Printer Friendly

Gender, structural factors, and credit terms between Canadian small businesses and financial institutions.

Do credit terms differ between female and male small business owners? This question, fraught with emotional overtones and inconclusive findings, has received considerable attention in the public press and in academic research. If, indeed, differences do exist, to what extent are structural differences in borrowers' eligibility accountable? Conversely, what role (if any) does systematic gender discrimination play in determining any discrepancies? The aim of this study is to report on research that sheds further light on these issues. To this end, this study has three objectives.

The first is to identify factors that are most closely associated with credit terms. Potential structural differences include the form, size, industry, and track record of an enterprise. These are some measures of the risk associated with lending to a business. The second objective is to determine if factors such as these risk measures are also correlated with gender of the owner. Gender-related disparities in business attributes, quite independently of systematic gender discrimination, may lead to discrepancies in credit terms, discrepancies that may appear as if gender-related. The third objective is to investigate whether or not credit terms differ between female-owned and male-owned enterprises after accounting for structural differences in business attributes. This study investigates five credit terms: rates of loan turndowns, rates of requests for spousal co-signature, ratios of collateral to line of credit, ratios of amount received to amount applied for, and interest rates.


Three streams of research literature relate to the topic of this study: the general literature on lending criteria; the literature on women and small business; and the integrated literature.

The general literature on bank lending reveals financial institutions to be low-risk lenders (Thornton, 1981; Poapst, 1981; Grant, 1986), a finding of little surprise. Several researchers have investigated the relationship between measures of business risk and credit terms (for example, Grant, 1986; Wynant & Hatch, 1991). One of the few consistent findings was that firm size plays a central role in determination of access to and terms of credit. For example, Haines, Riding, and Thomas (1989) found that the size of the firm was the one variable among many potential determinants that correlated with interest rates on bank loans.

Other research on lending criteria and the bank - small business interface includes that of Thornton (1981), who found that the quality of financial management was a factor in lending decisions. Fertuck (1981) noted that small businesses may receive more favorable credit terms from small community-based financial institutions such as credit unions and caisses populaires.(1) In general, however, this literature has not explicitly allowed for the possibility that gender might be one of several factors which, acting jointly, determine terms of credit. The singular increase in women-owned businesses in recent years has also given rise to a burgeoning literature on business problems that are specific to women. This second stream of literature has identified legal status, age, industrial sector, level of sales, and rate of sales growth as some factors that differ systematically between men and women business owners (Schwartz, 1979; Hisrich & Brush, 1986a, 1986b; Litton, 1987; Collom, 1982).

This literature also implied that gender-based differences on these factors may account for some portion of the financial difficulties that women small business owners often perceive (for example, Kryzanowski & Bertin-Boussu, 1984; Stevenson, 1986). This stream of literature, however, has not determined whether or not the credit experiences of female and male business owners actually differ. Many of the studies in this stream of the literature, while citing gender-related problems with respect to credit, did not compare the experiences of women business owners with benchmark samples of men business owners.

A third segment of the literature not only identified gender as a potential determinant of credit terms but also attempted to control for factors that may confound the analysis of differences in the credit experience of female and male owners. Appendix 1 provides a tabular sampling of the findings of this stream of literature.(2) The current state of research about this issue is best illustrated by a consideration of recent studies by Buttner and Rosen (1988, 1989, 1992) and Riding and Swift (1990).

Buttner and Rosen (1988) found that loan account managers associated the traits usually identified with successful small business owners more closely with male business owners than with women owners. Subsequently (1989), they attempted to discover whether or not sex stereotypes actually influence loan rejections. Their results indicated that neither gender of applicant, type of loan account manager, or presentation format of application played a role. These findings, that bank loans account managers did not treat men and women small business borrowers differently, was also consistent with the work of Riding and Swift (1990), who came to similar conclusions based on a sample of Canadian businesses.

Riding and Swift noted that, on average, women-owned businesses were younger and smaller than those owned by men. To the extent that age and size of firms were measures of financial risk, it was unclear to what extent gender, or age and size, might account for differences in borrowing experiences. In order to determine the extent to which gender played a role in the determination of terms of lending, researchers face the primary challenge of disentangling the effects of gender from the effects of variables associated with gender. Studies that do not take into account the potentially confounding effects of financial variables that are themselves correlated with gender are then susceptible to a "missing variables" error. This type of error leads to biased estimates and inappropriate conclusions.

Accordingly, Riding and Swift (1990) used a matching technique to obviate this problem. Each woman respondent to the survey was matched with one or more male respondent who resembled her in terms of business size, geographical location, age, sector, and form of business. Riding and Swift found that after controlling for these variables no statistically significant difference remained in five out of six terms of credit, between women- and men-owned firms.(3) The work of Riding and Swift would have been strengthened had multivariate analysis been used. However, the sample available to them included too few women business owners to permit the use of multivariate techniques.

In their 1992 study, Buttner and Rosen attempted to explain the paradox of this body of research. On the one hand, they cited the consistent allegations by female owners that gender bias exists. Conversely, Buttner and Rosen noted the statistical studies that document that the terms of credit to male and female business owners do not differ. They offered three hypotheses, each of which could potentially explain the incongruity:

* that female business owners were more likely to underestimate the difficulty inherent in securing capital;

* that female owners were more likely to attribute rejection to gender bias than to the shortcomings of the business; or,

* that male owners were more active than females in the search for capital.

Based on their survey data, none of these hypotheses were supported, so the dilemma remains.

This study differs from previous research in three ways.

1. This study is based on the largest sample of lending experiences of women owners reported to date. This large sample admits the use of methodologies that were not available for previous studies.

2. This study presents a theoretical model of potential determinants of credit terms, a model that distinguishes between subjective and objective measures of business risk and one that includes more measures of business attributes than has any previous study. The model developed here does not assume that the credit decision-making process is standardized. It recognizes that both structural factors and systematic gender discrimination may determine credit terms; that is, the model is multivariate, allowing for the association of gender, credit terms, and business risk to be considered simultaneously.

3. The study employs powerful and easily interpretable techniques that are robust against statistical assumptions.


As Buttner and Rosen (1992) noted, the literature includes a pervasive belief that gender discrimination leads to differences between male and female business owners on terms of bank financing. Terms of credit on which differences may exist include loan turndown rates, requests for co-signature, ratio of amount received to amount requested, and interest rates. Hence, these measures were selected as dependent variables for this study.

Traditional financial wisdom posits five sets of structural factors that lenders use to measure the risk associated with loan applications. These have been termed the "five C's" of commercial lending (Ross, Westerfield, Jordan, & Roberts, 1993, pp. 690-692):

1. Capacity is the extent to which an organization is able to meet its obligations as they fall due.

2. Capital relates to the amount of equity investment made by the owner(s).

3. Collateral relates to the value of assets available to secure the loanable funds against liquidation or default. Firms in sectors such as manufacturing are believed to possess more pledgable assets than firms in sectors such as retailing and services (Thornton, 1981).

4. Character includes the track record of the business and its owners. Age of business, years of managerial experience of owners, and level of financial management and expertise are among the variables which have been used to measure character.

5. Conditions refer to the proprietary nature of the product or service, the size of the market, and the industrial climate.

Even a priori, it is clear that there are associations among the five Cs of commercial lending. For instance, the number of owners that a firm has (capital) may also have an impact on its asset structure (collateral). To the extent that the owners have a diversified set of skills, the number of owners can also influence the extent to which the firm is professionally or adequately managed (character). Firms that have employed professional financial managers may demonstrate fiscal responsibility in terms of character, capital, capacity, and collateral. Character could also have an impact on any of the five Cs. The education and experience that the owner has may have an impact on the equity (capital) invested in the business. In short, some of the elements of the five Cs do not fit nicely together under one particular category. This underscores the need for techniques that can distinguish among the various sets of variables and take their correlations into account.

Figure 1 is a diagrammatic representation of the model on which this study hypothesizes. It identifies the five Cs of commercial lending as overlapping elements in the credit-granting decision. However, the model also includes demand-side features that may moderate measures of business risk. Systemic economic, political, and social forces may also be responsible for differences in business attributes that might otherwise seem to be gender-related.

In addition, potential gender discrimination is represented in the model by a decisional prism. If gender bias is absent, only structural differences in business attributes would account for differences in credit terms. However, just as white light is fragmented into different colors when it passes through a prism, so may credit terms differ between female and male small business owners if systematic gender discrimination exists. The overlapping measures of business risk and credit terms, respectively, are meant to represent the interactive nature of the model.

Accordingly, research has five challenges to meet if the issue of gender discrimination is to be resolved.

1. It must determine whether or not differences actually exist in terms of credit received between female and male small business owners.

2. It must investigate whether or not such differences can be accounted for by differences in the characteristics of the two groups of borrowers.

3. Research must examine the extent to which lending criteria are standardized or consistently applied in evaluating credit applications for female and male small business owners.

4. Contingent on these findings, research should explore the plausibility that female and male small business owners may view the role of the banker differently.

5. Since multiple criteria measures are needed for a comprehensive investigation, to minimize the likelihood of spurious empirical results, tests should either be conducted at lower levels of significance or multivariate approaches should be considered for analyses.

To meet this research agenda, the credit experience of both groups of small business owners, female and male, should be evaluated under identical research conditions. That is, the two groups should be investigated within the same study to facilitate comparisons. Such an agenda provides rationale for the stated objectives of this study.


The results of this study were based on data collected from a national survey conducted by the Canadian Federation of Independent Business of its members in late 1990. This population may not be representative of the population of all Canadian small business owners. Firms that join the CFIB have reached a point in their development such that they are able and interested in joining this organization. Accordingly, the findings of this study might not be generalizable to particularly small or newly formed firms; however, the findings are probably representative of the population of established small businesses.

The first section of the questionnaire sought background information on the business, on up to three of the principal owners, and on the financial manager of the enterprise. The second section solicited data on the terms of credit the respondent had received and also asked the respondent about the type and amount of assets that could be used as security. The third section gathered data about bank shopping behavior, customer satisfaction, and the perceptions of the respondent about the role of the banker. The fourth section was an attempt to gather objective financial information. However, few respondents were forthcoming with these data; the data that were received in this section were not sufficiently standardized to permit inclusion.

A total of 14,980 questionnaires were mailed on June 1, 1990, with a cut-off date for returns of September 15, 1990. The sampling procedure was a stratified one: questionnaires were sent to all 5,246 women-owned businesses on the CFIB membership list; 9,734 men-owned businesses were selected randomly. In total, 2,763 (759 female and 1,974 male) responses were received, a response rate of 18.44 percent. Firms in the agricultural sector were excluded from this study, yielding an overall working sample of 758 and 1,907 responses from female and male respondents, respectively. As is normally the case with survey research, not all respondents answered all the questions.(4)

To minimize non-respondent bias two measures were taken (Davis & Cosenza, 1988). First, prior to mailing, the questionnaire was pre-tested. Second, salient characteristics of respondents and those of the general small business population were compared. This comparison confirmed the expectation that the CFIB sampling frame tended to exclude very small and start-up firms. In addition, the CFIB working sample, relative to the general population, was over-represented in manufacturing and retailing. In contrast, the Statistics Canada database features a greater proportion of firms in the construction and transportation sectors. This disparity is arguably a consequence of the stratification which ensured an adequate number of women business owners.

Four credit terms were of interest. The loan turndown decision was a dichotomous variable: ultimately, a loan either gets rejected or accepted. A request for spousal co-signature was also a binary variable: it was either required or not. It was anticipated that interest rates and the ratio of amount received to amount applied for would be continuous variables and that they would be normally distributed. While these variables were arguably continuous, the ratio of amount of loan requested to the amount applied for was not normally distributed. This skewness, which proved immune to any transformation, indicated that this variable ought best be treated as a categorical variable. Consequently, the ratio of amount received to amount requested was treated as a binary variable and was assigned a value of one if a firm got less than 90 percent of the loan for which it had applied and a value of zero otherwise.(5)

Thus, this study involves a determination of whether or not gender is among the variables that are determinants of three binary variables (turndowns, co-signatures, amounts received/requested ratio) and of the continuous normally distributed interest rate variable. Potential explanatory variables include: measures of capital (form of business, number of owners); capacity (annual level of sales, number of full-time equivalent workers); conditions (rate of annual growth in sales, industrial sector, geographical location, high-technology component of product and process); character (financial and managerial background, training and skills of the owners, age of business, credit track record of the respondent, gender of the owners and financial managers); and collateral (respondents were asked to indicate whether or not each of the 12 categories of assets were available as collateral).

In order to ensure robustness of the findings, all analyses were carried out for each of three alternative definitions of the gender of the owner. The first sub-sample included all firms, and gender was based on the reported gender of the primary owner. The second sub-sample consisted of firms in which a majority of the owners were the same gender. The third sub-sample of firms involved only those firms whose owners were all of the same gender. This sample is the most stringent definition of gender of ownership; it obviates any confounding effect of joint female and male ownership. Predictably, this group is smaller than either of the other two groups of firms.


This study used three different techniques to investigate the association of gender, business risk, and credit terms: logistic regression; multiple linear regression; and logistic regression analysis of covariance with propensity scores.

Logistic regression was used for two purposes: to identify structural gender-based differences in business attributes, and to determine factors influencing loan rejections, ratio of amount received to amount requested, and requests for spousal co-signature.

Stepwise logistic regression is a multivariate technique that may be used to determine the association between a binary dependent variable and several independent variables that are measured on different scales. It is robust against the normality assumption and permits the use of independent variables on different scales. Also, it takes the correlation among the independent variables into account. This is because it employs the principle of linear combinations where association among several variables are considered. The logistic regression equation (for loan turndowns, as an example of the general principal), is given by:

Probability of Loan Turndown = [e.sup.f{x}]/1 + [e.sup.f{x}]

where the f{X} represents a linear combination of variables which, in concert, can be argued to have an effect of loan turndowns. For example, the f{X} could take a form such


f{X} = [a.sub.0] + [a.sub.1](SIZE OF FIRM) + [a.sub.2](EARNINGS GROWTH RATE) + [a.sub.3](GENDER) + . . .

In the event that the coefficient of gender, here [a.sub.3], is statistically significant, the conclusion that gender affects the probability of a loan rejection would then be supportable.

Multiple linear regression was employed to analyze the association between interest rates and measures of business risk. Because of the stratified nature of the sample, weighted least-squares regression was employed instead of the more commonly used ordinary least squares. Again, the use of stepwise regression enabled the inclusion of those variables most closely associated (in the statistical sense) with the dependent variable. Collinearity was thereby minimized.


Differences in Lending: A First Glance

Access to credit was measured by two variables. The turn-down rate (defined as the ratio of rejected loan applications to the total number of loan applications) and the approval rate (defined as the ratio of the dollar value of loans approved to the total dollar value of loans sought).

The terms of credit were also measured by two variables: the interest rate on loans (expressed as the number of percentage points above the prime rate) and the requirements, if any, for spousal and/or other co-signature. Additional data of interest were: the expressions of satisfaction with the terms of lending (measured on a 9-point scale, the percentage of owners who responded that they were dissatisfied is reported), and the frequencies of defaulting on loans and of exceeding the limits of lines of credit.

Tables 1, 2, and 3 report these data for term loan applications, new line of credit applications, and applications for increases in existing lines of credit, respectively. These tables break down the terms of and access to credit by size of firm and by gender of the principal owner. Only those firms that had applied for credit within the three years previous to the survey are reported.

These tables demonstrate that smaller businesses appear to have less access to bank [TABULAR DATA FOR TABLE 1 OMITTED] credit than do larger firms. Moreover, credit that is received by the smallest firms seems to be on terms that are more severe than the terms of credit accorded relatively larger small businesses. With few exceptions, turndown rates, approval rates, interest rates, and requirements for cosignatories are consistently least favorable for the smallest firms. Larger SMBs appear to have easier and more reasonable access to bank credit. This is particularly evident in applications for increases in lines of credit.

The data also suggests that banks are more selective in advancing credit to relatively smaller businesses. Turndown rates on loan applications are higher yet smaller businesses default less frequently on loans than larger businesses. Moreover, smaller firms tend to stay within their credit limits to a greater degree than larger firms.

The level of dissatisfaction with banks felt by all small business owners is also noteworthy. The data show that women business owners are, in general, less satisfied than men owners. This is an issue to which this study will return. Again, the smaller the firm, the more often the owners expressed dissatisfaction about the terms of lending.

These findings, however, are not unequivocal about whether or not gender of the owner is a partial determinant of the terms of credit, even within the micro-business sector. This is because the gender of the principal owner is highly correlated with the size of the firm: women business owners tend to be associated with smaller firms. Even within the micro-business sector, this confounding of size with gender continues: 58 percent of the firms with annual sales of less than $100,000 were owned by women compared with 38 percent of firms with annual sales of $100,000 to $200,000. Therefore, methodologies that are able to disentangle the potential confounding of firm size [TABULAR DATA FOR TABLE 2 OMITTED] [TABULAR DATA FOR TABLE 3 OMITTED] and gender are required. Likewise, business attributes that are also correlated with gender need to be identified.

Gender-related Differences in Business Attributes

Appendix 2 lists the variables employed as independent variables in the modelling of the theoretical paradigm presented in Figure 1. This table describes each variable and presents the salient descriptive statistics for each variable, broken down by gender of principal owner. In order to identify which of these, if any, are systematically associated with the gender of the owner, logistic regression was employed. The variables that logistic regression found to be statistically significant are listed in Table 4. According to these findings, women business owners when compared with men owners are more likely:

* to have applied for term loans;

* to be found in retailing;

* to be their own financial managers;

* to have completed more years of education;

* to be sole owners; and,

* to have personal real estate, bonds, and securities available as collateral.

In contrast, relative to businesses owned by men, women-owned businesses tend:

* to have lower sales volume;

* to have fewer years of managerial experience;

* to be less likely to report having either business real estate or machinery available as collateral;

* to be less likely to report that their products or services are of the high-technology genre;

* to be less likely to report having a personal automobile available as collateral;


* to be less likely to have an employee, rather than an outside accounting firm, perform the financial function.

The entry of all the dummy variables that express rates of sales growth is an indication that the sales growth rates of women-owned businesses are less stable than those of men-owned firms.

Of all these variables, years of managerial experience was the variable that was most closely associated with gender of owner(s). This, and other systemic differences between men- and women-owned firms (being younger, smaller, concentrated in retailing, etc.), indicates that, from the perspective of a banker, the average women-owned firm is riskier than the average firm owned by men. It remains to be seen to what extent such systemic differences might account for any differences between terms of credit to the two groups of firms and to what extent gender-based discrimination might play a role. In other words, it is necessary to ascertain the role of each of the potential determinants of terms of credit and whether or not, after allowing for these systemic factors, terms of credit still differ between men and women business owners.

Factors Influencing Terms of Credit

In this section, the nature of the relationship between each of the four credit terms and structural factors is explored. Logistic regression analysis was used to examine the association between loan rejections, ratio of amount received to amount requested, and requests for spousal co-signature. Multiple linear regression using weighted least squares was used to estimate the relationship between structural factors and interest rates.

Panel A of Table 5 summarizes the findings of the logistic regression estimation of the determinants on loan turndowns.(6) The variables that proved to be most closely associated with loan turndowns were: depth of financial management, age of business, and rate of growth in sales. The finding that firms in the manufacturing sector had better access to capital is at variance with past research (Haines, Riding, & Thomas, 1989). No main effect of gender was found. Gender of majority owner had low t-statistics (0.88, 0.31, and 0.26) across the first, second, and third sub-samples, respectively. As indicated in the table, the likelihood of loan rejections was highest for firms that lacked professional financial management, that were relatively young, and whose annual rate of growth had been declining. Geographical location was found to be a significant determinant in one sub-sample, but this finding was not consistent across definitions.

Variables that seemed to be most closely associated with ratio of amount of credit received to amount of credit requested were number of owners and type of assets available as collateral. As Panel B of Table 5 indicates, firms that were most likely to receive less than 90 percent of the amount of credit they requested were those with more [TABULAR DATA FOR TABLE 4 OMITTED] than two owners and those that cited having other business assets available as collateral. It was a rather surprising finding that firms with fewer than three owners were more likely than their counterparts to be given 90 percent or more of the amount of credit they requested. This seemed to contradict the suggestion of past research (Thornton, 1981; Wynant & Hatch, 1991) that banks would be more willing to provide credit to firms with more than two owners or to those that do not exhibit the characteristics of "Ma and Pa" stores.

From Panel C of Table 5, factors most closely associated with requests for spousal co-signature included: the employment status of the financial manager; the type of financial institution approached for credit; whether or not personal real estate or other business assets were available as collateral; the age of the firm; and the level of education of the principal owner.

Requests for spousal co-signature were more likely for firms that reported having part-time financial managers; approached either Canadian chartered banks or Trust Companies for credit; reported having personal real estate, other personal assets, and other business assets available as collateral; were unequally owned; and reported having an employee or a principal owner as financial manager, at least part-time. However, the older a firm, and the more education its owner has, the less likely that it would be requested to provide spousal co-signature.

Across the three different samples of firms examined, variables most closely associated with interest rates were number of full-time equivalent workers, rate of growth, and level of sales. None of the other variables was related with interest rates.

As Table 6 indicates, there was an inverse relationship between rates of interest charged on loans and capacity. The more full-time equivalent workers a firm had, the lower the rate of interest charged on its loan. Likewise, the higher the volume of sales of a firm, the lower the rate of interest charged. However, firms with declining rates of growth were charged higher rates of interest than their counterparts. These results concur with past research findings, which indicated that capacity, as represented by size, has a strong influence on rate of interest charged on loans.

From Tables 5 and 6, it was seen that in no instance was the term of credit or the likelihood of a loan turndown correlated with the gender of the owner of the firm. The results of logistic and linear regression models revealed no appreciable difference in credit terms after structural factors were controlled.(7) Credit terms did not seem to differ between female and male business owners. This result, however, seemed to be at variance with the considerable amount of prior research and with anecdotal accounts which suggest discrimination. The section which follows attempts to reconcile this divergence.

Interpersonal Dimensions of the Banking Relationship

In order to test the importance of interpersonal factors and to assess the satisfaction felt by small business owners, factor analysis was employed to help identify which [TABULAR DATA FOR TABLE 5 OMITTED] rattributes of the banking relationship were important to small business owners. The survey asked respondents to rate, on a three-point scale ("not important," "important," "very important"), each of nine specific aspects of their banking experiences. These nine items, with a breakdown of the responses, are listed in Table 7.
Table 6

Determinants of Interest Rates: Regression Results

Significant        Coefficient        Standard          Level of
Variables           Estimate           Error          Significance

Constant              2.630            0.320         [less than] 1%
WORKR                -0.377            0.072         [less than] 1%
DECLIN               -0.224            0.089         [less than] 1%
SALES                -0.074            0.021         [less than] 1%

In order to identify the causes underlying these feelings of what is important, a factor analysis was carried out, revealing two significant factors. The rotated factor matrix and the eigenvalues are also listed in Table 7. The first attribute of the banking relationship with small business owners was associated with responses to the first five items in Table 7; this factor could be seen as being related to the "business-related" aspects of the banking relationship. The second factor was associated with items 6, 7, and 8 of the survey. These relate to the interpersonal relationship with the loans account manager. From Table 7, it is seen that this dimension of the banking relationship was highly regarded by small business owners.

In addition to being asked to rate the importance of the nine items listed above, respondents were also asked to rate the performance of their respective bankers on each of the nine items. Again, a three-point scale was used ("poor performance," "acceptable performance," "good performance").

In order to quantify the satisfaction that owners felt about the two dimensions of what owners deemed to be important, indices were created corresponding to each factor. Each index value was computed for each respondent by multiplying the level of importance of each item (0 = not important; 1 = important; 2 = very important) by the degree of perceived performance expressed on the item by the respondent (-1 = poor performance; 0 = acceptable; 1 = good performance). The products were then summed across the number of items. Mathematically, then, the j'th respondent's measure of satisfaction with respect to the "business-related" factor [[BUSSAT.sub.j]] is given by:

[BUSSAT.sub.j] = [Sigma][[I.sub.i].sup.*][P.sub.i] i = 1, 2, 3, 4, 5

where [I.sub.i] and [P.sub.i] are, respectively, the j'th respondent's assessment of importance of and their banker's performance on the i'th attribute.

A similar expression involving items 6, 7, and 8 defined the value of an index of satisfaction with respect to the interpersonal dimension [[PERSSAT.sub.j]]; a third such index involving item 9 (venture capital) measured satisfaction with the banker as a supplier of venture funding. Thus, each of these indices measures the j'th respondent's satisfaction for the factor in question, where satisfaction is weighted by the levels of importance attached to each item by the respondent. Kronbach's alpha for these indices were 0.89 for the "interpersonal" satisfaction measure and 0.75 for the "business" factor.

Table 8 presents the means of each of these measures of satisfaction across gender [TABULAR DATA FOR TABLE 7 OMITTED] lines. The statistical significance of each pairing of satisfaction scores is also reported. It was clear from these results that women business owners, when compared with men counterparts, come away from the banking experience with much less of a sense of having been treated with respect and are less comfortable about their banking experience. It also appeared that these feelings pervade their satisfaction with the "business" dimension; women owners seem to be less satisfied, on average, than male business owners with both aspects of the banking relationship.


Previous research and anecdotal evidence has suggested that financial institutions treat female and male small business owners differently. It has not yet been resolved to what extent systemic factors account for any observed differential treatment. This study found that, after accounting for structural differences between male and female business owners, no difference remained in the rate of loan rejections; nor did any differences persist in other objective measures of terms of credit.

This study has shown that measures of business risk are closely associated with [TABULAR DATA FOR TABLE 8 OMITTED] credit terms: firms with declining rates of growth are more likely to have their loans rejected; depth of financial management and the track record of the firm seem to be the variables that are most closely related with loan turndowns. The entry of variables representing industrial sector and geographical location may be an indication that conditions faced by the firm also play an important role in credit decision making. To secure access to credit, owners should strive for professional management of their financial function.

Firm size, the existence of liquid assets, professional financial management, type of financial institution approached for credit, average rate of annual sales growth, capacity (sales), conditions (rate of growth), and collateral (type of assets) seem to play important roles in determining terms of credit.

It was found that gender is closely related with most, if not all, of the above factors. In general, women-owned businesses are smaller (that is, they have lower volumes of sales). Women small business owners have (on average) less capacity, less capital, a narrower range of collateral, and an unproven track record or character relative to male counterparts. This may have an adverse effect on their perceived capacity to service or to repay their loans. Hence, they may in fact experience greater difficulty in obtaining credit, on average. Ceteris paribus, however, the terms of bank credit did not differ between female and male small business owners. In general, all firms with low volumes of sales are likely to encounter such difficulty and to face high interest rates.

Various analyses also indicated that women small business owners are overrepresented in retailing compared to men-owned businesses. With little variation, firms in retailing are known to have few pledgable assets. In case of liquidation, the assets of a retailing concern have low resale value. That is, the amount realized may not suffice as security for a new line of credit.

It was also found that women small business owners have less managerial experience than men owners. Most small firms fail before their fifth year. While numerous factors may lead to demise, lack of managerial experience plays a role. The experience that, on average, women business owners lack may have an adverse effect on how their character is judged. Their firms might be perceived as being more risky or prone to failure.

Women tend to have completed more years of education than male counterparts. This could have a positive effect on their character, possibly a balance to their lack of managerial experience. However, women business owners were less likely than men to have professional degrees. Lack of specialization may raise perceived risk associated with women-owned businesses.

Women small business owners are more likely than men to be sole owners, firms generally known to be under-capitalized relative to firms with more owners. However, in addition to financial capital, human capital is often a problem for such firms. Firms with several owners have access to a diversity of skills and talents that may elude other firms.

Women were more likely than men to report having personal assets as collateral. In contrast, men were more likely than women to report availability of business assets as collateral. Personal assets are more likely to be shared than business assets. Consequently, owners with personal assets may be more likely to be requested for spousal co-signature than owners with business assets.

Of particular concern were the findings that women were more likely than are men to perceive that they were not given due respect by financial institutions, that they did not think that their account managers were easy to talk to, and that they were more likely than men to report that they were not made comfortable by financial institutions. This dissatisfaction with the interpersonal dimension of the banking relationship may explain the dilemma that confronted Buttner and Rosen in their 1992 study. This finding explains why women business owners tend to express feelings of gender bias when, in fact, the terms of lending were not significantly different from those advanced men business owners. To improve relations with their female business clients, financial institutions may need to reassess their marketing strategies to women as well as their policies regarding the hiring and training of account managers. However, banking institutions, through standardization and centralization of the lending decision, appear to have mitigated successfully the potential role of discrimination.



1. Caisses Populaires and Credit Unions are Canadian Bank-like institutions that specialize in personal banking. In several Canadian provinces they are not permitted by law to be involved in business banking. In general, such institutions are co-operative and are usually associated with other institutions (for example, caisses populaires have historically evolved from associations with parish churches in Quebec and credit unions tend to be associated with such institutions as the civil service, firefighters' associations, etc.).

2. One of the attributes of this segment of the literature is the variety of methodologies, insofar as findings sometimes appear to be methodology-dependent. For example, Wynant and Hatch (1991) and Buttner and Rosen (1989) rely exclusively on univariate statistical techniques. However, this approach is limited in at least two important respects: such methods ignore the interrelationship among the independent variables; also, while individual univariate analyses may not prove significant, a single multivariate analysis might (Stevens, 1986). This is due to the fact that multivariate techniques can consider linear combinations of the independent variables which univariate tests ignore. Inconsistencies also arise because of other methodological approaches. These include the tendency to ask leading questions; lack of male controls; and conclusions based exclusively on the subjective perceptions of women small business owners.

3. In particular, Riding and Swift found

apparent gender-related differences in . . . rates of loan approvals, cosignature requirements, requirements for loan collateral, and interest rates on loans and lines of credit . . . can be accounted for by the differences in the characteristics of male- and female-owned businesses

Only in the requirements for collateral for lines of credit did Riding and Swift note any residual difference between the treatment of male and female owners. This is not unexpected given the high likelihood of spurious results applied to repeated univariate tests. Moreover, this difference could have alternative explanations. Structural factors (such as years of managerial experience, asset structure, and financial background) were not included in their research and are possible explanations for the discrepancy. Nonetheless, this finding has been misinterpreted in both the popular media (Languedoc, 1988) and in the academic literature (Belcourt, 1991) as proof that gender discrimination exists.

4. It is worth noting that the response rates from men- and women-owned businesses differed to a statistically significant extent. As noted in previous research (and confirmed by the findings of this work) aggregate women-owned businesses tend to be smaller and newer than businesses owned by men. In general, women owned businesses to make as much use of financial professionals. Accordingly, this difference in response rates is not unreasonable.

5. This is a reasonable dichotomization. This distribution of the ratio of amount received to amount applied for was such that 85.2 percent of the respondents received more than 90 percent of the amount requested. To all intents and purposes, they obtained that which they had sought. Conversely, 11.6 percent of the respondents received less than 20 percent of the amounts they had sought. Thus, only 3.2 percent of the respondents received loans of more than 20 percent but less than 90 percent of what they had sought.

6. Table 5 presents findings for only one of the alternative definitions of the gender of firm ownership: exclusively women-owned firms vs. exclusively men-owned firms. Findings were robust across all three definitions of gender of ownership.

7. It should be noted that logistic regression and linear regression were not the only techniques employed in this research; subclassification on propensity scores was also employed, with similar results. The findings of the entire analysis are reported in Fabowale (1991). Briefly, for each of the analyses described here, two separate sets of propensity scores calculated. The first set of propensity scores was the expected probability that a firm's primary owner would be a woman based on its vector of systemic structural factors. The second set of propensity scores was the expected probability that a firm's primary owner would be a woman based on its vector of credit terms. Quintiles of the first set of propensity scores representing measures of business risk were created. The means and standard errors of the propensity scores that represented terms of credit were then calculated for female and male small business owners within each quintile. Subsequently, the average gender-related difference in credit terms was then tested for statistical difference. It was found that when gender was used as a dependent variable, only the propensity score on measures of business risk seemed to be strongly associated with gender. The variables that expressed the terms of credit did not prove significant.


Belcourt, M. (1991). From the frying pan into the fire: Exploring entrepreneurship as a solution to the glass ceiling. Journal of Small Business and Entrepreneurship, 8(3), 49-55.

Belcourt, M., Burke, R., & Lee-Gosselin, H. (1991). The glass box: Women business owners in Canada. Ottawa: Canadian Advisory Council on the Status of Women.

Buttner, E. H., & Rosen, B. (1988). Bank loan officers' perceptions of the characteristics of men, women, and successful entrepreneurs. Journal of Business Venturing, 3(3), 249-258.

Buttner, E. H., & Rosen, B. (1989). Funding new business ventures: Are decision makers biased against women entrepreneurs? Journal of Business Venturing, 4(4), 249-261.

Buttner, E. H., & Rosen, B. (1992). Rejection in the loan application process: Male and female entrepreneurs' perceptions and subsequent intentions. Journal of Small Business Management, 30,(1), 58-65.

Collom, F. (1982). Canadian women owner managers. Working Papers, Small Business Secretariat, Department of Industry, Trade, and Commerce, Government of Canada, Ottawa.

Davis, D., & Cosenza, R. (1988). Business research for decision-making. PWS-Kent: Boston.

Fabowale, L. (1992). An empirical investigation of financial institutions' terms of credit to female owners of small businesses. Master's thesis, Management Studies, Carleton University.

Fertuck, L. (1981). Small business financing and non-bank financial institution volume I: Text: A study for the small business financing review. Toronto: Facsym Limited.

Grant, M. (1986). Update: Small business and banks. Canadian Banker, 3(2), 12-16.

Haines, G., Riding, A., & Thomas R. (1989). Small business loan turndowns by Canadian chartered banks. Carleton University School of Business Working Paper Series.

Hisrich, R., & Brush, C. (1986a). Women and minority entrepreneurs: A comparative analysis. Working Paper, The University of Tulsa.

Hisrich, R., & Brush, C. (1986b). The woman entrepreneur: Starting, financing and managing a successful new business. Lexington, MA: Lexington Books.

Kryzanowski, L., & Bertin-Boussu, E. (1984). Equal access to credit: Lenders' attitudes towards and applicant's sex and marital status. International Journal of Women's Studies, 4(3), 213-233.

Languedoc, C. (1988). Women have tougher time getting bank credit: Study: Financial Post, December 19, 3.

Litton, M. (1987). Women mean business: Successful strategies for starting your own business. Toronto: Key Porter Books. Orser, B. J, Riding, A., & Swift, C. (1993). Banking experiences of Canadian micro-businesses. Journal of Enterprising Culture, 1 (3), 321-345.

Orser, B. J., Foster, M., & Mueller, R. (1991). A pilot study of Canadian perspectives on sex role stereotypes and requisite managerial characteristics. Best Paper Proceedings, Women in Management Division, Administrative Sciences Association of Canada Annual Conference.

Pellegrino, E., & Reece, B. (1982). Perceived formative and operational problems encountered by female entrepreneurs in retail and small firms. Journal of Small Business Management, 20(2), 11-24.

Poapst, J. (1981). Financing gaps. Small business financing and non-bank financial institution volume I: Text: A study for the small business financing review. Toronto: Facsym Limited.

Riding, A., & Swift, C. (1990). Women business owners and terms of credit: Some empirical findings of the Canadian experience. Journal of Business Venturing, 5(5), 327-340.

Ross, S., Westerfield, R., Jordan, B., & Roberts, G. (1993). Fundamentals of corporate finance: First Canadian edition. Irwin. Homewood, Ill.

Schwartz, E. (1979). Entrepreneurship: A new female frontier. Journal of Contemporary Business, Winter, 47-76.

Stevens, J. (1986). Applied multivariate statistics for the social sciences. Hilsdale, N J: Lawrence Erlbaum Associates.

Stevenson, L. (1986). Against all odds: The entrepreneurship of women. Journal of Small Business Management, 24(4), 30-36.

Thornton, D. (1981). Small business financing policies and practices: An interview survey of managers. Small business financing and non-bank financial institutions. Toronto: Facsym Ltd.

Wynant, L., & Hatch, J. (1991). Banks and small business borrowers. London, Ontario: The Western Business School, The University of Western Ontario.

Lola Fabowale conducted thesis research that provided the data on which this study is based. She graduated from Carleton University in 1991.

Barbara Orser is Professor of Business Management at Ryerson Polytechnic University in Toronto and a Research Associate of the Bradford University Management Centre in the United Kingdom.

Allan Riding is Professor of Finance at Carleton University in Ottawa, Canada.

The authors wish to acknowledge the generous and gracious assistance of the Canadian Federation of Independent Business (CFIB) in this research. The CFIB kindly provided access to data from their survey of banking experiences and were proactive in suggesting this as a topic worthy of investigation. The text of the paper is based on the thesis research of Lola Fabowale. The authors acknowledge the assistance of Dr. Roland Thomas, Dr. Lorraine Dyke, Dr. Louise Heslop, and Dr. Ian Lee, all members of the faculty of the Carleton University School of Business, for their advice and support. The editorial assistance of Miss Diane Ryerson is gratefully valued. Errors and omissions that remain are contribution of authors.
COPYRIGHT 1995 Sage Publications, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1995 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:includes appendices
Author:Fabowale, Lola; Orser, Barbara; Riding, Allan
Publication:Entrepreneurship: Theory and Practice
Date:Jun 22, 1995
Previous Article:Understanding the small enterprise financial objective function.
Next Article:The entrepreneurial decision: economic theory and empirical evidence.

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |