Discussion of Earnings at Risk (EAR) An Effective and Understandable Methodology for IRR Management. Randy C. Thompson, Ph.D.

Transcription

1 Discussion of Earnings at Risk (EAR) An Effective and Understandable Methodology for IRR Management Randy C. Thompson, Ph.D. CEO - TCT Risk Solutions, LLC August 7, 2014 Value-at-risk has become such a widely adopted risk measurement tool that it can seem heretical to question the relevance of the VAR concept. But some managers in the corporate and government sectors have done just that, arguing the VAR models as used by financial intermediaries are counter-productive to their own identification and management of risk. Introduction and Overview Justin Myatt (2003) Identifying, measuring and managing interest rate risk has become a critical issue for credit unions. Following deregulation in the early to mid-1980s, credit union balance sheets have become more complex and susceptible to the unpredictability of interest rate risk. Regulators have emphasized measurement and management of interest rate risk both in focused regulatory changes and in the day to day practices of examination staff. There is no argument that increased scrutiny of the impact of interest rate risk on credit unions is warranted. The changing nature of credit union balance sheets with increasing percentages of long-term assets makes them more susceptible to various risks such as re-pricing, yield and basis. However, while the need to measure and manage IRR is well established, the methodology that is most appropriate and or useful is less well determined. However, scholars have questioned the reliability and validity of VAR modeled risk assessments. Reed (1997), Hoppe (1999) and Simons (2000) identified the following shortcomings with VAR. First, there is no standard agreement in respect to VAR methodology. Second, VAR seems to be designed to assess risk of investment trading of short-term liquid assets in a natural market condition. Third, a few studies show that different VAR methodologies yield different results despite the fact that the same stat and time period are use. In response to these noted deficiencies in VAR multiple authors such as KPMV Risk Publications (1997), Schachter (1997) and Jones, Oncu and Sheikh (2000) suggest that although VAR can provide an efficient (single figure) measure of IRR, other types of measures (such as 1 P a g e

2 EAR) should be used to engender more reliable risk management practices. (As reported in Muresan and Danila 2004). As noted in the quote above, value at risk (VAR) has become widely accepted as an IRR tool. Its application in banks is understandable. Value at risk models are widely used by banks and other companies with stock ownership held by discreet investors. The valuation of investor stock is derived from a value at risk model. For this reason the use of VAR models is most appropriate for banks. Credit unions are not structured like banks with discreet investor owners. Credit unions are cooperatives with shared ownership. The valuation of the balance sheet for purposes of establishing stock value is irrelevant for credit unions. Equity, or capital, is driven by cash flow and earnings. Since equity is a principle indicator of credit union soundness its derivatives are an essential ingredient for analyzing risk. Positive earnings are the source of equity growth and losses are the cause of equity reduction. Hence quantifying and measuring the impact of IRR on earnings, and equity are of paramount importance to credit unions. To regulators and credit union leaders alike, the determination of cash flow, and it impact on earnings should be of paramount importance. Discussion of the Earning at Risk (EAR) Model A relevant and meaningful way to quantify IRR at a credit union is Earnings at Risk or EAR. EAR measures the quantity by which net income is projected to decline in the event of an adverse change in prevailing interest rates. EAR provides a robust measure of a credit union s exposure to adverse consequences from changes in prevailing interest rates. Practitioners report that EAR is an objective, statistically derived risk metric that measures risk with 97.5% level of confidence. (Vanguard 2008) Earnings at risk (EAR) measures the quantity by which net income might change in the event of an adverse change in interest rates. It is a risk measurement which is closely linked with value at risk (VAR) calculations. The difference is that while VAR looks at the change in the entire value over the forecast horizon, EAR looks at potential changes in cash flows or earnings. Both models rely on a determination of cash flows. The difference is that VAR discounts the cash flows using present value calculations while EAR applies the cash flows to simulate the effect of interest rates changes on the actual earnings of the credit union. In addition, EAR allows the credit union to quantify and measure the effect of non-interest income and operational expenses on the credit union over the horizon of the analysis in concert with the cash flow analysis. To illustrate the basic implementation of an EAR model, let's assume we have a simple credit union that has only one asset, a 10 year U.S. Treasury bond yielding 4%. Now, let's assume that this credit union has only one deposit, a six-month CDs currently priced 1%. The base net interest income equals the 4% yield on the bond less the 1% cost of the deposit, or 3%. This becomes our base net interest income, or margin. 2 P a g e

3 Now, let's assume an interest rate shock of +100 basis points. If rates go up 100 bp in an immediate shock, the yield on our 10 year bond will not change. However, after six months, when we renew the CD it will now cost 2%. That means that for the first 12 months, we have six months of net interest income at 3% (unchanged from the base) and then six months with a net interest income of 2% (4% bond yield less 2% CD cost). This means that in a rates up 100 basis point scenario, the twelve-month net interest income in the up 100 basis point scenario is now reduced to 2.5%. Similarly, if rates go up 200, 300, and 400 basis points in an immediate shock scenario, our net interest income will decline an additional 50 basis points for each 100 basis points of rate shock increase. These additional analyses provide a simulation of anticipated cash flow and earnings which are easy to read and understand. Proponents of VAR regularly point out that it is robust because it takes into account longer-term cash flows in the calculation of present value. That is true. However, EAR also takes the longterm cash flows into account in this way. Assets are re-priced consistent with the amortization pattern established in the loan contract. For example if the credit union mentioned above funds a 60 month auto loan, every month the borrower will return roughly 1/60 th of the principle in the scheduled monthly payment. This amount is now available to fund new loans at the prevailing interest rate. The composite of monthly payments is the amount that may be used to fund new loans and increase the weighted average loan rate. With a 1 year loan, 1/12 th of the principle is returned each month. With a 30 year loan, 1/360 th of the principle is returned each month. As longer term loans are funded the rate at which the weighted average can be increased reduces. Hence, EAR shows the immediate effect on cash flow resulting from longer term loans. Consider the credit union we discussed above. Because the U.S. Treasury bond has a fixed rate for 10 years it will continue to earn 4% until maturity. Now assume that the credit union also has a 3 year auto loan written at 5% (Scenrio 1). The weighted average yield of the loan and bond is 4.5% in the first year. Each year 1/3 rd of the principle, or $3,333 is returned to the credit union. That means that if rates do increase by 1% the returned principle will fund loans at a rate of 6%. Over the three year period the total amount will re-price to 6% and the weighted average yield will increase to 5.0%. Of course this simple example assumes no prepayment of the principle balance. Now if the loan was a 10 year loan (instead of a 3 year loan) then only 1/10 th of the principle, or $1,000, would return each year in principle payments (Scenario 2). At the end of the first three years only $3,000 would re-price to the 6% rate and the weighted average yield will increase to only 4.65%. Hence, the longer term cash flows associated with long-term loans, have an immediate and direct effect on cash flow and earnings. So, the presence of longer term loans places a direct downward pressure on the potential yield increases resulting from any upward shock. This downward pressure is exerted over the life of the longer-term loans making the effect a multiple year phenomena. The table and graph on the following page give a visual representation of this effect. 3 P a g e

4 Scenario 1 Scenario 2 3 Year Auto Loan 10 Year Auto Loan Balance Yield Earnings Balance Yield Earnings Base Year T Bond $ 10, % $ $ 10, % $ Original Loan $ 10, % $ $ 10, % $ Total $ 20, % $ $ 20, % $ Year 1 T Bond $ 10, % $ $ 10, % $ Original Loan $ 6, % $ $ 9, % $ New Loan $ 3, % $ $ 1, % $ 60.0 Total $ 20, % $ $ 20, % $ Year 2 T Bond $ 10, % $ $ 10, % $ Original Loan $ 3, % $ $ 8, % $ New Loan $ 6, % $ $ 2, % $ Total $ 20, % $ $ 20, % $ Year 3 T Bond $ 10, % $ $ 10, % $ Original Loan $ % $ - $ 7, % $ New Loan $ 10, % $ $ 3, % $ Total $ 20, % $ 1,000.0 $ 20, % $ Earnings Comparisons: 3 Yr vs. 10 yr Loans $1,020.0 $1,000.0 $980.0 $960.0 $940.0 $920.0 $900.0 $880.0 $860.0 $840.0 Base Year Year 1 Year 2 Year 3 3 Year Auto Loan Earnings 10 Year Auto Loan Earnings Note: Assumes Constant balance of loans and investments 4 P a g e

5 Of course, actual earnings at risk calculations must include all line items on the balance sheet, as well as interrelationships between and among these items. This makes earnings at risk a much more complex calculation. However, it is still true that understanding the basics will help credit union leaders and examiners better understand the more complex results as well. Both state and federal regulations emphasize the importance of using IRR tools and models to quantify and measure risk and then to guide actions to manage it. We suggest that the EAR model is more intuitive in its analysis and more effective in developing strategies to mitigate IRR. Because this is a simulation of earnings that is tied to actual balance sheet behavior, it is easy to see the interrelationships between balance sheet alterations and the effects on earnings. For example, an increase in longer-term loans will slow amortization, reduce weighted yield increases and directly impact earnings and equity. The relationship of yield and earnings is something all managers and board members are accustomed to examining. To link EAR to IRR management is to take the familiar and apply it to the required. In a 2005 presentation Glacy and Sarna point out that EAR presents data in an accounting format consistent with GAAP. They also stated that EAR is a concise measure of downside risk that expresses results in accounting terms, captures all financial risks concurrently, proportionalizes exposures to various risks and enables a rich array of what if exercises. Using the simulations associated with EAR a credit union s leadership may run what-if simulations for different and distinct balance sheet alterations (e.g., increase real estate loans, longer term investments, or changing deposit mix) and see immediately the impact on IRR. If the primary goal of regulation is to increase board involvement in the management of IRR then EAR provides and efficient method to accomplish that goal. Credit union leaders are well versed in accounting and GAAP terms which makes EAR a more conversant model for credit union IRR management. Summary of EAR Output EAR reports show the interactions between the key elements of credit union earnings and the environmental interests that are changing as a measured by shocks. Some of these key elements are: weighted yield on loans, weighted yield on investments, cost of funds, net income/earnings, NII at risk, equity at risk and Equity Valuation (simulated equity based on earnings). By extending the analysis over a period of years the net effect of the shock on the both the base balance sheet and multiple iterations of the balance sheet through simulations. The annualized measures of yield in the base line case show the IRR in the current balance sheet. Simulations, or what ifs, can then be run and compared to the base line to measure the change in IRR that would be caused by the balance sheet alterations included in the simulation and create projections. 5 P a g e

6 The following tables show examples of the reporting of key elements in the base balance sheet. Based on the Weighted Average Maturity of loans a re-pricing schedule is derived resulting in a curve that represents the simulation of the weighted average yield of loans over the five years of the analysis. Average Yield on Loans Base Yr. Year 1 Year 2 Year 3 Year 4 Year 5 5% Upshock 5.65% 7.78% 8.83% 9.29% 9.63% 9.94% 4% Upshock 5.65% 7.35% 8.19% 8.56% 8.84% 9.08% 3% Upshock 5.65% 6.93% 7.56% 7.83% 8.04% 8.23% 1% Upshock 5.65% 6.08% 6.29% 6.38% 6.45% 6.51% -1% Downshock 5.65% 5.23% 5.02% 4.93% 4.86% 4.79% Based on the maturity of traditional investments and the cash flow of CMO type investments a re-pricing schedule is derived resulting in a curve that represents the simulation of the weighted average yield of loans over the five years of the analysis Average Yield on Investments Base Yr. Year 1 Year 2 Year 3 Year 4 Year 5 5% Upshock 1.28% 2.82% 3.91% 4.42% 5.03% 5.84% 4% Upshock 1.28% 2.51% 3.38% 3.79% 4.28% 4.93% 3% Upshock 1.28% 2.20% 2.86% 3.16% 3.53% 4.02% 1% Upshock 1.28% 1.59% 1.80% 1.91% 2.03% 2.19% -1% Downshock 1.28% 0.97% 0.75% 0.65% 0.53% 0.37% Using an econometric model employing the price elasticity of demand a re-pricing schedule for each unique type of deposits is created resulting in a curve that represents the simulation of the weighted average cost of loans over the five years of the analysis. Cost of Funds Base Yr. Year 1 Year 2 Year 3 Year 4 Year 5 5% Upshock 0.10% 4.44% 4.44% 4.44% 4.44% 4.44% 4% Upshock 0.10% 3.57% 3.57% 3.57% 3.57% 3.57% 3% Upshock 0.10% 2.71% 2.71% 2.71% 2.71% 2.71% 1% Upshock 0.10% 0.97% 0.97% 0.97% 0.97% 0.97% -1% Downshock 0.10% 0.03% 0.03% 0.03% 0.03% 0.03% 6 P a g e

7 By applying the derived rates for loans, investment and deposits against balances a simulated income statement is created for each year of the analysis. This income statement illustrates the change in interest income and interest expense from the base period that is caused by the interest rate shock. By adding other income and operating expenses (consistent with the accounting process) a simulated ROA is derived and the resulting impact on equity (equity valuation) is calculated. This process is repeated for each shock percentage over the five years of the study. These findings can then be compared to previous reporting periods to assess IRR trends for the credit union. Putting the results of an IRR analysis in accounting terms such as these creates specific advantages for the credit union. First it provides a commonality with the day to day accounting operations of the credit union. Second, it increases familiarity with IRR analysis by board and ALCO members. Third, it allows for regular and consistent what if simulations to guide credit unions in the management of the balance sheet. A final set of measurements associated with IRR that are provided by EAR are Net Interest Income (NII) at risk and Equity at Risk. This measurement shows the direct effect of the mismatched re-pricing curves on interest income at the credit union. In other words it provides a basic measure of actual IRR resulting from interest rate shocks. The next table shows a sample of this measure. Regulations required that credit union boards create policies with establish limits for measurements such as NII at risk. Management is then responsible to monitor, measure and report actual IRR measures in comparison to these limits to the board on a regular basis. 7 P a g e

8 Net Interest Margin Base Yr. Year 1 Year 2 Year 3 Year 4 Year 5 5% Upshock $ 1,973, $ 1,686, $ 1,634, $ 1,949, $ 2,197, $ 2,373, % Upshock $ 1,973, $ 1,743, $ 1,702, $ 1,954, $ 2,152, $ 2,293, % Upshock $ 1,973, $ 1,801, $ 1,770, $ 1,959, $ 2,108, $ 2,213, % Upshock $ 1,973, $ 1,916, $ 1,905, $ 1,969, $ 2,018, $ 2,053, % Downshock $ 1,973, $ 1,926, $ 1,850, $ 1,787, $ 1,737, $ 1,702, Interest Margin Percent Change Base Yr. Year 1 Year 2 Year 3 Year 4 Year 5 5% Upshock 0.00% % % -1.22% 11.34% 20.26% 4% Upshock 0.00% % % -0.98% 9.07% 16.21% 3% Upshock 0.00% -8.75% % -0.73% 6.80% 12.16% 1% Upshock 0.00% -2.92% -3.44% -0.24% 2.27% 4.05% -1% Downshock 0.00% -2.38% -6.25% -9.47% % % Conclusion Although Value-at-risk has become such a widely adopted risk measurement tool in banks and equity markets, its application in credit unions is questionable. As noted in the beginning paragraphs of this paper, VAR presents a number of deficiencies that minimizes it value to credit unions. In addition to the measurement and management of IRR, forecasting through EAR may be employed to evaluate concentration risk policies, liquidity risk, and investment practices. EAR shifts the focus from an estimation of market or liquidation value to a forecast of future earnings which will facilitate enhanced planning, budgeting and increased understanding of IRR; all of which will lead to enhanced management practices by credit union mangers and boards. 8 P a g e

9 Additional Notes: To assure statistical validity and reliability of the model, it is critical to use solid scientific design. To this end the EAR methodology holds the balance sheet constant thus isolating the interest rate shock as the sole independent variable in the analysis. In this way, IRR is measured for the existing balance sheet. Once the base line IRR is measured, alterations to the balances sheet (e.g., deposit runoff or growth, loan run off or growth, changes to the investment ladder, or changes to other income or operating expenses) may be modeled in a what if simulation. Output from these simulation may then be compared to the Base Case to quantify the total impact on IRR resulting from the alteration(s). Credit Union leaders should also note that because of the number of variables that can be applied to EAR modeling, no two models will look the same, even if applied to the same organization. It is therefore essential to ask for specific details when analyzing any EAR summary. EAR models will show analysis for any number of increases or decreases in interest rates. These may be shown as a single rise or fall (immediate shock), or gradual change (stepped shock) in rates. Very few organizations devise EAR models from scratch, but will rely on summary results generated by a computer. Credit union managers may well need to analyze EAR reports generated by specialized modeling tools and applications. 9 P a g e

10 Supplemental Table Comparison of Key Concepts for VAR and EAR Comparison of Risk Concepts and At-Risk Measures Concept Value at Risk Earnings at Risk Measurement Variable Market Value (a stock) Profit, earnings or cash flow (a flow) Example The market value of a portfolio (loans, investments, deposits The future cash flows from expected yields on assets and cost of funds Representation of Risk Variability of market value Variability of future cash flow and earnings Exposure analyzed At-risk quantification Source of variation (or risk factors Holding period for variation of prices Portfolio of assets and liabilities on balance sheet Difference between the market value of assets and liabilities Changes in interest rates affecting the net present value of assets and liabilities Single measure of current value based on discounted present value Future cash flows resulting from assets and liabilities on balance sheet Difference between current cash flows/earnings and future cash flow/earnings Changes in interest rates affecting cash flows and net earnings Multiple year (three to five) simulation of cash flows and net earnings 10 P a g e

11 References Glacy, J., & Sarna, C., Earning at Risk: Real-world Risk Management, Allstate Insurance symposium presentation, slide [online] Hoppe, R., It s Time We Buried Value at Risk. Risk Professional, July/August 1999, pp Jones, C., S. Oncu, and A. Sheikh. Developing and Implementing Risk Management Systems. BARRA Enterprise Risk Management Research Newsletter, no. 168 [online] KPMG Risk Publications, Financial Engineering Ltd, London, 1997 Muresan, E. R., & Danila, N., Using Earnings at Risk to Asses Risk of Indonesian Banks, Center for Public Research, Myatt, J., Other Alternatives to VAR, Risk Measurement, February 2003, pp Reed, N., Variations on a Theme: Understanding and Applying Value-at-Risk. KPMG Risk Publications, Financial Engineering Ltd, London 1997 Simons, K., Value at Risk New Approaches to Risk Management. In: VaR: Understanding and Applying Value-at-Risk. KPMG Risk Publications, Financial Engineering Ltd, London 1997 Vanguard, Investing [online] 11 P a g e