases. The pandemic
India. Of the 60 countries that have published quar-
and measures to contain its spread have disrupted
terly gross domestic product (GDP) data for the second
economic activity across the world, resulting in a
quarter of this year, only China and Vietnam saw posi-
severe global recession. All countries that already
tive growth. On average, countries lost 11.6Â percent of
Figure 1.1: COVID-19 infections are still rising and economic activity has collapsed, but growth is rebounding and financing conditions for EMDEs remain benign. A. COVID-19 infections are still growing.
B. GDP collapsed in most countries in 2020 Q2.
GDP growth and COVID-19 cases
New COVID-19 cases 7D MA, thousands New COVID-19 cases 7D350 MA, thousands
GDP growth 5 and COVID-19 cases 5 0
350 300
250 200 200 150 150 100 100 50 Jan-29Jan-29 Feb-07Feb-07 Feb-16Feb-16 Feb-25Feb-25 Mar-05Mar-05 Mar-14Mar-14 Mar-23Mar-23 Apr-01Apr-01 Apr-10Apr-10 Apr-19Apr-19 Apr-28Apr-28 May-07 May-07 May-16May-16 May-25 May-25 Jun-03Jun-03 Jun-12Jun-12 Jun-21Jun-21 Jun-30Jun-30 Jul-09Jul-09 Jul-18Jul-18 Jul-27Jul-27 Aug-05Aug-05 Aug-14Aug-14 Aug-23Aug-23 Sep-01Sep-01 Sep-10Sep-10 Sep-19Sep-19 Sep-28Sep-28
50 0 0
GDP growth GDP growth in 2020Q2 in 2020Q2 (y-o-y)(y-o-y)
300 250
CHN
VNM
0 0 -50
VNM2
4
2
4
CHN
6
8
6
8
-5 -10 -10 -15
10 RUS RUSUSA BRA USA
-15 -20
TUN BLZTUN BLZ
-20 -25 -25 -30
ITA ESP
IND
ESP
IND
PER
-30 -35
PER
ln(COVID-19for cases per million population) D. Financing conditions EMDEs remain benign. EMDE portfolio flows and EMBI spread USD, EMDEbillion portfolio flows and EMBI spread
Index, 60 50+=expansion
USD, billion 50
0
400 400
-90
-100
-90
-100
Composite PMI
SENTIX (RHS)
200 200
Portfolio flows Portfolio flows
Jul-20Jul-20
0
EMBI spread
Sep-20Sep-20
SENTIX (RHS)
-50 -50
Aug-20Aug-20
Composite PMI
Sep-20Sep-20
20 Aug-20Aug-20
-60 Jul-20Jul-20
30 Jun-20Jun-20
-60
May-20 May-20
30
Apr-20Apr-20
-30
Mar-20Mar-20
-30
40
Feb-20Feb-20
40
20
Basis points Basis 600 points 600
0
Jun-20Jun-20
0
May-20 May-20
0
50
Apr-20Apr-20
50
50
Mar-20Mar-20
30
-35
Feb-20Feb-20
60
ln(COVID-19 cases per million population)
Jan-20Jan-20
Index 30
10
BRA CHL CHL ITA
South Africa United States Brazil India Russia Mexico World South Africa United States Brazil India Russia Mexico a reboundWorld C. Monthly indicators suggest from an unprecedented low base. Global activity Index, 50+=expansion Index Global activity
Jan-20Jan-20
10
0
EMBI spread
Notes: A. The new cases are shown as seven-day moving average; B. The GDP growth was calculated using local currency units and the size of the bubbles is the average of the stringency index in Q2. Sources: A. Johns Hopkins University; B. World Bank; Johns Hopkins University; Hale et al. (2020); C. JPMorgan/IHS Markit and Haver Analytics; D. IIF, JP Morgan, Haver Analytics, and staff calculations
Beaten or Broken? Informality and COVID-19 Recent economic developments
Box 1.1 Both the spread of COVID-19 and related containment measures contributed to GDP losses The second quarter of 2020 was shaped by rising COVID-19 infections and lockdowns. During this quarter, COVID-19 infections picked up across the globe and most countries enacted stringent containment measures to control its domestic spread. The stringency of containment measures across countries can be compared with an index based on school closings, workplace closings, cancelation of public events, restrictions on gatherings, public transport closings, stay at home requirements, restrictions on internal movement, and international travel controls (Hale, Webster, Petherick, Phillips, and Kira 2020). Because of COVID-19 and the lockdowns, among 60 countries that already reported official GDP growth for the second quarter of this year, all but China’s and Vietnam’s economies contracted relative to the same quarter in 2019. For many it was the worst contraction ever recorded. Both COVID-19 infections and the stringency of the containment measures have had an impact. The decline in economic activity is correlated both with higher COVID-19 infections and more stringent containment measures (Table 1.1 Column 1). The decline in GDP – given the COVID-19 infection rate, the stringency of containment measures, and country characteristics – was smaller in more developed countries (Table 1.1 Column 2). One reason could be that those countries were able to adjust more smoothly to the pandemic and the containment measures, for example because in those countries more jobs can be done from home (see Chapter 3). Different from typical macroeconomic crises, services were hit badly. Consequently, countries with a larger share of their GDP generated by services had to deal with larger losses in GDP. A country generating 10 percent more of its GDP with services experienced a 3.3 percent larger contraction. Due to border closures, one may have expected that countries usually exporting more also contracted more, but there is no evidence for that. These results also hold when the growth rate in the quarter before is included as a control (Table 1.1 Column 3). The global collapse of economic activity is hence not just a consequence of the “great lockdown” but also of the faster spread of COVID-19 during this quarter. This has important implications for current activity and the recovery. Containment measures have been relaxed, which will support the economic rebound. However, COVID-19 is still spreading rapidly in many countries and there will be economic impacts so long as the pandemic is not under control (see Chapter 2). The stringency of containment measures, COVID-19 infections, and country characteristics together explain between 37 and 50 percent of the variation across countries’ growth rates in the second quarter. Despite this quarter being shaped by COVID-19 and lockdowns, there is hence still a lot of unexplained variation, showing that countries’ vulnerability to the pandemic was very heterogeneous.
Table 1.1: COVID-19 infections, containment measures, and country characteristics determined output in 2020 Q2.
GDP growth, y-o-y (1)
(2)
(3)
Log ( COVID-19 cases in Q2 per mill. pop)
-0.982** (0.378)
-1.073** (0.505)
-1.455*** (0.468)
Stringency of containment measures in Q2
-0.205*** (0.0634)
-0.137* (0.0757)
-0.125* (0.0683)
Log GDP per capita (2017 USD PPP)
3.441** (1.692)
3.693** (1.528)
Share of services in GDP
-0.331** (0.135)
-0.260** (0.123)
Share of manufacturing in GDP
0.00796 (0.153)
-0.0496 (0.139)
Share of exports in GDP
-0.00589 (0.0268)
-0.01 (0.0242)
GDP growth Q1 2020
0.802*** (0.225)
9.149* (4.791)
-10.02 (14.36)
-13.67 (12.99)
59
58
58
0.265
0.371
0.499
Constant Observations R-squared Notes: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1 Sources: Johns Hopkins University, Hale et al. (2020), World Bank, and staff calculations.
11
12
Beaten or Broken? Informality and COVID-19 Recent economic developments
their output compared to the year before. In South Asia, only a few countries publish quarterly GDP figures, and apart from India publication lags are sig-
COVID-19 hit South Asia late but hard
nificant. In the region, quarterly GDP from April to June is only available for India, where gross domestic
Governments across South Asia reacted resolute-
product (GDP) declined by 23.9 percent year-on-year.
ly to contain the pandemic. When the first cases
Private sector activity in manufacturing and services
of COVID-19 were detected in South Asia, coun-
fell by almost 30 percent. The only other country re-
tries quickly enacted strict measures to contain
porting a similar decline was Peru, which has one of
the domestic spread of COVID-19 (Figure 1.2.A).
the highest per capita infection cases in the world.
Following the example of many advanced economies, countries implemented travel restrictions,
Recent monthly economic indicators suggest a
border closures, and lockdowns. The travel restric-
gradual recovery. Purchasing Manager Indexes
tions have halted tourism and labor outmigration.
(PMIs) are monthly economic indicators based on
In addition, border closures severely disrupted
surveys of private companies that enquire, among
supply chains and trade throughout the region.
other things, about the status of new orders, out-
In some countries related logistical difficulties
put, and employment. A value above 50 indicates
and repatriations of foreign workers impacted
an improvement; a value below 50 a deterioration.
construction. The lockdowns depressed domestic
The global composite indicator was above 50 at the
supply and demand, as businesses were unable to
beginning of the year but collapsed subsequently.
operate and consumers curbed expenditures, trig-
It troughed in April, with a never recorded low of
gering a massive contraction in output and impos-
26 (Figure 1.1.C), when many countries enacted very
ing significant social hardship on poor and vulner-
strict containment measures, and substantial parts
able households – specifically urban migrants and
of the world were under lockdown. With the re-
workers in the informal economy (see Chapter 3).
strictions eased subsequently, the composite index
The lockdowns had far-reaching consequences in
bounced back and surpassed 50 in July, suggesting
other areas of life as well. For example, education
that the situation is improving, albeit very gradu-
came to a standstill and efforts to teach children
ally. The SENTIX, a monthly economic indicator
during school closures proved challenging. The
based on investor confidence, dropped with the on-
estimated costs of the school closures in terms
set of the pandemic and troughed in April as well.
of learning and earning losses are substantial (see Box 1.2).
Benign financing conditions for EMDEs provide a silver lining. When the COVID-19 pandemic spread
The spread of COVID-19 infections in South Asia
to more and more EMDEs in March, capital flows
has been heterogeneous across countries. It is not
reversed and interest rates for EMDE bonds in-
clear whether lockdowns can effectively mitigate
creased, as reflected in a rising Emerging Markets
a pandemic in countries with a large share of ur-
Bond Index (EMBI) spread (Figure 1.1.D). Howev-
ban poor and densely populated cities. In some
er, concerns about an imminent EMDE crisis un-
cases, they may even be counterproductive. While
wound in April and financing conditions improved
some countries were successful in controlling the
again. Significant quantitative easing in advanced
pandemic, others were not. Bhutan and Sri Lanka
economies, combined with an expectation that the
avoided large-scale domestic transmission and ex-
economic fallout from the pandemic will be con-
perienced only very small numbers of infections per
trolled, translates into relatively benign financing
capita. Despite comparable measures, cases surged
conditions for EMDEs so far. However, capital flows
in Bangladesh and Pakistan. However, after spik-
are inherently volatile, and a future reversal remains
ing between June and July, they have fallen subse-
a major risk (see Chapter 2). Investor sentiments
quently. In Bangladesh, where recorded infections
can be erratic, and a wider and longer than expect-
declined slower than in Pakistan, the government
ed spread of COVID-19 in EMDEs could trigger a
has now decided to gradually suspend COVID-19
reassessment. In the medium-run, if the economic
treatment in the dedicated COVID-19 public hos-
recovery between advanced economies and EMDEs
pitals because of a shift to home-based care and a
is not synchronized, for example because advanced
declining number of hospitalized patients. In In-
economies are accessing a vaccine for COVID-19
dia, Maldives and Nepal, however, the number of
first, monetary policy normalization in advanced
recorded cases is still rising rapidly (Figure 1.2.B).
economies could reverse capital flows again.
While cases in India were initially concentrated in
Beaten or Broken? Informality and COVID-19 Recent economic developments
Figure 1.2: All countries enacted strict measures to contain the spread of COVID-19; some succeeded, but cases are still surging in India and Nepal. A. All governments enacted strong containment measures.
B. Covid-19 infections declined in some countries but are still rising in India and again in Nepal.
New COVID-19 cases per million population in South Asia
Stringency of containment measures Index 100
7D MA
7D MA
80
300
70
80
250
60 60
200
50 40
40
150
30
100
20
20
50
10 0
Afghanistan Nepal
Bangladesh Pakistan
India Sri Lanka
India Maldives (RHS)
13-Sep
27-Sep
30-Aug
2-Aug
16-Aug
5-Jul
19-Jul
7-Jun
Bangladesh Pakistan
21-Jun
10-May
24-May
12-Apr
26-Apr
29-Mar
1-Mar
0 15-Mar
13-Sep
27-Sep
16-Aug
30-Aug
19-Jul
02-Aug
05-Jul
21-Jun
07-Jun
10-May
24-May
12-Apr
26-Apr
15-Mar
29-Mar
01-Mar
0
Nepal
Notes: A. The index is based on school closings, workplace closings, cancelation of public events, restrictions on gatherings, public transport closings, stay at home requirements, restrictions on internal movement, and international travel controls.; B. The new cases are shown as seven-day moving average. Sources: A. Hale et al. (2020); B. Johns Hopkins University and staff calculations.
Table 1.2: Not all South Asian countries were hit equally strong; in all of them the death rate is relatively low, but so is testing. Afghanistan
Bangladesh
Bhutan
India
Maldives
Nepal
Pakistan
Sri Lanka
China
United States
Total confirmed cases
39,268
363,479
282
6,312,584
10,291
77,817
312,806
3,380
90,528
7,233,043
Total cases per million people
1,008.7
2,207.1
365.5
4,574.3
19,038.3
2,670.7
1,416.1
157.8
62.9
21,851.9
Active cases
4,985
82,637
60
942,164
1,142
21,830
8,825
134
370
3,256,873
Death rate
3.7
1.4
0.0
1.6
0.3
0.6
2.1
0.4
5.2
2.9
Test per infected
2.8
5.4
485.6
12.0
14.9
13.1
11.3
85.7
15.7
14.1
25.5
7.7
52.7
250.0
20.6
Test per infected (in last 7 days)
9.7
Notes: The death rate is the ratio of recorded death due to COVID-19 to all recorded infections. Data as of September 30. Sources: National health ministries.
a few large and densely populated cities, COVID-19
Due to supply constraints, social stigma, and a
is now spreading in almost every state and across
young population (more likely to exhibit asymp-
smaller towns, villages, and rural areas as well. In
tomatic infections), testing is low and the actual
some states with high infections, health care sys-
extent of COVID-19 infections is highly uncertain.
tem capacity constraints are becoming a concern,
Most South Asian countries fare poorly on test-
especially in more rural areas. Maldives has the
ing for COVID-19. The benchmark range for ade-
highest number of confirmed cases per capita, and
quate testing set by the World Health Organization
new infections are still high. In Nepal, cases were
(WHO) is between 10 and 30 per confirmed case. In
initially restricted to areas bordering India but later
South Asia, only Bhutan and Sri Lanka, which have
surged also in urban areas. After peaking in June,
very few cases per capita, have tested far above that
new infections declined to very low levels in July.
range since March. Maldives and India are with-
However, Nepal has entered a second wave that is
in the range, though at its lower end. In Pakistan,
much more severe than the first.
testing was low initially, but it increased and was
13
Beaten or Broken? Informality and COVID-19 Recent economic developments
Box 1.2 Learning and related income losses due to school closures in South Asia are huge Temporary school closures in all South Asian countries have had major implications for students. They have kept 391 million students out of school in primary and secondary education, further complicating efforts to resolve the learning crisis. While most governments have made enormous efforts to mitigate the impact of school closures, it has been difficult to engage children through remote learning initiatives. This is resulting in enormous dropouts and substantial learning losses, which will have a lifetime impact on the productivity of a generation of students. The pandemic may cause up to 5.5 million students to drop out from the education system. The impact on learning is equally enormous. Most school systems closed in March, and—though there are important exceptions—countries are starting to reopen or have already opened their schools. Children have been out of school for approximately 5 months. Being out of school for that long means that children not only stop learning new things, they also forget some of what they have learned. The projected learning loss for the region is 0.5 years of learning-adjusted years of schooling (LAYS), falling from 6.5 LAYS to 6.0 LAYS, an enormous setback from recent advances in schooling (Figure 1.3.A). This figure already takes mitigation into account, including the likely effect of remote learning.
Figure 1.3: School closures across South Asia will result in learning and lifetime earning losses. A. South Asia has lost 0.5 years of learning…
B. … and may lose 5 percent in lifetime earnings.
Earning loss Billion US$, 2017 PPP
Pre-crisis Learning and COVID-19 loss learning adjusted years of school 9
700
Pre-COVID LAYS
8 7
600
6
500
5 4
400
3
300
2 1
200
0
100 Learning loss due to COVID-19
SA R
a nk
an
l pa
ist
La Sri
Pa k
Ne
ia Ind
n uta Bh
ng
lad
an Ba
ist an
SA R
a nk La Sri
l
an ist Pa k
pa Ne
ia Ind
n uta Bh
esh lad
ist
an
ng Ba
an
esh
0
-2
Afg h
-1
Afg h
14
Notes: No data available for Maldives. Learning-adjusted years of school are calculated by multiplying the estimates of expected years of school by the ratio of most recent harmonized test scores to 625. The wage calculation is based on lifetime income, using an annual return to an additional year of schooling of 8 percent, correcting for labor force participation and adult mortality. Source: Updated regional estimates from Azevedo, J.P. et al. (2020).
Future earning losses related to the learning setback are substantial. To estimate the long-term economic impact of these learning losses, one can project the effect of this learning loss on future earnings. Based on country data on household labor incomes, the average child in South Asia may lose USD 4,400 in lifetime earnings once having entered the labor market, equivalent to 5 percent of total earnings. These projections are based on what we currently know about returns to schooling, using the reduced level of learning caused by the crisis. Summing these numbers for all children in South Asia (and correcting for current labor force participation and adult survival rates for each country), the region stands to lose USD 622 billion from the school closures in the present scenario, or up to USD 880 billion in a more pessimistic scenario. While the regional loss is largely driven by India, all countries will lose substantial shares of their GDP (Figure 1.3.B). For reference, note that South Asian governments spend only USD 400 billion per year in total on primary and secondary education. The total loss in economic output from the current closures is hence substantially higher than what countries currently spend on education.
Beaten or Broken? Informality and COVID-19 Recent economic developments
adequate in September. Afghanistan and Bangla-
and manufacturing consequently collapsed. In
desh are testing far below the range, which makes
countries with fiscal years ending in the sum-
their recorded numbers particularly uncertain.
mer months, official GDP growth has decelerated
Low testing is both a result of supply constraints
significantly.
and subdued demand for testing. Due to the social stigma associated with an infection, many prefer
• Visitor arrivals: With travel restrictions in place
not to get tested in the first place and only seek
across the world, visitor arrivals have fallen to al-
medical attention when the symptoms are severe.
most zero since the end of March (Figure 1.4.A),
And since South Asia’s population is on average
impacting significantly the smaller countries
very young, infections are more likely to cause no
with large tourism sectors (see Box 1.2). Tourism
or only mild symptoms. Moreover, testing fees and
inflows in Maldives, the country most dependent
distrust in the testing facilities result in subdued
on tourism, remained anemic even after borders
demand in some places. Recent surveys based
reopened in mid-July. Only 13,787 tourists visit-
on testing random samples of the population for
ed between July 15 and September 15, a 95 per-
coronavirus antibodies suggest that official num-
cent year-on-year decline. There are only very
bers are grossly understating the actual spread of
few international commercial flights compared
COVID-19. In Afghanistan, a survey conducted by
to before the pandemic, and half of all resorts
the government and the WHO suggests that more
remain shut.
than a third of the population could have been
• Mobility: With the enactment of lockdowns, mo-
already infected and more than a half of the res-
bility declined sharply across South Asia. Pres-
idents of Kabul. In Pakistan, a government survey
ence at workplaces declined steeply (Figure 1.4.B)
concludes that 300,000 people may have been in-
as people stayed more at home. At the end of
fected in Islamabad alone (a prevalence of 14.5 per-
March, presence at the workplace was between
cent). In Delhi, a study showed that 22 percent of
40 percent to 80 percent lower than normal. It
the people have coronavirus antibodies, and a na-
dropped the least in Afghanistan and the most in
tional survey conducted between May 11 and June
Sri Lanka and Nepal. It recovered subsequently
4 suggests that the ratio of infections to registered
across the region, but it is still around 20 percent
cases may have been around 82. In Dhaka city, one
below normal levels in most countries, and near-
in every 10 people may have been infected already.
ly 40 percent below in Nepal due to new contain-
Even if the actual infections are much higher than
ment measures.
the registered cases, however, it seems unlikely
• Services: Following the stringent containment
that South Asian countries are already close to the
measures, activities in tourism, travel, trade and
levels that would trigger herd immunity. In ad-
transport have been severely disrupted, result-
dition, the evidence that those who recover from
ing in a near collapse in certain services such as
COVID-19 develop long-term immunity is not
hotels, restaurants, aviation, and trade. The Ser-
very solid. Hence, there is significant uncertainty
vices PMI, which is available only for India and
about the future development of the pandemic in
Sri Lanka, fell to unprecedented levels (Figure
South Asia, and the only relative certainty is that it
1.4.C). While it recovered subsequently to 56 in
is unlikely to pass anytime soon.
Sri Lanka in August, it was still far below 50 in India.
The economic impact was sudden and steep
• Industrial production: In April, industrial production collapsed to around 40 percent of its pre-COVID-19 level in India, to 50 percent in Sri Lanka, to 60 percent in Pakistan, and to 70 percent in Bangladesh (Figure 1.4.D). When
The pandemic and related containment measures
restrictions were eased, industrial production
brought South Asia to a standstill. Travel restric-
firmed, but it remained subdued in July in
tions prevented travelers from reaching South
India.
Asia, with immense consequences for tourism
• GDP growth: Among South Asian countries, only
and related services. After lockdown measures
India already published GDP data for the sec-
were implemented mobility declined sharply,
ond quarter of this year. With a decline of 23.9
triggering
disruptions.
percent (y-o-y), India’s contraction is one of the
These in turn impacted incomes and amplified
largest among all countries in the world (see
risk aversion, which reduced demand. Services
above). In Bangladesh, Bhutan, and Pakistan, the
unprecedented
supply
15
Beaten or Broken? Informality and COVID-19 Recent economic developments
Figure 1.4 Economic activity in South Asia came to a near stand-still. A. Visitor arrivals in South Asia dropped to zero.
Nepal Nepal
Sri Sri Lanka Lanka
Maldives Maldives
00
-100 -100
India India (RHS) (RHS)
C. Services PMI indicators were at record lows.
Afghanistan Afghanistan Sri Sri Lanka Lanka
Bangladesh Bangladesh Nepal Nepal
13-Sep 13-Sep
-80 -80
16-Aug 16-Aug
200,000 200,000
30-Aug 30-Aug
-60 -60
19-Jul 19-Jul
400,000 400,000
02-Aug 02-Aug
-40 -40
05-Jul 05-Jul
600,000 600,000
21-Jun 21-Jun
-20 -20
07-Jun 07-Jun
800,000 800,000
10-May 10-May
00
24-May 24-May
Aug-20 Aug-20
Jul-20 Jul-20
Jun-20 Jun-20
May-20 May-20
Apr-20 Apr-20
Mar-20 Mar-20
Jan-20 Jan-20
Feb-20 Feb-20
50,000 50,000
1,000,000 1,000,000
12-Apr 12-Apr
100,000 100,000
20 20
26-Apr 26-Apr
150,000 150,000
1,200,000 1,200,000
29-Mar 29-Mar
200,000 200,000
Presence Presence at at workplace workplace Index Index
15-Mar 15-Mar
Total Total number number
250,000 250,000
00
B. Presence at the workplace declined massively.
01-Mar 01-Mar
Visitor Visitor arrivals arrivals in in India, India, Nepal, Nepal, Sri Sri Lanka Lanka and and Maldives Maldives Total Total number number
India India Pakistan Pakistan
D. Industrial production plummeted.
Purchasing Purchasing managers managers index: index: services services Index, Index, 50+= 50+= expansion expansion 70 70
Industrial Industrial production production Index Index (Jan (Jan 2020 2020 == 100) 100) 120 120 110 110
60 60
100 100
50 50
90 90
40 40
80 80
30 30
70 70
20 20
India India
Sri Sri Lanka Lanka
India India
Bangladesh Bangladesh
Pakistan Pakistan
Jul-20 Jul-20
Jun-20 Jun-20
May-20 May-20
Apr-20 Apr-20
Mar-20 Mar-20
Feb-20 Feb-20
Aug-20 Aug-20
Jul-20 Jul-20
Jun-20 Jun-20
May-20 May-20
40 40
Apr-20 Apr-20
00
Mar-20 Mar-20
50 50 Feb-20 Feb-20
10 10
Jan-20 Jan-20
60 60
Jan-20 Jan-20
16
Sri Sri Lanka Lanka
Notes: B. The decline refers to the change of visits and length of stay, compared to a baseline period. The baseline period is defined as the median value for the corresponding day of the week, during the 5-week period from January 3 to February 6. Holidays and weekends are linearly interpolated. For Afghanistan, data from May 19Â to July 2 is missing. Sources: A. Ministry of Tourism of India and Maldives; Ministry of Culture, Tourism and Civil Aviation of Nepal; Sri Lanka Tourism Development Authority; B. Google COVID-19 Community Mobility Reports; C. IHS Markit, Central Bank of Sri Lanka and CEIC; D. World Bank.
last fiscal year ended at the end of June and in
export restrictions disrupted industrial activity.
Nepal it ended July 15. In all of them the final
In Pakistan, real GDP growth (at factor cost) is
months dragged down growth. In Bangladesh,
estimated to have declined from 1.9 percent in
real GDP growth fell to an estimated 2.0 per-
FY18/19 to -1.5 percent in FY19/20, reflecting the
cent in FY19/20. On the demand side, exports
effects of COVID-19Â containment measures that
declined by 18.5 percent as external demand
followed monetary and fiscal tightening prior to
for readymade garments (RMG) plummeted. In
the outbreak.
Bhutan and Nepal, real GDP growth is estimated to have decelerated to 1.5 percent and 0.2 per-
Activity indicators can be built by combining in-
cent, respectively. Tourist arrivals dried-up and
formation from different high-frequency vari-
reduced foreign demand; shortages in critical
ables. Since the onset of COVID-19, many different
inputs (including foreign labor) and temporary
high-frequency indicators have been used to assess
Beaten or Broken? Informality and COVID-19 Recent economic developments
Box 1.3 Tourism in South Asia has been shattered but there are opportunities Figure 1.5: Tourism contributes significantly to GDP and employment in South Asia. Tourism contributions Percent 60
40
20
0
Maldives
Bhutan
Sri Lanka Contribution to GDP
India
Nepal
Pakistan
Bangladesh
Contribution to Employment
Note: The contribution to GDP includes direct as well as indirect contributions. Source: World Travel and Tourism Council.
Over the last two decades, South Asia has emerged as an attractive tourist destination due to its price competitiveness and diverse natural and cultural resources. Tourism has been an important driver of economic growth and job creation. According to the World Travel and Tourism Council (WTTC), tourism in South Asia contributed USD 234 billion (6.6 percent of the region’s GDP) in 2019. In Maldives, tourism contributed 56.6 percent of GDP (directly and indirectly) and 59.6 percent of employment (Figure 1.5). The tourism industry in Bhutan generated USD 88.63 million in 2019, contributing significantly to socio-economic development through revenue and foreign currency generation, as well as through job creation. In Sri Lanka, the tourism sector has grown rapidly in past decade, especially as a generator of jobs. From 2009 to 2019, the direct and indirect tourism employment has more than doubled. Eight out of ten tourism jobs were in hotels and restaurants. In India, tourism’s share in GDP has been declining over the last ten years, but the contribution of tourism to employment increased somewhat, from 10 percent in FY09/10 to 13 percent in FY18/19. The COVID-19 pandemic has triggered an unprecedented crisis in South Asia’s tourism economy. The WTTC estimates losses to amount to over USD 50 billion in the travel and tourism sector and that about 47.7 million jobs—many held by women and vulnerable groups working in the informal sector—are at risk due to the COVID-19 pandemic. The strict local containment measures and the pandemic’s impact on global travel have resulted in a significant decline in tourist arrivals in Bhutan, India, Nepal, Sri Lanka, and Maldives (Figure 1.4.A). The adverse impact on Maldives’ tourist inflows remains significant even after borders reopened in mid-July as international commercial flights have been slow to resume. Governments are providing support to the tourism sectors. In Bhutan, the Economic Contingency Plan (ECP) – aimed at helping priority sectors – provides support to tourism. Tax payments for tourism and related sectors (hotel, airlines and tour operators) are deferred until the end of the year, and waivers for rent payments and other charges for tourism related businesses leasing government properties (from April to December 2020) and free electricity and wi-fi charges for hotels (from July to September 2020) are provided. In Maldives, the Economic Recovery Plan includes offers financing to tourist resorts and guesthouses through loans from the Bank of Maldives, and in Nepal measures include a lending program for the tourism sector. In Sri Lanka, a six-month moratorium on bank loans for the tourism sectors was established. Recovery measures need to restore confidence and stimulate demand. As a first step, different forms of “tourism bubbles” have been proposed in Bhutan, India and Maldives. Building these safe zones bilaterally or among a group of countries with similar recovery trajectories could enhance regional collaboration and strengthen countries’ capabilities to safely manage the flow of tourists. In Maldives, the unique “one island, one resort” concept facilitates socially distanced vacations, but difficulties in resuming commercial flights and recent increases in domestic transmission pose challenges to attracting more visitors. Specific measures to mitigate health risks are crucial to keep travelers and workers safe and to reopen effectively. These may
17
18
Beaten or Broken? Informality and COVID-19 Recent economic developments
include hygiene protocols for hotels, restaurants, taxis and other tourism sector related public infrastructure and transportation, as well as requirements for regular cleaning of transportation, health screening and temperature checks at borders. South Asia offers multiple nature-based or ecotourism opportunities, including protected areas, culture tourism, nature resorts, adventure sports, and religious tourism. These could be promoted further through use of digital applications especially to promote intra-regional tourism. A regional effort could be undertaken to create a more enabling environment to support its growth. Interventions may include guidelines on designing and implementing safeguards and safety mechanisms, marketing, development of information and booking portals, and working with national and sub-national governments for appropriate policy changes. Due to new investment in the physical and technological infrastructure, greener value chains, and greater collaboration within destination management and regional partners, the COVID-19 pandemic could be a catalyst for the diversification of tourism products and services and a shift towards a more resilient and sustainable tourism industry in the longer term.
the evolution of economic output compared to
in June, and 15 percent below in July. Sri Lanka ex-
pre-COVID levels. In India, for example, electric-
perienced the largest drop of all countries in April
ity consumption is strongly correlated with eco-
but recovered faster than the others thereafter, as
nomic activity and available at daily frequency. It
a widespread domestic contagion was avoided.
was almost 30 percent below normal levels at the
In July, activity was lowest in Nepal, followed by
end of March and remained below normal levels in
India. The indicators based on fewer variables, to
August (see Box 1.3). While some of the variables
enhance comparability across countries, show a
have been studied for a long time, others like the
similar picture, although they suggest activity was
mobility data discussed above are new and directly
somewhat higher in India and Bangladesh (Fig-
related to the pandemic. Information from differ-
ure 1.7.B). The speed of the recovery has notably
ent variables can be combined in a simple activity
slowed in August across the region.
indicator with a statistical procedure that extracts their maximum common variance and combines
The relationship with GDP growth has not yet been
them into a common score. For each country two
established for all high-frequency variables enter-
such indicators are computed: one with a selection
ing the activity indicators. The activity indicators
of variables that seem important for activity in that
presented above have the crucial advantage that
country, and one with a reduced number of vari-
they include recently available variables that pre-
ables to enhance comparability of the indicators
sumably have a strong relationship with economic
across countries.
activity during the pandemic. However, they have three drawbacks: first, this relationship cannot be
Across the region, activity indicators troughed
properly estimated, as the time period for which
in April. When travel restrictions took effect and
some of the high-frequency indicators are available
countries enacted domestic containment mea-
is too short; second, the high-frequency variables
sures, activity started to decline. In March, activity
selected to enter the indicators can hence not be
in South Asia was only 80 percent of its January
chosen based on their past relationship with eco-
level (Figure 1.7.A). It fell the most in Sri Lanka,
nomic activity; and, third, the indicators above just
which introduced strict stay-home-orders and
extract common variance and hence ignore the
suffered particularly from the stop of tourist ar-
relationship of included variables with economic
rivals. Daily mobility and electricity consumption
activity conditional on the other variables includ-
data suggest a collapse after lockdown measures
ed (different variables may all contain the same
were enacted during the last week in March. In
information). Hence, they are complemented with
line, average monthly activity fell further in April,
formal Quarterly Economic Indicators based on a
to only 40 percent of the pre-COVID level. It fell
simple econometric model that provides an esti-
to levels between 30 and 40 percent of the pre-
mate of current economic activity. It is constructed
COVID level in all countries except Pakistan,
in three steps: first, many monthly indicators cov-
where activity was nearly 60 percent as high as
ering a reasonably long time period for South Asian
before COVID-19. Activity recovered across the
countries are collected; second, a statistical proce-
board subsequently. But it remained 40 percent
dure is used to select a limited number of variables
below pre-COVID levels in May, a quarter below
that together explain past GDP growth; and third,
Beaten or Broken? Informality and COVID-19 Recent economic developments
Box 1.4: Assessing India’s economic activity with daily electricity consumption In India, electricity consumption is strongly related to overall economic activity. Electricity is an input to activities throughout the economy, from industrial production to commerce and household activity, and changes in its consumption thus reveal information about these activities. It has a strong monthly relationship with other high-frequency indicators after detrending and seasonally adjusting them. Electricity consumption is strongly related to trade (both to exports and imports), to industrial production and similar activities, to traffic (whether from freight, cargo, or passengers) and even to tourist arrivals (Table 1.3).
Table 1.3: Electricity consumption in India is strongly correlated with economic activity. (1)
(2)
(3)
(4)
Trade
(5)
(6)
(7)
(8)
Generation/production
(9)
Traffic
(10) Tourism
Exports
Imports
IP
Auto
Steel
Textile
Freight
Cargo
Passenger
Foreign Arrivals
0.182***
0.227***
0.432***
0.241***
0.313***
0.216*
0.705***
1.165***
0.0413***
0.188***
N
85
85
85
84
85
66
85
85
85
84
R2
0.85
0.89
0.93
0.93
0.91
0.96
0.92
0.91
0.89
0.93
Coefficient
Note: All regressions are in logs and include a time trend and month fixed effects. * p<.10, ** p<.05 and *** p<.01 Source: World Bank (2020a).
Electricity consumption may vary for other reasons than seasonal patterns and changes in economic activity. For example, it tends to be lower at holidays and higher at very high temperatures. A recent World Bank Policy Research Paper estimates a daily electricity consumption model that takes these factors into account and can explain 90 percent of the daily variation in electricity consumption in India (Beyer, Franco-Bedoya, and Galdo 2020).
Figure 1.6: Electricity consumption dropped strongly in March and is still below normal levels. Deviation of electricity consumption from normal levels Percent 5 0 -5 -10
Monthly average deviations January -3.0 February 0.2 March -10.2 April -24.3 May -14.4 June -9.2 July -6.3 August -8.1 September -5.9
-15 -20 -25
27-Sep
7-Sep
17-Sep
18-Aug
28-Aug
29-Jul
8-Aug
9-Jul
19-Jul
29-Jun
9-Jun
19-Jun
20-May
30-May
30-Apr
10-May
10-Apr
20-Apr
21-Mar
31-Mar
1-Mar
11-Mar
10-Feb
20-Feb
21-Jan
31-Jan
1-Jan
11-Jan
-30
Notes: The line plots deviations of electricity consumption from the predictions of a model based on seasonal patterns, holidays, and temperature. The last observation is October 15. Source: Updated estimates of Beyer, Franco-Bedoya, and Galdo (2020).
The deviation of actual electricity consumption from normal levels (i.e. the model predictions) is a measure for the economic drag due to the COVID-19 pandemic. The first meaningful deviation from normal levels was on March 22, when India observed a 14-hour long curfew that the Government of India implemented in all major cities and 75 districts with COVID-19 cases (Figure 1.6). Electricity consumption dropped further over the next few days and especially after the national lockdown was implemented on March 25. It was nearly 30 percent below normal levels at the end of March and remained a quarter below normal levels in April. When some restrictions were eased in May, electricity consumption recovered, but it remained below normal levels. On average, it was 14 percent below normal levels in May. Since then, the monthly averages fluctuate around 6 and 9 percent below normal, suggesting a lingering drag on the economy.
19
Beaten or Broken? Informality and COVID-19 Recent economic developments
Figure 1.7: Activity indicators troughed in April, but recovery has been gradual; levels are still below preCOVID times in all countries.
0
0
Bangladesh
India
Nepal
Pakistan
Sri Lanka
Bangladesh
Nepal
Pakistan
Jun-20
20
May-20
20
Apr-20
40
Mar-20
40
Feb-20
60
Jan-20
60
Jul-20
80
Jun-20
80
May-20
100
Apr-20
100
Mar-20
120
Feb-20
120
India
Aug-20
B. Restricted indicators give similar results.
Activity indicator for South Asian countries (reduced analysis) Index (January 2020 = 100)
Jul-20
A. Activity indicators co-move strongly.
Activity indicator for South Asian countries (extended analysis) Index (January 2020 = 100)
Jan-20
20
Sri Lanka
Variables included in the Activity Indicators Type of analysis
Reduced analysis variables
Additional variables in extended analysis
Bangladesh
India
• Google Mobility: Grocery and pharmacy • Google Mobility: Retail and recreation • Google Mobility: Workplaces • Electricity generation • Government tax revenue, USD mln
• Google Mobility: Grocery and pharmacy • Google Mobility: Retail and recreation • Google Mobility: Workplaces • Electricity generation • E-way bill
• Government tax revenue • Ready Made Garment exports • Imports vehicles
• Car registrations • Exports non oil • Industrial Production • IPI: Infrastructure & Construction Goods • Manufacturing PMI • Services PMI • Petroleum consumption • Port cargo traffic • Rail freight
Nepal
Pakistan
Sri Lanka
• Google Mobility: Grocery and pharmacy • Google Mobility: Retail and recreation • Google Mobility: Workplaces • Visitor arrivals
• Google Mobility: Grocery and pharmacy • Google Mobility: Retail and recreation • Google Mobility: Workplaces • Passenger vehicle sales
• Google Mobility: Grocery and pharmacy • Google Mobility: Retail and recreation • Google Mobility: Workplaces • Tourism receipts
• Domestic credit • Exports
• Crude steel production • Domestic credit • Exports • Industrial Production
• New car registrations • Manufacturing PMI • Services PMI • Industrial Production
Notes: To construct the activity indicators, meaningful high-frequency indicators were selected and indexed to January; the loadings from a principal component analysis were used as weights to compute a common score (the activity indicator). Sources: Google COVID-19 Community Mobility Reports, CEIC, and staff calculations.
the model and up-to-date high-frequency are used
quarter of this year and a subsequent rebound
to project the current trajectory of economic ac-
(Figure 1.8). An interpretation of the decline and
tivity (see Appendix A). The Quarterly Economic
especially a comparison across countries needs to
Indicators allow an assessment of growth in those
consider potential biases related to the selected
countries with long publication delays or no quar-
variables (see Appendix A). That said, the indica-
terly GDP data at all, and to nowcast the economic
tors suggest a steep drop in output and year-on-
dynamics during the third quarter of this year.
year contraction in the second quarter of this year in all countries. The decline was the largest in Mal-
The Quarterly Economic Indicators of all coun-
dives, where tourism halted, and COVID-19 spread
tries suggest a contraction of output in the second
fast. The Quarterly Economic Indicators for India,
Beaten or Broken? Informality and COVID-19 Recent economic developments
Figure 1.8: Quarterly Economic Indicators suggest that all countries have contracted in Q2 and are now bouncing back. High frequency economic activity in South Asia Percent, y-o-y 10
4
20
5
2
0
0
0
-20
-2
-40
-5 -10
-4
-60
-20
-6
-80
-25
-8
-100
-15
-30
2019Q4
2020Q1
2020Q2
July/Aug
10
2019Q4
2020Q1
0 -5 -10 -15 2019Q4
2020Q1
2020Q2 Nepal
2020Q2 Sri Lanka
5
-20
-10
India
July
8 7 6 5 4 3 2 1 0 -1
July
-120
2019Q4
5
2020Q1
2020Q2 July/August
Maldives
4 3 2 1 0 -1 2019Q4
2020Q1
GDP growth
-2 2019Q4 2020Q2 July/Aug Bangladesh Quarterly Economic Indicators growth
2020Q1
2020Q2 Pakistan
July
Notes: 2020Q1, 2020Q2 and 2020Q3 are out-of-sample predictions. For Bangladesh and Pakistan, the model is first estimated with annual GDP data and then the coefficients are used to predict quarterly GDP. Sources: CEIC, Haver Analytics, World Bank, national sources, and staff calculations.
Sri Lanka, and Nepal fell by double-digits as well.
activity. In March, the sum of made-made lights
In India, the contraction of the Quarterly Indica-
emitted by South Asia was 5 percent lower than a
tor was 20.6 percent, nearly identical to the fall in
year earlier (Figure 1.9.A). It was 7.5 percent low-
officially reported GDP, which was 23.9 percent
er in April, more than 10 percent lower in May
lower than a year before. In Nepal and Sri Lanka,
and June, and rebounded after that. Changes in
the Quarterly Economic Indicators fell by 14.5 per-
nighttime light growth suggest an even larger
cent and 10.3 percent, respectively. In Bangladesh
impact of the COVID-19 pandemic. In May, for
and Pakistan, the declines seem much smaller.
example, growth in nighttime lights was 17 per-
However, since for them the models are based on
centage points lower compared to the average
annual GDP, the fall of economic output may be
growth in May over the last three years. Since ur-
underestimated. For July and August, the Quarterly
ban areas emit much more lights than rural ar-
Economic Indicators point to a rebound across all
eas, changes in overall lights are mainly driven
countries, but the implied output is still lower than
by developments in cities. Since nighttime light
last year in India, Maldives, Nepal, and Sri Lanka.
data are available at high spatial granularity, they can be used to examine the effects of COVID-19
Spatially heterogeneous COVID-19 impacts
at the district level. As an alternative measure to the growth of overall lights, one can average the growth of lights across districts in South Asia. In this case, each district has the same weight in the aggregate measure, which allows for a better
Nighttime lights observed from space can be used
tracking of developments in more rural districts.
to assess the economic impacts of COVID-19.
In line with the COVID-19 pandemic first hitting
While luminosity during evening hours has been
urban areas in most countries, this measure de-
increasing consistently in the past, the COVID-19
clined and troughed later (Figure 1.9.B). In July
pandemic has darkened South Asia since March.
and August, when activity in many cities recov-
Nighttime lights are detected by satellites, and
ered, the average nighttime light intensity across
changes are strongly correlated with economic
districts in South Asia was still 22 percent and
21
22
Beaten or Broken? Informality and COVID-19 Recent economic developments
Figure 1.9: South Asia has become darker since March; initially cities were impacted, later rural areas also. A. Change of overall lights in South Asia
B. Average district-level change of lights
Growth of nightime light intensity Percent, y-o-y
Growth of average nightime light intensity Percent, y-o-y
10
10
5
5
0
0
-5
-5
-10
-10
-15
-15
-20
-20
-25 -30
-25 February
March
April
Average 2017-2019
May 2020
June
July
August
Difference
-30
February
March
April
Average 2017-2019
May 2020
June
July
August
Difference
Note: The raw data is cleaned to minimize temporary lights and background noise following Beyer, Franco-Bedoya, and Galdo (2020). Sources: VIIRS-DNB Cloud Free Monthly Composites (version 1) made available by the Earth Observation Group at the National Geophysical Data Center of the National Oceanic and Atmospheric Administration (NOAA), World Bank, and staff calculations.
10 percent below its level a year earlier, respec-
• Changes in mobility: From “Facebook Data for
tively. This shows that the severe impact of the
Good”, which utilizes information about Face-
COVID-19 pandemic was felt across districts in
book usage in specific areas, one can assess
the region and not just in urban centers. In In-
changes in mobility during the COVID-19 pan-
dia, nighttime lights declined less in districts with
demic at high spatial granularity. In South Asia,
higher previous out-migration, presumably be-
apart from a very few districts, average mobility
cause it predicts the extent of return migration.
between March and August declined strongly
Early in the pandemic, millions of migrant work-
(Figure 1.10.B). In most districts in Bhutan and
ers moved back from cities to their home villages.
Pakistan, mobility declined by less than 20 percent (92 percent and 85 percent of the districts,
While the impact of the COVID-19 pandemic is be-
respectively). In Nepal and Bangladesh, which
ing felt across districts in South Asia, some districts
enacted much stricter lockdowns, mobility de-
are hit much harder than others. The COVID-19 in-
clined more than 20 percent in nine out of ten
fection rate varies both across and within countries.
districts. In India, mobility declined strongly
While mobility declined in nearly all districts, the ex-
nearly everywhere: for around a third of the to-
tent depends both on national containment measures
tal districts the average decline was between 20
and local COVID-19 infections. Average nighttime
and 30 percent, for half of them it declined be-
light intensity between March and August declined
tween 30 and 35 percent, and for 15 percent it
in more than three quarters of South Asia’s districts.
declined even more. This heterogeneity can be explained both with country fixed effects cap-
• COVID-19 infections: Despite limited testing espe-
turing national lockdowns and differences in
cially in rural areas, most districts in South Asia
overall infections, as well as by local COVID-19
have confirmed COVID-19 infections (Figure
infections capturing higher infection risks and
1.10.A). While the number of infections per mil-
local restrictions.
lion people varies strongly at the national level,
• Changes in nighttime light intensity: In more than
nearly all countries have districts with less than
three quarters of districts, the average nighttime
50 cases per 100,000 people and districts with
light intensity between March and August was
more than 250 cases per 100,000. In Bangladesh
lower in absolute terms compared to last year
and India, less than 10 percent of the districts
(Figure 1.10.C). In a fifth of the districts, night-
have less than 50 cases per 100,000 people. In
time lights declined by more than 15 percent
Nepal and Pakistan this is true for around a third
during this period. Districts with such a large de-
of the districts, and in Sri Lanka, which has very
cline are found in all countries and their share is
low total numbers, all districts have less than that.
highest in Bhutan, Bangladesh, and India.
Beaten or Broken? Informality and COVID-19 Recent economic developments
Figure 1.10 The pandemic is not concentrated locally but impacts most areas in South Asia A. COVID-19 hit most areas
Covid-19 cases per million population <427 428 - 809 810 - 1271 1272 - 2431 > 2432 No data
B. Mobility declined strongly
Travel range (Avg. March - August) < -0.36 -0.35 - -0.29 -0.28 - -0.21 -0.20 - 0.00 > 0.0 No data
C. Most districts became darker
Nighttime light intensity (Avg. y-o-y growth, March-August) < -15.0 -14.9 - -8.0 -7.9 - -1.5 -1.4 - 0.0 >0
Notes: A. COVID-19 infections are the number of recorded infectious cases at the end of August standardized by population; B. mobility is measured as the average number of Bing tiles (0.6 km x 0.6 km) a Facebook user was present in during a 24-hour period compared to pre-COVID levels; C. nighttime light intensity is defined as the sum of lights standardized by area. The raw data is cleaned to minimize temporary lights and background noise following Beyer, Franco-Bedoya, and Galdo (2020). Sources: Health Ministries and Disease Control Centers, Facebook Data for Good (//dataforgood.fb.com) movement range maps, VIIRS-DNB Cloud Free Monthly Composites (version 1) made available by the Earth Observation Group at the National Geophysical Data Center of the National Oceanic and Atmospheric Administration (NOAA), and staff calculations.
23
24
Beaten or Broken? Informality and COVID-19 Recent economic developments
Table 1.4: Indian districts with higher COVID-19 infections experienced larger declines in mobility and nighttime light intensity during the national lockdown. Nighttime light intensity
Mobility: movement range
(1)
(2)
(3)
(4)
-2.409*** (0.397)
-2.592*** (0.458)
-2.933*** (0.194)
-1.803*** (0.203)
Socio-economic controls
NO
YES
NO
YES
Observations
624
623
619
618
0.056
0.070
0.271
0.408
Log COVID-19 cases per population
R-squared
Notes: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1; estimation is for April 2020; socio-economic controls include the manufacturing and service shares as well as previous in- and out-migration. Sources: All those mentioned for Figure 1.10, Yi et al. (2015), World Bank, and staff calculations.
Some of the observed heterogeneity in nighttime
credit growth (Figure 1.11.A). In Pakistan, the policy
light changes across districts can be explained with
rate was reduced from 13.25 percent in February to
voluntary reductions in mobility due to higher local
7.0 percent at the time of writing. In India, the pol-
prevalence of COVID-19. With more registered cases
icy repo rate has been reduced from 5.15 percent
of COVID-19, the perceived local infection risk ris-
to 4.0 percent. In addition, fiscal policy has been
es and in response risk-aversion may prompt peo-
countercyclical. Despite sharply declining tax rev-
ple to either follow the containment measures more
enue, spending has either increased or only fallen
strictly or voluntarily change their behavior beyond
minimally (Figure 1.11.B). All countries authorized
the measures (e.g. reduce their mobility complete-
significant fiscal support measures quickly. In Af-
ly). One may hence expect the economic impact in
ghanistan, authorities are spending an additional
districts with a higher prevalence of COVID-19 to be
2.9 percent of GDP toward pandemic-related mea-
larger, even if the restrictions are the same. To test this
sures, including for a relief package benefitting 90
hypothesis, one can study Indian districts during the
percent of Afghan households. India’s initial fiscal
national lockdown, when restrictions were uniform
support measures included higher direct spending
across the country. As expected, districts with more
(about 1.7 percent of GDP), foregone or deferred
COVID-19 cases per capita experienced larger de-
revenue (about 0.3 percent of GDP), and measures
clines in mobility and nighttime light intensity (Table
unrelated to expenditures and revenue designed to
1.4). While less than 10 COVID-19 cases per million
support businesses and shore-up credit provision
residents were associated with a 3.7 percent points
(about 4.9 percent of GDP). Additional support was
larger decline in light intensity compared to districts
provided through changes to financial regulation.
without any cases, more than 50 COVID-19 cases per
All countries introduced mandatory credit repay-
million residents were associated with a 12.6 percent-
ment moratoria and relaxed provisioning rules for
age points larger decline. This has strong implications
non-performing exposures. In addition, all coun-
for the rebound of the economy. Without effectively
tries engaged in some form of liquidity support.
reducing the risk of a COVID-19 infection, voluntary
Some also put in place restrictions on the use of
reductions of mobility make it unlikely that the econ-
profits and resources (Bangladesh, India, Sri Lanka)
omy will return to full potential even when restric-
and eased limits on large exposures (India, Paki-
tions are relaxed. This may explain why the recovery
stan). In India, the risk weights for credit with public
has recently slowed in some parts of South Asia.
guarantees have been lowered. To support borrowers, countries provided loans (often through state-
Strong policy measures preserved macro-financial stability ‌
owned banks) to affected companies and sectors (Bangladesh, Bhutan, India, Maldives, Pakistan, Sri Lanka), subsidies to borrowers to facilitate repayments (India), and state guarantees on private-sector loans (Bangladesh, India, Pakistan, Sri Lanka).
In response to the economic turmoil, South Asian countries proactively stabilized economic activi-
Confronted with a shock of unprecedented scale,
ty through monetary easing, fiscal stimulus, and
South Asian countries have preserved macro-fi-
supportive financial regulation. Across the region,
nancial stability thus far. After large outflows in
central banks lowered their policy rate to support
March, capital flows were positive over the last
Beaten or Broken? Informality and COVID-19 Recent economic developments
Figure 1.11: Monetary and fiscal stimulus have supported economic activity. A. Policy rates have been reduced.
B. Fiscal policy has been countercyclical.
Policy rate Percent 15
Change in revenue and expenditure Percent, y-o-y 10 0
13
-10 11
-20 -30
9
-40 7
-50 -60
5
Bangladesh
India
Pakistan
Sep-20
Aug-20
Jul-20
Jun-20
May-20
Apr-20
Mar-20
Feb-20
Jan-20
-70 3
Sri Lanka
-80 Bangladesh (April)
India (July) Revenue
Maldives (July) Expenditure
Sources: A. Haver Analytics and national sources; B. CEIC and staff calculations.
months. With imports declining faster than ex-
account deficit narrowed from 1.7 percent of GDP
ports, given weak domestic demand and low oil
in FY18/19 to 1.5 percent in FY19/20, as a sharp de-
and commodity prices, South Asia’s terms of trade
cline in exports was offset by a – likely temporary
improved, and current account deficits narrowed
– surge in remittance inflows. In Sri Lanka the cur-
or turned to surpluses. As a result, international re-
rent account deficit is estimated to have narrowed
serves rose, which contributed to external stability.
in the first half of 2020 despite reduced receipts
Government bond yields remained constant or de-
from remittances and tourism, as stringent import
clined following monetary easing by central banks.
restrictions curbed imports. Trade disruptions in
While stock prices plummeted in March and early
Nepal led to a 19.7 percent drop in imports, sig-
April, they regained ground subsequently.
nificantly narrowing the current account deficit in FY19/20. The sharp drop in imports outweighed
• Capital flows: Capital flows to India and Pakistan
both a contraction in exports and a decline in re-
were positive over the last months (Figure 1.12.A).
mittance inflows. In Bhutan, imports fell more
As in many other EMDEs, India and Pakistan saw
than exports as well, which reduced the current
capital outflows in March, amid high uncertainty
account deficit to an estimated 14 percent of GDP
about the COVID-19 pandemic and its economic
in FY19/20 (down from 22.5 in FY18/19).
implications. However, following massive quan-
• Government bond yields: Due to strongly declining
titative easing in advanced economies, and espe-
interest rates in Pakistan, the yield of Pakistan’s
cially in the United States, capital inflows quickly
3-year investment bond nearly halved, from 12.0
resumed. In India, strong net foreign investment
percent in February to 7.2 percent in July (Figure
inflows increased foreign reserves to reach a re-
1.12.C). Yields also declined somewhat in Bangla-
cord high of USD 545 billion in the week that ended September 18.
desh and have been stable in the other countries. • Stock prices: Stock prices started falling in early
• Current accounts: Current account deficits nar-
March, in line with stock markets in advanced
rowed or turned into surpluses across the region
economies and before domestic containment
(Figure 1.12.B). In India, a large decline in imports
measures were enacted (Figure 1.12.D). Most
(both volume and prices) more than offset a drop
stock indices troughed in mid-April. They
in exports, so that the current account turned to
fell most strongly in India (losing 30 percent
a surplus in the first half of FY20/21. Similarly,
of their value compared to the beginning of
the current account deficit shrunk from 4.8 per-
March), and the least in Bangladesh, where they
cent of GDP in FY18/19 to 1.1 percent of GDP in
“only” lost 15 percent. They recovered subse-
FY19/20 in Pakistan, driven mainly by import val-
quently: in Sri Lanka they are now 7.8 percent
ues falling 19.3 percent. In Bangladesh, the current
below their levels at the beginning of March
25
Beaten or Broken? Informality and COVID-19 Recent economic developments
Figure 1.12: Portfolio flows have recovered, and current accounts improved; government bond yields are stable or declining and stock prices rebounded after large losses. A. Capital flows are positive again.
B. Current accounts improved.
India India
-2500 -2500
-6 -6
Aug-20 Aug-20
Jul-20 Jul-20
Jun-20 Jun-20
May-20 May-20
Apr-20 Apr-20
Mar-20 Mar-20
Feb-20 Feb-20
-20000 -20000
Pakistan (RHS) (RHS) Pakistan
C. Government bond yields are stable or declining.
Government Government bond bond yields yields (long-term) (long-term) Percent per annum Percent per annum
India India
Nepal Nepal
D. After falling strongly, stock prices regained ground.
Stock Stock prices prices Index Index (Mar (Mar 2020 2020 == 100) 100) 115 115 110 110 105 105
15 15
100 100 95 95
10 10
90 90 85 85
70 70 65 65
Pakistan Pakistan
Bangladesh Bangladesh Feb-20 Feb-20
Maldives Nepal Maldives Nepal Latest observation Latest observation
02-Mar 02-Mar 10-Mar 10-Mar 18-Mar 18-Mar 26-Mar 26-Mar 03-Apr 03-Apr 13-Apr 13-Apr 21-Apr 21-Apr 29-Apr 29-Apr 07-May 07-May 15-May 15-May 25-May 25-May 02-Jun 02-Jun 10-Jun 10-Jun 18-Jun 18-Jun 26-Jun 26-Jun 06-Jul 06-Jul 14-Jul 14-Jul 22-Jul 22-Jul 30-Jul 30-Jul 07-Aug 07-Aug 17-Aug 17-Aug 25-Aug 25-Aug 02-Sep 02-Sep 10-Sep 10-Sep 18-Sep 18-Sep 28-Sep 28-Sep
80 80 75 75
55
00
Bangladesh Bangladesh
2020Q2 2020Q2
-15000 -15000
2020Q1 2020Q1
-2000 -2000
-3 -3 -4 -4 -5 -5
-1500 -1500
2019Q4 2019Q4
-10000 -10000
-2 -2
2019Q3 2019Q3
-1000 -1000
2019Q2 2019Q2
-500 -500
-5000 -5000
00 -1 -1
2019Q1 2019Q1
00
2018Q4 2018Q4
500 500
00
22 11
2018Q3 2018Q3
1000 1000
5000 5000
Current Current account account balance balance Percent Percent of of GDP GDP 33
2018Q2 2018Q2
USD, USD, million million 1500 1500
2018Q1 2018Q1
Total Total portfolio portfolio flows flows of of India India and and Pakistan Pakistan USD, million USD, million 10000 10000
Jan-20 Jan-20
26
S&P500 S&P500 BSE: Sensitive Sensitive (Sensex) (Sensex) BSE: DSEX Index Index DSEX
Colombo Stock Stock Exchange Exchange (ASPI) (ASPI) Colombo Karachi Stock Exchange 100 Karachi Stock Exchange 100
Notes: B. Quarterly GDP for Bangladesh, Pakistan and Nepal were derived from annual GDP and assumed to be constant for all four quarters; D. Gaps in stock prices are due to market closures. Sources: A. IIF; B. Trading Economics, Haver Analytics and staff calculations; C. CEIC; D. Haver Analytics.
and in India they are 4.6 percent lower. In Paki-
recent economic difficulties and relaxed pruden-
stan and Bangladesh, they are now even above
tial regulation threaten financial stability. Due to
their early Match levels, by 3.4 percent and 7.6
loan moratoria introduced across the region, the
percent, respectively.
share of reported non-performing loans may remain stable for now; but these “bad loans” will
… but the situation is fragile, and policy tools have been exhausted
eventually erode capital buffers and are already impeding lending. The resolution of non-performing assets to foster credit growth, while limiting moral hazard and containing fiscal risks, will be difficult and require substantial policy dia-
The COVID-19 pandemic has exacerbated fi-
logue. Financial sector challenges are particular-
nancial sector vulnerabilities. In some countries,
ly severe in Bangladesh due to deviations from
Beaten or Broken? Informality and COVID-19 Recent economic developments
international regulatory and supervisory stan-
to the higher estimated fiscal deficit in Bhutan. In
dards, the absence of a bank resolution frame-
Nepal, lower revenues were partly offset by reduced
work, the introduction of interest rate caps, and
budget execution rates following disruptions asso-
weak governance in state-owned banks. Public
ciated with the pandemic. In Pakistan, the fiscal
banks can be used to support private credit in cri-
deficit narrowed somewhat, but less than planned
ses, but across the region public banks entered the
at the beginning of the year due to a fiscal stim-
crisis with weak balance sheets and severe gover-
ulus to fight the pandemic. The situation is most
nance issues (World Bank 2020b).
problematic in Sri Lanka and Maldives. In Sri Lanka, the fiscal accounts deteriorated in the first four
Rising inflation constrains future monetary eas-
months of 2020. Tax revenues fell short due to the
ing. In India, after reaching 4.8 percent in FY19/20,
fiscal stimulus package implemented in November
headline inflation averaged 6.7 percent during
2019, which included a reduction of the VAT rate
April-July 2020 due to strong supply-chain dis-
and an increase of the registration threshold, and
ruptions. After cutting the repo rate by a cumu-
severe disruptions in economic activity. As a result,
lative 115 bps between March and May and main-
despite a moderation in public investment, the
taining significant excess liquidity in the market,
overall budget deficit increased. Approximately 40
the Reserve Bank of India (RBI) paused further
percent of the deficit was financed by central bank
monetary easing in August. In Bhutan, headline in-
credit. In Maldives, fiscal imbalances have widened
flation accelerated to 7.6 percent in July 2020, driv-
significantly as well, as revenues and grants collect-
en by food prices and reflective of similar trends
ed between January and July halved compared to
in India, Bhutan’s largest trading partner. Despite
the corresponding period in 2019 while spending
weak activity, inflation rose also in Pakistan (most-
remained mostly constant (see Box 1.5).
ly due to rising food prices), so that the State Bank of Pakistan halted its determined easing cycle
Since the global financial crisis, there has been an
and kept the policy rate unchanged in September
increase in the share of debt from private financing
2020. Food prices increased due to supply chain
through bond markets. Since 2010, debt financing
disruptions also in Bangladesh, though non-food
through bond markets and commercial banks has
prices declined due to lower demand. In Nepal,
increased, while the share of official financing has
food prices increased first due to an export ban on
declined (Figure 1.14.A). For middle-income coun-
onions by India and later because of localized food
tries like India, relying more on international bond
shortages resulting from transport disruptions.
markets is a sign of healthy finances. For lower-in-
In Afghanistan, panic buying and import disrup-
come countries that access concessional financing
tions in March and April also triggered a significant
through the International Development Associa-
spike in food prices. As the government adopted
tion, however, official creditors have discouraged
administrative measures to prevent price gouging
the use of commercial debt and even set exposure
and distributed emergency wheat supplies, food
limits. One concern is that debt reduction initia-
inflation moderated to 12.8 percent year-on-year
tives could de facto lead to more lending on com-
as of end-June. The rising food prices had large
mercial terms, effectively creating a transfer from
distributional impacts and hit South Asia’s poor
official creditors to private creditors.
the hardest (see Chapter 3). Debt vulnerability is increasing in many countries Fiscal deficits are rising across the region amid col-
and especially in Sri Lanka and Maldives. In Sri Lan-
lapsing revenue. In Afghanistan, with the onset of
ka, the central government debt-to-GDP ratio rose
the COVID-19 crisis, revenue performance deteri-
to over 90 percent as of end-April 2020 (from 86.8
orated significantly and revenue estimates for 2020
percent at the end of last year), with more than half
were revised downward. Total domestic revenue
of the debt denominated in foreign currency (Fig-
collection at end-June was 20 percent lower than
ure 1.14.B). Citing limited fiscal buffers and exter-
the initial budget target. In Bangladesh, the fiscal
nal vulnerabilities, Fitch and S&P downgraded the
deficit in FY19/20 was estimated at 8.2 percent of
sovereign rating to B-. In Maldives, the total public
GDP, exceeding the budget target amid depressed
and publicly guaranteed debt rose significantly as
revenue, along with higher expenditures on social
well and is forecast to rise quickly (see Chapter 2).
protection programs and healthcare. In Bhutan
In both countries low international reserves are not
and Nepal, the fiscal deficit also increased. Addi-
providing an adequate buffer. In Sri Lanka, despite
tional COVID-19 related expenditure contributed
a swap facility of USD 400 million with the RBI, a
27
Beaten or Broken? Informality and COVID-19 Recent economic developments
Box 1.5 Worrying fiscal implications of shuttered tourism in Maldives Tourism is the main driver of Maldives’ government revenues. With the number of tourists increasing almost four times between 2000 and 2019, tourism-related revenues rose steadily from USD 65 million to USD 690 million and contributed 48.5 percent of total government revenues (excluding grants) last year. The main bulk of tourist-related revenues comes from the Tourism Goods and Services Tax (GST), as well as duties from imported food and fuel for tourist consumption. The introduction of new revenue sources such as green taxes (in 2015) and airport development fees (in 2017) have also contributed. The standstill of tourism has triggered a devastating impact on revenues. The COVID-19 pandemic led the Maldives to close its borders on March 27 and for the entire second quarter of 2020. As a result, tourism-related revenues fell by 77.3 percent (y-o-y) and total revenues excluding grants plummeted by 75.6 percent in the second quarter (Figure 1.13.A), making it impossible to meet the government’s revenue target of USD 1,900 million. However, even before COVID-19, the growth of tourism revenue had slowed. In 2019, even though visitor arrivals grew by 14.7 percent, revenue from the tourism GST remained nearly constant, which slowed overall revenue growth from 7.5 percent in 2018 to 2.8 percent in 2019. Given the record number of tourists, the stagnant revenue from the tourism GST could indicate some under-collection of taxes due to the use of online booking companies located offshore (IMF 2019).
Figure 1.13: The collapse of tourism has led to a large revenue shortfall. A. Revenue plummeted due to the collapse in tourism.
B. The revenue shortfall exceeds expenditure cuts.
Growth of tourist arrivals and revenue Percent, y-o-y
Revenue and expenditure from Jan to Aug 2020 Share of budgeted revenue/expenditure
40
0.5
20 0.4
0 -20
0.3
-40 -60
0.2
-80 0.1
p-1 9 No v-1 9 De c-1 9 Jan -20 Feb -20 Ma r-2 0 Ap r-2 0 Ma y-2 0 Jun -20 Jul -20 Au g-2 0
-100
Se
28
Tourism revenue
Total government revenue
Tourist arrivals
0 Revenue
Expenditure
Notes: A. Tourism revenue is the sum of tourism goods and services tax, green tax, airport service charge, airport development fee, rents from resorts, and an estimate of tourism-linked import duties. B. The budgeted revenue and expenditure is for the entire year. Sources: Maldives’ Ministry of Finance, Haver Analytics, and staff calculations.
Measures to mitigate the impact on fiscal and debt sustainability may be insufficient. The government has taken some steps to reprioritize public expenditure. In mid-March, the government announced cuts to recurrent and capital expenditures amounting to USD 65 million and provided a further USD 90 million in support to households and firms suffering from the effects of the pandemic. However, expenditures have not adjusted in line with the large shortfall in revenues. From January to August of this year, Maldives spent almost half of its planned expenditure for the year, which was 2 percent more than the same period in 2019. At the same time, it collected only a third of budgeted revenue, only 60 percent of the amount in the corresponding period of 2019. As a result, the overall fiscal deficit is estimated to widen significantly from 6.4 percent of GDP in 2019 to 20.5 percent of GDP in 2020. Moreover, Maldives remains at high risk of debt distress. Exacerbated by the current shock, total public and publicly guaranteed debt rose to USD 4.8 billion as of end-June 2020, a significant increase from USD 4.4 billion as of end-2019.
Beaten or Broken? Informality and COVID-19 Recent economic developments
Given the prolonged and uncertain nature of the COVID-19 shock, further adjustments to government expenditures may be necessary to mitigate fiscal and debt sustainability risks. Although borders have reopened to tourists since July 15, tourism has far from returned to normalcy. Only 13,787 tourists visited between July 15 and September 15, a 95 percent decline compared to last year. Capital spending increased by 16.7 percent y-o-y in the first half of the year, mainly due to land reclamation and harbor reconstruction projects. Larger reductions in spending, for example by delaying large public infrastructure investments that are not urgently needed, could help Maldives manage its mounting debt and fiscal challenges.
Figure 1.14: The share of debt financed through bond markets has increased and debt vulnerability is high in some countries. A. Debt financing through bonds has increased
B. Some countries face debt vulnerabilities
Total reserves (percent of external debt)
Long-term public debt as a share of GNI Average annual changes 0.3
500
0.2
400
0.1
Afghanistan
300
Nepal
0 200 -0.1
Bhutan
Bangladesh Maldives
100
-0.2
India 0
-0.3 Multilateral
Bilateral
Bonds 2010-2016
Commercial banks
Pakistan Other private
2016-2018
-100 -5
0
Sri Lanka
5 10 15 20 Short-term debt (percent of external debt)
25
Notes: B. The bubble size shows total external debt-to-GDP ratio in 2018 for Bhutan, Maldives, and 2019 for the rest. Sources: International Debt Statistics, World Bank, and staff calculation.
loan of USD 500 million from the China Develop-
from USD 311 million at end-January to USD 122
ment Bank, and a repo facility with the New York
million as of end-August, equivalent to only 0.5
Federal Reserve Bank for USD 1.0 billion, official
months of 2019 goods imports. To help maintain
reserves remain low relative to short-term external
exchange rate stability, the Maldives Monetary Au-
liabilities. In Maldives, as foreign exchange earn-
thority already activated a USD 150 million foreign
ings from tourism plummeted, usable reserves fell
currency swap with the RBI.
29
30
Beaten or Broken? Informality and COVID-19 Recent economic developments
Conclusion
second quarter of this year in all countries and a subsequent gradual recovery. The collapse in ac-
Amid a sudden and steep economic impact from
tivity was widespread within countries. Between
the COVID-19 pandemic, governments enacted
March and August, nighttime light intensity de-
strong policy measures to preserve macro-financial
clined in more than three quarters of South Asia’s
stability, but the situation is fragile. The COVID-19
districts and in August, the average nighttime light
pandemic is not yet under control in South Asia,
intensity was still 10 percent below its level a year
despite early containment measures. The crisis
earlier. Amid the economic turmoil, South Asian
brought South Asia to a near standstill. Travel re-
governments proactively stabilized economic ac-
strictions prevented travelers from reaching South
tivity through monetary easing, fiscal stimulus,
Asia and lockdown measures triggered massive
and supportive financial regulation. For now, mac-
supply disruptions. Information from high-fre-
ro-financial stability has been preserved. However,
quency variables, combined in activity indicators,
the situation is fragile amid weak buffers and ex-
show an unprecedented contraction in March and
hausted policy tools in some countries. Regulatory
April. Activity recovered subsequently across the
adjustments to the COVID-19 pandemic have ex-
region, but it remained below pre-COVID levels in
acerbated financial sector vulnerabilities, and fis-
August. High-frequency approximations of GDP
cal stimulus despite large revenue shortfalls have
suggest year-over-year contractions during the
resulted in rising fiscal deficits.
Box 1.6 Views from the South Asia Economic Policy Network The South Asia Economic Policy Network represents an attempt to engage more strongly with thinkers and doers across South Asia, to nurture the exchange of ideas and to foster learning from colleagues and counterparts in the region. Consisting of nearly 500 members, the Network includes researchers and experts from seven South Asian countries, selected based on peer recognition, recent conference presentations, and research outputs. Many of them are academics at renowned universities, others are researchers in central banks and think tanks, and some are affiliated with policy-making units. As in the last four editions of this report, a short opinion survey of Network members was conducted. The objective was to take the pulse of informed and dedicated experts about the economic developments in their countries. We also used this opportunity to understand their assessment of the COVID-19 pandemic, and how it impacts the informal economy. We received 70 completed questionnaires from 6 countries. Almost all respondents identified themselves as academics, around 85 percent as macroeconomists, two-thirds as policy advisors, and around 15 percent as policy makers.
Figure 1.15: Network members maintain that containment measures and border closures had the largest economic impact. How did each of the following factors contribute to the economic disruption since March of this year? Share of responses Containment measures (e.g. lockdown; quarantine) imposed in my country Border closures and travel restrictions imposed by my country Border closures and travel restrictions imposed by other countries Containment measures (e.g. lockdown; quarantine) imposed by other countries People changed their behavior due to the COVID-19 pandemic 0 A lot
10 A bit
20 30 Not at all
40
50
60
70
80
90
100
Source: South Asia Economic Policy Network, survey conducted for this report.
The economic disruption documented in this chapter is mostly attributed to lockdowns and border closures and many disapprove the authorities’ management of the crisis. This chapter discussed the stringent containment measures that South Asian countries imposed to control the domestic spread of COVID-19 (Figure 1.2). Over 75 percent of the respondents maintain that lockdowns and quarantine measures have strongly contributed to the economic disruption and the rest acknowledges some impact (Figure 1.15). Over
Beaten or Broken? Informality and COVID-19 Recent economic developments
half of the respondents assert strong impacts also of the travel restrictions, both for those imposed by their own country and those imposed by other countries. Somewhat contrary to the evidence presented in this chapter (Box 1.1 and Table 1.4), only 40 percent appreciate strong effects from behavioral changes, though nearly everyone agrees that they matter somewhat. Around two-thirds of the respondents believe that their respective governments are not doing the best at supporting economic recovery or containing the health crisis (Figure 1.16). The experts rate the management of the health crisis somewhat better than the management of the economic turmoil.
Figure 1.16: Around two-thirds of the respondents feel that the governments aren’t doing their best to contain the economic and health crisis. How do you think the authorities in your country are doing in Share of responses Supporting the economic recovery
Containing the health crisis 0
10
20
Very good job
30
40
Somewhat good job
50
60
Could do better
70
80
90
100
Not a good job
Source: South Asia Economic Policy Network, survey conducted for this report.
Respondents are concerned about the role of the informal sector but seem upbeat about the future opportunities of the digital economy. Very much in line with Chapter 3 of this report, most respondents contend that the large informal sectors in South Asia magnify the impact of COVID-19. Eight out of ten expect informality to either significantly or somewhat significantly amplify the economic costs of the pandemic as the informal sector is largely unregulated and unprotected (Figure 1.17). In addition, there is a strong sense that the current crisis will increase the size of the informal economy. In line, Chapter 3 documents that jobs created now are even more likely than before to be informal. Experts are upbeat about the opportunities of the digital economy but see challenges in the short run. Eight out of ten allege that digitalization already helps weathering the current crisis, but also that too few people have the skills to operate in the digital economy, which hence is increasing inequality. More than nine out of ten call for more investment to expand the digital infrastructure and emphasize the need to support education transferring the skills needed in the digital economy.
Figure 1.17: There is a strong consensus that a large informal sector in South Asia magnifies the impacts of COVID-19. Do you think that a large informal sector in South Asia magnifies or contains the impacts of the COVID 19 pandemic? Share of responses It significantly magnifies the impacts of COVID-19 because it is unregulated and unprotected It somewhat magnifies the impacts of COVID-19 because it is unregulated and unprotected It significantly mutes the impacts of COVID-19 because it can adapt easily to different circumstances of the pandemic It somewhat mutes the impacts of COVID-19 because it can adapt easily to different circumstances of the pandemic It makes no difference to the impacts of COVID-19 0
10
20
30
40
50
60
70
Source: South Asia Economic Policy Network, survey conducted for this report.
Experts rank an expansion of social assistance and higher cash-transfers as by far the most effective policy tools (Figure 1.18). It is followed by higher investment in health and public work programs, which more than four out of ten rank as either the highest or second-highest priority. Not too far behind, experts see merit in debt relief and credit extension to private formal firms. Very few of the experts rank more flexible labor laws a priority. In line with the top priority of South Asian experts, Chapter 3 highlights the need for providing better social protection to the ‘missing middle’.
31
Beaten or Broken? Informality and COVID-19 Recent economic developments
Figure 1.18: Expansion of social assistance is ranked the most effective policy tool for recovery, while more flexible labor laws are ranked lowest among South Asians. Which are the most effective tools policy makers can use to support the recovery? Please rank them (1=most effective; 5=least effective) Expand social assistance, increase cash transfers Increasing public investment in health sector Public works Debt relief and credit extension to private formal firms More flexible labor laws to encourage firms to hire and fire easily and enable workers to move to more productive industries/cities
0
1
10 2
3
20 4
30
40
50
60
70
80
90
100
5
Source: South Asia Economic Policy Network, survey conducted for this report.
Over the next six months, experts expect rising fiscal deficits and financial sector stress. The expectations of Network members regarding economic developments over the next six months are summarized in a single number, using so-called diffusion indices. For any indicator, a value above 50 indicates that an increase is expected, whereas a value below 50 corresponds to an expected decrease. The farther away the number is from 50, the greater the consensus among Network members that an important change is under way. Across all countries respondents strongly anticipate a continuation of the monetary policy easing with even lower interest rates and even higher inflation (Figure 1.19). Apart from Pakistan, they also expect a further deterioration of trade. Network members strongly agree that fiscal deficits will increase over the next six months and that financial sector stress will increase, which is line with the projections in Chapter 2. However, there is a silver lining. Compared to the expectations six months ago, when the COVID-19 outbreak had just hit the region, experts now seem a bit less pessimistic about the economic outlook. In line, around two thirds of respondents expect tourism to return in 2021. The survey also offered room to express general views on the current situation. In line with the focus of Chapter 3, experts across all countries are worried about the impact of the pandemic on livelihoods and inequality. Experts from India and Bangladesh, for example, expressed concerns regarding the rising inequalities in access to health, social safety and education.
Figure 1.19: Experts expect a continuation of the policy support and economic disruption. What do you expect to happen in your country within the next six months? Diffusion Index 100 90 80 Increase
70 60 50 40
Decrease
32
30 20 10 0
Real GDP growth
Headline inflation India
Interest rates
Volume of imports Volume of exports
Pakistan
Bangladesh
Fiscal deficit
Exchange rate
Financial sector stress
Others
Sources: South Asia Economic Policy Network and staff calculations. Note: The index is calculated as follows: Index=(P1*100) + (P2*50) + (P3*0), where P1 is the proportion of responses that report that the variable will increase, P2 is the proportion of responses that report that the variable will remain unchanged, and P3 is the proportion of responses that report that the variable will decline. The lines above the bars indicate the responses from six months ago.
Beaten or Broken? Informality and COVID-19 Recent economic developments
References Azevedo, J.P., Hasan, A., Goldemberg, D., Aroob Iqbal, S., & Geven, K. (2020). Simulating the Potential Impacts of COVID-19 School Closures on Schooling and Learning Outcomes A Set of Global Beyer, R. C. M., Franco Bedoya, S., & Galdo, V. (2020). Examining the Economic Impact of COVID-19 in India through Daily Electricity Consumption and Nighttime Light Intensity. World Bank Policy Research Working Paper, 9291. Hale, T., Petherick, A., Phillips, T., & Webster, S. (2020). Variation in government responses to COVID-19. Blavatnik School of Government Working Paper, 31. Hastie, T., Tibshirani, R., & Friedman, J. (2009). The elements of statistical learning: data mining, inference, and prediction. Springer Science & Business Media. International Monetary Fund. (2019). Technical Assistance Report—Reform Options to Strengthen Tax Policy. International Monetary Fund, Washington, DC. Meinshausen, N. (2007). Relaxed lasso. Computational Statistics & Data Analysis, 52(1), 374-393. World Bank (2020a). India Development Update. The World Bank. Washington, DC. World Bank (2020b). South Asia Economic Focus, Spring 2020: The Cursed Blessing of Public Banks. The World Bank. Washington, DC. World Bank. (2020c). Maldives Development Update: In Stormy Seas. The World Bank. Washington, DC. Li, Y., Rama, M., Galdo, V., & Florencia Pinto, M. (2015). A Spatial Database for South Asia. World Bank, Washington, DC.
33
34
Beaten or Broken? Informality and COVID-19 Recent economic developments
Appendix A. Quarterly Economic Indicators based on LASSO regressions With many variables at hand to track the GDP dynamics, it becomes an increasingly difficult task to select those that do a better job in explaining these dynamics in the short term. Fortunately, machine learning techniques can facilitate the model selection for GDP nowcasting. The Least Absolute Shrinkage Selection Operator or LASSO regression (Tibshirani 1996) is well suited for building simple nowcasting models and Quarterly Economic Indicators when the number of potential explanatory variables is large and the number of observations small. LASSO regression is a shrinkage method that aims to reduce the values of some coefficients of an ordinary least squares (OLS) regression toward 0. In other words, the LASSO regression reduces the number of explanatory variables included in the regression model, i.e. it drops variables containing little additional information. The advantage of this shrinkage method is that the estimated models exhibit more precise out-of-sample predictions than least squares estimates (since it does not overfit the training data). LASSO regression is based on a loss function that starts by minimizing the squared errors as an OLS regression but adds a penalty term that penalizes the sum of absolute values of the coefficients. The importance of the latter is controlled by a penalty parameter λ. The LASSO regression loss function is defined as follows:
ˆ ) L lasso (β
ˆ ) 2 + λ∑ mj‾1 |β ˆ |, = ∑ ni‾1 (yi − x ′i β j
where β ˆ is the vector of coefficients, yi is the dependent variable and xi the set of potential explanatory variables. The LASSO regression carries out the variable selection and parameter estimation simultaneously while keeping computational costs low (Hastie, Tibshirani, and Friedman, 2009). The parameter λ specifies the weight applied to the penalty term. When λ=0, the linear LASSO reduces to
the OLS estimator. As λ increases, the magnitude of all the estimated coefficients is shrunk toward zero.
ˆ increases with higher penalty terms. This shrinkage occurs because the cost of each nonzero β
The optimal penalty parameter λ is determined by cross-validation, which is a resampling technique. The
method divides the data set randomly into k different subsets (k is typically 10). Keeping one of the subsets
as the validation set, the model is trained over the remaining k-1 sets for a range of values for λ. It selects
the parameter that minimizes the average mean squared error (MSE), where the MSE is computed using each of the k subsets as a validation set for each possible value of λ .
The non-zero coefficients of the LASSO regression are biased because they are shirked towards zero. Hence, after the variables have been selected using the LASSO regression, the selected covariates are included in a linear model estimated by OLS, in line with the method known as relaxed LASSO (Meinshausen 2007). For India, Maldives, Nepal, and Sri Lanka, we use quarterly GDP series to select variables. Many monthly and daily variables that potentially predict GDP are collected and aggregated to quarterly frequency. In the case of Bangladesh and Pakistan, the selection is based on annual GDP series, since no quarterly series are available. For Afghanistan and Bhutan, the procedure is unable to identify meaningful high-frequency indicators. While in normal times it may be useful to consider both contemporaneous and lagged relationships, in an economic crisis like the current one a model considering only the former is more helpful.
Beaten or Broken? Informality and COVID-19 Recent economic developments
Both the number of variables considered as well as the number of variables selected vary across countries (Table 1.4). For example, 65 variables are considered for India, but only 19 for Maldives. The number of selected variables ranges from six for Sri Lanka to three for Maldives. The country-specific models have a reasonably good fit, with R squares ranging from 0.52 in the case of Nepal to 0.79 for India. In order to assess the current economic activity, the model is estimated until the last quarter of 2019 for those countries reporting quarterly GDP and until the last fiscal year for the others. The quarterly trajectory for 2020 is then projected out-of-sample based on these estimates and up-to-date high-frequency indicators. The underlying assumption is that the structural relationship between the explanatory variables and GDP growth did not change significantly during the COVID-19 pandemic. However, since the economic contractions were associated with unprecedented containment measures and larger than usual business cycle fluctuations, the results should be interpreted with caution. On the one hand, the Quarterly Economic Indicator may fall more than GDP, for example in countries depending a lot on tourism. For them, the estimated relationship between tourist arrivals and GDP in the past may exaggerate the effect from a complete collapse in tourist arrivals, which is unprecedented. On the other hand, the Quarterly Economic Indicator may understate the effect, for example because the estimated impact of changes in high-frequency indicators is relatively weak. This is a concern especially for Bangladesh and Pakistan, where the relationships are based on annual GDP. For India, the Quarterly Economic Indicator in the second quarter of this year dropped nearly as much as GDP. The former contracted by 20.6 percent (y-o-y), compared to 23.9 percent contraction of officially reported GDP. The most recent projection is for either July or July/August depending on the availability of the selected high-frequency indicators. If for one of these months only some variables are available, the missing ones are predicted based on those available.
Table A.1.1: Economic indicator models for South Asian countries Countries with quarterly GDP data
# variables considered
Variables selected by LASSO
R2 Time
Countries with annual GDP data
India
Maldives
Nepal
Sri Lanka
Bangladesh
Pakistan
65
19
22
33
33
37
• Cash on hand • Crude steel prod. • Industrial production • Pass. vehicle sales • Petrol. consumption • Rail freight
• "Exports • Imports • Visitor arrivals"
• • • • •
Broad money Exchange rate Exports Imports Foreign reserves
• Electricity • Exports • Industrial production • Policy rate • Remittances • Visitor arrivals
• Industrial Production
• Industrial production • Exports • Remittances
0.79
0.70
0.52
0.72
0.59
0.65
2008Q1 - 2020Q2
2012Q3 - 2020Q2
2005Q2 - 2020Q2
2011Q1 - 2020Q2
1997 - 2019
1996 - 2019
35
36
Beaten or Broken? Informality and COVID-19
2
Chapter
With a new world, a changing outlook
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Summary South Asia’s GDP is expected to contract 7.7 percent this year, by far the largest decline on record. All countries in the region find themselves in a dire situation. Maldives, heavily dependent on tourism, is expected to experience a contraction of 19.5 percent. India’s GDP is expected to contract 9.6 percent in the fiscal year that started in March. Even with the baseline forecast of a rebound next year, South Asia’s per-capita GDP in 2021 would still be 6 percent below its level in 2019. The forecast of the region’s GDP in 2021 is now 15 percent below what we forecast in the Fall of 2019, before the spread of COVID-19. This change in the forecast illustrates higher global contagion than expected earlier and the devastating economic impact of the pandemic. The impact on livelihoods will even be larger than the GDP forecast suggests. Unlike earlier recessions in which investment and exports led the downturn, consumption, traditionally the most stable component of demand, has been repressed. Private consumption in the region is expected to decline 10.1 percent this year and it is unlikely to recover quickly, even in a scenario of no further lockdowns. Livelihoods are further affected by a decline in remittances. This implies that the region will experience a sharp increase in the poverty rate. Uncertainty around the forecast is substantial. The pandemic is still not under control and the spread of COVID-19 could even accelerate in South Asia. External demand for the region’s exports is still volatile, as is the flow of workers’ remittances into the region. These all imply risks to the forecast, but the main risks lie in financing. Simulations presented in this chapter indicate that a sudden stop in external finance due to an increase in risk aversion by foreign investors could sharply cut governments’ ability to spend the money required to stimulate the economy, leading to a 5.2 percent average annual fall in output on top of the decline induced by the pandemic. A domestic debt crisis driven by the already fragile banking sectors with high levels of non-performing loans could also have quite severe effects on GDP, driving an additional 3.4 percent of GDP fall by 2025. The complicated balancing act for governments is to stimulate the economy and to keep debt sustainable. In such an environment effective spending, in terms of creating jobs and preventing financial distress, becomes even more important than
PHOTO BY: TIBY CHERIAN / SHUTTERSTOCK.COM
under normal circumstances and leads to a much faster recovery. In the long term, labor productivity is likely to have deeper scars the longer the crisis lasts. It will be increasingly difficult for large and small companies alike to avoid insolvencies and bankruptcies. A permanent loss of jobs will lead to a loss in skills. But there is a silver lining: the pandemic could spur innovations that improve South Asia’s future participation in global value chains, as its comparative advantage in tech services and niche tourism will likely be in higher demand as the global economy becomes more digital. Governments need to keep a close eye on these long-term trends so as to not lose the opportunity to build back better.
37
38
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Introduction
The pandemic could spur innovations that improve South Asia’s future participation in global
The COVID-19 crisis is not only a health crisis,
value chains, as its comparative advantage in tech
but also an economic one. The pandemic has had
services and niche tourism will likely be in higher
a devastating effect on economies and workers
demand as the world economy becomes more dig-
across South Asia. The 7.7 percent fall in regional
ital. In the long term, labor productivity is likely to
GDP forecast for 2020 would be by far the larg-
have deeper scars the longer the health crisis lasts
est decline on record. Unlike earlier recessions in
and the greater the share of output and employ-
which investment and exports led the downturn,
ment in each country that depends on sectors re-
consumption, traditionally the most stable compo-
quiring high social interaction.
nent of demand, has been artificially repressed and will take much longer to recover, even in a scenario
The importance of sound policies has become even
of no further lockdowns. The sharp decline in con-
more evident during the crisis. Governments in the
sumption is likely to drive an even greater increase
region are challenged with trying to decide how to
in poverty than suggested by the forecast of GDP.
allocate their scarce resources (across sectors, across
The global extreme poverty rate is expected to rise
large and small firms, and whether to spend early or
for the first time in over 23 years, and more people
hold on for when firms can rebuild). At the same time,
will be added to the ranks of the extreme poor in
they must carefully manage their budgets, as exter-
South Asia than in any other region in 2020.
nal financing availability is subject to uncertainties.
A first wave of the epidemic is still affecting South
The first section below describes the baseline fore-
Asia, so there is great uncertainty regarding when
cast. The second section presents simulations to
the pandemic will be controlled, and demand and
assess the risks to this forecast. The third section
productive capacity will face more normal con-
analyzes potential long-term impacts of the cur-
ditions. Neither a decline nor an improvement in
rent crisis. The chapter concludes with a section
external demand for the region’s exports is likely
that discusses policy options.
to have a major impact on growth in the region, because trade is a relatively small share of GDP in the two largest economies. Also, depressed energy
Growth in the region downgraded
prices tend to benefit these economies when global demand falls, while higher energy prices if the
Compared to the early months of the health crisis,
global recovery is stronger than expected tend to
there is now a better understanding of the impact
offset the gains. A decline in remittances would not
of the COVID-19 virus, but continued uncertainty
have a major impact on the regional aggregates,
about the economic outlook. With little understand-
but the decline in private consumption in some
ing of the epidemiological nature of the pandem-
of the remittance-dependent economies could be
ic spread and the effectiveness of specific policies,
severe.
many believed that an early and strict lockdown would be enough to contain the virus within a 1 to
The main sources of risk to the forecast lie in a
3-month period. Since then, it has become clear that
lack of financing shackling a fiscal stimulus. Sim-
the fight against the pandemic will take much lon-
ulations suggest that a sudden stop in external fi-
ger. Countries have tried to strike a delicate balance
nance due to an increase in risk aversion by for-
between opening the economy for business, taking
eign investors could sharply cut governments’
care of livelihoods and getting a handle on the con-
ability to spend the money required to stimulate
tagion rate, an effort hampered by the scarcity of
the economy in some countries, leading to a large
rapid and effective COVID-19 tests. These challeng-
fall in output on top of the decline induced by the
es are more difficult in South Asia, given weak pub-
pandemic. A domestic debt crisis driven by the
lic health care systems and most households living
already fragile banking sectors with high levels
on subsistence wages, than in most other regions.
of non-performing loans also could have quite
Nonetheless, there have been laudable examples of
severe effects on GDP growth and poverty levels.
effective containment in the region, such as Bhutan
However, the adoption of a larger fiscal stimulus
(with zero deaths from COVID-19), Sri Lanka and
owing to the greater availability of financing, if
specific regions in the larger countries. Still, the eco-
devoted to supporting viable economic activities,
nomic impact has been and will be more severe in
would speed recovery.
all countries than initially anticipated.
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
The forecasts for GDP for the region have been
the latter two economies since June (World Bank,
downgraded since June 2020, due in part to the
2020a).
effect of the lockdowns imposed in South Asia and across many trade partners, including high-in-
• In Afghanistan, real GDP is still expected to con-
come countries and the large downgrade of growth
tract by 5.5 percent in 2020, largely due to the
in India (Table 2.1). Restrictions on air transport,
impacts of the COVID-19 crisis on consumption,
international travel and mobility will hamper a
with a protracted recovery amid continued in-
full recovery in 2020 in key export sectors, such
security and uncertainty about the peace talks’
as information technology and business process
outcome.
management (IT-BPM), foreign-led construction
• The most devastating effects of COVID-19 in the
projects, tourism and remittances. The spread of
region will be borne by Maldives, where GDP is
the virus outside urban areas in India will make
projected to shrink by 19.5 percent in 2020 due to
future containment efforts more difficult. Both
the complete paralysis of the tourism sector in Q2
private consumption and investment will contrib-
2020 and the slow resumption of tourism since
ute to GDP decline in 2020 (Figure 2.1). Regional
borders reopened in mid-July. The forecast is for
GDP is expected to fall by 7.7 percent in 2020--af-
a rebound of 9.5 percent in 2021 under a scenario
ter growing over 6 percent a year in the previous
in which borders remain open and tourists gradu-
5 years. As a result, by 2021 per-capita income will
ally return, assuming a virus containment strategy
only be 94 percent of its level in 2019. (Figure 2.2).
that enables the return of some tourists in 2021 is successful (Box 1.3). Although medium- and long-
For countries that report GDP figures in calendar
term tourism prospects remain strong, visitor ar-
year--Afghanistan, Maldives and Sri Lanka—the
rivals are not projected to return to pre-pandemic
forecasts have been revised down significantly for
levels until 2023.
Table 2.1: The economic impact on South Asia will be more severe than initially estimated Real GDP growth at constant market prices, 2019 and forecasts (percent)
Fiscal year
2019
2020(e)
2021(f)
2022(f)
Revision to forecasts from June 2020 (percentage points)
Revision to forecasts from October 2019 (percentage points)
2020(f)
2020(f)
2021(f)
Calendar year basis South Asia region
January to December
4.1
-7.7
4.5
4.6
-5.0
-14.0
-2.2
Afghanistan
December to December
3.9
-5.5
2.5
3.3
0.0
-8.5
-1.0
Maldives
January to December
5.9
-19.5
9.5
12.5
-6.5
-25.0
3.9
Sri Lanka
January to December
2.3
-6.7
3.3
2.0
-3.5
-10.0
-0.4
Fiscal year basis
FY2018/19
FY2019/20(e) FY2020/21(f) FY2021/22(F)
FY2019/20(f)
FY2019/20(f) FY2020/21(f)
Bangladesh July to June
8.1
2.0
1.6
3.4
0.4
-5.2
-5.7
Bhutan
July to June
3.8
1.5
1.8
2.0
0.0
-5.9
-4.1
Nepal
mid-July to mid-July
7.0
0.2
0.6
2.5
-1.6
-6.2
-5.9
Pakistan
July to June
1.9
-1.5
0.5
2.0
1.1
-3.9
-2.5
FY2019/20 India
April to March
4.2
FY2020/21(f) FY2021/22(f) FY2022/23(f) -9.6
5.4
5.2
FY2020/21(f) -6.4
FY2020/21(f) FY2021/22(f) -16.5
-1.8
Note: (e)=estimate, (f)=forecast. For India, FY2020/21(e) runs from April 2020 through March 2021. June 2020 forecasts are from World Bank (2020e) and October 2019 are from October 2019 South Asia Economic Focus, World Bank. Pakistan was reported at factor cost. Source: World Bank
39
40
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Figure 2.1: South Asia’s per capita growth will fall amid a sizeable private consumption and investment-led downturn in 2020. Contribution to GDP growth in South Asia countries Percent, Percentage points 8 6 4 2 0 -2 -4 -6 -8 -10 -12 2007-2009
2015-2019
Private consumption
2020 (f)
Government consumption
2021 (f)
Gross fixed investment
2022 (f)
Net exports
GDP growth
Note: (f)=forecast. South Asia aggregates are converted to calendar year. The value of stacked bars for historical figures does not exactly sum to GDP growth due to inventory changes and statistical discrepancies. The blue shaded area is the baseline forecast. Source: World Bank and staff calculations
Figure 2.2: Income-per-capita in 2021 will remain 6 percent below 2019 estimates, reversing the trend gains made before COVID-19. South Asia real GDP per capita forecast Index, 2019=100 115 111
110 105 100
98 95
94
90 85 2016
2017
2018
Fall 2020 Forecast
2019 June 2020 GEP
2020 (f)
2021 (f)
Fall 2019 forecast
Note: South Asia aggregates are converted to calendar year. Source: World Bank and staff calculations
• Sri Lanka’s GDP will decline by 6.7 percent, with
better forecast estimates of the initial impact of
the crisis affecting all key drivers of demand: ex-
the shock in the recently ending fiscal year, but the
ports, private consumption and investment. The
economic downturn will continue to be reflected
current account deficit is expected to remain low
in the forecasts for the fiscal year ending in 2021.
(at 2.2 percent of GDP in 2020) thanks to low oil prices and strict import restrictions amid large foreign exchange shortages.
• In Bangladesh, which had been one of the fastest-growing economies in the world, GDP growth is projected to decelerate to 1.6 percent
Bangladesh, Bhutan and Pakistan report GDP in
in FY21. Private consumption growth is likely
fiscal years that run from July 1 to June 30, while
to remain subdued amid a projected decline
Nepal reports from mid-July to mid-July of the
in remittances and depressed wage income in
following year. These four countries already have
manufacturing and construction. Investment
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
and exports will suffer amid major uncertain-
FY22, assuming COVID-related restrictions are
ty about the resumption of demand for ready-
completely lifted by 2022, but mostly reflecting
made garments; demand in Europe and the
base effects.
United States is stabilizing, but the recovery is fragile. Moreover, while remittance inflows
The forecast calls for a short-term improvement in
have surged over the past three months, this
current account balances, while capital inflows are
may be the result of repatriated savings by
forecasted to remain positive in the baseline, bar-
returning overseas workers. Remittances are
ring any unexpected events. In 2020, most South
forecast to decline in FY21 with weaker demand
Asian countries will see an improvement in their
from migrant-receiving countries such as the
current account balance as a result of dried up de-
oil-producing Gulf states.
mand for imports, which are projected to fall by
• In Bhutan, economic growth is projected to
18.7 percent in real terms compared to a 13 percent
slow markedly, averaging 2.5 percent a year over
export decline (Table 2.2). An exchange rate depre-
the medium term, well below the pre-COVID
ciation in 2019 in Pakistan (which raised import
five-year average of 5.5 percent amid languish-
costs) and scarce foreign exchange in Sri Lanka
ing services. The growth deceleration in 2020
led to effective import compression even before
would have been steeper had new hydropower
COVID-19. For both countries, this effect is fore-
capacity not come onstream in August 2019. The
casted to continue throughout 2020. More favor-
slowdown in India is expected to depress manu-
able terms of trade due to lower commodity prices,
facturing and exporting industries, and the con-
as well as higher than earlier expected temporary
struction sector (which relies on Indian migrant
remittance inflows in mid-2020 for Nepal, Paki-
workers) is also likely to experience a protract-
stan, Bangladesh and Sri Lanka, lead to a more fa-
ed slowdown due to a limited pipeline of public
vorable outlook for the current account. Once the
sector infrastructure projects.
recovery of demand leads to import growth, there
• In Nepal, GDP is projected to expand by only 0.6
will be an inflection point as the current account
percent in FY21, from an estimated 0.2 percent
balance deteriorates, depending on the speed of
in FY20, as periodic and localized lockdowns
recovery of export demand. The baseline forecast
continue, and disruptions to tourism are expect-
also assumes that remittances will fall slightly in
ed to persist well into FY21. Remittance inflows
2021 as returning migrants have trouble finding
will remain close to FY20 levels due to lower
employment abroad, which will reduce the current
outmigration and weak economic activity in mi-
account further.
grant-receiving countries. A few key hydropower projects are expected to support industrial
Most countries will see slightly higher consumer
growth.
price inflation in 2020 relative to 2019. Some of this
• In Pakistan, economic growth is projected to
relates to sporadic shortages and bottlenecks that
remain below potential, at 0.5 percent for FY21
were created at the start of lockdown, but food pric-
compared to over 4 percent annual average in
es have not seen abnormal spikes since July. Agricul-
the three years to FY2019. This projection, which
tural production is expected to continue holding up
is highly uncertain, is predicated on the absence
well, although there is a risk of crop damage from
of significant infection flare ups or subsequent
the locust infestations in the northwestern part of
waves that would require further widespread
the South Asian region. The downward pressure
lockdowns.
from fuel prices as oil and commodity prices remain low amid weak demand will help create room
Finally, India’s current (FY21) fiscal year runs from
for monetary policy support going forward.
April 1, 2020 to March 31, 2021. That means that the most severe effects of the pandemic will be felt
Poverty is expected to rise, a reflection of the loss
in this fiscal year.
of livelihoods and employment that has devastated the region, which may not be fully reflected in the
India’s GDP is forecast to plunge in FY21 by 9.6
GDP numbers. Urban non-traded services were
percent (revised down since June from a 3.2 percent drop), reflecting the impact of the nation-
disproportionately affected: these sectors do not
al lockdown and the income shock experienced
have as high labor productivity as manufacturing
by households and small urban service firms.
and export sectors do, although they employ the
Growth is forecast to return to 5.4 percent in
vast amount of informal subsistence workers. This
41
42
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Table 2.2: All demand components are projected to decline in 2020 except government consumption, while imports fall faster than exports. South Asia forecast of real growth of GDP and demand components (percent) 2019
2020(f)
2021(f)
2020(f)
4.1
-7.7
4.5
4.7
Private consumption
4.7
-10.1
4.9
4.7
Government consumption
10.4
9.9
4.8
5.3
Gross fixed investment
-2.3
-14.9
6.2
6.2
Exports goods and services
-1.4
-13
5.9
8.2
Imports goods and services
-5.1
-18.7
9
10.4
GDP
Note: (f)=forecast South Asia aggregates are converted to calendar year. Maldives was excluded from the analysis because it does not publish demand-side national accounts. Source: Staff calculations
impact on informal workers also explains why a re-
of the extent to which the virus would spread, it
covery in GDP will not immediately translate into
would remain difficult to predict what that rate of
a recovery in livelihoods, so that the recession will
contagion implies for economic activity. The im-
undo at least three years of gains made in poverty
pact on economic activity will depend on various
reduction in the region (World Bank, 2020b). The
factors: (i) the type of economic activities and the
number of people in extreme poverty in 2020 in
extent to which they require social interaction; (ii)
South Asia (with less than $1.90 a day in PPP in-
how restrictive and effective containment policies
ternational dollars) was estimated to comprise 15.5
are; and (iii) whether the population is able or will-
percent of the total global extreme poor before
ing to follow the lockdown rules1. Even if South
COVID-19. Now South Asia’s new extreme poor
Asian countries are successful in containing the
due to COVID-19 will comprise more than half of
pandemic in 2020, they may be heavily depen-
the new global poor. This may be a conservative
dent—through trade, remittances or external fi-
figure given the recent downward revision to the
nancing—on economies that are not recovering as
growth forecast of India, the largest country in the
quickly.
region (see Chapter 4). Given this level of uncertainty, it is particularly im-
Simulating the risks to the forecast outlook
portant to explore alternative scenarios with differing assumptions for potential drivers of economic activity in the region. Our goal is to construct alternative scenarios that trace the impact of changes in exogenous drivers of growth (for example, changes
The forecasts are extraordinarily uncertain. The
in export demand or a shift in investors’ appetite
forecasts presented in the previous section are
for risk) on the outlook. To do this, we need to
based on extrapolations of high-frequency infor-
impose these exogenous changes into a model (we
mation, modeling exercises and assessments of
use a version of World Bank’s Macroeconomic and
policies in the pipeline to bring together all avail-
Fiscal Model-MFMod2), that takes account of how
able knowledge. However, there are still many un-
such changes would influence our official forecasts.
knowns. With a chronic lack of on-the-spot, freely
We first construct a baseline scenario, which is con-
available testing, and little understanding of the
sistent with the model relationships, with results
rate of asymptomatic transmitters in the popula-
as close as possible to the baseline forecast given
tion and the share of the population that has anti-
above. Then the alternative scenarios can be calcu-
bodies, it is impossible to predict the exact course
lated by altering exogenous elements of the model
of the virus (Stock, 2020; Baldwin and Di Mauro,
or the input data. The main elements are as follows
2020). Even if we could make an accurate forecast
(see Apendix A for details):
1 While there is some evidence from US states that those with stringent measures were eventually more successful in bringing the number of new cases down and keeping them that way, it was also evident that complete containment was going to be impossible with inter-state mobility (Hale et.al., 2020). 2 MFMod is a collection of standardized country models used for the Macro Poverty Outlooks at the Word Bank (see Burns et.al., 2019). Models are classic Keynesian (IS-LM) models with supply side, and well-defined cross-country trade, remittances and commodity interlinkages. The version used in the present analysis has been tailored to the South Asia region specifically.
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
• We first project the number of COVID-19 (henceforth denoted ‘COVID’) cases for each
it does not have national accounts on the demand side.
country based on an econometric model described in Box A2.1. Depending on the num-
The simulation results reflect the depth of the re-
ber of projected COVID cases (Figure A2.1), a
cession, which lasts into the medium-term. The
non-linear rule is applied to each country to
blue line in the left-hand panel of Figure 2.3 shows
denote the country-level impact of COVID on
the growth rate for South Asia under the baseline:
economic activity.
there is a 13.4 percentage points drop in GDP com-
• We then forecast the reduced domestic pro-
pared to the black line, which represents growth in
duction in each sector that directly results from
a no-COVID world (also simulated using the mod-
COVID. We know from high-frequency-da-
el, assuming a COVID shock value of zero). Al-
ta that some sectors, such as those that require
though GDP growth itself returns to similar levels
frequent face-to-face interactions, are affected
in 2022 and slightly surpasses the historic rate, this
more than others. Each of the 9 sectors’ end-July
is just a recovery from a very low base; it reflects
value is calculated based on high-frequency data
the assumption that the direct effects of COVID
using the principal components methodology
disappear (but not the knock-on effects). The right
employed for the monthly activity indicators in
panel shows the same simulation results but mea-
Chapter 1. The rate at which each sector is pro-
sures the GDP gap compared to the counterfactu-
jected to return to normal depends on the coun-
al of no-COVID. It shows that by 2025, the region
try-level COVID effect.
will have lost and still not regained an equivalent
• The domestic production estimates are adjusted
of 6.6 percentage points of pre-COVID GDP.
for the extent to which value added is produced for domestic demand or for export. The decline
Risks around the recovery in the external envi-
in export demand will depend on the course of
ronment cloud the outlook. Given the uncertainty
external demand, which itself depends on how
around the timing of the recovery, we simulate the
well trading partners have been able to control
possibility that the assumptions about the exter-
the epidemic. We also adjust the shock for a few
nal environment are too optimistic or pessimistic
country-specific effects that cannot be captured
relative to the baseline. A worse-than-forecasted
by the model. The COVID shock is assumed to
international downturn assumes that the recovery
lose steam in 2021 but still affect GDP and disap-
in other regions is delayed by one year compared
pear by 2022. The COVID shock is inputted to
to the baseline, perhaps because repeated COVID
MFMod, which then distributes this now lower
outbreaks further dampen consumer and business
GDP into demand components according to in-
confidence and economic activity in major ex-
put-output relationships.
porting countries outside the region. Oil and com-
• The baseline forecast roughly emulates the re-
modity prices are assumed to remain depressed:
sulting modeled GDP with three adjustments
Brent oil prices average USD 35/barrel in 2020
(Figure A2.2). First, the modeled GDP forecast
and $42 in 2021 before recovering in 2022. In ad-
for each country is converted to calendar year.
dition, remittance income is affected. We assume
Second, the effective fiscal impulse in the fore-
that 20% percent of migrants are unable to go back
casts is removed because we would like to model
to the host countries because of slow recoveries in
changes in the fiscal impulse. Finally, the base-
oil-producing states as well as anti-migrant policies
line forecast is extended to 2025 for the purpos-
in high-income countries.
es of the simulation. The impact on the forecast for regional GDP of a This generates the baseline scenario used in the
more depressed external environment is limited
simulations. Then the alternative scenarios can be
(Figures 2.3, lower bounds). For South Asia as a whole,
calculated by altering exogenous elements of the
the impact is small because the largest countries, In-
model or the input data. The baseline from the
dia and Pakistan, are not very open so that most of
model approach follows the forecast almost per-
the risk to the outlook comes from differences in the
fectly for 2020 and 2021 (Figure 2.3, red squares
growth of domestic demand (including investment
against blue line). We then construct a series of sce-
demand). Moreover, all regional countries are im-
narios with different assumptions for the evolution
porters of energy products, so depressed commodity
of external and domestic variables (see Table 2.3).
prices improve the terms of trade, partly offsetting
Maldives is excluded from the simulation because
the decline in demand for the region’s products.
43
44
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Table 2.3: Risks to the forecast and changes in circumstances are analyzed by simulating 6 scenarios Description of scenarios Scenario
Assumption behind scenario
Details of scenario construction
No-COVID
No-COVID counterfactual scenario
Uses World Bank Global Economic Prospects January 2020 forecast to 2021 (no COVID). Output gap is assumed to close after 2025.
Baseline
Forecast assuming health-related effects disappear and stringency measures are removed by 2022.
Constructed scenario using COVID shock, which approximates to forecast once converted to calendar year (Apendix A). COVID-related shocks progressively vanish in 2021 and 2022 with the magnitude being 25 percent of that of 2020.
Risks to the outlook: alternative scenarios assuming changes to the external environment
Worse international downturn
A protracted recovery: trading partner growth stalls amid uncontrolled pandemic abroad.
Extend baseline, but assumes postponed recovery in the rest of the world (0 percent growth in 2021 and similar recovery pattern starting 2022). Additional fall in commodity prices: oil price fall by 20 percent in 2020 and 5 percent in 2021 compared to baseline. Other commodities fall by 10 percent in 2020 and 2.5 percent in 2021 compared to baseline.
Worse international downturn and depressed remittances
In addition to the scenario above, remittances decline in line with lower oil prices and falling GDP in migrant host countries, while policies to reduce migrants reduces the stock of returning migrants by 10 percent in 2021. Inward-looking policies in advanced economies dry up grants for Afghanistan.
Worse international scenario for 2020 and 2021 and additional impact on remittance income. Secondary income and public transfers fall further for Afghanistan. Beyond 2021, the growth rate for remittances returns to baseline levels, but by a 20 percent lower base amid non-returning migrants (to oil-producing Gulf States, United States and Europe).
Faster international recovery
Trading partners are able to control the pandemic by 2021 (fast availability of vaccine, quick testing and improved COVID treatments).
Full return of world demand to no-Covid levels over 3 years starting 2021. Gap with respect to no-COVID situation is assumed to decline much more rapidly for export partners and migrant host countries, by 75 percent in2021 and continue to 2025)
Active scenarios: Simulated fiscal and financial shocks.
Financial crisis
Despite progress in containing the virus, bankruptcies amid pre-existing vulnerabilities in the banking sector and high levels of nonperforming loans lead to a financial crisis. With no credit to the private sector, private investment drops precipitously.
Extend baseline with an additional reduction in private investment and higher interest rates. The private investment shock in each country is calculated by applying the investment-to-GDP elasticity observed in India during the 2008 financial crisis--when private investment fell by 10 percent. The size of the fall is scaled by the share of nonperforming loans to total loans to the one observed in India in 2008, i.e. 2.4 percent. The shock is assumed to occur at end of 2021 and disappear in 2023.
Sudden stop of external financing
In addition to slow recovery in trading partner economies, vulnerabilities in the global financial system manifest and external creditors become highly risk averse. No new external financing available for South Asia except bilateral and multilateral creditors cover enough for debt service payments.
Extend worse international downturn scenario. Add limits on deficit financing (calibrated so that net financing is limited to baseline net domestic financing). Government expenditure cuts as a result allocated 60 percent to capital expenditures and 40 percent to expenditures on goods and services.
With available external financing, significant increased spending on health and revivalrelated programs.
Extend baseline, but with an additional fiscal stimulus of 5 percent of baseline GDP in 2020, 2021 and 2022. 60 percent of new fiscal expenditures are allocated to capital expenditures and 40 percent to related goods and services. The deficits are financed from domestic and external sources in the same proportionn as they have been historically.
Fiscal stimulus
Source: Staff calculations
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Figure 2.3: South Asia’s GDP growth is forecasted to plunge in 2020. Compared to a no-COVID counterfactual, the region is unable to completely recoup the loss over the medium-term (right figure). South Asia real GDP growth Percent
South Asia real GDP deviation from No-COVID scenario Percent 0
15 10
-5
5
-10
0
-15
-5
-20
-10
-25
-15
-30
2019
2020
2021
2022
2023
2024 No-COVID
2019
2025
2020
2021
2022
2023
2024
2025
Baseline, and forecast for 2020, 2021
Note: The blue shade confidence band refers to worse international downturn and depressed remittances and faster international recovery scenarios. Excludes Maldives. Source: Staff calculations using MFMod.
If a depressed international environment is ac-
fast as in the baseline and then gradually thereaf-
companied by lower remittance receipts, the im-
ter up to 2024 (details in Table 2.3). Progressive-
pact on some of the countries would be significant.
ly smaller waves of the pandemic outbreak and
The impact on the region as a whole is small be-
improved treatments lead to a faster lifting of re-
cause remittances comprise only 2.8 percent of
strictions, improving consumer confidence and
Indian GDP. However, private consumption falls
investment sentiment abroad. Oil and commod-
precipitously in Nepal, Pakistan, Sri Lanka and
ity prices bounce back to levels before the crisis
Bangladesh, where remittances are a large share
by the end of 2021 and remain there in the out-
of income (Table 2.4). Relative to baseline, GDP
er years. The results are shown in the upper band
would see an additional 2 percentage point decline
around the baseline (Figure 2.3). Again, the impact
in Bangladesh, 3.3 percentage points in Pakistan—
on the region is limited, particularly as the rise in
with some persistence as GDP does not revert
commodity prices offsets part of the gain from
over the 2023-2025 period---and 3.9 percentage
higher export demand. However, more favorable
points in Nepal. This also adds to the expected fall
international conditions in 2021 provide enough
in baseline private consumption, which is such an
of an impulse such that the difference between
unusual and devastating aspect of this crisis. There
the level of GDP in the simulation and the level if
are negative but insignificant additional impacts
the COVID pandemic had not occurred (the GDP
on the rest of the countries.
gap) improves permanently by about 2 percentage points of GDP to 2025. This happens because the
A front-loaded negative effect in 2021, just as econ-
scenario assumes full return of export demand to
omies are on the way to recovery, can also have dev-
the no-COVID levels. A more favorable external
astating consequences for welfare in the long-term.
environment also would improve external de-
A measure of welfare loss is the cumulative change
mand from India to the smaller countries such as
in private consumption over the 2020-2025 peri-
Bhutan, which could have additional second-order
od (in net present value terms). Compared to the
spillover effects (Box 2.1, World Bank 2020b). Nev-
baseline, a ‘worse external environment scenario
ertheless, in the current harsh economic context,
with reduced remittances’ leads to a relatively small
the scale of the impact of changes to assumptions
but negative loss in private consumption over the
about the international environment appears to be
medium-term in Nepal, Pakistan and Bangladesh
of second order and temporary compared to the
(Figure 2.4). This would also add to poverty since
COVID shock impact.
remittances are well targeted in terms of reaching poor households (Ratha, 2020). The loss is negligible for other countries (slightly positive for Sr Lanka).
Financial Shocks, Fiscal Effects, and External Indebtedness Even before COVID hit there were concerns that
On the other hand, an optimistic scenario could
non-performing loans (NPLs) and contingent lia-
materialize amid faster international recovery.
bilities in the region were stubbornly high, increas-
We assume external demand recovers twice as
ing the risk of a negative domestic financial shock.
45
46
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Table 2.4: Private consumption will fall by more in a scenario with depressed remittance inflows. In contrast, consumption held steady in the last global crisis. Change in private consumption, calendar year (%) Remittances to GDP in 2019 (%)
Annual change in 2008 during global financial crisis
2020 baseline scenario
2020 worse international decline and depressed remittances scenario
Nepal
27.3
3.5
-20.9
-24.3
Pakistan
7.9
1.5
-8.8
-11.1
Sri Lanka
7.8
7.5
-16.7
-17.2
Bangladesh
5.8
3.1
-14.2
-15.5
Afghanistan
4.6
-7.9
-7.1
-7.6
India
2.8
5.2
-19.0
-19.1
Bhutan
1.7
21.3
-2.2
-2.3
Maldives
0.1
South Asian countries
Note: Last two columns are the result of simulation exercises described in Table 2.3 Source: World Bank and staff calculations using MFMod.
Figure 2.4: If remittance inflows dry up, Nepal, Pakistan and Bangladesh see slightly worse welfare effects compared to baseline. Cumulative welfare loss change of a scenario with worse international downturn and depressed remittances compared to baseline scenario, 2020-2025 Percentage points 0.5 0.0 -0.5 -1.0 -1.5 -2.0 -2.5 -3.0 -3.5
Nepal
Pakistan
Bangladesh
Afghanistan
Bhutan
India
Sri Lanka
Note: Difference between cumulative private consumption change between 2020 and 2025 of the depressed remittances scenario, and the baseline scenario. Source: Staff calculations using MFMod
Hidden debts are also prominent in the region
We simulate a financial crisis in which private in-
(World Bank. 2020d). The financial system will
vestment collapses, which affects countries with
be depending on the swift recovery of large firms,
preexisting vulnerabilities. Private investment falls
particularly exporting firms. Rising NPLs could
by an amount proportional to the level of non-per-
create a cascading effect, reducing lending to other
forming loans (NPLs) for each country at the start
businesses and further jeopardizing the recovery.
of the COVID crisis, but calibrated to mimic the
Smaller, informal sector activities would also suf-
effect observed in India during the 2008 financial
fer indirectly through the dry-up of micro-credit .
crisis.4 We also assume that scarce private credit
Liquidity injections from the central banks could
leads to higher interest rates. The hypothetical fi-
counteract the cascading effect but cannot fully
nancial crisis hits in late 2021, so the largest impact
remedy the decline in credit supply.
comes in 2022. Most countries experience a double
3
3 There is already early evidence that micro loans and loans to SMEs, which are outside of the formal system and go in large part to consumption, are having higher default rates than before the pandemic (Rahman, 2020, Malik et.al., 2020). If these types of lending markets are also indirectly affected—say, because the monetary and fiscal authorities turn their attention to propping up the formal financial sector—the effect could be much worse because consumption recovery would be more protracted. 4 In 2008, the share of non-performing loans to total loans in India was 2.4 percent, less than a third of the share at end-2019. Yet private investment fell by almost 10 percent. Another difference of this simulated crisis to the 2008 global financial crisis is that the latter was externally induced and not domestic.
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Figure 2.5: Simulating risks: the ability of policymakers to manage the recession and cooperation of external financiers can make a significant difference in South Asia’s speed of recovery from the COVID-19 crisis. Deviation of GDP from No-COVID scenario Percent
Afghanistan
5
0.9
0
-4.5
-5
-4.6
-10 -15
-15
-20
-20
-25
-25
-30
-30 2020
2021
2022
2023
2024
2025
-5.5 -7.1 -9.9
2019
2020
2021
2022
2023
2024
2025
India
Bhutan
0
-1.8
-5 -10
-7.1
2019
Bangladesh
0
-5
-0.8
0
-4.3
-5
-7.8 -9.0
-10
-3.7 -6.3 -7.3 -9.7
-10
-15
-15
-20
-20
-25
-25
-30
-30 2019
2020
2021
2022
2023
2024
Sri Lanka
5
2025
2019
2020
2021
0
2023
2024
-5 -4.2 -6.4
-5 -10
-5.1 -9.0
-10
-9.3 -13.6
-15
-15 -17.5
-20 -25
2025
Nepal
0
4.9
2022
-20 -25 -30
2019
2020
2021
2022
2023
2024
2025
2019
2020
2021
Pakistan
2022
2023
2024
2025
South Asia region 0
0 -5
-5.7
-5
-10
-10.2 -12.1 -13.6
-10
-15 -20
-3.5 -6.6 -7.8 -10.0
-15 -20
-25
-25
-30
-30 2019
2020
2021
2022 Baseline
2023
2024
Financial crisis
2025
2019
2020
Sudden stop of external financing
2021
2022
2023
2024
2025
Fiscal stimulus
Note: The blue shade confidence band around baseline refers to worse international downturn and depressed remittances and faster international recovery scenarios. Maldives was excluded from simulations because it does not publish demand-side of national accounts. Source: Staff calculations using MFMod.
dip recession (Figure 2.5, red dotted line). Overall,
by 2025.5 There are exceptions though: for Bhutan
this is almost as sharp a fall of GDP as the COVID
and Bangladesh the effect persists into 2025, sug-
shock effect in 2020 for South Asia, although many
gesting a more permanent effect on productivity
countries close the gap with the baseline scenario
due to higher NPLs. Bhutan, India and Bangladesh
5 Laeven and Valencia (IMF WP/18/206) find that the fiscal cost in low- and middle-income countries of a financial crisis (the median) is about 10% of GDP, and on average they happened when NPLs to total loans spiked to over 14.4%. Most of them have continued effects that last up to 5 years from onset, and median output loss of 14%. We used these results to calibrate the shock.
47
48
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
experience important dips in growth: relative to
2020 almost doubles in some countries compared
baseline, GDP in 2022 would be 16.8, 8.0 and 6.8
to the baseline scenario (Figure 2.5, dark green
percentage points lower, respectively.
line). In Afghanistan, the scenario assumes that inward-looking policies in grant-giving countries
In the forecast for 2020 it is assumed that, faced
lead to a sharp drop in donor assistance in addi-
with shrinking fiscal space amid dwindling tax rev-
tion to the already forecasted drop in 2021. Given
enues, governments are re-prioritizing spending. In
the country’s dependence on external grants and
2020 budgets will be reoriented to spend on imme-
concessional loans, the level of GDP would barely
diate health crisis needs, including increased trans-
recover to its baseline level over the 5-year period.
fers to households, while maintaining prior levels of
India’s growth in 2020 would be 7 percent lower
non-discretionary expenditures such as public sector
than the baseline: although the gap would narrow
wages stable. In the baseline scenario, the assumption
slightly compared to baseline, it would still be 1.2
is that governments will be able to reorient spending
percent lower by 2025. Sri Lanka and Pakistan
starting in 2021 and towards rebuilding and recovery
would see growth in 2020 fall by an additional
of capital expenditures. But this may not happen un-
11.4 percent and 9 percent compared to the base-
less there is early and careful planning.
line scenario. Sri Lanka, Nepal and Pakistan’s GDP gap compared to baseline would persist over the
The possibility of insufficient external financing for
forecast period. In the baseline scenario, Sri Lan-
this increased spending is a major risk. Domestic fi-
ka and Pakistan already see a deviation from a no-
nancing should be available, as most central banks
COVID world as an ‘L’-shape, suggesting that their
in the region have already provided and should
pre-existing external financing challenges and high
continue to provide generous lines of credit to the
debt burden continue to play out in a post-COVID
government, with the view that the risk of inflation-
world. While this scenario is clearly a lower-bound
ary public spending amid the pandemic is small.
extreme case unlikely to materialize, it does bring
However, a greater risk for many governments
to light how much international financial cooper-
in the region is the reliance on external financ-
ation and expansionary monetary policies of ad-
ing. Most governments in the region understand
vanced and developing economies will matter for
that expansionary fiscal policy is the right remedy
external liquidity in South Asia. Prudent policy-
during this crisis but are weighing this against the
making in India, whose share of South Asia’s econ-
risk that already scarce financing could completely
omy is more than three-quarters, can also provide
dry up. This would create another risk: that post-
important spillovers and financing to the rest of
ponement of capital expenditures could lead to ex-
the region, particularly its small neighbors such as
actly the kind of procyclical policies that have ham-
Bhutan, Nepal and Maldives.
pered recoveries in the past (World Bank, 2020c). Early, proactive spending by governments comWe simulate a sudden stop scenario in which South
bined with the possibility of full access to interna-
Asian governments are only able to finance exter-
tional markets at reasonable rates could catalyze a
nal debt service—either from multilateral sources
faster recovery. This scenario assumes that govern-
or rollovers of other debt—but no new external fi-
ments undertake a faster expenditure switch from
nancing is available. As a result, spending is lim-
current transfers and supporting consumption (see
ited to what can be financed from depressed tax
Chapter 4) to activities that can help revive the econ-
revenues and some financing from the domestic
omy, such as temporary work programs. Moreover,
banking system. External capital markets have so
it assumes that external financing will be forthcom-
far functioned normally (Figure 1.1D). However,
ing at historical rates and terms (Table 2.3). Under
investors could become highly risk averse amid a
these assumptions, the recession would be much
protracted recession in their own countries or due
more muted than in the baseline scenario for all
to excess market volatility, which could lead to a
countries (Figure 2.5, light green line). Afghanistan
sudden stop.
Moreover, rising budget pressures
and Sri Lanka would be able to more than recover
have been accompanied by a new wave of sover-
GDP losses under the pandemic by 2025, and Bhu-
eign debt downgrades, surpassing peaks during
tan’s activity would be barely affected compared
prior crises (Bulow et.al., 2020).
to a no-COVID scenario. India and Bangladesh would experience a fall but would recover quickly (a
A negative external financing shock could have a
‘U-shaped’ recovery). These simulation results sug-
devastating effect on the region. The fall in GDP in
gest that the multiplier effect of public investment
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Figure 2.6: Prudent fiscal policies amid ample external financing significantly mitigate GDP loss. GDP annual average change from baseline Percentage points 10 5 0 -5 -10 -15 South Asia region
Pakistan
Afghanistan
India
Bangladesh
Sudden stop of external financing
Financial crisis
Nepal
Bhutan
Sri Lanka
Fiscal stimulus
Source: Staff calculations using MFMOD
on economic growth would be particularly strong in
that during the current crisis credit rating agencies
the first two years of the forecast period, as docu-
are procyclical in that they rate all countries more
mented by Beyer and Milivojevic (2020).
strictly during a downturn (Bulow et.al. 2020). In the last few months rating agencies downgraded
The simulations suggest that prudent fiscal and fi-
India, Sri Lanka and Maldives on COVID-related
nancial policies are important, but the availability
concerns. On the other hand, since the COVID cri-
of external financing makes a critical difference in
sis is global, investors may be less likely to discrim-
the rate of economic recovery in the region and
inate across international markets. Central banks
in minimizing the income loss. Figure 2.6 shows
need to be vigilant to the possibility of sudden
the average annual loss of GDP over the period
changes in external investor sentiments leading
in net present value terms. It shows that a sudden
to reversals of inflows that quickly create a sudden
stop would be devastating for most countries, but
stop (Al-Amine and Willens, 2020).
especially Sri Lanka. For the region, a sudden stop would lead to annual average GDP growing 5.2 per-
The baseline forecasts suggest that at least three
centage points slower than in the baseline (com-
countries are at high risk of debt distress, despite
pared to 3.4 percentage points slower than baseline
double-digit growth in external receipts in the five
in the financial crisis scenario). However, a fiscal
years ending 2018 (Figure 2.7). Looking at public
stimulus amid accessible external financing could
external debt in comparison to exports and remit-
have a very big impact in the medium-term, lead-
tance earnings, Pakistan and Sri Lanka had already
ing to growth of 4.1 percentage points faster every
crossed a major threshold before the crisis and
year on average. (Figure 2.5).
were already significantly vulnerable (see Chapter 1).6 The baseline forecast suggests that Sri Lanka’s external debt will become very high compared to
Managing debt well will be paramount in the short-term.
exports and remittance receipts. Pakistan will con-
There is great uncertainty about the flows of cap-
debt distress are believed to remain moderate in
ital over the 2020-21 period, particularly for India
Bhutan because most of the debt is linked to hy-
which is more exposed to international markets.
dropower projects financed by the Government
Generally during a crisis, creditors will become
of India and backed by intergovernmental agree-
more risk-averse, impacting the emerging market
ments. For Maldives this number is low as a share
premium and inciting capital outflows. For exam-
of exports, but the forecast is for a dramatic rise of
ple, the currency crisis in Turkey and Argentina in
public debt to GDP, while repayment issues may
mid-2018 reverberated across emerging markets,
arise given dwindling government revenues from
including India (ADB, 2018). There is evidence
tourism. (Box 1.5).
tinue to stay slightly above the threshold. Risks of
6 The IMF-World Bank Debt Sustainability Analysis framework has established an approximate threshold for the share of external debt to exports and primary income of 150 percent, over which debt of low-income countries is considered vulnerable (https://www.worldbank.org/en/programs/ debt-toolkit/dsa)
49
50
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Figure 2.7: External debt affordability has steadily deteriorated since the global financial crisis, which is expected to continue post-COVID General government external debt 2008-2019 and forecast Percent of exports of goods, services and remittances inflow 350 300 250 200 150 100 50 0
2008 Afghanistan
2009
2010 Bangladesh
2011
2012 Bhutan
2013 India
2014
2015 Maldives
2016 Nepal
2017
2018
Pakistan
2019 2020 (f)
2021 (f)
Sri Lanka
Vulnerability threshold
2022 (f)
Note: (f)=forecast. The blue shaded area is forecasted Source: International Debt Statistics, Macro Poverty Outlook, World Bank
There are two vulnerabilities to watch for:
unpaid loans in state-owned enterprises. Also, it is
(i) Despite important efforts to account for exter-
possible that arrears may be created between dif-
nal debt, domestic debt, particularly ‘hidden’
ferent levels of government, but this information is
or contingent debt, is always harder to track.
generally not reported (World Bank, 2020d). These
Domestic debt is also expected to grow in most
problems may be exacerbated by the fact that many
countries (Figure 2.8 in the baseline). This could
firms and banks were given moratoria to help them
be a vulnerability: Reinhart and Rogoff (2009)
through the crisis—in some cases the banks in-
show that in the five-year run-up to default in
volved already had high non-performing loans as
almost 90 episodes studied between 1827 and
happened in India and Bangladesh in 2018.
2003, domestic debt was growing faster than external debt, and that often the increase in domestic debt was ‘hidden’. On the positive side, according to the ‘fiscal stimulus’ scenario (lines in Figure 2.8) the debt to GDP ratio would
Longer-term effects: a preliminary assessment
not change much in Nepal and Sri Lanka. This is because the growth of GDP would be faster
There is no doubt that the pandemic will leave deep
than the growth in indebtedness. It also as-
scars in South Asia’s economy in the long-term,
sumes that Sri Lanka, would be able to access
but it is too soon to assess how deep. COVID-19
reasonable terms on new financing.
is an epidemiological disaster that will reduce the stock of human and physical capital in South Asia,
Other kinds of domestic financial stress can arise
and therefore potential output, in the same way
in crisis times, which can quickly exacerbate fiscal
that wars and natural disasters do. 7 For every clo-
problems. Even if a buildup in external debt does
sure of a business, reduced education or training
not materialize, many countries with low debt-to-
opportunity, or unused machine, the productive
GDP levels have domestic vulnerabilities that could
capacity of the economy is being eroded by the
transform into a public debt problem in the form
virus. Bankrupt firms and loss of installed capital
of contingent liabilities. Bangladesh, Pakistan and
will have a lasting effect on productive capacity.
Sri Lanka are expected to see a rise in domestic
The depth of those scars will depend on the length
debt in the baseline forecast (Figure 2.8). Moreover,
of the health pandemic, but also the way in which
rising non-performing loans in the domestic sector
governments and societies deal with it. Despite the
have been cited as an issue in Bangladesh, Bhutan,
huge uncertainty, we estimate the average growth
India and Pakistan. Most will see a rise in arrears of
of potential output comparing the baseline against
7 Potential output is defined as the maximum growth an economy can reach in each period in a sustained basis without creating inflation. This is associated with a level of value added per worker (labor productivity) and a more theoretically broad but difficult to measure concept: total factor productivity (TFP).
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Figure 2.8: Except for India and Maldives, domestic indebtedness is driving forecasted debt increases. Assuming no barriers in access to finance, debt ratios would improve. Domestic and external public debt to GDP: historical, forecasts and fiscal stimulus scenario Percent 180 160 140 120 100 80 60 40 20
India
Maldives
Nepal
Public external debt
Bangladesh
Public domestic debt
Pakistan
Public total debt
2020-2022
2010-2019
2020-2022
2010-2019
2020-2022
2010-2019
2020-2022
2010-2019
2020-2022
2010-2019
2020-2022
2010-2019
0
Sri Lanka
fiscal stimulus
Source: Macro Poverty Outlook and staff calculations using MFMod.
a no-COVID counterfactual for all South Asian
does not grow sufficiently in the outer years to gain
countries over the medium-term (except Maldives)
back all its per-capita productive capacity. Afghan-
using the results of MFmod.8
istan is unable to completely recuperate from the long-term damage (as its productive capacity per
We find that on average, most countries in the re-
capita was already very low). Human capital loss is
gion lose over 1 percentage points potential output
not accounted for here, but estimates suggest there
growth per capita as a result of the loss of productive
is already a loss in South Asia (Box 1.2).
capacity between 2020 and 2025 (Figure 2.9). While per-capita GDP begins to bounce back in 2021 (Fig-
In the long-term, labor productivity also suffers
ure 2.2), productive capacity is affected for much
important losses in a pandemic. The direct COVID
longer. Bangladesh could have an estimated output
shock for 2021-2025 is assumed to be dependent
gap 3 percentage points of GDP lower in per-capita
on the length and intensity of the COVID cases. A
terms due to COVID. Pakistan has a sharp dip and
recent World Bank productivity study (Dieppe et.
recuperates its losses due to COVID by 2023 but
al. 2020), using a local projections methodology
Figure 2.9: Potential output growth per person will be 1 percent lower on average over the medium-term. Estimated average annual growth in potential output per capita, 2020-2025 Percent 6 5 4 3 2 1 0 -1 -2 -3
India
Bangladesh
Bhutan
Sri Lanka
No-COVID (counterfactual)
Nepal
Pakistan
Afghanistan
Baseline scenario
Note: Baseline scenario as described in Table 2.3 using change in TFP estimates due to COVID from World Bank (2020a). Source: Staff calculations using MFMod. 8 To estimate TFP and capital deepening, aggregation is done through a standard Cobb-Douglas production function. The stock of capital evolves according to the usual law of capital accumulation and therefore progressively reflects changes in total investment. In the baseline simulations, TFP is assumed to be exogenous while structural employment is assumed to be unchanged, since COVID-related deaths represent an extremely low fraction of the labor force. As such, changes in potential output solely reflect the adverse investment shock induced by the COVID crisis.
51
52
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Figure 2.10: Labor productivity was higher in India and Sri Lanka before COVID struck, but COVID-19 is likely to negatively impact longer-term labor productivity in Sri Lanka the most. Contribution to labor productivity growth in PPP, 2012-2017 average Percentage points
Estimated long-term labor productivity loss resulting from COVID pandemic by sector, relative to 2017 Index, average global labor productivity loss=100 16
6 5
14
4
12
3
10
2
8
1
6
0
4
-1
2
-2
0 Bangladesh
Agriculture
India
Construction
Pakistan
Finance and busines services
Bangladesh
Sri Lanka
Manufacturing
Mining
India
Trade services
Pakistan
Transport services
Sri Lanka
Utilities
Total
Note: The labor productivity loss for each South Asian country is indexed by the total global labor productivity loss. The latter is set equal to 19.2 percent (6.4*3, equivalent to the cumulative effect on labor productivity of pre-COVID epidemics, but the effect lasts three times longer). This is multiplied by each country’s 2017 labor productivity (latest available number). The distribution of the loss across sectors in each country is calculated by multiplying the country’s labor productivity loss by each sector’s employment share of total. Other services, comprising mostly public services, are excluded from the totals. Source: Staff estimates using Dieppe et. al. (2020) dataset.
over horizons based on Jordà (2005), calculates that
are important contributors, in large part due to their
pre-COVID epidemics since 2000 have lowered la-
large share in total production. The COVID-19 pan-
bor productivity by a cumulative 6.4 percent after
demic would have a negative but different effect on
5 years. Although the longevity of epidemics since
labor productivity in each of the sectors, as we know
2000 has been 1 year on average, labor productiv-
that sectors where social interaction is needed will
ity itself can continue to fall over a longer horizon
be the last ones to come back to normal. As skills are
mainly through the adverse effects on investment
unused, many cannot come back to the labor market
and the labor force as well as delayed human cap-
and collaboration is impacted; it is as if every work-
ital accumulation and loss of skills from extended
er became 6.4 percent less productive over time. We
unemployment. As the authors admit, they have a
find that Sri Lanka, with the highest labor productiv-
very small sample of four epidemics, which makes
ity according to the most recent data before COVID,
the comparison difficult given that COVID-19 has
would consequently suffer the largest losses in the
been described as “the crisis of the century” (Re-
long term (Figure 2.10b). Weighted by the COVID
inhart and Reinhart, 2020). Our baseline scenario
shock in each sector and employment, agriculture
assumes that the brunt of the pandemic will be over
(in all countries) and manufacturing in Sri Lanka
in 2021 with lingering effects in 2022. In that case,
would contribute the most to the total loss.
the epidemic lasts 3 years (between 2020 and 2022), which is three times longer than the average dura-
The COVID-19 pandemic may well have a signifi-
tion of the sample of epidemics in the above-men-
cantly worse impact on labor productivity than
tioned study. That means that with lingering effects,
most previous natural disasters. There are three
the impact could be felt almost 15 years from now:
main reasons. First, the increased integration of
it makes sense to assume that the impact on labor
the global economy will amplify the adverse im-
productivity will be higher the longer the epidemic.
pact of COVID-19. Second, contagion prevention and physical distancing may render some activi-
We apply this factor to each sector but weighted by
ties, for example the hospitality sector, unviable
the impact of the COVID shock. Not surprisingly, the
unless they are radically transformed, which will
impact of the epidemic on labor productivity at the
take time. Even in less directly affected sectors like
sector level will be proportional to the labor produc-
manufacturing, banking and business, severe ca-
tivity before COVID. Figure 2.10 (left panel) shows
pacity underutilization lowers TFP while restric-
the contribution of each main sector to labor pro-
tions to stem the spread of the pandemic remain
ductivity growth in the large four South Asian coun-
in place. There-may be intra-sectorial shifts into
tries in the years leading up the pandemic. Services
low-productivity agriculture. Finally, disruptions
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Box 2.1. The Silver Lining: Can global value chains thrive in South Asia post-COVID? South Asia is not as integrated into global value chains (GVCs) as its East Asian neighbors and many other developing economies. GVC participation, a measure of trade integration in value-added terms, has consistently been lower than the global average and mainly driven by India’s participation (Figure 2.11, top left panel). Intraregional trade is very low, with the share of imported intermediates embodied in exports originating from South Asian partners as low as 3 percent (compared to 55 percent in ASEAN). Second, the overall trade restrictiveness index for South Asia countries, which captures the trade policy distortions that each country imposes on its import bundle, shows South Asia with the greatest effective protection compared to any other region (World Bank, 2019). The question then is whether the post-COVID world will provide an opportunity for South Asia to reorient its trade paradigm. GVCs are transforming as a consequence of the COVID-19 pandemic. While the continuing protectionist tendencies in some advanced economies and the trade redirection that began in 2019 because of the US-China trade and investment disputes are unrelated to COVID, they do make it difficult to predict the direction of the transformation of GVCs in the post-pandemic era. Nevertheless, over the medium-term experts have noted three accelerating trends post-COVID (UNCTAD 2020). First, participation in GVCs in advanced economies may weaken, and reshoring accelerate, because professionals cannot travel as freely as before COVID. In theory this could lead to less fragmented supply chains and more concentrated production--especially in higher-technology GVC-intensive industries catering to specialized consumers-- and set in motion increasing foreign divestment from developing countries1. On the other hand, replication capabilities that lead to a reallocation of manufacturing located closer to final consumers--which could accelerate as a result of COVID because of a desire to re-shore--could benefit highly populous economies such as India, Pakistan, and Bangladesh as well, precisely because their consumer market is growing much faster than in advanced economies. Second, diversification of suppliers to minimize supply interruptions given different and uncertain breakouts of the pandemic globally may affect some GVC-intensive industries. The garment sector already diversifies by making multiple orders of the same design from firms in different countries, and this trend is likely to increase post-COVID because buyers can ensure that interruptions due to a disaster in one location are hedged by having identical products sourced from other locations. Third, reduced air transport capacity may strengthen regional and local GVCs, affecting for example, perishables such as food processing industries and domestic tourism in South Asia. Finally, digitalization is likely to accelerate postCOVID, which will improve firms’ ability to coordinate supply chains and logistics from anywhere. History suggests that international production-sharing is unlikely to disappear, although global supply chains witnessed short-term interruptions at the height of the COVID-19 breakout. Some argue that producers will become more risk-averse and consider breaking with long-term international relationships ( Javorcik, 2020), but this presumes that specialized suppliers can be found locally at comparable cost, an unlikely situation for complex production processes2. Not only might self-sufficiency not diversify risks for future global and local shocks, but could have the opposite effect (Miroudot, 2020). Moreover, sectors with high GVC participation were no less likely than others to suffer from production bottlenecks, reinforcing the global nature of the shock (Maliszewska et al., 2020). This is also true for South Asian exports (Figure 2.11, top right panel). What might be revisited by multinational firms is the “just-in-time” production system which enables firms to minimize holdings of inventory due to sophisticated logistics networks. Now firms will need to build buffers against future shocks. There are no previous examples of such a global shock. Still, the available evidence of localized disasters suggests that firms have found ways to adjust and become more resilient. The Japan’s 2011 earthquake and the Thailand floods in 2011 led to short-term interruptions in supplies (Cavalho et.al. 2020), but firms adjusted by temporarily diversifying their suppliers (Matous and Todo, 2017). In the case of Thailand, multinational firms have since increased their production capacity in the country (Nakata et al, 2020; Cassar et al, 2017; Haraguchi et al, 2014). This GVC transformation that is likely to occur post-COVID, more regionalization, and the growing importance of services relative to manufactures, will provide a major opportunity for South Asia. Digitization and changes in working patterns in a post-COVID world will increase opportunities for countries with a comparative advantage in services provision. This bodes well for South Asian economies, as they have some of the highest rates of domestic value added of services embodied in total exports—in some countries, services like tourism and 1 3D printing is an example of such technology, as the same production process can be replicated in many locations through digital orders while allowing for some customization. However, large-scale production units of a uniform good are still cheaper to locate offshore. 2 The case for protecting the export of Personal Protective Equipment (PPE) like face masks continues to be at the heart of this debate. However, shortages were not due to supply chain malfunctioning but to a surge in demand that outweighed the global production capacity (Miroudot, 2020). Certainly, there is a case for governments maintaining stockpiles for future pandemics.
53
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
IT are directly and indirectly exported (Figure 2.11, bottom left panel). The key role of services in GVCs is likely to accelerate post-COVID for two reasons. First, the increasing use of remote work will help drive increasing demand for traded services (Baldwin and Forslid, 2020). Second, services production is increasingly embodied in manufactured goods, so the nature of service jobs that are offshored will also change. This was already a trend seen before COVID but is now likely to accelerate. Countries like India specialize in rapidly expanding skill-intensive sectors such as software development and professional services which are in large demand from multinationals (Figure 2.11, bottom right panel). Bangladesh, which principally exports readymade garments, also has a thriving local IT sector that is increasingly tapped by international firms, particularly in gig services (see Chapter 3). Finally, technological advances and the more intensive use of remote work following the COVID pandemic might allow countries to directly export the source of their labor advantage. Remote work will redefine the nature of GVC trade, as it allows people performing tasks for a firm from one country to be physically working in another country (Baldwin 2019). Wage differences and human capital shortages will create incentives for companies to hire more foreign-based service workers to perform different and new tasks online.
Figure 2.11: Though GVC participation has been low in South Asia, there are opportunities going forward: other than short-term disruptions, the size of a sector COVID shock and its GVC participation are uncorrelated. Moreover, services’ domestic value added in exports is high and India dominates in terms of jobs in specialized software FDI. GVC participation Ratio
COVID shock and backward GVC participation in South Asian sectors Index, 1=full-COVID effect
GVC 0.50participation Ratio 0.45 0.50 0.40 0.45 0.35 0.40 0.30 0.35 0.25 0.30 0.20 0.25 1990
COVID1.0 shock and backward GVC participation in South Asian sectors Index,0.81=full-COVID effect COVID shock COVID shock
54
1995
2000
2005
South Asia region World 2000 2005 India South Asia region without
2014
0.0
2014
East Asia average Services value added contribution to exports by country, 2019 OECD average East Asia average OECD average Bhutan Pakistan Bhutan Bangladesh Pakistan Sri Lanka Bangladesh India Sri Lanka Nepal India Maldives Nepal 0 10 20 30 40 50 60 70 Maldives share of total exports of goods and services (%)
0.1
0.2
0.3 0.4 0.5 GVC backward participation 0.2 0.3 0.4 0.5 India Bangladesh Sri Lanka GVC backward participation Pakistan Bhutan Nepal Maldives
0.6
0.7
0.6
0.7
Note: The red dashed line represents the linear trendline with the coefficient India Bangladesh Sri Lanka equal to 0.1, which shows the insignificant correlation between the GVC backward participation and the COVID shock. Bhutan Pakistan Nepal Maldives
Jobs created by FDI (non-video software)
Other Destination Countries Jobs created by FDI (non-video software) 6% Malaysia Philippines 3% Other Destination Countries Other South Asia 5% Malaysia 6% 1% 3% Philippines Other South Asia Singapore 5% 8% 1% Singapore South Korea 2%8%
80
10 20 30 40 50 60 70 80 Domesticshare services' value addedof goodsForeign services' of total exports and services (%)value added Domestic services' value added
0.1
-0.2
Note: GVC participation is the summation of backward and forward participation. World South Asia region Backward GVC participation refers to the ratio of the foreign value-added content South exports. Asia region withoutGVC Indiaparticipation corof exports to the economy’s total gross Forward Servicestovalue addedofcontribution to exports by country, 2019 responds the ratio the domestic value added sent to third economies to the economy’s total gross exports.
0
0.6 0.2 0.4 0.0 0.2 -0.2 0.0 0.0
1995
0.20 1990
1.0 0.6 0.8 0.4
South Korea China 2% 9%
India 66% India 66%
China 9%
Foreign services' value added
Note: Domestic services value added refers to value added contributed domestically to produce exports. Foreign value added is foreign services value added embodied in own exports. Source: World Bank (2019), ADB (2020), FDI Markets, Financial Times (2020)
Note: Share of new jobs created from greenfield investment in software publishing and custom computer programming service (excluding video games), 2015-2019
In sum, COVID has provided an opportunity to change course and increase GVC participation, but South Asian countries need to prepare to take advantage of these new opportunities. This can be done with improvements to logistics performance, lowering effective tariffs, improving regulatory environments, supporting innovation ecosystems, creating digital free-trade zones and expanding digital banking for small entrepreneurs who can now offer service to a larger global market.
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
to training, schooling and other education in the
hopefully a “better normal” that rewards resilience.
event of severe income losses, even once restric-
Policymakers need to take stock of their finances
tions are lifted, will also lower human capital and
and set the stage for a more efficient economy.
labor productivity over the long term. The main recommendations center around two But there are mitigating factors and open opportu-
themes. First, South Asian governments should
nities, not taken into account in the above analysis.
do everything they can to ensure that external fi-
In some dimensions, pandemics and epidemics
nancing is forthcoming while following sound
can accelerate productivity-enhancing changes
debt management practices. This is also a shared
such as investment in innovative types of train-
responsibility with the international community.
ing of more highly skilled workers (Bloom 2014).
Beyond that, the medium-term planning should
Moreover, the mitigation measures of COVID-19
focus on encouraging new endeavors and enabling
like social distancing may encourage investment
the process of creative destruction that is inevita-
in more efficient business practices. For example,
ble while not losing sight of long-term opportuni-
the greater reliance on digital technologies during
ties to build back better.
the pandemic is likely to lessen geographical barriers despite less air travel, which could increase the
• The extra time afforded by initiatives such as
region’s low participation in global value chains
the Debt Service Suspension Initiative (DSSI)
(Box 2.1). Surprisingly, foreign direct investment
and generous credits from other central banks
in India9 and Pakistan has surged so far in 2020,
will provide an opportunity for governments to
with early evidence in India of services develop-
take stock of their finances. While DSSI is un-
ment that could increase the productivity of firms.
likely to provide meaningful debt forbearance
FDI in services can also enhance the productivity
given the breadth of the pandemic, it may pro-
of other downstream sectors such as manufactur-
vide some breathing space for those that partic-
ing and exports, as has been documented in India
ipate to prepare for what may lie ahead. This is
over the last 30 years (Arnold et.al., 2014). There is
the time to increase transparency and reporting
already a high demand for gig-economy services
standards and reach out to creditors to ensure
from India and Bangladesh (Chapter 3).
there is good faith if a rollover or renegotiation is needed. Standstill is not a substitute for debt
The role of government and policy recommendations
restructuring. A good example is the “Punto Final” Program developed by Mexico during its debt restructuring, in which creditors were encouraged to participate in negotiations: those that came forth early were given incentives that
The effectiveness of public health policy and of
offset some of the write-down losses with Mexi-
government economic policies can also turn the
co. While none of the South Asian countries are
course of the downturn. The simulations above
in default, they can learn from the Mexican gov-
abstract from these issues. There have been dif-
ernment’s pro-active stance to reach a financial
ferences in responses by policymakers across the
resolution (Calomiris, 2020).
globe, and even within Indian districts or Pakistani
• A sound strategy would be if policymakers avoid
provinces. Figure 2.12 is a visual representation of
the temptation of temporarily propping up
how the combination of good health and economic
large enterprises and instead indirectly support
policies, as well as good leadership, can impact GDP
a transition towards activities that may be more
growth and economic recovery over the long-term.
viable in a post-COVID world. South Asian governments may not have broad unemployment
At this stage of the pandemic, it is still important to
benefits systems that help guide the recovery like
differentiate between short-term relief responses
high-income countries, but transfer and support
and setting the stage for the economic revival. Re-
policies that can help subsistence workers feed
structuring can be planned now as countries start to
themselves and their families during the transi-
reopen for businesses, and policies should support
tion can have a similar effect. At some point, the
firms’ and workers’ transition to a “new normal”,
private investment will take the driving seat of the
9 India’s investment promotion agency has seen a surge of foreign investment applications by firms post-COVID, many trying to de-risk the entire supply chain locally to get closer to the consumers (India Times, 2020).
55
56
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Figure 2.12: Policies matter for the long-term effects of the pandemic Possible course of post-pandemic long-term GDP growth as a function of health and economic policy effectiveness Better
Virus spread and public-health response
Virus contained but sector damage; lower long-term trend growth
Virus contained; growth returns
Virus recurrence; slow long-term growth insufficient to deliver full recovery
Virus recurrence; slow long-term growth; muted world recovery
Virus contained; strong growth rebound
Virus recurrence; return to trend growth; strong world rebound
Effectiveness of the public-health response Pandemic escalation; prolonged downturn without economic recovery
Worse
Pandemic escalation; slow progression toward economic recovery
Pandemic escalation; delayed but full economic recovery
Knock-on effects and economicpolicy response Effectiveness of government economic policy
Better
Source: McKinsey and Company (2020)
recovery. Governments should avoid subsidizing
provides immediate relief to the urban unem-
large firms directly or keeping large enterpris-
ployed and helps the recovery process. Over the
es afloat that will not be viable in a post-COVID
long term, it will help reduce initial fixed costs
world, but instead devote scarce funds to work
incurred by both national and international mi-
programs or business facilitation which also sup-
grants as well as create greater efficiency and
ports the private sector. For example, many gov-
transparency in the job-seeking process.
ernments in developing countries and the United
• A key challenge will be transforming the tourism
States have been under pressure to support large
sector, which will take many years to recover (Box
airlines, when support of the workers that are laid
1.3). A relatively innovative strategy has been to
off from those airlines instead of the companies
create ‘tourism bubbles’ or “free travel zones” in
themselves can give way to a faster transition. If
Asia--agreements with neighboring regions that al-
an activity comes back online amid renewed de-
low for travel across borders for non-essential trips
mand, firms in that sector will be able to rehire
without quarantining upon arrival. There are pro-
workers. Governments can enable the inevitable
posals underway in India, Bhutan and Maldives. If
reallocation of resources that will eventually take
implemented in tandem with other health precau-
shape through simplified regulations and as well
tions like more frequent testing for COVID-19, this
as continued support of lives and livelihoods.
could bring back a steady stream of visitors10. Tour-
• To help the recovery of jobs and consumption,
ism will face a difficult transition but is one of the
digital services for job seekers can benefit the
more promising and viable sectors in South Asia
poor and unemployed, including migrants. Sev-
in the long-term (World Bank, 2020e). Moreover,
eral apps and portals have been set up in India
South Asia offers multiple ecotourism and religious
(for example, Majajobs, Razgaar Bazaar, Pravas
tourism opportunities, which could be promoted
Rojgar) that seek to match potential employers
further through use of digital applications. Inter-
and employees. These could be replicated across
ventions may include guidelines on designing and
the region, both for local employment and mi-
implementing safeguards and safety mechanisms,
grants. Now potential employers and employees
marketing, development of information and book-
incur high costs as currently most placements
ing portals, and working with national and sub-na-
take place through references, which can limit
tional governments to harmonize policy changes.
options and result in sub-optimal matching of
• South Asia can be part of the solution to the
employees’ skills to jobs. This type of intervention
epidemic itself. Once a vaccine is developed,
10 For example, Emirates Airways required every passenger to be tested prior to flying, but also created incentives by providing free coverage of health care costs for any passenger that contracts COVID-19 on its flights.
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
depending on the technology, vaccine producers
developing world (Vaidyanathan, 2020). Serum
in the region, especially in India, are anticipated
Institute has already entered into two agreements
to play a pivotal role in providing doses for global
with funding from the Bill & Melinda Gates Foun-
immunization efforts against COVID-19. India is
dation to produce 200 million doses of vaccine
home to some of the largest vaccine producers in
candidates being developed by Oxford Universi-
the world, including the Serum Institute, which
ty/Astrazeneca and Novavax, respectively.
produces 1.5 billion doses of vaccine a year to pri-
• There is an opportunity to build back better.
marily support childhood immunizations in low-
Raising the quality and effectiveness of gover-
and middle-income countries. With its cost ad-
nance and improving the business environment
vantage and enormous scale, India is estimated to
while tackling climate change can encourage a
produce over 60 percent of vaccines used in the
faster rebound from disasters (Box 2.2).
Box 2.2 Green and Resilient Recovery in South Asia COVID-19 will profoundly transform South Asia for years to come, and South Asia may also change because the COVID-19 crisis has brought into sharp focus the possibility of unprecedented disruptions, including from climate change. Ignoring those threats is no longer a viable option. A first step to a more sustainable future is a smartly designed recovery program. Economic stimulus packages can include renewable-energy investments, climate-smart buildings, and resilient public transportation systems. During the last few months, carbon emissions have fallen, and pollution levels in the region are down. This happened in the general economic slump caused by the COVID-19 pandemic, offering the chance to shape economic recovery in ways that decouple economic growth from climate change. Most fiscal recovery programs following previous crises have tended to be carbon intensive. In 2009 the financial crises caused CO2 emissions to fall by 1.4 percent, but the following year they increased by 5.1 percent, much more than the rate of increase prior to the crises. Recovery policies can be designed to deliver both economic and climate goals. These could include: clean physical infrastructure investment in the form of renewable energy assets; improved battery storage technology and grid modernization; increases in efficiency through spending for renovations and retrofits, including improved insulation; improved heating and domestic energy storage systems; investment in building natural capital to address immediate unemployment from COVID-19 and structural shifts from decarbonization; climate-friendly agriculture; and clean R&D spending. Investing in resilience is crucial, as disruptions are extremely costly for governments, households, and the private sector. In response to calls from governments around the world, the International Energy Agency, in collaboration with the IMF, has recently produced a Sustainable Recovery Plan for actions that can be taken over the next three years (Figure 2.13). The plan identifies cost-effective measures that could be implemented during the specific timeframe of 2021 to 2023 and shows how governments have a unique opportunity today to boost economic growth, create millions of new jobs and put global greenhouse gas emissions into structural decline.
Figure 2.13: Long-term job opportunities in environmentally friendly activities abound. Jobs content of various clean energy investments Jobs per million dollars New nuclear New hydro Wind power CCUS Reduce methane emissions Unabated gas-fired power Unabated coal-fired power New grids Hydrogen production High-speed rail Exising grids Industry efficiency Urban transport infrastructure Solar PV Buildings efficiency retrofit Efficient new buildings
0
2
4
6 Construction
Source: International Energy Agency (2020).
8 Manufacturing
10 Total
12
14
16
57
58
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
References Al-Amine, R. and T. Willems, (2020). Investor Sentiment, Sovereign Debt Mispricing, and Economic Outcomes. IMF Working Paper No. 20/166, International Monetary Fund. Arnold, J., B. Javorcik, M. Lipscomb, and A. Mattoo. (2014). Services Reform and Manufacturing Performance: Evidence from India. The Economic Journal, 126: 1–39, February. Asian Development Bank. (2020). Key Indicators Database. https://kidb.adb.org/kidb Asian Development Bank. (2018). Asia Bond Monitor, November 2018. ADB. https://www.adb.org/publications/ asia-bond-monitor-november-2018 Baldwin, R. (2020, March 26). Testing for testing times: Fostering economic recovery and preparing for the second wave. VoxEU https://voxeu.org/article/testing-testing-times Baldwin, R. and Di Mauro, B. (2020, March). Economics in the Time of COVID-19: Introduction. A VoxEU and Center for Economic Policy Research (CEPR) book. Baldwin, R. (2019). The Globotics Upheaval. Oxford: Oxford University Press. Baldwin, R and R Forslid. (2020), Globotics and development: When manufacturing is jobless and services are tradable. CEPR discussion paper No. 14293. Beyer, R. and Milivojevic, L. (2020, September 17). Fiscal policy and economic activity in South Asia. Review of Development Economics. https://onlinelibrary.wiley.com/doi/10.1111/rode.12710 Bloom, N. (2014). Fluctuations in Uncertainty. Journal of Economic Perspectives 28 (2): 153–76. Bulow, J., C. Reinhart, K. Rogoff and C. Trebesh. (2020, Fall). The Debt Pandemic. Finance and Development, International Monetary Fund and World Bank. https://www.imf.org/external/pubs/ft/fandd/2020/09/debt-pandemic-reinhart-rogoff-bulow-trebesch.htm Burns, A., B. Campagne, C. Jooste, S. David and T. Bui. (2019). The World Bank Macro-Fiscal Model Technical Description. Policy Research Working Paper No. 8965. World Bank, Washington, DC. Cassar, A., A. Healy, and C. von Kessler. (2017). Trust, Risk, and Time Preferences after a Natural Disaster: Experimental Evidence from Thailand. World Development 94. Calomiris, C. (2020, May). The Unpayable Debts Problem: What to Do, and What Happens If We Don’t. World Bank presentation, June, http://pubdocs.worldbank.org/en/113381591034138447/Calomiris-Global-Debt-May-2020.pdf Carvalho, V., S. Hansen, A. Ortiz, J. R. García, T. Rodrigo, J. V. Rodríguez Mora and P. Ruiz (2020). Tracking the covid-19 crisis with high-resolution transaction data. CEPR Discussion Paper 14642. Chetty, R., Friedman, J. N., Hendren, N., and Stepner, M. (2020). How did covid-19 and stabilization policies affect spending and employment? a new real-time economic tracker based on private sector data. (No. w27431). National Bureau of Economic Research. Dieppe, A., S. Kilic Celik, and G. Kindberg-Hanlon. (2020). Global Productivity Trends. In Global Productivity: Trends, Drivers, and Policies. edited by A. Dieppe. Washington, DC: World Bank. Economic Times. (2020, September 5). India’s commitment to reform being taken seriously by foreign investors: FM Sitharaman. Economic Times India. https://economictimes.indiatimes.com/news/economy/policy/indias-commitment-to-reform-being-taken-seriously-by-foreign-investors-fm-sitharaman/articleshow/77949424.cms Hale, T., T. Atav, B. Kira, T. Phillips and A. Petherick. (2020). Variation in US states’ responses to COVID-19. Blavatnik School of Government Working Paper Series WP-2020/034, Oxford University Haraguchi, M. and U. Lall. (2014). Flood Risks and Impacts: A Case Study of Thailand’s Floods in 2011 and Research Questions for Supply Chain Decision Making. International Journal of Disaster Risk Reduction 14(3). International Energy Agency. (2020, June). Sustainable Recovery. World Energy Outlook Special Flagship Report IEA https://www.iea.org/reports/sustainable-recovery/evaluation-of-possible-recovery-measures#abstract Javorcik, B. (2020, April 2). Coronavirus will change the way the world does business for good. Financial Times Op Ed.. Jordá, Ó. (2005). ‘Estimation and Inference of Impulse Responses by Local Projections’, American Economic Review Vol. 95, No. 1, March (pp. 161-182) Malik, K., Meki, M., Morduch, J., Ogden, T., Quinn, S., & Said, F. (2020). COVID-19 and the Future of Microfinance: Evidence and Insights from Pakistan. Oxford Review of Economic Policy, Forthcoming. Maliszewska, M., A. Mattoo and D. Van der Mensbrugghe. (2020). The potential impact of COVID-19 on GDP and trade: A preliminary assessment. Policy Research Working Paper No. WPS 9211, World Bank.
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Matous, P. and Y. Todo. (2017). Analyzing the coevolution of interorganizational networks and organizational performance: Automakers production networks in Japan. Applied Network Science 2(5): 1-24. McKinsey and Company. (2020, September). COVID-19: Implications for Business. September Business Survey. https://www.mckinsey.com/business-functions/risk/our-insights/covid-19-implications-for-business Miroudot, S. (2020, June 18). Resilience versus robustness in global value chains, VoxEU.org. Nakata H., Y. Sawada and N. Wakamori. (2020). Robustness of Production Networks Against Economic Disasters: Thailand Case. Supply Chain Resilience. Anbumozhi V., Kimura F., Thangavelu S. (eds). New Jersey Department of Public Health. (2020). Travel advisory restrictions by state. https://covid19.nj.gov/ faqs/nj-information/travel-and-transportation/which-states-are-on-the-travel-advisory-list-are-theretravel-restrictions-to-or-from-new-jersey?utm_campaign=20200925_nwsltr_b&utm_medium=email&utm_source=govdelivery#direct-link Rahman, A. (2020, September). Small and Medium Firms during the COVID19 Pandemic: Evidence from Bangladesh. BRAC Institute of Governance and Development presentation, South Asia Economic Network Conference, World Bank. Stock, J. (2020, April 4). Data needs for shutdown policy. VoxEU, https://voxeu.org/article/data-needs-shutdown-policy Ratha, Dilip, (2020, February). Remittances: Funds for the Folks Back Home - Back to Basics. Finance and Development online, International Monetary Fund and World Bank. https://www.imf.org/external/pubs/ft/fandd/basics/remitt.htm Reinhart C. and V. Reinhart. (2020). The Pandemic Depression: The Global Economy Will Never Be the Same. Foreign Affairs, September/October 2020 https://www.foreignaffairs.com/articles/united-states/2020-08-06/ coronavirus-depression-global-economy Reinhart, C. and K. Rogoff. (2009). This Time it’s Different: Eight Centuries of Financial Folly. Princeton University Press. UNCTAD. (2020). World Investment Report 2020: International Production Beyond the Pandemic. United Nations, New York, and Geneva. Vaidyanathan, G. (2020, September 3). India will supply coronavirus vaccines to the world — will its people benefit? Nature. https://www.nature.com/articles/d41586-020-02507-x World Bank, (2020a, June). Pandemic, Recession: The Global Economy in Crisis. Global Economic Prospects. World Bank Group. World Bank. (2020b, April). The Cursed Blessing of Public Banks. South Asia Economic Focus. World Bank. World Bank. (2020c, October). Poverty and Shared Prosperity Report 2020: Reversals of Fortune. World Bank Group. World Bank. (2020d, forthcoming). Hidden Debt: Leveraging Public Capital Responsibly in South. World Bank, Washington DC. World Bank. (2020e, July). COVID-19 and Tourism in South Asia: Opportunities for sustainable regional outcomes’, World Bank Group Brief. World Bank (2019), “World Development Report 2020: Trading for Development in the Age of Global Value Chains” World Trade Organization. (2019). World Trade Report: The future of services trade. WTO.
59
60
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Appendix A. Procedure to estimate the impact of COVID on economic activity by sectors. A COVID shock by sector is defined as a supply-side exogenous shock, where reduced domestic production by each sector is assumed to be driven solely by the direct effects of COVID (the indirect economic effects are subsequently simulated by the model). The underlying country-sector shares for the COVID shock use the 2017-2019 value added data at the 35-sector level for all countries using the Multi-Regional Input-Output Table (MRIOT) database (ADB, 2020). We construct the COVID shock at a monthly frequency for 2020 and extrapolate to 2021 assuming it becomes smaller throughout 2021 (where the average level is 25 percent of the December 2020 effect), and goes to zero in December 2021 and beyond for all countries. It can take values between 0 (if there is no effect of COVID) and 1 (if the sector completely stops production). We summarize the procedure in three steps. In the first step, a regression of the number of daily COVID cases per capita for each country is estimated and the results extrapolated. Box A2.1 shows the estimation procedure results. In the second step, we develop a rule which maps the extent of the COVID-19 outbreak to the impact on economic activity, applied to each South Asian country as well as export partners. This will define the country-specific severity of the COVID-19 contagion on value added produced and consumed domestically. We ranked the sample of countries according to the projected number of new daily COVID-19 cases per 100,000 population for the last quarter of 2020—see forecast methodology in Box A2.1--and create an index between 0 and 1 according to the following criteria: • Countries with projected new average daily cases of 5 per 100,000 people or less on or before the last quarter of 2020 are assumed to see economic activities go back to normal levels by end-October due to lifting of restrictions, necessity or both. Within the region, Afghanistan, Bhutan, Pakistan and Sri Lanka had already reached that milestone by September 2020, while Bangladesh is forecasted to reach it as well. • For countries with projected new daily cases between 5 and 10 per 100,000 people, country domestic activity is expected to go back to normal more gradually, but normal activity is expected by end-December 2020. This is the case for Nepal. • For countries with projected new daily cases between 10 and 100 per 100,000 people, the direct shock persists at a rate proportional to the caseload into 2021, although it gets smaller. This is the case for India and Maldives. This abstracts from heterogeneity of the COVID impact within large countries like India • Beyond that the COVID shock on country domestic economic activity is assumed to continue to impact economic activity by the same amount as in the second and third quarters of 2020. None of the major South Asian trading partners are in this category. This will define the factor that measures the severity of the COVID contagion, S i for any of the i=1 to 93
countries in the sample.
In the third step, we estimate the gap in production for each South Asian country at the 9-sector production level (grouped from the 35-sector level). In turn, the share of production is disaggregated by destination--whether domestic or foreign—with the decomposition calculated using input-output relationships. • The production loss in each sector j that is consumed domestically, P rod _ Ldom i ,j, is proxied by high-fre-
quency indicators and measured by the average deviations of activities relative to the 2019 level using
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
principal components analysis discussed in Chapter 1. Agriculture is assumed to not be affected by COVID-19 (zero loss). Manufacturing loss is mostly obtained from indicators such as industrial production, and PMI. Many of the production loss of trade services’ activity indicators due to COVID-19 have been nowcasted using the Google mobility index. The associations are established in part using the regression analysis shown in table A2.2 but fitted for each South Asian country. At the initial stage the losses for public services, health and education provision, denoted as ‘other services,’ are assumed to be zero11. • The production loss that would have been destined abroad is assumed to be equal to the loss of exports relative to no COVID. This is weighted by the share of exports to all countries i multiplied by the COVID breakout severity Sk of all of country i’s export partners k (i ≠k). In other words, the impact on exports depends on the extent of the recovery of economic activity of external partners.
This provides a measure of the 2020 activity loss between March (when COVID was declared an international pandemic) up to July (the underlying data for nowcasting were not available beyond July). For January and February, the activity loss is assumed to be zero In the final step, we forecast the production loss out to December 2020 at the 9-sector production activity level for production consumed domestically and for exports. The supply-side COVID shock for country i and sector j, C OVID _ shocki j is thus: OVID _ shocki j = D C j * Prod _ L i (Si )+ ( 1 − Dj )* Prod _ L i + add _ factorj i dom,j x,j where Prod _ Ldom i ,j denotes the production loss due to COVID in country i sector j consumed or invested
domestically. In turn, P rod _ Lx i, j = ∑ i≠k Sk * expgrj k , denotes production loss in country i sector j’s exports to trading partner k.; and expgr is the decline in export demand growth as a result of COVID. Dj is
the average share of production of sector j consumed domestically in 2017-2019 (we only have data for 2018 for Afghanistan). We perform an out of sample projection based on the last observation where the implied growth forecast depends on the COVID contagion severity factor Si. For the production loss for
export (P rod _ Lx i, j ) the out of sample projection will be a weighted by the severity of contagion abroad
Sk (i ≠k).
This model-based GDP forecast is checked against our baseline GDP forecast by sector. If some exogenous event results in a large difference between the two, we add a third term a dd _ factorj i . This was the
case in three instances: (i) the forecasted 22 percent output growth of the new Mangdechhu hydropower plant in Bhutan, which offset negative effects; (ii) the expected one-time increase in trade services activity due to temporary remittance inflows in Bangladesh, Nepal, Sri Lanka and Pakistan; and (iii) the growth in foreign-financed infrastructure construction activity in Pakistan. If country-sector assumed effects were not captured, we include a sector-level add factor to the COVID shock for those sectors. These adjustments are very small in magnitude relative to the production loss effects, particularly the domestic effect. We obtain the 9 sector-level COVID shocks for each country i (8 estimated plus ‘other services’) and the model then aggregates using as weights the share of each sector’s value added in total GDP in 2017-2019 (ADB, 2020). Figure A2.2 shows the COVID shock for the South Asian countries in 2020 and the share of each sector’s value added in total. Countries with higher shares in GDP of community, trade and tourism services and construction, all of which require more social contact, also tend to have a higher COVID shock.
11 This is done because the model will endogenously change their value depending on whether the government is assumed to respond with either more or less spending.
61
62
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Box A2.1. Forecasting COVID caseloads and estimating services activity using the Google mobility index. In order to assess the evolution of COVID cases in South Asia, we regress the number of cases using some of the rule-of-thumb relationships espoused by public health experts as reflected in the following regression model:
log CPCit
= α log CPCit 2 + β log CPCi ,t−14 + µ D _ thresholdit + ɸ TestingRateit + γ StringencyIndexi ,t−30+ Constantit + ε it
We postulate that the log of number of reported COVID cases per capita in a given day t for country i will
depend on the number in the past few days to indicate persistence (log CPCi ,t−14). We add a squared term, log
CPCit 2 , to emulate epidemiological models that suggest that the relationship is concave, so the growth rate diminishes and tends to flatten (Baldwin, 2020). We control for other country-specific elements such as: (i) whether the virus is under control; if cases are above a threshold of 10 per 100,000 daily, the variable D_ threshold = 1, based on a rule of thumb used by public health experts1; (ii) the daily COVID tests per thousand population (TestingRateit ), since the reported caseload is dependent on the number of tests administered; and
(iii) S tringencyIndexi ,t−30, a proxy for the severity of pandemic-related measures imposed by the government in the previous 30 days. Table A2.1 show the main results. All else equal, a 1 percent increase in cases in the previous 14 days is associated with a 0.05 percent increase in new cases, and the higher the daily rate the slower the growth. All other coefficients are significant and of the expected sign. They show that, if cases are below the rate of 10 per 100,000, there is less testing, and stringency is high, the daily COVID caseload per capita will be lower. Figure A2.1 shows the actual and out-of-sample predicted caseloads for South Asian countries. One caveat is that if there is a breakout or unexpected surge (a non-linearity), the forecast error is higher. For example, the United States and India had a second surge, so the model under-predicts cases in the recent days in those countries. Likewise, Sri Lanka had an early localized surge but has since controlled the spread, so caseload is slightly overestimated. Second, the model forecasts reported cases, since we have scant information on actual caseload.
Table A2.1: Estimation results explaining the number of reported COVID cases Coefficient on variable. Dependent variable: Daily new cases per capita in log
log CPCit 2 (α)
log CPCi,t−14
it Dthreshold
TestingRateit
StringencyIndexi,t−30
(β)
(µ)
(Φ)
(ϒ)
Estimated value of coefficient
-0.039***
0.0487***
0.344*
0.0381***
-0.00121***
Note: Least Squares Panel with country fixed-effects and clustered standard error. ***, **, * denote significance at the 1, 5, and 10 percent levels, respectively. The estimation period is from January 1st to September 21st at daily frequency, with a sample of 93 countries from Our World in Data COVID-19 database. Testing rate for Bhutan is from official country data. The stringency index reflects strictness of government policies to contain the spread of spread of COVID-19 (Hale et. al., 2020), scaled to a value from 0 to 100 (100=strictest).
Even if the caseload per capita were the same across countries, the effect of the pandemic on activities maywould differ depending on each country’s wealth. To test this, we regress each of the six components of the Google mobility index for all countries at the daily frequency since March against per-capita income and COVID cases in the previous 15 days. A low mobility index number for in restaurants, for example, means that less people frequent restaurants compared to before COVID times. We also control for cellphone adoption since the mobility index is captured through cellphone location use. We also control for the stringency index, as measures such as lockdown reduce mobility. The main results are shown in table 2.3. The signs and significance of the variables is as expected: a higher prevalence of COVID is associated with greater mobility (because people are going about their daily business as if there was no virus).
1 For example, based on advice from their public health departments, US state governments restrict travel from residents of some other states (see for example New Jersey DPH, 2020), or decide on phased reopening at the local level based on the number of cases per 100,000 population. The most common thresholds used internationally are 10 per 100,000 or 5 per 100,000 population. The results do not change substantially if a different threshold is used.
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Figure A2.1: Most South Asian countries predicted to significantly reduce reported daily COVID caseload by end-2020 Actual and predicted COVID daily caseloads per 100,000 habitants in South Asia and selected trading partners Total daily cases per 100,000 habitants United States
3000 Nepal
Maldives
2500 India
2000 South Asia region
1500 1000 500
20
10-
Ma r -M ar 30 -M ar 9-A pr 19Ap r 29 -Ap r 9-M ay 19Ma y 29 -M ay 8-J un 18Jun 28 -Ju n 8-J ul 18Jul 28 -Ju l 7-A ug 17Au g 27Au g 6-S ep 16Se p 26 -Se p 6-O ct 16Oc t 26 -Oc t 5-N ov 15No v 25 -No v 5-D ec 15De c 25 -De c
0
Afghanistan Pakistan
Bangladesh Sri Lanka
Bhutan United States
India Germany
Maldives South Asia region
Nepal
Note: The shaded area is projected based on linear regression model with balanced sample of 107 countries. The last actual observation is September 30th, 2020. The United States and Germany are major trading partners. Source: Ourworldindata.org, World Bank Bhutan country office
Perhaps the most interesting result relevant for South Asia is that mobility of people to retail, groceries, pharmacy, transit and residential areas is not significantly different in rich or poor countries, all else equal. However, people in richer countries go to parks more frequently, possibly reflecting the fact that urban areas in rich countries have more parks that allow them to enjoy fresh air while socially distancing. Moreover, people in rich countries can avoid workplaces more, since workers are more likely to have the option to work remotely compared to workers in poor countries. Remote work is rare for many informal sector activities (chapter 3)2. It reinforces the fact that the virus does not affect all equally: people in lower-income countries, like in South Asia, are disproportionately affected because they are more exposed to the virus through their work in densely-packed urban areas and have less opportunities to enjoy fresh air.
Mobilityit
= α log CPCi ,t−15+ β StringencyIndexit + γ CellphoneAdoptioni + δ GDPcapitai + θ ( CellphoneAdoptioni * GDPcapitai )+ Constantit + ε it
Table A2.2: Government’s restrictive measures lead to a low mobility globally on average Mobility Index shows the number of visits and length of stay compared to no-COVID period. A lower number implies less mobility DEPENDENT VARIABLES
retail and recreation
grocery and pharmacy
parks
transit
workplace
residential
α log CPC i,t−15
2.580***
2.283***
4.943***
1.875***
1.331***
-0.794***
StringencyIndex i
-1.074***
-0.680***
-1.316***
-0.934***
-0.621***
0.336***
-0.000493
-4.58e-06
0.00253***
-0.000175
-0.000890***
8.01e-05
CellphoneAdoption i
-0.157**
-0.0956
-0.0652
-0.0902
-0.154***
0.0378
CellphoneAdop GDPcapita i tion i *
2.07e-06
-1.79e-07
-1.43e-05**
3.42e-07
4.56e-06***
-1.41e-07
13,066
13,066
13,066
13,066
13,066
13,035
88
88
88
88
88
88
GDPcapita i
Observations Number of countries
Note: Least Squares Panel with country random effect. ***, **, * denote significance at the 1,5,10 percent levels respectively. The estimation periods from January 1st to August 21st with daily frequency. Unbalanced sample with 88 countries. Google mobility data is taken from Google COVID-10 Community Mobility Reports. Note that low value of mobility index means there is less mobility relative to baseline (no COVID). Cellphone adoption is mostly insignificant but is negatively associated with mobility in retail, recreation and workplaces where there has been less activity.
2 This is also consistent with result from analyses of detailed US data that activity in poorer neighborhoods has been greater than in rich neighborhoods (Chetty et. al., 2020).
63
64
Beaten or Broken? Informality and COVID-19 With a new world, a changing outlook
Figure A2.2: The constructed COVID shock shows that countries with large services sectors are hit especially hard. Using the model to shock GDP, the resulting growth rates closely mimic the forecast, with variations dependent on the forecasted level of fiscal impulse of each country. COVID COVID shock shock and and share share of of economic economic activities activities in in GDP GDP on on the the supply supply side side Percent Percent 100 100 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 00 Bhutan Sri Lanka Lanka Afghanistan Bangladesh Bhutan Sri Afghanistan Bangladesh Community, trade trade and and tourism tourism Community, Government (incl. (incl. health health and and education) education) Government Transport (air, (air, land land and and water) water) Transport
Index, Index, 0=no 0=no COVID COVID 0.0 0.0 0.2 0.2 0.4 0.4 0.6 0.6 0.8 0.8 Pakistan Pakistan
Nepal Nepal
Agriculture and and natural natural resources resources Agriculture Manufacturing Manufacturing Utilities Utilities
India India
Maldives Maldives
1.0 1.0
Business and and finace finace services services Business Construction Construction Country-level Country-level COVID COVID shock, shock, 2020 2020 (RHS) (RHS)
Note: COVID shock is an indicator between 0 (no COVID effect) and 1 (economic activity is fully shut down). The public sector, including health and education, were not included at this stage.
Comparison Comparison of of constructed constructed baseline baseline with with forecast forecast outcomes outcomes assuming assuming different different degrees degrees of of fiscal fiscal impulse impulse Percent Percent 10 10 55 00 -5 -5 -10 -10 -15 -15
2020 2021 2020 2021 Afghanistan Afghanistan
2020 2021 2020 2021 Bangladesh Bangladesh
2020 2020
Constructed Constructed baseline baseline with with COVID COVID shock shock effect effect
2021 2021 Bhutan Bhutan
2020 2020
India India
2021 2021
2020 2021 2020 2021 Sri Lanka Sri Lanka
Baseline forecast forecast (including (including expected expected fiscal fiscal impulse) impulse) Baseline
2020 2020
Nepal Nepal
2021 2021
2020 2021 2020 2021 Pakistan Pakistan
Constructed Constructed baseline baseline assuming assuming no no fiscal fiscal impulse impulse
Note: Forecasts converted to calendar year. 2022 not included as outside outside of the full forecast period for some countries. Difference will be larger in 2021 if forecasted fiscal impulse varies. The dashes represent what growth would have been without a fiscal impulse. Source: World Bank and staff calculations
The constructed GDP from the model will differ from our baseline forecast in some respects. First, the model converts the baseline forecasts from fiscal to calendar year. Second, the model extends the forecast to 2025. More importantly, the shock was constructed assuming that the fiscal impulse is the same as under a no COVID counterfactual. The bars in the right-hand panel of Figure A2.2 show the modelled GDP growth forecast (neutral on fiscal impulse) for 2020 and 2021 against the forecast and a scenario in which there was no fiscal impulse. It shows that the constructed scenario follows the baseline very well for all countries. The difference between the constructed baseline and the forecast is explained by the implied fiscal impulse of the forecast. The ensuing lower supply-side GDP is distributed across demand categories—including imports—by MFMod.
Beaten or Broken? Informality and COVID-19
PHOTO BY: INDIAVIAMIRROR / SHUTTERSTOCK.COM
Recent economic developments
65
66
Beaten or Broken? Informality and COVID-19
3
Chapter
The impact of COVID-19 on the informal sector
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Summary The informal economy in South Asia has been hit hard by the COVID-19 pandemic. Many unorganized workers, self-employed people and microenterprises have experienced a large drop in earnings. This is suggested by model simulations presented in this chapter and early confirmations are found in rapid-response surveys. Informal self-employment has provided opportunities for some workers who have lost jobs to maintain livelihoods, but likely at much lower earnings levels than before the pandemic. A key reason for the dire situation in the informal sector is that the service sectors that were affected most by the lockdown are dominated by informality. Many informal workers also tend to be more exposed to the disease due to their involvement in the provision of services that require face to face interactions. Isolated, home-based work is not an option for them. With narrow internet penetration in the region, digital alternatives seem viable for only a small share of the informal sector. This is of grave concern, because more than three quarters of all workers in South Asia depend on income from activities in the informal sector. Informal workers and firms tend to have inadequate mechanisms for coping with short-term demand and supply interruptions, due to limited savings and constrained access to finance. While the poor have suffered severely during the crisis, many informal workers in the middle of the income distribution have experienced the greatest drop in earnings. Most of them are not covered by social insurance. A major threat to longterm income generation is that many viable small enterprises, including microenterprises, will fail because they lack the resources to survive over the next six months. The crisis lays bare complicated structural problems in the informal sector. The challenge for policymakers is not only to provide relief to the informal sector in the short run, but also to design more universal social protection systems in the longer run. More generPHOTO BY: SK HASAN ALI / SHUTTERSTOCK.COM
ally, policies should support productivity and human capital formation for the informal sector, rather than hoping that informal activities will gradually be absorbed by the formal sector. Informality has remained stubbornly constant in South Asia during the last decades, despite high overall economic growth. Digital technologies will play an important role in these reforms and could help the informal sector increase productivity and integrate better into markets, provided that access to internet and digital technologies is significantly broadened.
67
68
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Introduction
(vi) An expansion of the digital economy could help the informal sector over the long run by
As described in Chapter 1, the COVID-19 pandem-
reducing the amount of capital required to
ic and the containment policies to combat it have
start a business, facilitating matching in the
sharply reduced economic activity and household
job market, improving management practices
welfare in the South Asian region, and its effects
in firms, and helping firms connect to markets.
are far from over. The decline in demand and con-
However, the digital economy can also, in the
straints on production due to lockdown policies
short run, exacerbate inequality between the
have hit the informal sector, which accounts for
informal and formal sectors: higher income,
almost three quarters of employment in the re-
formal sector workers have greater potential
gion, particularly hard in comparison to the formal
for teleworking, few informal sector firms can
sector. This chapter reviews the impact of the pan-
take advantage of online platforms, and the
demic on the informal sector, and on the distribu-
“gig economy” is mostly for the better educat-
tion of income, using both preliminary data and
ed workers.
a simulation model to estimate the impact on income distribution and employment. It focuses on
The chapter begins with an analysis of recent
informal sector workers, a category that includes
household survey data demonstrating that in-
not only those working in informal firms, but also
formal workers are particularly affected by the
the self-employed and those employed in formal
COVID-19 pandemic. We then use a simulation
firms on an informal basis.1
model to assess ex-ante the impact of COVID-19 on informal sector workers and the distribution
The main messages which arise from this analysis
of income. The next section considers how as-
are:
pects of the digital economy affect vulnerable informal workers, followed by a discussion of key
(i) The COVID-19 impact is biased against infor-
policy-relevant aspects of informality in South
mality. Informal sector workers have suffered
Asia. The penultimate section considers the poli-
the largest declines in employment, and most
cy implications of this analysis, and a final section
of the households who have fallen into poverty
concludes.
during the pandemic are dependent on informal workers, largely daily casual wage workers from the middle of the income distribution. (ii) Many wage workers in both the informal and formal sectors who lost their jobs have turned to self-employment, but likely at much re-
COVID-19 is severely reducing employment and incomes, particularly in the informal sector
duced levels of earnings. (iii) The pandemic has severely affected both
New data reveal that the pandemic is sharply reduc-
the bottom and the middle of the income
ing wage employment in India, and that informal
distribution.
sector workers are most affected. Individual-lev-
(iv) Informality is heterogeneous and different
el panel data from the nationally-representative
policies are needed to assist different groups
Consumer Pyramids Household Survey (CPHS)
of informal firms and workers.
enables us to track the trajectory of individual labor
(v) The region has quickly expanded relief poli-
market participants before and after the COVID
cies, but some long-term challenges (e.g. the
lockdown.2 With some exceptions (see Box 3.2 fur-
limited coverage of social insurance and the
ther below) similar extensive data are not available
low productivity of informal firms) remain or
for the other South Asian countries (the section
have even been exacerbated.
following this one provides model estimates of the
1 Specifically, this chapter defines informal employment to include all individuals working in informal enterprises as owners, employees or contributing family members, as well as those employed in formal firms as casual or temporary workers without a formal contract (ILO, 2013). It defines informal enterprises as unincorporated enterprises owned by households, including those consisting of a single owner-worker (the self-employed). A key characteristic of such enterprises is that there is not a clear separation between the unit of production and its owner. 2 The analysis in this section is based on Bussolo, Kotia and Sharma (forthcoming). The CPHS is a panel survey conducted three times a year by the Center for Monitoring the Indian Economy (CMIE). Sample households are revisited every four months. For example, households visited in August 2019 (as part of Wave 17 of the CPHS) were revisited in December 2019 (as part of Wave 18) and April 2020 (as part of Wave 19). Unless otherwise mentioned, the figures and estimates discussed in this section are based on the sample of households that were surveyed in the month of August 2019, and then resurveyed in the months of December 2019, April 2020 and finally, August 2020. Details on the CPHS data analysis are presented in Appendix 3.A. Also, please see Vyas (2020) for a description of the CPHS and its execution during the COVID period.
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Figure 3.1: The early COVID-19 period in India saw a sharp increase in transitions out of employment Two-category Labor Market Transition Rates A. Post COVID (between December 2019 and April 2020) B. Pre COVID (between August 2019 and December 2019)
Aug−Dec 2019
Dec 2019−Apr 2020 100 Employment status, ending month (%)
Employment status, ending month (%)
100 80 60 40 20
0
80 60 40 20
0 Emp
Employment status, initial month
Emp
Unem/OLF Emp
Employment status, initial month
Unem/OLF
Unem/OLF
Note: These charts depict the rates at which working age individuals transition, or flow, from one labor force status to another between two periods (months). Each bar shows the breakdown (in percentage terms) of the ending period’s status (y-axis) for a group of individuals who had a specific status in the initial period (x-axis). For example, the first bar in Panel A shows that of all the individuals employed (Emp) in the month of December 2019 (the initial period), 43 percent were unemployed or out of the labor force (Unem/OLF) and 37 percent employed in the month of April 2020 (ending period). The second bar shows a similar breakdown for all the individuals who were Unem/OLF in the initial period. The same information is presented in the form of transition tables in Appendix 3.A. Source: Based on staff calculations using CPHS panel data on India.
probable impact of the pandemic on employment
The likelihood of unemployment post-COVID
in three countries).
was higher among those initially working in the informal sector (INF) than those initially in formal jobs (FOR); many of the latter kept working
The early phase of COVID-19 experienced a sharp drop in employment, and a shift from wage jobs to self-employment
by transitioning into the informal sector. Among
The early impact of the pandemic on employ-
is, unemployed or OLF) by April 2020 (Figure
ment has been unprecedented: 43 percent of
3.2, Panel A). The corresponding transition rate
those employed in December 2019 were either
for those initially in the formal sector is lower (31
unemployed or out of the labor force (OLF) in
percent). But another 31 percent of those initially
April 2020 (Figure 3.1, panel A). This is a stark
formally employed were in informal employment
departure from the pre-COVID survey wave be-
by April 2020. Transitions from formal to infor-
tween August 2019 and December 2019: about
mal employment do not occur at such high rates
95 percent of those employed at the beginning
in more normal times (Figure 3.2, Panel B). In gen-
of that period were still employed when it ended
eral, as suggested by panel data from a range of
(Figure 3.1, panel B).3
contexts, such high rates of transition out of for-
those employed in the informal sector in December 2019, over 44 percent were not employed (that
mal employment are rarely observed in low and Notably, this sharp drop in employment rates oc-
middle-income countries (see, for example, Gatti
curred during a mandated ‘lockdown’. Concerned
et al. 2014).
about a potential surge in COVID infections, the Government of India enacted a comprehensive na-
Within the informal sector, wage workers were
tional lockdown on March 25, 2020.4 This nation-
more susceptible to job loss in the early phase
wide lockdown lasted through the month of April
of COVID-19 than the self-employed (SE) were.
but was gradually eased in parts of the country
Among those in informal (that is, casual, daily or
starting May 2020.
temporary) wage jobs in December 2019, over 57
3 4
The detailed transition tables are presented in Appendix 3.A. Order. D.O.No.40-3/2020-DM-I(A), Ministry of Home Affairs, Govt. of India.
69
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Figure 3.2: The lockdown period also experienced high rates of transitions into informality Three-category Labor Market Transition Rates A. Post COVID (between December 2019 and April 2020) B. Pre COVID (between August 2019 and December 2019)
Dec 2019−Apr 2020
Aug−Dec 2019 100 Employment status, ending month (%)
Employment status, ending month (%)
100 80 60 40 20
80 60 40 20
0
0 FOR
INF Employment status, initial month
Unem/OLF
FOR
FOR
INF
INF Employment status, initial month
Unem/OLF
Unem/OLF
Source: Staff calculations based on CPHS Panel for India; see note in Figure 3.1 for explanations
Figure 3.3: Self-employment was more resilient than informal wage employment in the lockdown period Four-category Labor Market Transition Rates A. Post COVID (between December 2019 and April 2020) B. Pre COVID (between August 2019 and December 2019)
Dec 2019−Apr 2020
Aug−Dec 2019
100
100 Employment status, ending month (%)
Employment status, ending month (%)
70
80 60 40 20
80 60 40 20 0
0 FOR
SE Casual Employment status, initial month
FOR
Unem/OLF FOR
SE
Casual
SE Casual Employment status, initial month
Unem/OLF
Unem/OLF
Source: Staff calculations based on CPHS Panel for India see note in Figure 3.1 for explanations
percent were unemployed/OLF by April 2020
percent were in self-employment by April 2020
(Figure 3.3, Panel A). Those initially self-employed
(Figure 3.3, Panel A). This is an unusually high rate
in December 2019 had a lower (34 percent) chance
of transition from wage jobs to self-employment,
of being unemployed/OLF by April 2020. The
particularly for those initially in the formal sector
rates of transition into unemployment do not dif-
(compare with Figure 3.3, Panel B).
fer much across these two sets of workers in more normal times (Figure 3.3, Panel B).
for both formal and informal wage workers who
Aggregate employment levels are beginning to recover, but labor market churn and flows into self-employment remain unusually high
lost jobs. Both among those with formal and in-
CPHS
formal wage jobs in December 2019, about 20
that employment levels reached a low point in
Self-employment may have served as a “cushion”
data
from
June-August
2020
suggest
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Figure 3.4: Six months into the COVID crisis, employment levels are staging a recovery Two-category Labor Market Transition Rates A. Late COVID (between April 2020 and August 2020) B. Early COVID (between December 2019 and April 2020)
Dec 2019−Apr 2020
Apr−Aug 2020
80
80
Employment status, ending month (%)
100
Employment status, ending month (%)
100
60 40 20 0
60 40 20 0
Emp
Emp
Unem/OLF
Unem/OLF Employment status, initial month
Employment status, initial month Emp
Unem/OLF
Source: Staff calculations based on CPHS Panel for India
Figure 3.5: Six months into the COVID crisis, high rates of transition into informality continue Three-category Labor Market Transition Rates A. Late COVID (between April 2020 and August 2020). B. Early COVID (between December 2019 and April 2020).
Dec 2019−Apr 2020
100
80
80
Employment status, ending month (%)
Employment status, ending month (%)
Apr−Aug 2020
100
60 40 20
60 40 20 0
0 FOR
INF Employment status, initial month
Unem/OLF FOR
FOR INF
INF Employment status, initial month
Unem/OLF
Unem/OLF
Source: Staff calculations based on CPHS Panel for India
April-May 2020 – in the thick of the national lock-
– that between the months of April 2020 and Au-
down– and then started to recover. Labor market
gust 2020 – reveals how much the growth in unem-
transition rates between January 2020 and May
ployment has slowed down since early COVID-19
2020 are similar to those between December 2019
days. Among those who were employed in April
and April 2020 period, indicating that the crisis
2020, over 85 percent were still employed in Au-
of unemployment was still peaking in May 2020.
gust 2020 (Figure 3.4, Panel A), a much higher em-
But transition tables for more recent months look
ployment retention rate than that seen between
different.
December 2019 and April 2020 (Figure 3.4, Panel B). Moreover, a larger percentage of unemployed/
The most recent transition table possible given
OLF moved back into employment. More than
CPHS data availability as of writing this report
20 percent of those unemployed/OLF as of April
71
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Figure 3.6: Six months into the COVID crisis, self-employment was the most stable labor market category Four-category Labor Market Transition Rates A. Late COVID (between April 2020 and August 2020). B. Early COVID (between December 2019 and April 2020).
Apr−Aug 2020
Dec 2019−Apr 2020 100
100
60 80 40 60 20 40 200 FOR
SE Casual Employment status, initial month
SE Casual Employment status, initial month
Unem/OLF
SE COVID (August Casual Unem/OLF FOR SE Casual Late 2020) Pre-COVID (Dec 2019) Cross-sectional breakdown of the labor market Employment status, initial month Employment status, initial month 60 C. Late COVID (August 2020) D. Pre-COVID (Dec 2019) 60.5
Unem/OLF
0
Unem/OLF
EmploymentEmployment status, ending status, monthending (%) month (%) 200 40 20 60 40 80 60 10080
Dec 2019−Apr 2020
80 100
Source: Staff calculations based on CPHS Panel for India
FOR
60
FOR
0
EmploymentEmployment status, ending status, monthending (%) month (%)
Apr−Aug 2020
Late COVID (August 2020)
60
60.5
40
40 22.3 20 22.3
14.2
20 3.0
60.2
Pre-COVID (Dec 2019)
60
Percent of working Percentage of working population age population
Percent of working Percentage of working population age population
72
14.2
0
60.2
40
40
20.0
20
15.5 20.0
20
4.3
15.5
0 3.0
FOR
SE
0
Casual
4.3 Unem/OLF
0 FOR
SE
Casual
Unem/OLF
Source: Staff calculations based on CPHS Panel for India
2020 were employed by August 2020, an unusually
unemployment/OLF and into employment is
high rate of entry into employment.
much higher than it was in the early phase of COVID-19, most of this outflow is headed into
Nonetheless, the exodus from the formal sector
the informal sector.
has not abated. Among those still holding formal sector jobs as of April 2020, 51 percent were in
Transitions into self-employment remain at un-
informal employment by August 2020 (Figure
usually high levels. Nearly 30 percent of those still
3.5, Panel A). As a result, the April 2020 to Au-
in formal jobs in April 2020 were self-employed in
gust 2020 transition rate out of the formal sec-
August 2020, while another 22 percent were in in-
tor is nearly as high as it was in the early COVID
formal wage jobs (Figure 3.6, Panel A). Within the
phase (Figure 3.5, Panel B). The main difference
informal sector, the flow from informal wage jobs
is that the composition of the outflow from
into self-employment remains at a much higher rate
the formal sector has tilted away from unem-
than that in the reverse direction. Overall, self-em-
ployment/OLF and towards self-employment.
ployment is the only “stable” category in the labor
In addition, while the rate of transition out of
market.
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
As a result of this unprecedented pattern of employment transitions, six months into the COVID crisis, the overall composition of employment
Figure 3.7: Informal workers were significantly more vulnerable to loss of employment than formal workers were in the early phase of COVID-19 Probability of being employed Post-COVID: informal minus formal
in India has shifted noticeably towards self-em-
0.03
ployment (Figure 3.6). Consider how the size of each sector, measured in terms of its pre-COVID
0.01
share in total employment (as of December 2019), by nearly 30 percent and the informal wage sector by 8 percent, while the self-employed sector has grown by 12 percent. It is also notable that the share of the employed in the total working age population has not changed to the same extent as their composition. The post-COVID situation has evolved from a crisis of unemployment to one of low-quality employment.
Informal wage workers were the most vulnerable to a loss of employment in the early phase of COVID-19 in India The CPHS panel data suggest that informal workers – in particular, informal wage workers — were significantly more vulnerable to loss of employment than formal workers were during the early phase
0.0059 -0.0057
0.0024
0.00 Post x Informal
has changed: the formal wage sector has shrunk
0.013
0.02
-0.01
-0.0019
-0.02
-0.0082
-0.03 -0.04 -0.05
Daily Wage/Casual Self-Employed
-0.042
-0.06 -0.07 -0.08
-0.064 Apr
May
June
July
Note: Estimates based on differences-in-differences panel regressions, with industry-wave and district wave fixed effects included among the control variables. The green line shows the point estimates for the differential changes in employment probabilities post-COVID for those who were self-employed pre-COVID, compared to those in formal wage jobs pre-COVID. Each point estimate corresponds to a specific post-COVID month. The dotted lines depict the 95 percent confidence interval for the estimate. The blue line shows the corresponding estimates for those in informal wage jobs pre-COVID. The estimate for each month (of the year 2020) corresponds to a panel regression using data on the cohort of CPHS sample households that were revisited in that month. For example, the April estimates correspond to a regression on a panel of household surveyed in December 2019 and revisited in April 2019, while the estimate for May corresponds to a regression on a panel of household surveyed in January 2019 and May 2020. See Appendix 3.A for a technical description and regression tables. Source: Based on staff calculations using CPHS panel data on India.
of COVID-19. As discussed, the rates of transition into unemployment/OLF in the national lockdown phase (April-May 2020) were higher among infor-
this differential reduction in the probability of be-
mal workers. This differential tendency for job loss
ing employed post-COVID among informal wage
is explored further in regressions that compare
workers was 4.2 percentage points. The differential
changes in the probability of employment post-
between informal and formal wage workers nar-
COVID across those initially in formal and infor-
rowed down as the COVID crisis evolved, and was
mal jobs. The regressions control for differential
statistically not significant in June and July 2020.
changes in the probability of employment post-
In contrast to daily wage, casual, and temporary
COVID across industry, occupation and location.
workers, self-employed individuals do not show
These controls matter because the formal and in-
significant vulnerability post-COVID relative to
formal sectors have different industry, occupation
formal workers (Figure 3.7). These patterns are ro-
and location profiles. For example, the informal
bust to controlling for additional attributes such as
sector is more concentrated in sectors such as retail
occupation, education and caste.
and hospitality, which were hit hard by the lockdown (see model-based estimates, below). The regressions suggest that informal wage work-
The evidence from the CPHS suggests that labor markets remain in turmoil
ers were inherently more vulnerable than formal
Even though headline employment numbers may
employees to the early COVID-19 employment
be recovering, South Asian economies are not back
shock. Within the same industry and location
to normal. In particular, the fact that wage work-
(district), the decrease in the probability of being
ers, including formal wage workers, have moved
employed post-COVID, as observed in April 2020,
into self-employment at an unprecedented rate
was 6.4 percent points higher among those initially
reveals that a complex chain of adverse effects on
in informal wage jobs, compared to those initial-
labor demand by firms is still in play. Besides pre-
ly in formal wage jobs (Figure 3.7).5 In May 2020,
cipitating the lockdowns, the COVID-19 crisis has
5 Estimates based on differences-in-differences panel regressions using CPHS data with industry-wave and district wave fixed effects included among the control variables. See Appendix 3.A for a technical description and regression tables.
73
74
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
also hurt firms’ output demand, input supply, labor
that they were less likely to comply with mandated
supply and liquidity, while adding to their uncer-
lockdowns and related measures, whether out of
tainty (World Bank, 2020a).
desperation or due to a lack of proper information. Understanding this channel better could help de-
These channels of impact may have differed across
sign policies that make the recovery from COVID
industries, locations and over time. For example,
more robust by instilling better health and safety
firms more dependent on imported inputs from
measures in firms.
China were likely more vulnerable to external supply-chain disruption in the early stage of the crisis, while those exporting to OECD countries were more vulnerable to external demand shocks later on. In contrast, prior research would suggest that firms producing durables are more exposed to
Model simulations indicate that COVID-19 has particularly harmed informal sector workers
the ongoing domestic demand slump (Eaton et al. 2016; Levchenko et al. 2010). Small, informal firms
Complementing the still limited real-time data on
are particularly vulnerable to shocks as they tend
household incomes (see Box 3.2 for an example
to have less cash on hand and more limited access
of almost instantaneous data collection for some
to credit. For example, the “COVID-19 Business
specific subregions in Bangladesh), this section an-
Pulse Surveys”, conducted recently in 46 countries
alyzes the impact of COVID-19 with a macro-mi-
by the World Bank and the International Finance
cro simulation model that calculates the impact
Corporation, find that on average micro, small and
based on the characteristics of households before
medium-sized firms are significantly more likely
the pandemic and changes in aggregate employ-
than large-sized firms to fall into arrears in the
ment and prices during the pandemic. The ‘macro’
next six months (Apedo-Amah, M.C. et al., 2020).
sectoral employment losses are mapped to specific individuals using a Probit model, and the ensuing
The resilience of self-employment points to how
losses of labor income are then reflected in losses
individuals may have attempted to cushion the
of welfare at the household level (see Box 3.1 for a
immediate shock from COVID-19 in the absence
more detailed description).
of access to adequate safety nets. Potentially, several factors make self-employment a fallback option
In the three countries we examine, simulations
for those who have lost jobs. Such microenterprise
show that informal workers are over-represented
activity is less embedded in formal credit markets
among those likely to face job losses, especially in
and in complex supply chains than large formal
urban areas (top three graphs in Figure 3.8). Fig-
firms, and hence may be less exposed to credit and
ure 3.8 shows, for each percentile of the earnings
supply chain shocks emanating from COVID-19.
distribution – i.e. for each one percent of the pop-
Operating on a day-to-day basis with basic tech-
ulation earning an income ordered from the lowest
nologies and skills, such businesses may be more
to the highest group – the share of people likely
flexible in responding to demand shocks. There
to be newly unemployed and belonging to either
are also fewer entry barriers to microenterprise
the informal sector (green line) or the formal sec-
activity.
tor (dark blue line). While informal employment is 74 percent in urban India, the share among those
However, it is unclear if such marginal, necessi-
newly unemployed by the shock is 83 percent. The
ty-borne microenterprise activity can survive for
unemployment shock affects informal workers
long in the event of a prolonged downturn, or
across the full range of the earnings distribution,
form the basis of a thriving recovery. As discussed
but the negative slope of the incidence curves indi-
later in this chapter, the vast majority of microen-
cates that workers most affected were located, be-
terprises in South Asia already had very low val-
fore the job loss, at the low or middle percentiles
ue added per worker in pre-COVID times. The
of the distribution and high earners are relatively
downturn in demand in the current post-COVID
insulated from the shock.
environment may have reduced the income of the self-employed to even lower levels.
The distributional impact is more varied across countries in rural areas (bottom panel of Fig-
Another potential explanation of the resilience of
ure 3.8). Across the three countries, rural infor-
microenterprises in the early phase of COVID-19 is
mality is rife due to the large agricultural sector,
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Figure 3.8: Newly unemployed are mainly urban and informal workers India - Urban 20 18 16 14 12 10 8 6 4 2 0
1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97
1
Earnings percentiles (Poorest << Richest)
Percentage of workers (%) 10
15
25
32
41
54
63
73
81
87
95
99
1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97
Earnings percentiles (Poorest << Richest)
1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97
1
5
Earnings percentiles (Poorest << Richest)
Percentage of workers (%)
India - Rural 10 9 8 7 6 5 4 3 2 1 0
Percentage of workers (%)
Percentage of workers (%)
5
16 14 12 10 8 6 4 2 0
Earnings percentiles (Poorest << Richest)
Bangladesh - Rural 10 9 8 7 6 5 4 3 2 1 0
Pakistan - Urban 20 18
Percentage of workers (%)
Percentage of workers (%)
Bangladesh - Urban 20 18 16 14 12 10 8 6 4 2 0
11
18
25
32 43 52
61
71
76 84 92 96 100
Pakistan - Rural 10 9 8 7 6 5 4 3 2 1 0 1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93 97
Earnings percentiles (Poorest << Richest) Informal – newly unemployed
Earnings percentiles (Poorest << Richest)
Formal – newly unemployed
Note: The lines in each panel represents the ratios of new unemployed persons to total number of workers in each percentile. The lines are obtained using a polynomial regression of the actual ratios. Source: Authors calculations using data from Bangladesh (LFS 2015-16, left), India (PLFS 2017-18, center) and Pakistan (LFS 2017-18, right) and model simulations.
while the services sectors most hit by the crisis are
The rise in the food prices by more than non-food
less important. Pakistan is a slight exception, with
prices – linked to hoarding and food supply chain
seemingly a higher incidence of the shock in rural
disruptions – compounded the deterioration in in-
areas. However, this is most likely due to the sam-
come distribution owing to the pandemic’s impact
ple bias that does not capture the agricultural sec-
on informal workers. This relative price change has
tor when reporting wages (refer to Appendix 3.B).
a regressive impact since food is a large share of total consumption for the poor (as per Engel’s Law). A
Urban informal workers are particularly vulnerable
rise in the prices of food relative to non food by 11
because of their sectors of employment. High con-
percent in India, as estimated by the macro model,
tact-intensive urban services, like retail, transport,
thus has a further regressive impact on welfare.
accommodation, food, and tourism services were severely affected by the lockdowns, and on average,
The economic crisis is pushing many people into
90 percent of workers in these sectors are informal
poverty. The World Bank (2020b) (see also, Lak-
(Figure 3.9). These sectors have some of the highest
ner et al., 2019) estimates, on the basis of down-
shares of informal workers (apart from agriculture
side scenarios developed in June, that in the whole
and household activities--domestic work, subsis-
SAR region, between 49.3 and 56.5 million people
tence production of goods and services--where the
will have become poor by the end of 2020 com-
shares of informal workers are higher). Combined,
pared to a counterfactual without COVID-19. This
these sectors account for 13 to 20 percent of the
impact comprises the new poor, that is individuals
working age population and about 20 percent of
pushed into poverty by this crisis, as well as lost re-
labor income. While employment losses are clear-
duction in poverty, that is people who would have
ly concentrated in these most-affected sectors, the
escaped poverty had growth continued on its trend
overall ripple effects of the crisis are felt in other
before the pandemic. The increase in the num-
sectors as well.
ber of regional new poor will likely be even larger.
75
76
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
BOX 3.1: How to simulate the impact of the COVID-19 crisis The ex-ante assessment of the impact of the pandemic at the household level is carried out using a macro-micro simulation model (see Pereira da Silva et al., 2008; and Bourguignon and Bussolo 2013 for a survey of these methods). The objective is to assess how the crisis has affected the welfare of each individual in a
specific country. Starting with the micro model, real income per capita _ P at the household level can be used as the welfare indicator (Wh), and household per capita income (Yh) can be defined as the sum of household Yh
() h
members’ labor endowments (θ h,lwhere l represents the level of the skill of the worker which is linked to her education and sector-specific experience) rewarded by the market wages (wl ), and an exogenous income (Yh 0 ) as follows:
Yh = ∑ l θ h,l wl + Yh 0 The household-specific price index is for simplicity assumed to depend on the economy-wide prices of food
(pf ) and non-food (pnf ) items, weighted by the household consumption shares (φ h,f) of these consumption items:
Ph = p f φ h,f + pnf 1 − φ h,f
(
)
For each household, welfare effects can be approximated by the following expression: ∂ W _ ∂ Y ∂ Y ∂ W _ d Wh = _ ∂ θ , d θ h,l + _ ∂ Y { ∂ w d wl } + ∂ P dPh h
h
h
h
hl
l
h
h
This last equation determines changes in welfare as changes in household income and the household-spe-
cific price index. In the simulations, the budget shares φ h,f are kept fixed, and thus changes in the household price index depend only on changes of the food and nonfood economy-wide price indexes. Changes in household income are solely determined by changes in labor incomes and these, in turn, are allowed to
vary as a result of changes of workers endowments d θ h,l or, in the short term, by the intensity of the use of the endowment (a worker could lose her job), and the returns to labor in the different labor market seg-
(
)
ments (d wl ). A new household welfare aggregate is computed by adding the exogenous household income to the sum of simulated labor incomes for each member of the household (given her skill endowments and sector of employment) and deflating the new total household income by the new household-specific price index. In terms of welfare distribution, the initial distribution for ‘year’ t – representing the equilibrium before the COVID-19 shock hits the economy – for a population of N households can be written as:
Dt = { W1, t…W N ,t} = { f(Y1, t, P1, t)…f(YN ,t, PN ,t)}
(1)
The microsimulation consists of using new values for θ h,l, w l , p f and p nf , that represent the situation after the COVID has hit the economy, and equation (1) to compute new households’ real incomes and to generate a simulated new distribution:
ˆ t = { D W ˆ 1,t…W ˆ N,t} = { f(Y ˆ 1,t, P ˆ 1,t)…f(Y ˆ N,t, P ˆ N,t)}
(1’)
The macro model is used to compute the new endowments or, since these do not change in the short term, their aggregate intensity of use (employment and unemployment levels), and the other prices. The macro model used here is a global Computable General Equilibrium (CGE) model which includes all the individual countries in the South Asia region (with the exception of Bhutan and Maldives) and the other aggregated regional economies: Sub-Saharan Africa, Middle East and North Africa, Europe and Central Asia, Latin America and the Caribbean, East Asia and the Pacific, and the high-income countries. The model is implemented in comparative statics mode, and its main features (full documentation for the model is available in van der Mensbrugghe, 2013) are as follows. In each country, production is modeled using nested CES (Constant Elasticity of Substitution) functions that combine at various levels, with different substitution elasticities, intermediates and primary factors. Households’ consumption demand is derived from maximization of
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
household utility, whose argument is a composite good of imported and domestically produced varieties. Export supply is modeled as a Constant Elasticity of Transformation (CET) function. Producers decide to allocate their output to domestic or foreign markets responding to relative prices. The labor market specification is an important driver of the distributional results, so its specification calls for some clarification. Two types of labor are distinguished: skilled and unskilled. These categories are considered imperfectly substitutable inputs in the production process. Moreover, factor market segmentation is assumed: workers are not mobile across sectors. The labor market segmentation by skill level is a standard assumption, while the further segmentation by sectors is adopted to capture the very short-term impact, when workers cannot move freely across sectors. Wages are fixed in contracting sectors and are rising in expanding sectors, thus the adjustment to the shock is in quantities (unemployment) for the contracting sectors and in prices for the expanding ones. The COVID shock in the CGE model is simulated as the cumulative effect of seven separate shocks: i) all regions outside of South Asia are hit by a reduction of total factor productivity; ii) global fossil fuel prices drop by 10 percent and the global price of ores drops by 3 percent; iii) the cost of international trade rises; iv) the cost of tourism-related services is increased by 50 percent; v) South Asia countries are hit by a TFP reduction; vi) demand for services requiring face-to-face transactions is reduced by 15 percent and preferences for other goods and services are increased proportionately; and vii) domestic food supply chains are hit by disruptions. The first four represent the external shocks, and the next three represent the domestic shock. Note that these shocks are calibrated so that the GDP growth impacts (endogenously calculated in the model) are the same as those described in Chapter 2. For example, in the case of India, as shown in Figure 1, the comparison of the forecasts of GDP growth with COVID versus a counterfactual growth without COVID indicates an overall shock of 16 percent of GDP.
Figure 1: The growth loss due to COVID is significant Change in GDP forecasts
8
Real GDP growth ( % )
6 -16 pp
4 2 0 -2
FY2017
FY2018
FY2019
FY2020
-4 -6 -8 -10 -12 Without Covid-19
With Covid-19
Source: Chapter 2 in this report
The general equilibrium results from these simulations and, specifically, the overall reduction of income, unemployment by sector, and the food and non-food prices are used to shift the distribution Dt as described above. The sectoral unemployment results are mapped to specific individuals by using a Probit model estimated with household survey data. This probabilistic model uses characteristics like gender, age, education and household composition to predict an individual’s ties to the labor market, i.e. her employment status. Figure 2 below shows the estimation results for the case of India, highlighting clearly that, for example, higher-educated workers are more likely to be employed or, equivalently, are less likely to lose their job.1 In other words, many of the newly unemployed people will be the less educated workers. In the simulations, workers switch between employment and unemployment until the CGE-estimated unemployment levels of the 1 A validation of these estimates is provided by comparing them with those estimated from actual transitions from employment to unemployment in ex-post data (from the CMIE survey). The coefficients of predictors – gender, age, education and household characteristics – have the same signs and similar magnitudes. Geographic location is an exception; whereas individuals in urban India are more likely to be employed according to the PLFS 2017-18 sample, the direction is reversed in CMIE data for April 2020. This is coherent with the sectoral impact of the pandemic.
77
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
COVID scenario are achieved. Newly unemployed lose their labor income but, unless all other members of the household are also unemployed, their final per capita income is not equal to zero. At this point, the impact of the COVID shock can be assessed by comparing the standard inequality and poverty indicators of the initial distribution (Dt ) against those of the counterfactual distribution (D ˆ t).
Figure 2: Correlates of employment status Marginal effects from Probit regressions (95% Cls) Urban Female Currently Married Widowed Divorced Age Ageˆ2 Education Educationˆ2 Household Size -.05
0
.05
.10
.15
Marginal effects Source: Authors’ calculations using India’s PLFS (2017-18)
Figure 3.9: High-intensity, face-contact services employ informal workers in high proportions Informality by broad industry - SAR [country values weighted by labor shares] 100 94
89
89
88
91
89
85
81
80 Percentage share of total labor
66
95
83
61
62
60 51 40
48 35
33
39 33
33
20
13
Extraterritorial
Households activities
Other service
Arts, entertainment
Health & social
Education
Public admin.
Administrative
Professional
Real estate
Financial
Information & communication
Accommodation & Food
Transportation & storage
Wholesale & retail
Construction
Water
Electricity
Manufacturing
Mining
0 Agriculture
78
Note: Aggregates shown in the graph refer to the weighted average of four countries – Bangladesh, India, Pakistan and Sri Lanka. Source: Bangladesh LFS (2017), India PLFS (2017-18), Pakistan LFS (2017), Sri Lanka LFS (2015).
As explained in Chapter 2 of this publication, the
of many of the already poor falls further below the
growth outlook has further deteriorated in some
poverty line.
countries, even compared to the June downside scenario. Moreover, the estimate does not account
The pandemic has had a severe impact on the
for the increases in inequality described in this
middle of the income distribution as well. Workers
chapter. So, the region should anticipate a massive
in high-intensity face-contact sectors tend to live
increase in the number of poor, while the income
in urban households with members employed in
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Figure 3.10: Informal workers are found across the whole distribution of earnings Bangladesh 25
Percentage of workers
20 15 Affected sectors
10 Informal in -affected sectors
5 0
Formal affected sectors 1
3
5
8
12
14
19
22
30
37 49 51 58 60 69 Earnings Percentiles (Poorest<<richest)
75
82
87
89
92
94
96
98 100
India
Percentage of workers
25
Affected sectors
20 Informal in affected sectors 15 10 5 Formal in affected sectors 0
1
3
5
8 10 14 16 18 22 25 27 30 35 38 43 45 48 54 56 63 65 67 69 71 76 78 82 84 87 89 91 93 95 97 99 Earnings Percentiles (Poorest << Richest)
Percentage of workers
Pakistan 25 20 15 Informal in affected sectors
Affected sectors
10 5 Formal in affected sectors 0
1
3
5
7 10 12
1517 21 23 26 32 34 43 45 51 56 61 63 65 69 74 77 80 83 86 88 90 93 95 97 99 Earnings Percentiles (Poorest<<richest)
Source: Bangladesh (LFS 2015-16, left); India (PLFS 2017-18, center); Pakistan (LFS 2017-18, right)
different sectors, and thus they are not all at the
most affected by the pandemic (the red line), and
bottom of the distribution, where mainly rural ag-
the share of workers who are informal and working
riculture-dependent households are found. Infor-
in these sectors (green line) or formal and working
mal workers in these most affected sectors can be
in these sectors (blue line). Almost all the workers
found across the full distribution of earnings (see
in the bottom 50 percent of the earnings distri-
Figure 3.10 for Bangladesh, India and Pakistan).
bution in Bangladesh and India are informal. It is
Thus, the workers with the greatest probability of
only for the higher parts of the distribution that the
suffering unemployment and large losses in in-
shares of formal workers in these sectors become
come have varied income levels; large losses are
significant.
not restricted to the poor. Figure 3.10 shows, for each percentile of the earnings distribution – i.e.
The simulation model calculates that the fall in
for each one percent of the population earning
employment would have reduced consumption per
an income ordered from the lowest to the highest
capita in real terms by around 5 to 16 percent in the
group – the share of people working in the sectors
three countries in line with the difference in GDP
79
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Figure 3.11: Welfare losses of anonymous incidence curves are concentrated at the bottom BGD
PAK
−12
0
10
20
30
40
50
60
70
80
90
−6
−4
−2
0 0
100
−8
Percentage change in welfare (%)
−30
−12
−10
Percentage change in welfare (%) −25 −20 −15
Percentage change in welfare (%) −10 −8 −6 −4
−2
−10
IND
10
Earming percentiles: Poor <<Rich
20
30
40
50
60
70
80
90
100
0
10
Consumption percentiles: Poor<<Rich Uniform shock
20
30
40
50
60
70
80
90
100
Earming percentiles: Poor <<Rich
Employment + Food price shock
Note 1: The change in mean welfare (the average consumption per capita in India, and the average labor earnings per capita in Bangladesh and Pakistan) is the same, within each country, across the two lines presented in the graphs above. In the case of the straight line, the uniform shock, the welfare variable decreases by the same amount for all individuals, and this is the same decrease of the mean. The red line representing the employment and food price shocks emphasizes the distributional impact, as the average impact is the same as the straight line. Note 2: The red lines are produced with the LOWESS (Locally Weighted Scatterplot Smoothing) approach. This performs locally weighted regressions to generate a smoothed function. As it is based on many iterative model fits and not just one, it does not yield a unique confidence interval; the dots which represents the actual points show the dispersion around the line. Source: Authors calculations using data from Bangladesh (LFS 2015-16, left), India (PLFS 2017-18, center) and Pakistan (LFS 2017-18, right) and model simulations.
Figure 3.12: Non-anonymous incidence of welfare losses IND
PAK
0
10
20
30
40
50
60
70
80
Earning percentiles: Poor<<Rich
90
100
5 Percentage change in welfare (%) −10 −5 0 −20
−20
−12
−15
Percentage change in welfare (%) −18 −16 −14
Percentage change in welfare (%) −10 −8 −6 −4 −2
0
−12
BGD
−14
80
0
10
20
30
40
50
60
70
80
90
100
0
Consumption percentiles: Poor<<Rich Individuals in Informal HHs
10
20
30
40
50
60
70
80
90
100
Earning percentiles: Poor<<Rich
Individuals in Formal HHs
Note: As for Figure 3.11, the lines are produced with the LOWESS approach. Source: Authors calculations using data from Bangladesh (LFS 2015-16, left), India (PLFS 2017-18, center) and Pakistan (LFS 2017-18, right) and model simulations.
growth rates between the scenario with COVID-19
percentiles, is slightly misleading. Figure 3.11 com-
and that without (refer to Box 3.1). This loss, if it
pares the per capita consumption of each percen-
were affecting every one equally would be repre-
tile before and after the shock, irrespective of the
sented by a straight line, as in Figure 3.11. As can be
composition (or identity) of the households in the
seen, however, the poorer income groups in India,
percentiles, which is common practice in analyz-
Bangladesh and Pakistan suffer a greater fall in per
ing survey data.6 However, Figure 3.11 does not
capita consumption than the richer income groups
account for the fact that households move ranks,
do owing to the rise in unemployment. Similarly,
i.e. move across percentiles in the distribution.
in the three countries the rise in food prices hurts
If instead one calculates the change in per capita
the poor most.
consumption of the households based on their original percentile, the percentile they belonged
The impression given by Figure 3.11, that the
to before the shock, a different picture emerges (see
impact of the pandemic on the poorest is great-
Figure 3.12, which is based on the same informa-
er than the impact on households in the middle
tion as Figure 3.11).
6
The same approach is used in the calculation of the World Bank’s shared prosperity indicator.
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Workers and households in each percentile are not identical and are not equally hit by the employment shock. Within a specific percentile, some workers lose their jobs and consequently, all or a large share of their household incomes. These loss-
The informal sector cannot (especially in the short run) expect huge benefits from key aspects of the digital economy
es shift the ranking of the incomes of these workers and place them, after the shock, in lower per-
Three important innovations of the digital econ-
centiles. For example, in the case of India, almost
omy, teleworking, the “gig economy” and online
7 percent of workers who were, before the shock,
platforms, are likely to benefit higher-income
in percentiles 31 to 35 sustain substantial income
workers, most of them in the formal sector, much
losses of about 35 percent. These large losses push
more than informal workers.
these workers well below the average incomes of the poorest 10 percent, who also experience losses,
Teleworking
but not as large and thus move up in the income
helped formal workers, a divide that has become
(or consumption) ranks. This re-ranking results
particularly stark in the time of COVID-19.
in a shift in the composition of the percentiles.
specific occupations are amenable to teleworking
In the case of the anonymous growth incidence
given inherent task characteristics (Dingel and Nei-
curve of Figure 3.11, the lowest percentiles contain
man, 2020). For example, a computer programmer
the households who experienced the largest loss-
can work from home, but a construction worker
es, in this scenario households with members who
and domestic helper cannot. As illustrated in the
have lost employment, no matter where they were
case of India, the share of telework-friendly occu-
initially. The ensuing shift in composition of the
pations is significant only among workers located
percentiles tilt the line so that its upward slope ap-
in the upper reaches of the earnings distribution in
pears steeper compared with the non-anonymous
South Asia (Figure 3.13). In general, fewer than 10
curves in Figure 3.12. This is why the impact de-
percent of workers at the 70th and lower percentiles
picted in the anonymous incidence curve appears
of the earnings distribution can telework. Since
more regressive. Both results are valid, but in the
most of these are informal, this implies that only a
anonymous case the concern is on what happens to
small share of informal workers has access to tele-
the poorest, no matter who is at the bottom; in the
working possibilities.
technology has
disproportionally Only
non-anonymous case, the focus is to consider how each household is affected conditional on its initial
Given the disparities in access to digital technology,
economic position.
the percentage of informal workers who are actually able to work from home in South Asia is likely
Figure 3.12 also shows that informal workers, no
to be much lower than the potential upper bounds
matter at what level of income, suffer larger losses
shown in Figure 3.13. Even in occupations inher-
than formal workers, apart for the bottom 10 per-
ently amenable to telework, the actual possibility
cent in India (but, given the dispersion of the dots
of teleworking depends on the availability of com-
around the lines, the difference between formal
plementary factors (such as stable internet connec-
and informal individuals is not statistically signif-
tions) that may not be available at lower incomes.
icant at the bottom of the distribution). The online “gig economy” does offer considerable Indeed, preliminary survey data from India and
potential for informal jobs, but almost exclusive-
Bangladesh confirm the view that the economic
ly for educated workers. Employers worldwide are
impact of the pandemic was more likely to be re-
increasingly using online labor platforms to hire
ported by households in both the bottom and the
“online gig workers” for specific projects on hourly
middle of the income distribution. For India, Ber-
rates or piece rates. Millions of workers are thought
trand, Krishnan, and Schofield (2020) using CMIE
to participate in online gig works (Kuek et al., 2015),
data find that overall 84 percent of households
with the world’s two largest online platforms being
report suffering income losses. However, the re-
visited by about 200 and 75 million unique indi-
porting of losses is higher for the second and third
viduals per month, respectively (Kassi and Leh-
quintiles of the distribution, where more than 90
donvirta, 2018).
percent of household have experienced declines in income. The case of Bangladesh is described in
While online gig work can offer workers more flex-
Box 3.2.
ibility and higher returns by reducing job search
81
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
BOX 3.2: Early insights from Bangladesh - Informal workers and women are losing livelihoods, and considerable uncertainty remains * Panel data from Bangladesh provides early insights into the evolving labor market impacts of the COVID-19 crisis. These data are representative of some parts of the country which are particularly vulnerable to the crisis because of their density, for which baselines were collected before COVID. Follow up phone monitoring surveys were implemented in Dhaka, Chittagong City Corporations (conducted in June-July, 2020) and Cox’s Bazaar district (conducted in April-May 2020). Labor markets are a key channel through which welfare is affected, as labor incomes comprise more than 80 percent of household income for the poorest 40 percent of households (Hill and Genoni, 2019). A large share of Bangladeshi workers is engaged in sectors directly impacted by COVID-19. Compounded with pre-existing vulnerabilities and the absence of formal safety nets, households tend to manage income shocks with their own resources. According to the HIES 2016/17, about 25 percent of the population were living in poverty and another 54 percent could be considered vulnerable, as they had consumption levels very close to the poverty line (between the official upper poverty line and twice the line, refer to Figure 1 below).
Figure 1: Poverty and vulnerability by area (% of the population)
100
Percentage of population (%)
82
22
19
54
55
30
75
50
51
25 27
25
0
National
19
Rural Poor
Vulnerable
Urban Middle class
Source: Authors’ calculations using HIES 2016/17.
Early employment impacts, suggested by the rapid monitoring surveys, are large in terms of jobs losses, absenteeism, and reduced earnings, in a context of high uncertainty about jobs prospects. In poor and slum areas of Dhaka and Chittagong CCs, 23 percent of adults had stopped work between March 25, when the official COVID-19 lockdown was announced, and the time of the interview. Figure 2 (Panel A, left) shows that group of respondents that stopped actively working is composed of people expecting to resume work (32 percent searching for a new job (41 percent), or exiting the labor force (27 percent). Compared to Dhaka and Chittagong, Cox’s Bazar district is less urbanized, with its urban areas being located relatively far from concentrations of recently displaced Rohingya. Although close to 90 percent of the Bangladeshi living in Cox’s Bazar reported being employed during the lockdown (Figure 2, Panel B, right), these employment rates mask high rates of temporary absence from work. Almost 2 out of 3 adults who reported being employed were in fact not actively working in the 7 days before the survey. In contrast, during the baseline period (March to August 2019), temporary absence from work among the employed was less than 1 percent. Given the informal nature of jobs held by the majority of active and temporarily absent workers who report themselves as being employed, it is difficult to predict how fully this employment will translate into active jobs in the medium term. Reported income losses were widespread across all three areas, among those who retained employment. Monthly salaried wage workers in Cox’s Bazar have been relatively protected in terms of income losses, while daily and weekly wage laborers faced much higher losses in income.
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Figure 2: Employment indicators from Dhaka and Chittagong; and Cox’s Bazar district A) Dhaka and Chittagong - Employment status among respondents who stopped active work after March 25 (% of adults)
B) Cox’s Bazar district - Labor force indicators between baseline and follow up
100
100
20
34
24
29
75
50
25
32
41
24
42
41
39
29
42
39
Percentage of adults (%)
Percentage of adults (%)
27
95
89
75 51
50
42
25
38
5 0
0 All
Dhaka
Chittagong
Slum
Non-Slum
Temporarily absent
Unemployed
Labor force participation
Inactive
Employment rate
Baseline (Mar-Aug 2019)
11
Unemployment rate
Follow up (Apr-May 2020)
Source: Author’s calculations using the DIGNITY (Dhaka low Income area GeNder, Inclusion, and poverTY) survey and CITY (Chittagong low income area Inclusion, and PoverTy) survey follow-up from June-10 July 2020 (left); Cox’s Bazar Panel Survey (CBPS) follow-up from Apr-May 2020 (right)
Women have been disproportionately affected due to their overall lower participation in the labor market and their occupations. In Dhaka and Chittagong, women were more likely to leave the labor force, while in Cox’s Bazar, women have been more likely to look for work. In Cox’s Bazar, although unemployment rates increased across areas and gender, women in more urban, low-exposure areas were significantly more likely to become unemployed (Figure 3). However, this increase was not driven by job losses, but by new labor force entrants seeking jobs. In Dhaka and Chittagong, reductions in wages for salaried and daily workers were significantly higher for women, consistent with their high engagement in the garment sector and housemaid services, both of which have been severely impacted by COVID-19.
Figure 3: Cox’s Bazar district - increasing unemployment rates by gender and exposure area 16
Percentage of adults (%)
16 12
12
9 8
7
7 5
5
4
2
0 High exposure
Low exposure
High exposure
Male
Low exposure Female
Baseline (Mar-Aug 2019)
Follow up (Apr-May 2020)
Source: Cox’s Bazar Panel Survey (CBPS) follow-up from Apr-May 2020
The high level of job uncertainty in all three survey locations makes it difficult to infer the extent to which this crisis will translate into permanent job losses with longer-term consequences for poverty, food-security, and future earnings. There is some evidence of recovery in employment in ongoing second rounds of the phone surveys, although other dimensions such as earnings and certainty around earnings may take longer to recover, as there remains considerable uncertainty in the country on how widespread the health consequences of the pandemic will be, and whether more stringent social distancing measures will be enforced in the future. *This box was prepared by Luz Carazo, Maria Eugenia Genoni and Nandini Krishnan, and is based on “Losing Livelihoods: The Labor Market Impacts of COVID-19 in Bangladesh” (forthcoming)
83
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Figure 3.13: Teleworking status by wage percentile for India, formal vs informal
Percentage Of Workers (%)
84
50 45 40 35 30 25 20 15 10 5 0
Teleworkers
Informal Teleworkers 1
4
8
11
16
19
25
28
35
39
45
53
56
64
67
70
76
Formal Teleworkers 80
84
88
91
94
97
100
Earnings Percentiles (Poorest << Richest) Source: Staff calculation based on India’s PLFS 2017-18 and the occupational classification scheme in Dingel and Nieman (2020)
Figure 3.14: decomposition of OLI posts and workers India Bangladesh Pakistan United States Philippines United Kingdom Russia Ukraine Egypt Indonesia Sri Lanka Nepal Maldives Afghanistan Bhutan
United States United Kingdom Canada Australia India Germany Singapore Pakistan Bangladesh Sri Lanka Nepal Maldives Afghanistan 0
0.1
0.2 Share of posts
0.3
0.4
0
0.5
0.1
0.15 0.2 Share of workers
0.25
0.3
Note: The charts present OLI Post and Worker indices averaged over 2019 to calculate the share of posts and workers by country. Source: Staff calculations based on OLI (2019)
frictions, it also represents a new type of informal-
the total tasks posted while India (28 percent), Ban-
ity. Gig workers do not have formal employment
gladesh (12 percent), Pakistan (11 percent) and Sri
contracts or a long-term work relationship with its
Lanka (1 percent) account for more than half of the
associated protections (explicit or implicit). More-
workers on the online gig platforms (Figure 3.14).7
over, gig employers have considerable market
The occupational profile of online work suggests
power, which could allow them to extract a large
that it requires secondary school or higher edu-
share of the surplus generated by gig platforms
cation: most of the workers from South Asia (and
(Dube et al. 2020).
elsewhere) are engaged in tasks related to software development and technology, creative and
Online labor platforms increasingly allow edu-
multimedia industries, and sales and marketing
cated job seekers in South Asia to tap into foreign
support.8
markets for online tasks. Most job postings originate in high-income countries, while the majority
Online gig platforms may help workers cope with
of workers are from South Asia. As shown by data
domestic labor demand shocks, but in a limit-
from the Oxford Internet Institute’s Online Labor
ed sense. Since the demand for online gig jobs
Indices (OLI), the United States (39 percent), the
mainly originates from developed countries, it is
United Kingdoms (8 percent), Canada (7 percent)
largely insulated from purely domestic labor de-
and Australia (7 percent) account for two thirds of
mand shocks in South Asia. This can help domestic
7 Specifically, the OLI Index measures the number of new vacancies (tasks) posted on each day while the OLI Worker Index measures the number of workers active on the five online labor platforms in the 28 days. Counts of vacancies are normalized so that the value in May 2016 equals 100 index points. 8 Based on staff calculations using OLI.
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
0
100
Online labor index 200 300
400
Figure 3.15: OLI worker index, with SAR normalized to 100 in 2017
2017m10
2018m1
2018m4
2018m7
2018m10
2019m1
Rest of world
2019m4
2019m7
2019m10
2020m1
2020m4
2020m7
SAR
Source: Staff calculations based on OLI (2019)
workers in periods when domestic labor demand falls. Consistent with this hypothesis, cross-coun-
Table 3.1: Descriptive characteristics of firms with Facebook Business Page (percent shares) ROW
try panel regression analysis shows that country shares in the OLI worker index are countercyclical:
Total
(Mean)
that is, negatively related to domestic GDP growth, a proxy for domestic labor demand (see Table 3.C.1
SAR
Informal (if < 10 employees)
0.7
0.77
0.7
Engaged in international trade
0.29
0.32
0.29
Any ecommerce engagement
0.38
0.25
0.37
cantly even in normal times and declining precipi-
Access to credit
0.23
0.27
0.23
tously post-COVID (Figure 3.15).
Industry of operation Services
0.73
0.65
0.73
Manufacturing
0.14
0.17
0.14
informality is the result of limited firm capabilities
Construction
0.08
0.1
0.08
or human capital (as argued, for example, in Levy
Agriculture
0.05
0.09
0.05
Total Observations
63784
2018
65802
in Appendix 3.C). That said, the global gig economy itself seems to be highly volatile, with the total volume of online labor demand fluctuating signifi-
Finally, the growth of online platforms holds little promise for informal sector firms. To the extent that
2008 and Maloney 2004), the typical informal firm may have limited scope to benefit from online platforms. Firms that sell online are more likely to be
Source: Future of Business Survey 2019
large, export and cater to diverse markets (Kathuria,
e-commerce. Access to credit is low, but relative-
Grover, Perego, Mattoo, & Banerjee, 2019). Klapper,
ly higher in South Asian countries covered by the
Miller and Hess (2019) find that digital financial ser-
FBS (Bangladesh, India, Nepal and Pakistan).
vices make it easier for informal firms to register as formal businesses, but the benefits of formalization
Informal businesses on Facebook are more likely
may be most appealing to the larger informal firms.
to report that digital platforms (i.e. FBP) help their
These firms typically want to increase foreign sales,
businesses, and also tend to use them more often
purchase property and access credit; and even after
than formal firms (Table 3.2, Columns 1 and 2).
registering, they may require additional training.
They are also more likely to engage in e-commerce (Table 3.2, Column 5). Note, though, that this is a
Data from the “Future of Business” (FBS) survey,
self-selected sample of informal firms that have
which are representative of firms with an active
voluntarily signed up to FBP. It is also notable that
Facebook Business Page (FBP)–mostly small and
some of these tendencies are muted in the South
medium-sized— suggest that being on a digital
Asia region - only 30 percent of the population
platform does not eliminate key challenges for
use the internet; the global average is 50 percent
informal firms, such as expanding their market
(World Development Indicators, 2017). Although
reach or accessing credit. The rate of informality
digital technology may reduce some costs, it does
in the FBS sample is 70 percent, and most firms
not remove all barriers to informal businesses: in-
are in the services sector (Table 3.1). Despite hav-
formal businesses are still less likely to engage in
ing a digital presence on Facebook, only about a
international trade and have access to credit than
third of the firms engage in international trade and
formal firms (Table 3.2, Columns 3 and 4).
85
86
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Table 3.2: Regression results for firms with a Facebook Business Page Probit marginal effects Digital platforms help (1)
Frequent FB use for business (2)
Engages in intnl trade (3)
Access to credit (4)
Uses ecommerce (5)
Informal (if < 10 empl.)
0.0686*** (0.00670)
0.0689*** (0.00772)
-0.157*** -0.0069
-0.137*** -0.00727
0.0400*** -0.00841
SAR
-0.0281*** (0.00387)
-0.0839*** (0.00588)
-0.0673*** -0.00201
0.00216 -0.00158
0.0333*** -0.00135
Observations
36117
40306
48001
23450
24235
Pseudo R-squared
0.099
0.0821
0.0307
0.0465
0.0319
* p<0.10, ** p<0.05, *** p<0.010. Standard errors in parentheses are clustered at the country level. All regressions include country effects and control for firm level characteristics like age, gender and industry. Columns 1-2 use data from the survey round of 2019 winter and Column 3-5 use data from 2019 spring.
Understanding key characteristics of the informal sector is critical for policy effectiveness Government should take into account four aspects of the informal sector, including its pervasiveness and low value added per worker, its heterogeneity, the
Figure 3.16: Size of the informal economy Informal employment by region and decade (weighted avg by total labor force) 80 60 40
lack of social protection, and the barriers to formalization, in designing policies to support the sector. Three quarters (75 percent in 2017) of the region’s
20 0
workers are informal, the highest share among
ployed in formal firms. For example, in Bangla-
EAP
SSA 2000-2009
global regions (Figure 3.16). Note that a significant share of these informal workers is actually em-
SAR
LAC
MENA
ECA
2010-2017
Note: Graph excludes agriculture. Source: WDI.
desh, almost 16 percent of all the workers are informal workers working in formal firms. This share is
in larger informal firms add more value per capita
10 percent in the case of Pakistan, almost 8 percent
than those in informal microenterprises (Figure
in India, and about 3 percent in Sri Lanka. In other
3.18). Thus, the share of informal firms employing
words, the labor intensity of formal firms is differ-
more than 5 persons in the total revenue of the in-
ent if one includes or not the informal workers (for
formal manufacturing sector is much larger than
more details see Appendix 3.D).
their corresponding employment share (Figure 3.17). This heterogeneity within the informal sec-
Being concentrated in relatively labor-intensive
tor reflects variation in capital per worker, skills,
and low-skill intensive activities, the informal sec-
technology, scale economies and efficiency across
tor typically has lower low-value added per worker
informal firms of different sizes.
than the formal sector, and its share in total output is much smaller than that in total employment.
Similarly, while informal workers earn less than
In India’s manufacturing sector, for example, 85
formal workers on average, there is also significant
percent of employment is in informal firms, but
variation in their income levels. As illustrated in the
these jobs account for only 19 percent of the total
case of Pakistan and Bangladesh, there is a sizable
manufacturing sector revenue (Figure 3.17). The
presence of informal workers at every level of the
informal sector is critical to employment in South
distribution of earnings among South Asian work-
Asia precisely because it is labor intensive and low-
ers (Figure 3.19). Informal workers dominate the
skilled intensive.
lower half of the income distribution, but they are also found in large numbers in the upper half of the
Targeted policies are necessary to take into account
earnings distribution and are present even at its top
the heterogeneity of the informal sector. Workers
end.
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Figure 3.17: Size-wise breakdown of formal and informal firms in India’s manufacturing sector
8%
informal,5+
.6
3% 8%
Figure 3.18: Distributions of value added per worker in India’s formal and informal manufacturing sectors
29%
informal,1 formal
.4
62%
density
39%
informal,2−5
.2
81%
17%
34%
0
15% 4%
1%
Number Formal
Revenue Informal (>5) Informal (2-5)
Labor Informal (1)
4
6
8
10 12 14 ln(VA per worker)
16
18
20
Source: Staff calculations based on firm-level survey data from the Annual Survey of Industries and the National Sample Survey (NSS) of Unincorporated Enterprises 2015-16
Figure 3.19: Informal workers are found across the entire income distribution Bangladesh
Percentage share of workers (%)
100
Informal
80 60 40 20 Formal 0
0
3
6
9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87 90 93 96 99 Earnings percentiles (Poorer<<Richer)
Percentage share of workers (%)
Pakistan 100 80
Informal
60 40 20 Formal 0
0
3
6
9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87 90 93 96 99 Earnings percentiles (Poorer<<Richer)
Note: The green and blue lines depict the share of informal workers and formal workers, respectively, at every earnings percentile (x-axis). Source: Staff calculations based on Pakistan and Bangladesh Labor Force Surveys (LFS), 2017.
Informal workers are more vulnerable to unem-
financial buffers, likely causing faster layoffs and
ployment and less protected than formal sector
closures. In the farm sector, employment is almost
workers are. Informal workers are not covered by
entirely comprised of family workers and casu-
labor laws and social insurance schemes that ap-
al wage workers. In the non-farm sector, most of
ply to organized labor. Lacking formal contracts,
the workers are either self-employed or employed
they are more likely to be laid off in the event of a
in informal, unincorporated enterprises in which
negative shock. Furthermore, they are employed
there is not a clear separation between the unit of
disproportionally in smaller firms that have limited
production and its owner. The non-farm sector also
87
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Figure 3.20: Informality (stubbornly stable) over time 100 90 80 Percentage of total workers (%)
88
70 60 50 40 30 20 10 0
2008
2009 Bangladesh
2010
2011 India
2012 Maldives
2013
2014 Nepal
2015 Pakistan
2016
2017
2018
Sri Lanka
Source: ILOSTAT
includes a sizable number of workers who are em-
along the income distribution matter. Relief
ployed in formal firms under informal contracts.
measures should, for example, consider the high
Overall, informality “is more prevalent in smaller
vulnerability of casual and temporary wage work-
firms, more marginal locations, more rudimentary
ers. Expanding income assistance in the form of
activities, and among less educated people� (Loay-
transfers linked to a poverty threshold may be
za, 2018).
sufficient to protect some of these workers. And other forms of support, such as in-kind transfers
Finally, it is important to note for any discus-
or public works, may also be effective. However,
sion of how to prepare for future crises that it
since a large group of affected informal work-
is extremely difficult to achieve greater formal-
ers may not qualify for poverty-tested transfers,
ization. A view that formalization will gradually
there may be a need to loosen the conditions for
happen as incomes grow is contradicted by re-
eligibility. Different types of policy instruments
cent history. South Asia has experienced high
altogether are needed to support small informal
economic growth rates for at least two decades,
firms: here a great challenge is that of selection
but the share of informality has not changed,
and knowledge, in addition to budget resourc-
and in some cases has even increased, as shown
es. Successful policies need to support firms that
in Figure 3.20.
would be viable in the absence of the crisis, or even discriminate what specific constraint is af-
Supporting this large and diverse segment of the
fecting them, for example lack of capital or limit-
economy in South Asia is a crucial development
ed access to skills or to markets.
issue. A growth model that is not inclusive of the informal segment is ultimately not sustainable.
The second policy implication is that there is an
Moreover, the COVID crisis has quickly revealed
uneven playing field with respect to coping strat-
that existing disparities have been exacerbated and
egies, and this is linked to the institutional setting
risk leaving permanent scars. More fragmented so-
that is biased against informality. Formal workers
cieties will find it harder to shift to a sustainable,
or firms can access resources that are out of reach
inclusive economic path.
for their informal counterparts. For example, we showed that working from home is a realistic pos-
Policy implications
sibility only for a small share of workers, and these workers tend to be formal and concentrated at the top of the earnings distribution. Similarly, large
The analysis in this chapter has highlighted two
firms are more likely to have access to credit, ir-
relevant policy implications. The first is that the
respective of their profitability and other funda-
heterogeneity of informality and its incidence
mentals (Khwaja and Mian, 2005). In contrast, the
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
analysis illustrated that coping strategies for the
Table 3.3: Policy discussion, a simple framework Time frame Â
informal sector are more precarious. Self-employment tended to be resilient to the crisis or even operate as a buffer, as former wage workers become
Target
self-employed during the crisis. However, this resilience has been shown only in terms of employment status. While ‘not falling in unemployment’
Workers and households
may be read as resilience, income flows generated
Short run, relief
Long run, build better
Expansion of social assistance (food subsidies, public works, and cash transfers)
Towards universal social protection, expand social insurance; Closing digital gaps
Liquidity support (grants, credit)
Four main areas: 1) Capital 2) Access/matching with workers 3) Efficiency 4) Access to markets
by many informal self-employed during the crisis are close to subsistence. Access to credit or other forms of formal support, such as social insurance, is quite limited for these workers, and their own
Firms
savings are not sufficient. Indeed, only 30 percent of Indian households report being able to survive one month or more without additional assistance (Bertrand, Krishnan and Schofield, 2020).
Short-term and long-term policies in support of informal workers and households
This initial policy discussion can be organized ac-
In their short-term relief efforts, governments in
cording to Table 3.3.
the region have responded rapidly using their current systems (for a detailed example in the case
This table highlights that policy instruments should
of India, see Box 3.3). The COVID-19 response in
differ according to both their target beneficiaries
South Asia has mainly consisted of an expansion of
and the time frame. The short run measures aim
social assistance programs. This expansion, mak-
to tackle the immediate effects of COVID-19 by
ing up 76 percent of the social protection response
helping workers and firms stay afloat while longer
in the region (see Figure 3.21, from Gentilini et al.,
run policies would facilitate a recovery (Carranza et
2020), has been partly automatic, as people fell
al., 2020). Social protection policies are designed
below the eligibility thresholds, and partly dis-
to support workers and households and, while sup-
cretionary, by changing some of the rules. Social
port can be expanded to assist them in the short
insurance programs made up 20 percent of the
run, the crisis has also highlighted some long-term
response, while labor market programs accounted
challenges in the design of such policies. A similar
for 5 percent.
distinction can be made for policies helping firms. The next two subsections discuss these two sets of
Five of eight countries in South Asia have imple-
policies in more detail.
mented at least one cash transfer program or are
Figure 3.21: Social protection in COVID response, composition by region 100 90
7 9
18
11 20
6
4 16
14
18
80
32
23
70 Percentage share (%)
11
24
44
34
60 50 40
84
80
72
30
59
57
46
20
61 50
10 0 AFR
EAP
ECA Social Assistance
LAC
MNA Social Insurance
N. America
SAR
World
Labor Market
Note: Social assistance programs in South Asia include in-kind transfers such as food security programs as in Afghanistan, Bangladesh and Sri Lanka, and cash transfers. In principle these programs, which are not conditional on participation in the formal sector, might help address the impacts of the crisis on informal workers. Source: Gentilini et al. (2020)
89
90
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
BOX 3.3: Unpacking India’s COVID-19 Social Assistance Package* On 24th March 2020, the Government of India (GoI) ordered a nation-wide lockdown limiting the movement of 1.3 billion people as a preventive measure against COVID-19 pandemic. The resulting shock to economic activity threatened to drive millions to extreme poverty. In response, the GoI announced a substantial social assistance package – about one percent of India’s GDP – to provide immediate relief to the poor and vulnerable. The package, known as the Pradhan Mantri Garib Kalyan Yojana (PMGKY), extends the scope and raises the benefits of several existing social protection programs. Four components of the package stand out for their magnitude: food subsidies via the Public Distribution System (PDS), work guarantees under the Mahatma Gandhi National Rural Employment Guarantee Act (MGNREGA), cooking gas subsidies under the Pradhan Mantri Ujjwala Yojana (PMUY) and cash transfers to female bank account holders under the Pradhan Mantri Jan-Dhan Yojana (PMJDY). Taken together, these four components account for nearly 90 percent of the budget allocation (Figure 1). The implementation of the package is sensitive to the design features of each of the programs that are being scaled up. Implementation is also bound to depend on the resource envelope available to individual states, as well as on their administrative capacity. Therefore, the institutional setting is critically important to assess the amount of resources that will be spent on social assistance, and how it will be distributed across locations and households. Such assessment is difficult to conduct based on household surveys, as data on program coverage is limited, scattered across different survey instruments, and often dated. An alternative is to rely on administrative data from each of the four top programs. The largest component, PDS, covers about sixty percent of Indian families – nearly 200 million, with 2 million added since April 2020 alone. But the program data shows that not every family that is entitled to receive food subsidies under the PMGKY is doing so. It also shows that households are partially replacing subsidized food grains that they are entitled to under the National Food Security Act (NFSA) with free food grains under the PMGKY. Moreover, there is substantial variation in coverage, use and implementation across states. In the case of MGNREGA, an additional half a billion days of work was provided in the period between April and August 2020 relative to the same period in 2019. The budget allocation of the social assistance package was built on the assumption that many rural households would make use of the full 100 days of work they are entitled to. However, administrative data shows that the additional employment was mainly due to more households demanding work, not to more days worked per household. In fact, an analysis of the program data reveals that only 5 percent of the households with a job card worked for the full 100 days typically.
Figure 1: Program-wise Budget Allocation under PMGKY 12% 1% 6%
PDS MGNREGA
6%
PMJDY 50%
11%
PM-KISAN PMUJ Support to senior citizens, widows & Divyang Others
14% Source: Based on administrative data and authors’ calculations
The net transfers under the social assistance package can be computed considering features of program design and implementation like those described for PDS and MGNREGA, extrapolated to the entire fiscal year based on expected uptake. Overall, the transfers to households under the four top programs is about two-thirds of the budget allocation for the fiscal year 2020/21. While the transfers under the PMJDY and
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
MGNREGA are nearly 100 percent of the budget allocation, the transfers under the PDS and PMUY account for 60 percent and 39 percent of the budget allocation. The net transfer also varies considerably across India, ranging from less than 1 percent of aggregate household consumption in some states to 6 percent in others. Overall, the package is progressive in that households in poorer states receive a larger transfer. However, the package is also geared toward poorer and more rural states, whereas the crisis might have impacted informal workers in large metropolitan areas the most (Figure 2).
Figure 2: Household Transfers and COVID-19 Intensity 8000 MIZ CHH
7000
WB
RAJ Total transfers per household under PMGKY (INR)
6000
ODI KAR
UP MP
5000
AP BIH
JHA TRI
MEG 4000
MAH
TEL
HAR
GUJ
KER
HP 3000
SIK
GOA
TN
UTT 2000
1000 PUN
0 0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
Covid-19 cases per million (August 2020) Source: Based on administrative data and authors’ calculations *This Box is based on the background paper titled “Unpacking India’s Covid-19 Social Assistance Package” by Urmila Chatterjee and Deepak Varshney, forthcoming.
expanding cash payments through existing pro-
50 lakh poor families (5 million households) hit
grams in response to the pandemic. Overall, cash
hard by the crisis. Some countries are relaxing el-
transfer programs are projected to reach an addi-
igibility requirements. In Pakistan, eligibility for
tional 21 percent of the population in South Asia,
the means-tested “Ehsaas Kafaalat” cash transfer
on top of an already existing covered population
program is to be determined by using the nation-
of 15 percent (Gentilini et al., 2020). For example,
al socioeconomic database to relax eligibility re-
India has implemented a cash transfer of Rs 500
quirements during the pandemic. Similarly, the
($6.5) for 3 months to 200 million women with a
Indian states Delhi and Gujarat have expanded
Pradhan Mantri Jan Dhan Yojana (PMJDY) finan-
food security coverage to all residents, even those
cial inclusion account. In addition, Bangladesh is
in above-poverty line households.
giving Taka 2,245 ($30) to 5 million low-income families through mobile financial services.
The main requirement of these programs is identifying the poor, not whether a poor person works,
These programs are typically means-tested and
or used to work, in the formal or informal sector.
targeted to those with a low threshold income. For
In fact, these programs do not even need to iden-
example, Sri Lanka’s Samurdhi program, expand-
tify the shock, or the long-term reasons, that cause
ed its benefits to reach more vulnerable sections
someone to be in poverty. These are clear advan-
during the crisis. Similarly, in Bangladesh, Prime
tages, and the programs’ automatic and non-dis-
Minister Sheikh Hasina announced the disburse-
criminatory nature is important in assisting the
ment of Taka 1,250 crore cash assistance among
large group of people that the crisis has pushed
91
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Figure 3.22: Composition of social protection by component 100 90 80 Percentage of population (%)
92
1
70 60 50
20 61
40 30
19 30
0
42 31
20 10
4
21
16
3
27 14
9
EAP
ECA More than one SPL benefit
LAC
15
13
15
MNA
SAR
1 SSA
Only social assistance programs
Only social insurance programs
Source: ASPIRE database
into poverty. But, as shown by our analysis, the in-
highlighting the uneven institutional setting of
cidence of the shock is intense also for the people
current social protection systems. For example, in
above the poverty line, and informal workers in the
Nepal, individuals and firms get a 1-month exten-
middle of the distribution may have no access to
sion for loan payments if they have formal credit.
government relief support. This lack in coverage
And the Nepali government will pay one month of
can be quite concerning if some of these informal
Social Security Fund payments for both employee
workers are employing other people in their small
and employers associated with the Social Security
businesses.
Fund for formal workers. In India, the GoI amended EPFO regulations to allow workers to access a
In addition to the incidence issue just mentioned,
nonrefundable advance from their accounts. The
there may be other barriers to broad inclusion in
regulations will allow workers to withdraw 75 per-
these programs. Financial inclusion is one of these
cent or 3 months wages (whichever is lower). This
barriers. Many programs require recipients to have
also primarily benefits formal workers.
either a bank or mobile money account in order to receive funds. Yet, levels of financial inclusion
In the longer term, social protection systems and
in some South Asian countries remain persistent-
the financial sectors need to be redesigned to be
ly low, raising questions about how broadly these
more inclusive of the informal sector. The first way
programs will be able to reach those in need. For
to improve design of the social protection system
example, Bhutan’s Relief Kidu program website
is to expand its social insurance component. As
specifies that beneficiaries should either have a
shown in Figure 3.22, at one percent of the covered
bank account to facilitate receipt of the transfer
population, SAR has the lowest share of social in-
or collect via local administration. Similarly, Sri
surance coverage across the world. The main issue
Lanka’s “Samurdhi” allowance program requires
here is whether the current, limited coverage can
special bank accounts in the Samurdhi bank. Ban-
be extended as is to a larger group, or whether to
gladesh’s G2P programs require mobile money
design an alternative insurance system for the in-
accounts with bKash. In the Global Findex Survey
formal population. Antón, Trillo, and Levy (2012)
(2017), the share of individuals with a bank account
and Packard, et al. (2019) have an in-depth discus-
in the bottom 40 percent of the income distribu-
sion of the tradeoffs between a dual system and a
tion ranged from 14 percent in Pakistan and Af-
universal one.
ghanistan to above 70 percent in India and Sri Lanka. In Bangladesh, 21.2 percent reported having a
In the meantime, including the “missing middle”
mobile money account.
of informal workers could potentially be achieved by leveraging “big data”, such as mobile phone
In addition, informal workers have limited ac-
usage data, to better and more efficiently target
cess to other forms of government support,
those in need. Additionally, improving financial
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
inclusion – by also using non-bank financial insti-
economic
tutions, such as microfinance institutions – may fa-
South Asian governments introduced a range of
recovery
and
productivity
growth.
cilitate access to and participation in social protec-
measures to help firms tide over the temporary
tion programs and encourage resilience on its own.
shock. The types of assistance included grants,
By removing asymmetries in information, the
credit/credit relief, tax relief, assistance with pay-
technological revolution can deliver new types of
ing salaries of workers and assistance with or waiv-
social protection arrangements delinked from for-
er of utility payments. However, the determination
mal employment relationships of the past. These
of the ‘viability’ of firms (the selectivity issue) has
can be in the form of combined voluntary saving
been based on simplistic criteria, and the coverage
/ risk-pooling arrangements embedded in the gig
of small, informal enterprises by these short-term
economy platforms themselves.
measures has been a major concern. For example, measures such as tax relief and wage subsidies may
The World Bank and other international financial
have a limited impact on small enterprises that
institutions could have an important role in this
pay little or no tax and have few workers on formal
redesign effort. As discussed after the 2008 finan-
contracts.
cial crisis, the challenge is not to design the perfect social protection system against a specific crisis,
Our analysis suggests that six months into the cri-
but rather to design a system flexible enough to be
sis, wage employment in firms has only partially
effective for the current but also future crises. The
recovered. Liquidity support may still be need-
proposal to successfully achieve such redesign is
ed by many viable firms. A major challenge for
articulated in three parts (Kanbur, 2010 and 2012).
governments in South Asia is to plug the gaps in
First, given the uncertainty about the type of the
the coverage of small and informal firms that are
crisis (financial, climatic, civil unrest and displace-
mostly outside the tax and formal financial sys-
ment/migration, collapse of export prices, among
tem. Options include providing transfers through
others) and its timing (when the crisis happens and
suppliers, extending factoring and credit guaran-
how long it lasts), comprehensiveness and flexi-
tee mechanisms to informal firms, and supporting
bility are two key aspects to be considered in the
community-based financing institutions (World
redesign. It is also important to consider social
Bank, 2020a).
protection as a system rather than to evaluate separately its individual components. A potential way
Increasingly, short-term policies to address the
to do that is to conduct stress tests of the system
COVID-19 shock on firms need not just money,
against a range of crises. The results and recom-
but also more knowledge. The nature of the cri-
mendations could be gathered in a Social Protec-
sis is shifting, and it is no longer just a question of
tion Assessment Program (SPAP), similarly to how
supporting firms’ liquidity. Potentially, a range of
the resilience to shocks of the financial system is
market disruptions could be generating short-term
assessed in Financial Sector Assessment Programs
losses in productive firms. They include, for exam-
(FSAP) conducted jointly by the World Bank and
ple, issues with buyers, input supply chains and the
International Monetary Fund. Second, investment
labor supply. Each type of market failure or fric-
lending could finance improvements in the system
tion requires a different type of policy response
identified in the SPAPs, such as the potential gaps
and is not necessarily solved by just a grant or a
in coverage and difficulties in scalability. Third,
loan. It is therefore important to better understand
pre-qualified lines of credit could be set up to cover
the specific frictions and market failures affecting
automatically the needs of the redesigned system
firms, and to target the support accordingly.
when certain crisis triggers are breached. In the medium to longer term, keeping firms alive
Short-term and long-term policies in support of informal firms
ceases to be a suitable policy objective, and address-
Supporting firms in addressing their liquidity
reallocation will become increasingly important for
problems has been an urgent policy imperative
recovering from the crisis. Evidence from the US
during the stringent lockdown period (World Bank,
suggests that COVID-19 crisis has changed produc-
2020a). This was needed to avert the danger that a
tion and demand patterns and requires a realloca-
temporary lockdown would force viable firms to
tion of resources across firms to re-optimize their
close permanently and lose their organizational
use (Barrero et al., 2020). While many firms in the
capital, with adverse consequences for long-term
US are shrinking their workforce and even expect a
ing barriers to new business creation and resource
93
94
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
permanent reduction in their workforce, others are
impact on profitability, grants have been found
hiring workers as they expect a growth in demand.
to be effective in a range of contexts (Woodruff,
There are hints of a similar reallocation process in
2018; Jayachandran, 2020). For example, a one-
the labor market transition analysis discussed in this
time grant had long-lasting positive impacts
chapter, with recent months seeing unusually high
on microenterprises in Sri Lanka (de Mel et al.
rates of outflows and inflows into employment. Re-
2012b). With regards to strengthening business
lief measures that could inadvertently hamper real-
capabilities of microenterprise, classroom train-
location, such as subsidies for employees’ retention,
ing is less effective than one-on-one training pro-
should be phased out.
vided through consulting or mentoring (Woodruff, 2018).
With self-employment growing and becoming a key coping mechanism during the COVID-19 crisis, it is
The crisis has underlined the long-term value of
also increasingly critical to extend support to micro-
addressing barriers to the growth of small firms
enterprises and the self-employed to help them tide
in the informal sector. Simply making it easier
over the shock, and in the long run, raise their pro-
to register has limited impact on the productivi-
ductivity. Urgent support measures could include
ty and growth of informal enterprises (Bruhn and
one-time grants or microcredit. Evidence from the
McKenzie, 2014). Instead, policy needs to address
highly destructive 2004 tsunami in Sri Lanka sug-
supply and demand side market failures that keep
gests that microenterprise recovery following a ma-
small firms from realizing their growth potential
jor shock can take longer than a year, and that an
though systemic reforms, institutional strength-
early injection of capital can speed up the recovery
ening and targeted programs. Important supply
process significantly (de Mel et al. 2012a).
side constraints include access to credit, skilled workers, technology and organizational knowl-
In the even longer run (rebuilding better phase),
edge. The demand side also matters, as stronger
policymakers should consider strengthening fi-
linkages to markets can improve the productivity
nancial support and business training programs
of small firms by enabling competition and infor-
that can help improve the productivity of micro-
mation flows. It is also worth stressing that the evi-
enterprises. The form of support will matter. The
dence base on the growth of microenterprises and
evidence base on financial support to microen-
small firms is still evolving, and that investing in
terprises, while limited, favors grants over loans:
this evidence base in the context of South Asia is
while loans are generally found to have a limited
important.
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Conclusions The decline in demand and supply disruptions generated by the pandemic and the policies required to contain its spread have resulted in severe reductions in incomes in the South Asia region. Governments have begun to respond by expanding assistance to the poor and to affected firms. An effective policy response will require a clear understanding of which households and firms are most in need of assistance, and how to reach them. A key finding of this chapter is that workers in the bottom and the middle of the income distribution, mostly in the informal sector, experienced the largest loss in income. While policies are geared to assisting the poor, workers who had been in the middle of the distribution workers receive little in the way of social assistance because their previous incomes made them ineligible for these programs. At the same time, they lack access to the social insurance schemes and other forms of support available to formal sector workers. Finding ways to assist these workers will be critical to addressing the welfare losses from the pandemic. But the response to the pandemic cannot only focus on supporting incomes. The survival of many informal sector firms is threatened by what is hoped to be a temporary shock to their markets and access to supplies. These firms tend to be quite small and lack the savings and the financial access to keep afloat during this extended crisis. The likely disappearance of many otherwise viable firms as the crisis continues threatens the long-term welfare of informal sector workers and the productivity and growth of the economy as a whole. Providing effective and well-targeted assistance to such inherently productive firms while facilitating the entry of promising new firms will be a major challenge going forward. The pandemic highlights an important fact about South Asian economies. The informal sector dominates employment, but its productivity, and thus the incomes of workers, tends to be low. Improving these workers’ human capital and access to physical capital is key to maintaining high rates of growth. Thus, policies that focus on supporting productivity in the informal sector are critical to development, as well as to improving income distribution, in the region.
95
96
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
References Apedo-Amah, Marie Christine, Besart Avdiu, Xavier Cirera, Marcio Cruz, Elwyn Davies, Arti Grover, Leonardo Iacovone, Umut Kilinc, Denis Medvedev, Franklin Okechukwu Maduko, Stavros Poupakis, Jesica Torres and Trang Thu Tran. 2020. Businesses through the COVID-19 shock: Firm-level evidence from 46 countries. Working Paper, The World Bank. Barrero, J. M., Bloom, N., & Davis, S. J. (2020). COVID-19 is also a reallocation shock (No. w27137). National Bureau of Economic Research. Bertrand, Marianne, Kaushik Krishnan, and Heather Schofield. “How are Indian households coping under the COVID-19 lockdown? 8 key findings.” Rustandy Center for Social Sector Innovation Blog 11 (2020). Khwaja, Asim Ijaz, and Atif Mian. “Do lenders favor politically connected firms? Rent provision in an emerging financial market.” The Quarterly Journal of Economics 120, no. 4 (2005): 1371-1411. Bourguignon, F., & Bussolo, M. (2013). Income distribution in computable general equilibrium modeling. In Handbook of computable general equilibrium modeling (Vol. 1, pp. 1383-1437). Elsevier. Bruhn, M., & McKenzie, D. (2014). Entry regulation and the formalization of microenterprises in developing countries. The World Bank Research Observer, 29(2), 186-201. Bussolo, M., Kotia, A., Sharma, S. “Workers at risk: Panel data evidence on the COVID-19 labor market crisis in India”, forthcoming. Carranza, E., Farole, T., Gentilini, U., Morgandi, M., Packard, T., Santos, I., & Weber, M. (2020). Managing the Employment Impacts of the COVID-19 Crisis: Policy Options for Relief and Restructuring. de Mel, Suresh, Christopher Woodruff and David McKenzie 2012a. ‘Enterprise Recovery following Natural Disasters’, The Economic Journal 122 (559) 64-91 de Mel, Suresh, Christopher Woodruff and David McKenzie 2012b ‘One-time transfers of cash or capital have long-lasting effects on microenterprises in Sri Lanka’, Science 335 (6071) 962-966 Dingel, J. I., & Neiman, B. (2020). How many jobs can be done at home? (No. w26948). National Bureau of Economic Research. Dube, A., Jacobs, J., Naidu, S., & Suri, S. (2020). Monopsony in online labor markets. American Economic Review: Insights, 2(1), 33-46. Eaton, J., Kortum, S., Neiman, B., & Romalis, J. 2016. “Trade and the global recession”. American Economic Review, 106(11), 3401-38 Facebook, OECD and World Bank. 2019. World - Facebook: Future of Business Survey 2019. Gatti, Angel-Urdinola, Silva and Bodor (2014) Striving for Better Jobs: The Challenge of Informality in the Middle East and North Africa. World Bank, Washington D.C. Gentilini, U., Almenfi, M., Orton, I., & Dale, P. 2020. Social Protection and Jobs Responses to COVID-19. Goldfarb, Avi, and Catherine Tucker. 2019. Digital Economics. Journal of Economic Literature, 57 (1): 3-43. Hill, R., & Genoni, M. E. (2019). Bangladesh Poverty Assessment: Facing Old and New Frontiers in Poverty Reduction (No. 143200, pp. 1-83). The World Bank. International Labour Organization. (2013). Measuring informality: A statistical manual on the informal sector and informal employment. Geneva: International Labour Office. Jayachandran, S. (2020). Microentrepreneurship in developing countries (No. w26661). National Bureau of Economic Research. Kanbur, Ravi, 2010. “Protecting the Poor Against the Next Crisis,” Distinguished Lecture to Egyptian Center for Economic Studies, Cairo. Kanbur, Ravi, 2012. “Stress Testing for the Poverty Impacts of the Next Crisis,” in Ambar Narayan and Carolina Sánchez Páramo (eds.). Knowing, When You Do Not Know, The World Bank, pp. 50-55. Kässi, O., & Lehdonvirta, V. (2018). Online labour index: Measuring the online gig economy for policy and research. Technological forecasting and social change, 137, 241-248. Kässi, Otto, and Vili Lehdonvirta. “Online labour index: Measuring the online gig economy for policy and research.” Technological forecasting and social change 137 (2018): 241-248. Kathuria, S., Grover, A., Perego, V. M. E., Mattoo, A., & Banerjee, P. (2019). Unleashing e-commerce for South Asian integration. The World Bank. Klapper, L., Miller, M., & Hess, J. (2019). Leveraging Digital Financial Solutions to Promote Formal Business Participation. World Bank.
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Kuek, S. C., Paradi-Guilford, C., Fayomi, T., Imaizumi, S., Ipeirotis, P., Pina, P., & Singh, M. (2015). The global opportunity in online outsourcing. Lakner, C., Mahler, D. G., Negre, M., & Prydz, E. B. (2019). How much does reducing inequality matter for global poverty?. The World Bank. Leibovici, F., Santacreu, A. M., & Famiglietti, M. (2020a). Social distancing and contact-intensive occupations. On the Economy Blog, Federal Reserve Bank of St Louis. Leibovici, F., Santacreu, A. M., & Famiglietti, M. (2020b). How the impact of social distancing ripples through the economy. On the Economy Blog, Federal Reserve Bank of St Louis. Levchenko, A. A., Lewis, L. T., & Tesar, L. L. 2010. “The collapse of international trade during the 2008–09 crisis: in search of the smoking gun”. IMF Economic Review, 58(2), 214-253. Antón, A., Trillo, F.H. and Levy, S., 2012. The end of informality in México?: fiscal reform for universal social insurance (Vol. 1300). Washington, DC: Inter-American Development Bank. Loayza, N. (2018). Informality: Why Is It So Widespread and How Can It Be Reduced?. World Bank Research and Policy Briefs, (133110). Maloney, William. 2004. “Informality Revisited”, World Development 32(7): 1159-78. Pereira da Silva, L. A., Bourguignon, F., & Bussolo, M. (Eds.). (2008). The impact of macroeconomic policies on poverty and income distribution: macro-micro evaluation techniques and tools. The World Bank. Packard, T., Gentilini, U., Grosh, M., O’Keefe, P., Palacios, R., Robalino, D. and Santos, I., (2019). Protecting all: Risk sharing for a diverse and diversifying world of work. The World Bank. Van der Mensbrugghe, D. (2013). Modeling the global economy–Forward-looking scenarios for agriculture. In Handbook of Computable General Equilibrium Modeling (Vol. 1, pp. 933-994). Elsevier. Vyas, Mahesh, 2020. CPHS execution during the lockdown of 2020. Working Paper, The Center for Monitoring the Indian Economy. World Bank, 2016. World Development Report 2016: Digital Dividends. World Bank, Washington DC. World Bank, 2020a. “Assessing the impact and policy responses in support of private-sector firms in the context of the COVID-19 pandemic.” Policy Note. Finance, Competitiveness and Innovation Global Practice, The World Bank, Washington, D.C. World Bank, 2020b. “Poverty and Shared Prosperity 2020: Reversals of Fortune.” Washington, DC: World Bank. doi:10.1596/978-1-4648-1602-4. License: Creative Commons Attribution CC BY 3.0 IGO Woodruff, Christopher. 2018. Addressing constraints to small and growing businesses. International Growth Centre, London. www. theigc. org/wp-content/uploads/2018/11/IGC_ANDE-review-paper_finalrevised. pdf.
97
98
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Appendix A. Analysis of the Consumer Pyramids Household Survey (CPHS) of India The CPHS survey Consumer Pyramids Household Survey (CPHS) is administered on a panel of over 170,000 households across India thrice a year. The survey is typically conducted face-to-face but owing to the COVID lockdown in India after the third week of March, the face-to-face interview format was replaced with a telephonic one, allowing CMIE to continue gathering data. The response rate in comparison to the planned execution during the lockdown was a little over 60 per cent compared to over 95 per cent before the lockdown (Vyas, 2020). CMIE maintains that even with this reduced sample, their data is representative of the population across several dimensions. Notably, the rural-urban divide of the CPHS sample is typically about 37:63. In the first week of the lockdown (ending March 31), this shifted to 46:54 but was restored to pre-lockdown levels by week 3.
Sample selection The full CPDX sample of 174,000 households is surveyed over a four-month period, called a “wave”. Each wave of the survey is representative of the Indian population. Note that a household surveyed in one wave of the CPHS panel is resurveyed in the next wave approximately four months later. In this sense, the set of households that are covered in one full month of the survey can be considered as a monthly “cohort” that reappears in the CPHS panel after four months. The execution of the survey is planned and executed in a manner that the households surveyed each month are well distributed over the country. Hence, each monthly cohort gives a balanced picture of the country. India imposed its lockdown in the third week of March, which fell in the middle of wave 19 of the survey. To capture the impact of the COVID shock, we use wave 19 data of only those individuals who were interviewed in April 2020, i.e. after the survey. All transition tables and regressions are based on panel data observed for this “April 2020 cohort.” This cohort was surveyed earlier in August 2019 and December 2019. It was most recently surveyed in August 2020, during Wave 20. Given the balanced rollout of the survey, this sample spans 28 states of India. For the charts on the transitions between different labor force categories which are shown in the chapter, we use a balanced panel from August 2019 to August 2020, consisting of 20,126 individuals of the April cohort in the working age population. For the difference-in-difference regressions, we use a two-wave balanced panel for each of the four cohorts. CMIE provides weights that adjust for non-response. All figures and regressions use these weights to make our estimates representative of the working age population (individuals who are 15 years of age or more).
Variables and definitions We use data on employment status, employment arrangement, education, caste, district, industry, and occupation. A person is considered unemployed if she reports her employment status as “Unemployed, willing and looking for a job”. She is characterized as out of the labor force if she reports her employment status as “Unemployed, not willing and not looking for a job” or “Unemployed, willing but not looking for a job”.
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Labor market transition tables These tables show the “rates” at which working age individuals transition, or flow, from one labor force status to another between two periods, estimated using CPHS panel data on individuals’ labor market status in the initial and ending period. The rate of transition for a group of individuals who had the same status in the initial period is the breakdown (in percentage terms) of their ending period’s status.
Table 3.A.1: Labor market transition Aug-20
Apr-20
Formal
Self-employed
Casual
Unem/OLF
Total
Formal
37.38
28.90
22.01
11.71
100
Self-employed
4.26
68.97
14.35
12.42
100
Casual
3.78
27.76
49.72
18.74
100
Unem/OLF
0.98
10.15
10.87
78.00
100
Apr-20
Dec-19
Formal
38.30
21.19
9.82
30.69
100
Self-employed
4.00
55.90
6.23
33.87
100
Casual
3.53
18.82
20.31
57.34
100
Unem/OLF
0.73
3.00
1.65
94.62
100
Dec-19 Formal Aug-19
83.94
6.94
6.11
3.01
100
Self-employed
1.29
83.30
9.87
5.54
100
Casual
1.23
13.55
78.79
6.43
100
Unem/OLF
0.36
2.08
1.90
95.66
100
Difference-in-difference regressions We estimate the following regression specification: mployedit = α DWi + + β SEi + γ( DWi × POSTt ) + δ ( SEi × POSTt )+ η HighSchit + ρ Castei + ∑ θ t Xi + ε it E DWi is an indicator for being a daily wage (or casual or temporary salaried) worker, and SEi indicates
self-employed workers- both based on the baseline employment status. The omitted category is permanent (that is, formal) wage worker. P OSTt is a dummy variable that takes value 1 for April 2020 and all
following months. HighSchool is a dummy denoting workers with more than 10 years of education. Caste is a dummy equal to 1 if the caste category is OBC/SC/ST. ∑ θ t Xi indicate wave-interacted fixed effects. We use wave × occupation, wave × industry, and wave × district fixed effects. The above regression specifica-
tion is estimated for each of the four cohorts of households who repeatedly participate in the survey. The regressions are estimated separately for each cohort. For example, the regression using the households in the April cohort are estimated on a balanced panel for the December 2019 and April 2020 waves of this cohort. The main coefficients of interest are those on the interaction terms DWi × POSTt and S Ei × POSTt which
measure the differential change in the probability of employment post-COVID for the daily wage workers
and the self-employed, respectively, relative to formal wage workers. Tables 1-4 present detailed results for these regressions. Figure 3.7 in the chapter plots coefficients of DWi × POSTt and SEi × POSTt from column 1 of these tables.
99
100
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Table 3.A.2: Impact of COVID-19 on Employment (April cohort): DID estimates (1)
(2)
(3)
(DailyWage+Tem)
0.016*** (0.0021)
0.0067*** (0.0025)
0.0046** (0.0019)
Post X (DailyWage+Tem)
-0.064*** (0.011)
-0.017* (0.0098)
(Self Employed)
0.0073*** (0.0018)
Post X (Self Employed)
Table 3.A.3: Impact of COVID-19 on Employment (May cohort): DID estimates (1)
(2)
(3)
(DailyWage+Tem)
0.0083*** (0.0014)
0.0023* (0.0012)
0.0015 (0.0011)
-0.0013 (0.0094)
Post X (DailyWage+Tem)
-0.042*** (0.0095)
-0.0094 (0.0090)
0.0025 (0.0089)
-0.0010 (0.0039)
-0.0030 (0.0035)
(Self Employed)
0.0047*** (0.0012)
-0.00087 (0.0021)
-0.0017 (0.0020)
-0.0057 (0.011)
-0.010 (0.010)
0.0018 (0.0097)
Post X (Self Employed)
0.013 (0.0088)
-0.0073 (0.0085)
0.0038 (0.0084)
HighSchool
0.042*** (0.0047)
0.018*** (0.0043)
0.0089** (0.0040)
HighSchool
0.031*** (0.0042)
0.018*** (0.0043)
0.012*** (0.0042)
Caste
-0.026*** (0.0047)
-0.0043 (0.0042)
-0.0052 (0.0040)
Caste
-0.0066 (0.0041)
0.0022 (0.0039)
0.0041 (0.0039)
District X Wave FE
Yes
Yes
Yes
District X Wave FE
Yes
Yes
Yes
Industry X Wave FE
Yes
No
Yes
Industry X Wave FE
Yes
No
Yes
Occupation X Wave FE
No
Yes
Yes
Occupation X Wave FE
No
Yes
Yes
37007
37007
37007
42899
42899
42899
Observations
Observations
* p<0.10, ** p<0.05, *** p<0.010. Standard errors in parentheses are clustered at the individual level
* p<0.10, ** p<0.05, *** p<0.010. Standard errors in parentheses are clustered at the individual level
Table 3.A.4: Impact of COVID-19 on Employment (June cohort): DID estimates
Table 3.A.5: Impact of COVID-19 on Employment (July cohort): DID estimates
(1)
(2)
(3)
(1)
(2)
(3)
(DailyWage+Tem)
0.0017 (0.0022)
-0.00080 (0.0023)
-0.0012 (0.0023)
(DailyWage+Tem)
-0.010** (0.0049)
-0.00094 (0.0041)
-0.0023 (0.0049)
Post X (DailyWage+Tem)
-0.0082 (0.0091)
0.0096 (0.0092)
0.014 (0.0089)
Post X (DailyWage+Tem)
-0.0019 (0.0073)
-0.0066 (0.0068)
-0.0031 (0.0073)
(Self Employed)
0.00023 (0.0021)
-0.0019 (0.0035)
-0.0021 (0.0035)
(Self Employed)
-0.0093 (0.0057)
-0.014 (0.0086)
-0.015* (0.0087)
Post X (Self Employed)
0.0059 (0.0089)
-0.00020 (0.0095)
0.0034 (0.0091)
Post X (Self Employed)
0.0024 (0.0077)
0.0051 (0.010)
0.0082 (0.010)
HighSchool
0.010*** (0.0029)
0.0047 (0.0029)
0.0029 (0.0029)
HighSchool
0.00089 (0.0019)
-0.0015 (0.0021)
-0.0016 (0.0021)
Caste
-0.0056* (0.0030)
-0.0011 (0.0031)
-0.00052 (0.0029)
Caste
-0.0027 (0.0019)
-0.000048 (0.0019)
0.00011 (0.0019)
District X Wave FE
Yes
Yes
Yes
District X Wave FE
Yes
Yes
Yes
Industry X Wave FE
Yes
No
Yes
Industry X Wave FE
Yes
No
Yes
Occupation X Wave FE
No
Yes
Yes
Occupation X Wave FE
No
Yes
Yes
51284
51284
51284
42819
42820
42819
Observations
* p<0.10, ** p<0.05, *** p<0.010. Standard errors in parentheses are clustered at the individual level
Observations
* p<0.10, ** p<0.05, *** p<0.010. Standard errors in parentheses are clustered at the individual level
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
B. Technical note on macro-micro simulations The macro-micro simulations in this chapter use household data from labor force surveys for three countries – Bangladesh, Pakistan and India.
Generating earnings percentiles The Indian Periodic Labour Force Survey (PLFS 2017-18) reports employment information of respondents according to the usual status (principal + subsidiary) and current weekly status. We use the respondents’ current weekly status to match with earnings data. For regular wage/salaried persons in current weekly status information on earnings was collected for the preceding calendar month, for self-employed persons in current weekly status information on earnings was collected for the last 30 days and for casual labour information on earnings was collected for each day of the reference week. We compile all types of earnings into a single variable at a monthly frequency to generate rankings based on earning percentiles.
Dealing with missing observations We apply the same procedure to calculate monthly wage in Bangladesh labor force survey (2015-16) and Pakistan labor force survey (2017-18). However, we encounter the problem of non-response of earnings variable in both countries. We compare the subsample of wage-reporting workers and wage-missing workers to understand the sample bias. Wage-reporting workers are those employed workers who report their wage above zero. Wage-missing workers are those employed workers who do not report wage or report zero wage. In both Pakistan and Bangladesh, workers in the non-agricultural sector are more likely to report wage income. More than 60 percent of workers in the non-agricultural sector report wage in Pakistan, while only 12 percent workers in the agricultural sector do so. This implies that the sample of wage-reporting workers that we use to conduct earning rankings is biased towards non-agricultural workers.
Table 3.B.1: Description of Pakistan LFS (2017-18) data Wage-missing
Rural
Urban
Total
unskilled
3,168,768
3,375,757
6,544,525
skilled
2,043,859
4,650,453
6,694,312
Agri
unskilled
13,800,000
468,274
14,268,274
skilled
3,423,408
158,179
3,581,587
22,436,035
8,652,663
31,088,698
unskilled
4,070,372
5,908,776
9,979,148
skilled
2,832,001
7,900,943
10,732,944
unskilled
2,155,741
92,631
2,248,372
skilled
212,784
21,595
234,379
9,270,898
13,923,945
23,194,843
Non-agri
Total
Wage-reporting Non-agri
Agri
Total
101
102
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Table 3.B.2: Description of Bangladesh LFS (2015-16) data Wage-missing Non-agri
Agri
Rural
Urban
Total
unskilled
17,000,000
2,302,941
19,302,941
skilled
2,527,907
760,474
3,288,381
unskilled
17,700,000
450,563
18,150,563
skilled
1,712,277
67,177
1,779,454
38,940,184
3,581,155
42,521,339
unskilled
9,911,266
1,651,917
11,563,183
skilled
3,808,890
1,066,793
4,875,683
unskilled
3,881,749
61,195
3,942,944
skilled
110,581
5,592
116,173
17,712,486
2,785,497
20,497,983
Total
Wage-reporting Non-agri
Agri
Total
Calculating welfare rankings and food-shares Labour force surveys do not contain detailed data on food and non-food consumption. However, the Indian PLFS (2017-18) contains the household’s usual consumer expenditure (in Rs.) in a month calculated by asking respondents about (i) expenditure for household purposes, (ii) purchase value of any household durables and (iii) consumption from wages in -kind, home-grown stock and free collection. The sum of all three expenditure categories is reported. We use this total monthly consumption expenditure to generate household welfare rankings. In our simulations, we map the percentage change in household wages due to the employment shock to an equivalent percentage change in household welfare. In the case of Bangladesh and Pakistan, we use the labor force surveys to generate welfare rankings. However, one constraint is the lack of consumption variables in the labor force surveys. Alternatively, we use the wage variable from a subsample of wage-reporting workers to produce welfare rankings in both countries. We compare the distribution of wage per capita and consumption per capita. Mean and median wage are 4 times higher than mean and median consumption while wage per capita has similar mean and median value as consumption. The distribution of wage and wage per capita has a longer left and right tail than the distribution of consumption.
Table 3.B.3: Summary statistics (Pakistan and Bangladesh) observations
mean
median
wage per capita from LFS (2017-18)
140,198
4,220
3,106
consumption per capita from HH survey (2015)
157,636
5,052
3,931
PAK
BGD
wage per capita from LFS (2015-16)
58,820
2,619
1,875
consumption per capita from HH survey (2016)
185,115
3,800
3,040
Food shares are predicted by using a linear model on NSS (2011), which includes detailed consumption expenditure data, and applying it to PLFS (2017-18) data. We use total household consumption, household size, and the household head’s social group membership, employment status and education to predict food-shares of expenditure. The same procedure also applies to Bangladesh and Pakistan to calculate the food share variable.
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Work from home status The PLFS 2017 records individual’s occupations based on the National Classification of Occupations (NCO-04) at the 3-digit level. This is consistent with the International Standard Classification of Occupations (ISCO-88) at the Minor Group level (also 3-digt). Using the corresponding ISCO-88 codes for the occupations in the PLFS data, we map to the Standard Occupational Classification (SOC-00) system used by the US federal government. The SOC-00 and its 2010 version have been categorized by the Occupational Information Network (O*NET) with an indicator for being “suitable for teleworking or not”. Following a similar process of ranking individuals based on reported earnings, we summarize the share of workers who can potentially work from home along the whole earnings distribution and disaggregate by those sectors most likely to be affected by the lockdown. The same procedure applies to Pakistan and Bangladesh. Two caveats need to be mentioned: 1) Mapping from ISCO-88 to the O*NET classification is not one-to-one because the former is at the 3-digit level and the latter is at a much more granulated 7-digit level. We conduct some manual checks for consistency. 2) Classification of an occupation for “teleworking suitability” is likely to be very different for occupations based in the US vis-à-vis those in South Asia. We manually modify some occupations such as village-level administrators, teachers, etc. to approximate the context on the ground.
C. Can online gig work buffer domestic downturns? We use the country-year-quarter panel data between 2017Q3 and 2019Q3 to test the correlation between domestic economic shocks and domestic online gig jobs.
∆ labor shar e it = β 0 + α * ∆ log(GDP) it + γ i + ε it labor shar e itis the share of online gig workers in country iover total online gig workers in year quarter
t , log(GDP) itis log GDP in country i and year quarter t , and γ i is the country fixed effect. The underlying hypothesis that workers turn to online gig platforms to mitigate a negative domestic economic shock. Therefore, we expect the coefficient αto be negative.
The table below shows the estimation of the above equation. The change in the share of online gig worker is negatively correlated with domestic GDP growth, which is consistent with our hypothesis (Column 1). For the convenience of interpretation, we replace the dependent variable with the log change of online gig workers as a proxy for online gig worker growth in Column 3. A 1 percentage point drop in GDP growth is associated with 0.4 percentage points increase in online gig worker growth. On the other hand, the share of online gig posts and the growth of online gig posts is not significantly correlated with GDP growth (column 2 and column 4), suggesting the domestic online gig demand is not affected by domestic economic shocks.
Table 3.C.1: the share of online gig worker is negatively correlated with domestic GDP growth (1) D. worker share
(2) D. post share
(3) D. ln(worker)
(4) D. ln(post)
GDP growth
-0.371** (0.155)
-0.005 (0.159)
-0.406* (0.213)
-0.2 (0.202)
Country FE
Y
Y
Y
Y
Year-quarter FE
N
N
Y
Y
574
574
574
571
0.003
0
0.745
0.707
72
72
72
72
VARIABLES
Observations R-squared Number of countries Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
103
104
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
D. Informal employment in the region The tables below were obtained using the definition of informality adopted in this chapter and applied to data from Labor Force Surveys. The definition(s) of informality is based on the ILO practices which distinguish informal sector (a firm-based concept) and informal employment (a worker-based concept). A key feature to identify non-farm informal firms is that there is not a clear separation between the unit of production and its owner, often a household. Non-farm informal sector consists of unincorporated enterprises, household-run small firms. Informal employment is defined by considering both firm/worker status and relations of the worker with the firm, as shown in the table below: • Informal employment: all shaded cells in green • Employment in the informal sector (unincorporated firms with < 10 workers): cells 3 to 7 • Informal employment outside the informal sector: cells 1,2,8,9
Table 3.D.1: definition of informal employment Jobs by employment status
Production units by type
Own-Account Workers Informal
Formal
Employers Informal
Formal sector enterprise Informal sector enterprise
3
Households
8
4
Formal
Contributing Family Workers
Employees
Informal
Informal
1
2
5
6
Formal
7
9
These tables highlight several interesting facts, and more research will be needed in the future on the interaction of the formal and informal sector: • In some cases, formal firms are employing a higher share of informal workers than formal workers (Bangladesh in 2015-16, Pakistan 2001-02); • In Bangladesh, the share of informal workers has increased, but the share of informal firms (measured by the number of people these firms employ) has decreased. This is partly because of more informal workers in households, but also because the formal sector now employs more informal workers. • In Pakistan, the share of formal workers has increased much more than the share of formal firms (as measured by the number of people they employ). It seems that over time informal workers in the formal sector have received formal contracts.
Beaten or Broken? Informality and COVID-19 The impact of COVID-19 on the informal sector
Table 3.D.2: Informal Employment in the region Bangladesh, 2015-16 workers
Bangladesh, 2002-03
informal
formal
all
in formal firms
15.9
10.1
25.9
in informal firms
51.5
0.5
in households
22.1
all
89.4
workers
informal
formal
all
in formal firms
4.2
19.2
23.4
51.9
in informal firms
60.3
4.9
65.2
0
22.1
in households
11.2
0.2
11.4
10.6
100
all
75.7
24.3
100
informal
formal
all
Pakistan, 2017-18 workers
Pakistan, 2001-02
informal
formal
all
in formal firms
10.2
17.7
27.9
in formal firms
16.8
1.4
18.1
in informal firms
69.3
1.4
70.8
in informal firms
44.4
1.8
46.2
1.3
0.1
1.4
in households
35.7
0
35.7
80.8
19.2
100
all
96.8
3.2
100
formal
all
in households all
workers
Sri Lanka, 2015 workers
Sri Lanka, 2006
informal
formal
all
workers
informal
in formal firms
3.2
26.5
29.7
in formal firms
1.1
30.5
31.6
in informal firms
6.6
0.9
7.5
in informal firms
1.8
2.0
3.7
in households
60.8
2.1
62.9
in households
61.0
3.7
64.7
all
70.6
29.4
100
all
63.9
36.1
100.0
informal
formal
all
7.7
10.3
18
in informal firms
38.9
0.8
39.7
in households
42.1
0.1
42.3
all
88.8
11.2
100
India, 2017-18 workers in formal firms
105
106
Beaten or Broken? Informality and COVID-19
4
Chapter
South Asia country briefs
PHOTO BY: MANOEJ PAATEEL / SHUTTERSTOCK.COM
Beaten or Broken? Informality and COVID-19
107
108
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Afghanistan Afghanistan experienced moderate growth in 2019 as the agricultural sector recovered from the impacts of drought. However, the economy is estimated to have contracted sharply in the first half of 2020 due to economic disruptions associated with nation-wide lockdowns, border closures, and declining remittance inflows. Medium-term prospects are subject to high levels of uncertainty, related to the COVID-19 pandemic, peace talks and future international security and aid support. Given the shock to the economy poverty is expected to increase in 2020.
RECENT DEVELOPMENTS
Table 1
2019
Population, million
38.0
GDP, current US$ billion
19.3
GDP per capita, current US$
507.1
Poverty headcount ratio
54.5
a
School enrollment, primary (% gross)
72.5
Life expectancy at birth, years
64.5
a
b
Notes: (a) Afghanistan Living Condition Survey (ALCS) (2016-2017); (b) Most recent WDI value (2018). Sources: WDI, Macro Poverty Outlook, and official data.
Afghanistan recorded a current account surplus in 2019 and in the first half of 2020. In 2019, the
After a relatively strong performance in 2019
trade deficit narrowed to 30.4 percent of GDP
(3.9 percent growth), the economy contracted
(from 34.7 percent in 2018), as stronger domestic
in the first half of 2020. Over 2019, growth was
agricultural production drove a 7.1 percent drop
mainly driven by recovery in the agriculture
in imports. With large grant inflows, the current
sector (17.5 percent) following drought in the
account registered a small surplus of 0.6 percent
previous year. This good performance, together
of GDP in 2019. Over the first quarter of 2020 ex-
with a moderate expansion of industry (4.8 per-
ports grew by 11 percent (year-on-year) reflecting
cent), offset a decline in services (-1.4 percent).
the improved performance of air corridors, while
However, the economy contracted sharply in the
weak domestic demand led to a 14 percent de-
first half of 2020, largely reflecting the impact
cline in imports. In the second quarter of 2020,
of the COVID-19 crisis. Lockdowns hampered
both imports and exports fell precipitously giv-
domestic production and consumption (espe-
en border closures and disruptions to trade and
cially in urban centers). Border closures disrupt-
transportation, with greater absolute declines in
ed exports and supply chains, and remittances
imports driving an improvement in the trade and
declined significantly. While wheat production
current account balances.
grew significantly, driven by favorable weather conditions, this was insufficient to offset the
Fiscal imbalances that appeared in 2019 were aggra-
large negative impact of COVID-19 on other sec-
vated in 2020. Domestic revenues reached a record
tors of the economy.
high of 14.2 percent of GDP in 2019, largely driven by significant one-off revenues, including a transfer
Inflation was low in 2019 (averaging 2.3 percent)
of operating profits from the central bank (Afs 24
but it increased significantly in 2020. In March and
million, equivalent to 1.6 percent of GDP). However,
April 2020, panic buying and import disruptions
increased expenditures, mainly driven by salaries
drove a significant spike in food prices and led the
and wages, led to a deficit of 1.1 percent of GDP. With
government to adopt administrative measures to
the onset of the COVID-19 crisis, weak economic ac-
prevent price gouging and distribute emergency
tivity, disruptions to trade and compliance, revenue
wheat supplies. As a result, headline and food in-
performance deteriorated significantly and revenue
flation have since declined, standing at 5.3 percent
estimates for 2020 were revised downward by over
and 10 percent on average respectively, as of end
30 percent (from Afs 209 to 144 billion) in the budget
June.
mid-year review. Total domestic revenue collection
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Figure 1: Real GDP growth and contributions to real GDP growth
Figure 2: Actual poverty rates and real GDP per capita Poverty rate
Percent, percentage points 6
GDP per capita, USD
80
700
4
70
600
2
60
0
50
500
59 55
400
42 40
-2
30
-4
38
34
200
26
25
20
-6
10
-8 2016
2017
Agriculture
2018 Industry
2019
2020 f Services
2021 f
2022 f
Real GDP growth
300
42
36
100
2007
2009
2011
2013
National Poverty Rural Poverty
2015
2017
2019
0
Urban Poverty GDP per capita
Source: Macroeconomics Trade & Investment Global Practice.
Sources: WDI, NSIA, Official data, and ALCS (2016-2017)
at end-June reached Afs 74.7 billion, 20 percent low-
are projected to fall by 24 percent, imports (that
er than the initial budget target.
are significantly larger) are expected to decline by around 18 percent, reflecting border disruptions,
Poverty is believed to have worsened in 2019 surpass-
depressed domestic demand, and lower global oil
ing 54.5 percent (in household survey 2016-2017),
prices. With sustained foreign grants inflows, the
amidst continued violence and political uncertainty.
current account is expected to reach a surplus of
In the first half of 2020, with declining households
4.6 percent of GDP in 2020. However, it is project-
incomes due to economic hardship, higher food
ed to deteriorate over the medium term as grants
prices due to COVID-19, a significant fall in remit-
decline, to a deficit of 2.2 percent of GDP by 2022.
tances, and high returnee flows (mainly from Iran), poverty is estimated to have further increased.
With depressed revenues and higher expenditure needs, the fiscal deficit is projected to deepen to 3.4
OUTLOOK
percent of GDP in 2020. Over the medium-term, declining grants and overall growth weaknesses will constrain fiscal space, although the implementa-
Real GDP is expected to contract by 5.5 percent in
tion of VAT in 2022 should provide a partial offset.
2020, largely due to the impacts of the COVID-19 crisis. In following years, the pace of recovery is expect-
The economic contraction in 2020 is expected to
ed to be constrained in a context of continued insecu-
have significant adverse social impacts. World Bank
rity, uncertainties regarding the outcome of planned
micro-simulations suggest that the combination of
peace talks, and questions about the level and dura-
reduced incomes and higher prices could drive the
tion of international security and aid support.
poverty rate to as high as 72 percent, despite benefits from robust agricultural production to rural
External balances are expected to improve in 2020,
households. Over the medium term, the poverty
against a backdrop of declining trade activity. The
outlook hinges on the pace of economic recovery
trade deficit is projected to narrow to 26 percent of
and the continued provision of international aid
GDP down from 30.4 percent 2019. While exports
and humanitarian support.
109
110
Beaten or Broken? Informality and COVID-19 South Asia country briefs
RISKS AND CHALLENGES
limiting long-term economic damage, and establishing solid foundations for economic recovery. To
The main source of downside risk to the outlook
support households, the government should prior-
stems from possible further adverse COVID-19
itize: i) targeted social protection measures; and ii)
developments. Additional sources of risk include
ensuring the continued provision of basic services,
further political instability, a deterioration of se-
especially healthcare. To support the private sector,
curity conditions, uncertainties associated with the
priorities include: i) pursuing business regulatory
planned peace agreement with the Taliban, and
reforms to facilitate new investment; ii) expanding
precipitous reductions in aid flows. By contrast, on
access to credit; iii) ensuring the continued provision
the upside, a sustainable and credible political set-
of basic infrastructure; and iv) avoiding accumulat-
tlement with the Taliban could help boost growth,
ing arrears to private sector vendors. A clear com-
confidence and private investment.
mitment from international partners to continued grant support would help reduce uncertainty and
Given Afghanistan’s declining revenues and con-
improve confidence and investment, providing a
strained fiscal potential, public expenditures need
vital underpinning for Afghanistan’s recovery from
to be carefully directed to protecting the vulnerable,
the already-severe impacts of the COVID-19 crisis.
Table 2: Macro poverty outlook indicators (annual percent change unless indicated otherwise). 2017
2018
2019
2020 e
2021 f
2022 f
2.6
1.2
3.9
-5.5
2.5
3.3
Private Consumption
3.5
10.0
-2.0
-9.0
7.0
3.0
Government Consumption
3.3
-17.8
15.0
6.2
2.6
2.5
Gross Fixed Capital Investment
19.4
0.0
-15.3
-30.8
-3.6
4.1
Exports, Goods and Services
0.5
49.6
-6.3
-21.4
19.0
8.0
Imports, Goods and Services
9.8
13.1
-6.8
-16.3
13.0
3.0
2.3
1.2
4.4
-5.5
2.5
3.3
Agriculture
6.4
-4.4
17.5
5.0
3.5
4.0
Industry
9.2
11.1
4.8
-6.8
2.0
3.0
Services
-0.7
1.9
-1.4
-10.6
2.0
3.0
Inflation (Consumer Price Index)
5.0
0.6
2.3
5.0
4.2
4.5
Current Account Balance (% of GDP)
2.4
2.7
-0.1
4.6
-1.6
-2.2
Net Foreign Direct Investment (% of GDP)
0.2
0.5
0.0
0.0
0.0
0.1
Fiscal Balance (% of GDP)
-0.6
0.8
-1.1
-3.4
-2.2
-0.9
Debt (% of GDP)
6.6
5.8
6.6
8.6
9.6
9.6
Primary Balance (% of GDP)
-0.5
1.8
-0.4
-3.0
-2.2
-0.9
Real GDP growth, at constant market prices
Real GDP growth, at constant factor prices
Notes: e = estimate, f = forecast. Sources: World Bank, Poverty & Equity and Macroeconomics, Trade & Investment Global Practices.
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Bangladesh An extended national shutdown, a sharp decline in exports, and lower private investment reduced GDP growth in FY20 to an estimated 2.0 percent. While growth is expected to recover over the medium term, downside risks include a prolonged COVID-19 pandemic and financial sector fragility. The observed impact of COVID-19 on the labor market suggests that poverty is likely to increase significantly. Going forward, the implementation of the government’s COVID-19 response program will remain a paramount priority.
RECENT DEVELOPMENTS
Table 1
2019
Population, million
168.6
GDP, current US$ billion
302.7
GDP per capita, current US$
1795.5
International poverty rate ($1.9)
14.5
Lower middle-income poverty rate ($3.2)a
52.5
Upper middle-income poverty rate ($5.5)a
84.3
Gini indexa
32.4
School enrollment, primary (% gross)b
116.5
Life expectancy at birth, yearsb
72.3
a
Notes: (a) Most recent value (2016), 2011 PPPs. (b) Most recent WDI value (2018). Sources: WDI, Macro Poverty Outlook, and official data.
Real GDP growth fell to an estimated 2.0 percent
loans (NPLs) officially reached 9.2 percent of loans
in FY20 as COVID-19 and the global recession dis-
in June 2020 but are likely higher due to deviations
rupted economic activity in the second half of the
from international recognition, loss provisioning,
year. On the demand side, exports declined by 18.5
and capital calculation standards. The government’s
percent in FY20, as external demand for readymade
COVID-19 response relies heavily on commercial
garments (RMG) plummeted in the fourth quarter.
bank lending, supported by US$ 4.5 billion in BB re-
Private investment growth also slowed as financial
financing facilities and relaxed prudential require-
conditions deteriorated. Consumption, however,
ments. However, a new interest rate cap has disin-
was supported by a surge in remittance inflows,
centivized lending. Credit to the private sector grew
which partially offset lost labor income. On the
by 8.6 percent in FY20, while public sector credit
supply side, the industrial sector contracted, with
grew by 50.3 percent, driven by a rising fiscal deficit.
a severe decline in RMG manufacturing, while service sector growth decelerated due to disruptions in
The current account deficit (CAD) narrowed from
transport, retail, hotels, and restaurants. In the ag-
1.7 percent of GDP in FY19 to 1.5 percent in FY20,
ricultural sector, a resilient rice harvest was damp-
as a sharp decline in exports was offset by a surge
ened by losses in poultry, meat, and dairy due to
in remittance inflows. The lower CAD coupled
supply chain disruptions in the last quarter of FY20.
with increased external government borrowing resulted in a substantial balance of payments sur-
Inflation rose to 5.7 percent in FY20 from 5.5 per-
plus. The real effective exchange rate appreciated
cent in FY19, as food prices increased due to sup-
while the taka depreciated marginally in nominal
ply chain disruptions, while nonfood prices were
terms against the US dollar. Foreign exchange re-
dampened by weaker demand. In response to de-
serves remained adequate at US$ 36.4 billion or 7.2
teriorating economic conditions, Bangladesh Bank
months of imports at the end of FY20.
(BB) reduced the cash reserve ratio to 4 percent (from 5.5 percent) and the repo rate to 4.75 percent
The FY20 fiscal deficit is estimated at 8.2 percent
(from 6 percent). Monetary policy has been expan-
of GDP. Revenues were depressed by lower trade
sionary, with a 15.6 percent target for broad money
volumes and corporate profits while social protec-
growth announced in July 2020.
tion and healthcare expenditure rose. Public debt is estimated at 39.2 percent of GDP at the end of
The COVID-19 pandemic has exacerbated existing
FY20, and Bangladesh remains at a low risk of debt
financial sector vulnerabilities. Non-performing
distress.
111
112
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Figure 1: Real GDP growth and contributions to real GDP growth
Figure 2: Actual and projected poverty rates and real GDP per capita
Percent, percentage points
Poverty rate
15
100
Real GDP per capita (constant LCU) 80000
90
70000
80
10
60000 70 5
50000
60
40000
50 0
40
30000
30 20000
-5
20 10000
10 -10
2000
2003
2006
2009
2012
2015
Gov. cons.
Exports
GFCF
Private cons.
Imports
Statistical disc.
2018
2021
Inventories Real GDP growth
0
2005 2007 2009
2011
2013
2015
2017
2019
2021
0
International poverty rate
Lower middle-income pov. rate
Upper middle-income pov. rate
Real GDP per capita
Sources: Bangladesh Bureau of Statistics (BBS) and staff estimates.
Notes: See Table 2. Source: World Bank.
Job losses and temporary absences were widely re-
Inflation is projected to remain above target due
ported in representative surveys in poor areas of
to expansionary monetary and fiscal policies and
Dhaka, Chittagong and Cox’s Bazaar, with wide-
higher food prices. The CAD is expected to widen
spread uncertainty about employment and busi-
with a decline in exports, due to continued low ex-
ness prospects. Only 58 percent of active workers
ternal demand, and a decline in remittances, due
in poor areas of Dhaka and Chittagong thought
to the return of workers from overseas. The fiscal
they would be able to continue in their job or work
deficit is likely to rise as recurrent expenditure on
activity in the month following the survey. Women
social protection measures remains elevated in the
appear to have higher job and income losses than
near term, and capital expenditure increases in the
men, given higher employment in directly affected
post-COVID-19 recovery phase.
sectors such as RMG manufacturing. Poverty is expected to increase substantially in the short term, with the highest impact on daily and
OUTLOOK
self-employed workers in the non-agricultural sector and salaried workers in the manufacturing sector. GDP
Urban areas will continue to be disproportionately
growth is projected to decelerate to 1.6 percent in
affected, with an estimated 68 percent of directly af-
FY21 assuming that the impact of COVID-19 deep-
fected workers located in Dhaka and Chittagong.
Significant
uncertainty
notwithstanding,
ens. Private consumption growth is likely to remain subdued with depressed wage income and a decline in remittance inflows, while anemic private
RISKS AND CHALLENGES
investment is projected due to heightened uncertainty. Weaker demand and financing constraints
Downside risks to the outlook are substantial. Do-
may further reduce industrial production, while
mestic risks include additional waves of COVID-19
flooding in early FY21 may hamper agriculture
that may require renewed restrictions. In the govern-
production. However, GDP growth is projected to
ment’s COVID-19 response program, risks include
recover to 3.4 percent in FY22, supported by a re-
ineffective implementation of infection prevention
bound in export demand, remittance inflows, and
measures and limited operationalization of credit
public investment.
programs. In the context of COVID-19 disruptions,
Beaten or Broken? Informality and COVID-19 South Asia country briefs
fiscal risks may rise, particularly if tax reforms are
External risks also remain elevated. While external
delayed or infrastructure projects face cost overruns.
demand for RMG products is stabilizing, the recov-
Increased deficit financing from domestic banks may
ery is fragile. Lower oil prices may limit demand for
put upward pressure on interest rates and may fur-
Bangladesh’s overseas workforce in the Gulf region,
ther constrain credit to the private sector.
impairing remittance inflows. Also, continued appreciation of Bangladesh’s real exchange rate would
In the financial sector, challenges include devia-
adversely impact export demand and remittances.
tions from international regulatory and supervisory standards, the absence of a bank resolution
Going forward, the government’s COVID-19 re-
framework and weak governance in state-owned
sponse will remain a paramount priority, including
banks. The resolution of rising NPLs will require
testing, quarantining and treating patients and pro-
substantial policy dialogue to reduce credit risks,
viding economic relief to the poor and vulnerable.
limit moral hazard and manage fiscal risks. The in-
Other ongoing priorities include strengthening frag-
troduction of interest rate caps is an ongoing chal-
ile banks, diversifying exports, accelerating reforms
lenge, impairing the health of the banking sector.
in business regulation, and deepening fiscal reforms.
Table 2: Macro poverty outlook indicators (annual percent change unless indicated otherwise). 2016/17
2017/18
2018/19
2019/20 e
2020/21 f
2021/22 f
7.3
7.9
8.1
2.0
1.6
3.4
Private Consumption
7.4
11.0
3.9
2.1
2.0
2.7
Government Consumption
7.8
15.4
8.2
6.0
2.6
4.3
Gross Fixed Capital Investment
10.1
10.5
8.4
3.1
0.7
3.9
Exports, Goods and Services
-2.3
8.1
10.9
-18.5
-2.8
8.6
Imports, Goods and Services
2.9
27.0
-2.0
-11.6
-1.5
5.9
7.2
7.9
8.3
2.0
1.6
3.4
Agriculture
3.0
4.2
3.9
2.7
2.6
3.0
Industry
10.2
12.1
12.7
-0.1
-0.9
2.7
Services
6.7
6.4
6.7
3.2
2.9
3.9
Inflation (Consumer Price Index)
5.4
5.8
5.5
5.7
5.8
5.8
Current Account Balance (% of GDP)
-0.5
-3.5
-1.7
-1.5
-2.4
-1.9
Real GDP growth, at constant market prices
Real GDP growth, at constant factor prices
Net Foreign Direct Investment (% of GDP)
0.7
0.6
0.9
0.6
0.4
0.6
Fiscal Balance (% of GDP)
-3.4
-4.6
-5.4
-8.2
-8.8
-8.6
Debt (% of GDP)
31.0
31.9
33.1
39.2
45.9
51.4
Primary Balance (% of GDP)
-1.6
-2.8
-3.4
-6.0
-6.2
-5.6
International poverty rate ($1.9 in 2011 PPP)a,b
13.9
13.2
12.6
21.9
21.8
21.4
Lower middle-income poverty rate ($3.2 in 2011 PPP)a,b
51.4
50.2
48.9
59.1
59.0
58.5
Upper middle-income poverty rate ($5.5 in 2011 PPP)a,b
83.8
83.2
82.6
86.9
86.9
86.7
Notes: e = estimate, f = forecast. (a) Calculations based on SAR-POV harmonization, using 2010-HIES and 2016-HIES and fiscal year growth rates. Actual data: 2016. Nowcast: 2017-2019. (b) Projection using point-to-point elasticity (2010-2016) with pass-through = 1 based on GDP per capita in constant LCU. Sources: World Bank, Poverty & Equity and Macroeconomics, Trade & Investment Global Practices.
113
114
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Bhutan Real GDP growth is estimated at 1.5 percent in FY20 reflecting COVID-19 related disruptions, including in the tourism sector and industrial production. The fiscal balance deteriorated due to salary increases and higher government spending. Medium-term growth prospects are subdued and risks are tilted to the downside, particularly in the event of a large scale domestic outbreak of COVID-19. The poverty rate is expected to remain unchanged at 11 percent in 2020, reflecting lack of progress due to the pandemic.
Table 1
2019
Population, million
0.8
GDP, current US$ billion
2.5
GDP per capita, current US$
3288.6
International poverty rate ($1.9)a
1.5
Lower middle-income poverty rate ($3.2)a
12.2
Upper middle-income poverty rate ($5.5)a
38.9
Gini indexa
37.4
School enrollment, primary (% gross)b
100.1
Life expectancy at birth, yearsb
71.5
Notes: (a) Most recent value (2017), 2011 PPPs; (b) Most recent WDI value (2018). Sources: WDI, Macro Poverty Outlook, and official data.
RECENT DEVELOPMENTS The growth deceleration drove a reduction in Bhutan’s economy has been affected significantly by
trade activity and a narrowing of the current ac-
the COVID-19 crisis, with real GDP growth decel-
count deficit. Both exports and imports decreased,
erating to 1.5 percent in FY20 (from 3.8 percent in
in line with weak foreign and domestic demand,
FY19). Even though Bhutan managed to contain the
and disruptions to trade. Given that imports de-
number of domestic COVID-19 cases, the economy
clined significantly more than exports the current
was affected through two main channels: a decline
account deficit narrowed to 14 percent of GDP in
in the services sector as tourist arrivals dried-up,
FY20 (down from 22.5 in FY19).
and disruptions in industrial activities, reflecting reduced foreign demand, shortages in critical inputs
The fiscal balance, however, deteriorated, with
(including foreign labor), and temporary export
salary increases and additional COVID-19 related
restrictions. However, hydropower production and
expenditures driving the deficit to 3.1 percent of
exports increased in FY20 due to the on-streaming
GDP in FY20. Total spending is estimated to have
of the Mangdechhu project. On the demand side,
increased by 27.4 percent in FY20. Total revenues
consumption, public investment, and net exports
also increased, but to a lesser extent, and were
declined due to domestic containment measures,
driven by a one-off profit transfer from the com-
disruptions in public sector infrastructure projects,
missioning of the Mangdechhu hydropower plant.
and the lockdown in India -Bhutan’s largest trading
Non-hydro revenues declined with the discontinu-
partner-, which affected supply chains.
ation of excise duty refunds from India and lower tourism receipts. With a higher deficit, public debt
In spite of relatively low growth, headline infla-
is projected to have increased, albeit modestly (to
tion accelerated to 7.6 percent in July 2020, driv-
109.1 percent of GDP from 104.4 at end FY19).
en by food prices and reflective of similar trends in India. Asset quality in the financial sector de-
The poverty headcount, measured at $3.20 per day
teriorated further. The Non-Performing Loan
per person (in 2011 PPP terms), is estimated to have
(NPL) ratio rose to 17.7 percent in March 2020,
decreased slightly, from 11.5 percent in 2018 to 11
up from 10.9 percent in December 2019. While
percent in 2019. Services sector workers in urban
this partly reflects seasonal fluctuations in NPL
areas, including many that directly or indirectly
cycles, the sector has been adversely impacted
depend on tourism, experienced jobs and earn-
by weak underwriting standards and supervision,
ings losses since the COVID-19 outbreak. However,
and the effect of COVID-19 on businesses and
tourism is highly concentrated in just a few dis-
households.
tricts with very low poverty while almost all of the
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Figure 1: Real GDP growth and sectoral contributions to real GDP growth
Poverty rate 80
Percent, percentage points 8
Figure 2: Actual and projected poverty rates and real GDP per capita Real GDP per capita (LCU constant) 100000
7.4
7 6.2
6
90000
70
6.3
80000 60 70000
5 4
50
4.0
3.8
60000
3.8 40
3 2 1.5
50000 40000
30
30000
1 20 0
20000 10
-1 -2
2013/14
2014/15
Taxes/subsidies Agriculture
2015/16
2016/17
Services Manufacturing
2017/18
2018/19 f 2019/20 f
Non-manufacturing (incl. hydro) GDP growth growth
0
10000
2007
2009
2011
2013
2015
2017
2019
2021
0
International poverty rate
Upper middle-income pov. rate
Lower middle-income pov. rate
Real GDP per capita
Notes: Bhutan reports data on fiscal year (FY) basis. The fiscal year runs from July 1 though June 30. Source: Government of Bhutan and staff calculations.
Note: See Table 2. Source: World Bank.
poor live in rural areas, primarily engaged in sub-
investment and hydropower projects. However, the
sistence agriculture, and are thus relatively shield-
fiscal deficit is projected to increase to 6.7 percent
ed from the economic fallout of the pandemic.
of GDP in FY21 before gradually decreasing over the medium term. This trend reflects upward pres-
OUTLOOK
sure on expenditures to implement the COVID-19 recovery package and downward pressures on non-hydro revenues from weak economic activity.
Economic growth is projected to slow markedly, averaging 2.5 percent a year over the medium term,
The pandemic is expected to significantly slow
well below the pre-COVID-19 five-year average of
down the pace of poverty reduction. The poverty
5.5 percent. Tourism is expected to recover only
headcount rate (at $3.20 per day) is projected to
gradually, given that travel restrictions in Bhutan
remain unchanged at 11 percent in 2020. Unem-
will likely continue until at least early 2021, delay-
ployment will likely remain high, particularly in
ing a rebound in overall services sector growth. The
tourism related activities, though temporary cash
slowdown in India is expected to depress manufac-
support through the Druk Gyalpo’s Relief Kidu
turing and exporting industries, and the construc-
should help mitigate the impact of earnings losses.
tion sector is also likely to experience a protracted
Reduced demand for agricultural products could
slowdown due to a limited pipeline of public sec-
lower exports and hurt agribusinesses. Elevated
tor infrastructure projects. Finally, hydropower
food prices could disproportionately impact poor
production is expected to pick up marginally in
households since not all of their food requirements
FY21, providing limited support to industry sector
are met by own production. This could exacerbate
growth. Inflation will likely remain elevated in the
high levels of pre-existing malnutrition and should
short term because of localized food shortages re-
be closely monitored.
sulting in higher prices, but a moderation in prices is expected in the medium term.
RISKS AND CHALLENGES
Relative to previous years, the current account deficit is expected to remain lower in the medium term
Given the unpredictability of the pandemic’s
mostly on account of subdued imports for public
course, there is a high degree of uncertainty over
115
116
Beaten or Broken? Informality and COVID-19 South Asia country briefs
the ultimate growth and poverty trajectory. The
the decline in excise duties and grant financing in
most acute risk to the outlook is a large-scale do-
the medium term.
mestic outbreak of the virus leading to prolonged mobility restrictions. The materialization of finan-
The immediate opportunities are to prevent a
cial sector contingent liabilities is another poten-
large-scale community transmission of the vi-
tial risk, which could strain government finances.
rus and to ensure that the national response to
Other domestic risks include lower-than-expected
COVID-19 is well coordinated across sectors. There
hydropower production and delays in the imple-
is also an opportunity to accelerate the policy re-
mentation of revenue measures, particularly the
forms required to boost private sector job creation
goods and services tax, which are critical to offset
and economic diversification.
Table 2: Macro poverty outlook indicators (annual percent change unless indicated otherwise). 2016/17
2017/18
2018/19
2019/20 e
2020/21 f
2021/22 f
6.3
3.8
3.8
1.5
1.8
2.0
Private Consumption
0.0
10.1
6.0
1.5
2.0
3.5
Government Consumption
4.3
3.7
4.5
6.0
5.0
4.0
Gross Fixed Capital Investment
4.4
-3.6
7.2
-19.5
-10.4
-5.6
Exports, Goods and Services
0.4
5.5
-5.7
-3.9
-9.6
3.1
Imports, Goods and Services
-5.3
3.6
4.4
-20.2
-15.4
-1.6
6.0
3.3
4.2
1.5
1.8
2.0
Agriculture
3.6
3.7
3.8
2.5
3.1
3.5
Industry
4.7
-1.2
-1.6
3.7
2.0
1.5
Services
8.2
7.9
9.9
-0.6
1.3
2.2
4.3
3.7
2.8
3.2
5.0
2.8
Current Account Balance (% of GDP)
-23.9
-19.1
-22.5
-14.0
-13.8
-12.3
Fiscal Balance (% of GDP)
-4.8
-2.8
0.7
-3.1
-6.7
-5.3
Debt (% of GDP)
111.8
110.5
104.4
109.1
108.8
109.1
-3.5
-1.5
1.6
-2.1
-5.6
-3.7
1.5
1.4
1.3
1.3
1.3
1.2
12.2
11.5
11.0
11.0
10.8
10.2
38.9
37.8
37.1
37.0
36.9
36.0
Real GDP growth, at constant market prices
Real GDP growth, at constant factor prices
Inflation (Consumer Price Index)
Primary Balance (% of GDP) International poverty rate ($1.9 in 2011 PPP)
a,b
Lower middle-income poverty rate ($3.2 in 2011 PPP)
a,b
Upper middle-income poverty rate ($5.5 in 2011 PPP)
a,b
Notes: e = estimate, f = forecast. (a) Calculations based on SAR-POV harmonization, using 2017-BLSS and fiscal year growth rates. Actual data: 2017. Nowcast: 2018-2019. Forecast are from 2020 to 2022. (b) Projection using neutral distribution (2017) with pass-through = 0.7 based on GDP per capita in constant LCU. Sources: World Bank, Poverty & Equity and Macroeconomics, Trade & Investment Global Practices.
Beaten or Broken? Informality and COVID-19 South Asia country briefs
India India’s economy had been slowing prior to the COVID-19 pandemic. The spread of the virus and containment measures have severely disrupted supply and demand conditions. Monetary policy has been deployed aggressively and fiscal resources have been channeled to public health and social protection, but additional counter-cyclical measures will be needed, within a revised medium-term fiscal framework. Despite measures to shield vulnerable households and firms, the trajectory of poverty reduction has slowed, if not reversed.
KEY CONDITIONS AND CHALLENGES
Table 1
2019
Population, million
1371.3
GDP, current US$ billion
2862.3
GDP per capita, current US$
2087.3
International poverty rate ($1.9)
22.8
Lower middle-income poverty rate ($3.2)a
62.4
Gini indexa
35.4
School enrollment, primary (% gross)b
113.0
Life expectancy at birth, yearsb
69.4
a
Notes: (a) Most recent value (2011), 2011 PPPs.; (b) WDI for School enrollment (2017); Life expectancy (2018). Sources: WDI, Macro Poverty Outlook, and official data.
contain the health emergency. To mitigate its impact on the poorest, it was complemented by social protection measures; to ensure that businesses could maintain their operations, the Reserve Bank
India emerged from the Global Financial Crisis
of India (RBI) and the government also provided li-
(GFC) with stressed balance sheets of banks and
quidity and other regulatory support. Nonetheless,
corporates, depressed private investment, and
there was a massive contraction in output and poor
weaker exports growth. Efforts to deal decisive-
and vulnerable households experienced significant
ly with nonperforming assets in the banking sec-
social hardship – specifically urban migrants and
tor, strengthen the insolvency framework, and
workers in the informal economy.
improve the governance of public sector banks were only partially successful. Thus, in the period following the GFC, growth was driven mainly by
RECENT DEVELOPMENTS
private consumption. From FY09 to FY18 annual GDP growth averaged 6.7 percent (or 5.3 percent
In the first quarter of FY21 (India’s fiscal year is
per capita).
from April 1 to March 31) economic growth contracted by an unprecedented 23.9 percent (year-
After FY17, during which the economy grew at 8.3
on-year). On the demand side, private consump-
percent, growth decelerated in each subsequent
tion and investment contracted sharply. On the
year to 7.0, 6.1 and 4.2 percent. This was on account
supply side, industrial and services output fell by
of two mutually reinforcing dynamics: emerging
38 and 21 percent, respectively.
weaknesses in non-bank financial companies (a major source of credit growth, making up for risk
After reaching 4.8 percent in FY20, headline infla-
aversion from banks) and slowing private con-
tion averaged 6.6 percent, during April-July 2020,
sumption growth.
given supply-chain disruptions. The RBI cut the repo rate by a cumulative 115 bps between March
Thus, the impact of COVID-19 materialized against
and May, while maintaining significant excess li-
a backdrop of (i) enduring fragility in the financial
quidity in the market, and then paused further eas-
sector, (ii) slowing overall growth, and (iii) limited
ing in August.
fiscal buffers. The response of the government of India to the COVID-19 outbreak was swift and com-
During the first quarter of FY21, the current ac-
prehensive. A strict lockdown was implemented to
count turned to a surplus, as a large decline in
117
118
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Figure 1: Real GDP growth and contributions to real GDP growth
Figure 2: Actual and projected poverty rates and real GDP per capita Poverty rate
Percent, percentage points 15
Real GDP per capita
80
120000
70
10
100000
60 5
80000
50 0
40
60000
30
-5
40000 20
-10
20000 10
-15 2012
2013
2014
2015
Other Net exports Final consumption
2016
2017
2018
2019
2020e
Gross fixed capital formation Real GDP growth
0
2004
2011
International poverty rate Real GDP per capita
2018
0
Lower middle-income pov. rate
Sources: Central Statistics Office and staff calculations.
Notes: See Table 2. Source: World Bank.
imports more than offset a drop in exports. With
the last week of August than in the months leading
significant net foreign investment inflows, foreign
up to the lockdown. They also point to increased
reserves reached USD 534.5 billion at end-July,
vulnerability: 11 and 7 percent of urban and rural
equivalent to more than 13 months of FY20 im-
individuals, respectively, who recently identified
ports. Following a sharp depreciation in March,
themselves as “employed” performed zero hours
the rupee has gradually regained its value against
of work in the week prior to the survey. Data on
major currencies but remains slightly weaker than
the government’s rural workfare program show
at the start of the year.
that demand for casual work increased 66 percent y-o-y in August 2020. Between the last four
The growth slowdown in FY20 and the contraction
months of 2019 and May-August 2020, the pro-
in early FY21 have impaired revenue collection.
portion of people working in urban and rural areas
Thus, after increasing to 7.6 percent in FY20 (from
fell by 4.2 and 3.8 percentage points, respectively.
5.4 percent in FY19), the general government defi-
Overall, the pandemic has likely raised urban pov-
cit is believed to have increased further during the
erty, creating a set of “new poor” characterized by
first half of FY21.
non-farm employment and secondary or tertiary education.
Available household survey consumption data indicate that the poverty rate declined from 22.5 percent to values ranging from 8.1 to 11.3 percent,
OUTLOOK
between 2012 and 201711. More recent household survey data22 indicate significant disruptions to
Growth is expected to contract sharply in FY21
jobs due to COVID-19 that likely boosted the pov-
(by 9.6 percent in a baseline scenario), reflect-
erty rate, with 2020 rates back to levels overserved
ing the impact of the national lockdown and the
in 2016. These surveys suggest the labor force par-
income shock experienced by households and
ticipation rate was 3.2 percentage points lower in
firms. However, there is substantial uncertainty
1 The point estimate for 2017 is 10.4. The confidence interval reflects the degree of uncertainty associated with different statistical methods used to estimate poverty in the absence of recent household survey data. As documented in Box 1.3 of the Poverty and Shared Prosperity report (2020), there are other additional sources of uncertainty that are not reflected in this range of estimates. 2 From the Centre for Monitoring Indian Economy (CMIE).
Beaten or Broken? Informality and COVID-19 South Asia country briefs
related to (i) the course and duration of the pan-
The COVID-19 shock will lead to a long-lasting in-
demic, (ii) the speed at which households and
flexion in India’s fiscal trajectory. Assuming that the
firm behavior will adjust to the lifting of lock-
combined deficit of the states is contained within
downs, and (iii) a possible new round of counter-
4.5-5 percent of GDP, the general government fis-
cyclical fiscal policy. Thus, there is a wide con-
cal deficit is projected to rise to above 12 percent
fidence interval around the baseline projections.
in FY21 before improving gradually. Public debt is
Growth is expected to rebound to 5.4 percent in
expected to remain elevated, around 94 percent,
FY22, but mostly reflecting base effects, while po-
due to the gradual pace of recovery.
tential output is expected to remain depressed in the medium-term. Inflation is expected remain
Policy interventions have preserved the normal
around the RBI’s target range mid-point (4 per-
functioning of financial markets thus far. However,
cent) in the near-term.
the demand slowdown could lead to rising loan delinquencies and risk aversion. Recent RBI analysis
Weak activity, domestically and abroad, will de-
indicates the gross nonperforming loans to asset
press both imports and exports. Thus, the current
ratio of scheduled commercial banks may increase
account is expected to reach a surplus of 0.7 per-
to 12.5 percent by March 2021 (from 8.5 percent in
cent of GDP in FY21 and is projected to gradually
March 2020).
return to a deficit in later years.
Table 2: Macro poverty outlook indicators (annual percent change unless indicated otherwise). 2017/18
2018/19
2019/20
2020/21 e
2021/22 f
2022/23 f
7.0
6.1
4.2
-9.6
5.4
5.2
Private Consumption
7.0
7.2
5.3
-13.2
6.1
5.5
Government Consumption
11.8
10.1
11.8
10.5
5.5
5.9
Gross Fixed Capital Investment
7.2
9.8
-2.8
-16.2
7.8
6.7
Exports, Goods and Services
4.6
12.3
-3.6
-12.0
7.3
8.5
Imports, Goods and Services
17.4
8.6
-6.8
-20.0
12.3
12.0
6.6
6.0
3.9
-9.6
5.4
5.1
Agriculture
5.9
2.4
4.0
4.0
3.5
3.5
Industry
6.3
4.9
0.9
-20.0
5.5
5.0
Services
6.9
7.7
5.5
-7.4
6.0
5.7
Inflation (Consumer Price Index)
3.6
3.4
4.8
3.8
4.0
4.0
Current Account Balance (% of GDP)
-1.8
-2.1
-0.8
0.7
0.0
-0.5
1.1
1.1
1.5
1.1
1.3
1.5
Fiscal Balance (% of GDP)
-5.8
-5.4
-7.6
-12.4
-10.9
-8.9
Debt (% of GDP)
69.8
67.5
72.2
90.4
93.5
94.1
Primary Balance (% of GDP)
-1.1
-0.9
-2.8
-7.0
-4.4
-2.1
International poverty rate ($1.9 in 2011 PPP)a,b
10.4
9.2
8.3
11.1
10.0
9.0
Lower middle-income poverty rate ($3.2 in 2011 PPP)a,b
44.9
42.4
40.9
46.2
43.9
41.9
Real GDP growth, at constant market prices
Real GDP growth, at constant factor prices
Net Foreign Direct Investment (% of GDP)
Notes: e = estimate, f = forecast. (a) Calculations based on SAR-POV harmonization, using 2011-NSS-SCH1 and fiscal year growth rates. Actual data: 2011. Nowcast: 2012-2019. Forecast are from 2020 to 2022. (b) Projection using neutral distribution (2011) base on HFCE with pass-through .733 (rural) and .559 (urban) up to 2015, and .67 for 2016-17. GDP pc in constant LCU with pass-through = .67 for 2018-23. " Sources: World Bank, Poverty & Equity and Macroeconomics, Trade & Investment Global Practices.
119
120
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Maldives Maldives is expected to face its deepest recession in history. GDP is projected to contract by 19.5 percent in 2020 and to rebound by 9.5 percent in 2021, largely on account of base effects. The poverty rate is expected to increase to 5.6 percent in 2020, given widespread income losses. Fiscal vulnerabilities, already high prior to the pandemic, have been exacerbated by further external non-concessional borrowing. Postponing non-essential spending, especially on infrastructure, is critical to restoring fiscal and debt sustainability.
RECENT DEVELOPMENTS
Table 1
2019
Population, million
0.5
GDP, current US$ billion
5.7
GDP per capita, current US$
10710.0
Upper middle-income poverty rate ($5.5)
3.4
Gini indexa
31.3
School enrollment, primary (% gross)b
97.1
Life expectancy at birth, yearsb
78.6
a
Notes: (a) Most recent value (2016), 2011 PPPs. (b) WDI for School enrollment (2017); Life expectancy (2018). Source: WDI, World Bank, and official data. WDI, Macro Poverty Outlook, and official data.
of capital goods, diesel, and food and beverages consumed by tourists. However, merchandise exports
The COVID-19 pandemic has paralyzed the Mal-
fell even more, by 35.4 percent y-o-y, over the same
divian economy through its impact on tourism.
period. This was mostly due to lower re-exports of
Real GDP contracted by 5.9 percent year-on-year
jet fuel from muted air traffic, but also to lower fish
(y-o-y) in Q1 2020. The shock is expected to be
exports following weak demand from Europe and
larger in Q2 due to border closure and stringent
lower yield during the monsoon season.
mobility restrictions. Tourism inflows remained anemic even after borders reopened in mid-July.
As foreign exchange earnings from tourism plum-
Only 13,787 tourists visited between July 15 and
meted, usable reserves fell from USD 311.3 million
September 15, a 95 percent y-o-y decline; the av-
at end-January to USD 122.3 million as of end-Au-
erage number of daily international commercial
gust, equivalent to 0.5 months of 2019 goods im-
flights has declined to four (compared to 40 before
ports. Nonetheless, Maldives maintains a de facto
the pandemic) and half of all resorts remain shut.
stabilized exchange rate arrangement. To help pre-
Construction, the other main driver of growth,
serve exchange rate stability, the Maldives Mone-
also slumped due to logistical difficulties and re-
tary Authority activated a USD 150 million foreign
patriations of foreign workers following COVID-19
currency swap with the Reserve Bank of India.
outbreaks. Fiscal imbalances have widened significantly. AlWith overall depressed activity, price controls on
though the state collected only USD 471 million
staple food, and additional subsidies on utility bills,
in revenues and grants from January to July (half
prices fell by an average of 4.0 percent y-o-y in Q2
the amount in the corresponding period of 2019),
2020. The deflation was more pronounced in MalĂŠ
there was no commensurate adjustment in spend-
than in the atolls. Credit to the private sector grew
ing. While there was some degree of fiscal consoli-
by 6.9 percent y-o-y in Q2 thanks to the deploy-
dation on the recurrent side, capital spending grew
ment of relief loans to businesses. Non-performing
by 16.7 percent y-o-y in the first half of the year,
loans remained stable at 9.3 percent of total gross
mainly due to land reclamation and harbor recon-
loans as of July, reflecting the loan moratoria an-
struction projects. Total spending amounted to
nounced in March.
USD 948 million over January to July, only 1.2 percent less than over the same period in 2019. Total
Weak activity also led the goods trade deficit to nar-
public and publicly guaranteed (PPG) debt rose to
row. Merchandise imports fell by 29.8 percent y-o-y
USD 4.8 billion as of end-June 2020, a significant
from January to July 2020, driven by lower imports
increase from USD 4.4 billion as of end-2019.
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Figure 1: Public and publicly guaranteed debt
Figure 2: Actual and projected poverty rates and real GDP per capita
Percent of GDP 120
Poverty rate (%) 6
Growth of real GDP per capita 15 10
100
5 5
80
4 0
60
-5
3
-10
40
2 -15
20
0
1
Jun'17
Dec'17
Jun'18 Dec'18 External debt Guaranteed debt
Jun'19 Dec'19 Domestic debt Total PPG debt
Jun'20
0
-20
2016
2017
2018
2019
2020
2021
2022
-25
Poverty Rate, $5.5/day (PPP)(lhs) GDP Per Capita Growth, constant LCU (rhs)
Sources: Ministry of Finance data. 2019 and 2020 denominators are World Bank estimates.
Sources: 2016 data is World Bank estimate from HIES 2016, all other years are World Bank projections.
With the shutdown of tourism due to COVID-19,
22.5 percent of GDP as expenditures have not suffi-
it is estimated that about 22,000 Maldivians have
ciently adjusted to lower revenue levels, which are
suffered job and/or income losses in the sector, not
projected to halve. As a result, PPG debt is expected
including seasonal workers, third-party providers
to increase from 77.7 percent of GDP in 2019 to 120
of ancillary services and guesthouse employees. To
percent of GDP in 2020.
mitigate the impact, monthly income support was provided to about 9,000 Maldivians in Q2, with
COVID-19 will likely cancel the gains in pover-
an extension of the scheme in Q3. Based on data
ty reduction from the last five years. The poverty
collected in 2016, poverty was estimated at 2.5 per-
rate is projected to increase to 5.6 percent in 2020
cent in 2019 (using the international poverty line of
(measured at USD 5.50 a day in PPP terms) and to
USD 5.50 per day in PPP terms).
decline very slowly thereafter. Even in the baseline scenario, the poverty rate would still be higher in
OUTLOOK In a baseline scenario that assumes borders remain
2021 than in 2016.
RISKS AND CHALLENGES
open and tourists gradually return, GDP is projected to shrink by 19.5 percent in 2020. Thereafter,
Should a ‘second wave’ materialize and prevent
it is expected to rebound to 9.5 percent growth in
tourists from visiting in Q4 2020, the recession
2021, largely on account of base effects and the ex-
would be even more severe. While the unique “one
pected resumption of tourism once a COVID-19
island, one resort” concept facilitates socially-dis-
vaccine is commercially available, tentatively by
tanced vacations, difficulties in resuming commer-
mid-year. However, real GDP is forecast to remain
cial flights and recent increases in domestic trans-
below 2019 levels until 2023.
mission pose challenges to attracting more visitors. Although medium- and long-term tourism pros-
Lower remittance outflows and some import com-
pects remain strong, visitor arrivals are not pro-
pression should help to narrow the current account
jected to return to pre-pandemic levels until 2023.
deficit to 19.5 percent of GDP in 2020. However,
Meanwhile, the potential closure of tourist estab-
the fiscal deficit is projected to more than triple to
lishments and other small businesses could result
121
122
Beaten or Broken? Informality and COVID-19 South Asia country briefs
in permanent supply-side losses, hurting long-
The COVID-19 shock has shed renewed light on the
term growth.
importance of strengthening the Maldives’ resilience to external shocks. Although there are plans
Against this backdrop of slower growth and lower
to develop agriculture and fishing to diversify the
revenues, addressing core spending needs will be a
economy, the scarcity of arable land is a binding
challenge. Greater fiscal prudence would help ad-
constraint. Focusing on higher value-added finan-
dress fiscal and debt sustainability risks. In particu-
cial and business services could create good jobs,
lar, large public infrastructure investments that are
but the growth of these sectors is currently con-
not urgently needed in a context of weak aggregate
strained by a shortage of local skills. Investing in
demand could be postponed.
human capital, including by retraining and upskilling workers, can help Maldives build back better.
Table 2: Macro poverty outlook indicators (annual percent change unless indicated otherwise). 2017
2018
2019
2020 e
2021 f
2022 f
Real GDP growth, at constant market prices
6.8
6.9
5.9
-19.5
9.5
12.5
Real GDP growth, at constant factor prices
6.7
6.9
5.2
-19.5
9.5
12.5
Agriculture
8.3
4.8
3.2
1.8
3.5
4.5
Industry
10.7
10.5
4.5
-2.7
4.8
5.5
Services
6.0
6.5
5.5
-23.2
10.9
14.3
2.8
-0.1
0.2
0.1
0.5
1.0
-21.5
-28.0
-25.7
-19.5
-17.6
-15.6
Net Foreign Direct Investment (% of GDP)
9.7
10.8
15.7
5.0
7.5
10.6
Fiscal Balance (% of GDP)
-3.2
-4.7
-6.4
-22.5
-19.6
-16.6
Debt (% of GDP)
64.6
73.1
77.7
120.0
127.7
128.8
-1.6
-2.9
-4.6
-20.8
-17.7
-15.0
3.2
2.9
2.5
5.6
4.1
2.7
Inflation (Consumer Price Index) Current Account Balance (% of GDP)
Primary Balance (% of GDP) Upper middle-income poverty rate ($5.5 in 2011 PPP)
a,b
Notes: e = estimate, f = forecast. (a) Calculations based on SAR-POV harmonization, using 2016-HIES.Actual data: 2016. Nowcast: 2017-2019. Forecast are from 2020 to 2022. (b) Projection using neutral distribution (2016) with pass-through = 0.87 based on GDP per capita in constant LCU. Sources: World Bank, Poverty & Equity and Macroeconomics, Trade & Investment Global Practices.
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Nepal Nepal’s economy came to a standstill in FY20 with negligible growth of 0.2 percent, a deceleration in the service sector and a contraction in industrial activity due to COVID-19. Trade disruptions resulted in a collapse in imports and narrowed the current account deficit, but the fiscal balance deteriorated. In the medium term, the economy is expected to recover only gradually as the pandemic-induced disruptions recede. Poverty is expected to increase in the short term.
Table 1
2019
Population, million
29.9
GDP, current US$ billion
30.6
GDP per capita, current US$
1023.4
International poverty rate ($1.9)
15.0
Lower middle-income poverty rate ($3.2)a
50.9
Upper middle-income poverty rate ($5.5)a
83.0
Gini indexa
32.8
School enrollment, primary (% gross)b
142.1
Life expectancy at birth, yearsb
70.5
a
Notes: (a) Most recent value (2010), 2011 PPPs.; (b) WDI for School enrollment (2019); Life expectancy (2018). Source: WDI, Macro Poverty Outlook, and official data.
KEY CONDITIONS AND CHALLENGES
impacted production across all sectors of the economy and exacerbated pressure on livelihoods and an already stressed domestic labor market. In re-
In the three years prior to the pandemic, the econ-
sponse, the government adopted health measures,
omy grew at an average rate of 7.3 percent, against a
and expanded safety net programs and conces-
backdrop of political stability and policy emphasis
sional loan facilities.
on investment, productivity, and effective public institutions. Still Nepal’s economy remains hampered by structural challenges, including heavy
RECENT DEVELOPMENTS
reliance on remittance-fueled consumption, and higher concentration of employment in agricul-
Economic growth slowed to 0.2 percent in FY20,
ture. In addition, an ambitious decentralization re-
mainly due to a national lockdown in response to
form initiated in FY18 has resulted in higher fiscal
the pandemic. Service sector growth deteriorated
deficits and constrained service delivery, reflecting
to an 18-year low of 0.7 percent, as tourism arriv-
the weak capacity of the new local administrations.
als stopped and domestic transport and wholesale
Addressing these challenges will require improving
and retail trade were disrupted. Industrial growth
the quality of policymaking in the federal context,
contracted, and capacity utilization fell from 80 to
investing in human capital and skills, strengthen-
46 percent because of shortages in production in-
ing the business climate, and improving budget ex-
puts and labor. Agricultural growth also decelerated
ecution toward closing critical infrastructure gaps.
sharply as market access and labor mobility became constrained. On the expenditure side, a contraction
The COVID-19 crisis has derailed the growth mo-
in private consumption and investment was mar-
mentum and exacerbated structural vulnerabil-
ginally offset by positive contributions from gov-
ities. Travel restrictions have halted tourism and
ernment expenditures (on wages and COVID-19 re-
labor outmigration with ripple effects on domestic
lated expenditures on health and social assistance)
employment, remittances, and private consump-
and net exports (driven by lower imports).
tion. Trade with India, Nepal’s main trading partner, remains impaired as both countries struggle to
Average inflation reached 6.2 percent in FY20. Food
contain the pandemic. The nationwide lockdown
price inflation spiked in early FY20 following an
123
124
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Figure 1: Real GDP growth and contributions to real GDP growth Percent, percentage points 10
Figure 2: GDP growth is expected to decelerate significantly compared to projections before COVID-19 outbreak Percent 8
8 6 6 4 4 2
2
0
0
-2 FY12
FY13
FY14
FY15
FY16
FY17
Agriculture Service
FY18
FY19
FY20e
Industry GDP growth
-2 FY19
FY20e
Pre-Covid-19 Covid-19 Downside
FY21f
FY22f
Covid-19 Baseline Covid-19 Upside
Source: Central Bureau of Statistics and staff calculations.
Sources: Central Bureau of Statistics and staff calculations.
export ban on onions by India and remained ele-
external concessional loans. Foreign exchange re-
vated because of localized food shortages resulting
serves increased to $11.6 billion, equivalent to 12.7
from transport disruptions. Prudential indicators of
months of imports.
banking and financial institutions remained within regulatory targets: non-performing loans stood at
The fiscal deficit increased in FY20 as trade restric-
1.9 percent, capital adequacy at 14.2 percent, and
tions caused a 5 percent contraction in tax revenue.
liquidity at 27.9 percent. As a part of COVID-19 re-
The fall in tax revenue was partly offset by low bud-
lief measures, the central bank relaxed regulatory
get execution, with total expenditures remaining
requirements for banking and financial institutions,
close to FY19 levels in nominal terms. As a result, the
reduced the targeted interest rate, and increased the
fiscal deficit widened only marginally, to 3.2 percent
size of the refinancing facility. Despite these mea-
of GDP in FY20. Public debt increased to 38.3 per-
sures, deposit growth exceeded credit growth for
cent of GDP in FY20, compared to 30.1 percent in
the first time in five years: the credit to core capital
FY19, reflecting the adverse growth dynamics.
and deposits ratio fell to 69.6 percent, well below the 80 percent regulatory limit.
COVID-19 related disruptions in livelihoods and the contraction in households’ consumption are
Trade disruptions led to a 19.7 percent year-on-
expected to have affected the poor, vulnerable, and
year drop in imports in FY20, significantly nar-
households engaged in informal sector activities
rowing the current account deficit to 0.9 percent of
disproportionately. Therefore, poverty is likely to
GDP. The sharp drop in imports outweighed both
increase in 2020. However, outdated poverty data
the contraction in exports - reflecting lower exter-
(from 2010/11) makes projection-based estimates
nal demand for Nepali goods and a shutdown of
of poverty imprecise. The 2017/18 national labor
tourist arrivals – and a 3.4 percent decline in remit-
market data also predates COVID-19. But, new data
tance inflows due to lower outmigration and weak
from an ongoing regional survey on the labor im-
economic activity in migrant receiving countries.
pacts of COVID-19 will help track the evolution of
The current account deficit was mostly financed by
job losses and recoveries in the last quarter of 2020.
Beaten or Broken? Informality and COVID-19 South Asia country briefs
OUTLOOK
expected into FY22, assuming a gradual retreat of the pandemic.
Growth is expected to remain subdued in FY21 and FY22. Under a baseline scenario, GDP is pro-
Inflation is expected to accelerate slightly to 6.5 per-
jected to expand by only 0.6 percent in FY21,
cent in FY21, reflecting a gradual recovery in glob-
as periodic and localized lockdowns continue.
al oil prices and higher domestic food prices, due
Disruptions to tourism are expected to persist
to slow agricultural growth and periodic disrup-
well into FY21. Although key hydropower proj-
tions to supply chains. The current account deficit,
ects, such as the Upper Tamakoshi Project, are
meanwhile, is projected to remain close to its FY20
expected to support industrial growth, demand
level, with a gradual rebound in imports offset by
and trade are likely to remain subdued, further
increased remittance inflows and merchandise ex-
depressing industrial sector growth in FY21. The
ports. However, remittances are likely to remain
agricultural sector is expected to pick up, given
below pre-COVID-19 levels until the end of FY21.
favorable monsoons, but could be impacted by a
Thereafter, the current account deficit is expected
shortage of fertilizers and labor. On the demand
to gradually widen to 3.1 percent of GDP in FY22, as
side, private consumption is expected to support
domestic demand for imports rebounds. The fiscal
growth, but investment and exports will likely
deficit is expected to widen to 6.4 percent of GDP
remain subdued. In a downside scenario, should
in FY21 due to additional recurrent spending on
COVID-19 persist, continued disruptions and
COVID-19 relief and recovery measures, stable cap-
weak subnational capacity to implement relief
ital spending, and subdued tax revenue collection.
spending could weaken growth to 0.1 percent in FY21, likely increasing poverty. But if an effective
Risks to poverty will increase with declining remit-
vaccine becomes available, growth could recov-
tances and lagged effects from extended lockdowns
er to 2 percent in FY21. A protracted recovery is
on informal sectors.
Table 2: Macro poverty outlook indicators (annual percent change unless indicated otherwise). 2017
2018
2019
2020 e
2021 f
2022 f
8.2
6.7
7.0
0.2
0.6
2.5
Private Consumption
2.6
3.3
5.5
-3.1
-0.9
2.3
Government Consumption
10.5
13.5
7.8
38.7
-2.2
-6.7
Gross Fixed Capital Investment
44.3
18.1
5.0
-46.1
-1.2
40.1
Exports, Goods and Services
11.3
6.2
4.7
-16.6
-12.2
6.1
Imports, Goods and Services
27.2
16.5
7.8
-21.7
-5.6
12.7
7.7
6.1
6.6
0.2
0.6
2.5
Agriculture
5.2
2.8
5.1
2.3
2.1
2.9
Industry
12.4
9.6
7.7
-5.9
-4.1
3.2
Services
8.1
7.2
7.3
0.7
1.0
2.1
Inflation (Consumer Price Index)
4.5
4.2
4.5
6.2
6.5
6.3
Current Account Balance (% of GDP)
-0.4
-8.1
-7.7
-0.9
-1.0
-3.1
Fiscal Balance (% of GDP)
-3.1
-6.6
-2.8
-3.2
-6.4
-5.8
Debt (% of GDP)
26.1
30.1
30.1
38.3
43.6
46.7
Primary Balance (% of GDP)
-2.7
-6.1
-2.2
-2.5
-5.6
-4.9
Real GDP growth, at constant market prices
Real GDP growth, at constant factor prices
Notes: e = estimate, f = forecast. Source: World Bank, Poverty & Equity and Macroeconomics, Trade & Investment Global Practices.
125
126
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Pakistan Pakistan’s economy has been severely affected by measures taken to contain the pandemic. Economic activity contracted and poverty is expected to have risen in FY20, as monetary and fiscal policy tightening, earlier in the year, was followed by lockdowns. Growth is expected to gradually recover but remain muted, given heightened uncertainty and the resumption of demand compression measures. A possible resurgence of the infection, widespread crop damage due to locusts and heavy monsoon rains pose major risks to the outlook.
Table 1
2019
Population, million
204.7
GDP, current US$ billion
278.2
GDP per capita, current US$
1359.5
International poverty rate ($1.9)
4.0
Lower middle-income poverty rate ($3.2)a
35.1
Upper middle-income poverty rate ($5.5)a
75.6
Gini indexa
33.5
School enrollment, primary (% gross)b
94.3
Life expectancy at birth, yearsb
67.1
a
Notes: (a) Most recent value (2015), 2011 PPPs; (b) Most recent WDI value (2018). Sources: WDI, Macro Poverty Outlook, and official data.
RECENT DEVELOPMENTS also shrank given weaknesses in global trade and Real GDP growth (at factor cost) is estimated to
domestic demand. In contrast, government con-
have declined from 1.9 percent in FY19 to -1.5
sumption growth rose, reflecting the rollout of the
percent in FY20, the first contraction in decades,
fiscal stimulus package to cushion the effects of the
reflecting the effects of COVID-19 containment
pandemic.
measures that followed monetary and fiscal tightening prior to the outbreak. To curtail the spread
Despite weak activity, consumer price inflation
of the infection, a partial lockdown - that included
rose from an average of 6.8 percent in FY19 to an
restrictions on air travel, inner-city public trans-
average of 10.7 percent in FY20, due to surging
port, religious/social gatherings and the closure
food inflation, hikes in administered energy prices,
of all non-essential businesses and schools - was
and a weaker rupee, which depreciated 13.8 per-
imposed in March and gradually eased from May
cent against the U.S. dollar in FY20. With elevated
2020 onwards. This disrupted domestic supply and
inflationary pressures, the policy rate was held at
demand, as businesses were unable to operate and
13.25 percent from July 2019 to February 2020 but
consumers curbed spending, which specifically af-
was subsequently lowered to 7.0 percent over the
fected services and industry. The services sector is
remainder of FY20, to support dwindling activity
estimated to have contracted, by over 1 percent,
and as inflationary expectations fell amid the pan-
while industrial production is expected to have
demic. The central bank also implemented multi-
declined even more, due to the high policy rates
ple measures to provide liquidity support to firms.
prior to the pandemic and plunging domestic and
At end-FY20, the banking system remained well
global demand thereafter. The agriculture sector,
capitalized, however upticks in non-performing
partially insulated from the effects of containment
loans were beginning to erode capital buffers.
measures, is estimated to have expanded modestly over the year.
The current account deficit shrank from 4.8 percent of GDP in FY19 to 1.1 percent of GDP in FY20, the
On the demand side, private consumption is esti-
narrowest since FY15, driven mainly by import val-
mated to have contracted in FY20, as households
ues falling 19.3 percent (Figure 1). Total export val-
reduced consumption amid the lockdown and
ues also contracted 7.5 percent due to weak global
dimmer employment prospects. Similarly, with
demand. Despite the global downturn, workers’ re-
heightened uncertainty, disrupted supply chains
mittances increased relative to FY19, underpinning
and a global slowdown, investment is estimat-
a wider income account surplus. Meanwhile, high-
ed to have fallen drastically. Exports and imports
er net foreign direct investment, and multilateral
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Figure 1: Twin deficits and real GDP growth Percent
Figure 2: Public debt
Percent of GDP
6
10
Percent of GDP 100
8 4
6
80
4
2
2 0
60
0 -2
-2
40
-4 -6
-4
20
-8 -6
-10 FY17
FY18
FY19
Real GDP growth (lhs) Fiscal balance (rhs)
FY20-e
FY21-f
FY22-f
Current account balance (rhs)
0
FY17
FY18
FY19
Guaranteed debt External general government debt
FY20 e
FY21 f
FY22 f
Domestic general government debt Public debt
Notes: Pakistan reports data on fiscal year (FY) basis. The fiscal year runs from July 1 through June 30. Sources: Ministry of Finance and staff estimates.
Sources: Ministry of Finance, State Bank of Pakistan and staff estimates.
and bilateral disbursements, more than offset a de-
human capital challenges. The closure of educa-
cline in portfolio flows, leading to a larger financial
tion institutes has impacted more than 50 million
account surplus. The balance of payments conse-
students, while access to essential healthcare like
quently swung to a surplus of 2.0 percent of GDP
prenatal/postnatal services and immunization has
in FY20, and official foreign reserves increased to
been disrupted. All these challenges disproportion-
US$13.7 billion at end-June 2020, equivalent to 3.2
ately affect poor and vulnerable groups, including
months of imports.
women and girls.
In FY20, the fiscal deficit narrowed to 8.1 percent of GDP from 9.0 percent in FY19. Total revenues
OUTLOOK
rose to 15.3 percent of GDP due to higher non-tax revenue, as the central bank and the telecommuni-
While domestic economic activity is expected
cation authority repatriated large profits. Despite
to recover, as lockdown measures are lifted, with
reforms, tax revenues slipped to 11.6 percent of
a gradual decline in active COVID-19 cases, Pa-
GDP, with lower economic activity and larger tax
kistan’s near-term economic prospects are sub-
expenditures. Expenditures rose mainly due to a
dued. Significant uncertainty over the evolution
fiscal stimulus package valued at around 2.9 per-
of the pandemic and availability of a vaccine, de-
cent of GDP, while public debt, including guar-
mand compression measures to curb imbalanc-
anteed debt, increased to 93.0 percent of GDP by
es, along with unfavorable external conditions,
end-FY20 (Figure 2).
all weigh on the outlook. Economic growth is projected to remain below potential, averaging
The economic contraction is likely to have a sig-
1.3 percent for FY21-22. This baseline projection,
nificant impact on poverty. Lockdown measures
which is highly uncertain, is predicated on the
have severely affected non-farm sectors that pro-
absence of significant infection flare ups or sub-
vide livelihoods to the most vulnerable segments
sequent waves that would require further wide-
of the population, particularly in urban areas. With
spread lockdowns.
government estimates of pandemic job losses at approximately 14 million, poverty is expected
The current account deficit is expected to widen
to increase for the first time in two decades. The
to an average of 1.5 percent of GDP over FY21-22,
pandemic is also expected to exacerbate Pakistan’s
with imports and exports gradually picking up as
127
128
Beaten or Broken? Informality and COVID-19 South Asia country briefs
domestic demand and global conditions improve.
RISKS AND CHALLENGES
The fiscal deficit is projected to narrow to 7.4 percent in FY22, with the resumption of fiscal consoli-
There are considerable downside risks to the out-
dation and stronger revenues driven by recovering
look with the most significant being a possible re-
economic activity and critical structural reforms.
surgence of the infection, triggering a new wave
Expenditures will remain substantial due to size-
of global and/or domestic lockdowns and further
able interest payments, a rising salary and pension
delaying the implementation of critical structural
bill, and absorption of energy SOE guaranteed
reforms. Locust attacks and heavy monsoon rains
debt by the government.
could lead to widespread crop damage, food insecurity and inflationary pressures, and livelihoods
Given anemic growth projections in the near term,
for households dependent primarily on agriculture
poverty is expected to worsen. Vulnerable house-
could also be negatively impacted. Finally, external
holds rely heavily on jobs in the services sector,
financing risks could be compounded by difficul-
and the projected weak services growth is likely to
ties in rolling-over bilateral debt from non-tradi-
be insufficient to reverse the higher poverty rates
tional donors and tighter international financing
precipitated by the pandemic.
conditions.
Table 2: Macro poverty outlook indicators (annual percent change unless indicated otherwise). 2016/17
2017/18
2018/19
2019/20 e
2020/21 f
2021/22 f
5.6
5.8
1.0
-1.5
0.5
2.0
Private Consumption
8.5
6.2
2.9
-1.1
1.5
2.4
Government Consumption
5.3
8.6
0.8
5.6
0.1
1.9
Gross Fixed Capital Investment
10.3
11.2
-12.8
-17.9
-6.9
1.2
Exports, Goods and Services
-0.6
12.7
14.5
-8.6
-0.7
3.8
Imports, Goods and Services
21.2
17.6
4.3
-10.5
-0.7
4.3
5.2
5.5
1.9
-1.5
0.5
2.0
Agriculture
2.2
4.0
0.6
1.2
1.0
2.3
Industry
4.6
4.6
-2.3
-5.0
-2.4
1.7
Services
6.5
6.3
3.8
-1.3
1.2
2.1
Inflation (Consumer Price Index)
4.8
4.7
6.8
10.7
9.0
7.0
Current Account Balance (% of GDP)
-4.0
-6.1
-4.8
-1.1
-1.5
-1.7
Net Foreign Direct Investment (% of GDP)
0.8
0.9
0.5
1.0
0.7
0.8
Fiscal Balance (% of GDP)
-5.8
-6.4
-9.0
-8.1
-8.2
-7.4
Debt (% of GDP)
70.0
75.2
89.8
93.0
93.5
93.8
Primary Balance (% of GDP)
-1.5
-2.1
-3.5
-1.8
-1.8
-1.3
Real GDP growth, at constant market prices
Real GDP growth, at constant factor prices
Notes: e = estimate, f = forecast. Source: World Bank, Poverty & Equity and Macroeconomics, Trade & Investment Global Practices.
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Sri lanka COVID-19-related disruptions will lead to a sharp contraction in Sri Lanka’s economy in 2020. A significant increase in poverty is expected due to widespread jobs and earnings losses. With already narrow fiscal buffers before the pandemic, additional spending and reduced revenues due to COVID-19 will put additional pressure on fiscal sustainability. Over the medium term, growth is expected to recover slowly. However, macroeconomic vulnerabilities will remain high, with depleted fiscal buffers, high indebtedness and large refinancing needs.
Table 1
2019
Population, million
21.8
GDP, current US$ billion
84.0
GDP per capita, current US$
3861.0
International poverty rate ($1.9)
0.9
Lower middle-income poverty rate ($3.2)a
10.8
Upper middle-income poverty rate ($5.5)a
41.7
Gini indexa
39.8
School enrollment, primary (% gross)b
100.2
Life expectancy at birth, yearsb
76.8
a
Notes: (a) Most recent value (2016), 2011 PPPs.; (b) Most recent WDI value (2018). Source: WDI, Macro Poverty Outlook, and official data.
RECENT DEVELOPMENTS
financial businesses deteriorated, reflecting the impact of decelerating loan recoveries and shrinking
Sri Lanka’s economy was already showing signs of
margins.
weakness before the COVID-19 pandemic. After growing by 2.3 percent in 2019, the economy con-
The current account deficit is estimated to have
tracted by 1.6 percent y-o-y in the first quarter of
narrowed in the first half of 2020. A reduction in
2020. The contraction, a first in 19 years, was driv-
imports due to severe import restrictions and low
en by weak performances of construction, textile,
oil prices is likely to have offset reduced receipts
mining and tea industries. Against this backdrop,
from remittances, tourism, tea and textiles. Fol-
the COVID-19 health crisis is believed to have im-
lowing heavy depreciation pressures in March, the
pacted economic activity severely. High frequency
Sri Lankan Rupee (LKR) stabilized in the second
indicators suggest that growth has faltered in the
quarter, as import controls helped the current ac-
second quarter, as curfews impeded economic ac-
count. Official reserves, estimated at USD 7.4 bil-
tivity and global demand remained weak. More-
lion as of August, remain low relative to short-term
over, the closure of airports to tourists between
external liabilities. Included in reserves is a swap
April and September brought tourism activity to
facility of USD 400 million with the Reserve Bank
a standstill.
of India and a loan of USD 500 million from the China Development Bank. The central bank also
Weak demand has kept inflation in check thus far
secured a repo facility with the New York Federal
in 2020, creating room for policy support. An-
Reserve Bank for USD 1.0 billion as a contingency
nual average inflation measured by the Colombo
measure. A Eurobond of USD 1.0 billion is matur-
Consumer Price Index was 4.8 percent in August
ing in October 2020.
2020 despite high food inflation. This allowed the central bank to reduce policy rates by 250 basis
Fiscal accounts deteriorated in the first four months
points and the reserve ratio by 300 basis points
of 2020. Tax revenues fell short due to weak col-
over the first seven months of 2020. The central
lection of value-added, income, and import taxes.
bank also introduced a refinancing facility and a
The fiscal stimulus package implemented in No-
credit guarantee scheme to encourage commer-
vember 2019 -which included a reduction of the
cial banks to increase lending. Despite these mea-
VAT rate and an increase of the registration thresh-
sures, private credit growth remained subdued in
old, as well as import controls-, and slow growth
the first half of 2020. Asset quality and earnings of
contributed to the reduction in tax collection
129
130
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Figure 1: Real GDP growth and contributions to real GDP growth (production side)
Figure 2: Actual and projected poverty rates and real GDP per capita
Percent, percentage points
Poverty rate
10
70
Real GDP per capita (constant LCU) 500000
9
450000 60
8
400000
7
50
350000
6
300000
40
5
250000 4
30
200000
3
150000
20
2
100000
1
10 50000
0 -1
0 2011
2012
2013
2014
Agriculture Industry
2015
2016
Services Net taxes
2017
2018
2019
Real GDP growth
Source: Department of Census and Statistics and staff calculations.
during this period. As a result, despite a modera-
0 2006
2008
2010
2012
2014
International poverty rate Lower middle-income pov. rate
2016
2018
2020
2022
Upper middle-income pov. rate Real GDP per capita
Note: See Table 2. Source: World Bank.
OUTLOOK
tion in public investment, the overall budget deficit increased by 24 percent in the first four months of
The COVID-19 crisis has substantially clouded the
2020, year-on-year. Approximately 40 percent of
outlook and exacerbated an already challenging
the deficit was financed by central bank credit. The
macroeconomic situation. The economy is expect-
central government debt-to-GDP ratio rose to over
ed to contract by 6.7 percent in 2020, with all key
90 percent (from 86.8 percent at end 2019), with
drivers of demand affected: exports, private con-
more than half of the debt denominated in foreign
sumption and investment. The current account
currency. Citing limited fiscal buffers and exter-
deficit is expected to remain low (at 2.2 percent of
nal vulnerabilities, Fitch and S&P downgraded the
GDP in 2020) thanks to low oil prices and strict im-
sovereign rating to B-.
port restrictions, which should largely offset the reduction in receipts from garment exports, tourism
The economic effects of COVID-19 will have sig-
and remittances. However, refinancing require-
nificant welfare implications. Poverty measured
ments will be high, with annual foreign exchange
using the $3.20 poverty line (in 2011 PPP) is esti-
debt service requirements estimated at 7-8 percent
mated to have declined from 9.4 percent in 2018 to
of GDP over 2020-2022. The fiscal deficit is pro-
8.9 percent in 2019. However, the COVID-19 crisis
jected to expand further to over 11 percent of GDP
is believed to have caused sharp jobs and earnings
in 2020, driving an increase in debt levels.
losses. Informal workers, about 70 percent of the workforce, are particularly vulnerable as they lack
Reflecting these challenges, the $3.20 poverty
employment protection or paid leave. The appar-
headcount is projected to increase from 8.9 per-
el industry, which employs about half a million
cent in 2019 to 13 percent in 2020. During the
workers, has reportedly cut significant jobs. While
lockdown, the government extended temporary
agricultural production is expected to be largely
cash support to Samurdhi households, including a
undisrupted, weak external demand likely impact-
large number of which were on the waitlist. But the
ed export-oriented subsectors and wages. High
program is not well targeted and benefit amounts
food price inflation, which remains at double-dig-
are inadequate. Construction and services sectors,
its, is disproportionately affecting the poor who
including tourism, have been important sources
spend a larger share of their budget on food.
of jobs growth in recent years and the outbreak
Beaten or Broken? Informality and COVID-19 South Asia country briefs
will likely harm the prospects of many low-skilled
outlook. In turn, a longer downturn could push many
workers. The government is employing 60,000
small and medium enterprises from illiquidity to in-
graduates and 100,000 individuals from low-in-
solvency, and the poverty rate could rise even high-
come families to support livelihoods, but this will
er as more people suffer income losses. Low growth
remain insufficient and add further strain on pub-
would also put additional strain on public finances.
lic finances. A fall in remittances could adversely impact some poor households that rely on them as
Sri Lanka is also highly exposed to global finan-
an important source of income.
cial conditions, as the repayment profile of its debt requires the country to access financial markets frequently. A high deficit and rising debt levels
RISKS AND CHALLENGES
could further deteriorate debt dynamics and negatively impact market sentiment. Thus, Sri Lanka
A longer than expected outbreak of COVID-19, that
will need to strike a balance between supporting
would extend the horizon and depth of related
the economy amid COVID-19 and ensuring fiscal
economic disruptions, is a key risk to the baseline
sustainability.
Table 2: Macro poverty outlook indicators (annual percent change unless indicated otherwise). 2017
2018
2019
2020 e
2021 f
2022 f
3.6
3.3
2.3
-6.7
3.3
2.0
Private Consumption
3.6
3.7
2.9
-6.7
3.3
2.0
Government Consumption
-6.0
-5.1
9.6
3.8
4.1
2.1
Gross Fixed Capital Investment
6.1
-1.3
4.1
-15.1
3.4
2.6
Exports, Goods and Services
7.6
0.5
7.1
-34.8
3.3
4.6
Imports, Goods and Services
7.1
1.8
-5.8
-29.2
2.8
3.8
3.6
3.7
2.2
-5.7
3.2
2.0
Agriculture
-0.4
6.5
0.6
1.0
2.0
2.0
Industry
4.7
1.2
2.7
-6.1
3.2
1.9
Services
3.6
4.6
2.3
-6.3
3.3
2.0
Inflation (Consumer Price Index)
6.6
4.3
4.3
4.9
4.9
5.0
Current Account Balance (% of GDP)
-2.7
-3.2
-2.2
-2.2
-2.8
-2.9
Real GDP growth, at constant market prices
Real GDP growth, at constant factor prices
Net Foreign Direct Investment (% of GDP)
1.5
1.7
0.7
0.3
0.4
0.5
Fiscal Balance (% of GDP)
-5.6
-5.3
-6.8
-11.1
-8.8
-8.4
Debt (% of GDP)
77.9
83.7
86.8
102.0
106.0
110.3
Primary Balance (% of GDP)
0.0
0.6
-0.8
-4.3
-2.1
-1.6
International poverty rate ($1.9 in 2011 PPP)a,b
0.8
0.7
0.6
1.8
1.7
1.6
Lower middle-income poverty rate ($3.2 in 2011 PPP)a,b
10.0
9.4
8.9
13.0
11.9
11.3
Upper middle-income poverty rate ($5.5 in 2011 PPP)a,b
40.2
38.8
37.6
43.5
41.8
40.6
Notes: e = estimate, f = forecast. (a) Calculations based on SAR-POV harmonization, using 2016-HIES.Actual data: 2016. Nowcast: 2017-2019. Forecasts are from 2020 to 2022. (b) Projections for 2020 are from a microsimulation. Source: World Bank, Poverty & Equity and Macroeconomics, Trade & Investment Global Practices.
131
132
Beaten or Broken? Informality and COVID-19 South Asia country briefs
Beaten or Broken? Informality and COVID-19
PHOTO BY: MAHENDRA N PARIKH/ SHUTTERSTOCK.COM
South Asia country briefs
133