What determines firms’ access to credit in the absence of effective economic institutions: evidence from China

The existing literature suggests that economic institutions determine the allocation of resources for economic growth. As an important counterexample, although China has one of the world’s fastest-growing economies, its legal and financial systems are underdeveloped. With evidence from China, the author confirms that government intervention positively and causally determines firms’ access to credit. The author further provides evidence that government intervention enables firms’ profit through facilitating access to credit. This evidence confirms that the mechanism of government intervention allows firms’ access to credit and then enables the firms to obtain relatively large profits. Ultimately, this paper reveals that, in the absence of effective economic institutions, government intervention determines firms’ access to credit. JEL O17 G21 G28 C51


Introduction
The existing literature suggests that economic institutions determine the allocation of resources for economic growth (North 1990). The endogenous growth theory states that -the fundamental explanation of comparative growth is differences in [economic] institutions‖ (Lin et al. 2010: 49). As an important counterexample, although China has one of the fastest-growing economies in the world, its legal and financial system is underdeveloped (Allen et al. 2005). Given that China has among the weakest protection of property rights (La Porta et al. 2004), what determines the allocation of credit by financial intermediaries in this rapid economic growth? In microeconomic terms, what determines firms' access to credit from financial intermediaries?
This microeconomic question presents the existing literature with a challenge.
The recent literature on law, finance, and economic growth explains the impact of economic institutions on economic growth, but it offers no insight into our research question. Allen et al. (2005) explain China's growth puzzle by revealing the important contribution of the informal sector on that growth, while Ayyagari et al. (2010) and Linton (2006) analyze formal and informal financing. Other scholars investigate the impact of a firm's government connection (e.g., Cull et al. 2015) or bank relationships (e.g., Ongena et al. 2011) on its corporate finance. 1 However, these scholars do not discover what drives the formal financing mechanism when legal institutions are un- 1 The government or political connection is only one type of government intervention. The endogenous connection is one that a firm deliberately constructs with governments or politicians while the exogenous connection is represented by the state share. These two connections represent the institutional property of a firm in the short or long run, respectively. This paper is motivated to explore how government intervention functions in place of a market system. Thus, we need to capture the general government intervention issue (instead of government connections), which is commonly measured by the interaction of the surveyed firm with governments (see Section 3). developed. Even when scholars (e.g., Friedman et al. 2000) explain the underground economies or government intervention (e.g., Chen et al. 2011;Easterly and Levine 1997;La Porta et al. 1999;Shleifer 1997), they do not explore how financial resources (i.e., capital) are allocated when legal systems are ineffective.
Moreover, the existing institutional literature can explain the failure of institutional reforms, but it neglects financial development in the absence of effective legal institutions. In fact, economic institutions have attracted economic scholars only since Coase (1937). The follow-up economics of contract (e.g., Cheung 1970Cheung , 1974Cheung , 1983Cheung , 1998) and transaction cost economics (e.g., Williamson 1976Williamson , 1979Williamson , 1985Williamson , 2000 explain the effect of formal institutions and relational contracts, respectively. The endogenous growth theory (e.g., Acemoglu et al. 2002Acemoglu et al. , 2005Barro 1990;Beck et al. 2005; Claessens and Laeven 2003) emphasizes the impact of economic institutions on economic growth. Recently, legal origin theory (e.g., La Porta et al. 1997Porta et al. , 1998Porta et al. , 2008 and moral economics (e.g., Alesina and Giuliano 2015; Tabelini 2008) enrich the impact of economic institutions with consideration of legal origin and culture, respectively. These institutional theories all neglect the topic of resource allocation without an effective legal system. This paper posits that government intervention replaces legal institutions in allocating capital (La Porta et al. 1998). Legal institutions in developing countries have been underdeveloped for a long time, but local governments control the financial system and tend to facilitate investment in economic growth. For example, China's local governments construct strategic alliances with financial intermediaries (e.g., banks) and firms in making investment decisions. The strategic alliance guarantees the promotional effect of government intervention on firms' access to finance (Wang 2007).
To reveal the resource allocation function of government intervention, we first confirm that government intervention enables firms' access to credit in China. We then show that government intervention enables firms to obtain relatively large profit.
We further confirm that government intervention has a promotional effect on profit through access to credit. Ultimately, we reveal a mechanism of government intervention that facilitates firms' access to credit and then profit. Our findings are robust to the potential endogeneity issue, different estimation methods, and different types of standard errors. This paper contributes to the institutional economics in the micro government intervention issue. Because the public choice school exposes the impact of corruption (e.g., Alesina et al. 1992;Becker 1983;Tanzi and Schuknecht 2000), the academic world consistently views government intervention as a -grabbing hand.‖ With reference to the -Washington consensus,‖ government intervention is considered almost equivalent to corruption (see International Monetary Fund 2002;World Bank 1997, 2004. The Washington consensus has been objectively challenged by Hopkin and Rodriguez-Pose (2007). Specifically, governments act as a grabbing hand only within a particular institutional structure (Di Tella and Schargrodsky 2003;Goel and Nelson 1998;Mauro 1998;Tanzi and Davoodi 1997)  This paper also contributes to the literature on corporate finance. To the best of our knowledge, we offer the first attempt to explore the micro effect of government intervention on a firm's formal financing, not to mention firms' access to credit. Many scholars of development economics or institutional economics study government structure (e.g., Di Tella and Schargrodsky 2003; Van Rijckeghem and Weder 2001), but they overlook government intervention in formal economies. In particular, the resource allocation function of government intervention discussed in this paper can be generalized for all developing countries. For example, it helps explain the failure of the Russian government's economic reforms. The Russian government adopted radical reform policies in property rights protection but also lost control over its economic system. Accordingly, the new (and strange) economic institutions and government fail to properly allocate Russian economic resources.
Section 2 designs a theoretical framework to explain the helping hand of government intervention in firms' financing. Section 3 explains our data and variables.
Section 4 presents the model, and Section 5 reports our empirical results on the promotional effect of government intervention on firms' access to credit. Section 6 concludes.

Background: Financial intermediaries and firms
China's financial system consists of financial intermediaries, 2 stocks, bonds, and venture capital, but Chinese firms rely heavily on loans for financing. First, since its inception in 1992, China's stock market has grown rapidly, but the scale is still small relative to that of the banking sector (Allen et al. 2005 Moreover, all of China's financial intermediaries are controlled by the government in whole or in part. The government partially owns and fully manages the banking sector and other financial intermediaries (Ayyagari et al. 2010). The country has 2 Financial intermediaries are also called financial institutions. Note that -institutions‖ as in -financial institutions‖ differ from -institutions‖ as in -economic institutions.‖ The former mean intermediaries as organizations; the latter involve an abstract structure for property rights protection and contract enforcement. To avoid expository confusion, we use -financial intermediaries‖ instead of -financial institutions‖; we only use -institutions‖ to mean -economic institutions.‖ only -two nominally private banks, both [of which] are dominated by state shareholders and management‖ (Linton 2006: 4). Thus, government intervention can affect financial intermediaries' credit supply.
These two facts drive China's local governments to construct strategic alliances with financial intermediaries and firms (Wang 2007), which has two consequences.
First, local governments have the power to affect firms' financing. Local governments have been empowered through three phases of administrative decentralization, one of which delegated to them administration over financial intermediaries and other organizations. Moreover, local governments have an incentive to facilitate firms' access to credit. Before 2012, every chief official in local government had to achieve a GDP growth target determined by the government department above it. Hence, facilitating investment is the most important goal of local governments. One illustrative clue is that banks were under so much pressure from local governments to support investment that 30-40% of bank loans were nonperforming in 2006 (Economist Intelligence Unit 2006).
Second, financial intermediaries are willing to support investments that involve government intervention because they are assured of being bailed out by local governments. In the entire history of the People Republic of China, only one bank has been permitted to go bankrupt. Thus, financial intermediaries undertake no business risk due to government intervention. For these reasons, local governments, firms, and financial intermediaries have mutual incentives with respect to the extension of loans; in effect, a strategic alliance among these three parties is generated through the mechanism of government intervention.

Theoretical effects and hypotheses
Through this alliance structure, government intervention promotes firms' access to credit in the following way. First, local governments contribute -critical inputs‖ (Byrd 1990;Chang and Wang 1994;Naughton 1992Naughton , 1994 such that government intervention reduces the financial cost of firms, thus firms will seek government intervention to gain access to credit. Second, government intervention can also effectively constrain the financial cost in the broad sense because it reduces ex ante and ex post uncertainty. As Cai et al. and continuously expanded after the fiscal decentralization empowering local governments to control financial intermediaries (Oi 1992(Oi , 1999Qian and Weingast 1996;Wong 1992). Based on the foregoing, we posit the following hypothesis.

Hypothesis 1: Government intervention enables firms' access to credit.
We distinguish between real access to credit and the ability of a firm to obtain access to credit. Real access is indicated by the existence of loans to a firm from financial intermediaries, whereas the ability to access is inversely reflected by the ratio of collateral required for loans. A higher collateral ratio is required when a firm has a lower ability (or potential) to obtain access to loans. To further reveal the mechanism of government intervention, we need to investigate the effect of government intervention on a firm's performance. Given the importance to firms of access to credit (Allen et al. 2005; Ayyagari et al. 2010), we expect to find a positive relationship between government intervention and firm profit.

Hypothesis 2: Government intervention facilitates profit for firms.
Moreover, considering that government intervention reduces firms' financial cost in a narrow sense or in a broad sense, it should help firms to access credit and then to generate better performance than other similar firms without that access.
Hypothesis 3: Government intervention enables profit for firms through facilitating their access to credit.
Hypotheses 1 predicts a causal effect of government intervention on firms' access to credit. We also propose Hypotheses 2-3, and these three hypotheses together provide a coherent theory of the government intervention mechanism. In other words, Hypotheses 1-3 examine whether access to credit is a significant mediator variable for the promotional effect of government intervention on firm profit. Thus, we posit two aggregate hypotheses as follows.

Data and variables
The data on Chinese firms come from the World Bank Investment Climate Sur-vey, undertaken in 2005. 3 The survey samples from the universe of registered businesses and follows a stratified random sampling methodology. The survey comprises 12,400 firms located across 120 cities in 30 provinces. The firms surveyed are in 30 types of manufacturing industries. All variables are updated to 2004. The descriptive statistics and the correlation matrix for the main variables are reported in Tables 1 and   2, respectively.

Access to credit (dependent variable)
We construct a dummy variable to measure whether the firm has access to credit. Alternatively, we use the ratio of collateral value (relative to the potential loan) to measure the ability of a firm to obtain access to credit. The collateral ratio inversely reflects the trustworthiness of the firm according to financial intermediaries. As Hypothesis 1.2 suggests, the collateral ratio should negatively respond to government intervention.
In a comparison between these two variables, the dummy for loans objectively reflects real access to finance, whereas the collateral ratio inversely indicates the ability of the surveyed to obtain access to credit.
Government intervention and access to credit are distributed differently across cities. To save space, we report only the top ten and bottom ten cities with government intervention and access to credit, respectively. As Table 3 shows, in the top ten cities, government intervention reaches around rank 3, whereas the dummy for loans equals almost 1 and the collateral ratio is at least 7.387. However, in the bottom ten cities, government intervention reaches only rank 1 or 2. Similarly, the dummy for loans is available for less than 40% of firms and the collateral ratio is only 2 or 3. Simply speaking, Table 3 shows that government intervention and access to credit vary greatly across cities in China.

[Insert Table 3 about here]
Comparing Table 3 with Table 1, we find little firm heterogeneity at the city level. Table 1 shows that the largest value of government intervention is 8, but the largest value of government intervention at the city level (in Table 3) is only 3.16. Similarly, the maximum value of the dummy for loans (or collateral ratio) is 1 (or 10) across firms, but the variable of access to credit at the city level is at most 0.92 (or 8.771).
Therefore, individual firm properties are not negligible, which also motivates us to including the following control variables.

Control variables
We include two types of control variables, firm characteristics and CEO characteristics. With respect to firm characteristics, we first control for firm age. The survey provides the establishment year of the firm, so we can obtain the firm age in 2004.
Second, we use the log of total income to control for firm size, as other scholars (e.g., Cai et al. 2011;Lin et al. 2010) have done. In particular, we select the total income value in 2003 to reduce potential reverse causality. Third, we also control for whether the firm has export sales and the ratio of the state share to the foreign share in the ownership structure. Export firms tend to benefit from preferential policies and then have better access to credit. The state share represents the firm's relationship with the government in terms of receiving benefits (Cull et al. 2015), whereas the foreign share can indicate benefiting from a preferential government policy (Lemoine 2000). However, these two ratio variables can also be negatively related to access to credit be-cause a firm with a state share or a foreign share can rely more on informal financing, such as trade credit, because of government support (Ayyagari et al. 2010).
With respect to CEO characteristics, we design three variables. First, we control for whether the CEO is appointed by the government. A firm with a government-appointed CEO can also strategically use informal financing to rely less on access to credit thanks to its close government ties. Moreover, we control for the education and tenure of the CEO, respectively. These three characteristics should be beneficial to firm access to credit and to firm performance because they create social capital for firms and thus aid them in gaining support (Narayan et al. 2000;World Bank 1998).

The model
We test the relationship between government intervention and a firm's access to credit by estimating the following equation: where refers to access to credit, specifically the dummy for loans or the collateral ratio, is government intervention, and represents the matrix of control variables. We use the Probit method to regress the dummy for loans and the Tobit method to regress the collateral ratio, which has no negative values. To reduce the potential for omitted variables, we also control for city-and industry-fixed effects. We estimate Equation (1) using two types of standard errors. First, we use robust standard errors to avoid the heterogeneity issue. Second, we use cluster standard errors to reduce the heterogeneity issue across different firm groups. Considering that we have controlled for the city-fixed effect, we control for cluster standard errors at the county level.
Despite the issue of omitted variables, we admit the coefficient of interest may be biased by potential reverse causality. We use the IV probit method (Rivers and Vuong 1988) or IV Tobit method to estimate the following equations.
in Equation (2) is the fitted value of , which is estimated from Equation (3). in Equation (3) is government intervention in public security matters. 4 For expository convenience, we call it -public security intervention.‖ The investment climate survey asks the firm manager how many days the firm needs to spend on interaction with the government regarding public security matters. We use the response to measure public security intervention. According to our definition, is positively related to government intervention ( ). The former reflects government intervention in public security matters while the latter indicates general government intervention.
Moreover, is irrelevant for loans between firms and financial intermediaries.
Thus, can be a suitable IV for . In IV estimations, we also use robust standard errors or cluster standard errors.
To test Hypothesis 2, we originally estimate the following equation.
Considering the clear reverse causality between firm profit and government intervention, we then estimate the following equation with an IV.
where is fitted with the variable according to Equation (3); is an indicator reflecting that the firm's per capita profit is larger than the median value of a firm's per capita profits in the same city and industry. As a reference, we also use the indicator that firm profit is not smaller than the median value. Specifically, is defined as follows. where , , represents a firm, city, and industry, respectively. For later regressions, our dependent variable is the profit dummy, . Considering that the mean value may be biased by data skewness, we select the median value instead of the mean value as a benchmark in the definition of . In these estimations, we also use robust standard errors or cluster standard errors.
To test Hypothesis 3, this section examines whether government intervention has an indirect promotional effect on a firm's profit through the firm's access to credit.
Considering that Equations (1) and (4) have tested the effect of government intervention on access to credit and profit, we only need to examine whether access to credit significantly explains profit with government intervention included in the regression.
where is the variable of access to credit (the dummy for loans or the collateral ratio) and is government intervention. We use robust standard errors or cluster standard errors as before. The significance of the mediator variable requires that, first, despite in Equation (1) and α in Equation (4), in Equation (5) is significant; second, in Equation (2) and in Equation (5) are different.
To objectively measure the significance of the difference between in Equation (2) and in Equation (5), we refer to z-tests in the frameworks of Baron and Kenny (1986), Goodman (1960), and Sobel (1982, respectively. z 1 = * 2 * 2 + 2 * 2 + 2 * 2 (6) z 2 = * 2 * 2 + 2 * 2 − 2 * 2 (7) where is the marginal effect of government intervention at its average value and is the marginal effect of access to credit at its average value (i.e., average partial effect; see Wooldridge 2002). and are the standard deviation of government intervention and access to credit, respectively.
Because the Probit (or Tobit) method used in Equations (2), (4), and (5) follows the maximum likelihood estimator, the inclusion of the mediator variable will alter the coefficient of . The change in coefficients includes differences in effects and differences in scale parameters. To exclude the difference due to the rescaling problem, we adopt average partial effects to calculate the z-test score as Woodridge (2002) suggests.

Government intervention and access to credit
We report baseline estimates for Equation (1) in Table 4. As Table 4 shows, the coefficient of government intervention is positively (or negatively) related to the dummy for loans (or the collateral ratio). In particular, the coefficient is highly significant at the 1% level. This finding confirms Hypotheses 1.1 and 1.2, thereby support-ing Hypothesis 1.

about here]
Moreover, all coefficients of control variables are significant except firm age.
Among the significant control variables, the state share, the foreign share, and government appointment of the CEO are negatively (or positively) related to the dummy for loans (or collateral ratio); the others are positively (or negatively) associated with the dummy for loans (or collateral ratio). The corresponding signs associated with the dummy for loans are the opposite of the one for collateral ratio. The opposite signs are also expected. The negative (or positive) coefficients of the state share, the foreign share, and having a government-appointed CEO reflect that these three variables raise the probability that firms will have real loans and reduce the collateral ratio for potential loans, respectively. Generally speaking, these three variables constrain firms' access to credit.
To deal with the endogeneity issue, we conduct IV estimations according to Equations (2) and (3). We report first-stage estimates in Columns 1 and 2 in Table 5; to offer additional insight, we also report reduced-form estimates in Columns 3 and 4 in Table 5.
As Columns 1 and 2 in Table 5 show, the coefficient of public security intervention is positively and significantly related to government intervention. The p-value of the chi-square test equals zero, which indicates that data for the first-stage estimates have a good fit. Reduced-form estimates (Columns 3-6 in Table 5) show that public security intervention has a strong positive (negative) relationship with the dummy for loans (or the collateral ratio). Both the first-stage and reduced-form estimates are stable across the various specifications.

about here]
We reports second-stage estimates in Table 6. As Table 6 shows, the coefficient of government intervention is significant in an IV framework. As found earlier, government intervention is positively related to the dummy for loans and negatively associated with the collateral ratio, respectively. Moreover, all coefficients obtain the same signs as in Table 4. Simply speaking, our IV estimates have the same findings as the basic estimates in Table 4. Therefore, our estimates are robust to the potential endogeneity issue.
[Insert Table 6 about here]

Government intervention and profit
To test Hypothesis 2, we conduct estimations according to Equation (4)' and report IV estimates 6 in Table 7. As Table 7 shows, the variable is positively and significantly related to the variable across different specifications. This indicates that government intervention enables firms to obtain a sizable profit, which supports Hypothesis 2.

about here]
Moreover, all control variables are significant. Among these control variables, firm age, state share, and having a government-appointed CEO are negatively related to the profit probability ( ), whereas the others are positively associated with the profit probability. In comparison to regressions in Table 6, all control variables have the same signs for profit and access to credit except firm age and the foreign share. These consistent signs indicate that these control variables have the same effects on profit and access to credit.
The difference in sign for firm age (or the foreign share) between Tables 6 and 7 is actually logical. First, an older firm has more social capital that it can use for access to external resources, but it has a weaker response to the external environment (e.g., Argyres and Silverman, 2004) and tolerates poorer performance. Thus, an older firm has good access to credit even though it has low profit. Second, the foreign share allows some managerial advantages (e.g., Organization for Economic Cooperation and Development 2002; Shleifer and Vishny 1986) but creates disadvantages in terms of access to external finance (e.g., Linton 2006). That is why having a foreign share enlarges firms' profit but tends to constrain a firm's access to credit.

Government intervention, access to credit, and profit
To test Hypothesis 3, we conduct estimations according to Equation (5) and report the results in Table 8. As Table 8 shows, the coefficient of access to credit is highly significant. Moreover, the government intervention coefficient is still significant but changes after access to credit is controlled for. Furthermore, all control variables used in Table 7 are significant and obtain the same signs as in Table 8. This suggests that our estimates are highly robust.

about here]
To objectively identify the significance of the mediation effect of access to credit, we conduct z-tests in the frameworks of Baron and Kenny (1986), Goodman (1960), and Sobel (1982. We report the z-tests in Table 9. Panel A reports z-test results for the regressions under the Probit method, whereas Panel B presents z-test results for those under the Tobit method. As Table 9 shows, with each type of standard errors (robust or clustered), estimation method (Probit or Tobit), and z-test version, the z-test result is 1.960 or greater. Namely, all z-test results confirm that the mediator variable of access to credit is significant at least at the 5% level. Specifically, when access to credit is captured by the dummy for loans, significance even reaches the 1% level, whereas significance still reaches 5% when access to credit is captured by the collateral ratio.
[Insert Table 9 about here] The findings from Tables 8 and 9 confirm that government intervention enables a firm's profit through the firm's access to credit (including the dummy for loans or the collateral ratio). Put differently, Hypothesis 3 is not rejected.
As mentioned earlier, our results confirm that government intervention causally and significantly facilitates a firm's access to credit and profit, respectively. We also document that government intervention has an indirect promotional effect on a firm's profit through the firm's access to credit. In sum, our empirical findings jointly confirm that government intervention enables a firm's access to credit and then the firm's profit. Therefore, our results jointly support Aggregate Hypothesis 1.

Conclusions
The existing literature suggests that economic institutions determine the allocation of resources in economic growth. As an important counterexample, China has the largest and one of the world's most rapidly growing economies, but its legal and financial systems are underdeveloped. This paper explored what determines firms' access to credit in countries whose legal and financial systems are underdeveloped.
To capture firms' access to credit, we measure real access to credit and a firm's ability to obtain access to credit. Real access is indicated by the existence of loans to a firm by financial intermediaries, whereas the accessibility can be inversely reflected by the ratio of collateral required for the potential loans. Moreover, government intervention in a firm is measured by the number of interaction days a general or deputy general manager needs to spend on communications with the government per month.
With these measures, we first confirm the causality between government intervention and firms' access to credit. The correlation between government intervention and access to credit may be affected by endogenous bias, but our results are still robust after using an exogenous instrumental variable to remove the potential endogeneity bias. In particular, our estimates confirm a causal promotional effect of government intervention on firms' access to credit.
Second, our IV estimates confirm that government intervention increases the probability that a firm will obtain high profit per capita relative to other firms in the same industry and city. Third, our estimates show that the variable of access to credit is still significant and that the government intervention variable coefficients change significantly. Put differently, government intervention has an indirect effect on firms' profit through their access to credit. In particular, all these results are also robust to different types of standard errors and different estimation methods.
The three procedures together investigate the mediator effect of a firm's access to credit on the relationship between government intervention and the firm's profit. With reference to z-tests, our evidence confirms that government intervention enables firms' access to credit and then further enables the firms to obtain large profits. Therefore, our work reveals the resource allocation function of government intervention involved in a firm's access to credit in the absence of effective legal institutions.     We capture access to creditwith financial loans or collateral ratio, which are the dependent variables. The former (Columns 1-2) is measured by whether the surveyed firm has loans from banks or other financial intermediaries. The latter (Columns 3-4) is measured by the ratio of the potential collateral relative to the loans obtained from financial intermediaries. The variable of interest is government intervention (in business affairs), reflected by days the general manager or the deputy general manager spends on government assignments and communications per month. Despite control variables, we also control for city and industry fixed effects. With each estimation method, we use robust standard errors or clustered standard errors. Standard errors of estimate are given in parentheses. *, ** and *** denote significance at the 10%, 5% and 1% level, respectively. # Considering that we have controlled for city fixed effects, we control for the cluster standard errors at the level of county. The instrumental variable used in this regression is public security intervention. It is measured by days the firm needs to spend on the interaction with the government on public security matters (e.g., public security intervention). Government intervention (in business affairs) reflected by days the general manager or deputy general manager spends on government assignments and communications per month. Loans are measured by whether the surveyed firm has loans from banks or other financial intermediaries. Collateral ratio is measured by the ratio of the potential collateral relative to the loans obtained from financial intermediaries. For each estimation method, we use robust standard errors or cluster standard errors. Standard errors of estimate are given in parentheses. *, **, *** denote the significance at the 10%, 5% and 1% level, respectively. # Considering that we have controlled for city fixed effects, we control the cluster standard errors at the level of county. The instrumental variable used in this regression is public security intervention. It is measured by days the firm needs to spend on the interaction with the government for the public security matters (e.g., public security intervention). Government intervention (in business affairs) reflected by the number of days the general manager or deputy general manager spends on government assignments and communications per month. Loans are measured by whether the surveyed firm has loans from banks or other financial intermediaries. Collateral ratio is measured by the ratio of the potential collateral relative to loans obtained from financial intermediaries. For each estimation method, we use robust standard errors or clustered standard errors. Standard errors of estimate are given in parentheses. *, **, *** denote the significance at the 10%, 5% and 1% level, respectively. # Considering that we have controlled for city fixed effects, we control the cluster standard errors at the level of county. The dependent variable is the probability that a firm obtains a large profit, = 1 . Put differently, the dependent variable reflects whether the firm's profit is larger (or not smaller) than the median one in the same district and city. The variable of interest is the variable of government intervention in general affairs. Considering that there is clear reverse causality between government intervention and firm profit, we adopt the IVProbit method. In particular, the instrumental variable is government intervention in public security matters (i.e., public security intervention). For these estimations, we use robust standard errors or clustered standard errors. Standard errors of estimate are given in parentheses. *, **, *** denote significance at the 10%, 5% and1% level, respectively. # Considering that we have controlled for city fixed effects, we control the cluster standard errors at the level of county. The dependent variable is the probability that the firm obtains a large profit, = 1 . Put differently, the dependent variable reflects whether the firm's profit is larger (or not smaller) than the median one in the same district and city. The variable of interest is government intervention. For these estimations, we use robust standard errors or clustered standard errors. Standard errors of estimate are given in parentheses. +, *, **, *** denote significance at the 15%, 10%, 5% and 1% level, respectively. # Considering that we have controlled for city fixed effects, we control for the cluster standard errors at the level of county.