Luận án Tiến sĩ Kinh tế Phát triển của Saini Das - Đại học Oklahoma 2021

UNIVERSITY OF OKLAHOMA GRADUATE COLLEGE ESSAYS IN DEVELOPMENT ECONOMICS A DISSERTATION SUBMITTED TO THE GRADUATE FACULTY in partial fulllment of the requirements for the Degree of DOCTOR OF PHILOSOPHY By SAINI DAS Norman, Oklahoma 2021 ESSAYS IN DEVELOPMENT ECONOMICS A DISSERTATION APPROVED FOR THE DEPARTMENT OF ECONOMICS BY THE COMMITTEE CONSISTING OF Dr. Joan Hamory, Chair Dr. Daniel Hicks, Co-Chair Dr. Kevin Kuruc Dr.

Pallab Ghosh Dr. Kirsten de Beurs © Copyright by SAINI DAS 2021 All Rights Reserved. Acknowledgments It did not start out easy. Nevertheless, the writing of this dissertation was made a fullling experience by a number of individuals.

To them, I owe my deep gratitude. I thank my advisors Joan Hamory and Daniel Hicks for their passionate guidance along the way. I thank Joan for showing me the value of rigorous and meticulous research, for helpful feedback in my later projects, and giving me the opportunity to work alongside her in collaborative projects. To Dan, I owe my rst foray into the world of development, for patiently molding (and sometimes steering me away from!) half-formed ideas and being an enthusiastic supporter of my career in economics.

I also thank my professors in the University of Calcutta who stimulated in me the idea of graduate school. I have been fortunate to be surrounded by some amazing women trailblazers- my advisor Joan, my sister Roshni, my aunts Rita and Rina. Thank you for pushing the envelope, so I can live by example. I especially want to thank my sister, for being by my side all the way to graduate school.

I would not be here today without her showing the beacon. I also want to thank my entire extended family for coming together to support each other during this very tumultuous past year. Their perseverance was my strength. Each of my committee members played a signicant role in the development of my chap- ters through conversations and feedback.

I thank Kevin Kuruc, for many a lively discussion that sparked new ideas. Pallab Ghosh, for feedback that enhanced my projects and Kirsten de Beurs for showing new avenues for research. I especially want to thank my ancé Robert Press for being my sounding board during the endless hours spent in coding. But mostly, I iv thank him for toning down my crazy.

Finally, I want to thank my parents Pranab and Chhabi Das, for their tireless support, unwavering trust, and an undaunted motivation to inspire and support me at every turn. I would not be where I am or who I am without their constant enthusiasm for the last couple decades. v Contents Acknowledgments iv List of Figures ix List of Tables xi Abstract xii 1 The Heat is On: The Long-term impact of Heat Stress during Primary School Leaving Examinations in Indonesia 1 1.1 Education in Indonesia .2 Temperature, Humidity and Cognitive Performance .2 Indonesia Family Life Survey .4 Academic Performance and Temperature in the Short Run .5 Long-run consequences of Exposure to High Temperature and Humidity .2 Reduced Form Results .3 Instrumental Variable results .6 Chronic Heat Stress. 28 2 Environmental Factor Endowments, Technological Diusion and Inequal- ity: Evidence from the Green Revolution in India 42 2.2 Agricultural Context and Background .1 Cultivation of Rice and Wheat in India .1 Impact of Productivity on Employment and Wages .2 Impact of Productivity on Welfare .6 Drivers of Growth in Rice versus Wheat districts .1 Dierences in forms of Investments and Factor Accumulation .2 Dierences in Patterns of Adoption .3 Other potential drivers of Agricultural Investments .7 Instrumental Variables Strategy: Agroclimatic Conditions.

68 vii 3 Family Ties and Coping: Evidence from Idiosyncratic and Aggregate shocks in Indonesia 74 3.1 Aggregate and Idiosyncratic Shocks .2 Formation of Family Network in IFLS .1 Dierences in coping strategies .2 Evolution of networks. 90 Appendix 100 viii List of Figures 1.1 Variation in Wet bulb temperature by geographic coverage of test takers .2 Variation in Wet bulb and Dry bulb temperature .3 Non-linear eect of Wet bulb temperature on test scores by subject .4 Impact of temperature on students at various quantiles .5 Reporting Bias in test scores .1 Growth in yields and rate of adoption and post Green Revolution .2 Variation in Yield in Rice and Wheat districts .3 Capital Intensiveness of the new technical change .4 Labor Intensiveness of the new technical change. 89 ix List of Tables 1.1 Summary Statistics: Student Characteristics in IFLS and Weather Variables 31 1.1 (contd) Summary Statistics: Short and Long Term Achievements by sex .2 Short Term Eect of Temperature on Ebtanas Test Scores .3 Heterogeneity in the Short Term Eect of Wet Bulb Temperature on Ebtanas Test Scores .4 Robusness Check 1: Report Cards presented to Interviewer .5 Reduced Form Estimates of the Eect of Wet Bulb Temperature on Long Term Education Outcomes .6 Reduced Form Estimates of the Eect of Wet Bulb Temperature on Long Term Marital Outcomes .7 Reduced Form Estimates of the Eect of Wet Bulb Temperature on Long Term Labor Market Outcomes .8 Reduced Form Estimates of the Eect of Wet Bulb Temperature on Long Term Labor Market Outcomes .9 Placebo Checks: Impact of temperature in months after exam .2 Eect of Productivity Increase on Employment and Wages .3 Eect of Productivity Increase on Welfare measures .4 How much can the adoption of HYV explain the Increase in Productivity? 71 1.5 Pattern of Changes in Land Use and Land Holdings .6 IV estimates of the eect of Productivity Increase on Employment and Wages 73 3.1 Descriptive Statistics by Household Types .2 Descriptive Statistics on total Network Links .3 Results: Aggregate Shock and Networks .4 Results: Idiosyncratic Shock and Networks .5 Results: Aggregate Shocks and Network Formation .6 Results: Idiosyncratic Shocks and Network Formation .1 IV Estimates: Long Term Education Outcomes .2 IV Estimates : Long Term Labor Market Outcomes .3 IV Estimates of the Eect of Test Scores on Labor Market Returns .4 IV estimates: Marital Outcomes .5 Impact of Hot weather during the previous Dry Agricultural Season .6 Chronic Heat Stress: Impact of hot months in preceding School years .1 Eect of Productivity Increase on Employment and Wages using State Fixed Eects .2 Eect of productivity in Neighboring Districts that switch .3 Eect of Productivity on Employment and Wages w/o Bengal and Gujarat 105 xi Abstract My dissertation chapters study the impact of environmental factors on outcomes of human functioning. The rst chapter studies the impact of heat stress during a primary school- leaving exam in Indonesia on outcomes in later life.

The Ebtanas are a national standardized test that students must take to gain entry to secondary school. Using individual-level data on test scores, I rst show that cognitive performance during the test is aected by heat stress. Impacts are heterogeneous across dierent subjects, with math and science being the most heavily impacted. Next, I show that disruptive weather conditions during the Ebtanas have compounding negative eects on a wide range of long-term achievements such as adult educational attainment, labor market participation and entry to the marriage market.

A 1 ‰ increase in temperature in the month of exam leads to 1.53% fewer years of education, 2% fewer hours worked and a 2% higher probability of being married by 18 for women. These ndings stress that even examinations conducted during early or mid adolescence may have impacts that persist through adulthood. The second chapter studies the repercussions of an agricultural productivity shock for labor market outcomes and inequality in India. I show that the increase in productivity had heterogeneous impacts on technological diusion and local labor market outcomes.

In wheat growing areas, the productivity increase was followed by investments in labor-saving tech- nology, demonstrated by an increase in the use of tractors. Rice areas in contrast, invested more heavily in labor-enhancing technology such as fertilizers, creating new opportunities xii for application of labor. These shifts exacerbated inequality in wheat districts while re- ducing inequality in rice districts. I show that these results are robust to xed eects and instrumental variables strategy.

These ndings demonstrate that driven by dierences in environmental factor endowments, a productivity shock can have heterogeneous impacts on agricultural labor markets and inequality. Chapter 3 examines the ecacy of genealogically constructed networks in sharing risk under aggregate versus idiosyncratic income shocks in the context of geographically split-o families in Indonesia. While informal transfers are eective in sharing aggregate risks, they are ineective in an idiosyncratic shock. Plausible reasons include a higher probability of repayment as well as greater economies of scale from resource pooling.

We show that idiosyn- cratic shocks induce households to make long-term and costly changes to their household structure. We demonstrate this in the formation of new split-o families over time who reside outside the district. Our ndings further reveal that controlling for shocks to members in a family network is an important source of omitted variable bias in empirical estimations of the impact of shocks on informal transfers. xiii Chapter 1 The Heat is On: The Long-term impact of Heat Stress during Primary School Leaving Examinations in Indonesia 1.1 Introduction The relationship between environmental conditions and human functioning has received con- siderable scrutiny in academic circles recently (Dell et al.

A branch of this literature has specically focused on the eect of temperature on individual productivity and human capital production. Much of this eect is derived from contempora- neous weather variations and are thus temporary in nature (Gra Zivin et al. (2018); Garg et al. Less research has focused on long-term persistence of these ad- verse outcomes well into adulthood.

Identifying to what extent random weather realizations induce long term changes to an individual's human capital accumulation is crucial from a public policy perspective directed towards adaptive behavior. The idea that an environmental condition such as heat stress can lead to temporary dis- 1 ruptions in cognitive functioning is not new. In a laboratory setting, even minor uctuations in temperature in the range of 2-5 degrees Celsius over normal core body temperature has been shown to induce non-trivial changes in physiology such as mental confusion, memory loss and loss in reaction time (Hancock (1986); Enander and Hygge (1990)). In economics, the few studies that document the direct cost of cognitive disruption in high stakes examina- tions primarily focus at the level of college admissions (Ebenstein et al.

(2016); Park (2017); Gra Zivin et al. The long term toll of such disturbances in high-stakes exams conducted earlier in life is unknown. This is vital given that typically, less developed countries have a very low rate of grad- uation from high school. According to a recent report for Indonesia, only 34.6% of the population have completed education at the higher secondary level (World Bank (2018)).

Of this, an even smaller proportion would have actually taken the examination for entry to college. As such, for over 65% of the population, decisions about adult human capital for- mation is path-dependent on circumstances (including examinations) occurring much earlier in life. Ideally, in this scenario, evaluating the ramications of heat stress during a standard- ized national test occurring earlier in adolescence would yield more broadly generalizable estimates. This paper addresses this need by studying the impact of heat stress during a primary school-leaving exam in Indonesia on outcomes in later life.

The Ebtanas are a national standardized test in Indonesia that students must take to gain entry to secondary school. Taken in early to middle childhood, a typical student is between 12 or 13 years of age. Each student is tested on a range of subjects such as mathematics, science, social science, language and religion, each of which require varying levels of cognitive capability. Using rich data on self reported test scores from the Indonesian Family Life Survey (IFLS), I evaluate the sensitivity to dierent levels of temperature and humidity for each of these subjects.

I show that, similar to previous studies, and consistent across dierent specications, 2 math and science test scores are most impacted by high temperatures.