by Dr. Ronald L. Olson, Chairman, 1998 | |

Introduction | |

In my previous article, “Without Risk Benchmarks, Risk Measures Don’t Make Sense” (hereafter, referred to as the “Benchmarks” article), I suggested that the industry needs benchmarks or standards for measuring interest-rate risk; that any one of several different approaches can be acceptable; and that unique assumptions will customize standardized measurements to individual institutions. Despite this range of choice, industry-wide agreement on definitions and methods is important. In this article, I show how standardized measurements make it possible to look at interest-rate risk in an organized way.To manage interest-rate risk, banks must first measure interest-rate risk. A standard measurement model gives bankers the ability to learn one accepted way to measure risk. It gives the ability to communicate with others. It gives the ability to measure interest-rate risk in a way that is consistent with measurements of other financial risks. It gives each individual institution a perspective of trends over time. It gives the ability to compare the various ways bankers manage interest-rate risk. And it makes it possible to combine data to serve as industry-wide benchmarks. The data presented here for Illustration Bank A is for December 31, 1996. The trend data for Illustration Bank B, Illustration Bank C, and for industry-wide peer groups are for the six quarters December 31, 1995, through March 31, 1997. The exhibits in this article are summarized from more detailed data available in our quarterly Industry Reports. All of the data presented here and on the web page are the results of a standard measurement model applied consistently over time. This model, based on call report information, is processed centrally using standard inputs (unique data and assumptions in a uniform format) and rigorous audit checks to provide measurement consistency from one institution to the next. The model produces measurements of financial risk, trend analyses of historical performance, financial forecasts, and peer group statistics. Report output from the model provides interest-rate-risk measures; Camels rating ratio information; risk-measurement trends; fair value for assets, liabilities, and equity; and alternative simulations for risk management. | |

Interest-Rate-Risk Measures | |

Today, most analysts and bankers concur that interest-rate-risk measurement includes fair values, duration, net-income forecasts, and rate-shock simulations. Interest-rate-risk measurements are shown in exhibits in the “ As discussed in the “
Net-interest earnings at risk is the In Peer Group C appears to have the least amount of risk to earnings from interest-rate movements. The mean percentage drop that potentially would occur from rate swings is lowest in Group C (-4.2%), and the highest percentage drop within Group C (-15.4%) is lower than the high percentage in Group A or Group B.
Equity value at risk is the potential In
Duration, or interest-rate elasticity (IRE), is an estimate of the amount of time until all the cash flows of a financial instrument (or portfolio of similar financial instruments) will be re-priced to current market rates. It can be approximated by interest-rate elasticity, the percentage drop of present value that is expected if rates rise by 100 basis points; therefore, the absolute value of interest-rate elasticity can be used as an estimate of modified duration. For example, the mean IRE of total securities for Group C is -2.9%. This same number can be read as 2.9 years as an estimate of modified duration. | |

Camels Rating Information | |

On December 9, 1996, the federal bank regulators adopted the revised Uniform Financial Institutions Rating System (UFIRS, or Camels, rating system). These new rules came after 17 years under the Camel rating system. The new rules added a sixth component, sensitivity to market risks; revised the definition of each component to include consideration for risk management; and established a requirement that each of the six components must be scored separately with a 1 – 5 rating (1 being the best). While the standard measurement model described here was originally created for interest-rate-risk measurement and determining fair values, it also provides ratio information for assessing the five financial components of the Camels system (all except management, which is qualitative). By using the standard measurement model to analyze all the financial risks at the same time, an institution ensures the compatibility of definitions and can obtain peer group statistics. The five financial components, --capital adequacy, asset quality, earnings liquidity, and sensitivity—are illustrated in Bank A has an excellent capital position. Bank A has a higher-than-peer average-equity-to-assets percentage and higher risk-based capital. The detailed version of Bank A has excellent asset quality. Bank A has no nonperforming assets, and a 1.7% allowance for loan losses. The average peer bank has fewer reserves and significantly more problem assets. Bank A enjoys a higher market-value premium in its investments portfolio and a higher present-value premium on loans. Bank A has no problem assets, good reserves, good economic values (market values), and no concentration problems. Bank A’s earnings performance is very good, especially its expense control. Bank A has very good ratios of return on assets and return on equity. The net-interest margin is just slightly under the peer group average. The good earnings come from a good operation efficiency. The banks’ liquidity position is excellent. The bank has a loan-to-deposit ration slightly higher than peer and a larger-than-peer percentage of assets invested in available-for-sale securities. Its primary funding is from deposits, and it uses a lower level of purchased funds relative to peer averages. Thus, if funds are needed for lending or deposit withdrawals, Bank A can meet its needs. Bank A’s interest-rate-risk exposure is low. The net-interest earnings at risk and the equity value at risk are both at or below peer average. All interest-rate elasticity measures (duration) are at or near the peer average, except the 4.1 years duration in the investment portfolio. The long investments are probably attempts to improve the yield on earning assets and present no liquidity or asset-quality problems. The bottom line for Illustration Bank A is that it is a strong candidate for a 2 or a 1 Camels rating on each of the five quantitative measures. A more complete examination will provide the final determination and give the information necessary for the management rating. Although the examiners make their own evaluation, the summary and detail reports of Camels rating ratios will give the banker an early indication of what may result from a regulatory review. | |

Risk-Measurement Trends | |

To give perspective to current measurements of any type of risk, analysts must understand historical trends. Understanding interest-rate risk is no exception. Any current measurement may be good or bad, depending on the measured value in the recent past. For the most recent quarter, it is important to examine how an individual bank compares to its peers as discussed above and shown in The earnings-at-risk measurement for Bank B has hovered around the 5% level for the past six quarters. The consistency of the measure is positive. As interest rates have fluctuated, Bank B has responded in ways that have held the risk exposure consistent from quarter to quarter. In contract, Banc C’s measurement for earnings at risk has hovered around the 1% to 2% level. Bank C’s interest earnings-at-risk exposure is substantially less than Bank B, but the exposure over time seems to move in the opposite direction from Bank B’s exposure. The economic equity value-at-risk measurement for Bank C has hovered above 14% during the past six quarters; for Bank B, the measurement has been in the 9% range in recent quarters. Bank C’s higher economic equity value at risk seems to offset its lower earnings at risk. The opposite is true for Bank B. A review of the interest-rate elasticity, or duration, of Bank B’s assets and liabilities is consistent with the equity-value-at risk analysis. Bank B has shorter assets and comparable or longer liabilities. The securities portfolio duration, for example, in Bank B is slightly less than two years and the securities portfolio duration in Bank C is more than five years. The trend analysis of interest-rate-risk measurements reveals that these banks, on the average, are performing consistently over time. While the average consistency was probably caused somewhat by reasonably stable economic activity over the past few years, not all banks performed at the average. A review of the detail data for individual banks in our quarterly Industry Reports reveals that some were more volatile than others. One can conclude, therefore, that some banks have had more interest-rate risk. A basic principle of economic analysis is that high risk is associated with high volatility, and vice versa. | |

The Fair-Value Juggernaut | |

Both the Financial Accounting Standards Board (FASB) and the International Accounting Standards Committee (IASC) are moving toward full market-value accounting. In general, the desire for the use of market values instead of the existing historical cost base is driven by the need to observe, analyze, and understand risk (market, price, and interest-rate risks in particular). The time of full market-value accounting appears to be close at hand. Bankers, accountants, and analysts must agree upon a standard approach to determining fair value. Agreement about concepts, methods, and techniques is especially needed for financial instruments that are not actively traded in a public financial market. Trading market values or acceptable reference market values are available for investment securities and a few other items, such as derivatives, traded long-term debt, and some loans. For non-maturity deposits and many loan portfolios (particularly in the vast majority of smaller banks), traded market values are not available. In June 1997, the FASB moved toward resolving questions about methods for determining fair value in an exposure draft of a proposed Statement of Financial Accounting Concepts, “Using Cash Flow Information in Accounting Measurements.” In summary, “This Statement provides a framework for using future cash flow as the basis for an accounting measurement….It also provides a common understanding of the objective of present value in accounting measurements.” For fair values based on discounted cash flows, there must be agreement among bankers, analysts, accountants, and regulators about the concepts, assumptions, and methods used to estimate the cash flows and the discount rates. The model used here (and described in the “ Fair values that are computed, instead of taken from the market, must be verifiable and believable. The present-value computations must be repeatable by objective their-party referees, the fundamental data used by the computations must be verifiable, and the assumptions must logically fit with industry knowledge and norms. The standard measurement model presented here produces a full present value balance sheet and can produce detail reports about the computations, data, and assumptions. In addition, computed fair values must be consistent with information about financial market prices and interest-rate movements over time. The standard measurement model presented here produces fair values consistent with market data. The data in
The first data presented in As interest rates rose during early 1996, bond prices fell. The premiums of market values over cost also fell. As rates flattened or fell slightly during the fourth quarter 1996, bond prices rallied some and premiums increased. The mean value for the premium was 1.1% for fourth quarter 1995 and at 0.4% for the first quarter 1997. The lowest premium in an individual bank was actually a loss of -4.7% in the second quarter of 1996; the highest premium was 20.6% in the first quarter of 1997.
When the present-value premiums of loans, as report in The cash flow estimates of the current model are based upon current balances, fixed/variable-rate classifications, amortization definitions, maturity dates, and prepayment assumptions. The discount rates are daily rates from the U.S. Treasury yield curve adjusted by a risk premium that reflects allocated operation expenses and a credit-quality factor. The discount rates have passed the test of bankers as reasonable estimates of current offering rates. The resultant fair-value estimates have passed the test of auditors who were faced with implementation of SFA no. 107, “Disclosures about Fair Value of Financial Instruments.”
The present-value premiums increased from on year-end to the net for the median and the mean for the peer group. This increase in deposits premiums primarily reflected the increase in interest rates on the long end of the U.S. Treasury yield curve during 1996. The pattern of the mean of the deposit premiums is the mirror image (reverse) of the pattern of the premiums for loan present values and premiums for investment securities market values, that is, when asset values go up, liability values go down. In addition to the verifiability of the data, assumptions, and computations, the model produces fair values consistent with information about actual purchases and sales of bank deposits and branches.
A casual review of 54 branch or bank sales between 1989 and 1996 The deposits premium paid on the above transactions was higher (on average) for transactions executed without the assistance of the RTC. The mean of the deposit premium for the non-assisted transactions was 5.78%. The mean of the deposit premium for the assisted transaction was 2.69%. It can be concluded that the RTC-assisted transactions were made an arm’s length value, which represented the cost of alternative funding (or “bond value of deposits”). Further, it can be concluded that a 3.0% incremental franchise-value premium exists over and above the alternative funding cost, as evidenced by the unassisted free-market negotiations. Although there are timing differences in the numbers (and therefore different market conditions), the 2.69% mean of the reported non-assisted sales is similar to the mean reported in
One of the criticisms of market-value accounting has been that it would create a confusing and irrelevant volatility of equity. Such does not appear to be the case, however. For the present value of equity premium over book, the peer group statistics in The model determines the present value of equity (economic value of equity) by subtracting the present value of liabilities from the present value of assets. The collective result of the present-value calculations does not take into account the value of the franchise but only considers the economic present value of the future cash flows. Franchise value, on the other hand, reflects the added business value of the customer base and other market considerations above and beyond the net valuation of discounted assets and liabilities. As with the deposit present-value premium compared to book the cash flow discounting process does not attempt to evaluate the value that may be assigned to the portfolio due to unique markets, unique customer relationships, supply quantities, or alternative demands. The economic value of equity may not reflect market capitalization value. The trading price of common (or preferred) stock on any day is influenced by many things other than fundamental economic value of the sum of the net present value of the collected individual portfolios. On the other hand, the economic value of equity provides a rational focus for evaluating the potential impact of adverse-rate and financial-market price movements. | |

Uses of Standard Measurement Model for Risk Management | |

Comprehensive risk management has become the banking industry’s current byword. During 1996, the bank regulatory agencies took an important, new initiative: the regulatory risk-assessment system. For this system to simplify exams, enhance safety and soundness, and improve overall performance of the industry, three prerequisites must be met: - There must be commonly accepted definitions.
- Regulators must follow through with implementation at the field examination level.
- Bankers must implement risk-management systems (to the extent that effective systems are not in place.)
Budget planning, forecasting, and what-if simulations are important parts of a risk-management system. Today many banks have a comprehensive budget-planning system, or at a minimum, executives, staff, and board members annually develop some notion about the coming year’s income, expense, and growth opportunities. The budget becomes a base for measurement of progress and budget variances. Computerized, mathematical models that produce forecasts and what-if simulations are also important parts of risk assessment and decision making.
Risk management is focused on controlling future losses. So, current measured positions that serve as benchmark estimates of future outcomes are important. A financial forecast provides a framework for evaluating potential decisions that will affect the future. Some banks develop a flat forecast of balance-sheet amounts and constant interest rates. Others use business plans and budgets as a base forecast for a risk-management system. To be useful, forecast calculations must, first, adhere to the accounting principles of debit, credit, balance, accruals, and cash flow. The idea of using statistical trend projections performance characteristics has given way to more detailed approaches. Separation of current balance maturity and re-pricing characteristics from new value assumptions provides the capability to calculate alternative cash flows. Second, models must provide interest income and expense calculations that are tied, either directly or indirectly, to the U.S. Treasury yield curve. The yield curve reflects the financial markets, which drive interest income and expense, asset and funding volumes, and pricing decisions for loans and deposits. The risk measurements shown in the exhibits of the
Interest-rate-risk measurement requires some type of what-if analysis. Once a bank has a base forecast, it can pose what-if questions. The analyst can change one or more assumptions and see how the base forecast changes. Typical what-if questions involve different interest rates, different growth assumptions, or different product mixes. Different product characteristics also give rise to what-if questions. Financial instruments, whether capital market instruments or loans and deposits, incorporate many different characteristics: fixed or floating interest rates, spread definitions, maturity characteristics, amortization techniques, options—the list goes on. Because of this complexity, all forecast explicitly or implicitly make assumptions about the characteristics of financial instruments. What-if simulations are useful in fine-tuning a forecast, a budget, or a plan. What-ifs are also useful in evaluating potential changes to balance sheets. Since risk is the potential harm or injury that may occur in the future, what-if calculations are needed for risk assessment. | |

The Industry Should Establish Standards | |

To manage risk effectively and efficiently, banks regulators, and analysts must agree on benchmarks or standards. A standard measurement model can produce information that is useful in the risk-management process. The exhibits here show information from 600 banks using consistent definitions, inputs, and assumptions and demonstrate that risk measurements can be made reasonably standard. The time to move forward with standard approach is now. |