Essays in Applied Economics and Machine Learning

My dissertation concentrates on using methods in applied economics and machine learning to study causal relationships in the context of health economics and monetary policy. In the first two chapters I study the effect of telehealth use on medical spending and utilization in the United States. Leveraging administrative healthcare data, I observe that patients start using telehealth when they experience a severe health shock, and that telehealth users are quite different from non-users. In the first chapter, to identify the causal effect of telehealth usage in the presence of non-random adoption, I use state telehealth parity law changes as an instrument for usage. I find that parity laws have a very small effect on average telehealth use. However, by using a machine learning method for heterogeneous treatment effects, I am able to focus on the group that is most affected by parity laws. Among the most affected subgroup, I find that patients who shift to telehealth use due to parity laws experience a reduction in medical spending by 33.7 percent and a reduction in the probability of an emergency room (ER) visit by 4.6 percentage points. In the second chapter I focus on the effect of telehealth use on chronic disease management for patients with serious mental illnesses (SMI). I use random forest proximity matching to construct a control group for telehealth users and then use difference-in-differences for estimation. I find that, while recovering from a health shock, telehealth users have a significantly higher number of visits and higher drug spending on SMI-related diagnoses. However, a higher number of visits does not result in an overall increase in medical spending, due to telehealth visits being cheaper than in-person visits. In the third chapter, using data on 68 countries from 1974 to 2010, I study how the effect of monetary policy shocks on exchange rates varies by development status and capital market openness. Theoretically, there is ambiguity surrounding the effect of openness on the ability of money to influence the exchange rate and empirically, in light of the recent study by Hnatkovska et al., (2016), there is a contrast between developed and developing countries in terms of the response of nominal exchange rates to monetary policy shocks. My results indicate that more financially open, developed economies experience larger appreciation as a result of monetary tightening, whereas openness does not seem to affect the relationship between exchange rates and monetary policy for developing countries.