Essays in financial economics

Abstract

This dissertation consists of three independent chapters that each shed light on a particular aspect of the banking sector. The first chapter is joint work with Agostino Capponi and Joseph E. Stiglitz (both of Columbia University), and it is motivated by the fact that banks usually hold large amounts of domestic debt which makes them vulnerable to their own sovereign’s default risk. At the same time, governments often resort to costly bailouts when their banking sector is in trouble. We investigate how the network structure and the distribution of sovereign debt ownership within the banking sector jointly affect the optimal bailout policy under this “doom loop”. We argue that rescuing banks with high domestic sovereign exposure is optimal if these banks are sufficiently central, even though that requires larger bailout expenditures than rescuing otherwise identical low-exposure banks. Our model illustrates how the “doom loop” exacerbates the “too interconnected to fail” problem. The second chapter, written jointly with Melina Papoutsi (European Central Bank), studies potential side-effects of large exposure disclosure requirements for banks. Banks in the euro area must inform banking supervisors about exposures to individual counterparties that exceed 10% of the bank’s capital. Using a new granular dataset that combines banks’ loan and security exposures, we employ a Regression Discontinuity (RD) model to test whether banks pass on the cost of complying with the large exposures framework to borrowers above the threshold in the form of higher interest rates. We find no statistically significant evidence of bunching of exposures below the threshold, but we present preliminary evidence in support of a sizable interest rate premium for large exposures. Our results suggest that, ceteris paribus, when a firm becomes a large exposure for its bank, its interest rate increases by 6–8 basis points on average, which implies an increase in interest cost for the average loan of 3.5–4.5%. The third chapter investigates the nature of the constraints banks face in extending their balance sheet to originate loans. Many papers in the banking and finance literature (and even more so in macroeconomics) either postulate or empirically find a positive causal link between banks’ deposit and loan growth. A naive interpretation of this result is that deposit inflows increase the amount of funds available to a bank for lending purposes and therefore trigger additional lending. However, this interpretation is at odds with banks’ ability to create money-like liabilities at will. This paper argues that if an inflow of deposits triggers additional lending, it must be because it relaxes some previously binding constraint. Otherwise, the bank could just as well have increased lending in the absence of deposit inflows. In this paper, I empirically explore three possible channels that could explain the observed positive relation between deposit and loan growth: Relaxation of either (a) a dynamic leverage constraint, (b) a dynamic liquidity constraint, and (c) a “hot potato effect”.