RSS | Register/註冊 | Log in/登入
Site search:
Home>FFTC Document Database>Extension Bulletins>The Role of Information in Lending and Credit Risk Management
facebook分享
The Role of Information in Lending and Credit Risk Management
John B. Penson, Jr.
Regents Professor and Stiles Professor of Agriculture
Texas A&M University
College Station, Texas USA, 2004-09-01

Abstract

This Bulletin discusses the role of information technology (IT) in agricultural lending, particularly in view of the globalization of economic and financial relationships in recent years which brings with it additional risks and uncertainties. It particularly underscores the need for an early warning system for agricultural loans and the incorporation of this information into decisions regarding individual loan requests and portfolio management strategies. The concept of stress testing is combined with a proposed early warning system based upon generally accepted economic concepts to quantify the probability of cash flow stress for benchmark enterprises. This stress will be related to the probability of the borrower covering his/her operating costs and total costs, and the implications for the exposure to credit risk and portfolio risk. This system can be seen as an extension to quantitative history-based credit scoring models currently used by lenders, and can be enhanced to include the probability of default.

Abstract

Introduction

Information has traditionally played a major role in lending decisions. The manner in which this information is stored, accessed, and analyzed, however, has changed markedly over the last ten years. Information technology has played a major role in this transformation. Largely gone are the days where lending decisions were made by a loan committee based upon hard copy documents stored in file cabinets. Internally maintained electronic databases, coupled with acquisition of a borrower's credit history from external credit repositories and credit scoring models, have revolutionized the loan decision-making process. The result is a more efficient decision process, both in terms of the resources required to make the decision, and the time interval necessary to provide the decision to the borrower. But are we using information technology to its fullest potential? And are existing computer-generated credit scores sufficient basis for reaching a decision?

Evaluation of a borrower's credit worthiness has typically rested upon historical information gleaned from credit bureau reports, credit scoring models, and current financial statements. The underlying assumption often voiced is that if the borrower could make payments on schedule in the past, then he/she must represent good future credit risk as well. Many credit scoring models in use today, however, have yet to be tested in a depressed farm economy absent of huge government subsidies. These models are based upon weights assigned to specific risk factors reflecting how well the applicant did in servicing existing debt commitments during a specific historical time period when government subsidies made the difference between profit and loss. Agricultural lenders, however, should account for likely future economic trends as well as historical performance when evaluating loans, particularly loans involving the purchase of fixed assets that have little or no alternative use in other sectors, and typically decline in market value during economic downturns.

The globalization of economic and financial relationships in recent years brings with it additional risks and uncertainty. When coupled with the unique risks facing agricultural lenders, the use of historical information is still necessary, but is no longer sufficient. The evolving agricultural lending environment raises serious doubts as to whether past term debt repayment performance is a good predictor of future debt repayment capacity. The argument applies to the bank's entire current agricultural loan portfolios as well. There is a substantial difference in the behavior of risk factors evaluated during normal business conditions, and the behavior of these factors during financial crises.1

Risk to a financial institution manifests itself in a variety of ways, including interest rate risk, liquidity risk, investment risk, operating expense risk, collateral risk, and credit risk, to name a few. The focus of this paper is upon early signals of credit risk, and the implications for specific segments of the bank's portfolio. Specifically, this paper underscores the need for an early warning system for agricultural loans and the incorporation of this information into decisions regarding individual loan requests and portfolio management strategies. The concept of stress testing is combined with a proposed early warning system based upon generally accepted economic concepts to quantify the probability of cash flow stress for benchmark enterprises. This stress will be related to the probability of the borrower covering his/her operating costs and total costs, and the implications for the exposure to credit risk and portfolio risk. This system can be seen as an extension to quantitative history-based credit scoring models currently used by lenders, and can be enhanced to include the probability of default. It represents a means to mitigate future credit and supervisory risk.

Reliance on Historical Information

Three types of information typically support a lender's evaluation of a borrower's future debt repayment capacity:

  • (1) Historical financial indicators derived from previous annual financial statements are used along with information from credit bureau reports to assess the overall historical financial performance of the borrower. These indicators focus on past trends in the borrower's liquidity, solvency, efficiency, profitability and timeliness of loan payments. The credit bureau reports ignore information on the borrower's assets and profitability.2
  • (2) Comparative financial indicators available from public and private sources, or from internally developed benchmarks, help evaluate a firm's financial performance relative to other firms operating in the same industry. These indicators are often compared to the institution's underwriting standards to assess the creditworthiness of the borrower when applying for the loan and, in many cases, are used to price the loan.
  • (3) Credit scoring models, like the FICO scores available from Fair/Issacs, are based upon historically-determined weights assigned to specific historical factors, and are widely used to separate good loan applications from potential problem loans in the consumer lending and home mortgage lending markets (Hubbard and Gregg 2001). Their principal focus is on how well the applicant did in servicing existing debt commitments during a specific historical time period.

It is important at this point to clearly state that there is absolutely nothing wrong in reviewing a borrower's past loan payment record and how a set of financial indicators have either trended over time or stack up against similar firms. Historical and comparative analysis is important. Furthermore, credit scoring models have merit, although these models as applied in agriculture typically: (a) place more weight on equity than cash flows, (b) do not differentiate between sources of income and their potential variability, and (c) do not incorporate analyses of future profitability and debt coverage.

The "missing link" in traditional quantitative credit analysis therefore is not the coverage of historical information, but rather the almost complete lack of forward information utilized in the process. This is particularly important for longer-term strategic borrowing and lending decisions.

Too much has changed and will continue to change in global agricultural markets to rely solely on the "rearview mirror" approach to lending when going down the road with a borrower. Lenders need to look out the "windshield" as well to see where their farm customers are going and what key forks in the road may mean for problem loan exposure.

This requires the stress testing of plausible magnitudes associated with risk factors affecting borrower behavior and performance. These factors include changes in macroeconomic policy at both home and abroad, changes in international trade policies, changes in domestic farm commodity policy, and trends in global production and consumption patterns. These sources of stress will ultimately affect domestic commodity prices for crops and livestock, and hence the term debt repayment capacity of farm borrowers.3

Regulatory Perspectives

Regulators of farm lending institutions are aware of the need to stress test debt repayment capacity. During the height of the economic uncertainties facing U.S. agriculture in the late-1990s, the Office of the Comptroller of the Currency, the agency of the U.S. Treasury with responsibility for regulating and supervising national banks, issued a handbook to guide the actions of bankers and examiners in assessing credit risk associated with agricultural loans.4 As a part of this guidance, examiners were urged to take historic, current and prospective economic conditions into account when assessing the longer-term quality of agricultural loans. It also urged bankers to fully understand and actively manage their agricultural portfolios.

The Farm Credit Administration (FCA), an agency with the responsibility of regulating and supervising the banks and associations that comprise the Farm Credit System in the U.S., included language in their examination manual that urged member associations in a rapidly changing and competitive lending environment to manage their loan portfolios proactively.5 Both documents point to the need to monitor a variety of external factors that may affect future credit quality.6

The phrases "prospective economic conditions", "longer-term quality of agricultural loans", and "manage their loan portfolios proactively" are key phrases that signaled a realization that economic conditions in agriculture are affected by a variety of external risk factors that require ongoing analysis. It is significant that both documents underscored the need to combine historical analysis with pro forma analysis in assessing credit risk at the account and portfolio levels in an increasingly uncertain agricultural lending environment.

What Is an Early Warning System?

Rose (1995) states that "Not only is it useful to look at historical data in a source and uses of funds statement, but it is also extremely important to estimate the business borrower's future sources and uses of funds and its statement of financial position". It was also suggested that a lending institution would be well advised to carry out a simulation analysis of the customer's future financial condition, assuming an array of different environments, and seeing what the consequences are for the business's pro forma balance sheet, income statement, and sources and uses of funds statement. Rose (1995) further states that "Armed with this information, the loan committee can move toward a more satisfactory credit decision based on its assessment of the most likely future conditions."

Fraser, Gup and Kolari (1996) state that "Analyzing past financial data may not be adequate to make determinations about a borrower's ability to pay back loans in the future." They argue that, "By evaluating the current financial condition and using pro forma statements and sensitivity analysis, credit analysts have a better estimate of the borrower's ability to repay a loan."

The design and implementation of an early warning system should embody the pro forma stress testing efforts referred to above. The system should look at the probability of cash flow stress for key enterprises over a multi-year forecasting interval rather than a multi-year historical interval. In other words, today's portfolio and loan applications should be evaluated for their risk potential over the next five years, say, rather than the last five years. This is the time interval over which loan payments will be made.

What does an early warning system for agriculture look like? Can economic theory give us a hand in designing the system? What do we want the system to tell us? And how do we validate the early warning system once it has been estimated?

Proposed Early Warning System for Agriculture

The foundation of an early warning system for future nonperformance should rest on generally accepted economic principles. This section discusses the procedures for developing an early warning system for agricultural loans and portfolios in the United States.

Economic Foundation

Economic theory of the firm suggests that the breakeven (zero accounting profit) for the firm occurs where marginal revenue is equal to marginal cost. In Fig. 1(929), this occurs at quantity Q3 if the market price of the commodity is equal to P3. At this level of production, the farm is just covering its total costs. Average profit per unit of output (P3 _ ATC) at Q3 would be equal to zero. The farm would just cover its total cash costs at P2 by producing quantity Q2, and would just cover its variable costs (costs that vary with the level of production) by producing quantity Q1 if the price fell to P1.

The prospect of a zero net cash income (the price of the commodity is less than P2 when the farmer is producing Q2) implies the enterprise cannot contribute to the farm's ability to cover its scheduled principal payment in the current year. And if the price is less than P2, the firm cannot meet its total cash costs, which includes scheduled interest payments. In both cases, the farmer also is not covering the depreciation of productive assets, which has longer run implications for the soundness of the loan.

The design of the early warning system for agriculture proposed in this paper is built upon the cost concepts illustrated in Fig. 1(929). The focus of this system is on the key commodities in the agricultural loan portfolio, and the primary commodity in the farmer's operations. Rather than simulate each and every loan in the portfolio or each loan application, benchmark farms representing the typical cost structure for different sizes and locations of commodities in the bank's agricultural loan portfolio can be used to identify specific enterprises likely to experience cash flow stress over the foreseeable future.

Development of System

We can begin to develop an early warning system for agricultural loans and portfolios in the United States by using cost of production budgets for specific commodities assembled by extension economists in each state at the sub-state level. The state of Texas, for example, is segmented into 12 production districts. Cost of production budgets for major commodities produced in each of these Texas districts are available for different sizes of operations. This helps capture the effects of economies of size in the cost of production. These historical budgets, expressed on an average per acre or per head basis, can be updated with projected unit input costs for fuel and other inputs over a forecasting interval.

Similarly, the commodity price can be projected over the same forecasting interval by modeling the factors affecting demand and supply. Thus, rather than observing past prices and average cost of production when assessing accounting profit, one can project these observations over a desired time interval.7 Taking both projections together, we can determine whether or not an enterprise is projected to cover its cash cost of production.

Probability of Covering Costs

The approach taken in this paper is to project the commodity prices and unit input costs based upon random values for the exogenous variables in the econometrically estimated equations. Simulation of a multi-market set of simultaneous equations would capture the correlation or relationship between the commodities and inputs (Penson and Taylor 1992).

Use of a random number generator to assign values in the exogenous variables over a sufficient number of observations would permit derivation of a cumulative density function (CDF), illustrating the probability that the benchmark farm will not be able to cover its cash operating costs (net cash income is negative) and its total costs (net income is negative). Fig. 2(1013), for example, suggests that an enterprise is in a strong position to cover its cash operating costs since the right-hand CDF for coverage of operating costs lays far to the right of zero. The benchmark farm will contribute $170 per acre to debt repayment capacity from a cash flow perspective. However, the left-hand CDF for coverage of total costs shows there is roughly a 20 percent probability that the enterprise will not be able to cover its total accounting costs in 2005.

The closer the right-hand CDF for coverage of operating costs is to zero, the greater the current cash flow stress for the enterprise, the greater the cash flow stress for the farmer, and the greater the credit risk exposure for the lender. Once this process has been repeated for all the primary commodities in the bank's agricultural loan portfolio over a forecasting interval (say 2005-2010), the bank's portfolio database can be sorted by primary commodity of all borrowers (or at minimum all borrowers with a large exposure or debt outstanding) in specific locations. This would show the amount of loans outstanding associated with each primary enterprise. Those enterprises with CDFs lying near or to the left of zero have a high probability of experiencing cash flow stress in the near future, and should be closely monitored.

One can also determine the percentage share of the portfolio likely to be under increasing risk over time. For example, if a loan portfolio with $4 billion contains 15 percent of loans where wheat is identified as the primary commodity and the CDF for benchmark wheat farms indicates a high probability of building cash flow stress over the next five years, the early warning system would indicate potential cash flow stress associated with $600 million in current loans. A proactive bank management team can take steps to deal with this potential problem by: (1) rationing the quantity of loans made to wheat farmers, (2) charging a risk premium on long term loans to wheat farmers to compensate for the impending risk, (3) requiring additional security when making loans to wheat farmers, or (4) all of the above (Penson 1998). Regulators, armed with the same information, are more apt to begin examining large loans to wheat producers with a debt-to-asset ratio above levels known to result in nonperformance (Penson 1998).

Validation of the early warning system can be accomplished by back casting over a recent historical period. This would demonstrate how well the system did in identifying portfolio segments that actually came under increased financial stress and needed restructuring of their existing debt, and hence represented a heightened credit risk to lenders.

Event Stress Testing

The Basel Committee on global financial system defines stress testing as "a generic term describing the various techniques used by financial forms to gauge the potential vulnerability to exceptional but plausible events." The new Basel Capital Accord mentions that banks adopting an internal rating based (IRB) approach for calculating capital requirements must undertake sound stress testing procedures which "should involve identifying possible events or future changes in economic conditions that could have unfavorable effects on a bank's credit exposures and assessment of the bank's ability to withstand such changes."

This same system can also be used to evaluate the effects of a specific event in the magnitude, for example, of the Asian financial crisis in the late 1990s (Penson,1999; 2000). Examples include what might happen if a substantial outbreak of mad cow disease occurred in the United States or if the soybean rust outbreak in Brazil resulted in bank failures in Brazil and hence, in nonperforming loans at U.S. banks. While these events might not eventually occur, bank management and regulators would know more about the ability of the entire portfolio to perform under specific stressful conditions. Analysis of external factors through stress testing can help management assess the downside risk associated with the future debt repayment capacity and the ability of existing underwriting standards to correctly separate potential good loans from bad loans.

The goal of an early warning system is to recognize potential cash flow stress and adverse credit risk signals before actually making the loan or, in the case of an existing borrower or portfolio segment, before it is too late to take proactive measures.

Probability of Default

An enhancement to the early warning system proposed above would involve projecting the probability of default or nonperformance for a specific segment of the portfolio (e.g., borrowers with wheat as their primary commodity).9 Based upon historical observations of nonperformance (e.g., loan volume past due by 90 days or more) by primary enterprise and location, one can econometrically estimate the relationship between this loan volume and net cash income for the primary enterprise along with other variables, including data from credit report bureaus mentioned earlier. One can then generate a CDF for the probability of loan default based upon random events affecting the enterprise's net cash income over a forecasting interval.

Those portfolio segments with a high probability of default would thus require closer attention along the lines suggested above for the early warning system (quantity rationing, price rationing, and/or additional security of loans). Those portfolio segments with a low probability of default would represent potential marketing opportunities. An accurate forecast of the probability of default will also help bank management determine the appropriate level of loan loss reserves.

Summary and Conclusions

The information technology revolution, coupled with the gains in efficiency gained in loan approval processes, make the development of an early warning system and event stress testing possible. An early warning system can help management identify pockets of potential risk in time to be proactive in its lending and portfolio decisions. Major regulators in the United States, including the Federal Reserve System, the Federal Deposit Insurance Corporation, and the Office of the Comptroller of the National Bank Administration, all have early warning systems of one type or another that focus on bank portfolios as a whole. The particular features of the system recommended in this paper are designed specifically for agricultural loan portfolios at the bank level. At a minimum, such a system should be based upon generally accepted economic concepts, be pro forma or forward looking rather than historical in nature, capture the correlation among enterprises, and be validated before it is implemented. Extension to capture the probability of default is recommended where data are available.

Agricultural lenders in the United States have avoided greater levels of nonperforming loans due to the levels of government subsidies to farm program participants. Government payments as a percentage of net farm income reached as high as 60 percent during the financial crisis besetting U.S. agriculture in the 1980s, and approached nearly 50 percent in the late 1990s. The potential for more restrictive government spending limits in light of the current federal budget deficits underscores the need for an early warning system like that proposed in this paper to warning against future credit and supervisory risk.

Finally, advances in information technology permit more streamlined approaches to loan and portfolio analysis than were practiced in previous generations. The application of this technology, however, requires that banks gather the necessary data from borrowers and other sources to fully utilize its potential. The design of the agricultural loan database and supporting databases should contain current data on the economic performance of borrowers if lenders are to successfully deal with the many risks faced in a rapidly changing global economic environment.

References

  • Agricultural Lending, published by the Office of the Comptroller of the Currency, Washington, D.C., 1999.
  • Basel Committee on Banking Supervision, Overview of the New Basel Capital Accord, Bank for International Settlements, April 2003.
  • Examination Manual, published by the Farm Credit Administration, McLean Virginia, 1998.
  • Fraser, D.R., B.E. Gup and J.W. Kolari. 1996. Commercial Banking: The Management of Risk, West Publishing Company, Minneapolis/St. Paul.
  • Hubbard, M. and S. Gregg. 2001. NextGen FICO Scores: More Predictive Power in Account Management, FairIssac Corporation.
  • Penson, J.B., Jr. and C.R. Taylor. 1992. "United States Agriculture and the General Economy: Modeling Their Interface." Agricultural Systems Journal, Volume 39, No.1.
  • Penson, J.B., Jr. 1997. "Incorporating Forward Information in Credit Analysis." Journal of Agricultural Lending, Volume 11, Issue 1.
  • Penson, J.B., Jr. 1998. "Loan Pricing and Increasing Risk." Journal of Agricultural Lending, Volume 11, Issue 3.
  • Penson, J.B., Jr. 1999. "Stress Testing Agricultural Loans", Journal of Lending and Credit Risk Management, Volume 82, Number 4.
  • Penson, J.B., Jr. 2000. "Stress Testing Agricultural Loan Portfolios", Journal of Lending and Credit Risk Management, Volume 83, Number 1.
  • Rose, P.S. 1995. Commercial Bank Management, Third edition, The Irwin Series in Finance, Richard D. Irwin Publishing Company, Chicago. pp.596-598.
  • Stam, J.M. and B.L. Dixon 2004. "Farmer Bankruptcies and Farm Exists in the United States, 1899-2002", Agriculture Information Bulletin No. 788, Economic Research Service, U.S. Department of Agriculture.

Footnotes:

1 For background information on U.S. farm bankruptcies, see Stam and Dixon (2004).

2 There are three major credit bureaus in the United States that record all loan and credit transactions of borrowers. Lenders can obtain reports from these bureaus that indicate the amount of loans outstanding, the timeliness in which payments are made, and other information pertaining to a borrower's use of credit.

3 This in turn will also affect the valuation of productive farm assets pledged as collateral for term loans.

4 See Agricultural Lending, published by the Office of the Comptroller of the Currency, Washington, D.C., 1999.

5 See section EM-310 of the Examination Manual, published by the Farm Credit Administration, McLean Virginia, 1998.

6 The OCC guide refers to interconnected risks that affect agriculture, including weather and disease, crop prices (and livestock prices), overseas markets, financial markets and an overabundant supply. The FCA manual states the "volatile commodity prices, global economic conditions, the agricultural credit market, pricing practices and services of competitors, interest rates, farm programs, existing and proposed regulations, technology, climatic conditions and commodity markets should be analyzed" (Section EM-310, page 3 of 19).

7 Accounting profit excludes a number of cost items normally included in the definition of economic profit, such as opportunity cost and the value of unpaid operator and family labor. While these noncash costs are appropriate in may settings, our focus here is on the ability of the enterprise to contribute to the debt repayment capacity of the farm, and potential cash flow stress.

Index of Images

Figure 1 Cost Concepts for Early Warning System for Agriculture

Figure 1 Cost Concepts for Early Warning System for Agriculture

Figure 2 2005 Coverage Per Acre (Head)

Figure 2 2005 Coverage Per Acre (Head)

Download the PDF. of this document(633), 192,450 bytes (188 KB).