SME Credit Availability Around the World: Evidence from the World Bank Enterprise Surveys

Transcription

1 Evidence from the World Bank s Enterprise Survey SME Credit Availability Around the World: Evidence from the World Bank Enterprise Surveys Rebel A. Cole Dept. of Finance Florida Atlantic University Boca Raton, FL USA coler@fau.edu Phone: Andreas Dietrich Institute of Financial Services IFZ Lucerne University of Applied Sciences, Grafenauweg 10, 6304 Zug, Switzerland, andreas.dietrich@hslu.ch DRAFT: Dec. 31, 2017

2 SME Credit Availability Around the World: Evidence from the World Bank Enterprise Surveys ABSTRACT: In this study, we use data from the World Bank s Enterprise Surveys of 133 countries over the period from to test the importance of governance to the availability of credit. We model the credit-allocation process for SMEs into a sequence of three steps. Based upon these three steps, we classify small businesses into four groups based upon their credit needs. In a first step, we analyze which firms do, and do not, need credit. The No-Need firms have received scant attention in the literature even though they typically account for more than half of all small firms. We find that a No-Need firm is more likely to be owned by a foreigner; and is more likely to be classified as a manufacturing or chemicals company, and that firms located in developing, but not developed, countries with better governance are less likely to need credit. In a second step, we analyze Discouraged firms those which reported a need for credit but failed to apply because they feared being turned down or thought that interest rates and collateral requirements were too unfavorable. Like No-Need firms, Discouraged firms have received scant attention in the literature. Discouraged firms typically outnumber firms that apply for and are denied credit. Among firms that need credit, we find that a Discouraged firm is both younger and smaller; is much less likely to be organized as a corporation; and is less likely to run by an experienced management team. We also find that firms located in countries with better governance are less likely to report that they are Discouraged, both in developed and in developing countries. In our third step, we analyze firms that applied for credit and either were turned down (Denied firms) or were extended credit (Approved firms). Among firms that apply for credit, we find that a Denied firm is both younger and smaller; is less likely to be organized as a corporation; and is less likely to be run by more experienced management team. We also find that firms located in countries with better governance are less likely to report that they were Denied credit, both in developed and in developing countries. From this evidence, we conclude that country-level governance plays a critical role in the availability of credit to SMEs in both developed and developing countries. Key Words: availability of credit; denied credit; discrimination; discouraged firm; entrepreneurship; governance; small business; WBES JEL Classification: G21, G32, J71, L11, M13

3 SME Credit Availability Around the World: Evidence from the World Bank Enterprise Surveys 1 Introduction Among small and medium enterprises ( SMEs ) around the world, which ones need credit and which ones get credit? The answer to this query is of great interest not only to SMEs, but also to SME creditors, customers and suppliers, as well as to policymakers and to the economies in which these firms operate. Also important is how country-level governance affects the availability of credit to SMEs, as policy-makers can implement changes that improve governance, thereby affecting the availability of credit. In this study, we analyze data from a series of surveys conducted by the World Bank in 133 countries during to provide new evidence on how to answer these questions. The availability of credit is one of the most fundamental issues facing a small business; therefore, it has received much attention in the academic literature (see, e.g., early work by Petersen and Rajan, 1994; Berger and Udell, 1995; and Cole, 1998). Since the Great Financial Crisis began in 2008, the issue of credit constraints faced by SMEs has been a major concern of policymakers seeking to increase employment and improve economic growth. However, many small firms indicate that they do not need credit ( no-need firms) while others indicate that they needed credit but did not apply for credit ( discouraged firms). With a few notable exceptions, starting with Cole (2009), Brown et al. (2011), and most recently, Cole and Sokolyk (2016), the existing literature largely has ignored no-need firms. Discouraged firms those that do not apply for credit because they expect to be turned down have received somewhat more attention in the literature (see, e.g., Han et al. 2009) than no-need firms, but not nearly as much as firms that actually apply for credit. Many of the studies that have analyzed discouraged firms (see, e.g., Gropp et al 1997; Berkowitz and White, 2004; and Berger et al., 2011) pool them with firms that actually - 1 -

4 applied for, but were denied credit, making it difficult to draw clear inferences from these studies. In this study, we analyze these four groups of firms to shed new light upon their similarities and differences around the world. We utilize data from the World Bank s Enterprise Surveys ( WBES ) over the period from to estimate a sequential set of three logistic regression models. First, a firm first decides if it needs credit ( no-need firms versus need firms). Second, a need firm decides if it will apply for credit ( discouraged firms versus applied firms ). Finally, an applied firm applies for credit and then learns from its prospective lender whether it is successful in obtaining credit ( approved firms versus denied firms). Beginning in 2006, the World Bank implemented a Global Methodology for its SME surveys, which was designed to ensure a consistent definition of the population, a consistent methodology of implementation and a common core questionnaire. This methodology enables researchers to compare the results of surveys across countries and years. 1 Hence, the results of our study provide politicians, policymakers, and regulators with new insights on how to tailor macroeconomic policy and regulations to help small businesses obtain credit when they need credit. An important caveat is that our results are descriptive based upon cross-sectional surveys so our findings should not be interpreted as providing the basis for causal inference. Our main findings can be summarized as follows. Overall, we find that a no-need firm is more likely to be owned by a foreigner; and is more likely to be classified as a manufacturing or chemicals company, and that firms located in developing, but not developed, countries with better governance are less likely to need credit. 1 In addition, we use a set of year dummies and a set of macro-economic variables to control for differences across time and economic development

5 Among firms that need credit, we find that a discouraged firm is both younger and smaller; is much less likely to be organized as a corporation; and is less likely to run by an experienced management team. We also find that firms located in countries with better governance are less likely to report that they are discouraged, both in developed and in developing countries. Among firms that apply for credit, we find that a denied firm is both younger and smaller; is less likely to be organized as a corporation; and is less likely to be run by more experienced management team. We also find that firms located in countries with better governance are less likely to report that they were denied credit, both in developed and in developing countries. Why are these issues of importance? Small businesses are critical to economic growth and employment. In the U.S., for example, the government reports that small firms account for over half of all private-sector employment and produce almost two-thirds of net job growth. 2 Outside the U.S., Ayyagari et al. (2011) analyze data from 99 countries surveyed during and find that small firms with less than 250 employees account for about almost two-thirds of all employment and almost 90% of job creation. The importance of small firms in less developed countries where publicly traded firms are less prominent is all but certain to be even larger. Therefore, a better understanding of who needs credit and who gets credit can help policymakers to take actions that will lead to more jobs and faster economic growth. We contribute to the literature in at least four important ways. First, we provide the first rigorous analysis of the differences in our four types of firms non-borrowers, discouraged borrowers, denied borrowers and approved borrowers for a large international sample of countries around the world. So far, researchers only have analyzed SMEs in the 2 See, Frequently Asked Questions, Office of Advocacy, U.S. Small Business Administration (2010). For research purposes, the SBA and Federal Reserve Board define small businesses as independent firms with fewer than 500 employees

6 U.S. (Cole, 2009) and Europe (Brown et al., 2011). Moreover, our analysis spans the period before, during and after the GFC, so we can shed new light on how the crisis impacted the availability of SME credit. Second, we provide an analysis of credit availability that properly accounts for the inherent self-selection mechanisms involved in the credit application process: who needs credit, who applies for credit conditional upon needing credit, and who gets credit, conditional upon applying for credit. Most previous researchers except for Cole (2009) and Brown et al. (2011) have ignored firms that do not need credit, and many have pooled discouraged and denied firms. We find significant differences across these groups around the world. Hence, our results shed new light upon the credit-allocation process. Third, we provide evidence from the WBES on the availability of credit to small businesses using the World Bank s global methodology. This contributes to the growing literature on SME finance that has emerged from the cross-country World Bank surveys of SMEs, including Beck et al. (2005, 2006, 2008); Chakravarty and Xiang (2009); De la Torre et al. (2010 ); Ayyagari et al. (2011); Brown et al. (2011); Hanedar et al. (2014) and Love and Peria (2015). Our study, however, focuses only on countries surveyed using the global methodology. Fourth, we contribute to the literature on how governance affects the availability of credit to SMEs (Beck et al., 2005, 2006, 2008; Love and Peria, 2015). We find that firms located in countries with better governance are less likely to need credit; are less likely to be discouraged from applying for credit when they need credit; and are less likely to be denied credit when they apply for credit. In section 2, we briefly review the literature on the availability of credit, followed by a description of our data in section 3 and methodology and our variables in section 4. Our results appear in section 5 and we provide a summary and conclusions in section

7 2 Related Literature The literature on availability of credit to SMEs dates back at least to Wendt (1947), but really came into prominence following the release of a series of nationally representative surveys of SMEs in the U.S. conducted by the Federal Reserve Board beginning in the late 1980s. More recently, international interest in this area has grown following the release of a series of international surveys of SMEs conducted by the World Bank. In general, these studies focus on how the opacity of SMEs affects their ability to obtain new financing when they need credit to fund their existing operations or new projects. 2.1 Studies using the Federal Reserve s Survey of Small Business Finances In a seminal article, Petersen and Rajan (1994) analyzes data on loan rates from the 1987 iteration of the Survey of Small Business Finances (SSBF) for evidence on how relationships influence the availability of credit to small firms. The authors find that the length of a relationship between a borrower and her bank decreases the rate she is charged. Berger and Udell (1995) also analyzes data on loan rates from the 1987 SSBF, but focused on lines of credit at small business in order to provide more compelling evidence on the importance of relationships. The authors also find that the length of a relationship between a borrower and her bank lowers the spread charged by the bank on her credit line. Cole (1998) is the first study to analyze data from the 1993 SSBF and is the first to focus on determinants of the loan turndown decision rather than of loan rates. Cole finds that a pre-existing relationship between a prospective borrower and her bank increases the likelihood that her bank approves the loan application, but also finds that the length of that relationship is not important. Chakraborty and Hu (2006) also uses data from the 1993 SSBF to analyze how relationships affect a lender s decision to secure lines of credit and other types of loans with collateral. These authors find that the length of relationship decreases the likelihood of - 5 -

8 collateral for a line of credit, but not for other types of loans. Previously, Berger and Udell (1995) had shown that longer relationships reduced the likelihood of collateral being required for lines of credit, using data from the 1987 SSBF. Cole (2009) is the first study to focus on no-need firms, and to separate firms into the four categories that we also use. He analyzes data from the 1993, 1998 and 2003 iteration of the SSBFs and finds that no-need firms look very much like approved firms and that discouraged firms differ from denied firms in a number of significant ways. Han et al. (2009) analyzes discouraged SMEs in the U.S., using data only from the 1998 SSBF. These authors find that both the demographics of the entrepreneur, such as age and personal wealth, and of the business, such as size and use of financial products, influence discouragement. They also find that riskier borrowers are more likely to be discouraged, which they interpret as an efficient self-rationing mechanism. Chakravarty and Yilmazer (2009) uses data from the SSBFs to provide evidence on the availability of credit to discouraged, denied and approved firms, but ignore noneed firms that constitute more than half of all SSBF firms. These authors find that various measures of the strength of the relationship between a firm and its prospective borrower are associated with a higher likelihood of applying for credit and, conditional upon applying a higher likelihood of obtaining credit. Cole and Sokolyk (2016) extend Cole (2009) by focusing on no-need and discouraged firms. They find that about half of all small firms report no need for credit even in times of recession, and, in a counterfactual exercise, that one in three discouraged borrowers would have received credit had they applied for credit

9 2.2 Studies using the World Bank s SME Surveys The World Bank has catalogued more than 350 studies that use data from its SME surveys; 3 consequently, we will review only some of the most prominent of published studies. Beck et al. (2005) uses data from World Bank surveys of more than 4,000 SMEs in 54 countries to analyze whether financial, legal and corruption obstacles affect firm growth rates. These authors find that it is growth of the smallest of firms that are consistently most affected by all three types of obstacles. Beck et al. (2006) uses data on more than 10,000 firms in 80 countries to examine financial obstacles faced by SMEs. This study finds that older, larger, and foreign-owned firms report fewer financing obstacles; and that institutional development is the most important factor in explaining cross-country differences in financing obstacles. Beck et al. (2008) uses data on more than 3,000 firms from 48 countries to analyze how financial and institutional development affects firm-level financing at SMEs. This study finds that, in countries with poor institutions, firms use less finance, especially from banks; and that small firms, in general, use less bank finance. De la Torre et al. (2010) uses data from World Bank s SME surveys to provide evidence on bank involvement in SME finance. This study finds that all sizes of banks cater to SMEs, and that large banks have comparative advantages in offering many products. In the study closest to ours, Brown et al. (2011) follows the general methodology of Cole (2009). Their firm-level data come from the 2004/2005 and 2008 waves of the Business Environment and Enterprise Performance Survey (BEEPS). These authors look at 20 countries in Eastern and Western Europe prior to, whereas we look at 133 countries around the world after, implementation of the Global Methodology that ensures consistency across surveys. They find that small and financially opaque firms are less likely to apply for credit, 3 See

10 while firms with more financing needs are more likely to apply for credit. Most interestingly, they also find that firms applying for credit rarely are denied credit. Chakravarty and Xiang (2013) use data on more than 8,000 firms from ten countries to analyze discouraged firms. Their firm-level data come from the 2002/2003 waves of the Business Environment and Enterprise Performance Survey (BEEPS). These authors find that discouraged firms differ across developed and developing countries; and that larger firms, more transparent firms, and firms with stronger banking relationships are less likely to be discouraged. Hanedar et al. (2014) use data from the World Bank Business Environment and Enterprise Performance Survey (BEEPS) to investigate what determines collateral requirements for SMEs in less-developed countries. Not surprisingly, they find that firm-level variables, especially measures of risk, are more important than country-level variables in determining collateral requirements. La Porta and Shleifer (2014) use data from the WBES and other sources to establish a set of stylized facts about the informal economies in developing countries. They find that the informal economies: are large, accounting for as much as half of the total economy in the poorest countries; have low productivity as compared to the formal economies; are largely disconnected from formal economies; and shrink as countries grow and develop. Love and Peria (2015) uses data from the WBES to analyze how bank competition affects the access of SMEs to finance. They find that low levels of competition reduce such access, and that the impact depends upon the existence and coverage of mechanisms for sharing credit information, such as public credit registries. 3 Data To conduct this study, we use data from the World Bank s Enterprise Surveys. The World Bank conducted these surveys in 135 countries between 2006 and 2014 and provides - 8 -

11 117,358 firm-year observations. In order to focus on SMEs and to be consistent with the studies that analyze the SSBFs, we only use data from firms with less than 500 employees according to the definition of SMEs in the U.S. We also delete data from two countries for which we cannot obtain governance data. This reduces our final sample to 106,611 firm-year observations from 133 countries over the period. Appendix Table 1 identifies the number of observations in our sample by country and shows the year of the WBES for each country. The WBESs collect information about the business environment, how it is perceived by individual firms, how it changes over time, and about the various constraints to firm performance and growth. In each country surveyed, the WBESs collect firm-level data on a representative sample in the non-agricultural, formal private economy in the manufacturing, services, transportation, and construction sectors; the surveys explicitly exclude firms in the public-utilities, government-services, health-care, and the financial-services sectors. Besides the consistent definition of the population, the methodology of implementation and core questionnaire have served as foundation for the so-called Global Methodology under which the various surveys have been conducted. Because the standardized approach of the Global Methodology was implemented beginning in 2006, it is possible to compare the surveys across countries and years. The data collected through the surveys include quantitative as well as qualitative information. World Bank representatives conducted face-to-face interviews with company owners and managers in order to gather information on their firms and the business environment in each country. The surveys address a broad range of topics, such as general information on the company, infrastructure, services, crime, finance, and labor. One key limitation of the WBESs is the fact that most of the data gathered in the survey are based on subjective perceptions of the owners and managers of the firms, with the exception of some company figures. Another is the fact that the WBESs do not provide some - 9 -

12 key information about firms that typically are required by banks when a company applies for a loan performance indicators, such as the profitability, debt-equity-ratio, margins, etc. are not included in the data. Even so, the WBESs provide detailed information about each firm's most recent borrowing experience. This includes whether or not the firm applied for credit and, if the firm did not apply, did it fail to apply because it feared its application would be rejected (discouraged borrowers). We also use data from the World Bank s Worldwide Governance Indicators, which provides annual country-level indicators of governance for There are six governance indicators: (1) voice and accountability, (2) political stability and absence of violence/terrorism, (3) government effectiveness, (4) regulatory quality, (5) rule of law, and (6) control of corruption. These six aggregate indicators are based on World Bank staff s analysis of hundreds of underlying variables that come from a numerous sources. Each indicator is expressed in standard normal units ranging from -2.5 to These six indicators also are highly correlated within each country, with correlation coefficients greater than 0.6. Consequently, we calculate an annual country-level average of the six, which we label GovIndex. Finally, we would like to control for country-level differences in the supply of credit. To accomplish this, we gather data from the World Bank s World Development Indicators database on bank credit to the private sector as a percentage of GDP, which we label PrivCredit. 5 4 Methodology and Model Specification 4.1 Model Specification In order to analyze characteristics of firms that need credit, apply for credit and get credit, we follow the methodology of Cole (2009) and Cole and Sokolyk (2016). First, we 4 These data are available at 5 These data are available at

13 classify firms into one of four categories based upon their responses to questions regarding their most recent loan request during the previous years: (1) No-Need Firms: the firm did not apply for credit during the previous three years because the firm did not need credit. (2) Discouraged Firms: the firm did not apply for credit during the previous year because the firm feared rejection, even though it needed credit. (3) Denied Firms: the firm did apply for credit during the previous three years but was denied credit by its prospective lender(s). (4) Approved Firms: the firm did apply for credit and was approved for credit by its prospective lender. Once we have classified our sample firms, we calculate descriptive statistics for each group of firms and test for significant differences across categories, e.g. differences between no-need firms and need-credit firms. We also conduct multivariate tests on the data, estimating logistic regression models that explain the sequential selection of the loan application and approval process. First, a firm decides whether it needs credit. We include firms from all four groups in this analysis, and define Need Credit as equal to zero for noneed firms and a value of one to all other firms ( discouraged, denied, and approved firms). Second, a firm that needs credit decides whether to apply for credit. We exclude noneed firms from this model and define Apply for Credit as equal to zero for discouraged firms and equal to one for firms in one of the two groups that applied for credit ( denied and approved firms). Third, a firm that decides to apply for credit is either approved or denied credit by its prospective lender. In this stage of the model, we include only those firms that applied for credit and define Get Credit as equal to zero for denied firms and equal to one for approved firms

14 We estimate this three-step sequential model using a set of three weighted logistic regression models. We use logistic regression because our dependent variables are binary (i.e., they take on a value of zero or one), so that ordinary least squares is inappropriate. We use the sampling weights provided by the World Bank because the samples are stratified random samples rather than simple random samples, so that certain types of firms were oversampled and have different selection probabilities. We cluster robust standard errors at the country level. In addition to analyzing our full sample, we also analyze separately firms from developing and developed countries. We categorize countries in developing and developed countries based upon the World Bank income classification. Developing countries include the countries classified as low-income and lower middle-income, while developed countries include those in the categories of upper middle-income and high-income. In line with the results of Beck et al. (2008), we expect firms in developed countries to face fewer difficulties in obtaining a credit, as the financial sector in these countries is usually further developed. 4.2 Dependent variables In this section, we explain in detail our classification criteria for each borrower type with reference to specific WBES questions. No-Need: We classify a firm as No-Need if (i) it reported that it did not apply for credit during the last complete fiscal year and (ii) answered No need for a loan - establishment has sufficient capital when asked about the main reason of absence of a credit application on question K.17 of the WBES ( ) questionnaire. All other firms are classified as Need-Credit firms. Discouraged: We classify a firm as Discouraged if (i) it is first classified above as Need-Credit firm and (ii) it reported that it did not apply for credit during the last complete fiscal year and (ii) it answered with

15 Application procedures for loans or line of credit are complex, Interest rates are not favorable, Collateral requirements for loans or line of credit are unattainable, Size of loan and maturity are insufficient, Did not think it would be approved, or Other, when asked about the main reason of absence of a credit application on question K.17 of the WBES ( ) questionnaire. We classify all other Need-Credit firms as Applied firms, as they did apply for credit. Denied: We classify a firm as Denied if (i) we classify it above as an Applied firm because it reporting that it applied for credit during the last complete fiscal year and (ii) it did not report having credit at the time of the interview. These firms are identified using questions K.16 (application of credit during last complete fiscal year), K.8 (existence of credit at this time), and K.10 (year of approval of existing credit) of the WBESs ( ). Approved: We classify a firm as Approved if (i) we classify it above as an Applied firm because it reported that it applied for a loan during the last fiscal year, (ii) it reported that it had credit at the time of the interview, and (iii) that it reported that a credit was granted to the firm during the last complete fiscal year or in the current year by its prospective lender(s). 4.3 Independent variables For explanatory variables, we generally follow the existing literature on the availability of credit, which hypothesizes that a lender is more likely to extend credit to a firm when that firm shares characteristics of other firms that historically have been most likely to repay their credits. We expect that the same set of characteristics should explain no-need firms relative to need firms; applied firms relative to discouraged firms; as well as approved

16 firms relative to denied firms. For No-Need firms, we assume that they do not apply for additional credit because they are operating at their optimal capital structure. Consequently, we look to the pecking-order theory (Myers and Majluf, 1984; Myers, 1984) and trade-off theory (Kraus and Litzenberger, 1973) of capital structure to explain differences in No-Need and Need-Credit firms Firm Characteristics Firm characteristics include public reputation as proxied by firm age; firm size as measured by the total number of employees; the legal form; and firm industrial classification as measured by a set of dummy variables. We expect that the age of a firm, measured by the number of years since the firm started its operations, is a positive influence on the availability of credit. Older firms are thought to be more creditworthy because they have survived the high-risk start-up period in a firm s life cycle and, over time, have developed a public track record that can be scrutinized by a prospective lender. Hence, firm age can be used as a proxy for the firm s reputation. Beck et al. (2006) argue that older firms report fewer financing issues. Larger firms, as measured by number of total employment, are thought to be more creditworthy because they tend to be better established and typically are more diversified than are smaller firms. Beck et al. (2006) and Aterido et al. (2007) find that micro and small firms face more obstacles in accessing finance than do large firms. We use the logarithm of the number of employees. We also use information on legal form of organization to classify firms as corporations, proprietorships and partnerships. We classify firms by industry using a set of dummy variables: one each for construction, 6 food, textiles, 7 manufacturing, 8 retail_wholesale trading companies, chemicals, 6 Construction includes firms classified in the construction and transportation industries

17 metals, transport, services and other services. Firms in chemicals, construction, manufacturing, metals and transport industries are thought to be more creditworthy because they typically have more tangible assets that can be pledged as collateral than do firms in more service-oriented industries Owner Characteristics Our vector of owner characteristics includes variables such as the experience of the top manager in this sector, the gender of the owner as measured by dummy variables for female- or male-owned firms; and dummy variables for domestic-owned or foreign-owned firms. We include a variable measuring the experience of the top manager in this sector in years. The more experienced is a top manager, the better is her track record and, thus, the better is her creditworthiness expected to be. The gender of the firm owner variable takes into account whether there are any females amongst the owners of the firm. We have no expectations regarding indicators for firms with female owners, even though Muravyev et al. (2009) argue that the probability that female-managed firms obtain credit is lower than with male-managed firms. We include this variable in an effort to ascertain whether minority-owned firms are experiencing disparate outcomes in the credit markets relative to firms whose controlling owners are males. Furthermore, we include a dummy variable regarding foreign and domestic ownership. The company is foreign-owned if private foreign individuals, companies or organizations own 50% or more of the firm. The results of this variable might heavily depend on the country of the owner and the country in which the firm is acting. Generally, we expect 7 Textiles includes firms classified in the textiles, leather and garments industries. 8 Manufacturing includes firms classified in the metals & machinery, electronics, chemicals & pharmaceuticals, wood & furniture, non-metallic & plastic materials, auto & auto components, and other manufacturing industries

18 that a lender perceives a domestic owner to be more creditworthy because it should be easier for a lender to gather information on domestic than on foreign owners; in addition, it should be easier to collect unsecured debt from a domestic owner Market/Environmental Characteristics We include two country-level: govindex and bankcred, which we defined earlier. 9 We expect that firms are more likely to be discouraged when governance, such as the level of corruption, is worse, and we expect that firms are more likely to be denied when governance, such as the rule of law, is worse. We have no a priori expectations about how governance would affect the need for credit. We expect that credit supply as measured by bank credit to the private sector as a percentage of GDP should positively affect the need for credit, and reduce both discouragement and denials of credit. Table 1 gives a brief overview and description of the variables used in our analysis. 5 Results 5.1 Descriptive Statistics and Univariate Results Table 2 presents descriptive statistics for the full sample, and, separately, for firms that need credit and for firms that have no need for credit, along with t-tests for differences in means of these two groups. We first describe the full-sample means and then discuss the differences in the means of the t-tests Full Sample The averages for our full sample appear in column 2 of Table 2. The average firm in our sample has been in business for 16.6 years. By employment size, the average firm in our sample (median) has fewer than 20 employees. By legal form of organization, 47 percent of 9 We do not include GDP per capita (or other country-level governance variables) in our regressions because these variables are highly correlated. For example, in our sample of countries, the governance index and GDP per capita have a highly significant correlation of

19 the firms organize as corporations while 54 percent organize as proprietorships (35%), partnerships (16%), or other (3%) 10. By industry, 24 percent of the firms are each in manufacturing and services, 4 percent are in chemicals, 8 percent are each in food, textiles and metals, 18 percent are in retail and wholesale trade, 1 percent in each construction and transport and 4 percent in other services. Regarding the owner characteristics, the top manager has, on average, 17.6 years of experience in this sector. By ownership, 92 percent of the firms are domestic, while the remaining 8 percent are foreign-owned. Foreign ownership refers to the nationality of the shareholders. If the primary owner is a foreign national resident in the country, it is still a foreign owned firm. At least one of the owners is a female at 29 percent of the firms. The average Governance Index of the 133 countries in the sample is , while the average Bank credit to GDP ratio is 38.9 percent. Looking at the distribution by survey year, 13 percent of the firm-year observations come from 2006, 8 percent from 2007, 3 percent from 2008, 16 percent in 2009, 13 percent from 2010, 3 percent from 2011, 6 percent from 2012, 23 percent from 2013 and 15 percent from Because of the standardized approach of the Global methodology implemented in 2006, it is possible to compare the various indicator sets across various countries and different years. 10 The missing 3% answered don t know and other

20 5.1.2 No-Need Firms versus Need Credit Firms In columns 3 and 4 of Table 2 are the averages for our No-Need and Need subsamples, respectively; the difference in mean appears in column 5, followed in column 6 with a t-statistic for a significance test on the difference in means shown in column 5. Most of the firm characteristics are significantly different for the subsamples of firms that need credit ( discouraged, denied, and approved ) and firms that have no need for credit. Overall, 56 percent of all companies needed credit while 44 percent did not. This result is very similar from the U.S., where Cole (2009) reports that 55 percent of all companies needed credit. When compared to a firm with no need for credit, a firm needing credit is less likely to be organized as a corporation (46% vs. 47%) and a partnership (15% vs. 16%), and more likely to organized as a proprietorship (34% vs. 36%). By industry, it is less likely to be in the construction (10% vs. 11%), retail/wholesale trade (17% vs. 19%), services (23% vs. 26%) and transport (1% vs. 2%) sectors, and more likely to be in the manufacturing (25% vs. 23%), chemicals (5% vs. 4%), food (8% vs. 7%), metals (8.5% vs. 8%), and textiles sectors (8% vs. 6%). By ownership characteristics, a firm in need of credit is more likely to be domesticowned (94% vs. 91%) and female-owned (30% vs. 27%), and has a less experienced management (17.3 years vs years). By market characteristics, a firm in need of credit is more likely to be located in a country with a lower Governance Index (-0.41 vs ) and a lower ratio of Bank Credit to GDP (37.8% vs. 40.3%) Discouraged versus Applied Firms Table 3 presents averages for all firms that need credit, and then, separately, for firms that applied for a credit and for discouraged firms, i.e., for firms that needed credit but did not

21 apply because they feared rejection, along with a t-test for differences in means of these two groups. Overall, 53% of the 62,928 firms that needed credit were discouraged. This is significantly more than in the U.S. (28%; Cole, 2009), Western Europe (17%; Brown et al., 2011) and also Eastern Europe (40%; Brown et al., 2011). Relative to a discouraged firm, an applied firm has significantly more employees, is older (17.5 vs years), and more likely to be organized as a corporation (62% vs. 32%). By industry, an applied firm is more likely to be in the chemicals (5% vs. 4%), food (9% vs. 8%), and services (24% vs. 23%) sectors, and less likely to be in the manufacturing (24% vs. 25%), construction (0.6% vs. 1.3%), metals (7% vs. 9%), Retail_Wholesale (16% vs. 17%), textiles (8% vs 9%) and transport (1% vs. 2%) sectors. By owner characteristics, an applied firm has more experienced management (19 vs years), is more likely to be foreign-owned (7.2% vs. 5.7%), and is more likely to have a female owner (35% vs. 26%). By market characteristic, an applied firm is significantly more likely to be located in a country with a higher Governance Index value (-0.26 vs ) and higher bank credit to GDP ratio (42% vs. 35%) Approved Firms versus Denied Firms Table 4 presents descriptive statistics for firms that applied for a credit, and then, separately, for firms that were turned down ( denied firms), and for firms that received credit ( approved firms), along with the difference in these two means and a t-test for significant differences in means. Overall, 30% of the firms applying for credit were turned down by their prospective lenders. By comparison, Cole (2009) reports that only 11% - 22% of the U.S. firms applying for credit were turned down. Hence, credit around the world appears to be tighter than in the U.S

22 When compared with a denied firm, we find that an approved firm is significantly older (18.1 vs years) and larger. Furthermore, approved firms are more likely to be organized as corporations (66% vs. 51%) and less likely to be organized as Partner (11.4% vs. 14.1%) and as Proprietorship (19% vs. 25%). By industry, firms that are granted a credit are less likely to be in the manufacturing (23.5% vs. 25%), construction (0.5% vs. 0.9%), services (23% vs. 26%), transports (0.7% vs. 0.9%) and retail/wholesale trade (16% vs. 17%) sectors, and more likely to in the chemicals (6% vs. 4%), food (9% vs. 8%), textiles (8% vs. 7%) and metals (8% vs. 7%) sectors. By owner characteristics, an approved firm has significantly more managerial experience (19.7 vs years), is significantly more likely to be domestic-owned (9.3% vs. 9.2%) and is significantly more likely to have a female owner (36.8% vs. 31.0%). By market characteristics, an approved firm is significantly more likely to be located in a country with a higher Governance Index (-0.19 vs ), and a higher bank credit to GDP ratio (43.6% vs. 36.6%). 5.2 Multivariate Results Need-Credit Firms vs. Need-No-Credit Firms In Table 5, we present our multivariate results from estimating the logistic regression models explaining who needs credit, as described in section 4.2. The dependent variable No- Need is equal to one if the firm indicated that it did not need credit and equal to zero for firms indicating a need for credit (including discouraged, denied, and approved firms). In column 2, we present results for the full sample, while in columns 3 and 4, we present results for our developing and developed subsamples, respectively. Rather than present logit coefficients, which are uninformative other than sign, we exponentiate these coefficients to obtain the odds ratios associated with each variable. This enables us to talk about the magnitude of effects. An odds ratio greater than 1.00 indicates that a firm with that

23 characteristic is more likely to need no credit, whereas an odds ratio less than 1.00 indicates that a firm with that characteristic is less likely to need no credit. Alternatively, one divided by the odds ratio for being a no-need firms gives the odds ratio of being a firm that needs credit. In Column 2, we show the results for firms that need credit versus firms that do not need credit. Among the firm characteristics, only industry indicators for manufacturing and chemical firms are statistically significant. Firms in these two industries are about 30% more likely to need credit than are firms in other industries. Among the owner characteristics, foreign-owned firms are about 35% less like to need credit than are domestic-owned firms. Neither of our market characteristics is significantly different from zero at standard levels of significance but the odds ratio of GovIndex is 1.21, indicating that firms located in countries with better governance are less likely to need credit (p-value=0.12). Among our year indicators, the odds ratios for are in the range of indicating that firms were about 50% more likely to need credit during these crisis years, but only the coefficients for 2008 and 2010 are statistically significant at the 10% level or better. Column 3 presents the results for our developing country subsample. Among the firm characteristics, the coefficient on the indicator for corporations is significant and its associated odds ratio indicates that corporations are about 30% more likely to need credit than firms with other legal forms of organization. This is consistent with the limited liability provided by the corporate form of ownership. Six of the industry indicators are statistically significant. Firms classified as manufacturing, chemicals, construction, metals, services and textiles are more likely to need credit. For all but services, this result is consistent with the greater amount of tangible assets in these industries; for services, this result is counterintuitive and suggests that factors other than tangibility of assets are driving the need for credit at service firms. Among the owner characteristics, foreign-owned firms are much less likely to need credit than are domestic-owned firms and female-owned firms are more likely

24 to need credit than are male-owned firms. Both of our market characteristics is significantly different from zero at the 10% level of significance. The odds ratio of GovIndex is 1.36, indicating that firms located in developing countries with better governance are less likely to need credit (p-value=0.04). The odds ratio of BankCred indicates that firms located in countries where banks provide more credit to the private sector are less likely to report a need for credit, most probably because their credit needs already have been met. Among our year indicators, the coefficients for 2008, 2013 and 2014 are highly significant and the associated odds ratios are in the range of , indicating that firms were much less likely to need credit during these years. This finding suggests that the financial crisis had a very different impact on firms in developing countries as compared with firms in developed countries. Column 4 presents the results for our developed country subsample. Among the firm characteristics, the coefficient on our firm size proxy is significant at the 10% level and its associated odds ratio indicates that larger firms are significantly more likely to need credit than are smaller firms. This finding is consistent with what Cole (2009) reports for SMEs in the U.S. Only two of the industry indicators are statistically significant the same as in the full sample manufacturing and chemicals. Among the owner characteristics, foreign-owned firms are much less like to need credit than are domestic-owned firms and female-owned firms are more likely to need credit than are male-owned firms. Neither of our market characteristics is significantly different from zero at standard levels of significance but the odds ratio of GovIndex is 1.15, indicating that firms located in countries with better governance are less likely to need credit. Among our year indicators, the odds ratios for are in the range of indicating that firms were about 50% more likely to need credit during these crisis years, and the coefficients for these year indicators are statistically significant at the 10% level or better. In combination with the results for year dummies from the developing country subsample, this shows that the full

25 sample results for the year dummies are driven almost entirely by the developed country subsample Discouraged Firms vs. Applied Firms In Table 6, we present our multivariate results from estimating the logistic regression models explaining which firms reported being discouraged from applying for credit. The dependent variable Discour is equal to one if the firm indicated that it needed credit but was discouraged from applying and equal to zero if the firm indicated that it needed credit and that it applied for credit (including denied and approved firms). In column 2, we present results for the full sample, while in columns 3 and 4, we present results for our developing and developed subsamples, respectively. Among the firm characteristics, the coefficient on our firm size proxy is significant at the 1% level. Its odds ratio of 0.55 indicates that larger firms are significantly less likely to be discouraged from applying for a credit than smaller firms. Older firms are significantly more likely to be discouraged to apply for a credit. This is in contrast to the opposite relation reported by Cole (2009) for U.S. firms. The coefficient of our indicator for corporate legal form of ownership is statistically significant and its associated odds ratio indicates that corporations are about 40% less likely to be discouraged from applying for credit. Four of our industry indicators are statistically significant. Firms classified as manufacturing, construction, food and textiles are more likely to be discouraged from apply for credit. Among our owner characteristics, firms with more experienced management are less likely to be discouraged borrowers than firms with a management with less experience. Among our market characteristics, GovIndex is statistically significant at better than the 0.1% level. The associated odds ratio of 0.44 indicates that firms located in countries with a better governance are much less likely to be discouraged from applying for a credit. A one

26 standard deviation improvement in governance reduces the odds of being discouraged by about 55 percent. Among our year indicators, the coefficients for 2007 and 2014 are highly significant, and the odds ratios indicate that firms were much more likely to be discouraged from applying for credit during these years. Columns 3 and 4 of Table 6 present results for our regression model when we split our sample into developed and developing countries, respectively. As in our full model, we find a positive relation between firm size and applying for a credit in both subsamples, with the odds ratio indicating a stronger relation in developed (0.54) than developing (0.63) countries. Furthermore, the coefficient of the legal form of corporations is statistically highly significant in the full model and in the separate estimations for developing and developed countries with similar odds ratio between 0.56 and We find some differences between our subsample of developing and developed countries for the industry indicators. In developing countries, firms classified as chemicals and transport are less likely to be discouraged while firms classified as construction are more likely to be discouraged to apply for a credit. In developed countries, industry indicators for manufacturing, construction, food and textiles are statistically significant. Firms in these four industries are more likely to be discouraged from applying for credit than are firms in other industries. Among the ownership characteristics, firms with a more experienced management are less likely to be discouraged firms in developed countries. In developing countries, ownership does not have an influence on whether a firm applies for a credit or if it was discouraged and did not apply. By market characteristics, we find a negative relation between the Governance index variable and the discouraged firms in both developed and developing countries. The odds ratios of 0.56 and 0.40 indicate that firms located in countries with better governance are