Skip to content

Roundup from ISI 2019

In December 2019, IDinsight Financial Inclusion team members attended the Indian Statistical Institute (ISI) conference on Economic Growth and Development. Manager Andrés Parrado presented a paper on the results of sending SMS reminders to banking agents in rural India. A future blog post will discuss these findings in detail. In this post, IDinsight attendees share takeaways from the conference including findings on health, education, history and other economics topics.

Banner for the 2019 Indian Statistical Institute (ISI) Conference. ©IDinsight/Siobhan McDonough

To those not familiar with India’s academic landscape, the ISI conference is the economics conference to attend. In 2019, for example, about 174 papers were presented in 64 parallel sessions and three plenary lectures, covering both empirical and theoretical work across every field of economics. This post showcases the diversity of topics, highlighting public employment programs, girls’ education, history of urbanization, and more. It’s the Indian equivalent of the American Economic Association’s annual meetings, but without the stresses of job recruiting. Instead, you find Indian and international economists having long conversations over lunch in the winter-sunlit gardens of the Indian Statistical Institute.


Over two days, I attended lectures ranging from intergenerational mobility in India to the effect of local ambassadors on the uptake of digital financial products in Peru. Of these sessions, two stood out. In “General Equilibrium Effects of (Improving) Public Employment Programs: Experimental Evidence from India,” Karthik Muralidharan shared findings from a large-scale randomized experiment (19 million people!) illustrating the effects of improving India’s public employment program National Rural Employment Guarantee Scheme (NREGS). The reform introduced a biometric card system to deliver payments to beneficiaries who had been part of NREGS. The authors find that the biometric card system “raised low-income households’ earnings by 13%, with 90% of this gain coming from non-program earnings, driven by increases in both market wages and private-sector employment.”

Now, the working paper version of this report has been out for a while. But it is worth emphasizing two striking aspects of this work. First, the gains in income stem from improvements on an existing program, illustrating that tweaking an already existing program can be a cost-effective way of improving service delivery. Second, and a key point that Muralidharan emphasized during the presentation, is that the vast majority of gains come from private sector employment. The positive spillovers of the program are incredible. The program is able to raise market wages when there are only a handful of other employers (or in more technical terms, improving NREGS raises the outside wage in what is a monopsonistic environment). As the authors themselves highlight in their paper, the policy implication is that well-implemented welfare programs can address more structural market failures when reaching low-income people and thereby increasing well-being.

Another interesting talk was titled Parental Investments and Early Childhood Development: Short and Long Run Evidence from India . This research speaks to the effects of India’s Integrated Child Development Services (ICDS), one of the largest early childhood development programs in the world. ICDS provides education and early childhood resources to children under the age of six. Saravana Ravindran uses a differences in differences strategy based on the varying times at which the program was implemented to estimate its effect on parental investment (such as the amount of food, education, or additional remedial classes that parents purchase for their children). He concludes that “children exposed to the ICDS program in India were significantly less likely to be malnourished and more likely to be able to read and do math.” Nevertheless, he points out that these effects might crowd out parents’ investments in the nutrition of siblings, particularly girls. These results suggest that welfare programs should carefully consider the potential negative spillover effects in contexts where social norms favour males.


The most interesting session I attended explored how gender inequality within households affects education in India. Geeta Gandhi Kingdon and Sandip Datta, in their paper “Gender Bias in Intra-Household Allocation of Education in India: Has it Fallen over Time?” analyzed how families in rural areas enrol and financially invest in boys’ versus girls’ education. In 1995 these inequalities were largely due to higher enrollment of boys than girls; by 2014, these enrollment gaps had decreased (and had almost disappeared for primary school students). In 2014 the inequalities that persist are primarily due to families choosing to send their sons to fee-charging private schools, and their daughters to tuition-free government schools. Differences in educational spending, once girls and boys are enrolled in school, have actually increased from 1995–2014, most notably for children ages 15–19.

But who are these households spending money on sons and not daughters? Ashwini Deshpande and Apoorva Gupta looked further into the causes of these within-household gender gaps in education in “Nakusha? Son Preference, Resource Concentration and Gender Gaps in Education.” They theorized reasons a family might allocate educational resources differently between genders — from bias against girls to families forced to invest the only education money they have to the child they think has the best chance to succeed.

They found the increase in the gender gap in quality of education over the past three decades is the largest in families with “unwanted girls” — families who keep having girl children until they get a boy, for example, a family with three girls and then a boy. The second-largest gender gap comes from families motivated by “resource concentration” — families who did not demonstrate gender bias in the patterns of children they had, but who may believe their sons have a higher likelihood to succeed in a patriarchal society.

Unfortunately, these papers show that even while gender discrimination in terms of non-enrollment of girls and sex-selective abortion is decreasing, it is occurring (and even increasing) in other ways. Women’s labour force participation rates in India remain low, and many families continue to value and invest in boys more. The authors of the second paper conclude there is an “urgent need for mainstreaming gender concerns in educational and labour market policies, as well as highlight the multitude of positive contributions by Indian women in multiple fields to dispel the notion that investment in the quality of daughters’ education is a waste of resources.”


I’m interested in history and urbanization, so I attended the “History: Urban Settlement and Institutions” session. Aliz Toth (Stanford University) has assembled comprehensive data on the location and size of urban centres, trade routes, and political borders in medieval India. She uses this data to study the effects of political fragmentation on urban growth in medieval India (paper here). She finds that, unlike western Europe, fragmentation didn’t lead to much warfare, and hence didn’t reduce urban growth all that much.

In the same session, Abhimanyu Gupta (Essex University) used (really) rich archaeological data from the Greek island of Antikythera (famous for the Antikythera mechanism, apparently the world’s first computer) to show that the canonical urban economics model — the monocentric city model, which predicts that the rich live closer to the urban centre — also held true in Ancient Greece (paper here, joint with Jonathan Halket, Texas A&M University).

In the next session, on Institutions, Kishore Gawande (UT Austin) shared research on three Chinese rebellions that occurred between 1851 and 1880: the Taiping Rebellion, the Nian Rebellion and the Dungan Revolt. In this paper with Hua Cheng (Nankai University), Gawande argues that these rebellions helped to build local fiscal capacity, because the Qing dynasty empowered generals to collect taxes locally to suppress these rebellions. This local state capacity is highly persistent: Chinese counties with more intensive rebellions in the second half of the 19th century collected higher local tax revenue per capita in 2000. The authors use the severity of the rebellions (measured in terms of population loss) as an instrumental variable for state capacity, and show that higher local state capacity causes local economic growth in China.

Finally, the next day, I attended a fascinating presentation by Abhiroop Mukhopadhyay (ISI Delhi) on how the partition of India may have lowered trust and social cohesion in the areas that received partition refugees. With Latika Chaudhary (Naval Post Graduate School) and Prasad Bhattacharya (Deakin University), he shows this using sample survey data from the WHO’s Study on Aging and Adult Health (SAGE), which was a bonus since I’d never heard of this data source before. That was all the time I had to reflect on the thumb of history, before I had to head back to work to try and improve policymaking in the present.