with Sascha Steffen
We study the role of Payment-in-Kind (PIK) provisions in private credit markets as a substitute for bank-provided liquidity. Using novel loan-level data from U.S. Business Development Companies (BDCs), we show that borrowers without access to bank credit lines often rely on PIK features to manage liquidity shortfalls. These features allow borrowers to defer interest payments, effectively providing contingent financing during periods of distress or high interest rates. We find that PIK usage strongly predicts future credit deterioration, delinquency, and bankruptcy—especially for borrowers lacking private equity sponsors or where lenders do not hold equity claims. A simple model highlights the agency conflicts inherent in deferred interest and identifies contractual mechanisms that mitigate these risks. At the lender level, we show that increased PIK usage constrains BDCs' portfolio and dividend growth and is associated with tighter bank-imposed covenants. Our findings reveal how nonbank lenders adapt liquidity provision and the associated risks to the financial system.
I study how a firm's organizational structure impacts capital allocation within the booming private debt industry. Focusing on business development companies (BDCs), important nonbank lenders, I document that perpetual-life BDCs lend more bilaterally to smaller and riskier borrowers while finite-life BDCs participate in larger deals with multiple lenders. This two-sided endogenous matching can arise from a search-theoretic model, where the lender's perpetual-life incentivizes both counterparties to engage more in bilateral lending. With perpetual-life lenders, the borrower trades off a lower loan rollover risk for a higher coupon rate. Empirically this effect is 37 basis points based on an instrumental variable estimate. Stronger lending relationships also provide borrowers more stable credit and support employment growth.
with Basile Dubois
We examine the effects of quantitative easing (QE) on bank lending in the Eurozone. QE has substantially increased central bank reserves held by commercial banks and raised the volume of short-term wholesale deposits, which made bank funding less stable. Basel III regulation complicates how large volumes of excess reserves and short-term wholesale deposits influence bank lending. We develop a structural model incorporating imperfect competition in credit and deposit markets and regulatory costs that escalate as banks approach minimum requirements. This framework allows us to quantify how excess reserves contribute to regulatory costs. In France, QE increased the marginal cost of long-term lending by 16 basis points in Q4 2021. Counterfactual analysis indicates that if the aggregate reserves had been maintained at their 2019 level of 2 trillion euros instead of 4 trillion euros in Q4 2021, aggregate bank lending would have been approximately 5% higher.
with Petri Jylhä
First draft March 2020
This draft July 2024
The market for auction rate securities collapsed in February 2008, significantly increasing some closed-end funds' cost of borrowing. The affected funds reacted by moving to a leverage-constrained funding structure. We present a model that explains this fund behavior and then use the event as a quasi-natural experiment to study empirically how leverage constraints affect investors' portfolio choices. Consistent with our model's predictions, we show that becoming leverage-constrained results in an increased appetite for systematic risk: in the months following the shock, the affected funds increased their portfolio betas by buying significantly more high-beta stocks than their unaffected peers.
This draft May 2024
Household wealth effects are heterogeneous between asset classes due to concentrated asset ownership between household groups with plausibly different marginal propensities to consume. However, wealth effect estimates between asset classes across studies are hard to compare. With tax data I construct a new county-level data set on U.S. household asset and debt positions and estimate wealth effects on payroll and employment simultaneously for five important asset classes. Using Bartik instruments for identification, I find large dynamic wealth effects from local house price shocks and mortgage rate shocks, and small effects from stock market wealth shocks. A model with heterogeneous agents motivates the empirical analysis.
This draft July 2024
I document that in several countries the ratio of equity wealth to other (non-equity) wealth (EO-ratio) has moved in low-frequency cycles between 1873 to 2020. First, I find that a high level of EO-ratio strongly and robustly predicts low future stock market returns and vice versa, and using two novel present value decompositions I show that both differences in valuation- and payouts contribute to these cycles---but differently in the U.S. versus abroad. Second, I build a quantitative macro-finance model with limited participation, redistributive income shocks, and inflation and show that the model and these factors help explain these cycles.
with David Chambers, Elroy Dimson and Antti Ilmanen (Annual Review of Financial Economics - 2024)
The literature on long-run asset returns has continued to grow steadily, particularly since the start of the new millennium. We survey this expanding body of evidence on historical return premia across the major asset classes – stocks, bonds, and real assets – over the very long-run. In addition, we discuss the benefits and pitfalls of these long-run datasets and make suggestions on best practice in compiling and using such data. We report the magnitude of these risk premia over the current and previous two centuries, and we compare estimates from alternative data compilers. We conclude by proposing some promising directions for future research.
with 300+ coauthors (Journal of Finance - 2024)
In statistics, samples are drawn from a population in a data-generating process (DGP). Standard errors measure the uncertainty in sample estimates of population parameters. In science, evidence is generated to test hypotheses in an evidence-generating process (EGP). We claim that EGP variation across researchers adds uncertainty: non-standard errors. To study them, we let 164 teams test six hypotheses on the same sample. We find that non-standard errors are sizeable, on par with standard errors. Their size (i) co-varies only weakly with team merits, reproducibility, or peer rating, (ii) declines significantly after peer-feedback, and (iii) is underestimated by participants.