Capital Controls and Income Inequality1 Zheng Liu1
Mark M. Spiegel1
1 Federal 2 Shanghai
Jingyi Zhang2
Reserve Bank of San Francisco
University of Finance and Economics
IMF-OECD-World Bank Conference, September 25, 2020
1 The views expressed herein are those of the authors and do not necessarily reflect the views of the Federal Reserve Bank of San Francisco or the Federal Reserve System. Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
1 / 16
Recent studies link capital surges to income inequality Liberalizing capital controls found to exacerbate income inequality in EMEs (Furceri and Loungani 2018) Theoretical explanations of the channels between capital flows and income inequality are scarce in literature Assessment of impact of capital account policy complicated by financial frictions and presence of other policy distortions Policymakers’ view on capital controls has evolved Surges seen as destabilizing If flows are transitory, then “...use of capital controls—–in addition to both prudential and macroeconomic policy—–is justified as part of the policy toolkit to manage inflows.” (Ostry, et al. 2010)
Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
2 / 16
Paper addresses capital account policies and income distribution in a GE framework OLG model of small open economy Heterogeneous agents (households and entrepreneurs) Intermediation by costly banks Capital account restrictions: taxes on inflows and outflows Examine both short-run and long-run relation between capital controls and inequality
Main findings: Long-run steady state: relaxing controls on either inflows or outflows reduces income inequality Short-run transitions: shocks that boost inflows exacerbate inequality; shocks that induce outflows lower inequality
Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
3 / 16
Confirm SR predictions in cross-country panel
87 EMEs from 2000-2018 Examine impacts of private inflows and outflows on income distribution, measured by GINI Instrument through changes in 2-year treasuries interacted with �remoteness,� proxied by great-circle distance from New York
Results show statistically and economically significant impact of private inflows (+) and outflows (-) on income distribution Robust to a large variety of sensitivity tests
Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
4 / 16
OLG model of small open economy Two types of agents: households (θ) and entrepreneurs (1 − θ) Household consumes, works, and saves (i.e., deposits in domestic or foreign banks) when young and consumes accumulated assets when old Entrepreneur consumes, works, invests, and borrows (from domestic or foreign banks) when young; consumes net worth when old
Domestic bank takes deposits from households and lends to entrepreneurs, subject to intermediation costs Government taxes earnings on both capital inflows and outflows: capital controls Production function Yt = AKt1−−1α (Hht + Het )α Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
5 / 16
Households (H) Utility function y o Uht = ln(Cht ) + β ln(Ch,t +1 )
Budget constraints y Cht + Dt + Bftd = wt Hht + Γht , o ∗ d Ch,t +1 = Rt Dt + (1 − τd )Rt Bft + Th,t +1 − Γh,t +1 .
where Th,t +1 denotes bank dividends and government transfers and Γh,t +1 denotes bequest Capital outflow tax creates wedge between domestic deposit rate R and world rate R ∗ Rt = (1 − τd )Rt∗
Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
6 / 16
Entrepreneurs (E) Utility function y o Uet = ln(Cet ) + β ln(Ce,t +1 )
Budget constraints 2 It I¯ − Kto = wt Het + Bet + Γet , Kto K¯o h i = qtk+1 (1 − δ) + rtk+1 (Kto + It ) − Rlt Bet + Te,t +1 − Γe,t +1 .
y Cet + qtk Kto + It +
o Ce,t +1
Ωk 2
Capital stock follows the law of motion Kt = (1 − δ)Kt −1 + It where Kt ≡ Kto + It denotes end-of-period capital stock
Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
7 / 16
Banks and Foreign Investors Competitive banks take deposits Dt from H and make loans Bt to E Rlt Bt = Rt Dt t Financial intermediation costs (Curdia-Woodford 2016): Ξ( B Yt )Yt b Profits are returned as dividends (Πt ), where Bt b Yt Πt = Dt − Bt − Ξ Yt
Bank optimization implies a credit spread 0 Bt Rlt = Rt 1 + Ξ Yt Foreign investors break even:
(1 − τl )Rlt =
Rt∗ Φ
Bftl Yt
where τl is tax on foreign earnings and Φ(·) is external country risk premium Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
8 / 16
Steady state analytic results
We set bequests and transfers to 0 to simplify and obtain the following analytic results: 1
2
3
↑ inflow taxes (τl ) → ↑ the interest rate (Rl ) and ↓ aggregate income (Y ) ↑ outflow taxes (τd ) → ↓ the interest rate (Rl ) and ↑ aggregate income (Y ) ↑ in either inflow or outflow taxes ↓ the ratio of household-to-entrepreneur capital income, exacerbating income inequality
Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
9 / 16
Transition dynamics following temporary declines in R ∗ Foreign interest rate
0
Capital inflow
12
Capital outflow
5
Bank loans
25
10
-2
20 0
8
15
-4 6
-5
10
-6 4
5 -10
-8
2
-10
0 0
5
10
Domestic deposit rate
0
0 -15 0
5
10
Domestic lending rate
0
-5 0
5
10
Credit spread
2.5
0
5
10
Capital price
30
2
-2
-2
20 1.5
-4 -4
1
10
-6 0.5 -6
-8
0 0
-10
-8 0
5
10
Output
0.4
-0.5 0
10
Investment
1
0.3
5
-10 0
5
10
Net import
40
0
40
0.8
20
30
0.6
0
20
0.4
-20
10
0.2
-40
5
10
Bank profits
0.2
0.1
0
0 0
5
10
0
-60 0
5
10
-10 0
5
10
0
5
10
Rl declines, raising q k , and stimulating I and Y Outflows also decline, lowering R, hurting households Inflows increase, while outflows decrease, raising income inequality Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
10 / 16
Empirical implications
Model predictions: 1
Shocks that increase capital inflows raise inequality (E benefits more than H)
2
Shocks that increase outflows reduce inequality
3
Sensitivity to flows may depend on savings rate and labor income share
Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
11 / 16
We test these predictions in a cross-country panel 87 EMEs from 2002-2018 1
Income distribution measured by GINI coefficient
2
Private capital flows from Lane and Milesi-Ferretti (updated)
3
Exclude OFCs
Endogeneity an issue 1
IV with 2-year treasury interacted with distance to NYC as first instrument
2
Need 2 instruments for both inflows and outflows; also use 3 regional dummies, ASIA, AFRICA, and WESTHEM
Also include battery of conditioning variables in 2nd stage Standard errors clustered by year Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
12 / 16
Baseline specification
Baseline specification GGINIi,t = c + β 1 PINFLOWSi,t + β 2 POUTFLOWSi,t + βXi,t + θt + ei,t GGINI : YoY changes in Gini coefficients (YoY changes) PINFLOWS: (∆ national liabilities − gov. borrowing)/GDP POUTFLOWS: (∆ national assets − ∆ official reserves)/GDP Xi ,t is vector of conditioning variables: CAPOPEN, TRDOPEN, LOWCORR, GDPCAP, POP
Also consider a specification with net private inflows alone
Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
13 / 16
Baseline regression results Dependent variable: PINFLOWS POUTFLOWS
(1)
(2)
0.107*** (0.042) -0.263*** (0.100)
NPINFLOWS Observations CLR P-value
(3) 0.083*** (0.028) -0.315*** (0.056)
0.141*** (0.031) 968 12.76 0.01
(4)
968 12.12 0.01
(5) 0.116*** (0.026) -0.338*** (0.109)
0.086*** (0.024) 1,165 14.00 0.01
(6)
1,165 13.60 0.01
0.112*** (0.023) 968 13.07 0.01
968 12.37 0.01
One std ↑ in gross inflows raises Gini by 1.35 percentage pts One std ↑ in gross outflows reduces Gini by 1.56 percentage pts One std ↑ in net inflows raises Gini by 1.80 percentage pts Conditioning variable coefficients in paper Similar results with conditioning variables dropped Col (3) and (4)) full sample (1,165 obs) Col (5) and (6) baseline sample (968 obs) Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
14 / 16
Optimal capital account policy following R ∗ declines Benchmark policy
Optimal inflow tax
Optimal outflow tax
(1)
(2)
(3)
(4)
(5)
(6)
(7)
ω
0.5
0.3
0.5
0.7
0.3
0.5
0.7
τl1 τl2 τd1 τd2
10.17% 10.17% 1.64% 1.64%
15.35% 27.07% 1.64% 1.64%
10.17% 10.17% 22.81% 10.07%
10.17% 10.17% 8.68% 1.74%
10.17% 10.17% −30.98% −27.27%
Optimal capital flow tax rates 18.43% 22.60% 1.64% 1.64%
20.69% 19.16% 1.64% 1.64%
Planner ↑ SR inflow tax τl1 to curb inflows; ↑ ω leads to stronger tightening LR inflow tax τl2 also ↑, but production and household income ↓, ⇒ ↑ ω implies smaller increase in τl2 Optimal SR outflow tax τd1 ↑, ↓ domestic rates and ↑ loan demand. ↑ ω ⇒ ↓ τd1 (or even negative) LR outflow tax τd2 much lower than SR; benefits H more than E Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
15 / 16
Conclusion In a small open economy with heterogeneous agents and financial frictions, capital account liberalization impacts income distribution In the long run, permanent reductions in taxes on both inflows and outflows raise household income share and reduce inequality In the short run, changes in inflows and outflows have opposite effects on inequality: inflows raise inequality but outflows reduce it Temporary declines in world interest rate lead to surges in inflows, skewing distribution in favor of entrepreneurs Tightening inflow restrictions mitigates this effect
Model’s predictions about short-run effects of capital flows on income inequality are supported by data.
Liu, Spiegel, and Zhang
Capital Controls and Income Inequality
September 3, 2020
16 / 16