Financial crises are often followed by debates about whether bankers’ incentives helped create distress in the financial sector.  We contribute to this debate by documenting the extent to which bankers’ pay contains prudence-related targets, the association between those targets and other incentives, and how the targets affect future bank risk-taking. We may be the first to posit that bankers are paid to safeguard the safety of their banks, and we examine whether these incentives are effective.

We refer to prudence-based compensation targets as “pay for prudence” (PfP). PfP rewards managers for outcomes that lower credit risk, which is a central objective of bank supervision. For example, managers may be paid bonuses when they achieve a satisfactory regulatory rating, reduce non-performing loans, minimize loan losses, or maintain high credit quality.

As in other industries, a bank’s shareholders and board of directors provide incentives for management to maximize firm value. While bank shareholders may prefer risky actions that are likely correlated with increases in shareholder value, they must also gauge the likelihood of regulatory intervention. These conflicting preferences create a trade-off between risky (but potentially value-maximizing) projects and regulatory intervention.

We construct a novel dataset of PfP using the compensation disclosures of all publicly traded bank holding companies (BHCs) in Def 14A filings. We search within compensation-related discussions and identify contract terms, performance vesting provisions, and bonuses that are contingent on prudence-related terms, using a series of regular expressions based on a library of prudence-related terms from the Federal Reserve Board’s Commercial Bank Examination Manual (CBEM). We create two measures of PfP designed to capture the extensive margin of PfP use. First, a “concrete” PfP discussion contains actual targets and the levels managers must achieve to receive incentive compensation. Second, a “detailed” PfP discussion discloses the metrics considered but may not disclose an actual target. In addition to these two proxies, we document instances of “vague” PfP discussions in which the bank simply lists a PfP measure among several other financial metrics considered without further clarification. These vague PfP targets may only provide weak incentives or signals. Therefore, we focus the majority of our analysis on concrete and detailed PfP targets.

Using our novel dataset, we document that PfP has existed since the beginning of our sample period in 2001.​ Detailed PfP and concrete PfP, though initially rare, reached 10 percent and 8 percent utilization across all banks by 2005 and then at least doubled over the second half of that decade. Thereafter, detailed PfP and concrete PfP both remained relatively steady at approximately 20 percent of banks through 2017. Vague PfP was significantly more common early in the sample period, reaching a peak of 67 percent in 2004, but fell precipitously with only 38 percent of banks disclosing vague PfP by the end of our sample period. Thus, we document that while the aggregate use of PfP decreased over time, the frequency of both detailed and concrete PfP goals actually increased.

This descriptive analysis yields two important insights. First, PfP use substantially predates the financial crisis. Second, the sea change in PfP specificity, from vague to concrete and detailed, aligns with the implementation (2006) and enhancement (2010) of the Compensation Discussion and Analysis (CD&A) section in Def 14A filings and the financial crisis of 2007–2009. This trend also persists to the end of our sample period, even though the PfP-related requirements of the Troubled Asset Relief Program have long expired, and the compensation-related provisions of Dodd-Frank remain unimplemented.

Having documented the prevalence of PfP, we studied the association between PfP use and equity-based compensation incentives for risk-taking. Prior research suggests that equity-based incentives align managerial actions with shareholders’ objectives, but such alignment may induce the types of risk-taking that regulators find most concerning. ​Coles et al. (2006)​ and ​Shue and Townsend (2017)​ link these incentives to increased leverage. ​Armstrong and Vashishtha (2012)​ show that equity risk-taking incentives lead managers to pursue strategies that expose their firms to systematic risk, which they can hedge, and not idiosyncratic risk, which they cannot hedge, and ​Armstrong et al. (2022)​ show that these incentives are associated with an increase in systemic risk. In other words, equity-based incentives for risk-taking can lead banks’ managers to increase leverage and systemic risk, exactly the types of risks that concern regulators most.

We expect that bank boards design compensation contracts to consider both shareholders’ desire to encourage value-enhancing risk while discouraging imprudent risks and the potential for costly regulatory intervention. Consistent with this, we find a positive association between the use of PfP goals and managers’ equity compensation. These results are not subsumed by controls for performance, size, or banks’ focus on traditional banking activities. We also find that the association between equity compensation and PfP use systematically varies with measures of bank risk. Specifically, the association between equity compensation and PfP use is increasing in return variance and bank leverage. This suggests that banks put in place guard rails to complement equity incentives when insolvency risk is higher.

We next study the relation between PfP and observable measures of bank risk. If PfP fosters prudent behavior, PfP will be negatively associated with bank risk.​ Alternatively, PfP may be window dressing intended to distract regulators or other stakeholders. Under this view, PfP does not deter risk-taking and may even be associated with greater risk-taking. We find that PfP is associated with lower risk-taking. Specifically, both detailed and concrete PfP are associated with lower tail risk and fewer bad loans. We also find that the profitability of PfP banks is not significantly different from non-PfP banks on average.

We extend this analysis by examining cross-sectional variation in the association between PfP and bank performance. Our principal-agent framework predicts that PfP may be more beneficial when external stakeholders have stronger incentives to intervene in firm operations. For example, when banks are in financial distress, PfP may better align managerial actions with the preferences of stakeholders who prefer stability. To test this, we measure three bank characteristics related to instability: size, regulatory capital, and core deposits.​ We find that the positive association between PfP and performance is stronger among small banks, banks with low regulatory capital, and banks that are more reliant on non-core deposits.

Next, we examine whether PfP is associated with regulatory intervention. If PfP encourages prudent practices or signals banks’ commitment to prudence, then we expect PfP banks to be downgraded less frequently by regulators. Alternatively, banks may use PfP to appear prudent when in fact they are not. In this case, we expect that banks that use PfP are more likely to be downgraded by regulators. We find a negative association between PfP and the likelihood of regulatory downgrades. Moreover, this result is stronger among small banks and poorly capitalized banks. These results reinforce the idea that PfP is associated with more prudent business practices.

Finally, we examine bank operations and find that PfP is associated with more diverse loan portfolios and that this diversification is partially achieved by reducing exposure to real-estate lending. Increased diversification through decreased specialization in real estate is likely perceived positively by bank regulators tasked with ensuring safety and soundness (OCC, 2020​), as prior research suggests a link between real estate lending and heightened risk-taking (​Costello et al., 2019​; ​Granja et al., 2017​). However, shareholders likely prefer banks to increase profitability by specializing in specific loan categories (​Stomper, 2006​; ​Rossi et al., 2009​; ​Berger et al., 2010​; ​Tabak et al., 2011​), including lucrative real estate lending.

Collectively, our findings suggest that PfP complements risk-taking incentives, and that PfP is associated with lower bank risk. From the perspective of shareholders, the costs of PfP include limiting banks’ appetite for pursuing risky but positive net present value projects (such as real estate lending) and foregoing the benefits of specialization. Importantly, the voluntary nature of PfP adoption does not allow us to isolate the costs and benefits of PfP. In the absence of mandatory PfP adoption, we acknowledge that the banks that adopt PfP are those for which PfP provides net benefits.

The recent failures of Silicon Valley Bank, Signature Bank, and First Republic Bank highlight the importance of our study to researchers and regulators. PfP is conspicuously absent from the compensation discussions in Silicon Valley Bank’s proxy statements, and while First Republic Bank’s proxy statements include concrete PfP targets, Signature Bank’s proxy statements only provide vague discussion of PfP. These anecdotes suggests that PfP is not a panacea and raise important questions for future research about the role of boards and compensation practices in mitigating the risks associated with bank failures.

This post comes to us from Salman Arif at the University of Minnesota’s Carlson School of Management, John Donovan at the University of Notre Dame, Yadav Gopalan at Indiana University’s Kelley School of Business, and Arthur Morris at the Hong Kong University of Science and Technology. It is based on their recent paper, “Pay for Prudence,” available here.