Monday, August 27, 2012

Incentivizing Calculated Risk-Taking: an Experiment with Commercial Bank Loan Officers. By Martin Kanz and Leora Klapper

Incentivizing Calculated Risk-Taking: an Experiment with Commercial Bank Loan Officers. By Martin Kanz and Leora Klapper
Mon, Aug 27, 2012    08:42am

In the aftermath of the global financial crisis, there has been much criticism of compensation practices at banks. Although much of this debate has focused on executive compensation (see the recent debate on this blog), there is a growing recognition that non-equity incentives for loan officers and other employees at the lower tiers of a bank’s corporate hierarchy may share some of the blame — volume incentives for mortgage brokers in the United States that rewarded high-risk lending at wildly unsustainable terms are a particularly striking case in point.

The view that excessive risk-taking in the run-up to the crisis had its roots in flawed incentives at all levels of financial institutions — not just at the top — has made inroads in policy circles, and has been reflected in efforts to regulate how banks can pay their loan officers. Well-intentioned as these efforts may be, they mask the fact that providing performance incentives in lending is, in fact, a very difficult problem. Assessing a borrower’s creditworthiness requires a complex tradeoff between risk and return; it contains an inherent element of deferred compensation and requires the interpretation of a noisy signal about an applicant’s actual creditworthiness. Whether and how performance incentives work in this setting is unclear: the limited evidence that exists about the impact of performance pay on employee behavior comes from the labor economics literature and suggests that — even in simple production tasks — the behavioral response to incentives tends to be much more complex than a simple mapping from stronger incentives to greater effort and performance.

So how does “pay-for-performance” affect the risk-appetite and lending decisions of loan officers? In a recent paper, coauthored with Shawn Cole of Harvard Business School we designed a field experiment with real-life loan officers to examine the impact of performance incentives on loan officer behavior. Working with a number of leading commercial banks in India, we recruited more than 200 loan officers with an average of more than ten years of experience in banking and brought them to a behavioral economics lab. In the lab, participants were asked to evaluate a set of loan applications under different, exogenously assigned incentives. This cross-over between an actual field experiment and a controlled lab setting allowed us to study risk-taking behavior using a real life population of highly experienced loan officers, while being able to get detailed measurements of risk-assessment and risk-taking behavior — the kind of data that would usually only be available from a lab experiment.

We deliberately set up our experiment in an informationally challenging emerging credit market — the Indian market for unsecured small enterprise loans. Borrowers in this market typically lack reliable credit scores and an established track record of formal sector borrowing. This generally rules out the use of predictive credit scoring and other advance loan approval technologies, making banks particularly reliant on the risk-assessment of their frontline employees. The credit files that our loan officers evaluated in the experiment consisted of actual loan applications from small enterprises applying for their first formal-sector loan. Each loan was matched with ten months of repayment history from the lender’s proprietary database (not surprisingly, more than 90% of defaults occur in the first three months of a loan’s tenure). This allowed us to compare the actual outcome of the loan with the loan officer’s decision and risk assessment in the experiment and to offer incentive payments based on the profitability of lending decisions loan officers took in the lab.

The reassuring news is that basic incentives seem to work quite well in lending. We find that pay for performance (incentives that reward profitable lending and penalize default) indeed induces loan officers to exert much greater effort in reviewing the information that is presented to them. This is all well, but the real question is whether this translates into improved lending decisions. One common concern with performance pay in lending is that stronger incentives may indeed make loan officers much more conscientious, so conscientious in fact that they may shy away from risks that would be profitable from the viewpoint of the bank and simply stop lending. In our experiment, we find this not to be the case: when loan officers faced high-powered incentives, the probability that they would approve a non-performing loan was reduced by 11% while overall lending went down by only 3.6%. In other words, more stringent incentive schemes actually made loan officers better at identifying and eliminating bad credits from the pool of loan applicants. Profits per loan increased by up to 4% over the median loan size and by more than 40% compared with the case when loan officers faced volume incentives.

These strong results highlighting the negative impact of volume incentives are in line with much recent evidence using observational data (Agarwal and Ben-David 2012; Berg, Puri, and Rocholl 2012). So is pay-for performance the solution to all of a bank’s internal agency problems? Unfortunately not. In an additional set of experiments, we varied the time horizon of the loan officer’s compensation contract — an important second dimension of the incentive scheme over which a bank typically has control. Interestingly, our results show that performance incentives quickly lose their bite as they are deferred even by a couple months. Given that in real life performance pay typically occurs in the form of a quarterly or annual bonus, this casts some doubt on the wisdom of trying to fix agency problems within financial institutions with monetary incentives alone. Interestingly, however, deferred compensation also makes permissive incentive schemes less tempting and can attenuate many of the negative effects of volume incentives. Some direct advice that comes out of this finding is that if a bank finds it necessary to provide volume incentives, it can limit the potential damage through deferred compensation.

Perhaps most interestingly, the results from our experiment also show that incentives affect not only actual lending decisions, they also distort loan officers’ subjective assessment of credit risk. Put simply, we find that when participants faced incentives that emphasize lending volume over loan quality, they started viewing their clients’ creditworthiness through rose-colored glasses. They inflated internal risk ratings — which were neither seen by any supervisor nor tied to incentives — by up to .3 standard deviations for the same loan. This finding resonates with the psychological concept of “cognitive dissonance” (Akerlof and Dickens 1982) and is in line with behavioral economics explanations that have tried to make sense of seemingly irrational behavior in sub-prime lending prior to the crisis, which are nicely summarized in a recent article by Nicholas Berberis (2012) from the Yale School of Management.
What are we to take away from these results? The question of how to better align private incentives with public interest is a major unresolved policy question that has arisen from the global financial crisis. Our experiments provide some of the first rigorous evidence on the link between performance pay and behavior among loan originators, which we hope will be a first step that can help tackle this important issue from the angle of corporate governance —– with the ultimate aim of making compensation policy a more effective component of a bank’s risk management mechanisms. Much work has recently gotten underway in this exciting research agenda, but it is clear that much more evidence is needed to translate these findings into meaningful policy prescriptions. To contribute to this agenda, we are currently working on a number of follow-up experiments to more fully understand the behavioral and psychological implications of the problem of incentives and individual risk-taking. Stay tuned.


Agarwal, Sumit, and Itzhak Ben-David. 2012. “Do Loan Officer Incentives Lead to Lax Lending Standards?” Ohio State University, Fisher College of Business. Working Paper WP-2012-7.

Agarwal, Sumit, and Faye H. Wang. 2009. “Perverse Incentives at the Banks? Evidence from a Natural Experiment.” Federal Reserve Bank of Chicago. Working Paper WP-09-08.

Akerlof, George A., and William T. Dickens. 1982. “The Economic Consequences of Cognitive Dissonance.” American Economic Review 72 (3):307–19.

Baker, George, Michael Jensen, and Kevin Murphy. 1988. “Compensation and Incentives: Practice vs. Theory.” Journal of Finance 43 (3):593–616.

Bandiera, Oriana, Iwan Barankay, and Imran Rasul. 2007. “Incentives for Managers and Inequality among Workers: Evidence from a Firm-Level Experiment.” Quarterly Journal of Economics 122 (2):729–73.

_____. 2009. “Social Connections and Incentives in the Workplace: Evidence from Personnel Data.” Econometrica 77 (4):1047–94.

_____ Team Incentives: Evidence from a Firm Level Experiment. Journal of the European Economic Association, forthcoming.

Barberis, Nicholas. 2012. “Psychology and the Financial Crisis of 2007-2008.” In Financial Innovation and the Crisis, edited by M. Haliassos. Cambridge, MA: MIT Press.

Berg, Tobias, Manju Puri, and Jorg Rocholl. 2012. “Loan Officer Incentives and the Limits of Hard Information.” Duke University Fuqua School of Business Working Paper.

Measuring Systemic Risk-Adjusted Liquidity (SRL) - A Model Approach. By Andreas Jobst

Measuring Systemic Risk-Adjusted Liquidity (SRL) - A Model Approach. By Andreas Jobst
IMF Working Paper No. 12/209
Aug 2012

Summary: Little progress has been made so far in addressing—in a comprehensive way—the externalities caused by impact of the interconnectedness within institutions and markets on funding and market liquidity risk within financial systems. The Systemic Risk-adjusted Liquidity (SRL) model combines option pricing with market information and balance sheet data to generate a probabilistic measure of the frequency and severity of multiple entities experiencing a joint liquidity event. It links a firm’s maturity mismatch between assets and liabilities impacting the stability of its funding with those characteristics of other firms, subject to individual changes in risk profiles and common changes in market conditions. This approach can then be used (i) to quantify an individual institution’s time-varying contribution to system-wide liquidity shortfalls and (ii) to price liquidity risk within a macroprudential framework that, if used to motivate a capital charge or insurance premia, provides incentives for liquidity managers to internalize the systemic risk of their decisions. The model can also accommodate a stress testing approach for institution-specific and/or general funding shocks that generate estimates of systemic liquidity risk (and associated charges) under adverse scenarios.


A defining characteristic of the recent financial crisis was the simultaneous and widespread dislocation in funding markets, which can adversely affect financial stability in absence of suitable liquidity risk management and policy responses. In particular, banks’ common asset exposures and their increased reliance on short-term wholesale funding in tandem with high leverage levels helped propagate rising counterparty risk due to greater interdependence within the financial system. The implications from liquidity risk management decisions made by some institutions spilled over to other markets and other institutions, contributing to others’ losses, amplifying solvency concerns, and exacerbating overall liquidity stress as a result of these negative dynamics. Thus, private sector liquidity (as opposed to monetary liquidity), which is created largely through banks and other financial institutions via bilateral arrangements and organized trading venues, is invariably influenced by common channels of market pricing that can amplify cyclical movements in system-wide financial conditions with the potential of negative externalities resulting from individual actions (CGFS, 2011).

The opportunity cost of holding liquidity is invariably cyclical, resulting in a notorious underpricing of liquidity risk, which tends to perpetuate a disregard for the potential inability of markets to sustain sufficient liquidity transformation under stress. Banks have an incentive to minimize liquidity (and mitigate the opportunity cost of holding excess liquidity in lieu of return-generating assets) in anticipation that central banks will almost certainly intervene in times of stress as lenders-of-last-resort. Even without central bank support, liquidity risk is most expensive when it is needed most while generating little if any additional return in good times. While central banks can halt a deterioration of funding conditions in order to maintain the efficient operation of funding markets (see Figure 1), prevent financial firms from failing, and, thus, limit the impact of liquidity shortfalls on the real economy, their implicit subsidization of bank funding accentuates the magnitude of liquidity risks under stress. Central bank measures during the credit crisis have further reinforced this perception of contingent liquidity support, giving financial institutions an incentive to hold less liquidity than needed (IMF, 2010a).

Current systemic risk analysis—as a fundamental pillar of macroprudential surveillance and policy—is mostly focused on solvency conditions. Disruptions to the flow of financial services become systemic if there is the potential of financial instability to trigger serious negative spillovers to the real economy. Macroprudential policy aims to limit, mitigate or reduce systemic risk, thereby minimizing the incidence and impact of disruptions in the provision of key financial services that can have adverse consequences for the real economy (and broader implications for economic growth). Substantial work is underway to develop enhanced analytical tools that can help to identify and measure systemic risk in a forward-looking way, and, thus, support improved policy judgments. While systemic solvency risk has already entered the prudential debate in the form of additional capital rules that apply to systemically important financial institutions (SIFIs), little progress has been made so far in addressing systemic liquidity risk.

In contrast, proposals aimed at measuring and regulating systemic liquidity risk caused by the interconnectedness across financial institutions and financial markets have been few and far between. Systemic liquidity risk is associated with the possibility that maturity transformation in systemically important institutions and markets is disrupted by common shocks that overwhelm the capacity to fulfill all planned payment obligations as and when they come due. For instance, multiple institutions may face simultaneous difficulties in rolling over their short-term debts or in obtaining new short-term funding (much less longterm funding). However, progress in developing a systemic liquidity risk framework have been hampered by the rarity of system-wide liquidity risk events, the multiplicity of interactions between institutions and funding markets, and the conceptual challenges in modeling liquidity conditions affecting institutions and transactions separately or jointly. The policy objective of such efforts would be to minimize the possibility of systemic risk from liquidity disruptions that necessitate costly public sector support. While a financial institution’s failure can cause an impairment of all or parts of the financial system, firms are not charged for the possibility that their risk-taking affects the operation of the financial system as a whole. In fact, individual actions might cause losses elsewhere in the system through direct credit exposures and financial guarantees, forced asset sales, and greater uncertainty regarding mutual exposures (possibly in combination with greater risk aversion of investors), which increases the cost of funding for all financial institutions. These “negative externalities” impose costs to the system, which increases the greater the importance of a single institution to the system (“too-important-to-fail”) and the higher the level of asymmetric information as coordination failures accentuate the impact of common shocks. Thus, more stringent prudential liquidity requirements, much like higher capital levels, might be beneficial ex ante by creating incentives of shareholders to limit excessive risk-taking, which would otherwise increase the potential loss in case of failure (Jensen and Meckling, 1976; Holmstrom and Tirole, 1997). However, certain liquidity standards might also encourage greater concentrations in assets that receive a more favorable regulatory treatment based on their liquidity characteristics during normal times (which remains to be tested during times of stress).

A number of prudential reforms and initiatives are underway to address shortcomings in financial institutions’ liquidity practices, which have resulted in more stringent supervisory liquidity requirements. Under the post-crisis revisions of the existing Basel Accord, known as Basel III, the Basel Committee on Banking Supervision (BCBS, 2010a, 2010b and 2009) has proposed two quantitative liquidity standards to be applied at a global level and published a qualitative guidance to strengthen liquidity risk management practices in banks. Under this proposal, individual banks are expected to maintain a stable funding structure, reduce maturity transformation, and hold a sufficient stock of assets that should be available to meet its funding needs in times of stress—as measured by two standardized ratios:

*  Liquidity Coverage Ratio (LCR). This ratio is intended to promote short-term resilience to potential liquidity disruptions by requiring banks to hold sufficient highquality liquid assets to withstand the run-off of liabilities over a stressed 30-day scenario specified by supervisors. More specifically, “the LCR numerator consists of a stock of unencumbered, high-quality liquid assets that must be available to cover any net [cash] outflow, while the denominator is comprised of cash outflows less cash inflows (subject to a cap at 75 [percent] of total outflows) that are expected to occur in a severe stress scenario (BCBS, 2011 and 2012b).”
*  Net Stable Funding Ratio (NSFR). This structural ratio limits the stock of unstable funding by encouraging longer term borrowing in order to restrict liquidity mismatches from excessive maturity transformation. It requires banks to establish a stable funding profile over the short term (i.e., the use of stable (long-term and/or stress-resilient) sources to continuously fund short-term cash flow obligations that arise from lending and investment activities). The NSFR reflects the proportion of long-term assets that are funded by stable sources of funding with maturities of more than one year (except deposits), which includes customer deposits, long-term wholesale funding, and equity (but excludes short-term funding). A value of this ratio of less than 100 percent indicates a shortfall in stable funding based on “the difference between balance sheet positions after the application of available stable funding factors and the application of required stable funding factors for banks where the former is less than the latter (BCBS, 2011 and 2012b).”

However, these prudential measures do not directly targeting system-wide implications.  The current approach assumes that sufficient institutional liquidity would reduce the likelihood of knock-on effects on solvency conditions in distress situations and complement the risk absorption role of capital—but without considering system-wide effects. Larger liquidity buffers at each bank should lower the risk that multiple institutions will simultaneously face liquidity shortfalls, which would ensure that central banks are asked to perform only as lenders of last resort—and not as lenders of first resort. However, this rationale underpinning the Basel liquidity standards ignores the impact of the interconnectedness of various institutions and their diverse funding structures across a host of financial markets and jurisdictions on the probability of such simultaneous shortfalls. Moreover, in light of the protracted adoption of both the LCR and the NSFR (whose implementation is envisaged in 2015 and 2018, respectively) and the associated risk of undermining timely adjustment of industry standards, Perotti (2012) argues for strong transitional tools in the form of “prudential risk surcharges.” These would be imposed on the gap between current liquidity positions of banks and the envisaged minimum liquid standards at a level high enough to compensate for and discourage the creation of systemic risk in order to ensure early adoption of safer standards while offering sufficient flexibility of banks to chart their own path towards compliance.

An effective macroprudential approach that targets systemic liquidity risk presupposes the use of objective and meaningful measures that can be applied in a consistent and transparent fashion (and the attendant design of appropriate policy instruments).  Ideally, any such methodology would need to allow for extensive back-testing and should benefit from straightforward application (and avoid complex modeling (or stress-testing)).  While it should not be too data intensive to compute and implement, enough data would need to be collected to ensure the greatest possible coverage of financial intermediaries in order to accommodate different financial sector characteristics and supervisory regimes across national boundaries. In addition, the underlying measure of systemic risk should be time-varying, and, if possible, it should offset the procyclical tendencies of liquidity risk and account for changes to an institution’s risk contribution, which might not necessarily follow cyclical patterns. Finally, it would also motivate a risk-adjusted pricing scheme so that institutions that contribute to systemic liquidity risk are assigned a proportionately higher charge (while the opposite would hold true for firms that help absorb system-wide shocks from sudden increases in liquidity risk).

In this regard, several proposals are currently under discussion (see Table 1), including the internalization of public sector cost of liquidity risk via insurance schemes (Goodhart, 2009; Gorton and Metrick, 2009; Perotti and Suarez, 2009 and 2011), capital charges (Brunnermeier and Pedersen, 2009), taxation (Acharya and others, 2010a and 2010b), investment requirements (Cao and Illing, 2009; Farhi and others, 2009), as well as arrangements aimed at mitigating the system-wide effects from the fire sale liquidation of assets in via collateral haircuts (Valderrama, 2010) and modifications of resolution regimes (Roe, 2009; Acharya and Oncu, 2010). In particular, Gorton (2009) advocates a systemic liquidity risk insurance guarantee fee that explicitly recognizes the public sector cost of supporting secured funding markets if fragility were to materialize. Roe (2009) argues that the internalization of such cost would ideally be achieved by exposing the lenders to credit risk of the counterparty (and not just that of the collateral) by disallowing unrestricted access to collateral even in case of default of the counterparty. In this way, lenders would exercise greater effort in discriminating ex ante between safer and riskier borrowers. Such incentives could be supported by time-varying increase in liquidity requirements, which also curb credit expansion fueled by short-term and volatile wholesale funding and reduce dangerous reliance on such funding (Jácome and Nier, 2012).

In this paper, we propose a structural approach—the systemic risk-adjusted liquidity (SRL) model—for the structural assessment and stress testing of systemic liquidity risk.  Although macroprudential surveillance relies primarily on prudential regulation and supervision, calibrated and used to limit systemic risk, additional measures and instruments are needed to directly address systemic liquidity risk. This paper underscores why more needs to be done to develop macroprudential techniques to measure and mitigate such risks arising from individual or collective financial arrangements—both institutional and marketbased—that could either lead directly to system-wide distress of institutions and/or significantly amplify its consequences. The SRL model complements the current Basel III liquidity framework by extending the prudential assessment of stable funding (based on the NSFR) to a system-wide approach, which can help substantiate different types of macroprudential tools, such as a capital surcharge, a fee, a tax, or an insurance premium that can be used to price contingent liquidity access. 

The SRL model quantifies how the size and interconnectedness of individual institutions (with varying degrees of leverage and maturity mismatches defining their risk profile) can create short-term liquidity risk on a system-wide level and under distress conditions. The model combines quantity-based indicators of institution-specific funding liquidity (conditional on maturity mismatches and leverage), while adverse shocks to various market rates are used to alter the price-based measures of monetary and funding liquidity that, in turn, form the stress scenarios for systemic liquidity risk within the model (see Table 2 and Box 2). In this way, the SRL model fosters a better understanding of institutional vulnerabilities to the liquidity cycle and related build-ups of risks based on market rates that are available at high frequencies and which lend themselves to the identification of periods of heightened systemic liquidity risk (CGFS, 2011).

This approach forms the basis for a possible capital charge or an insurance premium—a pre-payment for the contingent (official) liquidity support that financial institutions eventually receive in times of joint distress—by identifying and measuring ways in which they contribute to aggregate risk over the short-term. Such a liquidity charge should reflect the marginal contribution of short-term funding decisions by institutions to the generation of systemic risk from the simultaneous realization of liquidity shortfalls. Proper pricing of the opportunity cost of holding insufficient liquidity—especially for very adverse funding situations—would help lower the scale of contingent liquidity support from the public sector (or collective burden sharing mechanisms). The charge needs to be risk-based, should be increasing in a common maturity mismatch of assets and liabilities, and would be applicable to all institutions with access to safety net guarantees. Since liquidity runs are present in the escalating phase of all systemic crises, our focus is on short-term wholesale liabilities, properly weighted by the bank's maturity mismatch.