volume 57 article #2

Lending Relationships, Customer Loyalty, and Competition in Agricultural Banking

Peter J. Barry, Paul N. Ellinger, and LeeAnn McEdwards Moss

Peter J. Barry is a professor of agricultural finance, Paul N. Ellinger is an assistant professor of agricultural finance, and LeeAnn McEdwards Moss is a graduate student and research specialist, all with the Center for Farm and Rural Business Finance, Department of Agricultural and Consumer Economics, University of Illinois at Urbana-Champaign.

Abstract

Data from a recent survey of midwestern agricultural banks are used to investigate the influence of the competitiveness of lending markets in agriculture on the relationships between lenders and borrowers. The effects of market competitiveness assessments, potential products and services lenders could offer to their borrowers, and other institutional characteristics are investigated for farm real estate and non-real estate farm borrowers using an ordered probit regression model. The empirical results clearly indicate an inverse relationship between competition and borrower loyalty, which serves as a proxy for the lender-borrower relationship.

Key words: agricultural bankers, competition, lender-borrower relationships.

Article <top>

Increases in competition generally are considered to improve market efficiency, diminish excessive profits, and allow firms to operate closer to optimal size and minimum long-run average costs of production. Greater competition is conditioned by widespread availability of high-quality information, numerous buyers and sellers, a homogeneous product, unfettered entry and exit from the industry, and producers and consumers motivated, respectively, to maximize profits and utility. Common stocks, bonds, and other financial assets with numerous transactions in well-developed secondary markets are considered to have highly competitive and efficient markets. In contrast, business and consumer loans made in more isolated markets may experience less competition and lower efficiency.

In some cases, however, greater competition may be disadvantageous. Petersen and Rajan (1995) cite the potentially adverse effects of competition on longer term lender-borrower relationships. Supported by empirical evidence for small, nonfarm businesses, Petersen and Rajan argue that the diminished capacity for the formation of lender-borrower relationships in more competitive markets will hamper newer firms or financially stressed firms that require time to establish or restore profitability. Less competitive markets better enable lenders to grant short-run concessions in financing terms to these disadvantaged firms while adjusting financing in more favorable times in order to share in the future surplus of the borrowing firm and maintain profitable, long-term loan performance in present-value terms. Higher competition, according to Petersen and Rajan, precludes these elements of lender flexibility, and may require newer or financially stressed firms to compete for funds on the same (or even more stringent) terms as established, successful firms.

In assessing agricultural finance markets, it is plausible that farmers prefer access to credit that is competitively priced, versatile in use, reliably available over time, accompanied by useful financial services, and matched well with business cash flows (Barry, Ellinger, Hopkin, and Baker). In the U.S., these credit preferences have been consistent with the role of the specialized Farm Credit System (FCS), government farm loan programs, and the historically significant role of localized, community-oriented agricultural banks in strong agricultural regions of the country. These institutional arrangements have tended to accommodate farmers’ interests, foster close ties between agricultural lenders and rural borrowers, and contribute to information-intensive, lender-borrower relationships. As Petersen and Rajan (1995) suggest, however, the changing competitiveness of rural financial markets could have both advantageous and disadvantageous effects for rural borrowers.

Our goal in this study is to test the hypothesis that lenders’ perceptions of customer loyalty, serving as a proxy for lender-borrower relationships, are significantly influenced by the competitive-ness of lending markets, taking account of institutional and market characteristics. The data come from a survey of agricultural bankers in three midwestern states of the U.S. and from bank call reports (Moss, Barry, and Ellinger). An ordered probit model is the appropriate econometric technique, based on the discrete and ordinal nature of the loyalty variable in the survey. In the following sections we review related literature, rationalize loyalty as a proxy for relation-ships, describe the survey and analytical approach, and present the empirical results and consider their implications.

Related Studies <top>

As past studies have shown, relationships between banks and their customers are inherently linked to competition, although the focus of the linkage has changed over time. In a 1963 study, Hodgman introduced the customer relationship concept by showing the importance of demand deposits as a source of a bank’s capacity to lend and invest and the resulting importance of the bank’s relationship to loan customers who hold demand deposits. Wood’s 1975 publication extended Hodgman’s customer-deposit relationship to multi-periods by showing how a liberal lending policy may induce increases in future deposits that can, in turn, be loaned or invested. Wood also added the customer-loan relationship which suggests that a bank’s current lending policy influences its future loan demands. In 1978, Barry applied these concepts to quantitatively estimate the strength of the loan-deposit relationships.

In a 1990 analysis, Sharpe considered asymmetric information as a determinant of customer relationships attributable to a bank’s monopoly power over its established, higher-profit borrowers who become "informationally captured" by the bank. The adverse efficiency consequences of this informational imperfection are reduced by implicit contracts arising from the bank’s efforts to create a reputation as a reliable lender. The terms of such contracts are dependent on the degree of informational advantage, reputational perceptions, and other determinants of customer profitability to the bank. Sharpe contrasted his ex-post information-driven relationship theory to an alternative justification suggested in the late 1970s by Wachter and Williamson, based on the existence of ex-ante relationship-specific capital investment created by the pre-loan customer evaluation.

In 1996, Hanson, Robison, and Siles examined the lender-borrower relationship in terms of the social capital created between the two parties, as a result of various forms of social closeness in business, community, and personal dealings (e.g., business colleagues, friends, family members). Their results of bankers’ surveys in Michigan indicate that the lender-borrower relationship at a specific institution is increasingly solidified as social capital and closeness increase.

Petersen and Rajan also considered the interactions between lender-borrower relationships and financial markets in two recent publications. Their 1994 Journal of Finance article utilizes data from a national survey of small, nonfarm businesses to determine that lender-borrower relationships may have significant effects on credit availability and less significant effects on credit costs. The authors’ 1995 Quarterly Journal of Economics study uses the same data base to test the interactions between lending competition and the availability and cost of credit for young or financially stressed borrowers, both of which are hypothesized to benefit from stronger lender-borrower relationships in more concentrated markets. Their results are summarized as follows:

Young firms in concentrated markets receive more institutional finance than do similar firms in competitive markets. As firms get old, the difference in the relative firm borrowing from institutions disappears. Young firms who get institutional loans are more indebted in concentrated markets than in competitive markets, but this pattern reverses for older firms. Creditors seem to smooth interest rates over the life cycle of the firm in a concentrated market, charging a lower-than-competitive rate when the firm is young and a higher-than-competitive rate when the firm is old (Petersen and Rajan 1995, p. 439).

These findings support the authors’ theory that competition and longer term lender-borrower relationships are not necessarily compatible.

The Approach Used in This Study <top>

Our study is motivated by Petersen and Rajan’s (1995) work, but with a different data-generation approach based on self-assessments by surveyed agricultural bankers about the degrees of lending competition in their market place and the strength of loyalty of their farm real estate and non-real estate farm borrowers. Thus, customer loyalty serves as a proxy for the relationship between the lender and a borrower, and the effectiveness of this proxy choice should be considered further.

Relationships are defined by Petersen and Rajan as "close ties" between the borrower and the lender, while loyalty (according to Webster’s Collegiate Dictionary, 1991 edition) has elements of faithfulness and allegiance by one party to another. Thus, relationship and loyalty are closely related, but not identical concepts. Intuition suggests that lender-borrower relationships will be stronger as loyalty increases, although relationships still may exist in the absence of loyalty due to the informational advantages cited earlier, the scarcity of lenders available in specific financial markets, or other factors. The plausibility of the positive linkage between loyalty and relationships is strengthened by the responses to this survey in which bankers ranked "relationship with loan officer" as the most important factor determining customer loyalty. (Other elements in order of importance are staff knowledge of agriculture, the rural attitude of the bank, and the stability of the institution and its staff.) Thus, measures of customer loyalty may serve as the strongest proxy for lender-borrower relationships, in comparison to Petersen and Rajan’s use of such indirect proxies as length of borrowing, use of debt from financial service providers, and the number of institutions from which a firm borrows.

Specifically, the bankers were asked in the survey to rate the loyalty of their farm real estate and non-real estate farm borrowers on a five-point Likert scale (with 1 = lowest loyalty and 5 = highest loyalty). The bankers also were asked to estimate how competitive other lenders in the market place are with their bank, and the importance of interest rates to this competition. Degrees of competition and the importance of interest rates also were elicited using the five-point Likert scale approach (where 1 = lowest importance and 5 = highest importance). Other questions in the survey addressed the importance of various products and services banks could offer to their customers, the banks’ financial and operating characteristics, market size, holding company affiliation, and the banks’ outlook for financing agriculture.

Survey Procedures and Respondents <top>

A total of 982 banks in Illinois, Indiana, and Iowa were surveyed in the summer of 1995.¹ Agricultural banks in the three-state region were defined as having a ratio of agricultural loans to total loans exceeding 25%, or agricultural loan volume exceeding $2.5 million. A total of 222 banks responded to the survey for an overall response rate of 23%. The response rates by state were 23.5% from Illinois, 24.4% from Indiana, and 21% from Iowa.

¹A copy of the survey is available from the authors upon request.

Profiles of the responding banks indicate that about 73% had total assets of less than $100 million, while 45% had assets less than $50 million. Asset size ranged from $6.2 million to $2 billion, with a mean of $106.1 million and a median of $55 million. Eighty-eight percent of the banks reported a local lending market having an average radius of 29 miles, in contrast to a regional, statewide, or national market. The average agricultural loan ratio of the respondent bank was 31%.² Thus, the population of banks surveyed and banks responding to the survey are dominated by smaller rural banks in local markets heavily involved in financing agriculture.

²The mean values for asset size and agricultural loan ratio for the responding banks were not signifi-cantly different from the mean values of the population of surveyed banks at the 95% confidence level.

Borrower Loyalty and Competition <top>

The distributions of the bankers’ ratings of borrower loyalty are shown in Table 1, indicating relatively few incidences of "low loyalty," but higher degrees of dispersion between the high loyalty (4 and 5) and medium loyalty (3) ratings. The results also show a slightly higher loyalty rating for non-real estate borrowers relative to real estate borrowers. The latter finding likely reflects a more frequent and more information-intensive set of contacts between lenders and borrowers for non-real estate loans than for real estate loans.

Distributions and mean values for the competitiveness ratings and the importance of interest rates to competition are shown in Tables 2 and 3, respectively. The respondents, on average, considered other commercial banks equally as competitive as the Farm Credit System in farm real estate lending, but significantly (at the 99% level) more competitive in non-real estate lending—perhaps reflecting the differences in market shares of these institutions in the respective lending markets.³

³The FCS holds 31.4%, 21.5%, and 27.4% of the farm real estate debt in Illinois, Iowa, and Indiana, respectively. Commercial banks hold 37.5%, 32.3%, and 31.5% of the farm real estate debt in Illinois, Iowa, and Indiana, respectively (U.S. Department of Agriculture 1997). Also, the FCS involvement in non-real estate farm lending is relatively minor in two of the surveyed states. At year end 1994, for example, the FCS shares were 6.85% in Illinois, 6.25% in Iowa, and 16.27% in Indiana, compared to a 16.17% share nationwide (U.S. Department of Agriculture 1996).

 

Moreover, machinery dealers were rated by the bankers as similarly competitive to FCS institutions in non-real estate lending. Interest rates are of highest average importance for competition with the FCS and other banks for farm real estate loans, while interest rates are of highest average importance for competition with other banks, machinery dealers, and the FCS in non-real estate farm lending (Table 3).

Regression Analysis <top>

The empirical analysis utilizes an ordered probit technique to regress each respondent’s loyalty ratings for farm real estate (Model 1) or non-real estate (Model 2) farm borrowers against three groups of independent variables representing different attributes of the respondent bank. The models have the following conceptual form, with linear, additive relationships among the respective independent variables in each group:

where y is the ordered, discrete loyalty variable having a value of 3 for the high loyalty ratings of 4 and 5, a value of 2 for medium loyalty, and a value of 1 for the low loyalty ratings 1 and 2. Due to the low number of observations in the highest and lowest categories, and to aid in the interpretation of the ordered probit model, the dependent variable for each regression was reduced from five classes to these three classes.

The G₁, G₂, and G₃ notations represent the three sets of independent variables appropriate to each of the two models. The group G₁ (market competition) independent variables describe the bankers’ assessments of the competitiveness of alternative lenders in their market area—the fundamental issue motivating this study. Survey respondents rated other commercial banks and the Farm Credit System as the two strongest competitors in their farm real estate lending markets, in terms of mean competitiveness ratings. They rated other commercial banks and machinery dealers as their two strongest competitors in non-real estate farm lending.⁴ These market competition variables are included in the farm real estate and non-real estate models, respectively. As established earlier, the anticipated relationship is higher loyalty associated with lower competition.

⁴The non-real estate competitiveness rating for the Farm Credit System also was evaluated in the probit model. However, it did not exhibit a significant relationship in the non-real estate lending model.

Group G₂ (products and services) contains a set of products and services banks could employ to maintain competitive agricultural lending positions and build greater borrower loyalty. In the survey, the bankers were asked to rate the importance of 21 of these services on a 1­5 scale. Univariate analysis and practical significance were considered in identifying candidate variables for inclusion in each regression model, since incorporating all the variables would result in multicollinearity problems. The farm real estate borrower model includes the banks’ rating of the importance of: financial management and consulting services, insurance products, a specialized agricultural program, farm management services, estate management services, and interest rate adjustments for prepayment risks. The non-real estate farm borrower model includes the importance of long-term service from the same loan officer and the use of guaranteed loans. The anticipated relationship is that greater importance of services is associated with more loyal borrowers.

Bank attributes group G₃ (other institutional characteristics) includes other bank characteristics: multi-bank holding company affiliation and bank size (in total assets). A dummy variable for holding company affiliation distinguishes between banks affiliated with multi-bank holding companies (given a value of one), and independent banks or those affiliated with single-bank holding companies (assigned a value of zero). Bank size is measured by log of total assets (in $000s). Among these institutional variables, a weaker relationship is expected for larger and affiliated banks, reflecting their larger market areas and the likelihood of greater uniformity in lending policies.

The use of the ordered probit model reflects the discrete and ordinal nature of the borrower loyalty variable in each regression (Zavoina and McElvey). Caution must be used in interpreting coefficients from an ordered probit model (Greene). Only the signs of the changes in the highest and lowest categories can be interpreted unambiguously from the model coefficients. Thus, marginal elasticities are often reported. In this study, we focus the discussion on the marginal effects of high loyalty.

The results of the two regression models are presented in Table 4.⁵ A positive (negative) coefficient for an independent variable indicates that an increase in the variable will result in an increase (decrease) in the probability of being in class 3 (high loyalty) and a decrease (increase) in the probability of being in class 1 (low loyalty). The directional effect on class 2 cannot be explicitly determined through the sign of the regression coefficients (Greene). Furthermore, the marginal effects of the independent variables on the probability of being in the three classes are not equal to the coefficients.

⁵ Two additional and separate models were evaluated that included only the significant variables from each model. All variables remained statistically significant and their coefficients were the same sign as in the reported models.

The marginal effects of the independent variables in the real estate and non-real estate lending models are reported in Tables 5 and 6, respectively. Since many of the regressor variables are discrete values (ratings), the marginal effects reported in Tables 5 and 6 are the changes in the probability of being in a loyalty class resulting from a one-unit (rating level) change in the specific variable while other values are held at their sample mean values.⁶ For example, the -11.59% value for competitiveness of the FCS in farm real estate lending in the high loyalty column of Table 5 indicates that the change in probability of being in the high loyalty class decreases 11.59% as the banks’ rating of the competitiveness of the FCS in farm real estate lending increases from 3 to 4. The one-unit increase in the competitiveness of the FCS in farm real estate rating increases the probability of being in the low and medium loyalty classes 1.70% and 9.89%, respectively. The values of the marginal effects in each row of Tables 5 and 6 will sum to zero following from the requirement that probabilities must sum to one. Thus, the values in Tables 5 and 6 can be used to interpret the directional effects and the practical significance of the independent variables, while the results in Table 4 can be used to show their statistical significance.

⁶ The one-unit changes in the discrete values are evaluated between the two discrete classes closest to the sample mean for the discrete variable. One-unit changes at the extreme values also were evaluated. No qualitative differences in the results occurred.

 

The competitiveness of the FCS in the farm real estate lending model has a significant (at the 99% level) negative effect on the loyalty of these banks’ farm real estate borrowers (Table 4). Although the influence of the competitiveness of other commercial banks is also negative as anticipated, it is not statistically significant. Among the products and services, the banks’ rating of the importance of financial management and consulting services and the banks’ rating of the importance of offering a specialized agricultural program to customers have a significant positive effect on their loyalty. The coefficients of all other products and services variables are not statistically significant. Bank size does not exhibit a significant effect on farm real estate borrower loyalty. However, affiliation with a multi-bank holding company does have a negative relationship, indicating affiliation with a multi-bank holding company reduces the probability of high customer loyalty by 12.83% (Table 5).

In the non-real estate farm borrower loyalty model (Tables 4 and 6), market competition is also an important influence on borrower loyalty. The loyalty of non-real estate farm borrowers is significantly affected (90% level) by the competitiveness of other commercial banks. The banks’ rating of the importance of long-term service from the same loan officer is positively and significantly related to loyalty. Bank size and holding company affiliation are not significantly related to customer loyalty.

Concluding Comments <top>

In general, the empirical results reported here are consistent with the anticipated inverse relationships between market competition and borrower loyalty, which serves as a proxy for the strength of the lender-borrower relationship. Bankers in more competitive farm real estate and non-real estate lending markets tend to have less loyal customers, irrespective of other institutional and market characteristics. These results about lending relationships and competition in agricultural lending are consistent with those of Petersen and Rajan (1995). Data limitations, however, do not allow explicit evaluation of the impacts on younger or financially stressed borrowers.

Several potentially offsetting factors can be considered in assessing the implications of the loyalty/competition link in agricultural lending. Despite greater competition, evidence also suggests that rural financial markets are more concentrated and less competitive than their urban counterparts (Collender; Barry and Ellinger). Agricultural and rural business lending may represent niche markets for many community banks in which specialization is conducive to relationship building, targeted skills in financial analysis, and the types of informational advantages cited earlier by Sharpe. The negative effects of greater competition on borrower loyalty also may be mitigated by an increased service orientation tailored to the characteristics of different types of borrowers.

Finally, the Farm Service Agency (formerly the Farmers Home Administration) and various state credit programs have responded to the needs of high-risk, often younger but potentially creditworthy farmers through direct loans and guarantees of loans made by commercial lenders. Thus, lender-borrower relationships may continue to play important roles in more competitive, but not perfectly competitive, agricultural lending markets.

References <top>

Barry, P.J. "Rural Banks and Farm Loan Participation." Amer. J. Agr. Econ. 60(May 1978):214­24.

Barry, P.J., and P.N. Ellinger. "Liquidity and Competition in Rural Credit Markets." In Financing Rural America. Kansas City, MO: Federal Reserve Bank of Kansas City, 1997.

Barry, P.J., P.N. Ellinger, J.A. Hopkin, and C.B. Baker. Financial Management in Agriculture, 5th ed. Danville, IL: Interstate Publishers, Inc., 1995.

Collender, R. "Can Federal Action Improve Efficiency in the Market for Farm Loans?" Agr. Info. Bull. No. 724-01, U.S. Department of Agriculture,

Economic Research Service, Washington, DC, March 1996.

Greene, W.H. Econometric Analysis, 2nd ed. New York: Macmillan Co., 1993.

Hanson, S.D., L.J. Robison, and M.E. Siles. "Impacts of Relationships on Customer Retention in the Banking Industry." Agribus.: An Internat. J. 12(January/February 1996):27­35.

Hodgman, D.R. "Commercial Bank Loan and Investment Policy." Bureau of Economic and Business Research, University of Illinois, Urbana-Champaign, 1963.

Moss, L., P.J. Barry, and P.N. Ellinger. "Financing Agriculture: Competitive Challenges for Commercial Banks." Center for Farm and Rural Business Finance, University of Illinois, Urbana-Champaign, March 1996.

Petersen, M., and R. Rajan. "The Benefits of Lending Relationships: Evidence from Small Business Data." J. Finance 45 (1994):3­37.

———. "The Effect of Credit Market Competition on Lending Relationships." Quart. J. Econ. 110(May 1995):407­43.

Sharpe, S.A. "Asymmetric Information, Bank Lending, and Implicit Contracts: A Stylized Model of Customer Relationships." J. Finance 45(1990): 1069­87.

U.S. Department of Agriculture, Economic Research Service. "Farm Business Economics Report, 1994." Pub. No. ECI-1994, USDA/ERS, Washington, DC, August 1996.

U.S. Department of Agriculture, National Agricultural Statistics Service. Agricultural Statistics, 1996. Washington, DC: U.S. Government Printing Office, 1997.

Wachter, M.L., and B. Williamson. "Obligational Markets and the Mechanics of Inflation." Bell J. Econ. 9(1979):549­71.

Wood, J.H. Commercial Bank Loan and Investment Behavior. New York: John Wiley and Sons, 1975.

Zavoina, R., and W. McElvey. "A Statistical Model for the Analysis of Ordinal-Level Dependent Variables." J. Mathematical Sociology 4(Summer 1975):103­20.

<top>

 

---

Send questions and comments to Faye Butts fsb1@cornell.edu

This page was last modified on: 02/10/04