We study bank responses to the Paycheck Protection Program (PPP) and its effects on lender balance sheets and profitability. To address the endogeneity between bank decisions and balance sheet effects, we develop a Bayesian joint model that examines the decision to participate, the intensity of participation, and ultimate balance sheet outcomes. Overall, lenders were driven by risk-aversion and funding capacity rather than profitability in their decision to participate and the intensity of their participation. Indeed, with greater participation intensity, banks experienced sizable growth in their loan portfolios but a decline in their interest margins. In counterfactual exercises, we show that the PPP offset a large potential contraction in business lending, and that bank margins would have fallen even more precipitously if lenders had not participated in the program. Although the PPP was intended as a credit support program for small firms, the program indirectly supported the margins of banks that channeled these loans.

This paper examines an episode when policy response to a financial crisis effectively incentivized financial institutions to reallocate their portfolios toward safe assets. Following a shift to a regime of enhanced regulation and scaled-down public assistance during the Savings and Loans (S&L) crisis in 1989, this paper finds that S&Ls with a high probability of failure increased their composition of safe assets relative to S&Ls with a low probability of failure. The findings also show a shift to safe assets among stock S&Ls relative to mutual S&Ls, thereby providing evidence of risk-shifting from equity-holders to debt-holders of stock S&Ls prior to the regulatory reforms. These findings show that credible signals to shareholders around government assistance will be crucial for the policies aimed at reducing moral hazard (such as the Orderly Liquidation Authority under Title II of the Dodd-Frank Act) to succeed. This paper identifies the effect of the policy change by developing a new Bayesian estimation method for causal studies.

No abstract

No abstract

No abstract