Published Papers

Abstract: We examine how substance-use-disorder (SUD) treatment providers respond to private-insurance expansions induced by states’ equal coverage (parity) laws for SUD treatment vis-à-vis general health care services. Economic theory suggests that such laws will lead to changes in providers’ behaviors. We use data on licensed specialty SUD treatment providers in the United States between 1997 and 2010 in a differences-in-differences analysis. During this period, 12 states implemented laws that require equality in coverage for SUD treatment. Following the passage of a state parity law, we find that providers are less likely to participate in public markets, are less likely to offer price discounts to patients, and increase the quantity of health care provided. Further, we find evidence of decreases in treatment intensity following passage of a parity law and of heterogeneity in effects across ownership status.

Work in Progress

Abstract: We present a parsimonious meta-learning approach to forecasting defaults with severe class-imbalanced data, as in the case of China bond market. We do so by optimally mixing sample wrangling techniques with ensemble learning algorithms that randomly create a sufficiently large number of balanced sub-samples from the original data. We find decent predictions with a good tradeoff between Recall and Precision. We further applied a cluster analysis to investigate performance by closely examining the false positives generated by our model and found such firms are, in fact, much ”closer” to defaulted than non-default firms, as measured by the Euclidean distance of their issuers’ important features. A key reason they have been kept afloat could be that these bonds are issued by firms that are too important to fail, which implies a bailout or restructuring by the State despite a dire financial situation and further demonstrates the robust and consistent predictive power of the model.

Abstract: In 2011, the Federal Deposit Insurance Corporation (FDIC) changed the relative return of holding reserves for U.S. branches of foreign banks compared to conventional domestic banks. The data show higher excess reserves held by U.S. branches of foreign banks following the policy change. This paper analyzes how that policy reform affected the federal funds market. To that end, it presents an equilibrium model in which different types of banks choose an asset portfolio. In the model, foreign and domestic banks face different tradeoffs when holding reserves. The comparison of the steady-states affirms that the foreign sector's portfolio choice contrasting that of the domestic sector may have implications on the outcome of the federal funds market and thereby total bank lending. 

Abstract: In July 2011, the Dodd-Frank mandated the FDIC to widen its assessment base from adjusted domestic deposits to total assets less tangible shareholders’ equity for the U.S. chartered banks. U.S. branches of foreign banks, established after December 1991, have no charter and no deposit insurance and are therefore not affected by this policy change. This paper documents the response of bank portfolio choice of deposits and excess reserves by exploiting the variation in policy effects on these two sectors. A difference-in-difference comparison suggests a decline of 0.16 reserves to assets of U.S. chartered banks compared to U.S. branches of foreign banks follow- ing the policy change.

Abstract: The paper evaluates monetary policy pass-through to deposit and lending rates given the competition across banks using the Dynamic Stochastic General Equilibrium (DSGE) model with sticky prices. While utilizing Boone's (2008) market power indicator, the data suggest that U.S. banks' market power is high in the deposit market and somewhat high in the loan market, with an incline in competition in both sectors in the last two decades proceeding 2001. The central assumption of the model is that the pass-through depends on competition across banks. It includes banks with imperfectly competitive markups for loans to firms, markdowns of deposit rates to consumers, and a monetary policy authority that can either change the federal funds rate or the spread between the federal funds rate and the rate paid on excess reserves. The model estimations align with the empirical evidence suggesting banks will compensate on loan spreads to avoid the contraction in lending caused by higher policy rates, while deposits will fluctuate less, and therefore spreads may increase when market rates increase.