This article is the second of a two-part series addressing credit risk

DOWN ON THE FARM Stress-Testing Net cash farm income of U.S. farmers in 1999, thanks to record level direct government payments received from Washington, was virtually identical to the $57.5 billion achieved in 1996. Back then, however, these payments were only 13% of net farm income. In 1999, these transfer payments represented a whopping 47% of net farm income! The fact that almost one-half of net farm income came from Washington rather than the marketplace last year underscores the weakened economic fundamentals in this sector. A variety of highly probable by John B. Penson Jr. This article is the second of a two-part series addressing credit risk in an increasingly uncertain agricultural lending environment. Emphasis in the first article was on the forces causing change in agriculture and the need to event stress test new term loan applications and debt restructuring requests to gain an understanding of the potential downside risks before the loan is made. This article focuses on event stresstesting an entire agricultural loan portfolio. adverse events over the next several years may continue to plague agriculture. A continued weak export market, higher interest rates, higher energy prices, and rising variable input prices are among the highly probable possibilities. How will these events affect the future debt repayment capacity of farm borrowers? Will a major overhaul of current U.S. farm policy help stabilize adverse trends? Which segments of agricultural loan portfolios will be most affected? These and other issues underscore the need for management to gain a full understanding of likely future trends in the 2000 by RMA. Penson is Stiles Professor of Agriculture, Texas A&M University. performance of their portfolio, and how specific events might alter these trends. Externalities Affecting Portfolio Performance Stress-testing of an agricultural loan portfolio should be conducted in an event-driven context, where the policies or events causing stress are clearly identified and where the reasons for any remedial actions to be taken are clearly understood. This requires a comprehensive management information system (MIS) that integrates a loan accounting system with current financial and credit 70 The Journal of Lending & Credit Risk Management February 2000

information for each account. The marriage of historical analyses with realistic projections of future scenarios provides boards of directors and management with the tools neces - sary to evaluate/predict potential risk exposure. Managers of agricultural loan portfolios may well be asked by their board of directors or regulator to address the implications of specific What if questions like those listed below. Suppose a board member of your institution read an article in the Wall Street Journal before coming to the meeting and wanted to know the implications the very first what if question continued global surplus production has for the performance of your institution s agricultural loan portfolio: 1. Production concerns: What if global surplus production for major crops continues to overhang commodity markets for another one or two years, keeping commodity prices at or below current depressed levels? What if excess global supply exists and local adverse weather conditions result in reduced yields not fully offset by crop insurance? 2. Farm policy concerns: What if pressures mount for change in farm commodity policy which reverts back to supply management policies embodied in previous legislation? What if direct government payments to farmers are reduced in a budget-cutting move accompanying proposed cuts in income tax rates? 3. Macroeconomic concerns: What if the Federal Reserve continues to raise interest rates in 2000 to ward off the unwanted inflationary pressures from a global economic bounce in export demand, rising energy prices, and an end to cheap imports? What if the Federal Reserve was only partially successful and we had the combination of higher interest rates and higher inflation? 4. Foreign trade concerns: What if the value of the dollar rose relative to currencies in client nations and competitor nations, making it relatively more costly to buy U.S. agricultural products? What if increasing concerns about overuse of chemicals (hormones in beef and genetically modified organisms or GMO in grain and oilseeds) further depress exports? What approach would you use to address the question raised by the board member? How long would it take to report back to the board? The technology is available to address each of these issues, so saying Who knows? is not an option! One thing in common among all of the events identified above is that they are all beyond the control of both borrowers and lenders. Furthermore, each has a probability of actually occurring, although some have a higher probability than others. Stress-testing for the effects of key external factors enhances management s ability to identify and control risk and to plan, prepare, and respond to real and potential portfolio threats. 1 Agricultural Loan Portfolio Stress-Testing Options Many agricultural lending institutions have computer programs that slice and dice their existing portfolio. These programs assess the existing portfolio concentration and, comparatively speaking, how well specific segments, or groups of accounts in the portfolio, have performed in the past. 2 While this may provide useful information to management, this is not a substitute for stress-testing expected future portfolio performance. Some institutions go one step further by assuming a specific decline in the price of a commodity or in the level of net farm income to see what this would have meant for debt coverage in their existing portfolio. This tactic, however, fails to account for the relationships among commodities and the reasons for a downturn in net farm income. Why did the price of corn decline or net farm income fall? Is this a one-time occurrence, or does it represent just the beginning of several years of adverse market conditions? Simply put, a five-cent decline in the price of corn does not take place in a vacuum. Other crop and livestock commodity prices are affected, as are farmland prices and other factors determining the overall health of customers over time! It is hard to be proactive in management if one does not understand the underlying fundamentals causing change. Ad hoc forecasts lack a systematic foundation across commodities and sectors and may well lead to substantial inconsistencies and significant error, particularly at the segment level. Furthermore, there is little or no linkage to specific events necessary to assess risk and risk 71

Ad Hoc Based upon assumptions or extrapolation of recent prices or net incomes. Lack of formal linkage to specific events or What if questions. Stress-Testing Choices management strategies in a meaningful manner. An alternative to ad hoc stresstesting is event stress-testing, where the reasons for a drop in the price of corn or level of net farm income are understood. Assessing the impact that the external factors listed earlier have on an agricultural loan portfolio requires access to structural econometric models tested over time as well as seasoned judgment on the part of the agricultural economists applying these models. The projections from this process represent sets of commodity prices, input prices, farmland values, and other sector and macroeconomic variables associated with a specific event, such as reverting to target price and set-aside concepts embodied in previous legislation, or some combination of events. This structural analytical framework should capture farm commodity market interrelationships, aggregate sector relationships, international trade relationships, and macroeconomic relationships. Most lenders admittedly do not have direct access to large-scale models of the domestic and international economies capable of projecting future trends in these variables, first under a most likely set of policies Event Driven Based upon structural econometric model projections of prices in a multi-market equilibrium framework. and events and then under alternative market scenarios. Yet, this is a necessary foundation to meaningful pro forma analysis and event stress-testing. One approach is to purchase a forecast of the most likely trends in farm commodity prices, input costs, and land prices from a reputable economic forecasting firm and then buy forecasts for alternative what if scenarios on an as-needed basis. Simulation of each account in the portfolio in a whole farm context based upon these projections would illustrate the downside risk associated with such key variables as nonaccrual loan volume and level of interest income over the next several years. Another approach is to use available software designed specifically to event stress-test agricultural loan portfolios. 3 Directly associated with specific alternative external forces, including a change in farm policy and global market events. Two Potential Perspectives There are two broad perspectives one can take when event stress-testing an agricultural loan portfolio. The first approach focuses on the future performance of specific segments of the portfolio singling out those groups of accounts falling below underwriting standards for a more in-depth analysis. The second approach focuses on the portfolio s contribution to the overall performance of the institution, illustrating how its profitability and other performance statistics are affected by agriculture-related externalities. Segment analysis. An agricultural loan portfolio is comprised of many segments or groups of homogeneous accounts reflecting unique commodity, location, and size characteristics. The more homogeneous the portfolio, the fewer the number of segments. A Corn Belt agricultural loan portfolio, for example, will be dominated by a relatively small number of commodities (corn, soybeans, hogs, and cattle). An east or west coast portfolio, on the other hand, will reflect a wide variety of fruits and vegetables. Not all segments of an agricultural loan portfolio will be identically affected by a weak export demand or by changes in farm policy. Accounts tied to export-sensitive commodities like wheat are more likely over the next few years to have increasing financial stress than commodities tied more closely to domestic demand. The goal is to identify which segments are more likely to exhibit increasing financial stress and require further attention. Focus at the segment level should illustrate the number of accounts and corresponding loan amounts outstanding in a particular segment that are projected to become nonaccrual as events unfold over time; it should also illustrate the year this would occur. For example, one may wish to assess all accounts located in Southeast Washington currently classified below acceptable in which wheat is the primary commodity. What we want to know here is which accounts are projected to fall below underwriting standards and will likely need carryover debt financing in light of cash flow deficits. Because of the pro forma nature of this analysis, the potential effects of 72 The Journal of Lending & Credit Risk Management February 2000

undesirable concentration can be visualized in time to consider the benefits from less concentration in a specific commodity, or to consider other strategies to offset increasing credit risk exposure. This analysis also should help illustrate which accounts in a particular segment would be most vulnerable to the stresses associated with specific external factors in product and input markets. Some crop producers will be hurt by weak export conditions while some livestock producers may benefit from the resulting lower feed grain costs. A chief credit officer may wish to further query the projected accounts in this segment to identify the credit officers associated with the accounts projected to be in a weakened condition. The credit officer should be made aware of the reasons for and level of severity, how this stress will occur over time, and the projected level of carryover debt financing required to cash flow operations. Both individuals will want to see the likelihood of these accounts overcoming this stress in the longer run as well. For example, will accounts classified today as special mention or substandard likely become nonaccrual over the next several years? Pro forma analysis at the segment level, combined with meaningful event stress-testing, also should help management assess the adequacy of an institution s underwriting standards for agricultural accounts. This includes not only the variables reflected in the underwriting process but the weights assigned to these variables as well. The fact that a particular set of variables and weights worked well in the past does not mean they will work well in a new lending environment characterized by greater price variability and a declining government safety net. Balance sheet indicators of solvency and liquidity, for example, may not pick up building financial stress as quickly as projections of debt coverage ratios or enterprise operating margins. 4 This same argument can be extended to risk-rating systems and loan-pricing sheets as well. The current variable weights and threshold score should be assessed in an event stress-test setting to see if they successfully identify accounts eventually projected to become nonaccrual over the next few years. Yet another dimension to segment analysis is assessing the risk and returns for loan accounts originated and maintained by a specific credit officer. To the extent that credit officer compensation is tied to measures of both quantity and quality, event stress-testing the segment of accounts administered by a particular credit officer to evaluate their potential risk and returns can provide a valuable insight as to credit quality. Institutional implications. Each of the what if questions raised earlier in this article also have implications for the performance of the portfolio as a whole and the financial credibility of the institution in the eyes of its stakeholders and regulator. The conclusions drawn at the portfolio level are not as clear-cut as they are at the segment level, since the negative effects of an external factor on one segment may have a positive effect on another segment of the agricultural loan portfolio. Simulation of all accounts in the portfolio will provide key information on the level of nonaccrual loans, interest income earned on institutional loans, and the need for adjustments to allowance for loan losses. This also will demonstrate the net effects of the existing portfolio concentration and implications for portfolio risk and return. An institution s business plan generally contains pro forma projections of the institution s financial statements over an intermediate time period under a most-likely scenario. Reflected in these statements are future trends in interest income, nonaccrual loans, and allowances for loan losses stemming from agricultural lending activities. The magnitude of this effect will depend upon the nature of the financial institution and the clientele it serves. Event stress-testing an agricultural loan portfolio at a commercial bank, for example, examines how alternative scenarios with a significant probability of occurrence would affect the agricultural loan division s contribution to the bank s capital, asset quality, earnings and liquidity. Event stress-testing an agricultural loan portfolio at a Farm Credit System (FCS) entity, on the other hand, involves projections of net cash provided by investing and financing activities in addition to the net cash provided from operating activities. The sensitivity of these sources and uses of funds to specific events obviously has a more pronounced impact on indicators of a FCS entity s capital, asset quality, earnings and liquidity because of its targeted lending authority. 5 The need to make provisions to loan losses to address potential downside risks in credit and collateral can also be tied to event testing. Focus here is on a systematic measure of an institution s near term 73

exposure to downside risks. Event stress-testing has the added benefit of helping explain the rationale underlying its allowance for loan loss level to the regulator. Missing data. An additional requirement for event stress-testing an entire agricultural loan portfolio is accounting for: A lack of full information necessary to simulate all accounts in the portfolio. The gap between annual remaining balances associated with existing accounts and the expected total loan volume in the agricultural loan portfolio over time. The graph to the right reflects a lack of information on existing accounts entered in a database in 1999 (the difference between the size of the portfolio and entered accounts) as well as the declining remaining balance on existing accounts). To account for these data gaps, one can profile the institution s existing portfolio and sample specific segments to identify representative or benchmark farms reflecting unique commodity, location, size and loan classification characteristics. These benchmarks and their corresponding relative importance weights, represent specific pools of loans in the portfolio and, as such, permit a portfolio-level assessment of credit risk exposure and indicate where further segment analysis of individual accounts could help direct proactive interaction with specific loan officers and existing customers. Summary Pro forma-based event stresstesting permits a proactive assessment of how certain events with a significant probability of occurring can affect specific segments of an institution s agricultural loan portfolio as well as the overall strength and performance of the institution itself. Agriculture today is more dependent than ever before on a growing world demand for food and fiber products to keep commodity prices above breakeven levels. There is a strong likelihood that we will continue to see considerable downside risk associated with commodity price spreads and asset values in agriculture. You can ignore the what if questions raised by board members and regulators. You can ignore the signals being sent by an increasingly global market place. In other words, you can ignore the benefits of event stress-testing at the account, segment, and portfolio levels, and continue to focus on past events when making decisions at each level. Or, as succinctly stated in a recent RMA publication: You can make the decision to invest in the people, processes, and tools that will ensure the marriage of risk management and business and lending strategy. By doing so, you will Size of portfolio not only be a survivor but a winner in the highly competitive financial services environment. Doing nothing or merely engaging passively in some solutions are not choices. Beating the Odds...A Community Banker s Guide to Risk Management Event stress-testing represents a viable risk management tool whose time has come. Combining the concept of pro forma analysis with simulation techniques and alternative projections from structural econometric models proven to signal distinct changes in domestic and global markets can make this a reality. Penson can be contacted at jpenson@tamu.edu. Notes 1 See page 18 of the booklet Loan Portfolio Management published in 1998 by the Farm credit Administration, regulator of the Farm Credit System. 2 The quality of this database, including the frequency at which information on specific accounts is refreshed, plays a key role in the meaningfulness of these efforts. For a further discussion of this point, see the article by Penson entitled Does Your Database Quack Like a Turkey? in the May 1999 issue of Ag Lender. 3 An example of existing event stress-testing software for agricultural loan portfolios can be seen at www.visionanalytics.com. 4 Many agricultural commodities also have unique financial characteristics that must be recognized in pro forma analyses of underwriting standards at the segment level. 5 The Farm Credit System, on page 20 of its recent booklet, Loan Portfolio Management, states the recognition of lessons from the past and use of historical credit analyses must be coupled with realistic projections of future scenarios. Benchmarks Constant portfolio Entered Accounts 1999 2000 2001 2002 2003 74 The Journal of Lending & Credit Risk Management February 2000