Political connections and informal financing: use of trade credit in China

ABSTRACT Using a dataset of listed non-state-owned enterprises in China, we show that politically connected firms have a greater difficulty in obtaining trade credit than non-connected firms. This credit discrimination against politically connected firms strengthens when firms locate in regions with lower-quality legal systems or suppliers have greater bargaining power. Further analysis suggests that the lower credit reputation of politically connected firms stems from suppliers’ concern about contract enforceability, that is, politically connected firms receive regulatory favours from government, which can create obstacles to the collection of trade credit through the legal system. Our study contributes to the existing literature by investigating the impact of political connections on informal financing activities, and by revealing the asymmetric influence of political connections on formal and non-formal financing channels.


Introduction
Politically connected firms are common globally and have attracted enormous research interest in recent years. Although previous studies have documented that political connections enable firms to obtain financing such as bank loans and equity financing more easily (Boubakri et al., 2012;Chen & Yuan, 2004;Piotroski et al., 2015), limited evidence has been produced to show how political connections can affect firms' access to more informal sources of financing, including trade credit. In the emerging market of China, even listed firms, which may find it easier to gain access to formal finance, rely heavily on trade credit as an alternative source of financing, as do unlisted firms (Cull et al., 2009;Lee & Stowe, 1993). This paper responds to this gap in the literature by investigating whether connected firms have better access to trade credit than non-connected firms in emerging markets.
We propose two competing hypotheses. On the one hand, it may be that connected firms receive more trade credit in line with their general operating advantages and better creditworthiness or because suppliers are willing to supply trade credit to them based on their operating advantages. On the other hand, connected firms may receive less trade credit because they have better access to other sources of finance (we refer to this as the demand-side formal financing argument) or because suppliers fear that regulatory favours provided to the client firm will result in limited debt contract enforceability if payment becomes overdue (we refer to this as the supply-side limited contract enforceability argument). Thus, whether connected firms receive more or less trade credit relative to non-connected firms is an open empirical question.
China is an ideal setting for investigating whether connected non-State-Owned Enterprises (non-SOEs) have better access to trade credit relative to non-connected firms. First, unlike a mature market economy, China, while one of the largest economies in the world, has impediments that are typical of emerging markets (Allen et al., 2005;Johnson et al., 2002). The cultivation of political connections is, on the one hand, an effective mechanism for overcoming market failure, helping non-SOEs access government-controlled resources and regulatory favours (Correia, 2014;Li et al., 2008). Thanks to this privilege, a non-SOE with political connections may be more competitive in obtaining trade credit from suppliers. On the other hand, the privilege of a non-SOE (i.e. regulatory favours) may also dampen the effectiveness of the legal system, and suppliers may face limited contract enforceability if payment becomes overdue and may obtain only limited protection from the legal system. These competing effects of political connections on the use of trade credit in a large but typical emerging market are our main interest.
Second, non-SOEs in China are suitable for the present study to explore in regard to their representativeness. According to the World Bank's Enterprise Surveys (2012), which were conducted on a sample that contained 90% non-SOEs, more than 22.4% of firms identify financial constraints to be a major obstacle for their operations. In medium-sized and small firms, this figure reaches 25%, and only 4.5% of the major investments of the surveyed firms, such as the purchase of fixed assets, are financed by bank loans. 1 Over the last few decades, non-SOEs have grown tremendously, with little support from formal financial markets, making trade credit even more crucial for their operations (Ge & Qiu, 2007;Lee & Stowe, 1993). Meanwhile, unlike SOEs, which are politically connected from the moment of their existence and face soft budget constraints, non-SOEs in China are relatively clean and are more suitable for the investigation of the role of political connections in financing.
We find that a firm's political connections result in a significant reduction in the trade credit ratio relative to non-connected firms, indicating that connected firms have greater difficulty obtaining trade credit. Our results further show that trade credit availability does not increase when firms' intrinsic demand for informal financing increase, so we rule out the formal financing argument and validate the limited contract enforceability argument that the effect is driven by the limited contract enforceability in connected firms. Moreover, the departure of senior executives is usually exogenous to factors affecting trade-credit taking (Fisman et al., 2012;Luechinger & Moser, 2014), this provides us with an exogenous shock to study the termination of political ties and thus mitigate the concerns over endogeneity. We employ PSM-DID approach and find a positive effect on trade credit after the termination of political ties, which thus means that a sudden halt in the political connection has a positive impact on a firm's trade-credit taking. Meanwhile, we also use two-stage least squares (2SLS) to address potential endogeneity issues and the results are robust. These findings together indicate that, contrary to existing conclusions on formal financing channels, politically connected firms receive less trade credit because suppliers fear being less able to enforce their debt contracts if payment becomes overdue.
To further validate our limited contract argument, we examine mitigating effects as corroborating evidence. First, we test the impact of the quality of legal environment on the relationship between political connections and trade credit and find that, for firms located in regions with higher-quality legal systems, the negative impact of political connections on the use of trade credit is partially alleviated. The implication is logically consistent with the literature, which documents the important impacts of legal enforcement on firm behaviour (Giannetti et al., 2011;La Porta et al., 1998).
Secondly, we analyse the composition of the overdue payment of accounts payable. The results indicate that connected firms in general keep accounts payable longer. On average, firms with political connections have a lower proportion of accounts payable (less than 1 year), higher proportion of late payments of accounts payable (more than 1 year) and overdue payments of accounts payable (more than 3 years), suggesting connected firms are more likely to substantially violate the due date. Then, we test the bargaining power of suppliers on the relationship between political connection and trade credit, as corroborating evidence. We find that in firms dealing with stronger bargaining power suppliers, the negative effect of political connections is pronounced, which indicates that supply-driven credit discrimination is more likely when suppliers have equally strong bargaining power over the politically connected customers.
Thirdly, we examine the market perception of politically connected firms when they are involved in contract violation cases compared to the perception of firms without such connections. We find that the cumulative abnormal return (CAR) is significantly negative in almost all window periods for non-politically connected firms, but for all politically connected firms, the CAR is nonsignificant. These tests further confirm the significant difference between connected and non-connected firms, indicating limited contract enforceability through the legal system related to politically connected firms.
Finally, we conduct other robustness tests in the following aspects: 1) analysing the effect of political connection on trade credit among state-owned enterprises (SOEs) and the subsamples with and without ownership change; 2) investigating the relationship between CSR activity in the form of charitable donations and trade credit by using different subsamples; 3) using different measurement of political connections and trade credit; 4) addressing potential omitted variable issue. Overall, these robustness tests further support our argument.
Our study contributes to the literature in several ways. First, by exploring the links between political connection and trade credit of firms, we add a new dimension to the literature on political connections. Although a few studies focus on the dark side of political connections, they provide evidence on the costs of developing political connections in China, such as incurring operating and investment inefficiencies (Pan & Tian, 2020;Piotroski et al., 2015), reducing accounting information quality (Chaney et al., 2011;Faccio, 2010), facilitating rent expropriation (Sun et al., 2016), and deteriorating corporate governance (Cao et al., 2017), but the benefits of doing so are more widely acknowledged. Most studies of which focus on how political connections facilitate financing activities and the kind of benefits that could ensue from formal financial intermediaries (Chen & Yuan, 2004;Claessens et al., 2008;Piotroski et al., 2015). Our study, instead, investigates the ongoing cost of having this specific asset (i.e. political connections). We argue that political connections could help firms win regulatory favours, but these benefits could dampen the enforceability of contracts with such firms and their creditworthiness from the suppliers' perspective.
Second, there is a strand of literature that emphasises the importance of the informal mechanisms of social capital in emerging markets (Ge & Qiu, 2007;Lee & Stowe, 1993;Petersen & Rajan, 1997). For example, a collection of studies suggest that access to external funding in China mainly depend on firm ownership and political connections also play an important role (Guariglia & Mateut, 2016;Li et al., 2008). We build on these studies and highlight the wide variation in the roles played by the same feature of social capital (i.e. political connections) in formal and informal channels of finance, focusing on non-SOEs in China, which are relatively clean samples and aren't influenced by the firms' ownership.
Third, our study complements research on the determinants of the use of trade credit (Carbo-Valverde et al., 2016;Cunat, 2006;Giannetti et al., 2011;Petersen & Rajan, 1997). In particular, we show that the acquisition of social capital (e.g. political connections) may incur side effects that reduce the use of trade credit.
The remainder of our paper is organised as follows: Section 2 introduces the institutional background, Section 3 provides a literature review and hypothesis development, Section 4 describes the sample and research design, Section 5 discusses the empirical results, Section 6 outlines the scope for further studies, Section 7 reports robustness checks, and Section 8 provides the conclusions.

Bank lending and trade credit in Chinese non-SOEs
When China launched its transformative reform to a market-based economy in 1978, the banking sector was deregulated and formally granted greater autonomy, and those banks became the commercial banks of today. However, despite the significant progress that has been made in economic reform and fiscal decentralisation over the last three decades, banks are still greatly influenced by central and local governments. According to statistics from the China Banking Regulatory Commission, in the 107 commercial banks operating in China, 2 as many as 85.05% of the controlling shareholders are state-owned entities.
This ownership structure may partially explain the financial constraints faced by the Chinese private sector (Garnaut & Huang, 2000;Ge & Qiu, 2007). Although the private sector formed about 60% of China's GDP, provided 50% of its tax revenue, and supplied 80% of its employment in 2019, only about 20% of private-sector funding resources came from banks (The report of China's private economy, 2019). Instead, non-SOEs have relied heavily on trade credit. According to statistics from the Chinese Academy of Social Sciences, in the year 2000, accounts payable represented as much as 14% of firms' total assets. Non-SOEs on average have 4.3% more in net trade credit than SOEs. In 2017, a study of cost reduction conducted by the Chinese Academy of Fiscal Sciences noted that the proportion of accounts payable in total assets in 2016 was 7.83% greater than in 2015, implying the importance and heavy use of trade credit in non-SOEs.

Political connections in emerging economies
In a transitional economy such as China, where market mechanisms are only partially utilised, the government retains significant power and critical control over resource allocation and regulation. In particular, the central government oversees national or transprovincial affairs, such as economic regulation, market supervision, social management, and provision of public services; local governments manage the equivalent types of matters that fall within their jurisdictions. To a large extent, local governments at the provincial level and below exercise control over the legal and financial systems within their boundaries. Firms' original registration locations indicate the provincial jurisdiction to which they are subject.
As a result of government interference, political connections are cultivated as an effective mechanism for reducing transaction costs (C. J. Chen et al., 2011), in stark contrast to the findings in the literature on law, institutions, finance, and growth (La Porta et al., 1998Porta et al., , 2000. Especially for non-SOEs, connections with the government can alleviate discrimination against non-SOEs and gain them access to governmentcontrolled resources at a lower explicit cost (Allen et al., 2005), even if the actual means are costly (e.g. direct lobbying, political contributions, and bribery; Correia, 2014;Lo, 2003).
There is a strand of literature that provides evidence on the costs of developing political connections in China, such as losing transparency, incurring operating inefficiencies, and diverging from value-maximisation objectives (Piotroski et al., 2015), but the benefits of doing so are more widely acknowledged. For example, a collection of studies suggest that firms with political connections can more easily obtain listing privileges in the security market, access to additional government subsidies and tax benefits, and stronger support from the legal system (C. J. Chen et al., 2011;Li et al., 2008;J. P. Fan et al., 2007).
Although political connections can bring multiple benefits to firms, the most important advantage seen in the literature is better access to formal financial resources (usually banks). However, few studies investigate the role of political connections in informal financing. Unlike banks, which are heavily influenced by central and local governments, trade credit depends more on particular suppliers' evaluations of their customers. The suppliers' perceptions of customers' political connections regarding repayment of trade credit could be quite different from those of other stakeholders (e.g. investors and banks). Taking account of the reliance on trade credit in non-SOEs, our study focuses on the effects of political connections on trade credit.

Literature review and hypothesis development
Suppliers provide trade credit when they deliver goods to customers but do not immediately collect the proceeds. Previous findings indicate that the amount of trade credit provided by firms is influenced by many factors, including financial constraints (Cull et al., 2009), access to bank loans (Claessens et al., 2008;Khwaja & Mian, 2005), investment opportunities, macro-financial policies (Sáiz et al., 2017), market competition (Petersen & Rajan, 1994), and default risk (Wu et al., 2014). Incorporating these findings into our analysis, we summarise the effects of political connections on trade credit relative to both demand and supply below.

Positive effects
Political connections are perceived to be valuable social capital (Bunkanwanicha et al., 2013;Faccio & Parsley, 2009;Fisman, 2001). The literature shows that connected firms are better able to obtain operating advantages: they are generally less vulnerable to enforcement of government regulations and can elicit regulatory favours, receive government subsidies and tax benefits, enter restricted markets and industries, and be offered preferential government bailouts in response to financial stress (Faccio, 2006;Li et al., 2008;Lin et al., 2015;Wu et al., 2014). These operating advantages help firms obtain more exclusive development opportunities, undertaking high-return projects, entering less competitive market sections, and reaching superior customers, thus stimulating their demand for financing, whether formal or informal.
Taking a supply-side view, political connections may signal a high level of operating premium, competitiveness, and potential. Previous works assign a value to political connections and confirm that markets, on average, perceive political connections to be a valuable resource. For instance, Fisman (2001), Faccio and Parsley (2009), and Bunkanwanicha et al. (2013) employ unique events that cause an exogenous shock to political connections validate the significant value that political connections can add to firms via changes in market pricing. Taken together, these findings indicate that political connections may improve the market perception of a firm, which then may add to the firm's creditworthiness and increase the willingness of suppliers to grant trade credit.

Negative effects
However, some specific operating advantages may also dampen the need for trade credit, such as better access to formal financing. Previous studies demonstrate that an obvious operating advantage of politically connected firms is their easy access to formal finance, such as loans from banks or other state institutions (Claessens et al., 2008;Khwaja & Mian, 2005;Li et al., 2008;J. P. Fan et al., 2007), as well as the privilege of being listed on securities markets (Boubakri et al., 2012;Chen & Yuan, 2004;Francis et al., 2009;Piotroski et al., 2015). Those findings implicate a substitution between trade credit and bank loans (Fisman & Love, 2003;Ge & Qiu, 2007). In other words, in relation to demand, politically connected firms can obtain formal financing to meet operational and transactional needs in a simple way, which results in a loss of demand for trade credit.
Another operating advantage that could have a negative effect on trade credit for politically connected non-SOEs consists of the regulatory favours that such firms can obtain. Earlier studies show that firms benefit from political connections in various ways, but recent studies focus on the regulatory favours that politically connected firms receive, which jeopardise the effectiveness of the legal system. By testing the effectiveness of enforcement actions performed by the United States Securities and Exchange Commission (SEC), Correia (2014) documents that politically connected firms are less likely to be subject to SEC enforcement action and face lower penalties when they are prosecuted. In emerging markets, such as China, Li et al. (2008) find that political connections afford firms greater confidence in the legal system and this effect is greater where institutions are weaker. Similarly, Xu et al. (2013) provide direct evidence for the effects of political connections on the efficiency of legal enforcement: they find that in China, the time spent identifying and prosecuting fraud for politically connected firms is usually longer than for firms that are not politically connected.
While banks can diversify risk by having a relatively large customer base, suppliers have only a small number of customers and cannot efficiently diversify their risk (Wu et al., 2014). Therefore, evaluating future default risk among their customers and the probability of retrieving the trade credit extended if default actually occurs are crucial factors in the decision whether to provide trade credit (Cunat, 2006;Fisman & Love, 2003;Petersen & Rajan, 1994). Guiso et al. (2004) point out that, once payment is overdue, the enforceability of a contract is largely determined by the suppliers' ability to retrieve the trade credit they provided. To a large extent, the enforceability of a contract usually depends on the effectiveness of the legal system. Johnson et al. (2002) report that improvements in the effectiveness of the courts can stimulate contracting with trade credit, which could encourage suppliers to 'try out new customers', especially when specific investments are needed. Therefore, from the supply side, suppliers may maintain a cautious attitude in extending trade credit to firms who enjoy legal favours. 3 Chinese media have reported much anecdotal evidence on the obstacles to achieving legal redress when politically connected firms refuse to pay their trade credit. For example, a long-outstanding payment order between Zhenxing Biopharmaceutical and Chemical Co., Ltd., and its supplier, Zhenjiang Hydraulic Co., Ltd; the marathon lawsuit, between Himin Solar Energy Group Co., Ltd., and its dealer, Jiuquan City Jiadian Trade Co.; and Greenland Holdings Group Co., Ltd., which was surprisingly publicly debt collected by several suppliers. Thus, operating advantage and better market perception add to politically connected firms' creditworthiness and increase supplier willingness to grant trade credit, but the better access to formal financing and regulatory favours that political connections bring may decrease the requirement for trade credit. Therefore, we propose two competing hypotheses below:

H1a: Politically connected firms obtain more trade credit from suppliers.
H1b: Politically connected firms obtain less trade credit from suppliers.

Demand or supply?
As mentioned above, the amount of trade credit taken by customers or provided by suppliers could be seen as a story of demand and supply. If taking trade credit and political connections are positively correlated, then operating advantages of political connections could be explained from either the demand side or the supply side.
However, if use of trade credit and political connections are negatively correlated, there are two plausible explanations that include demand and supply sides. On the demand side, the formal financing argument explains the negative correlation between political connections and trade credit relative to substitution. Connected firms' privileged access to formal financing allows them to obtain sufficient bank loans, and they exhibit less demand for trade credit (Ge & Qiu, 2007). From the supply side, the limited contract enforceability argument suggests that firms with political connections may be credit rationed by suppliers, as the regulatory favours received by politically connected firms could increase collecting costs and risk (Correia, 2014;Li et al., 2008).
Both perspectives predict a negative relationship between trade credit and political connections, but they use different interpretations. To determine the equilibrium and see which argument more convincingly explains the data, we identify the setting when firms are experiencing excessive demand for financing. If the formal financing argument explains the negative association between trade credit and political connections, we expect to observe firms to spontaneously switch from formal financing to informal financing (e.g. trade credit) as a supplement to the withdrawal of bank credit. In the setting where formal financing is difficult to obtain, we could observe that the negative relationship between trade credit and political connections is significantly weakened. However, if the limited contract enforceability argument has more explanatory power, we should expect that loss of formal financing would not change the supply-side factors that determine trade credit rationing. Hence, in settings with poor accessibility of formal financing, the negative association between trade credit and political connections would not be dampened. Therefore, we hypothesise the following:

Data and sample
The data are primarily drawn from the China Stock Market and Accounting Research (CSMAR) database. We exclude state-owned firms and focus on non-SOEs listed on the Shanghai and Shenzhen stock markets. The sample period ranges from 2003 to 2019. To anticipate potential error induced by outliers, we winsorise all variables at the top and bottom 1% and exclude observations with missing information, including cases without firm-specific data or cases with insufficient information on the background of the chair or CEO; firms in the financial industry. Our final sample contains 2,084 firms and 12, 839 firmyear observations.

Research design
The regression analysis we use to empirically test H1a and H1b is based on Equation (1): Where Tc is the trade credit, following Fisman and Love (2003), we calculate the use of trade credit (Tc) as the percentage ratio of accounts payable and notes payable to total assets, for the first measurement of trade credit. Because use of trade credit could depend on industry characteristics and customs (Giannetti et al., 2011), we adjust Tc using its industry median, denoted as Tc_adj, for the second measurement of trade credit. The independent variable Political i , t is a dummy variable that indicates whether the firm is politically connected. Following J. P. Fan et al. (2007), and C. J. Chen et al. (2011), we manually collect background information on top executives and measure firms' political connections through the lens of the firm's chair or CEO; if this person is a current or former officer in government or the military, the firm is considered politically connected. This includes working experience in central or local government departments, the military, the People's Congress, the People's Court and Procuratorate, or the Chinese People's Political Consultative Conference. μ i is the firm fixed effect and γ t is the time fixed effect.
Control is a set of control variables. Following the literature, we further control for firm size (Size), listed years (Listage), fixed assets ratio(Fix), leverage (Leverage), the percentage ratio of the loan to total debt (Loan), return on total assets (Roa), the percentage ratio of the Herfindahl index for the industry based on firms' sales (Comp), firm value (TobinQ), board size (Board), average age of board directors (Age), independent director ratio (Independent), the proportion of the first largest shareholder (Top), GDP growth rate (DGdp), and year (Year) dummies.
To test H2a and H2b, we introduce an interaction that examines whether politically connected companies are able to return to informal financing when their demand for formal financing cannot be met.
In Equation (2), we apply a micro-level variable (SAindex) to measure firms' intrinsic demand for informal financing. Following Hadlock and Pierce (2010), we first estimate the ordered logit model to predict constraints as a function of different quantitative factors, and then we construct the variable SAindex, which is 1 if the financial dependence index is above the median and 0 otherwise.

Descriptive statistics
Panel A of Table 1 shows the descriptive statistics for the variables. The mean value of the variable Political is 0.37, indicating that 37% of our sample firms are politically connected, consistent with Li et al. (2008). The mean value for trade credit (Tc) is 12.01%, and the difference between maximum and minimum trade credit is about 0.468, indicating a large variance in use of trade credit among firms. For the control variable, the mean value of Size is about 2.33 billion RMB (= e 21.57 ). The mean of Leverage is 42.23%, while the ratio of loans from banks accounts for 32.45% of total liability. Panel B of Table 1 presents the mean difference tests between politically connected and non-politically connected firms in use of trade credit. Specifically, the mean difference is significant at the 1% level, suggesting that non-connected firms receive more trade credit on average. And the difference in industry-adjusted use of trade credit is insignificant.  (1) and (3). In columns (2) and (4), we further include all control variables. And we adopt firm-fixed Note: * p < 0.1, ** p < 0.05, and *** p < 0.01. effects model in all regressions. The estimated coefficients associated with political connections (Political), are all negative and significant, indicating that managers' political connections on average decrease firms' use of trade credit. Beyond this, we find that firms with more collateral usually enjoy more trade credit, and the coefficient of Loan is significantly negative. Overall, the results in Table 2 support H1b. Table 3 provides the results of the examination of H2a and H2b. Columns (1) to (2) show the demand-side intrinsic need for formal financing, we find that the significance of political connections (Political) is negative and significant at the 1% level. More importantly, the interaction terms remain insignificant, implying that connected firms in the group with greater financial dependence will still encounter difficulty obtaining informal finance, even though their demand for finance is high. To some extent, these findings support the limited contract enforceability argument and suggest that politically connected firms' trade credit is still rationed by suppliers, even when the accessibility of formal financing is challenged. Note: T-statistics adjusting for clustering at the firm level in parentheses, * p < 0.1, ** p < 0.05, and *** p < 0.01.

Propensity Score Matching and Difference-in-Difference approach (PSM-DID)
While the main tests assume that political connection impacts the level of firm trade credit, the tests, however, do not take into account potential simultaneity between the two variables. In this section, we take advantage of exogenous events (e.g. termination of political connections) to mitigate that concern by employing a Propensity Score Matching and Difference-in-Difference approach (PSM-DID) procedure. The departure of senior executives is usually exogenous to factors affecting trade-credit taking (Fisman et al., 2012;Luechinger & Moser, 2014), this provides us with an exogenous shock to study the termination of political ties. In our sample, during the period 2003 to 2019, we could identify cases where the political tie was terminated. This subsample is our treatment group, in which these politically connected firms lose their political tie since politically connected senior executives leave his/her job. Before using Difference-In-Difference (DID) procedure to estimate, we use a propensity score matching (PSM) procedure, which allows us to identify a control group of firms which politically connected firms never lose their political ties. Since the firms' termination of political connection occurs in different years, we match each treatment firm with control firms year by year, following Blundell and Dias (2010). The propensity score is calculated by projecting the probability of a firm being treated on the same control variables except the macro variable DGdp. To ensure that the firms in the control group are sufficiently similar to those in the treatment group, we perform the nearest neighbour matching strategy with the common support required by matching each treatment firm with four control firms. In the matching process, we finally obtain 167 treatment firms and 473 control firms with 6228 firm-years. And after matching, the treatment and control groups appear to be largely indistinguishable in terms of the characteristics mentioned above (unreported for brevity).
Considering that the firms' termination of political connection occurs in different years, we employ the following multi-period Difference-In-Difference (DID) procedure to estimate.

TC i:t or Tc adj
Where Treat i,t takes a value of 1 if a firm is in the treatment group; and Post i,t is a time dummy with a value of 1 for the remaining years of losing political connection. The results from Equation (3) are presented in panel A of Table 4. The coefficients associated with the interaction variable, Treat×Post, are all positive and significant; suggesting that, when a firm switches from politically connected to non-politically connected status, the firm experiences an increase in trade credit, which means that a sudden halt in the political connection has a positive impact on a firm's trade-credit taking.

Instrumental variable estimation
It is possible that both political connection and use of trade credit are simultaneously determined by some underlying factors. To rule this out, we apply the 2SLS regression using instrument variables as a further robustness check. Following Cao (2013), we use two instrumental variables: one is lagged business volume of post and telecommunication services, and the other is the average percentage of political connected firms in each industry-year-region group. The region's economy affects government personnel decisions in China (Li & Zhou, 2005), which could further affect the relationship between government and firms. Post and telecommunications services which include express, telephone and network et al., to a large extent, reflect the development of local business, thus affecting the local officials' replacement. However, post and telecommunication service is an industry affiliated to the other economic sectors. Its business volume in last year is independent of firms' use of trade credit. Additionally, firms are more likely to develop political connections when the proportion of political connected firms in an industry-year-region is higher. Neither of these IVs is likely to have a direct effect on the firm-level use of trade credit, which makes these variables valid instruments. Panel B of Table 4 presents the IV regression results, we find a statistically negative effect of political connection on the use of trade credit, which means our main findings maintain when the potential endogeneity problem is accounted for.

Further evidence
The empirical results described above are evidence that politically connected firms may experience discrimination in the issuance of trade credit. In this section, to corroborate these results, we explore the sensitivity of the main results and the channel through which trade credit discrimination occurs.

Sensitivity of main results: quality of legal environment
The constitution of formal institutions is important for a firm's financing behaviour. La Porta et al. (2000) consider that the influence of the institutional environment, especially the state of law enforcement, is non-negligible. A recent paper emphasises how political connections win regulatory favours (Correia, 2014;Li et al., 2008) by dampening the effectiveness of legal enforcement, thus raising concern among private lenders regarding the enforceability of a contract when payment of trade credit is overdue. Following this logic, suppliers' concerns for their customers' ex post default risk reflect a variable pattern in legal environments of differing quality. Thus, we expect that, following our supply-side limited contract enforceability argument, credit discrimination against politically connected firms may be weakened in regions with sound legislation and effective law enforcement.
We use the Development of Intermediary Organizations (DIO) index to measure the quality of the legal environment in different provinces of China, based on the survey conducted by G. Fan et al. (2020). This index refers to the service conditions of lawyers, accountants, and other intermediary organisations, with higher scores indicating a higher-quality legal environment. The results are reported in columns (3) and (4) of Table 3, it is shown that the coefficient of Political is negative. And the coefficient of the interaction term (DIO × Political) is of more interest, which is positive and significant, indicating that the negative effect of political connection is mitigated by a relatively effective legal environment. Thus, this supports that political connections have a stronger impact on payables in regions with weaker legal institutions.

Use of trade credit in connected firms and the role of suppliers
In this section, we investigate the use of trade credit and the role of suppliers in connected firms, which could partially explain why politically connected firms exhibit worse credit reputations.

Payment of accounts payable in connected firms
While trade credit allows customers to delay payment to their suppliers for some period, this delay is usually restricted to the short term (3 months or less). When customers do not pay their accounts payable within a usual credit period, suppliers may not only suffer the cost of an immediate liquidity crisis but also face even more uncertain trade credit payments in the future. From the customer's perspective, late or overdue payment of accounts payable can lead to implicit costs such as deterioration of their credit reputation, making suppliers more cautious in their extension of trade credit. Therefore, we analyse the composition of overdue payments of accounts payable in politically connected firms. We use Late and Overdue to measure the late and overdue payments. Late i,t is the proportion of accounts payable more than 1 year past due to the total amount of accounts payable. Overdue is the proportion of accounts payable past due of more than 3 years to the total amount of accounts payable. And we also include Prompt to measure firms' on-time collection of accounts payable, which is the proportion of accounts payable of less than 1 year past due to the total amount of accounts payable.
Panel A in Table 5 reports the estimation results. As column (1) shows, the coefficient of Political is negative and significant at the 5% level, indicating that the accounts payable of less than 1 year past due is lower than for non-politically connected firms. In column (2) and (3), we find that accounts payable of over 1 year and 3 years past due in politically connected firms is significantly higher than in non-connected firms. These results suggest that politically connected firms usually exhibit longer withholding periods in the repayment of trade credit, which may lead to implicit costs such as deterioration in customer credit reputation and further result in suppliers being more cautious in extending trade credit.

The role of suppliers
The 'contract enforceability' argument assumes that the political risk to suppliers discourages their credit supply to the customers with political connections. However, supplydriven credit discrimination is more likely when suppliers have equally strong bargaining power over the politically connected customers. If suppliers have relatively strong bargaining power, then they can effectively hedge against the political risk of customers. Therefore, if the 'contract enforceability' argument holds, we would observe the main results to be stronger in the firms that are dealing with high bargaining power suppliers. We construct a dummy variable supply, which equals 1 if a firm's concentration of top five suppliers is above the median and otherwise 0 to evaluate suppliers' bargaining power. The results are reported in Panel B of Table 5. The coefficient of the interaction term is of more interest, which is negative and significant, indicating that the negative effect of political connection is pronounced in the sample with high bargaining power suppliers. Note: T-statistics adjusting for clustering at the firm level in parentheses,* p < 0.1, ** p < 0.05, and *** p < 0.01.

Market perception of contract violation in connected firms
A company may experience a negative market reaction when it discloses its involvement in a lawsuit, especially as a defendant (Sun & Zhang, 2006). La Porta et al. (2004) argue that, when defendants are politically connected, the courts usually have difficulty maintaining judicial independence. The influence of the government can cause courts to compromise their consideration of matters of justice (Glaeser & Shleifer, 2002). Recent studies also provide corroborating evidence for the regulatory favours that politically connected firms receive (Correia, 2014;Li et al., 2008). Thus, when firms are involved in lawsuits as defendants, investors may perceive companies with political ties as a special group, protected by the government, and those firms may thus experience less negative market reactions. Berkman et al. (2010) provide relevant evidence that market reactions in China to news of involvement in a lawsuit are much less drastic for firms with political ties than those involving non-connected peers. Following previous work, in this section we examine market perception of politically connected firms in a contract-violation lawsuit. When firms disclose contract-violation lawsuits, the market's perception of politically connected firms signals the degree of limited contract enforceability that may be relied upon in the legal system.We take the contract-violation case information for listed firms from the China Listed Firm's Litigation and Arbitration Research Database. Only the first record is used for firms that are involved in contract violation lawsuits more than once in a single year. Finally, we divide our sample into two groups, obtaining 17 politically connected firms and 63 non-politically connected firms.
We conduct an event study and calculate the CAR around the days contract violation cases are disclosed. The standard market model is used to compute the CAR for the relevant event windows: each window extends from (−1, 1) to (−10, 10) with 1-day intervals. Following G. Chen et al. (2005), we choose an estimation period of 250 days (−30, −280). Table 6 reports the results. For non-politically connected firms, the CAR is negative and significant for all windows, indicating a negative market reaction that follows the release of information on involvement in lawsuits. However, for all politically connected firms, the CAR is positive but not significant in the window (−1, 1) to (−10, 10). These results indicate that politically connected firms generally receive less negative reaction in the market when they are involved in contract violation cases, and thus they are likely to be perceived as a special group that is protected by the government and treated with regulatory favours. 4

Ownership change
As a comparison, we conduct additional analyses on the impact of political connection on the use of trade credit among state-owned enterprises (SOEs) and firms with ownership changes. Column (1) and (4) in panel A of Table 7 show the relation between political connection (Politcal) and trade credit (Tc) for the state-owned sample (Nsoe = 0) and the non-SOEs (Nsoe = 1) respectively. And we notice that the coefficient of Political dummy variable is negative and insignificant in the state-owned sample, compared with the significantly negative coefficient in the non-SOEs, suggesting that the negative effect of political connection doesn't exist in the SOEs. The behind reason may lie in that unlike non-SOEs, SOEs inherently have political ties and are endorsed by the state, which could weaken the negative effect of the political connection established through social capital.
Then the question arises: whether the effect of political connection on the use of trade credit changes if a state-owned firm becomes the non-SOE or vice versa. We define the variable Soe_chg as 1 if a state-owned firm becomes the non-SOE, and zero otherwise. Meanwhile, the variable NSoe_chg is a dummy with a value of 1 for the non-SOE becoming the SOE. The results are reported in panel A of Table 7. In columns (2) and (3), the coefficient of the variable Political is negative and statistically significant when SOEs become the non-SOEs, and still statistically insignificant in subsample of SOEs without ownership change. Similarly, columns (5) and (6) show the results in two sets of subsamples of non-SOEs without and with ownership change, indicating the effect of political connection becomes insignificant when non-SOEs change into SOEs. Overall, these findings further support our argument that political ties play a negative role in the use of trade credit for the non-SOEs who establish political connection through social capital.

Corporate social responsibility activities
Prior literature points out firms in China use corporate social responsibility (CSR) to build political networks (Lin et al., 2015). In this robustness analyses, we investigate the relationship between CSR activity in the form of charitable donations and trade credit by using  Note: T-statistics adjusting for clustering at the firm level in parentheses,* p < 0.1, ** p < 0.05, and *** p < 0.01.
the natural logarithm of amount of charitable donations plus 1 (Donation) as an alternative political connection measure. Panel B of Table 7 reports the results. Column (1) and (4) show the results for the non-SOEs (Nsoe = 1), indicating that Donation is positively associated with trade credit. This finding is in accord with existing literature (Zhang et al., 2014). Theoretically, the mechanism of CSR on increasing trade credit is mainly embodied in that engaging in CSR-related activities could enhance trust and reduce the information asymmetry between enterprises and suppliers; in turn, stakeholders are more likely to trust and cooperate with high-CSR firms since these are less likely to engage in rent seeking activities (Lins et al., 2017). Additionally, we examine such relation in the subsample of non-SOEs without and with political connections. The coefficients in columns (2) and (5) which provide the results in the subsample without political ties are still positive and statistically significant, while those in column (3) and (6) using the subsample with political connections become insignificant. These results indicate that political connections may reduce trust and dampen the positive effect of CSR on trade credit, which further verify our main findings about the dark side of political connections.

Different measurement of political connections and trade credit
In this subsection, we redefine our key variables. On the one hand, we expand the definition of political connection variable to include chairs and CEOs who have working experience in the People's Bank of China or in the provincial branches of the largest four state-owned banks based on our baseline definition. 5 On the other hand, we use alternative measurements of trade credit, which is defined as the percentage ratio of net trade credit (accounts payable + notes payable -accounts receivable -notes receivable), scaled by total assets. Besides, we further control the firm's power in the supply chain to avoid the omitted variable issue. Overall, our main findings maintain when these issues are accounted for (unreported for brevity).

Conclusion
In this paper, we investigate the impact of political connections on firms' use of trade credit. Using a sample of listed Chinese firms over the period from 2003 to 2019, we find that firms with political connections exhibit less use of trade credit. More importantly, we find that this credit taking for connected firms stems from suppliers' concerns over limited contract enforceability. To obtain corroborating results, we further find that in areas with a relatively ineffective legal system and in firms dealing with powerful suppliers, the negative impact of political connections on trade credit taking becomes stronger. Furthermore, we investigate why politically connected firms exhibit a low credit reputation. We provide evidence that firms with political connections on average have a higher proportion of late or overdue payment of payables. In addition, connected firms are perceived as being a special group protected by government and treated with regulatory favours, which creates obstacles to the collection of trade credit outstanding through the legal system.
Taken together, our findings suggest that, when politically connected firms are involved in contract violations, the foreseeable contract enforceability could be weakened, and thus this could dampen suppliers' confidence in doing business with new customers. Our study contributes to the literature by investigating the impact of political connections on informal financing activities and by revealing the asymmetric influence of political connections on formal and informal financing channels.

Disclosure statement
No potential conflict of interest was reported by the author(s).