A Loan in the Wild: A Study of Bank Lending Under the Troubled Asset Relief Program

Abstract

In response to the 2008 financial crisis, Congress gave Treasury Secretary Paulson $700 billion to purchase assets from American banks. The goal of the Troubled Asset Relief Program was to encourage economic recovery by stimulating bank lending. Paulson used the money to provide direct equity injections to banks. Previous studies have concluded that TARP did not lead to a meaningful increase in bank lending. However, these papers relied on a method of matched pairing to compare the lending patterns of TARP banks to the lending patterns of non-TARP banks. Other research has concluded that TARP banks are fundamentally different than non- TARP banks in ways other than just their TARP participation status. This limits the scope of research that relies on comparisons between the two groups and calls for an intragroup analysis of TARP associated lending. This paper exploits differences in lending patterns among publicly traded TARP participants to determine whether government capital injections increased bank lending. An ordinary least squares regression is employed to estimate the impact of the main variable (bailout size, adjusted by bank size) on the dependent variable (change in levels of pre- and post-TARP lending, adjusted by bank size). A collection of control variables allows for the identification of characteristics that made TARP banks more likely to have increased lending. A set of control variables that measure pre-TARP bank financial data is used to evaluate whether banks exhibited a character-driven or a situation-driven response to TARP. Under the theory of character-driven responses, banks that are aggressive pre- TARP lenders will use relatively more government-provided capital for lending. However, under a situation-driven response, banks that are more aggressive prior to TARP will encounter financial hardship and will increase lending less after receiving TARP. Evidence drawn from measures of bank capital adequacy, funding stability, and profitability supports the character-driven response to capital infusions. The model estimates a point value of 5.53 for the BailoutSize coefficient, which suggests that every dollar of government equity financed as many as $5.53 of new loans. However, that variable has a large standard error and only very weakly supports the notion that capital was leveraged into lending. Results from the variable tracking CEO departures determined a negative correlation between CEO departure and post-TARP lending. Since this study controls for bank performance, I theorize that a difference in the managerial style between incumbent CEOs and newly minted chief executives explains the divergence in lending patterns. A variable accounting for the condition in which CEOs depart provides evidence for that theory. Among banks that experienced executive turnover, those with forced CEO departures saw a larger decrease in TARP-associated lending than did those with CEO retirements. CEOs who replace a fired successor may be under increased pressure to curb risk. Or, bank directors might have removed CEOs specifically to replace them with a more risk-averse manager. Results also show that areas with a smaller rise in unemployment increased lending more than did banks in areas more stricken by recession and that smaller banks increased lending relatively more than did larger banks.