Trade Credit and the Supply Chain

Transcription

1 Trade Credit and the Supply Chain Daniela Fabbri Cass Business School City University London London EC1Y 8TZ, UK Leora F. Klapper * The World Bank Development Research Group Washington, DC [email protected] October 2011 Abstract: This paper studies supply chain financing. We investigate why a firm extends trade credit to its customers and how this decision relates to its own financing. We use a novel firm-level database of Chinese firms with unique information on market power in both output and input markets and on the amount, terms, and payment history of trade credit simultaneously extended to customers (accounts receivable) and received from suppliers (accounts payable). We find that suppliers with relatively weaker market power are more likely to extend trade credit and have a larger share of goods sold on credit. We also examine the importance of financial constraints. Access to bank financing and profitability are not significantly related to trade credit supply. Rather, firms that receive trade credit from their own suppliers are more likely to extend trade credit to their customers, and to match maturity between the contract terms of payables and receivables. This matching practice is more likely used when firms face strong competition in the product market (relative to their customers), and enjoy strong market power in the input market (relative to their suppliers). Similarly, firms lacking internal resources or external bank credit, are more dependent on the receipt of supplier financing in order to extend credit to their customers. * Corresponding author. We thank Mariassunta Giannetti, Vicente Cun at, Arturo Bris, Luc Laeven, Inessa Love, Max Macsimovic, Robert Marquez, Rohan Williamson, Chris Woodruff, and participants at the 2009 Financial Intermediation Research Society (FIRS) Conference on Banking, Corporate Finance and Intermediation in Prague, the Small Business Finance What Works, What Doesn t? Conference in Washington, DC, the Second Joint CAF- FIC-SIFR Conference on Emerging Market Finance in Stockholm, and the 4 th Csef-Igier Symposium on Economics and Institutions in Capri for valuable comments, and Taras Chemsky for excellent research assistance. The opinions expressed do not necessarily represent the views of the World Bank, its Executive Directors, or the countries they represent.

2 Large, creditworthy buyers force longer payment terms on less creditworthy suppliers. Large creditworthy suppliers incent less credit worthy SME buyers to pay more quickly CFO Magazine, April Introduction Supply chain financing (inter-firm financing) is an important source of funds for both small and large firms in developed and emerging markets around the world (e.g. Petersen and Rajan, 1997, Fishman and Love, 2003, Demirguc-Kunt and Maksimovic, 2002). An interesting characteristic of supply chain financing is that many firms use trade credit both to finance their input purchases (accounts payable) and to offer financing to their customers (accounts receivable). This pattern is widespread among both large public companies and small creditconstrained firms (McMillan and Woodruff, 1999; Marotta, 2005). These earlier results raise several questions related to capital structure decisions of the firm: For instance, why do listed firms with access to both private and public financial markets make large use of trade credit, if trade credit is presumed to be more expensive than bank finance? Alternatively, why do small credit-constrained firms decide to offer trade credit to their customers and how do they finance this extension of credit? Related literature has mainly focused on one-side of the trade credit relationship in isolation and has therefore been unable to fully address these questions. We take a new step in this direction, by looking at trade credit along the supply chain, i.e. we consider both sides of firm business the relation between a firm, its customers (product market) and its suppliers (input market). This unique perspective is crucial to highlight a novel link between the supply and the demand of trade credit. We use a large firm-level survey of Chinese firms, which is a unique source of data for at least two reasons. To our knowledge, this is the first firm-level data set providing detailed and 1

3 rich information on both the market environment and contract features of supplier and customer supply-chain financing. For example, it contains information on the amount, terms, and payment history of both trade credit extended by firms to their customers (accounts receivable), as well as the receipt of trade credit by firms from their own suppliers (accounts payable). Moreover, since only about 30% of firms in our sample have a line of credit, our data allows us to test the role of credit constraints in the firm decision to offer trade credit. Our empirical analysis provides a number of intriguing results. First, we document the importance of trade credit as a competitive gesture. Specifically, firms with weak bargaining power towards their customers measured as the percentage of sales to the largest customer or the number of suppliers of the firm s most important customer are more likely to extend trade credit, have a larger share of goods sold on credit and offer a longer payment period before imposing penalties. Similarly, firms facing stronger competition in the product market measured by the degree of sales concentration in the domestic market and the introduction of new products or lowered prices in the past year - are more likely to offer trade credit and provide better credit terms. These results hold after controlling for industry and firm characteristics including size, ownership, geographical location, market destination of the products and access to internal and external funds. Moreover, when investigating how firms finance the provision of trade credit, we find support for a supply-chain matching story: Firms are likely to depend on their own receipt of trade credit to finance the extension of trade credit and to match maturity between contract terms of payables and receivables. Specifically, we find very large and significant relationships between the decision to offer and the use of trade credit; the percentage of inputs and the percentage of sales financed by trade credit; the number of days extended to customers and the 2

4 maturity received from suppliers. This novel empirical result is robust to the inclusion of several product and market characteristics that proxy for the suitability of a product to receive trade credit in the up-stream and down-stream markets, and could provide an alternative explanation for the correlation between trade credit extended and received. Finally, we investigate in which circumstances firms are more likely to adopt a matching strategy. We identify two distinct patterns: First, this matching practice seems to be affected by the availability of internal and external funds. Firms with positive retained earnings and bank credit are less likely to rely on accounts payable. Second, this matching practice changes along the supply chain, i.e. it is more likely among firms that face stronger competition in the output market (firms that may need to offer trade credit as a competitive gesture) and enjoy stronger market power in the input market (firms that can demand favorable credit terms). This is a new finding that suggests a symbiotic relationship between a firm, its customers and its suppliers. Our paper contributes to the literature on trade credit along several dimensions. First, it provides an empirical test of theories that predict a relationship between the relative bargaining power between buyers and sellers and trade credit decisions. More specifically, our evidence provides support to the Wilner s (2000) argument that a customer obtains more trade credit if he generates a large percentage of the supplier s profits (i.e. the supplier s bargaining power is low). Our evidence is also in line with the prediction of Fisman and Raturi (2004) that trade credit is larger when the market of input suppliers is more competitive. Finally, Cun at (2007) provides an alternative hypothesis that trade credit is higher when suppliers and customers are in a sort of bilateral monopoly because of the high switching costs. In line with the bargaining power hypothesis, we find evidence that suppliers are less likely to offer trade credit when the good sold is tailored to the buyer specific needs. 3

5 Our analysis also complements and extends existing evidence. While empirical literature mainly focuses on the decision to extend trade credit and the links with the degree of competition in output markets (see among others McMillan and Woodruff, 1999; Fisman and Raturi, 2004, and Giannetti, Burkart and Ellingsen, 2011), we test the effect of competition on specific credit terms offered to customers, such as net days and early payment discounts and we document the relevance of competition also in the input markets for supply chain finance. Second, our matching story links our paper to the literature on working capital and risk management. The positive correlation between the decision to offer receivables and the decision to use payables is consistent with the recent focus of analysts on cash holdings, which may offer incentive to mangers to extend payment terms in order to maintain higher cash balances in working capital management (Bates, Kahle and Stulz, 2008). 1 At the same time, our evidence of a matching maturity strategy between the contract terms of payables and receivables supports the theoretical literature (Diamond, 1991 and Hart and Moore, 1991) and the empirical evidence (Guedes and Opler, 1996; Stohs and Mauer 1996; Demirguc-Kunt and Maksimovic 1999) on firm s matching maturity of assets and liabilities. We bring forward the idea that firms with no or limited access to internal resources and formal external financing match the maturity of assets and liabilities by using account payables as a risk management tool, i.e. to hedge their short-term receivables risk. Some suggestive evidence of a relationship between payables and receivables has been documented in other papers, but the lack of data on trade credit contracts simultaneously received and extended precluded a compelling analysis. Petersen and Rajan (1997) show that small and medium sized U.S. firms whose assets consist mainly of current assets demand 1 Furthermore, changes in inventory practices have led to firms holding more inventory (short-term assets), which would increase the demand for payables (short-term liabilities). 4

6 significantly more trade credit. Johnson, McMillan and Woodruff (2002) and Boissay and Gropp (2007) document that firms pass the delay of payments from customers to their suppliers. 2 We extend the previous literature by documenting that firms offer to their customers credit amounts and terms (e.g. number of days) that are similar to the ones they have been offered from their suppliers. Finally, our paper is related to the literature analyzing the role of trade credit in developing economies (McMillan and Woodruff, 1999; Fisman and Love, 2003; Johnson McMillan and Woodruff, 2004; Allen, Qian, and Qian, 2005, Cull, Xu and Zhu, 2007). The general idea in this literature is that trade credit can offer a viable substitute for formal bank credit in countries with low levels of financial development and/or in the early stages of transitioning towards a market based economy. For instance, Cull, Xu and Zhu (2007) test whether Chinese firms have indeed used trade credit as a substitute for the lack of formal institutions to stimulate growth, but they do not find supporting evidence. 3 Rather, they argue that the high growth rate in China is fueled by a growing private sector with increasing competitive pressure, which is likely to be an important motivation for the use of trade credit. This is in line with our results. Given that widespread competitive pressures across sectors is also a common feature of developed and market-based economies, our findings are likely not specific to China, but have more general implications. 2 Johnson, McMillan and Woodruff (2002) find that in European transition economies firms with a larger proportion of invoices paid by customers after delivery tend also to pay their suppliers late. Similarly, Boissay and Gropp (2007) document that French firms with defaulting customers are more likely to default on their suppliers. 3 They focus on differences in ownership structures, since in China it is not so much the lack of a formal financial system but rather its institutional bias in favor of state-owned enterprises that could give rise to trade credit among viable firms with restricted or no access to credit from state-owned banks. 5

7 The remainder of the paper is organized as follows. Section 2 provides the theoretical background and derives our testable hypotheses. Section 3 describes the data. Section 4 presents empirical results. Section 5 discusses some related issues and concludes. 2. Motivation and Hypotheses This section presents a conceptual framework and economic intuition for our hypotheses. Our reference unit is a firm buying input from its supplier (in the up-stream market) to produce goods to be sold to its customers (in the down-stream market). Our focus is on the firm s decision to offer trade credit to its customers (accounts receivable), having in mind that the same firm can itself take trade credit from its suppliers (accounts payable). 2.1 The Bargaining Power Hypothesis This hypothesis is motivated by previous theoretical literature showing that the decision to offer trade credit and the amount offered depend on the firm s relative bargaining power vis-àvis its customer. If a buyer depends too much on a seller, the seller is in a better position to force immediate payment. On the other hand, if a seller depends too much on a buyer, the buyer may be allowed to delay payment. For example, Wilner (2000) predicts higher trade credit when the customer s purchases count for a larger share of the firm s profits (i.e. the selling firm s bargaining power is low). The customer s relative bargaining power also depends on the degree of competition in the firm s product market. When the market is very competitive (and profit margins are low), the selling firm is in a weaker position to enforce payments and is therefore more willing to allow delayed repayments to attract new customers or to prevent existing customers from switching to a different supplier. The opposite is true when the supplier is a monopolist. A similar theory provided by Fisman and Raturi (2004) argues that when a market is 6

8 concentrated, customers have less incentive to take specific up-front investments to establish their creditworthiness, since they anticipate holdup problems in the future. It follows that trade credit is lower than in competitive markets where alternative suppliers are available. Another theory, related to the bargaining power hypothesis, but with an opposite prediction, is Cun at (2007). Motivated by high substitution costs, his model predicts higher levels of trade credit when competition is low (buyers and suppliers are in a sort of bilateral monopoly). According to Cun at (2007), intermediate goods tailored to the specific needs of the buyer make both suppliers and customers difficult to substitute. This theory has two implications for our analysis: First, it shows that product characteristics, like standardization, can affect the degree of effective competition faced by the supplier and, more generally, the bargaining power of the supplier vis-à-vis the customer. Second, in contrast to Wilner (2000) and Fisman and Raturi (2004), this theory predicts lower trade credit when competition is higher. Our empirical strategy aims to identify the effect of the firm s relative bargaining power on the decision to extend trade credit, the amount of trade credit offered and trade credit terms, by distinguishing between product sources of bargaining power, e.g. Cun at (2007), and market sources, e.g. Wilner (2000) and Fisman and Raturi (2004). We proceed by testing the following model: Account Receivable (AR) indicators = f {Firm characteristics, Bargaining Power (BP) indicators} (1) 2.2 The Supply Chain and Matching Maturity Hypotheses By offering credit to its customers, the firm postpones its receipt of cash payments and therefore needs to finance its own provision of credit. The firm can use internal resources, like retained earnings, or alternatively bank credit, if they have access to external finance. According 7

9 to previous literature (Frank and Maksimovic 2005, Giannetti, Burkart and Ellingsen, 2011), firms that are bank credit constrained should extend less trade credit. As an extension of this argument, credit constrained firms should extend shorter terms, as well as firms dependent on more costly financing, such as family and informal loans. Borrowing from the literature on the matching maturity of firm s asset and liabilities (Diamond, 1991 and Hart and Moore, 1991), we provide as an additional source of financing the "Supply Chain" hypothesis: Firms can finance their extension of trade credit (accounts receivables) with trade credit from their own suppliers (accounts payables). Trade credit can be an attractive source of funds for several reasons. First, trade credit is simple and convenient to use because it has low transaction costs. For example, in general no additional paperwork is required, as would be the case for a bank loan. Moreover, it is a flexible source of funds and can be used as needed. For instance, firms can pre-pay outstanding payables to suppliers if its customers remit early. An extension of the "Supply Chain" argument is that firms aim to match the maturity of their trade credit assets and liabilities (the Matching Maturity hypothesis). Once a firm has decided to use payables to finance its receivables, the firm must set specific credit terms. In principle, firms could try to extend the delay of payment of their inputs as far as possible. However this strategy can be expensive for the firm if a large discount for early payments has been offered or it can endanger the firm s supply chain if this request puts the supplier in financial distress. In both cases, a firm might have incentive to negotiate credit terms just long enough to finance its own credit extension, or equal to the maturity of its receivables. This would also allow the firm to eliminate the financial risk due to future changes in interest rates. This strategy implies that accounts payable can be used to optimally hedge receivables risk. We 8

10 empirically test the "Supply Chain" and Matching Maturity hypotheses, by estimating the following model: AR indicators = f {Firm characteristics, Financial characteristics, BP indicators, AP indicators} (2) 2.3 Interactive Effects Financial Constraints The availability of internal or external sources of funds is likely to affect the need to rely on the matching strategy. As an extreme case, payables might be the only source of funds available to a firm. We thus include the interaction between access to various sources of internal and external financing and accounts payable to test whether credit constrained firms or firms with less access to internal or external financing depend more on matching the terms of payables and receivables: AR indicators = f {Firm characteristics, Financial Characteristics, BP indicators, AP indicators, Interaction of Financial and AP indicators} (3) Bargaining Power As for the down-stream market, the ability of a firm to receive the desired amount and maturity of payables increases when its bargaining power vis-a-vis its supplier (up-steam market) is stronger. It follows that firms are more likely to match the terms of receivables with payables when they are in a strong position in the firm/supplier bargaining relationship, which again is more likely when the up-stream market is competitive. We therefore include the interaction terms of market structure and accounts payable to test whether firms operating in more competitive markets are more likely to match payables and receivables: AR indicators = f {Firm characteristics, BP indicators, AP indicators, Interaction of BP and AP indicators} (4) 9

11 3. Data and Summary Statistics We use firm-level data on about 2,500 Chinese firms, which was collected as part of the World Bank Enterprise Surveys conducted by the World Bank with partners in 76 developed and developing countries. 4 The dataset includes a large, randomly selected sample of firms across 12 two-digit manufacturing and service sectors. The surveys include both quantitative and qualitative information on barriers to growth, including sources of finance, regulatory burdens, innovations, access to infrastructure services, legal difficulties, and corruption. One limitation of the database is that only limited accounting (balance sheet) data is surveyed. In addition, in many countries (including China) the survey excludes firms with less than four years of age, in order to complete questions on firm performance and behavior relative to three-years earlier. We use the 2003 World Bank Enterprise Survey for China 5, which is the only country survey to include detailed questions on supply chain terms, as well as additional questions on the market environment, such as the number and importance of supplier and customer relationships. For the purpose of our analysis, the key questions regard the extension and terms of trade credit. Importantly, the survey asks both (i) whether firms offer trade credit to customers and (ii) whether customers accept trade credit from the firm. From our sample of 2,400 firms, 2,295 firms report whether or not they extend trade credit to their customers. 6 Table 1 shows variable names, definitions, and means for all variables. We include measures of trade credit, general firm characteristics, indicators of market power of the firm (relative to its customers and to its suppliers), financial characteristics, and indicators of the 4 The survey instrument and data are available at: 5 For additional information on the World Bank Enterprise Survey for China see Ayyagari, et al. (2010). 6 We also exclude from our sample 157 firms that provide financial services. 10

12 collateral value of goods sold and customer creditworthiness. Detailed summary statistics are shown in Table 2 (the full sample) and Table 3 (disaggregated by firms that do and do not offer trade credit). Table 4 shows a correlation matrix of our explanatory variables. 3.1 Trade credit variables Our main dependent variable is a dummy variable equal to one if the firm offers trade credit (accounts receivable), and zero otherwise (AR_d). We find that 39% of firms in our sample offer trade credit, and that the average percentage of goods sold on credit is 14% (AR_per); within the subsample of firms that offer trade credit, the average volume of credit extended represents about 35% of sales. On average, firms that extend trade credit offer customers about one month to pay: AR_days is a dummy variable equal to one if the firm offers a payment period equal or larger than 30 days, and zero otherwise. Finally, we find that 20% of firms that offer trade credit offer a prepayment discount on credit to their customers (AR_discount). Next, we examine firms use of trade credit from their own suppliers, accounts payable (AP). We find that 45% of firms use AP (AP_d), and the average percentage of supplies financed with credit is about 10% (AP_per); within the sample of firms that use AP, credit used equals about 20% of input purchases. 7 Similar to accounts receivable, the median term of payables is approximately one month (AP_days). We also find that about 7% of firms that use trade credit are offered a prepayment discount on credit from their suppliers (AP_discount). Table 3 shows that 62% of firms that extend trade credit to their customers receive credit from their suppliers, while only 34% of firms that do not extend credit use payables; this difference is 7 For the sample of firms with available balance sheet information, we confirm that firms using supplier financing report positive accounts payable and viceversa. We corrected AP_d in one case where the firm reported not purchasing inputs on credit but it had a positive value for accounts payable. 11

13 significant at 1%. Furthermore, a first glance at the data also shows significant differences in payment terms. For example, 36% of firms that extend trade credit to their customers are offered a payment period longer than 30 days before their own suppliers imposes penalties, while firms that do not extend credit receive shorter payment periods; this difference is significant at 1% Sample comparisons We perform a number of robustness tests to verify that our sample is nationally representative of China and we compare the use of trade credit in China with the one in developing and developed economies. First, we compare our sample to Cull et al. (2008), who use a large panel datasets of over 100,000 industrial Chinese firms and find that the average percentage of sales financed by accounts receivable in their sample of domestic private firms in 2003 is 18% (their dataset does not include additional contract information or information on accounts payables). This compares to 22% in our sample of firms (AR_per), which indicates that our random sample of Chinese firms is nationally representative. We also find that the use of trade credit in China is comparable to other countries. For example, using the complete World Bank Enterprise Survey database of over 100 developed and developing countries, we find that the average number of firms using trade credit for working capital or investment purposes (the only comparative variable available across countries) is 45% in China, and 51% for the complete sample. In addition, the average percentage of trade credit used for working capital purposes (averaged across firms that use trade credit) is 48% of total working capital financing in China, versus 44% for the complete sample. However, the use of trade credit in the U.S. is about two times the use in China. According to the Federal Reserve Survey of Small Business Financing (SSBF) database of U.S. 12

14 firms, about 65% of firms use trade credit to finance supplier purchases (20% more than in our sample of Chinese firms). In addition, the percentage of purchases made using trade credit is 20% among U.S. SMEs that use trade credit from suppliers (versus 10% in our sample of Chinese firms) and only 20% of firms that use trade credit are offered an early payment discount from their suppliers (in comparison to 7% in our sample of Chinese firms). 8 It is interesting to note that even in the U.S., about 80% of trade credit contracts do not include a pre-payment discount which suggests that trade credit might in fact be a relatively cheap source of financing across both developed and developing countries. We believe that our results shed light on trade credit behavior more broadly than the Chinese market. Unique features of the Chinese economy such as the bias towards stateowned banks and state-owned firms have been decreasing since 2001 (two years before our survey takes place) and we carefully address related potential biases. Moreover, there are no country-specific regulations on inter-firm financing. Finally, when we replicate our main results using data for Brazil, we find additional support for both the market power and the matching story found in China Firm and industry characteristics We include in all regressions some general firm characteristics, which are likely to be associated with trade credit. First, the log number of years since the firm was established (Age). Second, we use as a proxy for firm size the log number of total employees (including contractual employees) (Emp). All our empirical results are robust to using alternative measures of firm size, such as dummies indicating small, medium, and large firms. Likewise, we get similar results if 8 All U.S. data is cited from Giannetti, Burkart, and Ellingsen, Results available upon request. Unfortunately data for Brazil does not include as detailed information on supply chain contracts; therefore, we use data for Brazil as only a further robustness check for the evidence found in China. 13

15 we replace the log of total employment with the log of total sales, although we are less comfortable using accounting data because of the large number of missing observations and unaudited firms in our sample. Third, we include the log of the number of customers in the firm s main business line (N_customers), to control for any intrinsic relationship component of trade credit. This component would imply a mechanical relation between trade credit and number of customers that, if omitted, could bias our results Forth, we include a dummy variable equal to one if the percentage of the firm owned by foreign individuals, foreign investors, foreign firms, and foreign banks is greater than 50%, and equal to zero otherwise (Foreign). Fifth, we include a dummy variable equal to one if the percentage of the firm owned by the government (national, state, and local, and cooperative/collective enterprises) is greater than 50% (State). We include these ownership dummy variables to control for possible preferential access to financing from foreign and state-owned banks, respectively. It might also be the case that foreign and stateowned firms have preferential foreign and government product markets, respectively, and are not as sensitive to market competition. In our sample, 7% of firms are foreign owned, while 23% are state owned. All results are robust to using a 30% cutoff to identify foreign and state ownership. Sixth, we include a dummy variable equal to one if the firm sells its products abroad, and equal to zero otherwise (Export). We include this variable to control for possible differences in trade credit use among national and foreign customers. In our sample, 9% of firms are identified as exporters. We also include in all regressions 17 city dummies. Finally, we are concerned that trade credit patterns within supply chains might be driven by industry characteristics. For instance, there might be industry standards which set the percentage and terms of trade credit. If these standards are similar for the up-stream and downstream market or if firms have customers and suppliers in the same industry, then it is 14

16 likely that trade credit given and taken are correlated. To control for that, all regressions include 12 industry dummies that correspond to 2-digit NACE codes, which is the finest level of available sector classification. 10 Additional detailed information is available on the firm s main business line. However, this information is difficult to use in our regression analysis since it includes 1,818 descriptions and 99% of these classifications describe only one firm. Nevertheless, we studied the trade credit patters of firms within a few classifications with more than 10 firms and found no systematic patterns. For instance, 33 firms are classified as dress manufacturing. On average, 31% of firms extend trade credit and 53% of firms receive trade credit. For the 11 firms that extend trade credit to their customers, the percentage of sales offered ranges from 10% to 100% (the median is 20%); terms offered include 3, 4, 10, 30, 40, and 90 days; and three firms offer a discount. Examinations of additional narrow industry classifications found similar disparities across firms. 11 This exercise suggests that trade credit use and terms are not driven solely by product characteristics and further motivates our search for additional explanations. 3.3 Bargaining Power in the input and output markets Our next set of variables measures the bargaining power of the firm with respect to its customers. We distinguish between market and product sources of bargaining power. First, we measure the importance of the firm s largest customer with the percent of total sales that normally goes to the firm s largest customer (Sales_largest_cust). A higher value suggests that the firm s largest customer is important to its overall revenue and that the firm s bargaining power is weak, relative to its customers. 10 Similarly, Giannetti, et al. (2009) use 2-digit industry classifications to identify product specificity. 11 Additional industries available upon request. 15

17 Second, we measure the importance of the firm for its largest customer with the number of suppliers used by the firm s largest customer (No_supl_cust). If the customer uses many suppliers, it is less dependent on the firm i.e. ending the relationship poses less of a risk of a holdup problem and consequently less market power for the firm, relative to its customers. Third, we construct a Herfindal Index to measure the degree of competition in the domestic market for the main business line. Specifically, for the main business line we know the percentage of total sales in the domestic market supplied by the firm and by its main competitor and the total number of competitors in this market. Unfortunately this information is only available for a sub-sample of firms. Given that we do not know the market shares of all the competitors (but the main one), we assume that they are all equal. 12 This assumption implies that our continuous variable is an imprecise measure of domestic market concentration. As a result, we use a dummy equal to one if the Herfindal Index is larger than the median value and zero otherwise (Herfindal). A positive Herfindal implies a more concentrated market and therefore less competition. As a measure of bargaining power, we also include a dummy variable equal to one if the firm has introduced a new product (or service) or business line in the past year, assuming that this would require the firm to compete with a new product (New_product). Moreover, we proxy for broader changes in the competitive landscape with a dummy equal to one if on average, and relative to the average of the last year, the firm has lowered prices on its main business line, which we assume was done in response to greater competitive pressures in the market (Lowered_price). As shown in Table 3, in bivariate tests, firms operating in more competitive 12 The Herfindal Index (HI) is calculated as follows: HI =(x/100) 2 +(y/100) 2 +(Ncomp -1)*((100-x-y)/(100*(Ncomp- 1))) 2, where x and y are the firm and the main competitor market shares, respectively, and` Ncomp is the total number of competitors. 16

18 environments using all measures of market power are more likely to extend trade credit to customers. The relative bargaining power of the firm might also depend on the strength of the sellerbuyer relationship, which in turn can be related to product characteristics. n line with Cun at (2007), we measure the strength of the seller-buyer relationship with the percentage of sales made to the clients unique specification (Customized) or alternatively with the length of the relationship between the supplier and the customer (Age_rel_cust). The higher the percentage of sales of customized goods and the longer the trade relationship, the higher the switching costs for both supplier and customer and therefore the lower the competition (buyers and suppliers are in a sort of monopoly power). Finally, we measure the bargaining power in the input and output markets simultaneously. Our dataset includes unique qualitative and quantitative information on the bargaining power of manufacturing firms in relation to both their customers and their suppliers. First, we construct a dummy variable that measures the bargaining power of a firm relative to its suppliers (Main_customer), which is equal to one if the firm is the most important customer of its main supplier and zero otherwise. We compare this variable to the bargaining power of a firm relative to its customers, measured by whether the firm s largest customer accounts for more than 5% of sales (the median value). We also consider alternative thresholds like 10% or 30% of total sales. In our sample, 751 of 1,762 firms (43%) have weak bargaining power relative to their customers (Sales_largest_cust is larger than 5%), and 637 of 1,514 firms (42%) are in a strong position relative to their suppliers (Main_customer equals one). Firms with available information on both sides of the markets are 1,205: 29% of firms have weak bargaining power relative to their customers and strong market power relative to their suppliers; 37% of these firms have 17

19 weak bargaining power relative to both their customers and their suppliers; 14% of firms have strong bargaining power relative to their customers and strong market power relative to their suppliers; finally, 20% of these firms have strong bargaining power relative to their customers and weak market power relative to their suppliers. The correlation between the market power in the input and output markets is slightly negative ( ) but not significantly different from zero, suggesting that there is no correlation between the strength of the contractual position of a firm in their input and output markets. We use this information in two ways. First, we construct a new dummy variable (Bi_BP), to measure the bargaining power in the input and output markets simultaneously. As visualized in Fig. 1, we hypothesize that the most likely scenario for a firm to use its own payables to finance its receivables and match payment terms is in the case where a firm is in the strong position to demand trade credit from its suppliers, but must offer trade credit from a weak position relative to its customers ( Box 4 ). Hence, Bi_BP takes value equal to one if two conditions are satisfied: First, the proportion of total sales that normally goes to the firm s largest customer is greater than 5%, which indicates that the firm has weak bargaining power towards its customer (Sales_largest_cust is larger than 5%). Second, the firm is the most important customer of its main supplier, i.e. the firm has strong bargaining power towards its supplier. In the remaining cases ( Boxes 1, 2, and 3 ), the variable is assumed to be zero. Second, without assuming any particular order of the four markets in the table, we construct four different dummy variables: Bi_BP_d1 takes value equal to the one if the firm is located in Box 1 and zero otherwise; Bi_BP_d2 equal to one if located in Box 2 and zero otherwise; Bi_BP_d3 equal to one if located in Box 3 and zero otherwise, Bi_BP_d4 equal to one if located in Box 4 and zero otherwise. 18

20 Figure 1: Bargaining Power (BP) of the firm in up-stream and down-stream markets Low BP towards supplier Main_customer =0 High BP towards supplier Main_customer =1 High BP towards customer Sales_largest_cust<5% Low BP towards customer Sales_largest_cust>5% Box 1 Box 2 Box 3 Box Financial variables Next, we include various measures of financial liquidity. We use the percentage of unused line of credit, equal to zero if the firm does not have a line of bank credit (LC_unused), which is 7% on average. Less than 30% of firms have access to a line of credit from the banking sector, and on average, firms that have a line of credit have 26% unused. The low-level of financial access to formal credit market might suggest that many Chinese firms are credit constrained. We also include two dummy variables equal to one if the firm uses local or foreign bank financing (Bank) and family or informal credit (Fam/Informal). Finally, since we do not have balance sheet information on cash holdings, we measure the availability of internal resources by using a dummy variable equal to one if the firm uses retained earnings to finance working capital or investment (RE). Bivariate tests find that firms that extend trade credit are significantly more likely to have a larger unused line of credit and positive retained earnings. 4. Results Regressions are shown in Tables 5 to 9. All regressions control for general firm characteristics. Consistently, we find that larger firms are more likely to extend trade credit, which might be related to their longer and more established customer and supplier relationships. Moreover, younger firms are more likely to offer trade credit, which can be due to the fact that new firms face stronger competition when entering the product market. Our finding that larger 19

21 and younger firms offer more trade credit is not necessarily counterintuitive in a developing country context, where firms often remain small over time (and fail to grow as they age); for instance, the correlation between firm size and age in our sample is only 0.30 percent. 13 Firms are also more likely to offer trade credit the higher the number of customers. We find that firms where foreign shareholders (individuals or institutional investors or firms) have the majority of the firm equity have a higher probability to offer trade credit, while no relationship is found between foreign ownership and the amount of trade credit and credit terms. 14 There is also some evidence that exporting firms have a lower probability of offering trade credit; this might be explained by the difficulty in collecting late payments overseas or litigating in a foreign court. In general, however, we find no consistent significant relationship with state ownership. Table 5 Panel A shows that various measures of weaker market power have a positive and statistically significant effect on the decision to offer trade credit. For instance, the larger the percentage of sales to the largest customer, or the larger the number of suppliers of the firm s most important customer, or the lower the degree of concentration of the domestic market, the more likely are firms to extend trade credit. In addition, firms that have introduced new products or lowered prices in the past year are more likely to extend trade credit. The finding that firms introducing new products are more likely to extend trade credit could also be interpreted as evidence of the quality warranty theory, as suggested in Long, Malitz and Ravid (1993) and Klapper, Laeven, and Rajan (2011). Since a new product involves quality uncertainty, trade credit could be offered to give the buyer the possibility to check the product quality before the payment is due. Alternatively, in line with Lee and Stowe (1993), the 13 For additional discussion, see Klapper, et al. (2006), which shows that the relationship between age and size (measured by value added) is smaller in countries with weaker business environments. 14 If we identify foreign-owned firms using a 30% cutoff, we find a positive and significant relationship with the percentage of trade credit offered. 20

22 New_product variable could be interpreted as a proxy for complexity, suggesting that product risk and asymmetric information on the product market that increase with complexity could explain why firms offer trade credit. We can rule out these alternative interpretations by controlling for the percentage of sales carrying a warrantee (Warranty) and for the percentage of products that are certified (Certified). Both variables are likely to capture the need of quality checking and more generally the degree of asymmetric information between buyer and seller and its effect on trade credit. Results are reported in Table 5 columns (8) and (9) of Panel A. We find that New_product is still significant with the expected sign, while selling products with a warranty or selling certified products do not significantly affect the decision to offer trade credit. Panel B of Table 5 includes product characteristics as a determinant of trade credit use. According to Cun at (2007), suppliers extend more trade credit when intermediate goods are specific to the buyer s needs or when the seller-customer relationship is longer. In both cases, suppliers and customers are difficult to be substituted and thus create a sort of bilateral monopoly. First, we include a precise measure of product specificity - the percentage of sales of customized goods (Customized). 15 Column 1 shows that selling a larger proportion of customized goods does not have a significant effect on the decision to offer trade credit. The same result holds when N_customers is removed to take into account potential bias due to the correlation between the two variables (Column2). Selling customized goods could become relevant for trade credit decisions only when the percentage of customized goods is high enough, i.e. its effect could be non-linear. In Columns 3 and 4, we thus divide our sample in two sub-samples, using the percentage of sales of customized goods equal to 50% (corresponding to 40% of the 15 One could argue that the number of customers (N_customers), already included in our regressions, also measures the degree of product specialization/standardization. The idea would be that if a firm has only few customers, it might be providing a specific good or service to a larger customer. Under this interpretation, our previous results would suggest that firms supplying many goods (likely non-customized goods) face more competition and thus offer less trade credit, which is in line with our bargaining power hypothesis. 21

23 distribution) as a threshold. While in Column 3 the coefficient is not significant, in Column 4 it is statistically significant but with a negative sign: Firms selling customized products are less likely to offer trade credit, in line with our previous evidence and with our bargaining power hypothesis. Another possible explanation for the negative sign could be that the variable Customized captures the illiquidity of the good, rather than the lack of competition. If the good is made to the client s unique specification it can not easily resold in a secondary market. n that case, firms are less likely to offer trade credit, consistently with previous theories (Fabbri and Menichini, 2010) and evidence (Giannetti et al., 2011) on the role of collateral in supplier s lending. Next, we do not find evidence that longer seller-customer relationships affect trade credit supply (Column 5). Finally, one could argue that the significance of Sales_largest_cust is consistent with product specificity having a positive effect on trade credit supply, if more sales going to the largest customer implies more customized goods. Column 6 of Panel B shows that the effect of Sales_largest_cust still holds after controlling for the presence of customized goods and the length of the buyer-seller relationship. Table 6 investigates the effect of bargaining power on the percentage of sales financed with credit (AR_per) and on trade credit terms, including the payment period offered (AR_days) and the decision to offer a prepayment discount (AR_discount). Panel A uses the percentage of sales of the largest customer (Sales_largest_cust) as a measure of bargaining power and shows that when firms face an increase of competition in the product market, they allow customers to pay a larger share of sales on account, extend the payment period before imposing penalties, and are less likely to offer a prepayment discount, although this last effect is significant only at the 12 22

24 percent level. Panel B of Table 6 replicates the same analysis of Panel A using Lower_price as a measure of bargaining power. The results are qualitatively the same. Overall our findings are consistent with the idea that trade credit might be used as a competitive device to reduce actual competition or to prevent entry. In both cases, trade credit becomes crucial for the survival of the firm. A one standard deviation increment in product market competition (measured by Sales_largest_cust) increases the likelihood to offer trade credit (AR_dum) by about percentage points, which corresponds to 5% of the average likelihood. The same shock also increases the share of goods sold on credit (AR_per) by 2.7 percentage points, which corresponds to 19% of the average amount of trade credit offered. Both figures suggest that the economic effect of competition on trade credit supply is economically relevant. This would explain why even small firms without access to bank credit might still want to extend trade credit to their customers. Firms have different channels to finance the supply of trade credit, such as external financing (bank credit or informal sources), internal resources (retained earnings), or alternatively, credit from suppliers (accounts payable). Next, we examine the importance of a firm s access to finance from its own suppliers on its decision to extend credit to its customers, after controlling for other potential sources of accounts receivable financing. Table 7, Panels A and B, document the relevance of each source of financing. As shown in Panel A, unused bank credit lines (LC_unused) does not appear to have an effect on the likelihood to offer trade credit, the percentage of sales financed by trade credit, the length of the payment period, or the offer of a pre-payment discount. 16 We also include dummy variables indicating whether the firm uses retained earnings, bank financing, and informal or family sources of financing, and find 16 Note that the regressions using trade credit terms AR_days and AR_discount only include firms that offer trade credit (i.e AR_d=0). 23

25 consistently that even after including these variables, only accounts payable terms are significant. Our results are also robust to the inclusion of a dummy indicating positive profitability (not shown). 17 The most intriguing result of Table 7 is the robust finding that firms use accounts payable to finance the provision of accounts receivable and that firms match maturity of trade credit received from their own suppliers with the terms offered to their customers. We find very large and significant relationships between the decision to offer and the use of trade credit; the percentage of inputs purchased on credit and the percentage of goods sold on credit; and the number of days extended to customers and the ones received from suppliers. A discrete change in the decision to take trade credit (AP_dum) would double the firm s average likelihood to offer trade credit (AR_dum). A one standard deviation increment in the percentage of inputs purchased on credit (AP_per) more than doubles the average percentage of goods sold on credit. A discrete change in the likelihood to receive a payment period longer than 30 days from suppliers (AP_days) increases by 28% the firm likelihood to offer the same payment period to its customers (AR_days). These figures suggest that changes in the contract terms received by suppliers have a significant economic impact on the contract terms offered by firms to their customers. A possible concern about the findings in Table 7 could be that omitted product or market characteristics are likely to affect the suitability of a product for trade credit. If these characteristics are similar for the up-stream and down-stream markets, then trade credit given and taken are likely to be correlated. Our analysis already includes 12 sector dummies, 17 city dummies and a measure of bargaining power of the firm towards its largest customer (Sales_largest_cust). To address this issue further, we control for several product characteristics. 17 These results are also robust to the exclusion of accounts payable terms. 24

26 First, we add the percentage of customized goods (Customized). This variable could also be interpreted as a proxy for liquidity. In fact, a customized good is not easily resalable in a secondary market. Second, we control for the length of the buyer-seller relation (Age_rel_cust), and third for the easiness of ensuring quality (Warrantee). The results are shown in Panel C of Table 7: the matching story still holds. Although the decision to extend trade credit is not significantly dependent on the availability of internal and external financing, we do find that less liquid firms are significantly more likely to extend trade credit if they receive credit from their own suppliers. Table 8 shows our results. We find a significant effect of the interaction of the percentage of goods sold on credit and our dummy indicating the use of bank financing, suggesting that firms with access to bank financing rely less on accounts payable to finance the provision of trade credit to their own customers. Moreover, the interaction term between the use of account payable (AP_d) and the bank dummy is significant at 13%. Table 8 also includes the interaction of accounts payable terms and dummies indicating the use of retained earnings as a source of financing. Also, in this case, we find some interesting and significant terms: the coefficient on the interaction between the use of account payable (AP_d) and retained earnings (RE) is significantly negative, suggesting that firms with positive retained earnings rely less on accounts payable to finance the provision of trade credit to their own customers. All our results are robust to the inclusion of a measure of bargaining power and several product characteristics, as shown in column (2), (4), (6), (8) of Table 8. Similar results also hold if we look at trade credit terms (maturity and prepayment discounts, not shown). Finally, Table 9 Panel A, shows that the extent to which firms match accounts payable with accounts receivable depends on the bargaining power enjoyed simultaneously in the input 25

27 and output markets. In particular, the coefficients of the interaction term between our index of bargaining power in both markets - Bi_BP - and the decision to offer trade credit or the share of sales financed by trade credit AR_dum and AR_per are positive and statistically significant. These findings suggest that our supply chain hypothesis is most likely to hold when firms need to offer trade credit to their customers (as a competitive gesture), but have enough bargaining power with suppliers to set their own credit conditions. 18 Similar results are obtained when we control for product characteristics (see columns (3) and (4) of Panel A). One possible concern could be that this result is driven by the definition of the index of bargaining power. To address this issue we proceed in two ways. First, we use alternative definitions of the index (Bi_BP). While in columns (1)-(4) of Panel A, the threshold used to define firms with low bargaining power in the product market is Sales_largest_cust equal or larger than 5%, we use the 10% threshold in columns (5) and (6), and the 30% in the last two columns. The coefficient of the interaction terms retains a positive sign and is statistically significance in all specifications, with the significance increasing in the thresholds used. As a further robustness check, we substitute our index with a list of four dummies capturing firms located in each one of the four different market structures represented in Fig. 1. The dummy Bi_BP_d4 captures firms with low bargaining power in the down-stream market and high bargaining power in the up-stream market. We expect these firms to be more likely to match payables with receivables than firms in the other three markets. Results are shown in Panel B of Table 9. Columns (1) and (2) include the four dummies (Bi_BP_d1 is the default dummy) without the interaction terms. None of the dummies is significant at the standard levels. When we include the interactions terms (columns (3) and (4)), we find that they are all positive and 18 Notice that the variable Bi-BP is only available for the sub-sample of manufacturing firms. All results for all tables are robust for the subsample of manufacturing firms. 26

28 statistically significant, suggesting that firms are more likely to match payables with receivables when moving from market 1 to any of other three markets. However, the coefficient of the interaction terms with market 4 - AP_d*Bi_BP_d4 and AP_per*Bi_BP_d4 - has the highest economic and statistic significance. This finding suggests that firms with low bargaining power in the down-stream market and high in the up-stream one are the most likely to match maturities. This evidence is also robust to the inclusion of product characteristics (result not reported). Finally, we perform a series of important robustness tests. We find that our results still hold if we restrict the sample to only manufacturing firms, profitable firms (Profit_d=1), firms with state ownership (national, state, and local, and cooperative/collective enterprises) lower than 50% (State=0), or to non-exporter firms (Export=0). Our results are also robust to the inclusion of a dummy variable equal to one if the firm belongs to a government sponsored industrial park, science park, or Export Promotion Zone (EPZ). 19 Next, although our data set generally provides only cross-sectional firm-level data for one year, some questions in the survey refer to past firm activity. These questions allow us to control for some changes in firm policy that occurred in the past. For example, the variable New_product included in our regressions reflects whether a firm has introduced new products (or services) in the past year and may proxy for changes in the firm s investment policy. n addition, limited accounting information is available, both for the current and previous years. We use this information to control for potential idiosyncratic shocks at the firm level. For example, we construct a set of dummy variables to control for whether the firm increased sales or fixed assets in the past three years. These dummies are insignificant and do not affect our main results. 19 Regression results available upon request. 27

29 5. Discussion and Conclusion The evidence presented in the previous section raises some important questions. For example, why do we find, contrary to previous literature, that credit constraints (LC_unused) do not seem to affect the decision to offer trade credit either directly or indirectly through the interactions with accounts payable? A possible explanation could be that the percentage of unused credit lines does not necessarily capture the tightness of credit constraints. Firms are required to pay fees on the proportion of unused credit lines and therefore have incentive to reduce the unused portion. It follows that a fully used credit line does not necessarily identify a credit constrained firm. However, when instead we include a dummy indicating the availability of bank credit (both from local and foreign commercial banks), we find that firms with external bank finance rely less on accounts payable to finance the provision of trade credit, but again no direct and significant effect is found for the decision to extend credit or the amount of trade credit offered. Another possible explanation could be that trade credit is not necessarily more expensive than bank credit. The central question then becomes how costly trade credit is, relative to bank financing. Conventional wisdom is that trade credit is primarily a financing of last resort for firms that have exhausted or unable to access bank credit. However, some more recent papers challenge this view. For example, Giannetti, Burkart and Ellingsen (2011) document that a majority of the U.S. and European firms in their sample appear to receive cheap trade credit. Similarly, Miwa and Ramseyer (2005) argue that there is no evidence that sellers use extravagant cash discounts in Japan, while Marotta (2005) document that trade credit provided by Italian manufacturing firms is only slightly, if at all, more expensive than bank credit. Our findings are in line with this recent evidence. 28

30 Our result on the role of market competition raises another question: If competition is a main driver behind the decision to offer trade credit, why do firms not simply reduce the product price instead of offering a trade credit discount? A price reduction is not necessarily simpler than offering trade credit. A price reduction is observable by competitors and can trigger their immediate reaction by causing a price war that can be detrimental for the full industry. Trade credit is a less aggressive and more flexible instrument. The two instruments price reduction and supply of trade credit - could also be used as complement strategies. The finding in Table 5 - firms which lowered their prices relative to the previous year are more likely to offer trade credit - could also be interpreted along this line. To conclude, this paper uses firm-level data on about 2,500 Chinese firms to study the decision to extend trade credit. Supplier financing is often overlooked in the capital structure literature, although it is arguably the most important source of financing for small and medium sized enterprises particularly in countries with less developed financial and information systems. We show that firms use trade credit as a competitive gesture. More specifically, firms with weaker bargaining power vis-à-vis their customers, for example because of a highly competitive output market, are more likely to offer trade credit and better credit terms. Moreover, firms are likely to depend on credit from their own suppliers to finance the extension of trade credit to their customers and to match credit terms between accounts payable and receivable. This matching practice is largely used by illiquid firms and by firms facing high competition in both the output market (firms need to extend trade credit to win competitors) and in the input market (firms can get favourable terms from their suppliers). This evidence highlights the importance to go beyond a simple bilateral supplier-customer relationship towards a theory of trade credit where each firm is part of a supply chain financing. 29

31 References: Ayyagari, Meghana, Asli Demirguc-Kunt, and Voijslav Maksimovic, Formal versus Informal Finance: Evidence from China. Review of Financial Studies 23:8. Allen, Franklin, Jun Qian, and Meijaun Quian, Law, Finance and Economic Growth in China. Journal of Financial Economics 77, Bates, Thomas, Kathleen Kahle, and Rene Stulz, 2009, Why do U.S. Firms Hold so Much More Cash than They Used to? Journal of Finance 64:5, Boissay, Frederic and Reint Gropp, Trade Credit Defaults and Liquidity Provision by Firms, European Central Bank Working Paper n Brennan, Michael, Vojislav Maksimovic and Josef Zechner, Vendor Financing. Journal of Finance 43, Cull, Robert, Lixin Colin Xu, and Tian Zhu, Formal Finance and Trade Credit during China s Transition. Journal of Financial Intermediation. Cun at, Vicente, Trade Credit: Suppliers and Debt Collectors as nsurance Providers, Review of Financial Studies, 20. Demirguc-Kunt, Asli, and Vojislav Maksimovic, Institutions, Financial Markets, and Firms Debt Maturity. Journal of Financial Economics 54, Demirguc-Kunt, Asli, and Vojislav Maksimovic, Firms as Financial Intermediaries: Evidence from Trade Credit Data, World Bank Working Paper. Diamond D., Debt Maturity Structure and Liquidity Risk. Quarterly Journal of Economics, Fabbri, Daniela and Annamaria Menichini, Trade Credit, Collateral Liquidation and Borrowing Constraints. Journal of Financial Economics, 96, Fisman, Raymond and Inessa Love, Trade Credit, Financial Intermediary Development, and Industry Growth, Journal of Finance 58, Fisman, R., Raturi, M., Does Competition Encourage Credit Provision? Evidence from African Trade Credit Relationships. Review of Economics and Statistics 86, Frank, Murray and Vojislav Maksimovic, Trade credit, Collateral, and Adverse Selection. University of Maryland Working Paper. Giannetti, Mariassunta, Burkart, Mike and Tore Ellingsen, What You Sell is What You Lend? Explaining Trade Credit Contracts. Review of Financial Studies, 24,

32 Guedes, Jose and Tim Opler, The determinants of the Maturity of Corporate Debt Issues. The Journal of Finance 51, Hart, Oliver and John Moore, A Theory of Debt Based on the Inalienability of Human Capital. The Quarterly Journal of Economics 109, Johnson, Simon, John McMillan and Christopher Woodruff, Courts and Relational Contracts. Journal of Law, Economics and Organization18, Klapper, Leora, Luc Laeven, and Raghu Rajan, Financial Studies, Forthcoming. Trade Credit Contracts. Review of Lee, Yul W. and John D. Stowe (1993), Product Risk, Asymmetric nformation, and Trade Credit, Journal of Financial and Quantitative Analysis 28, Long, Michael S., Ileen B. Malitz and S. Abraham Ravid, Trade Credit, Quality Guarantees, and Product Marketability. Financial Management 22, Fisman, Raymond and Inessa Love, Trade Credit, Financial Intermediary Development and Industry Growth. Journal of Finance 58(1), Marotta, Giuseppe, Is Trade Credit More Expensive than Bank Credit Loans? Evidence from Italian Firm-Level Data. Applied Economics 37(4), McMillan, John and Christopher Woodruff, Interfirm Relationships and Informal Credit in Vietnam, The Quarterly Journal of Economics 114, Miwa, Yoshiro and J. Mark Ramseyer, Trade Credit, Bank Loan and Monitoring: Evidence from Japan. Working Paper. Petersen, Mitchell A. and Raghuram G. Rajan, Trade Credit: Theory and Evidence. Review of Financial Studies, 10 (3), Stohs, M., and Mauer, D., The Determinants of Corporate Debt Maturity Structure. Journal of Business, 69, Wilner, Benjamin S The Exploitation of Relationships in Financial Distress: The Case of Trade. Credit. The Journal of Finance 55,

33 Table 1: Variable Definitions and Mean Statistics Variable Name Definition Mean Measures of Trade Credit AR_dum Dummy (0/1), =1 if the firm offers credit to its customers (i.e. accounts receivable), =0 if the firm does not offer trade credit 0.39 AR_per The percent of monthly sales sold on credit AR_days Dummy (0/1), =1 if the firm offers a number of days larger than 30 (median) before imposing penalties to its customers, = otherwise (and =. if AR_dum is =0) AR_discount Dummy (0/1), =1 if the firm offers a pre-payment discount on credit to its customers, =0 otherwise (and =. if AR_dum is =0) 0.20 AP_dum Dummy (0/1), =1 if the firm uses supplier credit (i.e. accounts payable) to purchase inputs, =0 otherwise 0.45 AP_per The percent of inputs purchased on credit (based on period averages), = 0 if the firm does not use trade credit 9.58 AP_days Dummy (0/1), =1 if the number of days the firm is allowed to use the credit before its suppliers impose penalties is larger than (median). AP_discount Dummy (0/1), =1 if the firm received a pre-payment discount on credit from its suppliers, =0 otherwise (and =. if AP_dum is =0) 0.07 General Firm Characteristics Age Log number of years (+1) since the firm was established 2.57 Emp Log average number of total employees (including contractual employees) 4.94 Foreign Dummy (0/1), =1 if the percentage of the firm owned by foreign individuals, foreign institutional investors, foreign firms, and 0.07 foreign banks is greater than 50, =0 otherwise State Dummy (0/1), =1 if the percentage of the firm owned by the government (federal, state, local, and collective/cooperative 0.23 enterprises) is greater than 50, =0 otherwise Export Dummy (0/1), =1 if the firm is exporting, =0 otherwise 0.09 N_customers Log number of customers the firm supplies its goods

34 Indicators of (Weaker) Bargaining Power of the Seller (relative to its Customers) Sales_largest_cust Percent of total sales that normally goes to the firm s largest customer No_supl_cust Number of suppliers used by the firm s largest customer 9.49 Herfindal Dummy (0/1), =1 if the Herfindal Index in the domestic market for the main business line is greater than 0.21 (the median), = otherwise. New_product Dummy (0/1), =1 if the firm has introduced new products (or services) in the past year, =0 otherwise 0.42 Lowered_price Dummy (0/1), =1 if on average, and relative to the average of the last year, the firm has lowered prices on its main business line, = otherwise Customized The percent of sales made to clients unique specification (i.e. that cannot be sold to other clients) Age_rel_customer On average, how many years the firm has done business with its clients in its main business line Bi_BP Dummy (0/1), =1 if the percent of total sales that normally goes to the firm s largest customer is greater than 5% (i.e Sales_largest_cust>5%), and the firm is the most important customer of its largest supplier s, =0 otherwise LC_unused Bank RE Fam/Informal Financial Characteristics The percent of the firm s line of credit or overdraft facility that is currently unused (=0 if the firm does not have a line of credit or overdraft facility) Dummy (0/1), = 1 if the firm uses local or foreign bank financing for working capital or investment Dummy (0/1), = 1 if the firm uses retained earnings for working capital or investment Dummy (0/1), = 1 if the firm uses family or informal financing for working capital or investment Other Firm Characteristics Warrantee The percentage of sales carrying a warrantee 90.7 Certified The percent of products that are certified

35 Table 2: Summary Statistics See Table 1 for variable definitions. Variable Name Obs. Mean Std. Dev. Min Max AR_dum 2, AR_per 2, AR_days AR_discount AP_dum 2, AP_per 2, AP_days AP_discount Age 2, Emp 2, Foreign 2, State 2, Export 2, N_customer 1, Sales_largest_cust 1, No_supl_cust 1, Herfindal Lower_price 2, New_product 2, Customized 2, Age_rel_customer 2, Bi_BP 2, LC_unused 2, Bank 1, RE 1, Fam/Informal 1, Warrantee 2, Certified 2,

36 Table 3: Mean Differences, by Trade Credit Supply See Table 1 for variable definitions. t-statistics show the mean difference of firms that offer trade credit to customers versus firms that do not offer trade credit to customers. ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. Variable Name AR_dum=0 AR_dum=1 Sig. AP_dum *** AP_per *** AP_days *** AP_discount ** Age *** Emp *** Foreign *** State *** Export * N_customer *** Sales_largest_cust *** No_supl_cust *** Herfindal *** New_product *** Lower_price *** Customized Age_rel_cust *** Bi_BP *** LC_unused *** Bank RE *** Fam/Informal Warrantee *** Certified *** 35

37

38 Table 4: Correlation Matrix See Table 1 for variable definitions ***, ** and * indicate significance at the 1%, 5%, and 10% levels, respectively. Panel A: Explanatory Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) AP_dum(1) 1.00 AP_per (2) 0.51 *** 1.00 AP_days (3) *** 1.00 AP_discount (4) ** 1.00 Age (5) *** Emp (6) 0.11 *** 0.14 *** 0.16 *** *** 1.00 Foreign (7) 0.05 ** 0.12 *** *** 0.09 *** 1.00 State (8) *** *** *** 0.21 *** *** 1.00 Export (9) 0.04 * 0.07 *** *** 0.21 *** 0.36 *** *** 1.00 N_customer (10) 0.08 *** *** ** *** 1.00 Saleslargest (11) * 0.09 *** 0.13 *** ** *** *** 0.15 *** *** 1.00 No_supl_cust (12) * * 0.18 *** *** 1.00 Herfindal (13) ** * *** *** 1.00 New_product (14) 0.12 *** 0.14 *** 0.15 *** *** 0.23 *** *** 0.19 *** ** 1.00 Lower_price (15) 0.09 *** 0.11 *** 0.15 *** *** 0.06 *** ** 0.06 *** 0.06 ** 0.10 *** *** 0.22 *** 1.00 Customized (16) *** 0.09 * *** ** 0.12 *** *** 0.15 *** *** 0.34 *** *** *** 1.00 Bi_BP (17) 0.08 *** 0.16 *** 0.15 ** *** 0.07 *** * 0.11 *** ** 0.27 *** ** *** 0.15 *** 0.07 *** 1.00 LC_unused (18) 0.10 *** 0.16 *** 0.09 ** *** 0.08 *** *** 0.06 *** ** 0.16 *** 0.08 *** *** 1.00 Panel B: Dependent Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) AR_dum 0.27 *** 0.30 *** 0.23 *** 0.08 ** *** 0.09 *** 0.08 *** *** 0.04 * 0.13 *** *** 0.15 *** 0.18 *** *** 0.07 *** AR_per 0.23 *** 0.35 *** 0.20 *** *** 0.10 *** 0.09 *** *** 0.12 *** *** ** 0.14 *** 0.17 *** *** 0.08 *** AR_days 0.09 *** 0.07 * 0.14 ** *** *** * ** 0.16 *** 0.08 * ** AR_discount 0.06 * *** ** ***

39 Table 5: Trade Credit Supply and the Firm Bargaining Power (Panel A) The reported estimates are from logit regressions. AR_dum is a dummy indicating the use of accounts receivable. See Table 1 for variable definitions. All regressions include 12 sector dummies and 17 city dummies. Robust p-values are shown in parentheses, ***, **, * indicate significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) (4) (5) (6) (7) (8) (9) Sales<10% Sales>=10% AR_dum AR_dum AR_dum AR_dum AR_dum AR_dum AR_dum AR_dum AR_dum Age [0.02]*** [0.42] [0.31] [0.04]** [0.11] [0.16] [0.12] [0.16] [0.30] Emp [0.06]* [0.64] [0.23] [0.27] [0.14] [0.13] [0.27] [0.36] [0.33] Foreign [0.06]* [0.01]*** [0.79] [0.18] [0.02]** [0.03]** [0.02]** [0.03]** State [0.83] [0.79] [0.66] [0.72] [0.52] [0.91] [0.76] [0.79] [0.94] Export [0.11] [0.07]* [0.66] [0.11] [0.60] [0.04]** [0.06]* [0.06]* [0.07]* N_customers [0.32] 0.06 [0.32] 0.14 [0.03]** Sales_largest_cust 0.08 [0.06]* N_sup_cust 0.11 [0.09]* [0.32] 0.16 [0.08]* Herfindal [0.02]** Lower_price 0.53 New_product 0.21 [0.06]* 0.21 [0.06]* Warrantee [0.66] Certified 0.20 [0.09]* 0.4e-03 [0.76] Constant [0.03]** [0.03]** Observations 1,446 1, ,713 1,721 1,701 1,610 Pseudo R-squared

40 Table 5: Trade Credit Supply and the Firm Bargaining Power (Panel B) The reported estimates are from logit regressions. AR_dum is a dummy indicating the use of accounts receivable. See Table 1 for variable definitions. All regressions include 12 sector dummies and 17 city dummies. Robust p-values are shown in parentheses, ***, **, * indicate significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) Customized < 50% (4) Customized >=50% (5) (6) AR_dum AR_dum AR_dum AR_dum AR_dum AR_dum Age [0.22] [0.01]*** [0.23] [0.54] [0.08]* [0.05]** Emp [0.39] [0.01]** [0.51] [0.77] [0.14] [0.16] Foreign [0.02]** [0.06]* [0.31] [0.08]* [0.03]** [0.05]** State [0.85] [0.71] [0.70] [0.36] [0.78] [0.86] Export [0.07]* [0.06]* [0.45] [0.09] [0.06]* [0.10]* N_customers [0.07]* 0.14 [0.00]** 0.16 Customized 0.00 [0.62] [0.60] 0.00 [0.65] [0.08]* 0.00 [0.25] Age_rel_cust [0.61] 0.04 [0.45] Sales_largest_cust 0.07 [0.07]* Constant [0.30] Observations 1,653 1, ,011 1,714 1,383 Pseudo R-squared

41 Table 6: Trade Credit Supply and the Firm Bargaining Power (Panel A) The reported estimates are from OLS regressions in columns (1) and (2) and from logit regressions in columns (3)-(6). AR_per is the percent of monthly sales sold on credit. AR_days is a dummy indicating if the firm offers to its customers a number of days larger than 30 before imposing penalties. AR_discount is a dummy indicating if the firm offers a pre-payment discount on credit to its customers. See Table 1 for variable definitions. All regressions include 12 sector dummies and 17 city dummies. Robust p-values are shown in parentheses, ***, **, * indicate significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) (4) (5) (6) AR_per AR_per AR_days AR_days AR_discount AR_discount Age [0.03]** [0.02]** [0.11] [0.15] [0.92] [0.86] Emp [0.01]*** [0.08]* [0.06]* [0.19] [0.17] Foreign [0.41] [0.44] [0.32] [0.43] [0.62] [0.60] State [0.23] [0.21] [0.75] [0.73] [0.40] [0.43] Export [0.83] [0.88] [0.94] [0.89] [0.66] 0.64 N_customers 0.39 [0.35] 0.36 [0.85] [0.58] [0.49] 0.06 [0.45] 0.08 [0.43] Sales_largest_cust e e e e-03 [0.13] [0.09]* [0.12] [0.12] Customized 0.09 [0.88] [0.03]** 0.17 [0.13] Age_relation_cust e-04 [0.07]* [0.64] [0.99] Constant [0.95] [0.56] [0.41] 0.22 [0.86] Observations 1,459 1, Pseudo R-squared

42 Table 6: Trade Credit Supply and the Firm Bargaining Power (Panel B) The reported estimates are from OLS regressions in columns (1) and from logit regressions in columns (2 and (3). AR_per is the percent of monthly sales sold on credit. AR_days is a dummy indicating if the firm offers to its customers a number of days larger than 30 before imposing penalties. AR_discount is a dummy indicating if the firm offers a pre-payment discount on credit to its customers. See Table 1 for variable definitions. All regressions include 12 sector dummies and 17 city dummies. Robust p-values are shown in parentheses, ***, **, * indicate significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) AR_per AR_days AR_discount Age [0.11] [0.17] [0.72] Emp [0.02]** [0.14] [0.23] Foreign [0.20] [0.62] [0.88] State [0.15] [0.32] [0.43] Export [0.44] [0.16] [0.34] N_customers 0.04 [0.92] [0.17] 0.08 [0.19] Lower_price [0.12] [0.52] Constant 0.31 [0.95] Observations 1, Pseudo R-squared

43 Table 7: Trade Credit Demand and Supply (Panel A) The reported estimates are from logit in Columns (1), (3), (4), (5), (7) and (8), and OLS in Columns (2) and (6). AR_dum is a dummy indicating the use of accounts receivable. AR_per is the percent of monthly sales sold on credit. AR_days is a dummy indicating if the firm offers to its customers a number of days larger than 30 before imposing penalties. AR_discount is a dummy indicating if the firm offers a prepayment discount on credit to its customers. Columns (3-4) and (7-8) include only the sub-sample of firms that use accounts receivable. See Table 1 for variable definitions. All regressions include 12 sector dummies and 17 city dummies. Robust p-values are shown in parentheses, ***, **, * indicate significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) (4) (5) (6) (7) (8) Unused Line of Credit (LC_Unused) Bank Financing (Bank) AR_dum AR_per AR_days AR_discount AR_dum AR_per AR_days AR_discount L_Age [0.07]* [0.04]** [0.23] [0.78 [0.00** [0.17] [0.18] [0.30] L_emp [0.42] [0.08]* [0.18] [0.24 [0.42 [0.62] [0.07]* [0.16] Foreign [0.05]** [0.80] [0.77] [0.89 [0.13] [0.89 [0.60] [0.54] State [0.00] [0.34] [0.38] [0.28] [0.91 [0.19 [0.20] [0.77] Export [0.27] [0.61] [0.90] [0.37 [0.08]* [0.63] [0.73] [0.77] N_customers [0.39] [0.36] [0.25 [0.50] [0.58] [0.58] Sales_largest_cust [0.12] [0.01]*** [0.05]** [0.07]* [0.04]** [0.00** [0.16] LC_unused 9.55e [0.99] [0.86] [0.14] [0.51 Bank [0.39] [0.59] [0.84] [0.12] AP_dum AP_per AP_days [0.06]* [0.10]* AP_discount 0.61 [0.29] 0.52 [0.43] Constant [0.93] [0.66] [0.01]*** [0.25] [0.38] [0.38] Observations 1,360 1, Pseudo R-squared

44 Table 7: Trade Credit Demand and Supply (Panel B) The reported estimates are from logit in Columns (1), (3), (4), (5), (7) and (8), and OLS in Columns (2) and (6). AR_dum is a dummy indicating the use of accounts receivable. AR_per is the percent of monthly sales sold on credit. AR_days is a dummy indicating if the firm offers to its customers a number of days larger than 30 before imposing penalties. AR_discount is a dummy indicating if the firm offers a prepayment discount on credit to its customers. Columns (3-4) and (7-8) include only the sub-sample of firms that use accounts receivable. See Table 1 for variable definitions. All regressions include 12 sector dummies and 17 city dummies. Robust p-values are shown in parentheses, ***, **, * indicate significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) (4) (5) (6) (7) (8) Retained Earnings (RE) Family & Informal Financing (Fam_Inf) AR_dum AR_per AR_days AR_discount AR_dum AR_per AR_days AR_discount L_Age [0.06* [0.15] [0.09]* [0.21] [0.02]** [0.13] [0.12] [0.33] L_emp [0.78 [0.42] [0.13] [0.23] [0.55] [0.25] [0.09]* [0.25] Foreign [0.06]* [0.40] [0.36] [0.66] [0.17] [0.73] [0.60] [0.56] State [0.59 [0.76] [0.09]* [0.50] [0.96] [0.44] [0.08]** [0.42] Export [0.31 [0.75] [0.63] [0.93] [0.18] [0.87] [0.58] [0.98] N_customers [0.44] [0.20] [0.29] [0.39] [0.30] [0.26] Sales_largest_cust [0.05]** [0.01]*** [0.39] [0.15] [0.01]** [0.67] RE [0.39 [0.36] [0.47] [0.17] Fam/Informal [0.22] [0.91] [0.90] [0.40] AP_dum AP_per AP_days [0.05]** [0.10]* AP_discount [0.12] [0.31] Constant [0.98] [0.87] [0.99] [0.90] Observations Pseudo R-squared

45 Table 7: Trade Credit Demand and Supply (Panel C) The reported estimates are from logit in Columns (1), (3), (4), (5), (7) and (8), and OLS in Columns (2) and (6). AR_dum is a dummy indicating the use of accounts receivable. AR_per is the percent of monthly sales sold on credit. AR_days is a dummy indicating if the firm offers to its customers a number of days larger than 30 before imposing penalties. AR_discount is a dummy indicating if the firm offers a prepayment discount on credit to its customers. Columns (3-4) and (7-8) include only the sub-sample of firms that use accounts receivable. See Table 1 for variable definitions. All regressions include 12 sector dummies and 17 city dummies. Robust p-values are shown in parentheses, ***, **, * indicate significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) (4) (5) (6) (7) (8) Bank Financing (Bank) Retained Earnings (RE) AR_dum AR_per AR_days AR_discount AR_dum AR_per AR_days AR_discount L_Age [0.01]*** [0.09]* [0.18] [0.35] [0.06]* [0.15] [0.12] [38] L_emp [0.50] [0.36] [0.08]* [0.15] [0.88] [0.51] [0.11] [0.25] Foreign [0.21] [0.91] [0.69] [0.57] [0.11] [0.44] [0.55] [0.86] State [0.86] [0.21] [0.21] [0.91] [0.52] [0.74] [0.06] [0.57] Export [0.08]* [0.59] [0.80] [0.88] [0.31] [0.71] [0.68] [0.95] N_customers [0.54] 0.05 [0.63] 0.14 [0.27] [0.44] 0.05 [0.57] 0.29 [0.03]** Sales_largest_cust 0.01 [0.03]** [0.01]*** [0.20] 0.11 [0.07]* 1.47 [0.01** 0.27 [0.02]** [0.52] Customized 0.09 [0.25] 0.53 [0.50] [0.37] 0.32 [0.04]** 0.10 [0.19] 0.35 [0.65] [0.26] 0.33 [0.07]* Age_relation_cust 0.04 [0.56] 0.87 [0.18] [0.89] 0.00 [0.98] [0.21] [0.94] 0.09 [0.59] Warrantee 7.72e-04 [0.86] 0.02 [0.69] 0.00 [0.66] [0.32] [0.63] [0.93] 0.01 [0.54] [0.18] Bank [0.54] [0.66] [0.97] [0.12] RE [0.40] [0.36] [0.45] [0.35] AP_dum AP_per AP_days [0.08]* [0.05]** AP_discount [0.41] [0.14] Constant [0.41] [0.56] [0.65] [0.58] [0.83] [0.96] Observations Pseudo R-squared

46 Table 8: The Financing of Trade Credit Supply Is Trade Credit Demand More Important for Less Liquid Firms? The reported estimates are from logit (Colums 1, 2, 5, 6) and OLS (Columns 3, 4, 7, 8) regressions. AR_dum is a dummy indicating the use of accounts receivable and AR_per is the percent of monthly sales sold on credit. See Table 1 for variable definitions. All regressions include 12 sector dummies and 17 city dummies. Robust p-values are shown in parentheses, ***, **, * indicate significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) (4) (5) (6) (7) (8) Bank Financing (Bank) Retained Earnings (RE) AR_dum AR_dum AR_per AR_per AR_d AR_d AR_per AR_per L_age [0.01]*** [0.02]** [0.12] [0.13] [0.06]* [0.07]* [0.16] [0.15] L_emp [0.42] [0.47] [0.30] [0.39] [0.75] [0.83] [0.43] [0.53] Foreign [0.11] [0.13] [0.73] [0.79] [0.06]* [0.08]* [0.38] [0.42] State -3.85e [0.99] [0.95] [0.12] [0.12] [0.68] [0.68] [0.74] [0.72] Export [0.07]* [0.07]* [0.56] [0.52] [0.35] [0.34] [0.75] [0.72] N_customers [0.30] 0.54 [0.30] [0.42] 0.43 [0.43] Sales_largest_cust 0.11 [0.04]** 0.11 [0.06]* [0.06]* 0.11 [0.07]* [0.01]** Customized 0.08 [0.24] 0.43 [0.58] 0.09 [0.22] 0.29 [0.72] Age_relation_cust 0.04 [0.57] 0.76 [0.24] 0.07 [0.27] 0.83 [0.20] Warrantee 9.21e-04 [0.83] 0.01 [0.85] [0.66] -2.66e-03 [0.94] Bank 0.05 [0.77] 0.09 [0.65] 2.27 [0.24] 2.39 [0.22] RE [0.03]** [0.03]** AP_dum Bank *AP_dum [0.19] 2.66 [0.20] [0.14] [0.12] AP_per Bank *AP_per RE*AP_dum [0.02]** [0.02]** RE*AP_per [0.56] [0.56] Constant [0.70] [0.30] [0.98] [0.58] Observations Pseudo R-squared

47 Table 9: The Financing of Trade Credit Supply Is Trade Credit Demand More Important in Competitive Markets? (Panel A) The reported estimates are from logit (Colums 1, 3, 6) and OLS (Columns 2, 4, 6) regressions. AR_dum is a dummy indicating the use of accounts receivable and AR_per is the percent of monthly sales sold on credit. See Table 1 for variable definitions. All regressions include 12 sector dummies and 17 city dummies. Robust p-values are shown in parentheses, ***, **, * indicate significance at the 1%, 5%, and 10% level, respectively. Sales Threshold =5% Sales Threshold =10% Sales Threshold =30% (1) (2) (3) (4) (5) (6) (7) (8) AR_dum AR_per AR_dum AR_per AR_dum AR_per AR_dum AR_per L_Age [0.03]** [0.02]** [0.08]* [0.03]** [0.04]** [0.03]** [0.04]** [0.03]** L_emp [0.37] [0.08]* [0.67] [0.23]* [0.72] [0.50] [0.80] [0.55] Foreign [0.07]* [0.78] [0.05]** [0.63] [0.04]** [0.57] [0.05]* [0.63] State [0.95] [0.31] [0.88] [0.44] [0.98] [0.55] [0.90] [0.56] Export [0.23] [0.65] [0.21] [0.62] [0.15] [0.89] [0.17] [0.83] N_customers [0.42] [0.54] [0.22] [0.20] Sales_largest_cust 0.07 [0.09]* 1.08 [0.02]** 0.07 [0.09]* 0.99 [0.05]** 0.08 [0.14] [0.10] 1.61 AP_dum [0.14] AP_per Bi_BP [0.13] [0.52] [0.13] [0.22] [0.13] [0.22] [0.19] [0.56] Bi*AP_dum [0.08]* [0.05]** [0.02]** Bi*AP_per [0.02]** [0.01]*** [0.02]** Customized [0.13] [0.68] Age_rel_cust [0.66] [0.20] Warrantee [0.83] 0.87 Constant [0.68] [0.89] [0.03]** [0.36] [0.03]** [0.35] Observations 1,389 1,381 1,328 1,321 1,035 1,032 1,035 1,032 Pseudo R-squared

48 Table 9: The Financing of Trade Credit Supply Is Trade Credit Demand More Important in Competitive Markets? (Panel B) The reported estimates are from logit in Colum 1 and OLS in Column 2. AR_dum is a dummy indicating the use of accounts receivable and AR_per is the percent of monthly sales sold on credit. See Table 1 for variable definitions. All regressions include 12 sector dummies and 17 city dummies. Robust p-values are shown in parentheses, ***, **, * indicate significance at the 1%, 5%, and 10% level, respectively. (1) (2) (3) (4) AR_dum AR_per AR_dum AR_per L_Age [0.03]** [0.02]** [0.03]** [0.01]*** L_emp [0.64] [0.37] [0.64] [0.26] Foreign [0.03]** [0.48] [0.04]** [0.47] State [0.94] [0.52] [0.94] [0.46] Export [0.14] [0.99] [0.21] [0.88] N_customers [0.71] [0.52] Sales_largest_cust 0.11 [0.26] 1.16 [0.35] 0.01 [0.08]* 0.03 [0.48] AP_dum 1.52 [0.03]** 0.74 [0.03]** AP_per [0.16] Bi_BP_d [0.63] [0.99] [0.08]* [0.03]** Bi_BP_d [0.38] [0.30] [0.13] 1.52 [0.60] Bi_BP_d [0.51] [0.99] [0.02]** [0.51] Bi_BP_d2*AP_d 0.76 [0.02]** Bi_BP_d3*AP_d 0.81 [0.04]** Bi_BP_d4*AP_d 1.25 Bi_BP_d2*AP_per [0.78] Bi_BP_d3*AP_per 0.11 [0.37] Bi_BP_d4*AP_per 0.34 Constant [0.03]** [0.23] [0.07]* [0.22] Observations 1,004 1,001 1,004 1,001 Pseudo R-squared