Does Merchant Guild Culture Reduce the Cost of Debt? Evidence from China

ABSTRACT Using hand-collected data on merchant guild culture and a sample over 2003–2019, this study examines the effect of merchant guild culture on the cost of debt. Our findings reveal that merchant guild culture is negatively associated with the cost of debt, suggesting that merchant guild culture rooted in ancient China continuously affects corporate behaviour, builds the borrower-lender trust relationship by strengthening ethical standards of borrowers, and eventually reduces the cost of debt. The above findings are robust to a variety of sensitivity tests and using the PSM approach, two-stage IV regression procedures and regression discontinuity design to address the endogeneity. Moreover, channel tests show that merchant guild culture reduces the cost of debt by depressing the default risks. Furthermore, our main findings are more pronounced for firms in regions with lower extent of Marketisation and higher risk taking, firms with lower managerial ownership and lower ratio of independent directors.


Introduction
Debt financing affects corporate strategies and net incomes to a great extent (Graham et al., 2008;Kim, Song, et al., 2011).The cost of debt represents the default risks endured by lenders.A branch of extant literature finds that the cost of debt can be affected by the default risks and information asymmetry, which bring about agency conflicts between borrowers and lenders, more external guarantees and covenant restrictions (Anderson et al., 2004;Costello & Wittenberg-Moerman, 2011;Graham et al., 2008;J. Zhang, 2008;Kim, Simunic, et al., 2011;Kim, Song, et al., 2011).To mitigate the borrower-lender conflicts, prior literature has validated that accounting information, governance mechanisms, managerial education experience, external audit, physical distance, political rights and corporate social responsibility (CSR) investment are associated with the cost of debt (Anderson et al., 2004;B. B. Francis et al., 2013;Bharath et al., 2008;C. Li et al., 2019;Milani, 2014;Pittman & Fortin, 2004;Yu et al., 2013;Zhou et al., 2017).
CONTACT Qiao Lin xmdx_lq@126.comAccounting Department, School of Management, Xiamen University Paper accepted by Donghua Chen.
In addition to formal institutions, another branch of the existing literature has validated that religious social norms, Confucian culture and other cultural dimensions can bring out higher ethical standards, and thus affect corporate behaviour and financial consequences (e.g. higher audit quality and better governance mechanisms; C. Li et al., 2019;Du et al., 2020;Du, 2013Du, , 2015;;El Ghoul et al., 2013b).For instance, El Ghoul et al. (2013b) find that firms in more religious counties enjoy cheaper financing costs.C. Li et al. (2019) find that firms with religious owners invest more in socially responsible items, which leads to lower cost of debt.As a matter of fact, as an important informal institution, culture is extremely stable and remains fairly unchanged from generation to generation (Guiso et al., 2006;Williamson, 2000).In this regard, culture can shape individual ethics, strengthen business ethics, and mitigate managerial unethical behaviour (Du, 2015(Du, , 2016;;Du, Weng, Zeng, & Pei, 2017).
Specific to the effect of merchant guild culture on corporate outcomes, Du, Weng, Zeng, and Pei (2017) validate that merchant guild culture is negatively associated with owner-manager agency costs, implying that merchant guild culture rooted in ancient China has continuous impacts on contemporary enterprises, strengthens managerial ethical behaviour, and mitigates owner-manager agency conflicts.Xiu and Zhou (2018) and Kanagaretnam et al. (2019) find the positive effect of merchant guild culture on corporate philanthropy, supporting Du, Weng, Zeng, and Pei's (2017) argument -merchant guild culture is inclined to bringing out higher ethical standards.Nevertheless, specific to the cost of debt, extant studies have provided insufficient evidence on whether ethical standards embedded in merchant guild culture can strengthen the borrowers' ethical standards, mitigate the borrowers' unethical behaviour, motivate the lenders to trust the borrowers, and eventually reduce the cost of debt.To fill the above gap, our study focuses on the Chinese context to explore the impact of merchant guild culture on the cost of debt.
The cost of debt depends on the demand side and the supply side.On the one hand, borrowers (debtors) surrounded by merchant guild culture atmosphere can be affected by the core spirits of merchant guild culture (e.g.honesty, credibility and righteousness) to a great extent.In other words, merchant guild culture mitigates managerial misconducts and enhances ethics and credibility of borrowers.Thus, merchant guild culture plays an important role in depressing borrowers' credit risks.Moreover, culture can remain fairly stable and unchanged for thousands of years, and only slightly fine tune, suggesting that culture can be well inherited (Williamson, 2000).Thus, the ethics embedded in merchant guild culture still affects the Chinese society and infiltrates into commercial thinking.On the other hand, creditors need both hard and soft information to judge the default risks on debt contracting.In addition to hard information in financial statements, soft information is crucial for lenders to screen borrowers and improve loan decisions (Campbell et al., 2019;Cassar et al., 2015;Petersen & Rajan, 1994).Previous studies (C.Li et al., 2019;Du, Weng, Zeng, Chang, et al., 2017) find that CSR information can help creditors to reduce information asymmetry and better CSR performance ensures lower cost of debt.Clearly, borrowers' trustworthiness and ethics (honesty, credibility and righteousness) belong to soft information, which provide incremental judgement evidence for lenders.Thus, we predict that lenders applaud borrowers surrounded by merchant guild culture, which are generally regarded as being more honest and trustworthy, resulting in lower risks within borrowers. 1  Using a sample of 25,621 firm-year observations from the Chinese stock market over 2003-2019, this study examines the influence of merchant guild culture on the cost of debt.In brief, our findings reveal several aspects: First, using the geography-proximitybased variables of merchant guild culture based on data mining, our study reveals that merchant guild culture is significantly negatively related with the interest rate, the proxy for the cost of debt, suggesting that merchant guild culture can exert significant effects on contemporary enterprises and reduce the cost of debt.Specifically, lenders are inclined to trust debtors in regions with strong merchant guild culture -which is validated to be related to higher ethical standards, and thus lenders always charge lower compensation for the default risks.Second, our finding is robust to alternative proxies for merchant guild culture and the cost of debt.Third, our conclusion is still valid after using the propensity score matching (PSM) method, two-stage instrumental variable (IV) Tobit-OLS regression procedures and regression discontinuity design (RDD) to address the endogeneity.Fourth, merchant guild culture is significantly negatively (positively) related with corporate misconducts (social trust).Fifth, merchant guild culture increases the possibility of both credit loans and secured loans, but merchant guild culture has a more pronounced impact on credit loans than on secured loans.Sixth, channel tests show that merchant guild culture reduces the cost of debt by depressing the default risks.Lastly, the effect of merchant guild culture on the cost of debt is more (less) pronounced for firms in regions with lower (higher) extent of Marketisation, firms in regions with higher (lower) risk taking, firms with lower (higher) managerial ownership and firms with lower (higher) ratio of independent directors.
Our study contributes to prior literature in several ways.First, to our knowledge and literature in hand, our study is the first to examine the influence of merchant guild culture on the cost of debt, adding to the existing literature about the impacts of merchant guild culture on financial behaviour and corporate decisions (Du, Weng, Zeng, & Pei, 2017;Kanagaretnam et al., 2019;Xiu & Zhou, 2018). 2 Moreover, our study shed light on the negative association between merchant guild culture and default risks, enriching the literature about the influencing factors on default risks, especially the impact of informal institutions on default risks (Bhimani et al., 2010;Shih et al., 2021).
Second, our finding suggests that lenders regard borrowers with better ethics as those with lower default risks.In this regard, our study contributes to prior literature about the role of soft information in bank decisions, echoing that soft information is a useful channel 1 As Bertomeu and Marinovic (2016) and Campbell et al. (2019) note, it is difficult to verify soft information, but soft information can enable lenders to better screen borrowers and optimise loan decisions.In this regard, merchant guild culture itself, which is relatively easy to be captured and be verified, is not appropriate to be viewed as soft information.However, without doubt, the connotations of merchant guild culture, which includes 'honesty and credibility' and 'making profits by the righteousness' in our study, belong to soft information.Moreover, Cassar et al. (2015) validate that trustworthiness is typical soft information because it is unverifiable and is difficult to quantify and interflow with others.Lastly, as a matter of fact, Geographic-proximity-based variables in our study capture the merchant guild culture atmosphere around a firm, which can borrow the support from prior literature (DeFond et al., 2018;Du, Weng, Zeng, & Pei, 2017;El Ghoul et al., 2013a).Please refer to the subsection of 'Merchant Guild Culture (The Independent Variable)'. 2 Du, Weng, Zeng, and Pei (2017) find that merchant guild culture motivates managers to upgrade the operation efficiency, mitigates owner-manager agency conflicts, and reduces owner-manager agency costs.Kanagaretnam et al. (2019) and Xiu and Zhou (2018) find that merchant guild culture is significantly positively associated with corporate philanthropy.
Our study differentiates itself from Du, Weng, Zeng, and Pei (2017) in two major aspects.First, two studies have different research focuses.Du, Weng, Zeng, and Pei (2017) examine whether merchant guild culture can reduce owner-manager agency costs, but our study argues the mitigating role of merchant guild culture in lenderborrower information asymmetry, and further validates that merchant guild culture significantly reduces the cost of debt.Second, the logics and arguments in two studies are different.Focusing on similar governance structure between merchant guilds and contemporary firms, Du, Weng, Zeng, and Pei (2017) argue that merchant guild culture hinders managerial unethical behaviour and reduces owner-manager agency costs.In comparison, our study argues that the connotations of merchant guild culture such as 'honesty and credibility' and 'making profits by the righteousness' as soft information provide incremental and valuable information for lenders, facilitate lenders to judge the default risks, and eventually promote lenders to charge lower cost of debt with firms surrounded by merchant guild culture.
Third, our study contributes to prior literature about the macro-level impacts of historical culture on corporate financial decisions, which can inspire follow-up studies to continuously explore whether various cultural dimensions affect corporate decisions, as well as the impact of merchant guild culture in medieval Europe on corporate behaviour in contemporary European firms.
Lastly, our study can lend important support to Williamson's (2000) social institution analysis framework.Specifically, our study provides strong and consistent evidence that merchant guild culture (an informal institution) can serve as an important information source and an alternative mechanism to formal institutions under weak institutional environment.As a result, lenders applaud firms surrounded by strong merchant guild culture atmosphere by charging a relatively low cost of debt.
Section 2 introduces institutional background and develops hypothesis, Section 3 elaborates the sample, data resources, model and variables.Section 4 reports descriptive statistics, Pearson correlation analysis, main findings and robustness checks.Section 5 includes endogeneity (additional) tests and cross-sectional analyses.The final section is conclusions, managerial implications and limitations.

Institutional background: merchant guild culture in China
In China's Song dynasty, merchant guilds had already shaped their embryonic forms.However, before the Ming dynasty, most commercial activities were separately, wellknown as 'only merchant without guild' (H.-P.Zhang & Zhang, 1993).During the Ming dynasty, national unification, convenient transportation and the silver monetisation provided crucial conditions for the development of merchant guilds (J.Fan, 2006).After reaching the peaks in the Ming and Qing dynasties, merchant guilds were of great power and magnitude, and then established a series of rules or doctrines interior and exterior (K.Liu, 1988;X. Liu, 2018).Specifically, merchant guilds regarded honesty and credibility (Chéngxìn/诚信 in Chinese) as their core spirit (H.-P.Zhang & Zhang, 1993;X. Liu, 2018).This creed had instructed and taught merchant guild members to be honest and never to breach the contracts, which can motivate merchant guild members to be ethical and honest.Moreover, merchant guild culture was inevitably influenced deeply by Confucian culture (X.Liu, 2018). 3Confucianism regards commerce as a social activity rather than profit making merely, which imparts merchant guild members to hold a long-term business operation view (X.Liu, 2018).In reality, as the germination of enterprises, merchant guilds in both ancient China and medieval Europe were consisted of shareholders, staffs and managers (Du, Weng, Zeng, & Pei, 2017;Yang, 2005).Thus, inevitably, the lenderborrower relation widely existed in merchant guilds.
In ancient China, due to the traffic restriction, people lived in a relatively closed environment, which brought out obstacles for the migration across different areas and motivated strong clan system to be formed.After the canal was dredged and opened up in the Ming dynasty, transport infrastructure was improved, and thus merchants spontaneously united and formed merchant guilds in some regions to resist the risks of remote operation (Brook, 1981;J. Fan, 2006).Clearly, most merchant guilds were established based on the consanguinity and geographic proximity, so the patriarchal clan is used as the governance method (Cai et al., 2008;H.-P. Zhang & Zhang, 1993;K. Liu, 1988).Against this context, most employers and employees belong to the same clan and they are restrained by the clan rules.Encouragement and punishment are linked with the promotion of status and prestige in the clan rather than pure money incentive, which is an implicit contract to affect their career concerns. 4For example, the Hui (Anhui) adopts admonition, kneeling punishment, cudgel, and banish from the ancestral temple as the methods for violating the clan rules (Cai et al., 2008).Moreover, if a merchant had deceived customers, supplies or/and counterparties, he would lose the deals and discredit the whole merchant guild, and thus other members never to cooperate or trade with him.To sum up, the habit of abiding by the clan rules are integrated in merchant 3 Merchant guild culture was deeply affected by Confucian culture, but some differences exist between merchant guild culture and Confucian culture.First, honesty and credibility in merchant guild culture are not exactly the same with xìn in Confucian culture.Xìn in Confucianism means that individual should keep his words (i.e.match words to deeds; Tan, 1967).In comparison, honesty and credibility in merchant guild culture emphasise to abide by the contract, to treat others sincerely, never to cheat or deceive customers, never to cajole (H.-P.Zhang & Zhang, 1993;X. Liu, 2018), and to pay close attention to honesty (other than only keep one's word), broadening the connotation of xìn in Confucian culture.Second, an important spirit in merchant guild culture is struggling hard (艰苦奋斗), which is not included in Confucianism.According to Cai et al. (2008) and J. Fan (2006), population without enough farmlands impel the emergence of merchant guild and spur merchant guild members to struggle hard.Struggling hard as merchant guild culture teaches members to be industrious and endeavour to overcome difficulties in nature, geography and business (X.Liu, 2018).As a comparison, the core virtues in Confucianism, well-known as five Constants (Rén, Yì, Lǐ, Zhì, Xìn), imply the obligation of humaneness and altruism for others, the moral disposition to do appropriately, the norms of proper social behaviour, wisdom in thoughts and actions, and keeping one's word, respectively (Du, 2015(Du, , 2016;;Tan, 1967;Zhu, 2015).Lastly, merchant guild culture is affected by the clan culture (X.Liu, 2018).Most merchant guilds were established based on the consanguineous and geographic proximity due to the inconvenient transportation in ancient (H.-P.Zhang & Zhang, 1993).In this regard, the culture of sharing weal or woe was gradually formed to integrate into merchant guilds, which differentiate itself from Confucian culture. 4 Gibbons and Murphy (1992) argue that career concerns are stronger than incentive concerns such as bonus or awards.In this case, compared with money, the promotion of status or prestige in the clan may bring out better incentive effect.
guild culture, so the clan governance as an implicit contract among merchant members plays a vital role in inhabiting member misconducts. 5n addition to the clan system, a merchant guild also establishes its core spirits or ideas to guide the operation and get along with people, which gradually evolve into a unique merchant guild culture.An important cultural dimension within merchant guilds is wellknown as credibility and honesty, which is a consensus and can be generally accepted by all merchant guilds.Before the Ming dynasty, the descendants of merchants were forbidden to attend school, but this limitation was abolished by the Emperor Zhu, the first Emperor of the Ming dynasty (X.Liu, 2018).The Emperor Zhu opened a channel upward through Kējŭ-the most important system of selecting officials in ancient China.Consequently, Confucian culture as the core content in Kējŭ (the imperial examination) affects people, and further is gradually integrated into merchant guild culture.Thus, merchant guild culture is shaped by Confucian philosophy to some extent such as Yì (righteousness) and RénÀi (humanity) (X.Liu, 2018).More importantly, some merchant guilds that operated in a wide range disseminated merchant guild culture to the whole nation and even neighbouring countries (e.g.Japan, Russia).Hence, as Du, Weng, Zeng, and Pei (2017) note, merchant guild culture has gradually been accepted by more and more stakeholders such as customers, suppliers, shareholders and lenders, which cultivates a social atmosphere and constrains merchant guild members to behave themselves.
Similar to Confucianism (Du et al., 2014;Du, 2015), merchant guild culture as an informal institution is fairly stable (Williamson, 2000), and thus can be inherited in contemporary firms.Today, merchant groups (商会), which are composed of entrepreneurs with geographical proximity or blood ties and are similar to merchant guilds, still exert significant effect on corporate behaviour in China.

Merchant guild culture and the cost of debt
The cost of debt denotes the compensation for the default risks borne by lenders (Du, Weng, Zeng, Chang, et al., 2017), and thus can be viewed as a function of corporate default risks (Black & Scholes, 1973).As such, lenders need reliable and relevant information to mitigate information asymmetry and evaluate the default risks of borrowers.It is difficult to objectively verify the value of forward-looking assets, which brings noises to evaluate default risks (Bertomeu & Marinovic, 2016).Thus, lenders turn to seek for soft information to validate the credibility of hard information (e.g.ethical information; Cassar et al., 2015;Du, Weng, Zeng, Chang, et al., 2017).Specific to the borrower-lender relation, when a firm provides doubtful financial statements (hard information), lenders are apt to search soft information to reduce information asymmetry and evaluate default risks (Dhaliwal et al., 2014;Kim, Simunic, et al., 2011).In this regard, borrowers affected by merchant guild culture (e.g.making profit by righteousness; honesty and credibility) are more likely to have less managerial misconducts and better moral standards, which can not only reduce the default risks for borrowers themselves, but also serve as important soft information to motivate lenders to trust hard information in financial statements of borrowers.
The cost of debt can be determined by both borrowers (the demand side) and lenders (the supply side) (D'aurizio et al., 2015;Goss & Roberts, 2011;Petersen, 2004).From the demand side, firms with better moral standards have lower default risks and better reputation, which reduces the possibility of default, builds up positive images, and accumulates reputational capitals.Du, Weng, Zeng, Chang, et al. (2017) validate that firms with superior environmental performance are inclined to be credible, trustworthy and ethical.The reasons lie in that better environmental protection embodies an arduous and long-term process, leads to higher reputation, and eventually lowers the cost of debt.Dhaliwal et al. (2014) verify that CSR disclosure is negatively related to the cost of capital, implying that stakeholders cherish ethical behaviour.From the supply side, lenders tend to consider both hard and soft information to value the default risks on debt contracting.Kim, Simunic, et al. (2011) argue that voluntary audit is a strong commitment to better quality of financial statements, and further find a negative effect of voluntary audit on the cost of debt, implying that voluntary audit as soft information can facilitate lenders to make lending decisions.D'aurizio et al. (2015) show that family firms have higher non-monetary costs of default, motivating lenders to value this soft information.Thus, the cost of debt can be affected from both the demand (borrowers) and supply (lenders) sides.
Specific to the demand (borrowers) side, merchant guild culture reduces managerial misconducts (Du, Weng, Zeng, & Pei, 2017), enhances ethical standards and reputation of borrowers, and eventually mitigates the default risks.First, 'honesty and credibility' as the core spirit within merchant guild culture shapes honest personality and establishes ethical standards among members,6 which motivate members to abide by contracts and never to escape debts.Members in merchant guilds cherish the reputation and strictly comply with the contracts during all their lives, which allow them to keep footholds in the ancient Chinese society (Du, Weng, Zeng, & Pei, 2017).In this regard, honesty and credibility can mitigate managerial unethical behaviour and the risks of bankruptcy, promote the confidence of stakeholders, enhance their reputation, and eventually reduce the default risks.
Second, members in merchant guilds pay special attention to make profits by the righteousness (义利), which teaches managers and employees to do what he deserves to do and never to forget the righteousness during they make profits.This spirit restrains merchants from grabbing profits through unethical ways (e.g.unethical managerial behaviour or breaking the laws) (Du, Weng, Zeng, Chang, et al., 2017;X. Liu, 2018).As such, the core spirit of 'making profits by the righteousness' can win the trusts from outsiders and reduce the risks of violation.As Du, Weng, Zeng, and Pei (2017) note, after being rooted in hearts of merchant guild members, the spirit of 'making profits by the righteousness' can be passed down and transmitted from generation to generation, and further continuously affects corporate behaviour in contemporary China.Thus, the effect of 'making profits by the righteousness' on individual behaviour has gone beyond merchant guild members and been extended to stakeholders in contemporary enterprises.Thus, firms surrounded by strong atmosphere of 'making profits by the righteousness' can be viewed as having greater commercial ethics and reputation, and further lenders charge them with lower default risks.
Lastly, merchant guilds are well organised under the clan system (culture), attaching the special importance to moral education (H.-P.Zhang & Zhang, 1993;Yang, 2005).Once a member runs counter to ethical doctrines in the patriarchal clan culture, he will have to face with serious penalty, and even have no place to live in his hometown.Also, once a member discredits or goes against the core values within a merchant guild, then peers, clan members and his relatives will feel shame on him, even break off the relation with him.In other words, merchant guilds highly value 'honesty and credibility' and 'making profits by the righteousness', and thus members will restrain themselves from unethical behaviour because immoral behaviour may heavily impair reputation, relative status in the clan, living condition, and even the offspring (Cai et al., 2008;K. Liu, 1988;Moll-Murata, 2009).7 Consequently, merchant guild culture are firmly kept in mind by members and form a dominated social atmosphere, in which ethical concerns are emphasised, more trusts are given to merchant guild members, and eventually lower default risks are associated with lower cost of debt.
As for the supply side on debt contracting, lenders are inclined to care about the possibility of repayment of loan (the security of fund) and the compensation for default risks borne by them (lenders).In addition to hard information in financial statements, soft information is useful for lenders to mitigate borrower-lender information asymmetry (Campbell et al., 2019;Du, Weng, Zeng, Chang, et al., 2017;Kim, Simunic, et al., 2011;Petersen, 2004).Compared with hard information, soft information provides more reliable indication about risks (especially in crisis period; D 'aurizio et al., 2015;Jiangli et al., 2008).Soft information includes ethical considerations about borrowers, which can affect lending decisions to some extent (Dhaliwal et al., 2014;Du, Weng, Zeng, Chang, et al., 2017).In this regard, the connotations of merchant guild culture-'making profits by righteousness' and 'honesty and credibility' (H.-P.Zhang & Zhang, 1993)-can provide lenders with sufficient confidence to believe in hard information in financial statements of firms surrounded by merchant guild culture.8Moreover, lenders applaud and are inclined to keep a long-term borrowerlender relation with borrowers with great creditworthiness (Petersen & Rajan, 1994).Merchant guilds were formed based on the consanguinity and geography (H.-P.Zhang & Zhang, 1993;X. Liu, 2018), and thus merchant guild members can affect each other to a great extent.As such, the borrower-lender trust on the basis of creditworthiness can make profits from the stable relationship and even acquire more clients (Degryse & Ongena, 2005;Du, Weng, Zeng, & Pei, 2017).Consequently, it is a win-win choice for lenders to charge lower cost of debt with firms surrounded by merchant guild culture atmosphere and maintain the long-term borrower-lender strategic relation.
To sum up, the connotations of merchant guild culture ensure commercial ethics and reputation, which can reduce the default risks.More importantly, merchant guild culture atmosphere around a firm as soft information can help lenders to mitigate information asymmetry, strengthen borrower-lender trust relationship, and thus reduce the cost of debt.Thus, we formulate Hypothesis 1 (H1) as below: H1: Ceteris paribus, merchant guild culture is negatively associated with the cost of debt.

Sample and data sources
The initial sample includes all Chinese listed firms over 2003-2019.Then, we select the sample based on the following criteria: First, we eliminate firm-year observations pertaining to banking, insurance and other financial industries.Second, we exclude firm-year observations with negative net assets.Third, we delete firm-year observations with unavailable data on the cost of debt.Lastly, we delete firm-year observations with missing firm-specific control variables.The final research sample includes 25,621 firm-year observations, covering 3,335 unique firms.All continued variables are winsorised at the top 1% and the bottom 1% of their origin values.Panel B of Table 1 reports the sample distribution by year and industry, revealing no serious year or industry clustering in our sample.
Data sources are reported as below (see Appendix 1 for details): (1) The data of merchant guild culture (MGC_R) is computed and hand-collected based on a well-known equation from geographic information system (GIS), referring to extent studies (Du, Weng, Zeng, & Pei, 2017;El Ghoul et al., 2013a).( 2

The cost of debt (Dependent Variable)
The cost of debt (CD) is defined as the interest expense for the year divided by average short-term and long-term debts during the year (Pittman & Fortin, 2004).Extant literature also uses the excess interest of debt (EIR)-the interest expense for the year subtract the interest expense calculated by benchmark interest rate divided by average short-term and long-term debts during the year -as the proxy of cost of debt (Anderson et al., 2004;Bharath et al., 2008;G. Li & Liu, 2009).In response, our study adopts CD and EIR to conduct main tests and robustness checks, respectively.

Merchant guild culture (The Independent Variable)
Prior literature adopts the descriptive, mathematical or historical method to measure merchant guild culture (Grief et al., 1994;Pearson, 1994).In recent years, a branch of growing literature (Du, Weng, Zeng, & Pei, 2017) has constructed geographic-proximitybased merchant guild culture variables on the basis of data mining.Referring to Du, Weng, Zeng, and Pei (2017), our study employs Google-earth map, the geographic information system (GIS) and hand-collected original data to measure merchant guild culture.First, we hand-collect the address of each firm's office in a specific year on the basis of annual report and the China Stock Market and Accounting Research (CSMAR).And then, the Google-earth map is used to collect the longitude and latitude of each firm's office address in a specific year.
Third, referring to Du (2015) and Du, Weng, Zeng, and Pei (2017) and on the basis of the longitude and latitude of each firm (merchant guild), we adopt Equation (1) and Equation (2) to calculate the geographic distance between a firm's office address in a specific year and the location of each merchant guild.
In Equation (1), λ F and Φ F (λ M and Φ M ) denote the longitude and latitude of a firm (a merchant guild), respectively.Then we calculate the central angle α using Equation (2) and Equation (3).
Lastly, referring to Du, Weng, Zeng, and Pei (2017), we construct geographic-proximitybased variables of merchant guild culture, labelled as MGC_R (R = 100, 200 km; R is the threshold to discern the number of merchant guilds within a radius of R kilometres around a firm's office address).MGC_R (R = 100, 200 km) can capture the quasi-firm-level merchant guild culture atmosphere.9

Empirical model specification
H1 predicts that merchant guild culture reduces the cost of debt.To test H1, we employ Model (1) to link the debt of cost, merchant guild culture and other determinants:- In Model (1), the dependent variable is the cost of debt (CD) and the main independent variable is MGC_R (R = 100, 200 km).If the coefficient on MGC_R is negative and significant, H1 is supported.
A set of control variables are included in Model (1) to isolate the effect of merchant guild culture on the cost of debt.First, due to bank credit favouritism for state-owned enterprises (Brandt & Li, 2003;Ge & Qiu, 2007), we include an indicator of STATE into Model (1).Second, Anderson et al. (2004) argue that governance mechanisms are significantly related with the cost of debt, so the percentage of shares held by the largest shareholder (FIRST), the percentage of shares held by top managers (MAN_SHR), CEOchairman duality (DUAL), board size (BOARD) and the ratio of independent directors (INDR) are added into Model (1).Third, following Pittman and Fortin (2004), Model (1) includes an indicator of auditor reputation.Fourth, firm-specific financial characteristics affect the cost of debt (Bharath et al., 2008;Goss & Roberts, 2011;Qi et al., 2010;Zhou et al., 2017), so we add firm size (SIZE), the ratio of liabilities with interests (LEV), investment opportunity (TOBINQ), the total amount of property, plant and equipment scaled by total assets (PPE), return on equity (ROE), the ratio of operating cash flows to total assets (OCF) and an indicator for negative net profits (LOSS) into Model (1).Fifth, following Bharath et al. (2008), LISTAGE is included into Model (1) to isolate the impact of a firm's listing age on the cost of debt.Lastly, year and industry dummies are included.See Appendix 1 for variable definitions.

Descriptive statistics
Table 2 reports descriptive statistics.The mean value of CD (the dependent variable) is 0.0828, meaning that the interest rate of the debt is about 8.28%.This finding is similar to Pittman and Fortin (2004) and Zhou et al. (2017).The mean value of MGC100 (MGC200) is 0.2290 (0.6151), revealing the number of merchant guilds within a radius of 100 (200) kilometres around a firm's office address.
As for control variables, on average and approximately, 44.80% of sample firms are state-owned enterprises (STATE), the proportion of shares held by the largest shareholder (FIRST) is 35.18%,managerial ownership (MAN_SHR) is 9.91%, CEO-chairman duality (DUAL) exists in 22.96% of sample firms, corporate boards (BOARD) include nine directors (e 2.2697 ), the ratio of independent directors (INDR) is 37.01%, 6.31% of sample firms are audited by big4 accounting firms (BIG4), firm size (SIZE) is 4.03 billion RMB (e 22.1172 ), financial leverage (LEV) is 19.35%, the market value of assets to the book value of assets (TOBINQ) is 1.8698, the ratio of fixed assets to total assets (PPE) is 24.05%, returns on equity (ROE) is 5.93%, the ratio of operating cash flows (OCF) is 4.22%, 10.59% of firms have negative net profits (LOSS) and the listing age (LISTAGE) is 11.0220, respectively.

Pearson correlation analysis
Table 3 reports results of Pearson correlation analysis.As shown in Table 3, there is a significant and negative relation between the cost of debt (CD, the dependent variable) and merchant guild culture (MGC_R, R = 100, 200 km; the main independent variable) at the 1% level, indicating that merchant guild culture reduces the cost of debt.This finding provides preliminary support to H1.
Furthermore, CD is significantly positively correlated with BIG4, SIZE, TOBINQ, OCF, LOSS and LISTAGE, but CD is significantly negatively correlated with FIRST, MAN_SHR, BOARD, LEV, PPE and ROE.Above results suggest the necessity of including these variables into our regression.Moreover, the coefficients of pair-wise correlation among control variables are low, and non-tabulated results show that all the variance inflation factors of control variables are less than 10, implying no serious collinearity after including these control variables in our regression model simultaneously. 10

Main findings
H1 predicts the negative relation between merchant guild culture and the cost of debt.In Columns (1) and (2) of Table 4, the coefficients on MGC_R (R = 100, 200 km) are both significantly negative at the 1% level (−0.0077 with t = −5.57and −0.0041 with t = −5.92),suggesting that merchant guild culture enhances corporate ethics, provide valuable soft information for lenders, leads to higher extent of lender-borrower trust, and eventually lenders charge firms surrounded by strong merchant guild culture atmosphere with lower cost of debt.The coefficient estimations on MGC_R (R = 100, 200 km) reveal that one- Note: All the variables are defined in Appendix 1.
10 Serious multicollinearity may happen when the variance inflation factors are larger than 10 (Gujarati, 2004).
(3) standard-deviation increase in MGC100 (MGC200) leads to 0.32% (0.49%) reduction in the cost of debt (CD), equalling about 1.41% (0.80%) of the mean value of CD.Clearly, this amount is also economically significant.Above results, taken together, validate H1.As for control variables, it is noteworthy several aspects: First, STATE is significantly negatively associated with CD, revealing that state-owned enterprises have lower interest rate, echoing extant studies about credit discrimination in China (Brandt & Li, 2003;G. Li & Liu, 2009;Ge & Qiu, 2007).Second, ROE has significantly negative coefficients, suggesting that higher returns on equity can brings about lower cost of debt.Third, the coefficients on LOSS are significantly positive, indicating that firms with negative net profits are considered with higher risks and lead to higher cost of debt.Lastly, SIZE, TOBINQ, OCF and LISTAGE are significantly positively related to CD respectively, while MAN_SHR and LEV are significantly negatively related to CD.

Robustness checks using excess interest rate as the proxy for the cost of debt
Referring to prior studies (Anderson et al., 2004;Bharath et al., 2008;G. Li & Liu, 2009), excess interest rate (EIR) is adopted as the proxy of the cost of debt to re-test H1.In Columns (1) and (2) of Table 5, the coefficients on MGC_R (R = 100, 200 km) are both significantly negative at 1% level (−0.0084 witht = −6.69,−0.0047 witht = −7.46respectively), additionally supporting H1.  White (1980).All the variables are defined in Appendix 1.

Robustness checks using dummy variables of merchant guild culture
In Table 6, we construct11 two dummy variables of MGC_DUM_R (R = 100, 200 km), equalling 1 if there are one or more merchant guilds within the radius R kilometres around a firm's office address and 0 otherwise, as the proxy for merchant guild culture to re-test H1.In Columns (1) and (2) of Table 6, MGC_DUM_R (R = 100, 200 km) are significantly negatively associated with CD at 1% level (−0.0077 with t= −5.57, −0.0071 with t= −5.97 respectively), providing further evidence for H1.

Robustness checks using the reciprocal value of the distance between a firm and the nearest merchant guild
Following Du, Weng, Zeng, and Pei (2017), we define MGC_DIS as the reciprocal value of the distance between a firm's office address and the nearest merchant guild, and then retest H1.In Table 7, the coefficient on MGC_DIS is significantly negative at the 1% level (−0.0746 with t = −6.00),validating H1 again. 12 White (1980).All the variables are defined in Appendix 1.

Endogeneity tests using propensity score matching method
Hong and Chin (2007) and Hong (2014) argue that market sizes, labour force and traffic conditions are important determinants to affect a firm's choosing the office address.Du, Weng, Zeng, and Pei (2017) also argue that a firm is inclined to select its office address in a peaceful place with strong merchant guild culture.Consequently, the potential endogeneity between merchant guild culture and the cost of debt may arise because geographic-proximity-based variables of merchant guild culture are calculated on the basis of a firm's office address and the locations of merchant guilds.In response, we firstly adopt the propensity score matching method (PSM; Dehejia & Wahba, 2002) to mitigate the potential endogeneity.
In doing so, referring to Hong and Chin (2007) and Hong (2014), several firm-specific variables of FIRST, BOARD, SIZE, LEV, TOBINQ, LOSS and LISTAGE are included into the first stage regression to predict the likelihood that a firm is affected by merchant guild culture in the first stage of PSM (i.e.MGC_DUM_R; R = 100, 200 km).
Panels A and B of Table 8 report the results of the balancing effects of propensity score matching.As t-tests show, before conducting the propensity score matching, the differences in the mean values of most variables are significant between two subsamples.However, after conducting the propensity score matching, the differences in the mean values of all variables between two subsamples become insignificant, suggesting a relatively good matching work.In Panel C of Table 8 that reports the results of the first stage of PSM, all variables are significantly related with MGC_DUM_R (R = 100, 200 km).

Endogeneity tests using two-stage Instrumental Variable (IV) tobit-OLS regression approach
We further adopt two-stage instrumental variable (IV) Tobit-OLS regression procedures to control for the endogeneity issue.First, we identify the welfare level (WELFARE) in a province where a firm is located -the number of staffs in welfare institutions scaled by total population in a province -as the instrumental variable.Theoretically, social welfare is positively associated with benevolent (Wong et al., 2006), the righteousness spirit in merchant guild culture.However, no prior literature has confirmed that social welfare can affect the cost of debt.Second, in the first-stage Tobit regression, we examine the effect of WELFARE on MGC_R (R = 100, 200 km), by which we can obtain the fitted value of merchant guild culture (i.e.MGC_R*; R = 100, 200 km).
In Columns (1) and (2) of Table 9, WELFARE is significantly positively associated with MGC_R (R = 100, 200 km), consisting with our prediction.In Columns (3) and (4) of Table 9, the coefficients on MGC_R* are both significantly negative, providing additional support for H1.White (1980).All the variables are defined in Appendix 1.

Endogeneity tests using regression discontinuity design (RDD)
Next, we use the regression discontinuity design (RDD) to further address the endogeneity issue.In doing so, we chose the 'the Qinling Mountain and the Huai River' line as the geographic cut-off point (the QH line hereinafter).Historically, most merchant guilds emerged and the commerce was more prosperous on the south side of the QH line in the Ming Dynasty (Du, Weng, Zeng, & Pei, 2017;T. Zhang, 1974;X. Liu, 2018).Hence, theoretically, the influence of merchant guild culture on corporate behaviour on the south (north) of the QH line is more (less) pronounced.Referring to Y. Chen et al. (2013), we obtain the median value of the latitude of QH line (i.e.33.64°) and employ it as the dividing line (the latitude span is 33.03°-34.25°).Then, we calculate the difference in the latitude between a firm's office address and the QH line (labelled as LATDIF), and the negative (positive) value of LATDIF means that the firm is located on the south (north) side of the QH line.
Figure 1 displays the scatter diagrams for the cost of debt (CD) and the difference in the latitude (LATDIF).The black (red) dots denote the cost of debt for firms that are located on the south (north) side of the QH line.In Figure 1, a discontinuity in CD around the threshold of 'LATDIF = 0' exists, and further the fitted line for the subsample on the south (north) side of the QH line is lower (higher).Figure 2a (2b) displays the scatter diagrams for both MGC100 (MGC200) and LATDIF.In Figures 2a and 2b, there are clear discontinuities in MGC_R around the threshold of 'LATDIF = 0', and further the fitted lines on the south (north) side of the QH line are higher (lower), implying that firms on the south (north) side of QH line are affected by merchant guild culture to a greater (less) extent.
Following Y. Chen et al. (2013), we construct Models (2)-( 4) as below: In Models ( 2)-( 4), NORTH is an indicator, equalling 1 if a firm is located on the north of the QH line and 0 otherwise.f (LATDIF) is a polynomial in LATDIF.Also, other control variables are the same as those in Model (1).MGC_R* denotes the fitted value of merchant guild culture from Model (3).Table 10 reports the results of using regression discontinuity design.Referring to Y. Chen et al. (2013), we use several different samples (subsamples) with varied bandwidth to run Model ( 2  Table 11 reports the results after using two-stage instrumental variable (IV) Tobit-OLS regression approach in which NORTH and the polynomial in LATDIF are used as instrumental variables.In the first stage (MGC_R as dependent variables; R = 100, 200 km), the coefficients on NORTH are significantly negative, indicating that firms on the north side of QH line are less likely to be influenced by merchant guild culture.In the second stage, MGC_R* is significantly negatively associated with CD, validating H1 again.Above results       Note: ***, ** and * represent the 1%, 5% and 10% levels of significance, respectively (two-tailed).All reported t-statistics are based on standard errors adjusted by White (1980).All the variables are defined in Appendix.
suggest that our main findings remain unchanged after using regression discontinuity design to mitigate the endogeneity issue.

Additional tests after including economic, financial and law factors
In addition to merchant guild culture, other regional factors may affect the cost of debt.To better capture the effect of merchant guild culture on the cost of debt, we further control for economic (GDP), financial (FINAN) and law (LAW) factors in the regression model.GDP denotes GDP per capita in the province where a firm is located.FIN and LAW represent the financial index and the legal enforcement index in a region where a firm is located (Wang et al., 2018), respectively.In Columns (1) and (2) of Table 12, the coefficients on MGC_R (R = 100, 200 km) are significantly negative, validating our hypothesis again.Moreover, three added variables have not significant effects on the cost of debt.
Table 11.Endogeneity tests using two-stage instrumental variable (IV) Tobit-OLS regression approach -the difference in the latitude between a firm and the QH line as the instrumental variable.

Variable
The first stage The second stage The dependent variable: MGC_R The dependent variable: Cost of debt (CD) Note: ***, ** and *represent the 1%, 5% and 10% levels of significance, respectively (two-tailed).All reported t-statistics are based on standard errors adjusted by White (1980).All the variables are defined in Appendix 1.

Additional tests about the relation between merchant guild culture and honesty
Our study argues that honesty as the core spirit of merchant guild culture can shape a social atmosphere of credibility, which improves the borrower-lender trust relation and mitigates the probability of misconduct.To ensure our argument, we construct two additional variables for additional tests: (1) MIS, a dummy variable, equalling 1 if a firm misconducts in a specific year and 0 otherwise.(2) TRUST, the commercial credit index (G.Chen et al., 2017) in a city where a firm's office is located.
As shown in Columns ( 1) and (2) of Table 13 where MIS is used as the dependent variables, the coefficients on MGC_R (R = 100, 200 km) are both significantly negative, indicating that fewer misconducts happen in regions with strong merchant guild culture atmosphere.In Columns (3) and (4) of Table 13 where TRUST is the dependent variable, MGC_R (R = 100, 200 km) has significantly positive coefficients, suggesting the positive effect of merchant guild culture on the social trust, lending important support to our argument.Overall, results in Table 13 affirm that merchant guild culture can increase the honesty and credibility, and eventually lead to lower cost of debt.White (1980).All the variables are defined in Appendix 1.

Additional tests about the effects of merchant guild culture on secured loans and credit loans
Another potential explanation is that firms affected by merchant guild culture may acquire secured loans, and thus mitigate the cost of debt.In response, we introduce secured loans (credit loans) with the label of SEC_LOAN (CRE_LOAN)-a dummy variable, equalling 1 if a firm possesses secured (credit) loans, and 0 otherwise.For SEC_LOAN (CRE_LOAN) as the dependent variable in Columns (1) and (2) [Columns (3)-( 4)] of Table 14, the coefficients on MGC_R (R = 100, 200 km) are both significantly positive, suggesting that merchant guild culture increases both secured loans and credit loans.Also, the coefficient estimation in Column (3) [Column (4)] show that the marginal effect of MGC100 (MGC200) on CRE_LOAN is 3.14% (1.51%), which is greater than the marginal effect of MGC100 (MGC200) on SEC_LOAN is 2.28% (0.99%) in Column (1) [Column (2)], implying that merchant guild culture has a more pronounced impact on credit loans than secured loans.13

Channel tests using default risks as the mediator
Our study argues that merchant guild culture hinders mangers from unethical behaviour (Du, Weng, Zeng, & Pei, 2017), firms surrounded by strong merchant guild culture atmosphere have lower default risks, and thus default risks can serve as a channel for the negative relation between merchant guild culture and the cost of debt.In response, we construct two additional variables to test the mediating role of default risks in the relation between merchant guild culture and the cost of debt: (1) DISTRESS (a dummy variable; Altman et al., 2010;Guan et al., 2016), equalling 1 if a firm's z-score is less than 0.5 and 0 otherwise; (2) EDP, the naïve default probability (Garlappi & Yan, 2011).First, we examine the impact of merchant guild culture on default risks.Second, we add both MGC_R (R = 100, 200 km) and default risks (DISTRESS or EDP) into regression models to identify the moderating roles.Table 15 reports the results of channel tests using default risks as the mediators.As shown in Columns (1) and (3) of Panel A, the coefficients on MGC_R (R = 100, 200 km) are significantly negatively related to DISTRESS, implying that merchant guild culture reduces managerial misconducts and mitigates default risks.In Columns (2) and (4) of Panel A, MGC_R (R = 100, 200 km) remains significantly negative coefficients, and the coefficients on DISTRESS are positive and significant.Moreover, the Sobel Z (the last line) is significant at 1% level, revealing the moderating role of DISTRESS in the negative relation between merchant guild culture and the cost of debt.Similarly, in Panel B of Table 15, we use EDP as the proxy for default risks and obtain qualitatively similar results to those in Panel A of Table 15, additionally supporting the mediating role of default risks.

Cross-sectional analyses considering marketisation and region-level risk taking
To address whether the negative impact of merchant guild culture on the cost of debt depends on macro-level factors, we conduct two cross-sectional analyses as below.
The full sample is divided into two subsamples on the basis of the mean value of provincial marketisation index (MKT): (1) the high-MKT subsample; (2) the low-MKT subsample (G.Fan et al., 2011;Wang et al., 2018).In Panel A of Table 16, for both the high-MKT and low-MKT subsamples, the coefficients on MGC_R (R = 100, 200 km) are both negative and significant.However, the coefficient differences tests (the penultimate line of Panel A) reveal significant differences between two subsamples.Furthermore, the Chi 2 -tests reveal that the absolute value of the coefficients on MGC_R (R = 100, 200 km; the last line of Panel A) for the high-MKT subsamples are significantly smaller than those for the low-MKT subsamples, suggesting that the negative effect of merchant guild culture on the cost of debt is less (more) pronounced for firms in regions with higher (lower) marketisation.
Regional risk-taking level may result in managerial short-term behaviour (Christensen et al., 2018), which may bring about higher default risks.In response, we divide the full sample into two subsamples on the basis of the mean value of region-level risk-taking level, proxied by the lottery sales per capita in a province:  White (1980).All the variables are defined in Appendix 1.
(1) the high risk-taking subsample; and (2) the low risk-taking subsample.As shown in Panel B of Table 16, on the whole, the negative effect of merchant guild culture on the cost of debt is more (less) pronounced for the high (low) region-level risktaking subsample.

Cross-sectional analyses considering managerial ownership and the ratio of independent directors
Next, we conduct two cross-sectional analyses to address whether the effect of merchant guild culture on the cost of debt depends on governance mechanisms.White (1980).All the variables are defined in Appendix 1.
Managerial ownership can align the benefits between mangers and shareholders (Fu & Lei, 2011).In response, based on the mean value of managerial ownership (MAN_SHR), the full sample is partitioned into high-MAN_SHR and the low-MAN_SHR subsamples.In Panel A of Table 17, the coefficients on MGC_R are significantly negative in Columns ( 2) and ( 4), but MGC_R has insignificant coefficients in Columns (1) and (3), suggesting that the negative effect of merchant guild culture on the cost of debt is more pronounced for firms with low managerial ownership.
Independent directors can play an external supervisory role (Fich & Shivdasani, 2007;Ma & Khanna, 2013).In response, the full sample is divided into two subsamples based on the mean value of the ratio of independent directors (INDR): (1) the high-INDR subsample; and (2) the low-INDR subsample.As shown in Panel B of Table 17, on the whole, the negative impact of merchant guild culture on the cost of debt is more (less) pronounced for the low-INDR (high-INDR) subsample.

Conclusions, managerial implications, and limitations
Focusing on the impact of merchant guild culture on the cost of debt, our findings reveal that merchant guild culture is significantly negatively associated with interest rate, the proxy for the cost of debt.Our findings suggest that merchant guild culture enhances corporate ethics and provides valuable soft information for lenders, and thus higher extent of lender-borrower trust on the basis of merchant guild culture ensures lower cost of debt.Our study adds to a branch of very thin but growing literature about the impact of merchant guild culture at the firm level on corporate behaviour.
Our study has several managerial implications.First, our findings suggest that merchant guild culture (an informal institution) can shape social atmosphere of trust and honesty, which is beneficial to the borrower-lender trust relationship and leads to the reduction in the cost of debt.As such, merchant guild culture can serve as a potential channel for lenders to evaluate credit rating.
Second, our study reveals that the core spirits (e.g.honesty and credibility) and the governance methods (e.g. the patriarchal clan system) of merchant guild culture still play a critical role in contemporary China and affect corporate behaviour.On the one hand, regulators such as the China Securities Regulatory Commission (CSRC) and investors should pay close attention to the influence of merchant guild culture on bank decisions and corporate behaviour.On the other hand, our study, as well as Du, Weng, Zeng, and Pei (2017) and K. Liu (1988), motivates practitioners to attach the importance to the potential influence of merchant guild culture on corporate governance and financial behaviour.
Third, our findings can inspire managers to attach the important to cultural factors (e.g.merchant guild culture).Merchant guild culture is fairly stable, continuous and elusive for centuries, and thus can affect lenders and lending decisions.Thus, corporate culture after integrating merchant guild culture can earn a better borrower-lender trust, which is related with a lower interest rate.
Lastly, our study suggests that soft information is valuable for lenders (banks) to screen the borrowers and optimise their lending decisions.In this regard, lenders should attach the importance to soft information, especially corporate ethics such as honesty and creditworthiness, to improve the efficiency of risk management and reduce the nonperforming loan ratio.
Our study has several limitations that can be further addressed.First, due to data limitation, our study uses geographic-proximity-based variables to address the impact of merchant guild culture on the cost of debt.Geographic-proximity-based variables are relatively objective, but it captures merchant guild culture atmosphere around a firm, rather than merchant guild culture itself or the underlying merchant guild spirits of managers.In this regard, further research can employ questionnaire to obtain the data on whether merchant guild culture exists in a specific firm or whether managers are affected by merchant guild culture.And further, it is better to combine geographicproximity-based data and questionnaire data explore the influence of merchant guild culture on corporate decisions.Second, our findings are based on Chinese context, so our findings may undergo the challenges of generalising to other contexts.In response, researchers need to adopt the international settings to validate our findings.
) Based on the China Stock Market and Accounting Research (CSMAR) and the People's Bank of China, we calculate the cost of debt (CD) and excess interest rate (EIR).(3) The data on welfare (WELFARE) are obtained from the China Civil Affairs Statistical Yearbook.(4) The data on social trust (TRUST) are obtained from G. Chen et al. (2017).(5) Other data on control variables are obtained from CSMAR.
year observations pertaining to the banking, insurance, and other financial industries −755 Eliminate firm-year observations whose net assets or shareholders equity are below zero −576 Eliminate firm-year observations whose data on cost of debt (CD) are unavailable −7,930 Eliminate firm-year observations whose data on firm-specific control variables are unavailable − Note: ***, ** and *represent the 1%, 5% and 10% levels of significance respectively All the variables are defined in Appendix 1.
Panel C: The first stage of propensity score matching Variable The dependent variable: The dummy variable of merchant guild culture ( ): (1) the full sample; (2) the subsample in which the values of LATDIF are less than±10°; (3) the subsample in which the values of LATDIF are less than±5°; (4) the PSM sample based on the full sample; (5) the PSM sample on the basis of '(2)'; and (6) the PSM sample on the basis of '(3)'.Panel A of Table 10 reports the balancing effects of propensity score matching, suggesting a relatively good matching work.As shown in Panels B and C of Table10, the coefficients on NORTH are significantly positive in all columns, providing further support to H1.

Figure 1 .
Figure 1.The scatter diagrams for both the cost of debt (CD) and the difference in the latitude (LATDIF).Note: The pink line denotes the fitted line for the subsample on the south side of the Qinling Mountain and the Huai River; The blue line denotes the fitted line for the subsample on the north side of the Qinling Mountain and the Huai River.

Figure 2 .
Figure 2. (a) the scatter diagrams for both merchant guild culture (MGC100) and the difference in the latitude (LATDIF) (b) the scatter diagrams for both merchant guild culture (MGC200) and the difference in the latitude (LATDIF).Note: The pink line denotes the fitted line for the subsample on the south side of the Qinling Mountain and the Huai River; The blue line denotes the fitted line for the subsample on the north side of the Qinling Mountain and the Huai River.
of debt, measured as interest expenses divided by the average short-and long-term debts.CSMAR MGC_R= Geographic-proximity-based variables of merchant guild culture in China, measured as the number of merchant guilds within a radius of R kilometres (R = 100, 200 km) around a firm's office address.HandcollectedSTATE= A dummy variable, equalling 1 if the ultimate owner of a Chinese listed firm is a (central or local) government agency or government-controlled enterprise and 0 otherwise.CSMAR FIRST = The percentage of common shares held by the largest shareholders.CSMAR MAN_SHR = The total percentage of shares held by a firm's top management team.CSMAR DUAL = A dummy variable, equalling 1 if the chairman and the CEO are the same person and 0 otherwise.CSMAR BOARD = Board size, measured as the natural logarithm of number of directors in the boardroom.CSMAR INDR = The ratio of independent directors, measured as the number of independent directors scaled by the total number of directors in the boardroom.CSMAR BIG4 = A dummy variable, equalling 1 if a firm is audited by Big4 and 0 otherwise.CSMAR SIZE = Firm size, measured as the natural logarithm of a firm's total asset.CSMAR LEV = Financial leverage, measured as the debt with interests scaled by total assets.CSMAR TOBINQ = A firm's growth opportunity, measured as a firm's market value to its book value.CSMAR PPE = The total amount of property, plant and equipment scaled by total assets.CSMAR ROE = Return on equity, measured as a firm's net income scaled by total equity.CSMAR OCF = Operating cash flow, measured as a firm's net operating cash flow divided by total assets.CSMAR LOSS = A dummy variable, equalling 1 if a firm's net income is below zero in the specific year and 0 otherwise.CSMAR LISTAGE = The number of years since a firm's IPO.CSMAR MGC_DUM_R = A dummy variable, equalling 1 if there are one or more merchant guilds within a radius of R kilometres (R = 100, 200 km) around a firm's office address.Handcollected MGC_DIS = The reciprocal value of the distance between a firm and the nearest merchant guilds.Handcollected EIR = Excess interest rate, measured as the interest expense for the year subtract the interest expense calculated by benchmark interest rate divided by its average short-and long-term debt during the year. of staffs in welfare institutions scaled by total population in a province.CCASY NORTH = A dummy variable, equalling 1 if a firm is located on the north of the QH line (the Qinling Mountain and Huai River line), and 0 otherwise.Calculated LATDIF = The difference in the latitude between a firm's office address and the QH line (the Qinling Mountain and Huai River line).Calculated SEC_LOAN = A dummy variable, equalling 1 if a firm possesses secured loans and 0 otherwise.CSMAR CRE_LOAN = A dummy variable, equalling 1 if a firm possesses credit loans and 0 otherwise.CSMAR GDP = GDP per capita in the province where a firm is located.CSY FIN = The financial index in a region where a firm is located.Wang et al. (2018) LAW = The legal index in a region where a firm is located.Wang et al. (2018) DISTRESS = A dummy variable, equalling 1 if a firm's z-score is less than 0.5 and 0 otherwise.(Guan et al., 2016) Calculated EDP = Naïve default probability calculated following Garlappi and Yan (2011) Calculated CSMAR: China Stock Market & Accounting Research Database; CSY: China Statistical Yearbook; CCASY: China Civil Affairs Statistical Yearbook.

Table 4 .
The influence of merchant guild culture on the cost of debt.

Table 5 .
Robustness checks using excess interest rate as the proxy for cost of debt.

Table 6 .
Robustness checks using the dummy variables of merchant guild culture.

Table 7 .
Robustness checks using the reciprocal value of the average distance between a firm and the nearest merchant guild.

Table 8 .
Endogeneity tests using the propensity score matching approach.

Table 10 .
Endogeneity tests using regression discontinuity design.

Table 12 .
Additional tests after including regional economic, financial and law factors.

Table 13 .
The effects of merchant guild culture on corporate misconduct and social trust.Note: ***, ** and * represent the 1%, 5% and 10% levels of significance, respectively (two-tailed).All reported t-statistics are based on standard errors adjusted byWhite(1980).All the variables are defined in Appendix 1.

Table 14 .
Additional tests about the effects of merchant guild culture on secured loans and credit loans.

Table 15 .
Channel tests using default risks as the mediators.

Table 16 .
Cross-sectional analyses considering marketisation and region-level risk-taking.

Table 17 .
Cross-sectional analyses considering managerial ownership and the ratio of independent directors.