80
The Fast Track to New Skills
Box 2.4 Estimating Value Added Consider outcome Yijtk , where k refers to the outcomes of interest, i to the student, j to the program, and t to the cohort. Thus, the following can be posited: Yijtk = X i ' α k + Z ijt ' β k + u kj + δtk + ∈ijtk , where Xi contains student Systems Approach for Better Education Results (SABER) 11 scores as well as gender, age, parental socioeconomic status, and mother’s education; Z represents peer characteristics (average SABER 11 and proportion of peers’ mothers with at least a bachelor’s degree) in the student’s cohort; uj is a program fixed effect; δt is a cohort fixed effect; and ∈ijt represents unobserved individual-level characteristics that affect the outcome of interest. Thus, the vector X contains “individual characteristics,” and Z contains “peer characteristics.” The main parameter of interest is the set of program-level fixed effects, uj , which estimate the program-level contributions to student outcomes. X determines not only student outcomes, but also selection into programs. This is particularly true of SABER 11, which measures students’ academic readiness for higher education. Although many short-cycle programs are open access, others use SABER 11 as an admission criterion. Further, SABER 11 provides the student information on her abilities that she may use when choosing a program.
For wages, figure 2.9 further illustrates the variation in program contributions across and within fields. Most programs in engineering and architecture, health, and math and natural sciences make above-average contributions. In contrast, most programs in economics and business, agronomy, social science, and arts make below-average contributions. Once again, although some fields are more likely to deliver an above-average contribution than others, all of them—health in particular—display large within-field variation. To summarize, program value-added contributions vary across and within fields. The large within-field variation in program contributions implies that for a student seeking a program that will add much value to her human capital, it is not enough to choose a field with a high average contribution, as low-contribution programs exist even within seemingly “good” fields. From the point of view of the policy maker, the large within-field variation means she may need to monitor value-added contributions closely to identify programs throughout the distribution, especially at the lower end. What are the characteristics of institutions associated with high value-added contributions to wages? Correlations between program-level contributions and a set of institution- and program-level characteristics help to address this question. These suggest that three-year programs contribute more than two-year programs.22 And, consistent with some of the findings reported in the previous sections, technological institutes deliver higher wage contributions than universities.
