Research
Working Papers
On the Spatial Distribution of Colleges
(Job Market Paper)
In the United States, there is substantial cross-state variation in college (1) spending per student, (2) tuition, and (3) capacity. Empirically, I show that larger, more productive states have higher in-/out-of-state tuition, spend more per student, and offer less availability to out-of-state students. Motivated by these facts, I develop a heterogeneous agent lifecycle model embedded within a quantitative spatial framework to better understand cross-state patterns in higher education. Key features of the model include an endogenous distribution of college quality and firm activity across locations, along with the migration-based sorting of students and workers. I estimate the model to replicate cross-state dynamics of migration and education choices, wages, and college characteristics. Key parameters governing college quality and policy choices are calibrated and externally validated using causal, reduced-form evidence. Given the significant diversity and decentralized structure of the U.S. college system, I use the model to compute the optimal spatial distribution of college financing and availability at both the state and federal levels. For the federal policymaker, a key constraint is the historical distribution of colleges: the physical locations of colleges were determined more than 100 years ago, often far from centers of modern economic activity. For a fixed level of expenditures on higher education, optimal state policy increases aggregate welfare by 3.1%, and optimal federal policy increases welfare by 7.2%.
While the United States and Canada share many similarities, there are stark differences in their levels of income inequality and intergenerational earnings persistence. This paper investigates college quality distributions, tuition subsidies, student loan systems, and tax policies as possible sources of these differences. A heterogeneous agent model is developed where human capital investments occur over the lifecycle and across generations. The model is calibrated to the U.S. economy and is able to match key moments on intergenerational mobility, lifetime inequality, and higher education. The benchmark counterfactual exercise finds that the system of higher education accounts for approximately 22% of the differences in income inequality and 11% of the differences in intergenerational mobility between the U.S. and Canada. The distribution of college qualities accounts for the majority of differences in inequality, whereas its net effect on intergenerational mobility is small.
From Kindergarten to College:
The Impact of Education Policies over the Lifecycle
with Angela Zheng (submitted)
Across all education levels, recent policies aim to diversify the socioeconomic composition of student bodies. We study how these integration policies interact using a heterogeneous agent overlapping-generations model featuring multiple periods of human capital development. Households sort into public schools through housing location and into college via a competitive admissions process. The quality of schools and colleges is endogenous through peer effects. Key parameters linking college admissions to parent-child investments are identified using causal moments from the data. At the public school level, we model a rezoning policy that increases the proportion of low-income students at the highest-quality school. For college, we examine an income-based affirmative action policy. Public school integration weakens the link between residential location and school quality, increasing intergenerational mobility by 2.2%. On the other hand, the college policy decreases intergenerational mobility by 2.3%. When the high-quality college reserves seats for low-income students, college becomes more competitive, which increases income sorting at the public school level. An integration policy that combines public school rezoning and college affirmative action leads to minimal changes in upward mobility.
Declining Teen Employment: Causes and Consequences
with Alex Wurdinger
Teen employment in the United States has fallen by more than 50% over the last 30 years. We provide causal evidence attributing the majority of this decline to crowding out by adults. To determine the macroeconomic consequences of this decline, we begin by causally estimating the returns to teen employment. We find that teens who worked while in high school and did not attend university earn significantly higher wages and face lower unemployment rates later in life. Next, we develop a general equilibrium model and calibrate it to match our causal estimates. The model features adolescents choosing between accumulating human capital on-the-job or in school, and adults choosing teen vs. non-teen occupations. Using this framework, we simulate a shock to adult occupations that displaces teen workers. Teens crowded out by adults experience significant welfare losses, mostly due to a loss of work experience. Increased college attendance mitigates these negative effects. A decomposition exercise reveals that minimum wage levels play a key role in transmitting the impact of increased adult competition into falling teen employment. Finally, we show that education policies, such as optional vocational training are effective at reducing adverse effects for teens who have been crowded out of the labor market.
On the Changing Relationship
between Net Public Foreign Assets and Growth
with Juliana Gamboa-Arbelaez (submitted)
replication package
This paper documents novel stylized facts and illustrates a simple mechanism explaining patterns of net public foreign assets across countries and time. Previous literature found an unexpected negative correlation between growth and net public foreign assets from 1980 to the mid-2000s. Analyzing data up to 2019 we find that this result no longer holds. We document a significant reversal since 2004, with the correlation now zero or weakly positive. Empirically, we attribute this shift to a substantial substitution from public debt towards international reserves, particularly for slower-growing countries. Simultaneously, low-growth countries experienced heightened productivity volatility. Augmenting an open economy neoclassical growth model to include uncertainty, we demonstrate that this increased risk faced by low-growth economies explains 46% of the change in correlation.