جایگزینی ضد ادواری بین اعتبار تجارت و اعتبار بانکی

This paper explores the substitution relationship between trade credit and bank credit, and its counter-cyclic dynamic pattern through economic cycles. We propose a new theoretical model, using a mechanism design method, which predicts the substitution between the two credits and its counter-cyclic behavior, subject to the condition of technological efficiency not less than one. This model also helps explain the somewhat contradictory evidence in the literature, on the relationship between the two credits. We present empirical evidence on the substitution effect and its counter-cyclic behavior, by using a balanced panel data set of 284 listed Chinese companies for the period 1998–2006. We further find that the substitution behaves counter-cyclically with respect to the coincident macroeconomic indicator, namely, GDP. Our empirical analysis also has some new features such as treating endogeneity carefully and incorporating the lag-effect of trade credit coherently.

The existence of a bank credit channel and a trade credit channel of monetary transmission, and the substitution effect between the two, have been examined in a large body of literature. Despite the extensive empirical research on the relationship between these two channels, previous studies have left at least two major gaps. First, the empirical studies on the relationship between the two channels provide somewhat contradictory evidence, with most supporting the notion of substitution, while a few support that of complementariness. We therefore need a unifying theory to consider whether it is possible for such substitutability and complementariness to coexist and why substitution dominates complementariness. Second, the dynamic behavior of the relationship between trade credit and bank credit through different cycles should be exposed explicitly. It has been more than a decade since the issue was raised by Biais and Gollier (1997).1 Unfortunately, there remains minimal literature on this issue. Dealing with it, however, would not only reveal how interaction between these two channels reacts to the changing business cycle, but also shed new light on policy implications for both macroeconomic monetary policy and microenterprise management. Moreover, existing evidence on the relationship between these two channels mainly comes from the developed countries. More evidence from emerging markets is needed to enrich the literature and deepen our understanding of this relationship. In order to fill these research gaps, in this paper, we model the relationship between trade and bank credits, using a mechanism design method, and provide evidence on the substitution effect between these two channels and the associated counter-cyclic behavior, using a balanced panel dataset of listed companies from China. This country is, after all, a large, emerging and transitional economy and one which relies heavily on trade credit (Ge and Qiu, 2007). Theoretically, we model the relationship between trade and bank credits using a mechanism design method. Our model is novel in two respects. Firstly, it is capable of explaining the seemingly contradictory evidence on the relationship between the two credits. Our model reveals that it is the technological efficiency of the manufacturer, or of the credit receiver, that largely determines whether trade and bank credits are substitutes or complements. When the efficiency exceeds a small threshold value, which is an easy criterion to meet, substitution holds. Complementariness only exists when the efficiency is very low, and below the small threshold value, which constitutes a somewhat rare case. Our model thus presents a viable explanation of the co-existence of substitution and complementariness, but with the evidence in favor of substitution dominating. Secondly and more importantly, our model predicts the counter-cyclical pattern of the substitution effect when production efficiency is greater than one, which is also quite common in the real world. Furthermore, the pro-cyclical pattern of substitution is also shown to be possible, but very infrequent. Empirically, the unique data set used in this study, with a long sample period including both a slow-growth period and one of rapid-growth, 2 as well as some novel features of our econometric modeling, yield several advantages over existing studies. Our econometric modeling treats the endogeneity issue carefully, and incorporates the lag-effect of trade credit explored in Benishay (1968), but largely neglected in empirical investigations ever since. Our fixed-effect model estimates the substitution ratio between trade credit and bank credit as 14.27% through the course of cycles. However, the dynamic panel model, which incorporates the lag-effect of trade credit, yields a substitution ratio which declines to 12.05%. With the advantage of including both rapid-growth and slow-growth periods in our data set, we find sound evidence of the counter-cyclical pattern of the substitution, by estimating the time-varying coefficients model and the dynamic panel model with a cross term of the central regressor and economic-cycle indicator. We further find that the substitution behaves counter-cyclically with respect to the coincident macroeconomic indicator, namely, GDP, by means of a graphic illustration and non-parametric tests. Moreover, our empirical results are relatively robust with respect to different scaling methods for the dependent variable, different panel model specifications, and different industries. Furthermore, restricting our sample to companies with relatively stable corporate governance or with less access to bank credit, strengthens our empirical results. We also relate real output with trade credit and provide evidence to preclude concerns of a spurious substitution between trade and bank credits, which merely indicates the accumulation of accounts payable due to financial distress. We show that the substitution between trade and bank credits is economically significant. Sales decline further during bank credit shrinkage by around half of the size of the bank credit decline, if a firm cannot substitute trade credit for bank credit. Our paper makes a contribution to the literature in two respects. First, our theoretical model unifies the somewhat contradictory empirical evidence in the field and answers the question raised by Biais and Gollier (1997) a decade ago, by demonstrating the counter-cyclical relationship between trade and bank credits through economic cycles. Second, the empirical part of this paper adds some new elements to the literature. We present sound evidence on the counter-cyclical pattern of substitution, by using a unique data set from China. Ours is not the first study on the trade credit of Chinese firms. Cull et al. (2009) also analyze a data set of China’s manufacturing companies, covering the period 1998–2003, with information obtained from National Bureau of Statistics of China. They find that trade credit is a substitute for bank borrowing for those who find it difficult to obtain finance from banks. Another study from Ge and Qiu (2007) presents similar evidence, with an analysis of survey data of 2000 Chinese firms obtained from Chinese Academy of Social Science. They show that those non-state-owned enterprises which encounter difficulty in obtaining financing from banks rely heavily on trade credit. This provides indirect evidence on substitution hypotheses. The following salient features differentiate our empirical analysis from these two papers. First, we use a data set of China’s listed companies3 covering the period 1998–2006, which is longer than the above two studies. Cull et al. (2009) cover the period 1998–2003, and Ge and Qiu (2007) cover 1994–1999. Ours covers exactly one slow-growth and one rapid-growth cycle of the Chinese economy, which makes it possible to analyze the pattern of substitution through economic cycles. However, this is hard to accomplish within the time frame of the former two studies. Second, the endogeneity issue has been treated appropriately in this paper, while it was not dealt with explicitly in the former two studies. Thirdly, we pay special attention to the lag feature of trade credit by estimating dynamic panel models. Such a feature was explored by Benishay (1968), but has received little attention in the empirical literature ever since. We are among the first to re-activate this legacy and find that the lag component does exert a significant impact on the estimation results. The remainder of this paper is organized as follows. Section 2 surveys the relevant literature and Section 3 presents a mechanism-design model of the relationship between trade and bank credits. Section 4 describes our data set, econometric models and estimation strategies. Section 5 presents and interprets the empirical results and Section 6 concludes.

In this paper, we study, both theoretically and empirically, the substitution between trade credit and bank credit. Motivated by the co-existence of somewhat contradictory empirical evidence on the relationship between trade credit and bank credit, we set up a mechanism-design model to specify the conditions for substitution and complementariness. We find that, when production efficiency crosses a low threshold value, trade credit and bank credit tend to be substitutes. Otherwise, they are complementary, but this rarely occurs. Our model therefore provides a viable explanation of why the evidence on substitution in the literature dominates over that proposing complementariness. Furthermore, our model predicts that in most cases, the time-varying degrees of substitution between trade credit and bank credit behaves in a counter-cyclical manner. Empirically, we use a balanced panel data set drawn from listed Chinese companies to test the theoretical predictions on substitution and its counter-cyclical behavior. The feature of the data set covering a slow-growth and a rapid-growth cycle enables us to provide reliable evidence on both predictions. The substitution hypothesis is found ubiquitously by all kinds of model specifications in this paper. We also find solid evidence on the counter-cyclicality hypothesis, by estimating a time-varying coefficients panel model and a dynamic panel model with a cross-term of business-cycle indicators. The estimation results show that the magnitude of substitution during the slow-growth period (1998–2002) is significantly larger than during the rapid-growth period (2003–2006), and the coefficient of the cross-term of bank credit and the dummy variable of the slow-growth (rapid-growth) period is shown to be significantly negative (positive). Moreover, in order to exclude the possibility that the negative relationship between trade credit and bank credit reflects only the accumulation of unsolved accounts payable, but not really substitution in financing real production, we also run a regression of real output, represented by sales, to trade credit, with other relevant factors controlled. The estimation results show that trade credit does boost real output, which supports the notion that trade credit and bank credit are real substitutes in financing production activities in our case. Our findings highlight the counter-cyclical behavior of substitution between trade credit and bank credit. We thus explain the pattern of relationship between trade credit and bank credit through different cycles, as raised in Biais and Gollier (1997). Our results, from a large emerging economy, complement other prominent empirical studies (such as Peterson and Rajan, 1997, Nilsen, 2002, Danielson and Scott, 2004, Mateut et al., 2006, Love et al., 2007, Ge and Qiu, 2007, Cull et al., 2009, Ferri, 2009, Scott and Dunkelberg, 2010 and Lin, 2011; among others) to enhance our understanding of the relationship between the two most important short-term financing channels for firms. The investigation reveals various different behaviors of firms in using trade credit to offset bank credit constraints, when facing different phases of the economic cycle. Such a counter-cyclical offsetting is generally benign for the economy as a whole, because it is actually a spontaneous reliever of bank financing constraints during hard times and a self-generated smoother of liberal lending in good times. Hence, our results are also relevant to monetary policy-makers. They suggest that an aggregate credit perspective is necessary, when making decisions on monetary policy concerning credit supply. That is, the substitution effect of trade credit on bank credit, and its counter-cyclicality attribute should all be taken into account. Two caveats, both of which are unique to China, should be borne in mind in interpreting our empirical results. The first is that China did not experience a real recession during the period under study. This accounts to some extent for our finding that the difference in substitution effect over the rapid and slow-growth periods is not large, as shown in Table 7. The second involves the substantial influence of China’s government on the banking sector and the good relationships that China’s listed companies maintain with the banks. This somewhat attenuates the level of substitution, particularly during slow-growth periods. Therefore, had the contraction been deeper during the slow-growth period, and the banking sector more independent, the change in the substitution effect over different cycles might be more evident. This conjecture may apply to other emerging and transitional economies, an issue which remains open for further research. Although our paper sheds some new light on the relationship between bank credit and trade credit, there are many areas that could usefully be investigated in the future. For instance, our data include only large, publicly traded Chinese firms, and China is a large, emerging economy which arguably relies more on trade credit than the developed countries. Therefore, the patterns we observe may not generalize to SMEs or to firms in other countries. Thus, more research is needed to test whether the counter-cyclical pattern we find also applies to different firm sizes and countries. Furthermore, additional work is also needed to identify industry-specific cyclical substitution behavior, something which cannot be achieved through the present investigation, due to the limited sample size.