Essays in Empirical Macroeconomics and FinancePublic
This dissertation examines three distinct empirical questions in macroeconomics and finance. Chapter 1 studies the reasons why households file for bankruptcy. The debt relief households obtain in bankruptcy provides insurance against wealth losses, but also distorts borrower incentives to repay debt, discouraging lending. Understanding how bankruptcy filings respond to changes in cash-flows and "strategically" to relief generosity is important for assessing these trade-offs. This paper presents new evidence on the causes of bankruptcy using data on millions of mortgage borrowers. First, I exploit a kink in debt relief generosity induced by asset exemption laws in a regression kink design (RKD) to estimate a small positive effect of an increase in generosity on filing. Second, I exploit quasi-experimental variation in mortgage payment reductions to estimate a large negative effect of an increase in cash-flows on filing. The RKD isolates the strategic motive by holding wealth fixed and varying the payoff from filing while the payment reductions affect filing by increasing cash-flows that are not generally seizable in bankruptcy. Using a simple model of household bankruptcy, I show that the relatively weak strategic motive implies consumption gains to filers must be large but that other costs of bankruptcy, such as social stigma or from credit market exclusion, must also be large. My findings are consistent with a lack of insurance against cash-flow shocks driving bankruptcy and imply that increases in the generosity of bankruptcy only weakly incentivize further filing. Chapter 2 explores one channel through which financial crises can alter the strength of the credit channel of monetary policy. Analyzing microdata on the universe of US credit unions and exploiting plausibly exogenous variation in exposure to ABS markets, I find that asset losses among lenders increase the sensitivity of their lending to the policy rate. A 10 basis point fall in the two-year Treasury yield generates a 0.86 percentage point increase in quarterly lending growth when assets are unchanged. However, the same policy rate change leads to a 1.15 percentage point increase for a credit union experiencing a one standard deviation asset loss. This is a more than 20 percent difference relative to median lending growth. Additionally, a 10 basis point policy rate reduction lowers the effect of a one standard deviation asset loss from a 3.20 to 2.91 percentage point decrease in lending growth. These findings suggest that monetary easing is more potent among lenders with recently weakened balance sheets and there exists an additional benefit of monetary easing which reduces the contractionary effect of asset losses. Chapter 3 uses new data on the timing of sovereign defaults during 1869-1914 to quantify an informational channel of contagion via shared financial intermediaries. Concerns over reputation incentivized Britain’s merchant banks to monitor, advise, and occasionally bail out sovereigns. Default signaled to investors that a merchant bank was not as willing or able to write and support quality issues, suggesting that its other bonds may underperform in the future. In support of this channel, I find that during a debt crisis, a 5 percent fall in the defaulting bond’s price leads to a 2.19 percent fall in prices of bonds sharing the defaulter’s bank. This is substantial compared to the 0.24 percent price drop among bonds with different banks. Information revelation about financial intermediaries can be a powerful source of contagion unrelated to a borrower’s fundamentals. In modern financial markets, third parties such as credit rating agencies, the IMF, or the ECB could similarly spread contagion if news about their actions reveals information about their willingness to monitor risky borrowers or intervene in crises.