Academic Research

Composition Risk: How Debt Structure Impacts Debt Contracting (Job Market Paper)

Links: Github Google Drive

Committee: Stefano Giglio (co-chair), Gary Gorton (co-chair) and Heather Tookes

In a subordinated debt setting, the credit risk of a lender depends not only on the value of the defaulting firm but the relative sizes of each creditor's claim. I test if this channel of risk, which I title composition risk, is reflected in creditors' contracts. Using a novel source of data (assembled via a machine learning technique unique to this paper) on the universe of syndicated lending covenants, I perform one of the first apples-to-apples comparisons of the covenant packages of comparable bonds and loans issued by the same firm at about the same time. I document that bank lenders enjoy a significant covenant `protection premium' and that the composition risk implied by the debt structure of the firm is a strong determinant of the strength of this premium, even after controlling extensively for credit quality. As the size of the senior debt tranche grows to unity, senior bank lenders demand relatively stronger covenant protections while junior bondholders demand relatively higher credit spreads. Using a structural model of subordinated default, I show that this disparity in responses can potentially be explained by the low recovery risk tolerance of bank lenders.

Affirmative Covenants and Information-first Monitoring: Evidence from Syndicated Lending

Links: Github Google Drive

I explore the use of affirmative covenants (which stipulate actions firms must take) in syndicated lending contracts, their importance and how they differ from the traditionally-studied negative covenants. I conduct my analysis using a novel source of data on the originated contracts of syndicated lending agreements matched to their their renegotiations. From the data on originations, I conclude that affirmative covenants constitute a monitoring technology which banks use to ask for more or less information from their borrowers. Compared to negative covenants which are assumed to be a form of tripwire for lenders, affirmative covenants can be conceptualized as setting the sensitivity of these negative covenant tripwires. Using these observations, I hypothesize that banks monitor loans on an `information-first' basis, first asking for additional information before imposing more restrictions on firm actions. I test this in a sample of lending renegotiations and find that, compared to negative covenants, affirmative covenants are renegotiated first in time, first in default and first after technical defaults which are waived. The results suggest that affirmative covenants are a monitoring tool which is frequently adjusted to help banks manage the information asymmetry that develops over the life of a loan.

Cross-Debt Refinancings

With Heather Tookes and Alessio Saretto

Using a novel source of data on the use of proceeds for corporate bonds, we document the surprising fact that close to 50% of new bond issuance is for the purposes of refinancing existing debt. We explore the implications of this fact and document that a significant amount of this refinancing activity concerns what we call `cross-debt` refinancings where a loan is refinanced by a bond or vice versa. Cross-debt refinancings create significant changes in the firm's debt and maturity structure and are significant events in the transition from bank debt financing to bond debt financing. Cross-debt refinancings from loans to bonds are undertaken by firms which are older, have lower credit risk and which have higher cash flows. The opposite is true for refinancings from bonds to loans, suggesting that borrowers in distress not only tap existing lines of credit but use those existing banking relationships to refinance junior debt as well. The results are consistent with the life-cycle of debt hypothesis and provide direct evidence that corporate borrowings experience such a life cycle across debt types.

Following the Money: The effects of SNAP Benefits Changes on the Consumption Habits of Snap Beneficiaries

With John Morley

Links: Github Google Drive

This paper explores a proprietary point-of-sale large dataset of encompassing the individual grocery spending habits of some 43 million Americans — including some 13 million Supplementary Nutritional Assistance Program (SNAP, formerly known as the Food Stamp Program) beneficiaries — during the period of of 2009 to 2014. We attempt to answer the following question: How do changes in SNAP benefits effect the consumption patterns of SNAP beneficiaries? Specifically, we focus on the consumption of sugar sweetened beverages but also consider a spectrum of food categories. In 2009 the American Recovery and Reinvestment Act increased the maximum level of SNAP benefits by 13.6%, set to expire in 2013. Using a differences-in-differences estimation, we find that the 2009 increase in benefits increased consumption of sugar sweetened beverages among SNAP beneficiaries by 5.4% above the baseline (.65% absolute). We also find that the subsequent expiration of SNAP benefits in 2013 saw a respective decrease of -1.6% below the baseline (.18% absolute). Ours is one the first papers to ever to use point-of-sale grocery scanner data to examine spending habits among impoverished Americans and our sample size is an order of magnitude larger than any other paper to date.