How do firms choose between intermediary and supplier finance?This paper examines firms' short-term financing choices between intermediated loans and trade credit. I test two sets of empirical hypotheses: 1) hypotheses concerning the cross-sectional differences in the level of intermediary finance for firms that use different levels of trade credit and 2) hypotheses concerning the dynamics of trade credit growth. I find strong evidence that for firms with high agency costs, the use of trade credit facilitates access to conventional bank loans. The evidence is consistent with theories based on the signaling role of trade credit provision and suppliers' liquidation advantage.********** Suppliers regularly provide trade credit to their customers by selling them goods with a delayed payment. To their customers, trade credit is often an important source of external short-term financing. Official statistics indicate that at the end of 2005, US corporations had $6,029 billion outstanding in accounts payable, roughly half of their short- and long-term debt. (1) That same year, trade credit accounted for 43% of the total debt of UK corporations and for 44% of the total debt of Japanese corporations. (2,3) The economic significance of trade credit as a source of funding highlights the importance of questions regarding its use. In this paper, I examine the firm's choice between supplier financing and borrowing from a financial intermediary. Credit rationing models argue that when borrowing constraints from financial intermediaries are binding, firms increase their use of trade credit as a substitute source of financing. In contrast, my results confirm that trade credit is not a substitute for bank debt for firms with high agency costs. I find that trade credit and bank debt are positively correlated, and, as such, can be complementary sources of financing. This evidence is consistent with the signaling explanation of trade credit use, where the availability of supplier finance facilitates firms' access to bank debt and improves its terms. Previous studies have argued that there is a cost disadvantage to using supplier financing. Pricing trade credit is complicated as suppliers do not charge an explicit interest rate, so academics have relied on calculating an interest rate based on the hypothesized cash-discount terms of a two-part contract. (4) From the customer's standpoint, taking the cash discount results in zero cost borrowing for a period of time and reduces the cost of goods or services purchased. The cost of trade credit becomes high only when the firm foregoes the cash discount. Ng, Smith, and Smith (1999) demonstrate that the actual cost cannot be easily calculated as it depends upon how common the two-part contract is. In their sample, only a quarter of the firms offered a two-part contract. The size of late payment penalties and how widespread these are also affect the interpretation of the results. (5) In this paper, I examine the role of supplier finance in the firm's financing cycle without digression into the separate issue of pricing trade debt. I summarize the theoretical developments that have analyzed the interaction between trade credit and financial intermediary debt and formulate a number of hypotheses that I test empirically. My results are consistent with the signaling models of trade credit use. These models argue that in cases of default, suppliers have a liquidation advantage where they are able to efficiently redeploy repossessed inputs. Financial intermediaries are inefficient at redeploying assets, but have a lower cost of capital because they can raise funds in the deposit market. Antov and Atanasova (2011) argue that if the use of trade credit is perceived by financial intermediaries as a favorable signal regarding borrowers' creditworthiness, then firms that face high agency costs will use supplier debt to finance operations. Informational asymmetries and agency costs can impede the flow of funds to firms that need external financing. If entrepreneurs cannot credibly convey to outside financiers the creditworthiness of their projects, trade credit can be used as a means of acquiring reputation. In practice, the credit quality of many small businesses may be difficult to assess and monitor as most of the information is privately held. Suppliers can resolve, to some extent, these informational asymmetries. A firm's creditworthiness can be signaled to lenders and financial markets through the public disclosure of credit agreements, trade credit renewals, and an existing supplier-customer relationship. Adams, Wyatt, and Kim (1990) cite the Campeau Corporation bankruptcy case as a compelling example in which trade creditors were the first to detect financial distress and alert other lenders of this situation through their reluctance to provide seasonal inventories on credit. The main contribution of the paper is to test empirical hypotheses related to the interaction between trade credit and institutional financing using a large panel data set for the period 1998-2006. Both public and private companies are included in the sample and there is comprehensive coverage across industries. I find that for firms with high agency costs, increasing trade credit is associated with more institutional loans available to borrowers. For small firms, one standard deviation increase in the ratio of accounts payable to sales implies a 9.2% increase in the ratio of short-term bank debt to sales, which is large and economically significant relative to the mean of 12.6%. A similar economically significant effect is observed for young and risky firms. The fact that a firm is capable of obtaining and servicing supplier debt conveys important information to bank loan officers regarding its creditworthiness. Previous empirical tests of trade credit theories have produced mixed results. Long, Malitz, and Ravid (1993) do not find evidence in support of the notion that trade credit acts as a substitute for conventional loans. They confirm that less creditworthy firms do not apply to more creditworthy firms for financing when credit market constraints are binding. Alternatively, Petersen and Rajan (1997) find that firms use more trade credit when access to loans from financial institutions is not available, and firms with greater access to intermediated financing offer more trade credit. Likewise, Calomiris, Himmelberg, and Wachtel (1995) find that the issuance of commercial paper by high-quality firms to finance the extension of trade credit to lower quality firms is countercyclical. Finally, Alphonse, Ducret, and Severin (2006) provide evidence that small firms in the United States use trade credit to improve their reputation as borrowers. The difference in this paper is my focus on the dynamics of firms' choice of short-term external funding. I analyze the choice between two forms of short-term external financing, one of which is trade credit. The remainder of this paper is organized as follows. Section I formulates the testable hypotheses. Section II describes the data and the variables used in the empirical analysis providing summary statistics. Section III discusses the main results of the study. Then, I provide some extensions to the main results and consider the effect of late payment and financial distress. Robustness tests to the estimation results are discussed in Section IV. The paper concludes with a summary and discussion of future research opportunities. I. Literature Background and Empirical Hypotheses In this section, I present and contextualize the empirical hypotheses. I classify these hypotheses into two groups: 1) the cross-sectional level effect and 2) the time effect. The level effect refers to predictions concerning the cross-sectional differences in the level of intermediary financing for firms that use different levels of trade credit. The time effect concerns the dynamics of trade credit growth. A summary of the hypotheses is presented in Table I. A. The Level Effect First, I focus on studies that argue that suppliers finance their customers because they have an advantage over institutional lenders. Suppliers may have an advantage in information acquisition (Smith, 1987; Biais and Gollier, 1997) as they know the size and timing of orders and observe which firms are able to take advantage of early payment discounts. Even though banks can collect similar information via transactions accounts, Petersen and Rajan (1997) argue that suppliers obtain their information in a faster and cheaper way. Because of the very short-term nature of trade credit, suppliers may be better able to evaluate the credit risk of their customers than specialized financial institutions. Jain (2001) argues that trade credit mitigates the problem of asymmetric information and enables credit constrained firms to finance valuable projects. When suppliers have an information advantage, trade credit is a substitute for conventional loans for firms facing high agency costs. This advantage will be especially pertinent when high-risk customers regularly replace their inventories (high inventory turnover). Alternatively, suppliers may have a liquidation advantage in extending credit to high risk firms as they can seize delivered goods from customers in default and redeploy them efficiently. In Frank and Maksimovic (2004), suppliers know their customer base well, are able to reverse product specialization more efficiently, and, therefore, resell the goods at a higher price. When suppliers have a liquidation advantage over conventional lenders, there can be an important complementarity between the use of trade credit and bank loans for firms facing high agency costs. This complementarity affects both the access to and the cost of intermediated financing. The positive relationship between bank credit and trade finance is consistent with the Biais and Gollier (1997) argument that the extension of trade credit reveals favorable information to other lenders, thus increasing their willingness to provide loans. Giannetti, Burkart, and Ellingsen (2011) demonstrate that after controlling for firm creditworthiness, firms that use trade credit tend to borrow from a larger number of banks and also use geographically more distant banks. These firms are offered better deals and lower fees for their credit lines. These findings suggest that the supplier advantage differs in nature from the advantage of relationship banks. A bank-firm relationship confers an informational monopoly for the bank that restricts the firm's ability to secure funding from other sources. By contrast, trade credit seems to give rise to a positive informational externality that facilitates access to funding from other lenders. When trade credit is used by borrowers to signal high creditworthiness, it will be a complementary source of funding to intermediate financing. Another advantage that suppliers may have over institutional lenders is an ability to control the activities of their buyers. When there are limited alternative sources for input goods, the seller could credibly threaten to cut off future supplies as soon as the buyer fails to comply with the credit terms, whereas banks are constrained by bankruptcy laws from withdrawing financing. Cunat (2007) argues that if the supplier is important for the customer's future business, the buyer has an incentive to strategically default on the bank and not on the supplier. Suppliers may be less susceptible to the risk of strategic default as inputs are less liquid and, therefore, less easily diverted than cash (Burkart and Ellingsen, 2004). When differentiated goods are purchased, we expect that defaults related to the diversion of corporate resources are less likely to occur if the supplier grants the loan. Thus, if trade credit is used by suppliers to overcome moral hazard issues, it will act as a substitute for bank loans. Next, I consider the implications of models of trade credit as a means for price discrimination (Smith, 1987; Brennan, Maksimovic, and Zechner, 1988). In many countries, suppliers are legally constrained from varying credit terms across customers. They can use trade credit to increase their sales provided that the effective price is reduced for price-sensitive customers. The price discrimination models argue that the high price of foregoing a cash discount implies that only financially constrained buyers will take the extended trade credit. Nonconstrained firms will not use supplier financing. I expect that under this hypothesis, trade credit will act as a substitute for institutional loans and this effect will be stronger when competition in the input market is low. Petersen and Rajan (1997), Danielson and Scott (2004), and Atanasova (2007) have found evidence that trade credit is used as a substitute for bank debt when firms try to mitigate financial constraints. Finally, Long et al. (1993) offer a product quality based explanation of trade credit use. The supplier may have superior information regarding the input's true market value and to alleviate the customer's fears of being cheated, the supplier may grant the customer an inspection period before demanding payment. This offering of trade credit is a way to guarantee product quality by allowing the buyer to return inferior goods without paying. As the quality of differentiated products and services is more difficult to verify, implicit guarantees through trade credit should be more frequently offered for these goods and services. B. The Time Effect This subsection discusses the dynamics of trade credit growth. (6) Small and young firms often face high failure rates limiting their access to bank loan. They need to rely on their suppliers for financing. Suppliers' low liquidation costs are most valuable to entrepreneurs who have a high-expected probability of default. Franks and Sussman (2000) find that banks are very harsh in debt renegotiations with distressed small firms, whereas suppliers expand the amount of credit during the period of distress, even when it ends in formal bankruptcy. However, the firms that survive grow larger, generate more stable, positive cash flows, and their reputation provides easier access to reasonably priced bank loans. As businesses mature, the expected probability of default declines and the value of trade credit as a signal of their creditworthiness is reduced. At the macro level, Nilsen (2002) finds that during periods of tight credit and high default rates, small firms react by borrowing more from their suppliers. High failure risk may also imply that small and young firms, especially those that highly value control rights, worry about the liquidation of their venture in the event of future financial distress. If suppliers are more lenient than banks toward financially distressed firms, as argued by Wilner (2000) and Franks and Sussman (2000), entrepreneurs may prefer trade credit financing during periods of high default risk. Finally, Long et al.'s (1993) explanation of trade credit use can also be generalized in a dynamic context. If the quality of the supplier's input directly affects the customer's commercial success, bundling input sale and credit increases the supplier's incentive to provide high-quality goods. Therefore, the customer's probability of success is higher than if a bank was the creditor. As the supplier-buyer relationship develops, firms rely more on supplier reputation and no longer need to verify the quality of supplied goods. I expect the product quality motive will also become less important as a determinant of trade credit use over the firm's life cycle. II. Data Description The empirical analysis is based on an unbalanced panel of UK companies drawn from the Financial Analysis Made Easier (FAME) database from 1998 to 2006. FAME is a financial information database of all UK companies. (7) I follow the literature and exclude companies in the financial industry (standard industrial classification [SIC] 6000-6999) and service industry (SIC 7000-8999), firms with significant merger and acquisition activities, as well as all firm-year observations for which the variables of interest (accounts payable, bank loans, sales, and others) are missing. The final sample consists of 31 guarantees, 17 limited liability partnerships, 62 unlimited companies, 15,125 private limited companies, 847 public, not quoted, and 365 public, quoted companies. (8) During the sample period, 180 companies were in receivership at some point, 202 were in liquidation, and 868 companies were dissolved. Two particular characteristics of the sample make it very appealing for my analysis. First, more than 97% of the firms are not listed on an exchange and there are many small and young firms in the sample. In addition, using a panel allows me to control for the presence of firm-level heterogeneity and for the possible endogeneity of trade credit financing decisions. COMPUSTAT and other databases that consist of large publicly traded firms cannot be used to examine the role of trade credit in the firms' financing cycle as many large multinational firms use trade credit extensively for accounting and tax purposes. Alternatively, the US Federal Reserve Board National Survey of Small Business Finances (NSSBF) provides information regarding privately held small firms, but the survey is a random, independent cross section and cannot be used to study the dynamics of a firm's financing choices. Table II summarizes the relative size of trade credit for three different samples of firms: 1) my sample of 16,447 UK firms (FAME, 1998-2006), 2) a sample of 3,000 small US firms with less than 500 employees (NSSBF, 1998), and 3) a sample of 4,362 large US firms with total assets exceeding $250 million (COMPUSTAT, 1998-2006). (9) The table reports that there are similar patterns of trade credit use for the firms in FAME and NSSBE For these samples, trade credit accounts for roughly one-fifth of the total assets of a representative firm and one-half of its short-term debt. For large firms in COMPUSTAT, however, the size of trade credit relative to total assets and long-term debt is lower. These firms also have a much higher accounts payable to short-term debt ratios. Many large firms use trade credit not only as a financing alternative to conventional short-term loans, but also for cash management, accounting, and tax purposes. The sample probability of firm failure for my dataset is high, close to 8%. To avoid an attrition bias, I present the main estimation results using an unbalanced panel of active and failed firms. I also follow Love, Preve, and Sarria-Allende (2007) and run the estimations on two alternative balanced samples. The results remain the same indicating that the attrition bias is empirically unimportant in my setting. These robustness results are discussed in Section IV. Next, I discuss some summary statistics for the companies in my sample. Panel A of Table III confirms that the median firm has total assets of GBP 7 million and is 18 years old. The median firm has a trade debt payment period of about 31 days, its accounts payables represent 8.6% of sales, and its receivables represent about 13.4% of sales. (10) Accounts payable do not seem to exceed inventories suggesting that the median firm in the sample uses trade credit primarily to finance its inventories. Panel B of Table III reports summary statistics for the median firm when I split the sample by age and size. (11) There seems to be no difference in the payment behavior of firms related to their age or size as the statistical test for differences in medians is not significant at any conventional level. This is consistent with the evidence in Ng et al. (1999) that trade credit terms vary across industries, but not across firm characteristics. The table indicates that for young firms, payables exceed inventories suggesting that for these firms, trade credit has an additional role to that of financing inventories. Also, the young and small firms in our sample rely more heavily on trade credit and short-term bank debt to finance their operations. Smaller firms extend more trade credit and have larger holdings of liquid assets. Table III also demonstrates the proportion of small and young firms in the sample. The extensive coverage of small and young companies in the sample is an important strength of the study as I expect that the hypotheses formulated in the previous section to be particularly relevant for this segment of the corporate sector. An appendix at the end of the paper provides a list of the variables used in this study. The main variables of interest are BL_S, the ratio of short-term bank loans to sales, and TC_S, the ratio of accounts payable to sales. (12) I include two sets of interaction variables where: 1) TC_S is multiplied by a dummy variable related to the size, age, riskiness, or legal status of the firm, and 2) TC_S is multiplied by the relevant industry-specific variables identified in Section I. The industry-specific variables include the average failure rate as a proxy for industry risk, and the average liquidation cost and intangibility as proxies for the specificity of inputs. I treat the high market concentration industries as less competitive. The controls include firm characteristics that affect bank borrowing and trade credit availability. I consider variables that measure the informational opacity/transparency and the firm's ability to raise external funds. I expect that the greater the opacity of smaller, younger firms, the stronger the reliance on trade credit and the poorer the availability of conventional bank loans. The level of the firm's inventories, accounts receivables, and cash holdings is used as a proxy for the availability of collateral. Opler et al. (1999) found that cash rich firms have lower cash flows, are substantially smaller, and are in industries with highly volatile earnings. I use the total debt to total asset ratio to control for leverage. Finally, I include the industry averages for the variables discussed in Section I to control for common industry-specific effects in the use of trade credit and bank loans. III. Empirical Analysis This section discusses the estimation results. The amount borrowed via trade credit and the amounts borrowed via bank loans are choice variables in firms' financing decisions, and are, therefore, endogenous. Panel data estimation methods, like ordinary least squares (OLS) with fixed effects, are not appropriate for this set up. I address the problem of the simultaneity of the decision to borrow from a financial intermediary and/or borrow from a supplier by using difference generalized method of moments (GMM) estimation. A. The Level Effect The first set of results examines the relationship between levels of trade credit and intermediated finance. I estimate the following regression specification: [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1) where [alpha] is a firm-specific effect and e is an error term. I test whether the level of trade credit (TCS) has a significant effect on the level of bank loans (BL_S; i.e., [[beta].sub.1] [not equal to] 0) and examine the coefficients of the interaction variables to provide evidence for the validity of the hypotheses I formulated in Section I. In particular, I test the complementarity of trade credit and bank loans for high agency costs firms implied by the liquidation hypothesis, [[beta].sub.2] > 0, [[beta].sub.3] > 0, and [[beta].sub.4] > 0, versus the substitution effect implied by the information advantage, controlling advantage, and price discrimination hypotheses, [[beta].sub.2] < 0, [[beta].sub.3] < [[beta].sub.4] and [[beta].sub.4] < 0. Further, the supplier information advantage will be particularly important for industries with high inventory turnover (i.e., stronger substitution effect between trade credit and bank loans, [[beta].sub.6] < 0), whereas the liquidation hypothesis implies that for industries with high liquidation costs, the supplier advantage decreases (i.e., weaker complementarity effect, [[beta].sub.7] < 0). The controlling advantage hypothesis predicts a stronger substitution effect for concentrated industries with high failure rates, [[beta].sub.8] < 0, and the price discrimination hypothesis implies that for concentrated industries, there is a stronger substitution effect, [[beta].sub.9] < 0. Finally, the product quality hypothesis implies a stronger substitution effect for firms in industries with high levels of intangible assets, [[beta].sub.10] < 0. Table IV presents the estimation results for regression Specification (1). Column (1) presents the results from an OLS regression with fixed effects. The coefficient of trade credit to sales ratio is positive and significant suggesting that trade credit and bank loans are complimentary sources of funds for the firms in my sample. Columns (2)-(7) report the estimation results from difference GMM estimation. In most of the GMM regressions, the coefficient of the noninteracted variable, TC_S ([[beta].sub.1]), is not significant and its size is much smaller than the coefficients of the interacted variables. The results in Column (3) are consistent with the theories based on the supplier liquidation advantage. Although, the coefficient of TC_S is negative, very small, and not significant, the coefficients of the interaction variables for small, young, and low quality ([[beta].sub.2], [[beta].sub.3], and [[beta].sub.4]) are large, positive, and significant. There is no evidence, however, that the lack of access to public financial markets generates any differences in the demand for trade credit as the coefficient of TC_S*Listed ([[beta].sub.5]) is not significant at any conventional level. These results suggest that the positive effect of trade credit on bank loans is important only for firms with high agency costs. For these firms, important complementarities exist between trade credit and intermediated finance. The interaction term between industry liquidation cost and trade credit is significantly negative lending further support to the supplier's liquidation advantage hypothesis. The results in Column (2) do not support the supplier's information advantage hypothesis as the coefficient of TC_S*IndTurnover ([[beta].sub.6]) is not significant. Column (4) provides the estimation results for the controlling advantage hypothesis. The data do not support this hypothesis as the coefficient of TC_S*IndFailure*IndConcentration ([[beta].sub.8]) has the wrong sign. The results in Column (5) do not support the price discrimination and financial constraints hypothesis as the coefficient of TC_S*IndConcentration ([[beta].sub.9]) is not significant. However, I find support for the product quality hypothesis. Column (6) indicates that the coefficient of TC_S*IndIntangibility ([[beta].sub.10]) is negative and significant, suggesting that when product quality verification is important, trade credit acts as a substitute for bank loans. To minimize the problem of multicollinearity, Columns (2)-(6) test the impact of the interaction variables separately for each of the empirical hypotheses. Column (7) reports the estimation results when all the interaction variables are estimated together. Although, the coefficients of the industry interaction variables are not significant, the coefficients of the interaction variables of trade credit with size, age, and risk are still significant and positive. This suggests that the liquidation advantage remains important even for the most general specification of regression Specification (1). In all regressions, the estimated coefficients of the firm-specific control variables have the expected sign. Larger firms use significantly more bank loans as a percentage of sales. The coefficient of Pledgeable Assets (inventory and accounts receivable to sales ratio) has a significant positive effect on the level of bank loans, whereas leverage has a significant negative effect. Most of the coefficients of the industry-specific controls are not significant. (13) To summarize, the increase in trade credit to sales ratio is associated with an increase in the short-term bank debt to sales ratio for small, young, and risky firms. This effect is statistically significant and economically important. The fact that there is an important complementarity between trade credit and intermediated finance provides support for the theoretical models based on the suppliers' liquidation advantage. I also find evidence for the validity of the product guarantee hypothesis as my estimation results indicate that when the quality of products or services is difficult to verify, trade credit is used as a substitute to conventional loans. Next, I analyze the effect of financial distress on my results. I estimate Specification (1) using two different samples. The first sample includes firms that remained active during the sample period. The second sample includes firms that were in receivership and firms that were liquidated or dissolved during the sample period. 14 Column (1) of Table V presents the estimation results of the liquidation hypothesis for all active firms, whereas Column (2) reports the results for the firms in the sample that have failed. (15) Columns (3) and (4) provide estimation results of the product quality hypothesis for active and failed firms. The estimated coefficients for the sample of active firms are the same as my main results. For firms in financial distress, the quality motive for using trade credit is not important anymore. However, the liquidation advantage of the supplier is very important for these firms. The coefficients of trade credit interacted with the dummies for size, age, and riskiness remain positive and significant. Also for companies in default, Size, Cash, and Pledgeable Assets have a significant positive effect on the level of bank loans, whereas Leverage has a significant negative effect. I also evaluate the effect of late payment and cash discount on the main findings of the paper. Pike, Cheng, Cravens, and Lamminmaki (2005) discuss the trade credit practices of UK firms. Unlike in the United States, where firms look for early resolution of uncertainty through more stringent risk screening, greater reliance on payment on or before delivery, and the offering of two-part terms, UK firms place significantly less emphasis on early payment and two-part terms. To reduce risk exposure, UK firms undertake credit risk insurance and impose additional conditions in the trade credit contract, such as the retention of title clauses to ease recovery of salvageable assets, third party guarantees, and charges on fixed assets. I expect that in the UK context, the estimation results will remain robust in the effects of late payment and cash discounts. To examine whether buyers use intermediated funds to pay off outstanding trade credit and take advantage of cash discounts, I estimate Specification (1) for the subsample of companies that have a payables period of ten days or less. Column (1) of Table VI reports the results. The coefficient of TC S is not significant. This finding is consistent with Borde and McCarty (1998), who demonstrate that in reality, buyers do not borrow from financial institutions to take advantage of cash discounts offered to them by suppliers. According to Pike et al. (2005), the most common trade credit term that UK suppliers offer is net 30 days. Columns (2), (3), and (4) of Table VI demonstrate how my estimation results change when I consider late payment. Column (2) reports the results for a subsample of all firms that have a payables period longer than 30 days, whereas Column (3) provides the results for firms with a payables period longer than 90 days. Contractual payment periods and late payments will vary across industries. In Column (4), I report the results for a subsample of firms that pay later than the median firm in each split two-digit SIC code industry group. My results indicate that for these firms, trade credit acts as a substitute for bank loans as the coefficient of TC S is significantly negative and larger than the coefficients of the interaction variables. This is consistent with theories arguing that binding financial constraints are an important determinant in the use of trade credit. It is also consistent with the evidence in Cunat (2007) and Boissay (2006) suggesting that many suppliers continue to extend credit and supply goods to customers that pay late or have defaulted on their trade bills in the past. The option to pay late allows some firms to respond to short-term adverse liquidity shocks by passing them on to their suppliers. As a robustness check, I estimate Specification (1) for all firms with a payables period shorter than 30 days and all firms with a payables period shorter than the industry median. The estimated coefficients are the same as the main results. B. The Time Effect This subsection examines the time dimension of the hypotheses outline in Section I. The regressions I estimate have the following specification: [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2) where [alpha] is a firm-specific effect and [epsilon] is an error term. I test the null hypothesis that firm age has a significant effect on trade credit growth, [[beta].sub.1] 0, against the alternative of no impact, [[beta].sub.1] = 0. The financing advantage hypothesis implies that mature firms in high turnover industries will be less reliant on trace credit as a source of funding or [[beta].sub.6] < 0. Similarly, the liquidation advantage hypothesis suggests that as firms acquire more reputation, the value of trade credit as a signal is reduced. As such, I expect that the negative effect of high liquidation costs on trade credit growth will be lower for mature firms, [[beta].sub.7] > 0. The product differentiation hypothesis implies that as the supplier-buyer relationship gradually develops, firms rely more on supplier reputation and no longer need to verify the quality of supplied goods, [[beta].sub.10] < 0. Alternatively, the controlling advantage and the price discrimination hypotheses predict no significant impact of firm age on the dynamics of trade credit growth, [[beta].sub.8] = 0 and [[beta].sub.9] = 0. Table VII illustrates the estimation results for empirical Specification (2). The estimated parameter of Age is negative, although it is significant only in three out of the seven regression specifications. The effect of firm size on trade credit growth is significantly negative in all specifications. Conversely, leverage does not have a significant effect. The coefficients of accounts receivable and inventory growth are large and positive as expected because trade credit and inventories are related in the company's product/cash cycle. Finally, industries with high liquidation costs have lower trade credit growth, whereas industries with a high degree of product intangibility rely more on trade credit over time. Columns (2)-(7) report the coefficients of the interaction terms between age and the industry-specific variables. Column (2) provides support for the hypothesis that the financing advantage of the supplier is reduced as firms mature. The coefficient of Age*IndTurnover ([[beta].sub.6]) is significant and negative. Column (3) confirms that the coefficient of Age*IndLiquidation ([[beta].sub.7]) is also significant and negative. This result does not lend support for the dynamic version of the liquidation hypothesis as high liquidation costs continue to be a deterrent to trade credit use even as firm age increases. Columns (4) and (5) verify that the controlling advantage and price discrimination motives do not have a separate effect on trade credit growth via age as the coefficients of Age*IndFailure*IndConcentration ([[beta].sub.8]) and Age*IndConcentration ([[beta].sub.9]) are not significant at any conventional level. In contrast, Column (6) reports that the coefficient of Age*IndIntangibility ([[beta].sub.10]) is negative and significant implying that the importance of trade credit as a means of product verification decreases with firm age. Finally, Column (6) presents the estimation results when all the interaction variables are estimated together. Although some of the coefficients are only weakly significant, my results remain qualitatively the same even for the most general requirement of Specification (2). To summarize, my results indicate that as firms mature, the role of trade credit as a source of financing becomes less important. I do not find that the importance of supplier liquidation costs decreases with age. However, my results confirm that the information and product quality motives become less important as firms mature. Because I measure trade credit as a ratio of accounts payables to sales, the decline in trade credit used is not simply driven by a decline in sales. IV. Robustness Tests This section considers several robustness tests to my main results. The FAME data set covers the entire population of limited liability companies in the UK. Once a company undergoes any type of reorganization, it is immediately excluded from FAME. An attrition bias may arise if firms in receivership or liquidation and those that are dissolved or acquired were to have different trade credit policies than firms that remain active. My main results are based on an unbalanced panel to control for the possibility of such an attrition bias. Also, I follow Love et al. (2007) and create two balanced samples. The first sample contains all firms that are present in the database for at least five consecutive years. The second sample contains firms that are in the data set for at least seven consecutive years. The estimation results for Specification (1) are reported in Table VIII. Although some of the coefficients are not significant any more because of the reduced sample size, my main results remain the same as the coefficients of the interaction variables of trade credit with size, age, and risk are still significant and positive. This suggests that an attrition bias is not important for my sample. In my dynamic regressions, I have also considered additional specifications with lags of the dependent variable and with a sector-specific time trend. The main findings reported in Section III are robust across these specifications. For example, including the lagged trade credit growth in Specification (2) decreases the statistical significance as I lose about 35% of the observations, but the results are qualitatively similar with coefficients that have the same sign and are of similar magnitude. Overall, the robustness tests suggest that the empirical relationships documented in this paper are robust, whereas the testable predictions of the theoretical models in the current literature, when taken individually, cannot explain the rich empirical pattern of trade credit use. V. Conclusions In this paper, I examine how firms choose to borrow to finance their operations and report new empirical evidence regarding the use of trade credit as a source of short-term external finance. Models based on a simple insight that trade credit may serve as a means of acquiring reputation, predict that firm characteristics linked to the degree of information opacity (size, age, and riskiness) are associated with companies that use more trade credit. My results indicate that the relationship between trade credit and conventional loans is far more complicated than suggested by existing research in the area. It is likely that the overall costs of borrowing and the extent of credit rationing are key factors in determining whether trade credit and intermediated loans should be treated as complements or substitutes. My findings also have important policy implications. Policies that emphasize the development of technologies for the valuation and monitoring of informationally opaque firms might be more effective than policies of government credit guarantees. It would be interesting to further explore how information provisions in the credit market would affect firms' borrowing prospects and decisions.
Appendix: Variables Used in the Empirical Analysis
Quiscore is a measure of the likelihood of failure. The score is a
number from 0 to 100 and the bands 0-12 (high-risk band) and 21-40
(caution band) suggest significant risk of company failure. Industry
averages are constructed for every two-digit industry SIC code.
Variable Definition
BL_S Short-term bank loans (including overdrafts)
to sales ratio
TC_S Accounts payable to sales ratio
Interaction dummies
Small One if number of employees is less than 50,
zero otherwise
One if firm age is less than ten years, zero
otherwise
Low Quality One if Quiscore <40, zero otherwise
Listed One if the firm is publicly traded, zero
otherwise
Age Logarithm of one plus the firm's age
Size Logarithm of beginning of year total assets
Cash Cash and cash equivalents to sales ratio
Pledgeable Assets Inventory and accounts receivable to sales
ratio
Leverage Long-term debt to total assets ratio
Industry averages
IndFailure Industry failure rate measured by the
percentage of bankrupt firms
IndLiquidation Industry liquidation costs measured by the
percent of inventory that is finished
goods
IndConcentration Percentage of industry sales accounted for by
the largest four firms
IndIntangibility Industry intangible assets to total assets
ratio
IndInventory Industry inventory turnover measured by the
sales to inventory ratio
References Adams, P.D., S.B. Wyatt, and Y.H. Kim, 1990, "Trade Credit in a Theory of Finance: A Puzzle in a Competitive Capital Market," Presented at the Annual Meeting of the Financial Management Association in Orlando, FL, October. Alphonse, P., J. Ducret, and E. Severin, 2006, "When Trade Credit Facilitates Access to Bank Finance: Evidence from US Small Business Data," University of Valenciennes, Working paper. Antov, D. and C. Atanasova, 2011, "Trade Credit and Intermediary Finance: Theory and Evidence," Simon Fraser University, Working paper. Atanasova, C., 2007, "Access to Institutional Finance and the Use of Trade Credit," Financial Management 36, 49-67. Biais, B. and C. Gollier, 1997, "Trade Credit and Credit Rationing," Review of Financial Studies 10, 903-937. Boissay, F., 2006, "Trade Credit Defaults and Liquidity Provision by Firms," Working paper Series 753, European Central Bank. Borde, S. and D. McCarty, 1998, "Determining the Cash Discount in the Firm's Credit Policy: An Evaluation," Journal of Financial and Strategic Decisions 11, 41-49. Brealey, R. and S. Myers, 2000, Principles of Corporate Finance, 6th Ed., New York, NY, McGraw Hill. Brennan, M.J., V. Maksimovic, and J. Zechner, 1988, "Vendor Financing," Journal of Finance 43, 1127-1141. Burkart, M. and T. Ellingsen, 2004, "In-Kind Finance: A Theory of Trade Credit," American Economic Review 94, 569-590. Calomiris, C., C. Himmelberg, and P. Wachtel, 1995, "Commercial Paper, Corporate Finance, and the Business Cycle: A Microeconomic Perspective," Carnegie-Rochester Conference Series on Public Policy 42, 203-250. Cunat, V., 2007, "Trade Credit: Suppliers as Debt Collectors and Insurance Providers," Review of Financial Studies 20, 491-527. Danielson, M.G. and J.A. Scott, 2004, "Bank Loan Availability and Trade Credit Demand," Financial Review 39, 579-600. Frank, M. and V. Maksimovic 2004, "Trade Credit, Collateral and Adverse Selection," University of Maryland, Working paper. Franks, J. and O. Sussman, 2000, "An Empirical Study of Financial Distress of Small Bank-Financed UK. Companies: A Reassessment of English Insolvency Law," London Business School, Working paper. Giannetti, M., M. Burkart and T. Ellingsen, 2011, "What You Sell is What You Lend? Explaining Trade Credit Contracts," Review of Financial Studies 24, 1261-1298. Harhoff, D. and T. Korting, 1998, "Lending Relationships in Germany: Empirical Results from Survey Data," CEPR Discussion paper 1917. Jain, N., 2001, "Monitoring Costs and Trade Credit," Quarterly Review of Economics and Finance 41, 89-110. Long, M.S., I.B. Malitz, and S.A. Ravid, 1993, "Trade Credit, Quality Guarantees, and Product Marketability," Financial Management 22, 117-127. Love, I., L. Preve, and V. Sarria-Allende, 2007, "Trade Credit and Bank Credit: Evidence from Recent Financial Crises," Journal of Financial Economics 83, 435-469. Molina, C. and L. Preve, 2009, "Trade Receivables Policy of Distressed Firms and Its Effect on the Costs of Financial Distress," Financial Management 38, 663-686. Ng, C., J. Smith, and R. Smith, 1999, "Evidence on the Determinants of Credit Terms Used in Interfirm Trade," Journal of Finance 54, 1109-1029. Nielsen, J., 2002, "Trade Credit and the Bank Lending Channel," Journal of Money, Credit and Banking 34, 226-253. Opler, T., L. Pinkowitz, L.R. Stulz, and R. Williamson, 1999, "The Determinants and Implication of Corporate Cash Holdings," Journal of Financial Economics 52, 3-46. Petersen, M. and R. Rajan, 1997, "Trade Credit: Theories and Evidence," Review of Financial Studies 10, 661-691. Pike, R., N.S. Cheng, K. Cravens, and D. Lamminmaki, 2005, "Trade Credits Terms: Asymmetric Information and Price Discrimination Evidence from Three Continents," Journal of Business Finance and Accounting 32, 197-236. Smith, J.K., 1987, "Trade Credit and Informational Asymmetry," Journal of Finance 42, 863-872. Wilner, B.S., 2000, "The Exploitation of Relationships in Financial Distress: The Case of trade Credit," Journal of Finance, Vol. 55, No. 1, pp. 153-178. (1) Statistical abstract of the United States, http://www.census.gov/. (2) UK Office for National Statistics. (3) Financial Statements Statistics of Corporations by Industry, Ministry of Finance, Japan. (4) Trade credit practice, however, is pervaded by conventions and rules of thumb. The most typical terms of trade in various industries are usually published in handbooks. For example, shoe manufacturers commonly use "5/10, net 30" as terms of sale, whereas toy manufacturers generally sell goods on terms of "2/30, net 50" (Brealey and Myers, 2000). (5) For example, among the EU countries, only in Germany is a 2% penalty commonly granted for payment delay (Harhoff- Korting, 1998). Despite the introduction of the EU Directive on late payment in 2000, creditors" willingness to enforce the penalty rules is extremely low due to the direct costs involved in making use of the Directive and of the indirect ones created by the fear of hindering relationships between suppliers and customers (ECB Official Journal, 2008). (6) There are many studies on the cross-sectional determinants &trade credit use. Petersen and Rajan (1997) argue that the most important challenge for future research is to examine the determinants of trade credit use over time. (7) FAME contains data for both active and failed firms and data are available for, at most, ten years for each firm. Financial data dates back, at most, to 1997 if firms have their accounts filed in FAME up to 2006 and back to 1998 if they have filed accounts up to 2007. To avoid selection bias, the period of analysis is 1998-2006. (8) Note that in the UK, public refers to the legal status of the firm, whereas quoted refers to whether the firm is listed on a stock exchange. Private firms in the UK are necessarily unquoted, but public firms may be quoted or unquoted. In the UK, most public firms are not quoted. (9) Companies in the financial industry (SIC 6000-6999) and service industry (SIC 7000-8999) are excluded. Sample observations are winsorized at 1% and 99%. (10) The positive average net trade credit position is due to trade credit extended to households and companies, which are not part of the sample. (11) I use a standard classification of small and large firms in terms of number of employees. (12) The empirical results remain the same if the ratio of accounts payable to total assets is used. (13) These estimation coefficients are not reported, but are available upon request. (14) I have also estimated Specification (1) for all distressed firms using the definitions in Molina and Preve (2009). The results remain the same. (15) analyze the data for these firms just before they enter into receivership, are dissolved, or liquidated. I thank the anonymous referee, the seminar attendants, and conference participants at the University of York, Simon Fraser University, the 17 AFFi Conference, and the 21st ABF Conference for helpful suggestions. Christina Atanasova * * Christina Atanasova is an Assistant Professor at the Beedie School of Business at Simon Fraser University in Burnaby, BC, Canada.
Table I. Summary of the Empirical Hypotheses
Hypothesis Level Effect
Suppliers financing advantage:
1. Information acquisition Substitution: Trade credit mitigates
and financial constraints asymmetric information for
credit-constrained firms.
Supplier's advantage is greater
when inventory turnover is high.
2. Liquidation advantage and Complementarity: Trade credit
high agency costs reveals favorable information to
other lenders and increases their
willingness to provide loans. The
effect decreases as liquidation
costs increase.
3. Controlling advantage and Substitution: Trade credit mitigates
high default probability moral hazards as the seller could
credibly threaten to cut off future
supplies. The effect decreases as
the firm's financial health
strengthens and input market
competition increases.
Price discrimination and Substitution: The relatively high
financial constraints price of trade credit implies that
only financially constrained
buyers will take it up. The effect
decreases as input market
competition increases.
Product quality Substitution: When the supplier has
superior information about the
input's value, the customer may
be granted an inspection period
before payment. The effect if
stronger for
differentiated/intangible inputs.
Hypothesis Time Effect
Suppliers financing advantage:
1. Information acquisition Negative: As firms mature
and financial constraints and new sources of
information are available,
their reliance on trade
credit as a source of
funding decreases.
2. Liquidation advantage and Positive: As firms acquire
high agency costs more reputation, the value
of trade credit as a signal is
reduced. The negative
effect of supplier
liquidation costs is lower
for mature firms.
3. Controlling advantage and No impact: Firms that value
high default probability control rights prefer trade
credit financing because
suppliers are more lenient
than banks.
Price discrimination and No impact: Financially
financial constraints constrained firms borrow
more from their suppliers.
Product quality Negative: Over time, as the
supplier-buyer relationship
gradually develops, firms
rely more on supplier
reputation and no longer
need to verify the quality
of supplied goods.
Table II. Relative Importance of Trade Credit
The table presents the relative size of trade credit for three
different samples of firms: 1) my sample of 16,447 UK firms (FAME,
1998-2006),2) a sample of 3,000 US firms with less than 500 employees
(NSSBF, 1998), and 3) a sample of 4,362 US publicly traded firms with
total assets greater than $250 million (COMPUSTAT, 1998-2006). The
reported figures are the averages across the sample firms.
Sample Country Period Number of Trade
Firms Credit/Assets
FAME UK 1998-2006 16,447 0.20
NSSBF US 1998 3,000 0.17
COMPUSTAT US 1998-2006 4,362 0.08
Sample Trade Credit/ Trade Credit/
Long-Term Short-Term
Debt Debt
FAME 0.37 0.45
NSSBF 0.35 0.50
COMPUSTAT 0.26 2.23
Table III. Summary Statistics
The table presents summary statistics for 16,447 UK firms (1998-
2006). Days payable is the ratio of accounts payable to purchases
multiplied by 365. Panel B of the table reports p values for
differences in medians. The test employs chi-squared statistics.
Panel A
Mean Standard Deviation Median
Total assets (million GBP) 98.6 50.7 7
Age (years) 24.9 22.3 18
Days payable 45.8 942.2 31.3
Ratios (percentage of sales)
Accounts payable 12.6 103.2 8.6
Short-term bank debt 27.9 552.7 7.8
Long-term debt 33.7 363.3 4.6
Inventory 8.1 169.1 8.4
Accounts receivable 15.1 93.1 13.4
Cash holdings 22.1 2,281.2 2.7
Operating profit 0.1 98.8 2.8
Panel B
Age
Median firm <10 years [greater than or p Value
equal to] 10 years
Total assets (million GBP) 5.7 7.5 0.00
Days payable 31.8 31.2 1.00
Ratios (percentage of sales)
Accounts payable 8.7 7.6 0.04
Short-term bank debt 8.4 7.3 0.00
Long-term debt 6.8 8.5 0.00
Inventory 7.2 8.8 0.00
Accounts receivable 12.8 13.6 0.01
Cash holdings 2.8 2.7 1.00
Operating profit 2.8 2.8 1.00
Number of firms 4,209 12,238
Size
Median firm Small Large p Value
Total assets (million GBP) 2.9 336.6 0.00
Days payable 28.9 29.7 0.98
Ratios (percentage of sales)
Accounts payable 8.1 7.9 0.00
Short-term bank debt 9 6.8 0.00
Long-term debt 3.2 7.5 0.00
Inventory 9.2 6.9 0.00
Accounts receivable 15.1 10.5 0.00
Cash holdings 3.5 2.9 0.01
Operating profit 2.6 3.7 0.00
Number of firms 4,839 3,915
Table IV. The Level Effect
The dependent variable is BL_S (short-term bank loans to sales ratio).
Column (1) reports the estimation results from an OLS regression.
Columns (2)-(7) report the estimation results from difference GMM
regressions where the endogenous variables are instrumented with lags
two and three. Industry controls include IndInventory, IndLiquidation,
IndFailure, IndConcentration, and Indlntangibility. All regressions
are estimated with firm-specific fixed effects and year effects.
Standard errors robust to industry cluster effects are reported in
parenthesis.
Variable (1) (2)
TC_S 0.1341 *** 0.2062
(0.0116) (0.1702)
TC_S*Small
TC S*Young
TC_S*LowQuality
TC_S*Listed
TC_S*Indventory 0.0307
(0.0887)
TC_S*IndLiquidation
TC_S*IndFailure*IndConcentration
TC*IndConcentration
TC*Indintangibility
Age 0.0555 *** 0.0462 ***
(0.0150) (0.0148)
Size 0.0198 *** 0.0188 ***
(0.0072) (0.0072)
Cash 0.0155 *** 0.0241 ***
(0.0041) (0.004)
PledgeableAssets 0.5067 *** 0.5663 ***
(0.0080) (0.0080)
Leverage -0.1606 *** -0.1618 ***
(0.0154) (0.0153)
Industry control Yes Yes
N 16,447 12,424
Variable (3) (4)
TC_S -0.0152 -0.1376
(0.0255) (0.1532)
TC_S*Small 0.3922 **
(0.0182)
TC S*Young 0.6079 ***
(0.0272)
TC_S*LowQuality 0.1570 ***
(0.0314)
TC_S*Listed 0.6794
(0.4060)
TC_S*Indventory
TC_S*IndLiquidation -0.0530 ***
(0.0184)
TC_S*IndFailure*IndConcentration 0.0213 **
(0.0108)
TC*IndConcentration
TC*Indintangibility
Age -0.0069 -0.0350
(0.0068) (0.0486)
Size 0.0671 *** 0.0208
(0.0141) 0.0072
Cash 0.1218 *** 0.0090 **
(0.0042) (0.0041)
PledgeableAssets 0.7826 *** 0.5593 ***
(0.0086) (0.0080)
Leverage -0.1541 *** -0.1597 ***
(0.0144) (0.0153)
Industry control Yes Yes
N 12,424 12,424
Variable (5) (6)
TC_S 0.5220 ** -0.3596
(0.2240) (0.3370)
TC_S*Small
TC S*Young
TC_S*LowQuality
TC_S*Listed
TC_S*Indventory
TC_S*IndLiquidation
TC_S*IndFailure*IndConcentration
TC*IndConcentration -0.5345
(0.9812)
TC*Indintangibility -1.9459 ***
(0.0503)
Age -0.0338 -0.0431
(0.0248) (0.0474)
Size 0.0162 *** 0.0226 ***
(0.0072) (0.0072)
Cash 0.0247 *** 0.0069
(0.0040) (0.0040)
PledgeableAssets 0.5611 *** 0.6089 ***
(0.0079) (0.0081)
Leverage -0.1608 *** -0.1653 ***
(0.0152) (0.0152)
Industry control Yes Yes
N 12,424 12,424
Variable (7)
TC_S 0.4421
(0.3013)
TC_S*Small 0.1787 **
(0.0811)
TC S*Young 0.1624 **
(0.0794)
TC_S*LowQuality 0.2772
(0.0175)
TC_S*Listed -0.9278
(0.7329)
TC_S*Indventory 0.3189
(0.4455)
TC_S*IndLiquidation -0.6571
(0.8261)
TC_S*IndFailure*IndConcentration -0.1655
(0.2301)
TC*IndConcentration 0.0029
(0.0331)
TC*Indintangibility -0.0207
(0.0508)
Age 0.0719
(0.0266)
Size -0.0110
(0.0319)
Cash 0.1109 *
(0.0625)
PledgeableAssets 0.8077
(0.1951)
Leverage -0.1594 **
(0.0771)
Industry control Yes
N 12,424
*** Significant at the 0.01 level.
** Significant at the 0.05 level.
* Significant at the 0.10 level.
Table V. The Effect of Financial Distress
The dependent variable is BL S (short-term bank loans to sales ratio).
Columns (1) and (2) present results for all active firms and Columns
(3) and (4) results for all failed firms. All regression
specifications are estimated using difference GMM and include firm-
specific fixed effects and year effects. The endogenous variables are
instrumented with lags two and three. Industry controls include
IndInventory, IndLiquidation, IndFailure, IndConcentration, and
IndIntangibility. Standard errors robust to industry cluster effects
are reported in parenthesis.
Variable (1) (2) (3)
TC_S -0.1043 1.3579 *** -0.1488 *
(0.0262) (0.0343) (0.0938)
TC_S*Small 0.4019 *** 0.1546 *
(0.0226) (0.0815)
TC_S*Young 0.7381 *** 0.5737 ***
(0.0285) (0.1483)
TC_S*LowQuality 0.9755 *** 0.3303
(0.0326) (0.0932)
TC_S*Listed 0.0801 0.3682
(0.1418) (0.5167)
TC_S*IndLiquidation -0.5481 *** 0.1930
(0.1895) (0.1094)
TC*Indintangibility -1.9329 ***
(0.0512)
Age -0.0696 -0.0452 -0.0107
(0.0451) (0.0524) (0.0449)
Size 0.0105 0.0222 *** 0.0521 **
(0.0070) (0.0074) (0.0221)
Cash 0.1588 *** 0.0052 0.0356 ***
(0.0049) (0.0046) (0.0056)
PledgeableAssets 0.7900 *** 0.6055 *** 0.8031 ***
(0.0089) (0.0083) (0.0312)
Leverage -0.1542 *** -0.1691 *** -0.1276 ***
(0.0153) (0.0161) (0.0296)
Industry Controls Yes Yes Yes
N 11,522 11,522 902
Variable (4)
TC_S -0.2123
(0.3718)
TC_S*Small
TC_S*Young
TC_S*LowQuality
TC_S*Listed
TC_S*IndLiquidation
TC*Indintangibility -0.6278
(0.4908)
Age -0.0212
(0.0449)
Size 0.0468 **
(0.0222)
Cash 0.0334 ***
(0.0050)
PledgeableAssets 0.8108 ***
(0.0311)
Leverage -0.1268 ***
(0.0297)
Industry Controls Yes
N 902
*** Significant at the 0.01 level.
** Significant at the 0.05 level.
* Significant at the 0.10 level.
Table VI. The Effect of Cash Discount and Late Payment
The dependent variable is BLS (short-term bank loans to sales ratio).
Column (1) presents estimation results for firms with a payables
period less than ten days. Column (2) reports estimation results for
firms with a payables period longer than 30 days. Column (3) presents
estimation results for firms with a payables period longer than 90
days, and Column (4) reports estimation results for firms with
payables periods above the industry median period. All regression
specifications are estimated using difference GMM and include firm-
specific fixed effects and year effects. The endogenous variables are
instrumented with lags two and three. Industry controls include
IndInventory, IndLiquidation, IndFailure, IndConcentration, and
IndIntangibility. Standard errors are robust to industry cluster
effects are reported in parentheses.
Variable (1) (2)
TC_S -0.8947 -0.1781
(0.7205) (0.3011)
TC_S*Small 0.8746 ** 0.3917 ***
(0.3250) (0.0250)
TC_S*Young 0.3551 ** 0.2153 ***
(0.1726) (0.0321)
TC_S*LowQuality 0.1875 * 0.1010 ***
(0.1053) (0.0357)
TC_S*Listed 0.0786 0.5464
(0.3309) (0.4655)
TC_S*IndLiquidation -0.0973 -0.6353 ***
(0.8443) (0.0221)
Age -0.1772 -0.0506 **
(0.1160) (0.0205)
Size 0.0224 0.0119
(0.0524) (0.0104)
Cash 0.9248 *** 0.1051 ***
(0.1689) (0.0053)
Pledgeable Assets 0.6033 *** 0.8343 ***
(0.1748) (0.0110)
Leverage -0.0157 -0.1570 ***
(0.0103) (0.0214)
Industry controls Yes Yes
N 1,562 6,371
Variable (3) (4)
TC_S -0.4168 -0.4902 **
(0.5071) (0.2273)
TC S*Small 0.1241 ** 0.3397 ***
(0.0507) (0.1171)
TC_S*Young 0.1386 ** 0.4668 ***
(0.0511) (0.1132)
TC_S*LowQuality 0.1530 *** 0.2631 ***
(0.0314) (0.0739)
TC_S*Listed -0.1445 -0.3425
(0.2692) (0.6056)
TC_S*IndLiquidation -0.3497 *** -0.4771 ***
(0.0496) (0.0961)
Age -0.0359 * -0.0445 ***
(0.0189) (0.0132)
Size 0.0284 *** 0.0287 *
(0.0044) (0.0164)
Cash 0.0429 ** 0.0763
(0.0226) (0.0101)
Pledgeable Assets 0.4660 *** 0.4220
(0.0141) (0.0229)
Leverage -0.1421 *** -0.1510 ***
(0.0089) (0.0131)
Industry controls Yes Yes
N 1,072 6,196
*** Significant at the 0.01 level.
** Significant at the 0.05 level.
* Significant at the 0.10 level.
Table VII. The Time Effect
Dependent variable is ATC S (the change in accounts payable to sales
ratio). All regression specifications are estimated using OLS and
include firm-specific fixed effects and year effects. Industry
controls include IndInventory, IndLiquidation, IndFailure,
IndConcentration, and IndIntangibility. Standard errors robust to
industry cluster effects are reported in parentheses.
Variable (1) (2)
Age -0.0944 *** -0.0024
(0.0296) (0.0104)
Size -0.0195 *** -0.0185 ***
(0.0038) (0.0038)
ARec 0.1208 *** (0.1207 ***
(0.0021) (0.0025)
AInventory 0.3134 *** (0.3130) ***
(0.0028) (0.0028)
Age*IndTurnover -0.0313 ***
(0.0045)
Age*IndLiquidation
Age *IndFailure*IndConcentration
Age*IndConcentration
Age*Indintangibility
Leverage -0.0044 -0.0039
(0.0072) (0.0073)
Industry controls Yes Yes
N 14,882 14,882
Variable (3) (4)
Age -0.0164 * -0.0427 ***
(0.0097) (0.0093)
Size -0.0177 *** -0.0216 ***
(0.0037) (0.0037)
ARec 0.1197 *** 0.1170 ***
(0.0020) (0.0019)
AInventory 0.3178 *** 0.2938 ***
(0.0021) (0.0027)
Age*IndTurnover
Age*IndLiquidation -0.0121 ***
(0.0006)
Age *IndFailure*IndConcentration 0.2215
(0.3145)
Age*IndConcentration
Age*Indintangibility
Leverage -0.0060 -0.0034
(0.0072) (0.0070)
Industry controls Yes Yes
N 14,882 14,882
Variable (5) (6)
Age -0.0044 -0.0166
(0.00933) (0.0926)
Size -0.0210 *** -0.0189 ***
(0.0034) (0.0037)
ARec 0.1200 *** 0.1217 ***
(0.0028) (0.0020)
AInventory 0.2858 *** 0.2911 ***
(0.0029) (0.0028)
Age*IndTurnover
Age*IndLiquidation
Age *IndFailure*IndConcentration
Age*IndConcentration 0.3897 *
(0.2584)
Age*Indintangibility -0.1703 ***
(0.0042)
Leverage -0.0039 -0.0047
(0.0071) (0.0071)
Industry controls Yes Yes
N 14,882 14,882
Variable (7)
Age -0.0013
(0.0015)
Size -0.0194 **
(0.0036)
ARec 0.1168 **
(0.0021)
AInventory 0.2968 **
(0.0031)
Age*IndTurnover -0.0247 **
(0.0038)
Age*IndLiquidation -0.0129
(0.0268)
Age *IndFailure*IndConcentration -0.1123
(0.0929)
Age*IndConcentration -0.3853
(0.2985)
Age*Indintangibility -0.1123 *
(0.0629)
Leverage 0.0027
(0.0069)
Industry controls Yes
N 14,882
*** Significant at the 0.01 level.
* Significant at the 0.10 level.
Table VIII. The Level Effect for Two Balanced Samples
Dependent variable is BL_S (short-term bank loans to sales ratio).
Columns (1)-(4) present the difference GMM regression results for
firms with at least five consecutive year observations. Columns (5)-
(8) reports estimation results from difference GMM regressions for
firms with at least seven consecutive year observations. The
endogenous variables are instrumented with lags two and three.
Industry controls include IndInventory, IndLiquidation, IndFailure,
IndConcentration, and Indlntangibility. All regressions are estimated
with firm-specific fixed effects and year effects. Standard errors
robust to industry cluster effects are reported in parentheses.
Variables (1) (2)
TC_S -0.2791 -0.0239
(0.2046) (0.0502)
TC_S*Small 0.2422 ***
(0.0304)
TC_S*Young 0.4322 ***
(0.1657)
TC_S*LowQuality 0.3958 ***
(0.0295)
TC_S*Listed 0.3779
(0.4021)
TC S*IndInventory 0.0221 **
(0.0111)
TC_S*IndLiquidation -0.1838
(0.1599)
TC_S*IndFailure*Ind/Concentration
TC*IndConcentration
TC*Indlntangibility
Age 0.0506 *** 0.0170
(0.0163) (0.0155)
Size 0.0147 * 0.0126 *
(0.0078) (0.0074)
Cash 0.0236 *** 0.1621 ***
(0.0047) (0.0051)
Pledgeable Assets 0.6438 *** 0.6689 ***
(0.0114) (0.0110)
Leverage -0.1651 *** -0.1518 ***
(0.0163) (0.0154)
Industry Controls Yes Yes
N 10,769 10,769
Variables (3) (4)
TC_S -0.0339 0.5359
(0.0382) (0.4589)
TC_S*Small
TC_S*Young
TC_S*LowQuality
TC_S*Listed
TC S*IndInventory
TC_S*IndLiquidation
TC_S*IndFailure*Ind/Concentration 0.0177
(0.2387)
TC*IndConcentration -0.0035
(0.0030)
TC*Indlntangibility -1.2979 *
(0.6678)
Age -0.0187 -0.0151
(0.0168) (0.0162)
Size 0.0183 0.0185 **
(0.0013) (0.0078)
Cash 0.0021 ** 0.0015
(0.0010) (0.0047)
Pledgeable Assets 0.2481 *** 0.7161 ***
(0.0190) (0.0116)
Leverage -0.0607 *** -0.1681 ***
(0.0029) (0.0162)
Industry Controls Yes Yes
N 10,769 10,769
Variables (5) (6)
TC_S 0.3232 -0.0406
(0.3188) (0.0268)
TC_S*Small 0.2185 ***
(0.0231)
TC_S*Young 0.6778 ***
(0.0337)
TC_S*LowQuality 0.3689 ***
(0.0344)
TC_S*Listed 0.9760
(0.8471)
TC S*IndInventory 0.0446 ***
(0.0128)
TC_S*IndLiquidation -0.3150
(0.2105)
TC_S*IndFailure*Ind/Concentration
TC*IndConcentration
TC*Indlntangibility
Age 0.0512 -0.0430
(0.1919) (0.0586)
Size 0.0120 -0.0048
(0.0086) (0.0071)
Cash 0.1007 * 0.1922 ***
(0.0542) (0.0053)
Pledgeable Assets 0.9465 *** 0.9706 ***
(0.0155) (0.0126)
Leverage -0.1748 *** -0.1508 ***
(0.0183) (0.0150)
Industry Controls Yes Yes
N 8,884 8,884
Variables (7) (8)
TC_S -0.2544 0.0912
(0.3933) (0.0521)
TC_S*Small
TC_S*Young
TC_S*LowQuality
TC_S*Listed
TC S*IndInventory
TC_S*IndLiquidation
TC_S*IndFailure*Ind/Concentration 0.3294
(0.2570)
TC*IndConcentration -0.5338
(0.5731)
TC*Indlntangibility -1.1301
(0.7314)
Age 0.0353 * 0.0507
(0.0188) (0.1927)
Size 0.0057 0.0183 **
(0.0085) (0.0087)
Cash 0.1311 *** 0.0505
(0.0054) (0.0054)
Pledgeable Assets 0.9870 *** 0.9451***
(0.0147) (0.0147)
Leverage -0.1680 *** -0.1732***
(0.0179) (0.0184)
Industry Controls Yes Yes
N 8,884 8,884
*** Significant at the 0.01 level.
** Significant at the 0.05 level.
* Significant at the 0.10 level.
|
|
||||||||||||||||||
Printer friendly
Cite/link
Email
Feedback
Reader Opinion