Essays in Development Economics

Martin Wiegand

Research output: PhD ThesisPhD-Thesis - Research and graduation internal

102 Downloads (Pure)

Abstract

The three main chapters in this dissertation cover three different topics in development economics: education, poverty, and infrastructure. Each chapter contains relevant findings for the respective subject area and methodological contributions. Chapter 2 studies the impact of a conditional cash transfer (CCT) program on students’ decisions to continue school once the program ends. The CCT in question is Mexico’s PROGRESA, which covered students only until the end of middle school. I find that the program reduced the probability to transfer to high school afterwards by about 10 to 14 percentage points. The reasons appear to be behavioral: cash crowds out the intrinsic motivation for seeking education, and CCT programs ending early may signal that school is not worth it after a certain point. In addition, the program had positive spillover effects on students from better-off families who were not eligible, presumably raising their desire to distinguish themselves by staying in school. Identifying the effect on the transition probability poses a unique challenge: if PROGRESA has successfully kept students in middle school, then samples of middle school graduates who received the program are likely different from those who did not. To tackle this issue, I apply double machine learning—a recently developed method to identify treatment effects in the presence of many potential confounders. I extend the method to account not only for selection bias but also for non-random attrition. Chapter 3 is about participatory wealth rankings (PWRs)—a targeting method in which representatives of a community rank households by their wealth. I demonstrate how PWRs can be used to construct a welfare measure that reflects local perceptions of poverty, using data from a field experiment on targeting in Indonesia. The idea is to estimate the relationship between rankings and household characteristics via a rank-ordered logit model. The welfare scores predicted from this model can be used to compare households from different villages. I then estimate the impact of using this new welfare measure as targeting goal on program satisfaction of villagers and village leaders. I find that higher targeting accuracy, as measured by the rank-based welfare measure, significantly increases satisfaction. Furthermore, after controlling for targeting accuracy, the PWR does not lead to significantly higher satisfaction than consumption-based targeting methods. Lastly, targeting accuracy explains satisfaction outcomes better when it is measured against rank-based welfare rather than predicted consumption. This holds true even for communities where no PWRs had been conducted. The results show that taking local welfare perceptions into account leads to more satisfactory targeting outcomes, while the procedure of meeting and ranking households seems to matter little, if at all. In chapter 4, which is joint work with Eric Koomen, Menno Pradhan, and Christopher Edmonds, we study the impact of road development on household welfare in rural Papua New Guinea (PNG). Using two household surveys from 1996 and 2010 as well as corresponding road maps, we construct road quality variables for the route connecting households to urban areas. We use a correlated random effects model to account for unobserved location-specific effects that might influence both road quality and households’ well-being. Our results show that upgrading the roads leading to the nearest town increases average household consumption, housing quality, and school enrollment, and reduces reliance on subsistence farming. An analysis by subgroups shows that the effects on consumption and poverty are substantially higher for remote households. Furthermore, we apply generalized quantile regressions to look for effect heterogeneity along the distribution of consumption. The estimates indicate that upgrading dirt roads has a higher effect for the poorest households. The results imply that road infrastructure programs may be considered pro-poor policy measures.
Original languageEnglish
QualificationPhD
Awarding Institution
  • Vrije Universiteit Amsterdam
Supervisors/Advisors
  • Elbers, Chris, Supervisor
  • Pradhan, MP, Supervisor
Award date12 Oct 2021
Publisher
Print ISBNs9789036106641
Publication statusPublished - 12 Oct 2021

Keywords

  • development economics
  • conditional cash transfer
  • Progresa
  • machine learning
  • doubly-robust estimation
  • targeting
  • welfare measures
  • participatory wealth ranking
  • rural roads
  • poverty

Fingerprint

Dive into the research topics of 'Essays in Development Economics'. Together they form a unique fingerprint.

Cite this