Risky health behaviors, including substance use and risky sexual behaviors, are a major source of preventable deaths in the U.S. Many studies show a strong positive correlation between risky health behaviors with mental illness, but whether mental health illness promotes risky health behaviors remains unclear.  In this paper, I estimate the effect of depression on risky health behaviors at different stages of the life course using data from the National Longitudinal Study of Adolescent Health. To tackle unobservable confounders that could correlate with depression and risky health behaviors, and reverse causality that risky health behaviors could also shift one's depression;  I exploit variations in friend and family suicide attempts and a genetic score for depression as instrumental variables. I find one standard deviation increase in depression symptoms increases the engagement of individuals’ risky health behaviors including increasing the probability of having multiple sexual partners by 5% and the probability of smoking by 16%. These magnitudes are as big as the effect of obtaining a college degree on risky health behaviors. I also find a strong persistence of risky health behaviors from adolescence to adulthood. I further show these estimates are robust to individual fixed unobservables and relaxation of exclusion restriction assumption. 

The rise in social media use is considered a significant contributor to the increase in mental illness among young adults in the past couple decades, yet research is mixed. We provide new evidence on the causal effects of social media use on mental health of college students, exploiting unique, longitudinal data collected before the Covid-19 pandemic began and at two points during the pandemic. We find small insignificant effects of social media use 4 months into the pandemic during a period of social distancing, but large statically significant effects 18 months into the pandemic when colleges were mostly back in person. Using rich data on mental-health related behaviors, perceived stress and social support, we find some limited evidence that social media use may substitute away from activities that better support mental health. We find that the negative effects of social media use are primarily among socially isolated students. This heterogeneity helps explain the insignificant effects of social media use 4 months into the pandemic; social media use during the period of social-distancing increased more among those who were socially connected rather than those who were socially isolated.

This paper provides new insight into the role of non-traditional lenders -- peer-to-peer (P2P) lending platforms in the borrowing choices of small businesses during the Covid-19 crisis through the adoption of micro-level data from LendingClub. We found a substantial increase in demand for the loan from P2P lending (the loan amounts and the number of new loans) for small business purposes from P2P lenders after authorities implemented mobility restriction policies at the state level.

This paper provides new evidence on the causal effect of education on adult depression. I use data from The National Longitudinal Study of Adolescent to Adult Health and exploit genetic scores for education attainment as an instrument. I find that an additional year of education has a large and significant protective effect on mental health; one extra year of schooling reduces depression symptoms by 0.1 standard deviations and the probability of experiencing major depression by 2%. These effects are robust when relaxing some critical assumptions of the instrumental variables. I also explore the heterogeneous effects of education on mental health and find suggestive evidence that women gain more mental health benefits from an extra year of schooling but LGB individuals gain less. Mechanism results show that education could affect mental health outcomes from better labor market outcomes and health inputs.