United States v. Household Finance Corporation, Hfc American, Inc., and American Investment Company

SPRECHER, Circuit Judge. The issue raised by this appeal is whether the business of making direct cash loans by finance companies is a line of commerce within the meaning of section 7 of the Clayton Act, 15 U.S.C. § 18. More specifically, the issue is whether finance companies effectively compete with other financial institutions, e. g., banks, credit unions, and savings and loan associations, in the extension of loans. The district court found that personal loans from finance companies do not constitute a separate line of commerce. We hold, however, that the district *1257 court in reaching this conclusion both incorrectly applied legal standards and made clearly erroneous findings of fact. Accordingly, we reverse. I On January 9, 1979, the United States filed a civil action against Household Finance Corporation (HFC), one of its subsidiaries, and American Investment Company (AIC). The complaint alleged that a proposed merger between HFC’s subsidiary and American Investment Company would violate section 7 of the Clayton Act, 15 U.S.C. § 18. This violation was premised on the fact that the proposed acquisition would allegedly lessen substantially competition in the making of “direct cash loans” by finance companies in numerous sections of the country. On February 5 the parties agreed to the following stipulation for the purpose of narrowing the issues to be tried. 1. The only issue to be tried in this case is whether the business of making direct cash loans by finance companies is a line of commerce within the meaning of § 7 of the Clayton Act. 2. If the court finds that the business of making direct cash loans by finance companies is a line of commerce in the United States generally, the acquisition of AIC by HFC would constitute a violation of § 7 of the Clayton Act in various sections of the country and a permanent injunction against the acquisition of any stock in AIC by HFC directly or indirectly should be entered. 3. Such injunction should require divestiture of the existing stock ownership in AIC by HFC and prohibit the acquisition of any finance company assets of AIC by HFC directly or indirectly without the approval of the Department of Justice or of the court. 4. If the court finds that the business of making direct cash loans by finance companies does not constitute a line of commerce in the United States generally, the acquisition of AIC by HFC would not constitute a violation of § 7, and an order of the court dismissing this action should be entered. The issues having been thus narrowed, the district court conducted a 13-day trial. During this trial testimony was introduced from various government officials who regulate financial institutions, executives and managers of financial institutions, and professional economists. The district court then made its findings of fact and concluded that the making of direct cash loans by finance companies was not a separate line of commerce. The district court opinion begins by conceding that at one time consumer finance companies served a distinct consumer population consisting of “a lower-income and higher-risk consumer market than [was served by] the commercial banks, savings and loan associations, and other lending institutions.” However, the district court believed that recent expansions by other institutions in lending activity- with respect to this lower-income, higher-risk group “has created a very substantial overlap of a number of competitors seeking and serving the same consumer market as was once deemed to be the exclusive province of the consumer finance companies.” Various facts were relied on by the district court in support of this conclusion. First, the court cited statistics demonstrating the increasing involvement of commercial banks, credit unions, and savings and loan associations in the consumer installment market. Whereas the finance company market share, based on dollar amount of loans, decreased between 1967 to 1978 from 30.9 percent to 19.6 percent, the market share of banks increased from 41.7 percent to 49.9 percent and the market share of credit unions increased from 11.3 percent to 16.8 percent. The court also cited figures which, after adjustment for inflation, showed that real dollar amounts of bank credit during this period increased 99.8 percent; of finance company credit, 5.8 percent; and of credit union credit, 147.9 percent. The court sought to explain these alterations in market shares by finding that consumers were increasingly aware of in *1258 terest rates, thereby giving a competitive advantage to lower-cost forms of credit. Next, the court analyzed the competitive inroads made on consumer finance company business by credit cards and credit unions. Although the court made no findings as to how many customers of consumer finance companies could obtain, or had obtained, credit cards on which there were sufficient balances to meet the consumer’s credit needs, 1 the court found that these cards have had a “major” impact on the credit market and have served as “the vehicle for permitting commercial banks to move aggressively into the consumer credit market . .” Further, the court found that obtaining cash advances through credit cards was more convenient than, and a substitute for, obtaining a cash loan from a finance company. As to credit unions, the court found that they have experienced “impressive” growth and have been effective competitors with finance companies because of their lower rates as well as their ability to utilize payroll deductions. The court further found, however, that as much as 50 percent of the population had no access to credit unions. The court then proceeded to find that, in terms of several general indicia, finance company loans and bank loans were becoming increasingly similar. The average amount of finance company loans was found to be increasing while the higher average amount of a bank loan was decreasing. Likewise, finance company delinquency and charge-off rates were declining while the lower bank rates were increasing. Interest rates were said to be converging as well. Further, the credit scoring systems used by banks and finance companies to» determine loan eligibility overlap so that “finance companies in the main do not have higher or lower credit standard [sic] than other types of financial institutions.” The court also found a “substantial overlap between customers of finance companies and credit unions, thrift institutions, and retailers.” Finally the district court relied on findings by government agencies that banks and finance companies compete. Both the Federal Reserve Board and the Justice Department have taken the position in evaluating bank-finance company mergers that banks and finance companies compete. Based on these findings the district court concluded that the making of direct cash loans by finance companies was not a line of commerce within section 7 of the Clayton Act. Thus, under the terms of the stipulation, the proposed merger would not have an anticompetitive effect. Accordingly, the district court dismissed the complaint. II Although the district court did not explicitly set out the legal standard by which it should be determined whether the business engaged in by finance companies is a distinct line of commerce, it is clear what that standard should be. The Supreme Court has repeatedly held that a financial institution comprises a separate product market if it offers to a “significant” number of consumers a “cluster of products and services” that competing financial institutions do not. United States v. Connecticut National Bank, 418 U.S. 656, 664, 94 S.Ct. 2788, 41 L.Ed.2d 1016 (1974); United States v. Phillipsburg National Bank, 399 U.S. 350, 360, 90 S.Ct. 2035, 26 L.Ed.2d 658 (1970); United States v. Philadelphia National Bank, 374 U.S. 321, 356-57, 83 S.Ct. 1715,10 L.Ed.2d 915 (1963). Thus, for example, Connecticut National Bank concluded that banks were a separate line of commerce because they could provide services to commercial customers more effectively than savings and loan associations. The factors supporting this conclusion were the overwhelming dominance of banks in the commercial loan market (1 billion dollars versus 26 million dollars) and the inability of savings and loan associations, even under pending legislation, to offer checking accounts *1259 to commercial customers. 418 U.S. at 664-66, 94 S.Ct. 2788. Accordingly, if, as the government alleges, finance companies provide loans, counseling and other financial services to lower-income, higher-risk customers which other financial institutions do not, the standards set out by the Supreme Court would compel a finding that cash loans by finance companies constitute a line of commerce within the meaning of section 7. 2 Furthermore, Phillipsburg National Bank and Connecticut National Bank made it clear that the fact that banks and other financial institutions compete with finance companies in the provision of other services, such as the extension of loans to moderate or low risk consumers, is not relevant to the inquiry here. These other service areas may be relevant “in analyzing the effect on competition between a commercial bank and another type of financial institution,” Phillipsburg National Bank, 399 U.S. at 360, 90 S.Ct. at 2041, but in determining the effect of competition in mergers of one type of institution, it is the effect in any area of unique services that must be considered. 3 Phillipsburg National Bank, 399 U.S. at 360-61, 90 S.Ct. 2035; Connecticut National Bank, 418 U.S. at 663 n.3, 94 S.Ct. 2788. It should also be noted, as an introductory matter, that, although the Supreme Court may not have fully considered and finally concluded the matter, it has stated in two of the cases cited above that banks do not compete with finance companies in the extension of cash loans. Thus, in Philadelphia National Bank the Court declined to include banks and finance companies in the same market. The Court stated: Some commercial banking products or services are so distinctive that they are entirely free of effective competition from products or services of other financial institutions; the checking account is in this category. Others enjoy such cost advantages as to be insulated within a broad range from substitutes furnished by other institutions. For example, commercial banks compete with small-loan companies in the personal-loan market; but the small-loan companies’ rates are invariably much higher than the banks’, in part, it seems, because the companies’ working capital consists in substantial part of bank loans. 374 U.S. at 356, 83 S.Ct. at 1737-1738. 4 This passage was cited with approval in Phillipsburg National Bank, 399 U.S. at 361 n.4, 90 S.Ct. 2035. The record before us substantiates that banks continue to offer loans at a lower cost than finance compa *1260 nies, 5 thus making currently applicable the Supreme Court’s finding that personal loans made by banks are “insulated” from competition by finance companies and that the two thus constitute separate product markets. Applying the “unique product or service” standard, an examination of the record before us further supports the conclusion that the extension of cash loans by a finance company constitutes an independent line of commerce. We note first that very few of the district court’s findings affect this conclusion. Most of the district court findings merely indicated a trend for banks to increase their loan activity with respect to some customers who formerly had been serviced only by finance companies. As was made clear in Connecticut National Bank, a trend towards an overlap in services in and of itself is not sufficient to support a finding that sufficient overlap currently exists to preclude a finding that one type of institution offers a “unique cluster of products and services.” 6 Thus, although we do not dispute the district court’s findings of increasing market shares of banks, savings and loan associations, and credit unions or of converging interest rates, write-off rates and loan amounts, we do not believe these findings to be legally sufficient to support a conclusion that finance companies are not engaged in a distinct line of commerce. Only one finding was made by the district court that would satisfy the “unique service” standard. The district court stated: Although it may have been true in the past, statistics today demonstrate that there is no significant market segmentation on the basis of risk with regard to the making of loans by various types of financial institutions . . . . Finance companies, banks, and credit unions all use substantially the same criteria in making credit decisions, and finance companies in the main do not have a higher or lower credit standard than other types of financial institutions. Neither do finance companies lend to any significant percentage of their customers who would be ineligible to borrow from other types of institutions. An analysis of several studies available in the field reveals that significant percentages of finance company borrowers are eligible for or actually have non-mortgage loans at banks, that the majority of customers of consumer finance companies have bank credit cards. We believe, however, that the district court erred as a matter of law in concluding that no significant market segmentation exists. 7 The thrust of the district court’s conclusion, thus, is that although eoncededly some customers are uniquely served by finance companies, the number of such customers is insignificant. Obviously evaluating this conclusion is made difficult by the failure of the district court to specify precisely how large that group is or to detail the bases on which it concluded that this group is insignificant. Examining the record we find *1261 only the following bases for such a conclusion: two multivariate analyses of risk characteristics of bank and finance company customers, a comparison of several bank and finance company credit scoring systems, opinion testimony by bankers and finance company officials concerning credit practices, and figures showing the number of finance company customers with various types of bank credit. We will examine each of these bases in turn. First, the defendants introduced at trial two multivariate analyses of the risk characteristics of a sample of individuals who were borrowers at éither banks or consumer finance companies. In these analyses the authors examined several hypothesized variables of risk and attempted to determine if those variables were correlated with whether the individuals in the sample borrowed from a bank or a finance company. Based on that analysis a model was derived which would predict whether a given borrower, depending on his risk characteristics, was a borrower from a bank or a consumer finance company. That model was then applied to the sample from which it was derived to determine how well it would classify the sample members. In both studies, the authors noted some difficulty with the ability of such an equation to accurately classify members as bank or finance company borrowers. Neil Murphy, author of one of the studies, merely testified’ that the equation was “not able to classify the borrowers into categories terribly well.” Tr. 1508. Gregory Boczar testified that when his model equation predicted that a sample member was a bank debtor, it would be right nine out of ten times. However, when his equation predicted that a sample member would be a finance company debtor it would be right only one out of three times. (Tr. at 1126-28). Both authors therefore concluded that there was no risk segmentation by banks and finance companies. We do not believe that these studies support the proposition that there is not a significant class of consumers whose only alternative in the credit market is a consumer finance company. Certainly, the Boczar study demonstrates that a large number of consumer finance customers, specifically two-thirds, have risk characteristics which appear to make them theoretically eligible for bank loans. However, such a conclusion necessarily entails the related conclusion that one-third of those customers have characteristics which suggest that they would probably not be bank customers. Thus, this study does not negate the existence of a group of consumers whose credit needs are exclusively met by finance companies. Indeed, we do not understand Mr. Boczar’s conclusion that no market segmentation exists to negate the existence of this group given his definition of market segmentation. In his study, Mr. Boczar states that market segmentation hypothesis as follows: “[b]anks supposedly serve low-risk borrowers while finance companies cover the high-risk borrowers.” Thus after having found that finance companies contain a substantial number of low-risk borrowers in their customer base, Boczar appropriately rejected the market segmentation hypothesis. Boczar admitted that his rejection of the market segmentation hypothesis was largely premised on the mutual accommodation of low-risk customers by both banks and finance companies: “the results showed that there was substantial overlap in borrower characteristics between the two industries, largely due to the fact that finance companies served borrowers whose risk-related characteristics . . were similar to . [those] of bank borrowers.” (HX 2300). Thus, Boczar’s rejection of market segmentation does not entail a conclusion that certain higher-risk customers remain exclusively tied to finance companies. The Murphy study also does not support the position that there is no group of borrowers uniquely served by finance companies. The sample involved in this study was a group of borrowers in Maine that had previously been accepted for loans by finance companies but because of changes in Main law were not given loans by these companies. Murphy’s study then sought to *1262 determine whether postulated risk characteristics would predict whether these borrowers would ultimately be accommodated by another lending institution. The analysis proved that there was a statistically significant difference in the risk profiles between those customers able to receive loans and those that were not. This, of course, means that for statistical purposes the groups were not completely identical. Murphy, however, concluded that there was an extreme overlap between the group because he was only able to predict unsuccessful borrowers 64 percent of the time and successful borrowers 52 percent of the time. Nonetheless, these figures demonstrate that for unsuccessful borrowers almost two out of three had risk characteristics that predicted they would be unable to get a loan. Admittedly, these risk characteristics are less helpful in determining which borrowers will be successful, but this only means that financial institutions will not necessarily accommodate customers simply because they are low-risk. It is, for the purpose of our market analysis, only important that the study shows that previous finance company customers possessed risk characteristics making it unlikely that they would receive a loan from an institution. 8 The second possible basis for the district court’s conclusion was a study comparing credit-scoring systems. The defendants introduced a study in which recent applicants for loans from Household Finance were evaluated under the credit scoring system of the Security Pacific Bank and the VISA division of the First National Bank of Chicago. That study showed that 56 percent of the accepted HFC applicants would have satisfied the credit scoring requirements of Security Pacific and 35 percent would have satisfied the requirements of VISA. Of the declined applicants, 49 percent would have been acceptable to Security Pacific, whereas 28 percent would have been acceptable to VISA. (Tr. 1789-91). This study, however, runs directly contrary to the defendant’s position: 65 percent of those able to obtain loans at Household Finance apparently would be unable to receive a VISA card from the First National Bank of Chicago, and slightly under one-half would have been able to obtain loans from Security Pacific. Even though this study shows some overlap between credit standards, it nevertheless also shows a significant number of customers who would be served only by finance companies. The third basis for the district court’s conclusion was opinion testimony that banks and finance companies apply the same credit standards. Two bank officials from Pennsylvania banks testified that they believed that their banks applied the same credit standards as finance companies, although neither, of course, could testify as to how many other banks utilized similarly liberal standards. Indeed there was testimony from an executive vice-president of the First National Bank of Chicago that his bank’s cash loan program was principally directed at executives and professionals (Tr. 1402) and that his bank had recently tightened its credit card operations to include only those eligible for $1,000 credit limits, thereby excluding people previously eligible for credit. (Tr. 1374, 1390). The defendants also introduced testimony from the chief executive officer of Beneficial Finance Company that his company was applying similar standards as other financial institutions, but even he was forced on cross-examination to admit that some cus *1263 tomers are relegated to finance companies as their only source of loans. (Tr. 1305). Finally, the court also attempted to support its conclusion that there was no risk segmentation among financial institutions by finding that a significant percentage of finance company customers are eligible for or actually have non-mortgage loans at banks and that a majority of the customers of consumer finance companies have bank credit cards. Even accepting these findings on their face, they have little bearing on whether risk segmentation exists. Even if a “significant” percentage are eligible for bank loans, the remainder — conceivably also a “significant” percentage — are not. Similarly by the terms of these findings up to 50 percent of consumer finance company customers do not possess any bank credit cards. Notwithstanding that these findings are not relevant to whether these are indeed high-risk customers served only by finance companies, we cannot find support for these statements in the record. The defendants introduced a study of recent HFC applicants in a Chicago loop office which showed that of 50 accepted applications, only 36 percent had outstanding bank loans of any kind, only 14 percent had loans from credit unions and that none had any loans from savings and loan associations. A similar study by the defendants on a nationwide sample of 426 customers showed that more than 75 percent of accepted HFC customers did not have bank credit other than auto loans and that approximately 70 percent did not have even auto loans. Likewise, only about 25 percent of the accepted customers had loans from credit unions and less than 3 percent had loans from savings and loan associations. (HX 209; Defendant’s App. at 186). The chief executive officer of Beneficial Finance, another defense witness, testified that only 27.2 percent of Beneficial customers had loans from a bank and 16.4 percent had loans from credit unions. (Tr. 1317). The final evidence on this matter was the testimony by an officer of the target company in this case, American Investment Company, that 50 percent of American’s customers had bank credit and that at least 15 percent of those without such credit would not even obtain that credit under any circumstances. (Tr. 1620-21). These figures do not support the district court’s finding that finance companies do not lend to a significant percentage ineligible to borrow from other types of institutions, since even accepting the American executive’s admitted speculation that at least 15 percent of those without bank loans could not obtain them, this still concedes the existence of a cognizable number of borrowers relegated to consumer finance companies. The district court’s finding that a majority of consumer finance customers had bank cards is even more difficult to support. There was evidence that 80 percent of the customers at one office had bank cards; 9 however nationwide HFC figures showed that only 32 percent of Household’s customers had bank cards. (HX 209). Likewise, only 32 percent of Beneficial Finance customers have bank cards. (Tr. 1318). Not only was the evidence of non-segmentation inadequate, but also the record contained overwhelming evidence that risk segmentation does in fact exist among financial institutions. 10 This evidence took *1264 three forms: studies of eligibility of finance customers for bank loans, varying charge-off and delinquency rates between banks and finance companies, and studies of the demographics of borrowers at both institutions. We turn first to the four eligibility studies that were introduced. Dr. William Whitesell, an economics professor and former Federal Reserve Board economist, testified that he had done a study for a commercial bank seeking to acquire a finance company and found that 60 percent of the finance company customers would not have qualified for a loan at the bank considering the acquisition. (Tr. 115). An executive vice-president of CitiCorp testified that his examination of the files of 175 customers of Nationwide Finance Company, a subsidiary of CitiCorp, revealed that only 46 percent of them would be eligible for a loan from CitiBank, another subsidiary of CitiCorp. A study of Texas finance company borrowers, made by two economics professors revealed that, among the borrowers of small amounts, 25 percent had been refused loans by banks and 20 percent had been asked for too much collateral. (Tr. 486). Among the remaining borrowers, 30 percent replied that they had been refused by banks or that the banks required too much collateral. 11 The government also introduced a follow-up study on borrowers in Maine that had been accepted by finance companies for loans but later refused these loans because of a subsequent change in Maine law. Of those borrowers, 50 percent never received a loan from any source and only one-third of those who received loans (or about 16 percent of the entire sample) received a loan from a bank. Finally, it should be recalled that even the two defense studies, which we discussed earlier, lend some support to the conclusion that at least one-third of finance company customers have risk characteristics which suggest that they would be ineligible for bank loans. Thus, all these studies give a concrete demonstration of the existence of a significant class of borrowers, ranging from 20 to 50 percent of all finance company customers, who are able to obtain loans only from finance companies. Evidence of delinquency and chargeroff rates at banks and finance companies further supports a finding that finance companies offer services to a unique strata of higher-risk customers that are not serviced by commercial banks. The testimony at trial was unequivocal and uncontradicted that finance companies charged off somewhere around 2 percent of their loans in terms of total dollars outstanding whereas banks charged off approximately 0.7 to 0.8 percent of their personal loans. (Tr. 1055). Several witnesses testified that this differential indicates that consumer finance companies serve some customers that are a higher risk than those served by banks. 12 *1265 Finally, support for the existence of risk segmentation can be found in demographic studies conducted by the defendants. The defendants introduced evidence that showed that whereas the average income of those borrowers who exclusively used banks were $14,260, the average income for exclusive customers of finance companies was $9,050. Further, only 19.3 percent of consumer finance company customers were white collar, whereas white collar workers comprised 37.3 percent of bank borrowers. Conversely, only 17.5 percent of bank borrowers were blue collar, as opposed to 28.7 percent of consumer finance borrowers. (Tr. 832-33). In summary, we conclude that the record below supports the proposition that finance companies offer unique products and services to a class of higher-risk customers than are serviced by other financial institutions. All the evidence suggested that these customers constituted anywhere from 15 to 50 percent of the finance company clientele. The district court never disputed these figures or suggested that they were lower. Indeed, even if these customers constituted a far lesser percent of the finance company clientele, we would still find that group significant. Thus, the finding of the district court that finance companies do not have a significant unique clientele was erroneous, and accordingly the judgment below must be reversed. Reversed. Footnotes [1] The court did erroneously find, see pp. 1260, 1262-1263, that a majority of the customers of consumer finance companies possessed bank credit cards. However, it did not find what portion of those cards had sufficient remaining balances. [2] The availability of loans to high risk consumers has been explicitly recognized by the Supreme Court as a unique service: For some customers, full-service banking makes possible access to certain products or services that would otherwise be unavailable to them; the customer without significant collateral, for example, who has patronized a particular bank ... is more likely to be able to obtain a loan from that bank . Phillipsburg Nat’l Bank, 399 U.S. at 360, 90 S.Ct. at 2042. [3] Thus the Federal Reserve Board, the Comptroller of the Currency and the Justice Department have considered competition in these overlapping markets when analyzing the effects of mergers between different types of financial institutions, e. g., between banks and financial institutions. (HX 181, 255, 264, 299). In such mergers it is obvious that the important inquiry revolves around concentration in jointly provided services, since any lessening of competition will occur, if at all, only in those areas. It is this confusion about which sub-markets to consider which underlies most of the district court’s position that finance companies are not a distinct line of commerce because at least with respect to many customers finance companies and other financial institutions in fact “compete.” [4] The Court further elaborated this position in a footnote: As one banker testified quite frankly in the instant case in response to the question: “Do you feel that you are in substantial competition with these institutions [personal-finance and sales-finance companies] that you lend * * * such money to for [sic] loans that you want to make?” — “Oh, no, we definitely do not. If we did, we would stop making the loans to them.” (R. 298.) The reason for the competitive disadvantage of most lending institutions vis-á-vis banks is that only banks obtain the bulk of their working capital without having to pay interest or comparable charges thereon, by virtue of their unique power to accept demand deposits. 374 U.S. at 356 n.33, 83 S.Ct. 1738. [5] The National Commission on Consumer Finance found that the average interest rate charged by consumer finance companies was 26 percent, whereas 13 percent was the average rate for banks. (Tr. 625). No testimony was offered at trial to rebut these figures. [6] As the Court stated in Connecticut National Bank: To be sure, there is a large measure of similarity between the services marketed by the two categories of banks. In our view, however, the overlap is not sufficient at this stage in the development of savings banks . to treat them together with commercial banks . . . . Despite the strides that savings banks . . have made toward parity with commercial banks, the latter continue to be able to provide a cluster of services that the former cannot 418 U.S. at 663-64, 94 S.Ct. at 2793-94. [7] The defendants in their brief to this court characterize this conclusion as a “finding of fact” and argue that the district court may be reversed only if we find this conclusion to be clearly erroneous. We think, however, that whether the group of customers uniquely served by finance companies is significant enough to constitute a line of commerce within § 7 is a legal question. Certainly, the Supreme Court treated this as a legal question in Connecticut National Bank when it concluded simply that “the District Court overestimated the degree of competitive overlap . . 418 U.S. at 656, 663, 94 S.Ct. at 2793. [8] Both of these studies also appear seriously flawed in another respect. Obviously it is important that all major risk characteristics be utilized in the model, otherwise the predictive value of the model will be diminished and the degree of overlap between finance company and bank customers will be exaggerated. Boczar admits this in his study (HX 200 at 246), and Murphy conceded as much in his testimony at trial. (Tr. at 1547). Both studies left out a number of important risk variables. Neither included the borrower’s total outstanding indebtedness, a factor which Boczar indicated would have been helpful. (HX 200 at 249 n.8). Additionally, the Murphy study excluded the following risk variables: time on job, time in community home ownership, rent and mortgage payments. The evidence in this case clearly demonstrated that all these factors were frequently used in credit scoring models used by financial institutions themselves to measure risk. [9] This was the figure given at trial. Our examination however, of the exhibit on which this testimony was based reveals that this figure was erroneous. HX 210 showed that only 27 of 50 accepted Household customers, or 54 percent, had bank cards. Further, the office surveyed was located in the Chicago loop with a customer base of office and retail employees, a group that is not likely to be representative of HFC’s entire customer base. (Tr. 1419). [10] The rest of our analysis concerns only whether finance companies serve a consumer base which is not served by commercial banks. We have limited this analysis because the record clearly demonstrates that banks are the only substantial alternate source of credit for finance company borrowers. Bank credit cards, as we have found, are not a significant alternative since almost two-thirds of all finance company borrowers do not possess such cards. Likewise, credit unions are not a significant alternative since the district court found, and we accept its finding, that one-half of all consumers do not have access to such an institution. Finally, savings and loan associations are generally restricted by state law principally *1264 to loans relating to, and secured by, real estate. The record below illustrated that only a minute percentage of finance company loans fell within this category. [11] There was testimony at trial that customers who replied that the bank had required too much collateral had effectively been refused credit since that answer could have indicated that the bank had requested more collateral than the applicant could post. (Tr. 543). In any event, of this 30 percent, half were denied credit regardless of collateral. (Id.). [12] Two defense witnesses gave different constructions to the differentials. An executive from Mellon Bank attributed the difference, at least as far as his own bank and its finance company subsidiary were concerned, to lack of experience among finance company officers in granting loans. This does not deny, however, that finance company borrowers, for whatever reason, may turn out to be a greater risk. Gregory Boczar, a defense expert, characterized the difference as follows: there was a difference, no question about it, and . this is very judgmental, but it was my view that the numbers were not so different as to suggest that commercial banks were only serving the very well to do and the finance companies were just taking the poor fellow who can’t get a loan anywhere else. The issue, however, is not whether there is a total division between the types of customers of banks and finance companies but whether, notwithstanding any overlap in customer bases, finance companies service some customers unable to obtain bank credit. Mr. Boczar never denied that these figures certainly suggest the latter, less extreme position. Indeed, in one of the articles on which his testimony was based he concludes that “finance companies appear willing to service poorer risk borrowers than *1265 banks.” (HX 212). Finally, defense testimony suggested that the charge-off rates exhibit a converging trend. But even that testimony revealed that finance company charge-off rates had declined only to slightly below 2 percent.