Higher education subsidies are primarily distributed through need-based programs, without differentiating by college major. However, labor market outcomes vary significantly across majors. Science and Engineering graduates tend to earn the highest wage premiums and face the lowest unemployment rates, while there is a strong prior pattern of ability selection into these majors. I study the aggregate effects of higher education subsidies, taking into account key differences across college graduates by major. These differences include ability selection, patterns of skill formation, and frictions in post-college labor markets. I develop an equilibrium labor market search model with two-sided multidimensional heterogeneity and endogenous college and major decisions. In my model, individuals are initially sorted into college majors based on their multidimensional abilities (math, verbal, and social) and preferences. These decisions lead to differential human capital accumulation across all ability dimensions. I use data from the NLSY79 and O*NET to calibrate the model, which I then use to evaluate the effects of subsidies targeted at specific college majors. My findings indicate that Science and Engineering and Business and Economics majors demonstrate limited responsiveness to subsidies compared to Humanities and Social Sciences majors. This is because Humanities and Social Sciences majors tend to attract individuals who might otherwise opt out of college. The expenditure-neutral, welfare-maximizing subsidy scheme, which allows for differential subsidies based on college major while maintaining fixed total subsidy costs, leads to a 0.5% increase in overall welfare. This policy also results in a 35% increase in the number of Science and Engineering graduates.

Using establishment-level World Bank Enterprise Surveys, I document the following trends (i) the average tax noncompliance rate, defined as the ratio of unreported sales to total sales, decreases with GDP per worker, (ii) the tax noncompliance rate is size-dependent, i.e., small establishments conceal a higher fraction of their sales than large establishments, (iii) the level of this size-dependency diminishes as GDP per worker increases. To examine the implications of these findings for managerial quality and aggregate output, I develop a modified version of Lucas(1978) span-of-control model in which managers invest in their managerial skills and choose how much of their income to report to the government after considering the risk of getting inspected by tax officials. The results reveal that incomplete tax enforcement significantly diminishes economy-wide managerial quality, with the magnitude of this impact escalating with the level of size-dependency in tax noncompliance. For instance, transitioning from the benchmark economy, calibrated to U.S. data, to an economy similar to Brazil’s tax enforcement regime leads to an approximate 23% reduction in average managerial quality and roughly a 3% decrease in output.

We exploit data from the Mexican establishment census and show that informal plants grow along their life cycle at significantly lower rates compared to formal plants. To quantify the aggregate losses from the marked differences in growth rates between formal and informal plants, we develop a general equilibrium model where plants grow by investing on their productivity and informality arises as a result of incomplete enforcement. In equilibrium, informal plants exhibit flatter life cycle profiles to avoid detection and thus lower their tax burden. Full enforcement of the tax while reducing the rate to keep the tax-to-output ratio constant would increase aggregate output by 8 percent relative to the benchmark. Average plant size would increase by 36%, and the average plant would grow at a pace almost 50% faster.

We examine the interaction between unemployment benefits and subsistence self-employment in developing countries, where limited opportunities in formal labor markets push low-productivity individuals into self-employment. We develop a model that incorporates occupational choices, search frictions, and borrowing constraints, allowing individuals to choose between wage work, own-account work, and entrepreneurship. Our model highlights how low-wealth, low-ability individuals select into own-account work due to their inability to sustain the search for wage employment in a frictional labor market. Calibrating the model to Mexico, we explore the effects of introducing U.S.-style unemployment benefits. Our results indicate that if own-account workers can conceal their employment status, unemployment benefits lead to an excessive increase in own-account work and a reduction in aggregate output. However, when employment status is effectively monitored, unemployment benefits can reduce subsistence self-employment by 15 percentage points and generate significant output gains of nearly 2%.