Gaurav Khanna is an Associate Professor of Economics at UCSD's School of Global Policy and Strategy. His research covers a wide range of topics, including migration, education, conflict, international student movements, firm production, and more.
1. We usually like to start by asking the interviewee to tell us a bit about how they decided to become a development economist, and what drew them to the area of work that they focus most on. Can you tell us a bit about your pathway to becoming a development economist, and what particularly draws you to work in this field?
I grew up in Calcutta in the 1990s, as India was going through rapid structural and social change. As the country developed, it was hard not to notice that certain sections of Indian society were doing a lot better than others. My mother was a journalist, and we constantly discussed these contradictions around us, that still resonate with me today. For instance, while India was home to one-third of the world’s poor, Indians were also driving the high-tech boom in the US.
I took my first Economics class in college, in Delhi, where most Economics was some version of what the US would brand “Development Economics.” One summer was particularly eye-opening: we were studying social programs in various rural areas under the guidance of economist Jean Dreze. As we spoke with people in these villages, it became evident how deeply intertwined the issues were—lack of access to education, clean water, and political power combined to create a challenging existence. Over time, I gravitated to trying to understand this better in a way that could inform economic policy. What particularly draws me is the hope that some day maybe we can frame more effective policy in low-income settings.
2. You have worked on a wide range of topics in development. What do you see as the main thread or set of questions that link your work, or perhaps, what do you look for in deciding on what new papers to work on?
It’s a great question, because I wish I was more disciplined in not jumping between fields. It takes a lot of time to learn new tools and the literature, but at the same time, it does keep things fresh and exciting. More broadly, I often look for projects that have two criteria: (a) they address a national-level concern, and (b) have policy implications.
Early on, I started with focusing on how human capital can drive economic transitions, say by building new schools, expanding access to minorities through affirmative action programs, or migration for education purposes. I do think of migration as a core focus, because access to good jobs and opportunities in other places seems extremely transformative, even compared to other powerful development interventions I came across. Transformative, not just for the migrants, but also for economies they are migrating to, and migrating from. For instance, migrant origin areas benefit greatly from remittances, and skill development with the prospect of future migration.
At the end of the day, my goal is to get at the causal effects of large-scale government policies, which, unlike smaller researcher-led interventions, come with specific challenges, such as economy-wide spillovers, unintended consequences, and sporadic or poor-quality data.
3. Many of your papers often combine reduced form causal identification with structural modelling. How do you decide when to add a structural model to your papers, and what are a couple of examples of things that surprised you or that you learned when doing so?
I’m often just looking for the best tools to answer a question. Sometimes it’s just a simple cleanly-estimated causal reduced-form relationships. But other times, to make sense of the broader impacts, we may need a model. Models can be extremely useful in answering questions about spillovers and general equilibrium effects. Let me give you a few examples where the model helped me learn something new.
(1) GE effects. Our traditional Mincerian returns to schooling is where we take one student and give them one more year of schooling, and study their wage trajectory. But empirically, we estimated returns to schooling using variation from large-scale programs: tuition subsidies, school-building, compulsory schooling laws. All these programs don’t treat just one student, but rather an entire cohort of students. But if an entire cohort gets more education, that may depress the skilled wage, and raise the unskilled wage, and so change the returns to schooling. Reduced-form estimates may not recover a meaningful return, but a model can help do that, and derive the distributional consequences of such large-scale interventions.
(2) Dynamics. The prospect of migrating and earning a 600% higher wage in the US IT sector, induced students/workers in India to acquire computer science skills valued abroad. Yet, many could not migrate, and the Indian economy could have potentially suffered from an ‘over-supply’ of IT workers. Yet, those who didn’t migrate, and those who returned, over time developed the Indian IT sector, which drove Indian urban growth. This narrative needs a model as it depends on student expectations, innovation spillovers, and other dynamics.
(3) SUTVA and Spatial Spillovers. Amenities (like air quality) affects internal migration in China. This changes where people live, and so, where economic activity occurs. A reduced-form regression of pollution (X variable) on population (Y variable) may have SUTVA issues. If, say, because of a 1SD increase Beijing’s pollution, 100 people moved from Beijing to Shanghai, our reduced-form estimated effect of pollution would be a 200-person difference in population (100 less in Beijing and 100 more in Shanghai). With 300 Chinese cities, this exercise gets more complicated, but a simple model can help better interpret these reduced-form estimates.
(4) Network effects and Cross-sectoral Externalities. I want to understand the effects of a new cable-car system in the high-crime setting of Medellin, Colombia. But when we roll out a transit system, there are no real ‘control groups’, as all neighborhoods are indirectly ‘treated’. This is because they are part of a transit network, and jobs and people move to any potential ‘control’ neighborhoods too. An increase in crime, also makes it difficult for businesses to function (cross-sectoral externalities). Reduced-form estimates are confounded by these spillovers. A spatial model answers what happens to city-level crime with transit expansions.
A good model can generate a nice set of estimating equations that can cleanly be taken to the data, and we can use all our advances in causal inference to estimate. The model helps interpret the coefficient from the reduced-form exercise, and speak to economy-wide impacts.
4. Your network of co-authors is one of the most extensive I’ve seen for someone at your career stage, with an excellent set of over 40 co-authors covering many different topics. What has worked for you in forming so many collaborations, and what advice do you have for our early career researchers in networking and working with different collaborators?
I’m not sure I’m the best person to give advice, as I’m probably not very good at networking. Before getting into a ‘work relationship,’ it’s probably important to gauge excitement, as these are long commitments. I’ve been very fortunate to have some of the best possible co-authors one could ask for. My senior collaborators have been invaluable mentors, and I’ve learned all kinds of cool new tools from junior collaborators.
5. Some of your most cited areas of work comes from your work on international students in the U.S. higher education system. This work has found a wide range of impacts of this flow, from impacts on university revenues, U.S. labor markets, and innovation, to more surprising links such as your paper showing how the trade war with China is costing U.S. universities through Chinese families being less able to afford US tuition. You have also worked on higher education in India, showing that the roll-out of elite public universities had additional benefits in terms of higher educational attainment at lower levels of schooling. My sense is that spending on higher education in developing countries is often seen as regressive and having lower returns for development than expanding access and quality at the primary and secondary levels. But based on your work, do you see more of an argument for shifting funding to higher education in developing countries?
This is a difficult question, as higher education may have many benefits, but if it takes away funds from primary schooling, it may still be regressive. My goal is to understand education’s overall returns for development, including many of the unintended consequences. You already highlight some examples, but let me discuss two more. In a paper, I argue that affirmative action in higher education (and in government jobs), raises the returns to finishing secondary school for low caste students in India, and as a result, improves educational attainment at lower levels of schooling. So access to good colleges, and good jobs thereafter, raises aspirations and gains from education, driving more schooling at lower levels. Such dynamics raise the benefits of improved colleges.
Another paper is about India’s tech boom, which would be difficult to sustain without high quality Engineering colleges. Partly as a result of top Engineering colleges, India’s IT sector was the country’s fastest growing sector in the early 2000s, and India overtook the US in IT exports. These are large gains, but it may still be regressive to focus on higher education only (e.g., the IT boom was concentrated in urban English-speaking middle-class families).
On a final aside, I’m not sure “funding” is the only roadblock to human capital development in low-income countries, as there seems to be room for the same money to be better spent. This is where sensible policy-reform can really matter.
6. What current project are you most excited about working on? Any early findings or surprises to share?
During the Covid-19 pandemic, I was rather shocked to see how India’s economy collapsed. GDP growth was -7% in 2020/1, far lower than many other developing countries. Part of the reason was that supply chains collapsed during lockdowns, having hugely detrimental effects on livelihoods across the country.
While all my research till date had been around ‘human’ capital, I suddenly became interested in (less tangible) aspects of supply chain resilience. For instance, we argue that when a supplier could not supply inputs because of Covid-19 lockdowns, downstream firms found it difficult to find alternative suppliers, and the shock was propagated and amplified across the economy. Moving forward, this is going to be especially concerning with climate change driving increased flooding risk and other climate shocks. Downstream firms are less likely to source from suppliers in high-risk areas, and this exacerbates the distributional effects of climate change by reducing incomes in regions prone to more frequent shocks.
Here are our previous Six Questions with interview series:
· Six questions with Chris Udry
· Six questions with Rohini Pande
· Six questions with Mark Rosenzweig
· Six questions with Martin Ravallion
· Six questions with Andrew Foster
· Six questions with Tavneet Suri
· Six questions with Morgan Hardy
· Six questions with Oriana Bandiera
Join the Conversation