Gaps in SME Financing: An Analytical Framework

Debt Markets for Canadian SMEs

Overview

The two primary means of debt financing are lines of credit and business loans. The CFIB (2001) reports that almost 71.2 per cent of respondents hold a line of credit with their financial institution, and 41 per cent have a business loan. Other means of debt financing include commercial and personal, mortgages, personal loans, supplier credit, overdraft protection, and credit cards.

The tables that follow summarize the breakdown of the debt market for SMEs according to the various types of institutions and categories of borrowing. The purpose of these tables is to provide a quantitative sense of the structure of the debt market and how it is broken down across the three primary types of SME financing:

• commercial loans,
• non-residential mortgages, and
• lease contracts.

Table 2 illustrates shifts in lending patterns between 1994 and 1998 for the various groupings of institutional participants. Table 3 outlines the breakdown of 1998 lending by category of loan and institution type.

Table 2: Principal Sources of Debt for Canadian SMEs: 1998 vs. 1994^{Footnote 3}

Source	1994 ($ millions)	1998 ($ millions)	Change ($ millions)	Change (%)
Domestic Banks	44042	58232	14190	32%
Credit Unions & Caisses Populaires	14,093	16,153	2,060	15%
Specialized Finance Companies	7,928	17,410	9,482	120%
Crown Corporations	6,166	9,234	3,068	50%
Foreign Banks	4,404	3,692	-712	-16%
Trust & Mortgage Loan	4,404	2,442	-1,962	-45%
Life Insurers	3,523	2,639	-884	-25%
Credit Card Companies	3,523	5,778	2,255	64%
Total	88,084	115580	27,496	31%

Table 3: Types and Sources of Debt Used by Canadian SMEs^{Footnote 3}

Source	1998 ($ millions)	Proportion of Lending	Commercial Loans	Non- Residential Mortgages	Lease Contracts
Domestic Banks	58,232	50.4%	53,072	4,276	884
Credit Unions & Caisses Populaires	16,153	14.0	9,126	7,027	0
Specialized Finance Companies	17,410	15.1%	11,129	146	6,135
Crown Corporations	9,234	8.0%	9,234	0	0
Foreign Banks	3,692	3.2%	3,455	160	77
Trust & Mortgage Loan	2,442	2.1%	255	2,074	113
Life Insurers	2,639	2.3%	0	2,617	22
Credit Card Companies	5,778	5.0%
Total	115,580	100%	86,271	16,300	7,231

These tables convey that commercial loans from domestic banks, crown corporations, and credit cooperatives (Credit Unions & Caisses Populaires) are by far the dominant form of financing for SMEs. Collectively, these account for more than $71 billion and 62 percent of all debt financing in 1998. The second largest segment (more than $17 billion, 15 percent of lending during 1998) is that accounted for by commercial loans and leases from specialized finance companies. Nonresidential mortgages account for much of the remaining lending to SMEs.

These data also convey that the debt side of the market appears to have expanded substantially during the 1994-1998 period. This expansion is particularly evident in terms of commercial loans (additional $19 billion) and specialized finance companies (additional $9 billion). On the other hand, lending by Trust & Mortgage loan companies and by Life Insurer companies has shrunk considerably, reflecting changes in the non-residential mortgage market.

Commercial Bank Lending

According to data published by the Canadian Bankers Association as of the second quarter of 2001:

• more than $50 billion in bank credit facilities were reported as authorizations of less than $250,000 as of the second quarter of 2001. Of this total, $32.6 billion was outstanding, a 64.9 percent drawdown of authorized credit.
• The 789,000 customers represent an increase of 18.5 percent since the first quarter of 1996.
• The increase in authorized credit since 1996 Q1 was 5.7 percent and the $32.6 billion of credit outstanding was actually slightly less than the $33.5 billion outstanding as of the first quarter of 1996.

These data suggest, and Figure 1 demonstrates, that credit authorizations of less than $250,000 have been advanced to a greater number of clients but that, on average, SMEs have been making progressively less use of the credit available.

Figure 1: Measures of SME Bank Lending Activity

D

The special situation of lending to knowledge-based firms, given the BDC assertion of a gap, warrants specific attention. According to CBA data on loan authorizations of less that $250,000, the lending above includes credit authorizations of $806.4 million to firms in the KBI sector of which $429.4 million has been drawn down (53.25 percent).^{Footnote 4} It is worth noting that the draw down of authorized credit is much lower than the 64.9 percent rate reported above.

These data suggest that, on average, credit is available for firms that have been granted a credit facility.

The low drawdown of credit facilities suggests that, on average, for those firms that have been granted a credit facility additional capital is available. It will be seen that this finding is contrary to some views about what constitutes a capital market gap.

Wynant and Hatch (1991) undertook an exhaustive analysis of bank lending to small firms. They used bank-maintained loan files to document the credit decision process and terms of credit encountered by Canadian SMEs. Subsequently, Haines and Riding (1994), in a project supported jointly by Industry Canada, the Canadian Bankers Association, and the Canadian Federation of Independent Business (CFIB) also studied terms of credit to SMEs.

Ongoing periodic assessments of terms of lending between banks and Canadian SMEs have been conducted by both the CBA and the CFIB. On behalf of the CBA, Thompson Lightstone and Company (1996, 1997, 1998) have undertaken annual surveys of the Canadian debt market as it pertains to SMEs. In addition, the CFIB has also periodically sought to document terms of credit to SMEs. Both groups lack access to the quality of data that is available to Statistics Canada and are unable to achieve the magnitude of response rates typical of Statistics Canada surveying. Therefore, the SME FDI will be able to document, with high levels of reliability, the terms of credit available to Canadian SMEs and to gauge the relative importance of various forms of financial capital.

Research Task
The SME FDI should document terms of credit to Canadian SMEs and to provide breakdowns of credit policy (interest rates, collateral requirements, loan turndowns, etc.) across a variety of dimensions that include sector, size of firm, etc. Such breakdowns, however, do not provide definitive evidence of gaps or imperfections. This is because of the complex interactions of various elements in the credit decision.

These various analyses have also led to several analyses of the bank-SME interface. Published work by Fabowale et al. (1991a, 1991b, 1995), Feeney et al. (1999), Haines et al. (1991, 1993, 1998, 1999), Orser, Riding, and Swift (1993), Riding and Swift (1990), and Wynant and Hatch (1991) have used CFIB data or CBA to document the nature of the bank-SME interface. These works have found empirically that determinants of access to and terms of, bank lending include the size of firm or loan, the industry sector, and other factors. In general, these works have not accounted for the experience and skills of the business owners. Thus, there may be missing variables biases in these results.

Determinants of Terms of Credit

Several recent studies have documented determinants of access to, and terms of, credit for SME borrowers. Wynant and Hatch (1991) examined 1,539 bank credit files supplemented by responses from 371 owners of SMEs and 341 questionnaires administered to loan account managers. From these data, they investigated a broad range of issues related to the relationship between SME borrowers and bank lenders. From this analysis, they concluded that the following variables were each correlated significantly with loan turndowns:

• Age of business
• Provision of business planning information
• Length of time firm was owned by urrent owners
• Duration of relationship with the bank;
• Burden coverage ratio;
• Number of account managers;
• Size of loan facility.

In addition, they determined that the interest rate on loans were significantly correlated with each of:

• Industry sector;
• Age of business;
• Size of firm (level of sales);
• Bank's assessment of management capability;
• Length of time firm was owned by urrent owners
• Duration of relationship with the bank;
• Burden coverage ratio;
• Size of loan;
• Risk rating;
• Administrative effort.

In their analysis, Wynant and Hatch also reported that fees were correlated with each of: location (urban vs. remote), age of business^{Footnote 5}, size (level of sales), years owned by current owner(s), years of owner's management experience, duration of relationship with the bank.

While Wynant and Hatch's results are suggestive, they are not definitive. The analyses reported by Wynant and Hatch were largely univariate in nature. In other words, they correlated their various dependent variables (turndown, interest rate, fees) one at a time against potential determinants; for example they correlated interest rate on loan size (level of sales). They appeared to repeat this analysis for each pairing of dependent variable and potential determinant. This process of conducting multiple univariate tests presents two statistical problems.

First, multiple univariate tests can lead to conclusions of statistical significance when no correlations are present. For example, if one were to undertake ten successive tests, each at a five percent level of statistical significance, the probability of finding at least one significant test when only random error is present is 1-(0.95)¹⁰ = 40%. Second, univariate testing can mask the real relationships that might be at work. For example, Wynant and Hatch reported that interest rates were correlated with each of risk rating and burden coverage ratio. It is likely, however, that risk and burden coverage ratio are correlated with each other. Thus, the associations reported by the authors do not necessarily suggest causality.

To investigate determinants of terms of credit more rigorously, Fabowale, Riding and Swift (1991b) employed a weighted least squares multiple regression framework applied 1,831 responses to a CFIB questionnaire on access to capital. They found that interest rates on new term loans and new lines of credit were significantly related to the size of the firm (log of level of sales) and the size of the loan. In addition, they found that interest rates on new lines of credit were also correlated with the legal status of the borrower firm and the sector. Increases in existing lines of credit were related to the firm's level of sales, the number of employees, a location measure (urban vs. rural), and the size of the loan.

Using the same data, Haines, Riding, and Thomas (1994) found that significant determinants of loan turndowns (among 20 potential explanatory variables) were industrial sector (manufacturing vs, other sectors), whether or not the owner conducted personal banking at that bank, and whether or not the firm had a history of financial distress. Again using the 1991 CFIB data, Fabowale, Riding and Swift (1991a) found that the addition of gender of owner did not add any explanatory power to regression models of loan turndowns.

Subsequently, the CBA (1997, 1998) examined 'drivers of loan turndowns', also using a multivariate analysis context. They investigated hypothesized root cause of acceptance/turndown that included:

• Age of business;
• Sale volume;
• Number of employees;
• Region; industrial sector;
• Gender of owner;
• Ownership structure of business;
• Whether or not business was a franchise;
• Whether or not business was home-based;
• Whether or not business was rural-based;
• Geographic region;
• Whether or not business was tourism-related;
• Whether or customer was an existing bank client;
• Level of debt of the firm;
• Proportion of business' assets financed by banks;
• Type of loan requested;
• Type of credit facility requested;
• Purpose of the loan;
• Presence of spousal guarantees;
• Age of owner;
• Business experience of owner;
• Education of owner;
• Net worth of owner.

They found that significant correlates with loan turndowns included that the business was an existing borrower at the bank, size of firm, region, age of firm, net worth of owner, and type of request (formal, renewal).

These studies are among those that have statistically investigated determinants of access to, and terms of, bank credit using Canadian data. Additional studies include periodic assessments of the bank-SME interface undertaken by the CFIB, the Chamber of Commerce, and the then (1995) Canadian Labour Market and Productivity Centre) An overall perspective reveals that there is a lack of consensus about the findings. In part this results from the use of different measures of potential determinants of credit, sampling difficulties (includes difficulty of obtaining a sufficiently large sample to use a wide variety of categories, sampling biases, non-response biases, selection biases, and collinearity).

Specialized Finance Companies: Asset-based Financing and Leasing

According to the Canadian Finance and Leasing Association (CFLA),

The most frequent form of asset-based financing provided by specialized finance firms is a lease: "an agreement under which the owner of the equipment conveys to the user the right to use the equipment in return for a number of specified payments over an agreed period of time" (CFLA). The CFLA distinguishes two categories of leases. Capital leases are usually used to finance equipment for the majority of its useful life and "there is a reasonable assurance that the lessee will obtain ownership of the equipment by the end of the lease term."

The second type of lease, operating leases, are usually used to finance equipment for less than its useful life. At the end of the lease term, the lessee usually returns the equipment to the lessor without further obligation. In Canada, a wide variety of equipment can be leased. Leasing is used frequently for office, medical and dental equipment, computer equipment, trucks and trailers, construction equipment, railway rolling stock, busses, etc.

The CFLA quotes the International Finance Corporation (the private sector arm of the World Bank) to the effect that "asset-based financing and leasing are particularly attractive to small and medium-sized businesses (SME)". Leasing is attractive to some businesses because it frees up cash that would otherwise be tied up in fixed assets and that would not be available to finance other dimensions of the firm, such as working capital. In an environment where access to capital may be difficult, leasing may therefore provide a useful complement to traditional bank financing, augmenting the pool of available credit to SMEs. In addition, it is often asserted that lessors use different criteria for advancing credit by means of a lease than would be used for loans from commercial lenders. For example, the CFLA claims that

Considerable consolidation has characterized the supply side of the specialized finance market. Consequently, asset-based financing and leasing are becoming more standardized in terms of products and approaches. The CFLA asserts that the primary driver of this consolidation is the goal of achieving economies of scale. The result has been what the CFLA refers to as a polarization of the industry. At one extreme are well-capitalized lessors that operate on a national scale with an extensive infrastructure and which have good access to funding. At the other extreme are smaller, niche, participants who are either regionally based or focused on a particular segment.

The CFLA estimates that the asset-based financing and leasing industry financed as much as 25 percent of total business investment in machinery and equipment. The CFLA compares this to five percent of total business investment in machinery and equipment fifteen years ago. However, these estimates are based on surveys of members of the CFLA and there is little independent confirmation of the relative importance of asset based financing with respect to SMEs or of the growth of this form of asset acquisition. The Financing Data Initiative will clarify the importance of this segment of the market and document its usage across industrial sectors, sizes and stages of firm, and other dimensions of interest.

Research Task:
The SME FDI should be able to document the usage of asset based financing across sectors, geographic areas, sizes and stages of firms, and by other dimensions of interest. This will allow an assessment of the importance of this category of financing to SMEs and to particular categories of SMEs.

Hypothesis:
Asset based financing is equally employed by firms in all areas of the country, across all industrial sectors, and across all stages of business development.

Mezzanine Financing

Mezzanine financing is a means of raising growth capital for firms that are well beyond the start up stage but not yet ready to go public. Mezzanine financing usually involves the use of subordinated debt through private-placement transactions with institutional lenders, and equity investment. Typical deals sizes range from $1 million to $20 million. (Subordinated debt refers to loans secured by a general claim on assets, as distinct from claims secured by specific assets. In case of bankruptcy, lenders' holdings of subordinated debt rank next behind secured lenders and ahead of equity holders.

In a mezzanine deal, investment bankers look for strong profits and cash flows sufficient to finance the debt obligations. Part of the financing is typically in the form of subordinated debt borrowing and part is in the form of equity from the sale of stock in the business to the same investor from which debt is obtained. Usually, mezzanine financing is a transitional financing arrangement such that the mezzanine financier(s) are either bought out upon an IPO or the deals are refinanced after five to seven years.

Often, mezzanine financing is available from institutional venture capital companies. Approximately one-quarter of the full members of the Canadian Venture Capital Association list mezzanine financing or subordinated debt among the types of financing they provide. In addition, other firms specialize in mezzanine financing. Because of the size of the deals and the fact that most are private, no reliable data yet exist to document the importance of this segment of the marketplace or current trends. This leads to a research tasks and hypotheses that the SME FDI might investigate:

Research Task
The SME FDI should be able to document the usage of mezzanine financing across sectors, geographic areas, sizes and stages of firms, and by other dimensions of interest. This will allow an assessment of the importance of this category of financing to SMEs and to particular categories of SMEs.

Hypothesis:
Mezzanine financing is concentrated in firms that are at relatively advanced stages of development.

Hypothesis:
Mezzanine financing is equally employed in all areas of the country and across all industrial sectors.

Consequently, the SME FDI may be able to provide, for the first time, reliable estimates of the importance of mezzanine financing in Canada. It should also be able to document breakdowns of its usage by size of firm, stage of development, sector, geographic location, and other dimensions of interest.

Gaps in the Debt Market for SMEs

Gaps in the Credit Market: The Practical Perspective

The presence of gaps in the debt market for SMEs has been postulated for some time. For example, the perception of size-related gaps was part of the premise behind the Canadian Small Business Loans Act of 1961. Work conducted on behalf of the BDC identified four perceived gaps that relate to the debt markets in which SMEs participate.

The first three of these perceived gaps have in common that financial institutions are unable to adequately meet the needs of particular groups of potential borrowers: small firms, risky firms, and knowledge-based enterprises. Stated differently, the BDC findings suggest that firm size, knowledge orientation, and risk may be dimensions according to which lenders ration credit. Therefore, this empirical finding may be consistent with predictions from theories of information asymmetry.

Based on periodic surveys of its members, the CFIB (2001) also asserts the presence of gaps. It claims that one in five SMEs were unable to obtain the necessary level of financing and that a high proportion of these are "young high performers". The CFIB findings, while suggestive, are not conclusive. This is so for several reasons. First, the CFIB surveys are limited to its members, introducing a selection bias. Second, the CFIB analysis, even though based on 10,024 respondents represents a substantive non-response bias in that the survey was apparently distributed to the entire membership. If the CFIB membership totals more than 100,000 firms, the response rate to the survey is less than 10 percent. Non-response often signals that the topic is not central to those who chose not to fill in the questionnaire. Third, the level of analyses is generally limited to basic breakdowns, an approach that can result in misleading conclusions.

Gaps in the Credit Market: The Academic Perspective

Academic research may also suggest the presence of gaps in the debt market. The academic definition of a gap, however, is much more specific and consistent and is related to a sizable literature on credit rationing. This section attempts to capture the essence what Berger and Udell (1992, p. 1048) say is "... an entire generation of work on credit rationing based on an information-theoretic approach ... [about which] there remains little consensus".^{Footnote 6}

These theoretical models often assume that entrepreneurs possess salient private knowledge that is not shared with the lender or investor. Consequently, capital suppliers (lenders) cannot differentiate between a 'high-quality' firm and a 'low-quality' firm. Adverse selection can result. The capital provider therefore structures contracts to deal with adverse selection, potentially leading to credit rationing. Moral hazard is another potential consequence of information asymmetry and relates to the capital supplier's inability to control fully how the entrepreneur-borrower uses funds provided. Entrepreneurs can conceivably benefit economically by taking actions that hurt the capital supplier (for example, using borrowed funds to invest in higher risk projects than those approved by the lender).

To avoid this situation, capital suppliers can implement (costly) practices that discourage entrepreneurs from acting against the interests of the lender or investor, and these precautionary actions can lead to credit rationing. In their seminal article, Stiglitz and Weiss (1981) argue that banks may restrict lending by practicing capital rationing – denying credit to particular classes of borrowers.

In theory, information asymmetry can be mitigated in at least three ways, each of which reduces information opacity and provides salient information to investors and lenders:

• through a firm's ability to somehow signal its credit worthiness;
• by having developed a strong relationship with its lender; and,
• through due diligence and the lender's examination of the business plan and other lender requirements for documentation.

Some have argued that the availability of (and the willingness to use) collateral is one means of signaling creditworthiness (Bester, 1987), Chan and Kanatas. 1985; Toivanen and Cressy, 2000). Boot, Thakor and Udell (1991) find empirically that collateral is a powerful instrument for dealing with moral hazard.

If these theories are correct, then firms that are most likely to receive credit are those that are able to signal the creditworthiness of the firm, those that have an established relationship with the lender, and those that have been able to reduce information asymmetry through effective communications. This suggests the "pecking order" modeled in Table 4, according to which firms with the greatest degree of difficulty accessing debt capital would be new firms that lack collateral and effective management.

Table 4: Degrees of Creditworthiness According to Theories of Information Asymmetry

D

Hypothesis
Firms with least access to credit are those that do not have established relationships with lenders (e.g., new firms), those that are unable to provide collateral (a signal of creditworthiness), and those whose management are unable to communicate effectively the firm's creditworthiness.

Bridging the Academic and Practical Perspectives

It is not clear, from the theoretical academic literature, whether or not credit rationing is practiced nor the extent to which it is a problem. As Berger and Udell also note (1992, p. 1048):

Is it possible that if credit rationing is present that rationing might be based on such dimensions as size, risk, or knowledge – as suggested by the BDC? Hillier and Ibrahimo (1993, p. 287) state that:

While economic theory is consistent with the idea that observationally distinct classes of borrowers may be rationed, theory does not specify the categories across which credit rationing might occur (if it exists!). If credit is rationed, it therefore remains to identify the basis by which lenders discriminate among borrowers: on what dimensions do lenders deny loans, possibly leading to gaps based on credit rationing?

Potential Bases for Credit Rationing Gaps

Collateral. The studies summarized above suggest that lenders may be more likely to approve loans to firms that are able to provide collateral and to those firms that have established long term relationships with lenders. As suggested by the BDC the need for collateral may militate against technology based firms. Moreover, many technology-based firms are perforce relatively young firms and may not yet have developed relationships with bankers. Also, as Åstebro Johannsen and MacKay (2000) note, "…intellectual assets of high technology firms are more difficult to value than the brick and mortar of low technology firms". This "information opacity" may well make it relatively difficult (at least in theory) for knowledge-based firms to access debt. This argument, of course, begs the question of the extent to which debt is an appropriate source of capital for knowledge-based businesses. MacIntosh (1994) suggests that debt is not the most appropriate source of early-stage capital, a suggestion echoed by Leitinger and Schofer (2000) as well as by Brierley (2001).

Technology Orientation. Haines, Riding, and Swift (1993) used CFIB data to compare empirically loan turndown rates and terms of credit between technology-based firms and nontechnology-based companies. They concluded that, after allowing for traditional determinants of credit policy (sector, size of firm, age of firm, etc.), loan turndown rates and other dimensions of bank credit policy did not differ between the two categories of firms. From their analyses, they found that what appeared to be a gap based on technology orientation appeared to be more related to technology-based firms being small and young. However, it is not clear that this finding is robust across different definitions of "knowledge-based firm".

Growth may be another dimension that forms a basis for a gap based on capital rationing and for which a gap may be claimed. As Binks and Ennew (1996) argue, high growth firms may be more informationally opaque and may face a greater degree of difficulty obtaining financing, consistent with findings also reported by the CFIB (2001) to the effect that "young high performers" have the most difficulty accessing capital.

Gender may also be a dimension on which gaps may be present. Marleau (1995) contends that women business owners also face particular difficulty obtaining credit. However, Fabowale and her associates (1995) and Haines and his colleagues (1999) have found that rationing on the basis of gender may be more related to systemic attributes (e.g., sector, age of firm) of womenowned businesses.

Size of firm (or size of loan) may also be a dimension across which a financing gap may be present. This is suggested by the BDC and CFIB (2001) research and is also consistent with the theoretical and empirical findings of Toivanen and Cressy (2000). Toivanen and Cressy contend that credit policy is influenced by a fixed costs component associated with due diligence and monitoring. They argue that the size of this component relative to the size of the loan is a partial determinant of access to, and terms of, credit. Likewise, Wynant and Hatch (1991), Haines and Riding (1995), and Thompson Lghtstone (1996, 1997, 1998) also found empirical evidence that small firms face particular challenges with terms of credit and access to bank loans. In addition, of course, the understanding that small firms face disproportionate lack of access to credit is the basis for loan guarantee programs to small firms in many countries. It is not clear to what extent such schemes have mitigated access to credit on the basis of firm size. This forms a research task that the SME FDI might address.

Research Task
Data obtained under the terms of the SME FDI may be able to inform the extent to which the SBLA and the CSBFA achieve incrementality in terms of loans to smaller businesses.

Risk. Finally the BDC contends that risk may be a dimension across which a financing gap might exist. The theories of information asymmetry also suggest that credit rationing may reflect risk. However, the definition of risk in the theoretical may differ from that implicit in the BDC's research. Moreover, it can be argued that the role of the banks is in fact to discriminate based on risk. Thus, it is not clear whether a gap based on the dimension of risk is material to a societally efficient allocation of credit.

Empirical Findings about Credit Rationing and Market Imperfections

While the theories of credit rationing are highly developed, the same cannot be said about the state of empirical testing. As noted by Berger and Udell (1992, 1995) empirical studies of credit rationing requires micro data (firm-specific information). However, such data seldom document rejections of loan applications and researchers have resorted to analysis of such factors as the 'stickyness' of interest rates on loans.

Perhaps the most widely-cited analyses are those of Berger and Udell (1992, 1995) who conducted empirical tests of credit rationing theories, interest rates behaviour, collateral requirements, and business relationships with banks. They used a data set of about 3,400 telephone surveys conducted by the Federal Reserve Board and the SBA. Their work focused on the duration of a business-bank relationship as a measure of strength and observed the influence of 'stronger' relationships and contract terms. The study found that borrowers with longer banking relationships pay lower interest rates and are less likely to pledge collateral. These findings were both statistically and economically significant. The findings are consistent with the theoretical predictions of Petersen and Rajan (1994) and Boot and Thakor (1994) and support the literature on the role of banks as information producers.

Testing is complicated, however, because scholars have defined credit rationing in slightly different terms. For example Bester (1985, p.850) describes credit rationing as follows:

According to Jaffe and Russell (1976), however,

The two definitions are somewhat ambiguous because they can be interpreted in either of the two ways identified by Stiglitz and Weiss (1981). On the one hand, credit rationing can occur whenever there is a person/firm that is willing to accept the lender's terms of lending yet they are refused a bank loan. On the other hand, credit rationing is said to occur when certain categories of borrowers are systematically denied credit. Jaffee and Russell, however, suggest that size of loan is the issue. However, Jaffee and Russell's definition is not consistent with the data reported regarding credit facilities to SMEs. Recall that, on average, the draw down of credit facilities was low, particularly for KBIs. The lending data reviewed previously suggests that, on average, borrowers are able to avail themselves of more credit than they are using.

Therefore, it remains that credit rationing occurs when firms are refused credit that are willing to accept the required terms (interest rate, collateral, etc.) and which are capable of fulfilling such requirements. Thus, an empirical test for credit rationing must take into account firms' ability to meet terms of credit and then isolate the extent to which the credit decision is varied on the basis of the possible dimensions that may form a basis for credit rationing.

Another implication of credit rationing may provide an alternative avenue for empirical testing. One of the implications of credit rationing is that the law of the single price fails.^{Footnote 7} If, as Hillier and Ibrahimo (1993, p. 276) contend, we "can view the banks as the buyers of risky promises to pay and borrowers as the sellers of such promises", firms that have been turned down for loans would be indistinguishable from firms that have received loans if credit rationing obtains. Thus, a finding consistent with credit rationing would be that the attributes of firms that were turned down do not differ systematically from the attributes firms that received loans. This leads to the following hypothesis.

Hypotheses:
The attributes of firms that have been turned down for loans do not differ from those whose loan applications were successful.

In summary, there appears to be both theoretical and anecdotal evidence to suggest that certain classes of borrower firms may face relatively greater difficulty with access to credit. As Berger and Udell (1992) note, however, the theory-based literature lacks consensus and resolution must rest on empirical methods. To address the possibility that credit is rationed based on firm size, knowledge content, risk, gender, or other factors, the following testable hypothesis, based on both academic theory and the gaps postulated by the BDC, is advanced:

Hypothesis:
Financial institutions are unable to meet the needs of particular classes of borrowers. These classes include (but are not limited to) small businesses, risky firms, and knowledge-based enterprises.

The existence of gaps for these segments would be consistent with credit rationing and the presence of imperfections in the form of information asymmetry.

Another hypothesis arises from the fourth "gap" noted by the BDC:

Hypothesis:
Financial institutions do not provide flexible terms and conditions on their loans to SMEs.

Strictly speaking, this is not a gap in the sense of economic theory. The notion of "flexibility" is a highly subjective construct. At one extreme, this concept could be taken to mean that the loans provided by financial institutions are highly standardized: so called "cookie cutter" loans. The CFIB argues that such loans are a consequence of credit scoring, a practice that lenders are increasingly using. However, it can also be argued that credit scoring reduces the fixed cost component of lenders' due diligence and may therefore result in yet more credit being made available to SMEs. In a finding consistent with this argumentation, Frame, Srinavasan, and Woolsey (2001) conclude that "credit scoring lowers information costs between borrowers and lenders" and attribute an 8.4 percent increase in the portfolio share of small business loans to credit scoring.

Because of the subjective nature of the notion of 'flexibility', this hypothesis may not be directly testable from survey data and requires qualitative analysis of borrowers' perceptions of the nature of 'flexibility' and 'inflexibility' in the context of bank-SME relationships.

The Role of Loan Guarantees

It has long been perceived that smaller firms face disproportionately greater difficulty obtaining loans than larger firms. While the theoretical literature does not explicitly identify size of firm as a rationing criterion, numerous empirical studies have identified size of firm as one determinant of access to, and terms of, bank credit (Wynant and Hatch, 1991; Fabowale, Riding and Swift, 1991; CBA, 1998; CFIB, 1995). It is argued that loans that support the expansion of small enterprises may convey significant benefits to the borrowing firms and, through job creation and retention, to the rest of society. Consequently, governments and trade associations have often intervened in the credit markets by taking on the role of guarantor of loans that financial institutions advance to SMEs. For example, the Small Business Administration in the U.S. provides guarantees of loans made by banks to qualifying small firms. Similar schemes are in effect in, among other countries, Japan, the U.K., Korea, and Germany. In France, Spain, and other nations trade associations take on such roles.^{Footnote 8} Loan guarantee programs are designed in a variety of ways and often they do not appear to reflect guidance from economic theory or experience.

Loan guarantee programs have pre-dated the economic theory of credit rationing. Therefore, it cannot really be said that they represent a response to the issues that relate to credit rationing, yet the idea that a particular class of firms has difficulty with access to credit is consistent with the expectations of credit rationing. Vogel and Adams (1997) point out that, credit market imperfections do not provide a valid rationale for loan guarantee programs. However, they identify, two circumstances under which small firms may face disproportionate difficulty obtaining debt capital, situations that loan guarantees may address.

The first condition arises when the cost of lending to SMEs is too high to be economical for financial institutions. This high cost stems from two sources: the risk premium that lenders could expect and fixed costs of evaluation and monitoring. These are not imperfections in the market. These conditions are part of the normal way in which such markets operate. Nonetheless, the result may well be that SMEs face disproportionate difficulty with access to debt capital.

The second situation arises when lenders place importance on the availability of collateral and if small firms do not have sufficient collateral available in terms of the quality and quantity required. Again, this is not an imperfection in the credit market; rather, this condition is an aspect of the normal operation of credit markets. Experience suggests this situation –where lenders require collateral that SMEs do not have – is common to many young firms and, perhaps, to technology-oriented businesses. Again, the outcome may be disproportionate access to credit among small firms.

Thus, the observation that small firms face particular difficulty accessing capital may indicate credit rationing, relatively high fixed monitoring costs, lack of collateral, or combinations of these situations. To the extent that small firms are disadvantaged, the question of the extent to which loan guarantees redress this disadvantage is begged.

Several attempts have been undertaken, internationally, to measure the impact of loan guarantee programs (Pieda, 1992; Rhyne, 1988). In Canada, Equinox (1997) has found that the program is an extremely efficient mechanism of fostering job creation. However, two important elements of assessing the impact of loan guarantee programs remain problematic. The first is measuring the level of additionality (also known as incrementality) in the program. The second is deriving a benchmark, or control group, against which the expansion, survival, and job-creation benefits might be compared. It is likely that DFI will be able to inform these issues usefully.

One approach would be to compare firms with loan guarantees against those without loan guarantees against firm demographic data (size, age, stage of firm, etc.). In doing so, insights regarding incrementality could be derived. A second approach is to use the DFI baseline survey as the basis for longitudinal follow-up of a sub-sample of firms with loan guarantees as well as of a control sample of counterpart firms that did not use loan guarantees.

^{Footnote 3} Source: Conference Board of Canada, SME Debt Financing, as summarized in Secor (July 2000)

^{Footnote 4} The KBI sector, according to CBA data, comprises firms that are contained in a specified set of four-digit SIC codes.

^{Footnote 5} In the literature of entrepreneurship, there is a debate about determining the age of a business. Reynolds and Miller (1987) pose the problem as follows:

"When is the attempt to initiate an economic enterprise considered? When incorporation occurs? When business cards are printed? When loans are sought? When the first employees is hired? Each criterion has its own set of problems in defining a population of new firms."

Dennis, Dunkelberg, and Dial (1995) distinguish between 'substantive' and 'non-substantive' new businesses. They identify a 'substantive' business as one that has a business telephone number, is located outside the home, and has multiple owners. Thus the date of birth of a firm is not well defined. For example, the Statistics Canada LEAP file defines a new firm (an "entry") as a unit that reports paying one or more employees using a T4 form. While administratively this has a certain appeal and consistency, it is easy to imagine many situations where even substantive and established businesses may not count as enterprises by this definition. Therefore, researchers must understand how the way in which a question is posed on a survey may, or may not, lead to consistent and comparable responses. The year of birth of a firm, not to mention the perceived stage of development of an enterprise, may well depend on the eye of the beholder.

^{Footnote 6} The awarding of the 2001 Nobel Prize in Economics to Spence, Stiglitz, and Akerlof was in recognition of their contributions to the development of these theories of information asymmetry. For example, Stiglitz (with Weiss, 1981) demonstrated theoretically that if lenders are unable to distinguish good credit risks from poor credit risks, the debt market could collapse. In Akerlof's seminal 1970 article, he showed how the unequal distribution of information between buyers and sellers resulted in lower prices for the commodity in question (used cars in his case, but the principles he derived held wide applicability).

^{Footnote 7} The law of the single price states that "all objects with the same observational characteristics should sell at the same price" (Hillier and Ibrahimo, 1993, p. 286).

^{Footnote 8} Since its inception in 1961, the Small Business Loans Act (currently the Canadian Small Business Financing Act, CSBFA) provided federally-guaranteed term loans through approved lenders. To obtain a guaranteed loan, borrowers negotiate a loan with an approved lending institution. Within the terms of eligibility, lenders have full discretion regarding the loan decision and the invocation of the guarantee. The role of the Canadian government is passive. The CSBFA program provides exclusively for guarantees of term loans where the proceeds are used to finance land, premises, equipment, and certain other items. Proceeds may not be used to finance working capital, share acquisition, refinancing, and intangibles (including franchise and operating permits).