HOME Recent Events RMI in the News ALUMNI
  Issue 2 | Archive  

February 2010

Performance Management: Theory, Practice and the Reality

By Jeffrey R. Bohn, Head, Risk Appetite, Standard Chartered Bank, Singapore

More stringent regulation measures may have been in the media limelight recently, but new performance management metrics are key to a sustained recovery.

Given the recent financial crisis, there has been public outcry for more stringent regulation on regulatory capital requirements. This idea, though well-intended, is misguided: regulatory capital does not reflect diversification inherent in most large financial institutions' portfolios. Rather, regulatory capital can often be the result of committee-based compromises or even the result of the arbitrary whims of regulators. Excessive regulation may also lead to regulatory arbitrage where banks shift risks beyond the regulators' purview. Instead, I propose mitigating financial crises can be better handled through prudent use of new performance measures, which I will introduce here. I will also discuss some of the practical issues with implementing these new measures.

What is performance management in the context of managing financial institutions and why is it important? Performance management is the process of measuring performance with respect to the use of a financial institution's resources - especially its capital so that employees can meet organizational goals in an efficient and effective manner. Without proper assessment of the use of capital resources, bank employees will be motivated to take large bets that may not be in the best interests of the bank's shareholders.

Let's look at an example: Consider a large bank that bought an investment-grade tranche in a U.S. RMBS in 2006. These bank executives focused on higher spreads, higher returns on regulatory capital and higher accounting profits for 2006 without considering performance with respect to use of economic capital, which ultimately drives the bank's share performance. These economic realities came back to haunt this bank when RMBS tranche-values were marked down in 2009 and actual defaults spiked; the bank was suddenly faced with sizable losses on its RMBS investments. These RMBS investments were correlated with loans the bank had made to home-builders and real estate companies. Had the performance of the original investment been evaluated in terms of return on economic capital and economic profit, it is unlikely this bank would have made such sizable investments in RMBS and related securities. Since many U.S. banks followed this same strategy without proper recognition of the economic realities, the falling housing market ignited the current financial crisis. It almost goes without saying that banks, which avoided accumulating to much concentrated exposure have seen much better share performance in the past year.

Economic capital (EC) is a key feature of effective performance measurement and management. EC is the amount of risk capital required for a firm to cover its risks, such as market risk, credit risk and operational risk as a going concern over a certain time period with a pre-specified probability. EC allocated to a specific exposure reflects not just that exposure's risk, but also how this exposure's risk is correlated with other risks in the portfolio. EC can be considered the cushion the bank retains to absorb losses in the future. The distribution of possible portfolio losses determines how much EC is supporting a given portfolio. If banks reward short-term profits without recognition of how much EC is used to generate those profits, originators and traders will tend to focus primarily on the volume of transactions without concern for how these exposures will impact the economic performance of the portfolio in the future. EC can be incorporated into a performance measure such as shareholder value-added (SVA1), which reflects the net economic contribution of a transaction by taking net profits (after allocated costs) and further subtracting allocated EC multiplied by the cost of that EC. Using SVA, a bank can better align incentives with activities that increase bank-share value.

Theory vs. reality

While several approaches based on Value-at-Risk (VaR) models can be used to calculate EC, one particular approach known as expected shortfall works best in practice. Early EC models focused on the risk of losses beyond a specific threshold; however, no consideration was given to how much beyond that threshold these losses may go. With this type of EC model, a performance measurement system would have difficulty picking up exposures that have low probabilities of very high losses (possibly high enough to bring down the whole portfolio) well beyond a given threshold. Fortunately, expected shortfall (also known as CVaR) can be used to determine the average loss that occurs when overall portfolio loss exceeds a given threshold. Expected shortfall or CVaR captures the magnitude of losses in the tail region of the loss distribution.

In practice, choosing the right model can be the (relatively) easy part of choosing performance management measures. Building robust performance measures requires addressing data, validation and organizational challenges. While data is now widely available for estimating and testing probability of default (PD) models, other inputs to EC models such as LGD and correlations continue to suffer from lack of sufficient data. Even with sufficient data, default behavior is such that a good model may seem overly punitive more than 80% of the time. Banks may see many years of low losses and then one or two years (e.g. 2008-09) of unusually high losses. Bank executives have a tendency to discount the accuracy of a good model in low-loss years, which can create the seeds of future difficulties if the economic contribution of portfolio exposures is not accurately measured.

The following are a few suggestions for dealing with the practical challenges of implementing a strong performance management system:

1. Focus on minimum data requirements first (e.g. 1-year PDs mapped from an internal rating system, simple LGD estimates, etc.) in order to produce useful reports to develop support for the system early on. Many projects die an early death because nothing useful is produced for months or sometimes years.

2. Use best-practice EC models, which typically rely on simulation approaches. These models should also be subjected to sensitivity analysis to determine which data should be most carefully collected and analyzed.

3. Prioritize organizational acceptance over a model's theoretical elegance. Sometimes it is better to compromise on the details of the calculation such as using accounting-based numbers or EC approximations in order to push through a new performance management framework. That said, the output should generally agree with the more accurate model.

4. Avoid black-box models whether internally built or externally purchased. If you cannot see the detail behind the calculations and you cannot explain the conceptual approach to a quantitatively-informed executive, don't use that model.

5. Recognize that most EC models are still subject to model risk. Aside from more typical data and model issues, liquidity and jump risks are still not well-modeled constituting an ongoing source of model risk.

6. Emphasize model interpretation as an important approach to using output in a performance measurement system. No one should blindly follow model output in decision making.

7. Supplement quantitative input with qualitative analysis as no model can replace the experience and common-sense of risk managers. Allow for strategic priorities to override performance measures. Often, apparent disconnects between a performance measure and a strategic business arises from different time horizons used for the analysis. A business may appear a poor performer over a one-year time horizon, but much stronger over a three-year time horizon.

Performance management should align incentives with building share value and EC models are an important tool for good performance metrics. A good performance management system incorporates a measure like SVA, but still accommodates businesses with strategic importance and other organizational realities.

1 Shareholder value-added (also known as economic profit) = Revenue - Direct Costs - Allocated Costs - Expected Loss - (Allocated Economic Capital * Cost of Capital)
Net Operating Profit after Tax (NOPAT) = Revenue - Direct Costs - Allocated Costs - Expected Loss

Back To Newsletter

Published quarterly by Risk Management Institute, NUS
Editor: Ivy Wang (rmiwy@nus.edu.sg)