De-politicization and Corporate Transformation: Evidence from China

Abstract

1. Introduction

Politically connected board members create value for their firms by lobbying for cheap finance and bailouts, lucrative government contracts, and favorable regulations (Roberts 1990; Fisman 2001; Faccio 2006). A growing body of evidence documents that politically connected firms tend to operate inefficiently, especially in emerging market economies¹ (Johnson and Mitton 2003; Khwaja and Mian 2005; Li et al. 2008). For example, partially privatized Chinese firms perform worse when their CEOs are politically connected (Fan et al. 2007). In China, board members gain substantial and illegal private benefits from their political connections, even though these boards are supposed to protect small shareholders from being expropriated by controlling shareholders and managers (Syverson 2011; Giannetti et al. 2015). In Indonesia, politically connected firms operate less transparently than firms that do not have such connections (Leuz and Oberholzer-Gee 2006). Kornai (1992) and Shleifer and Vishny (1994, 1997) argue that in emerging market economies, removing politicians from firms can break the paternalistic relationship between firms and the state, and thus enable firms to reallocate their resources away from rent-seeking activities into activities that are rewarded in competitive markets, such as improving operating efficiency and innovation.

Only a few event-based studies, however, have examined how the removal of politicians from a firm affects its value and performance.² Thus, this article exploits the large-scale, unanticipated, and nontransient removal of politically connected independent directors from corporate boards in China. This reform was first announced on October 30, 2013, in the File 18 document. The massive wave of resignations of independent directors following the enactment is, to the best of our knowledge and based on consultations with industry experts, unprecedented.³ By June 2014, “tens of thousands of government officials…relinquished their external posts” (from both listed and nonlisted companies), “including more than 200 provincial and ministerial officials” (Xinhua, June 10, 2014). Regarding listed firms, File 18 applied to independent directors who were currently working or had previously worked for the government. Our hand-coded sample indicates that by the end of 2019, 928 independent directors in our sample had resigned from their posts because of File 18.

We first ask whether the value of politically connected firms increased or decreased following the announcement that politically connected directors would need to leave their boards. Because political connections are pervasive and notoriously entrenched in state-owned enterprises (SOEs), our analysis focuses on private firms. In principle, firm value could decrease or increase through what we denote as the “cost” and “dynamic-adjustment” channels. According to the cost channel, because politically connected firms receive benefits from connections such as access to cheap finance and regulatory concessions, the disruption of these connections would immediately drive down the value of connected firms. However, according to the dynamic adjustment channel, when the removal of political connections weakens inefficient political interference, firms can reallocate their resources to value-enhancing activities such as innovation and thus increase their value.

We estimate the value effect of the removal of politicians from boards in the short and long run, respectively. We first conduct an event study of the stock market response around the time of the announcement. Using a sample of more than 1100 listed private firms and a hand-coded dataset of corporate directors, we find that that for the five-day event window around File 18, the cumulative abnormal returns (CAR[−2,+2]) for connected firms fell by almost one percent relative to those of unconnected firms. The results remain similar when we focus on firms with and without director resignations ex post. Following Malmendier et al. (2018), we construct buy-and-hold returns (BHAR[−1,+3yr]) to capture the longer-term effect on firm value. However, the effect is not significant, which suggests that other factors offset the negative market reactions.

We then try to understand why connected firms initially lost value and then bounced back. We construct a 13-year panel dataset spanning the 2007–19 period to estimate the real effects of File 18. The identification is based on the director resignation events among firms. Using a difference-in-differences strategy based on a matched sample, we find evidence consistent with a “cost” channel, whereby connected firms lost the benefits of their connections and their value initially declined. Specifically, we find that firms that lost political directors experienced a temporary reduction in their access to cheap loans, as measured by an increase in the cost of debt.

We also find evidence of a “dynamic adjustment” channel, whereby firms successfully adjusted to a reduction in their political connections in ways that, over time, increased their value. Firms that had politically connected directors who resigned from their positions following File 18 increased their investment in R&D, suggesting that they engaged in more innovative activities. They imported more machinery from abroad, suggesting that they were upgrading their technologies; they also reduced their related party transactions (RPT), indicating that they became more transparent and reduced their tunneling activities. Finally, their total factor productivity (TFP) increased, suggesting that they operated more efficiently.

Our study is related to several quasi-experimental studies of corporate boards. Ahern and Dittmar (2012) analyze a law requiring that at least 40% of listed firms’ directors in Norway must be women. They find that the law caused firms to lose value in part because the women who came on board were less experienced and younger, and as a result, firms’ leverage and acquisitions increased while their performance deteriorated. Giannetti et al. (2015) find that when individuals with foreign experience join a firm’s board, the firm’s value and TFP increase, and, in time, its profitability. They find evidence suggesting that firms do better because their boards pick better policies, including less management of earnings (suggesting better corporate governance) and a more international focus, such as more exporting, a larger share of mergers and acquisitions with foreign partners, and more engagement with foreign investors. We document that firms’ profitability eventually increases and argue that this reflects the board’s adoption of a series of good policies.

Our paper complements those by Tang et al. (2016) and Xu (2017), who study the short-term impact of File 18 on firm value. Tang et al. find that ownership, regulation, and the importance of the departing director drive the decline in instantaneous value; Xu finds that firm values decline during the year of the enactment (2013) and one year after (2014) because firms operate passively to avoid accusations of corruption. Our results on market reactions are generally consistent with the findings of those studies. In contrast, we focus on director resignations to examine the longer-term realized impact of File 18.

Our study builds on Siegel’s (2007) study of how political connections affect the operating strategies of firms in South Korea. Siegel uses unanticipated and nontransient national-level political changes that drastically weaken or strengthen firms’ political connections. Whereas Siegel analyzes how firms’ cross-border activities are affected when political assets become political liabilities (and vice versa), our goal is to understand whether weakening political connections can cause connected firms to change their operating strategies and engage in activities that are rewarded in competitive market economies. Kim (2015) analyzes how firms in a developed economy, the United States, respond when one of its lobbyists plausibly has weaker connections because the US Congressional official for whom the lobbyist worked unexpectedly dies, resigns, or loses a close election.⁴

This study contributes to the literature on corruption. Fisman and Svensson (2007) show that firms that use bribes for their daily operations grow more slowly. Consistent with this finding, we demonstrate that the enactment of File 18 tends to encourage strategies that boost firms’ growth, including innovation, investment efficiency, and transparency, as opposed to rent-seeking activities. Goldman et al. (2013) document that politically connected board members pick procurement contracts from which they may enrich themselves but that leads their companies to underperform.

The next section describes File 18; Section 3 describes the sample construction; Sections 4 and 5 study the impact of File 18 on the value and activities of firms; and Section 6 concludes.

2. File 18

The Chinese government under President Xi Jinping has focused on the perceived abuse of political connections. A well-known example is how the government blamed the misuse of political connections for the catastrophic explosion of a warehouse in the northern port city of Tianjin in August 2015. More than 100 people were killed, roughly 700 were injured, and more than 17,000 homes were destroyed in the explosion. The warehouse held roughly 700 tons of poisonous sodium cyanide.⁵ Safety regulations require such sites to be at least a kilometer from residential areas, yet in Tianjin, some apartment buildings “were only 700 meters away” from the exploding warehouse (Bloomberg-Business, August 18, 2015). Xinhua, the state-run news agency, reported that two controlling shareholders of the warehouse had strong political connections, which allowed them to ignore government regulations (World-Post, August 16, 2015). Immediately after the explosion, top leaders vowed to hold a thorough investigation, and at the end of August, 12 people were arrested and 11 officials and port executives were accused of dereliction of duty or abuse of power (BBC, August 17, 2015; Reuters, August 27, 2015).

In addition to the tragedy in Tianjin, it has been documented that political connections play an important role in other sectors of China, such as mining industries (Fisman and Wang 2015), the real estate industry (Fang et al. 2014; Gao et al. 2016; Chen and Kung 2018), and academia (Fisman et al. 2018).⁶ There is also a strong perception that the misuse of political connections on company boards is prevalent (Xinhua, June 10, 2014). Since 2002, China’s security regulatory commission has required one-third of the boards of directors of publicly listed companies to be independent. This means that an independent director is not allowed to work for the company that he or she oversees and is not allowed to hold more than 1% of the company’s shares or be one of its top-10 largest shareholders. Because controlling shareholders in China have highly concentrated stakes, increasing the number of independent directors was expected to help protect minority shareholders (see Clarke 2006). According to Xinhua, however, “instead of inviting qualified and competent professionals to serve on their board, public firms have been inclined to reserve such seats for prominent government officials and scholars in exchange for their political and social influence.”

In October 2013, the Central Organization Department of China issued a directive, referred to as File 18, that banned government officials from taking external positions to control the misuse of political connections in the economy. Specifically, File 18 required independent directors who are current politicians or who retired from government official posts within the previous three years to resign from their positions. Former or current officials who wish to stay in their positions must seek special approval and receive no salary. Appendix A shows the full content of File 18. According to government estimates, by June 2014, 40,700 government officials had given up their external posts, suggesting that File 18 was effectively enforced.⁷ According to Xinhua, 173 provincial or higher level officials had left their jobs in the company, and 56 were still in the process of departure by June 2014 (Xinhua, June 22, 2014).⁸

We base our identification on the resignations of politically connected directors of publicly traded firms following File 18. Consistent with the intent of File 18 and the literature (e.g., Fan et al. 2007), an independent director is classified as politically connected if he or she held a position as a government official in the three years preceding the enactment of File 18. We consider a broad classification of government officials including mayors, secretaries of municipal party committees, directors of city-level bureaus, current members of the National Peoples’ Congress or Chinese Peoples’ Political Consultative Conference, members of the Peoples’ Procuratorate, and members of the Peoples’ Court. We also include independent directors who held positions with political rank and political power, such as managers of SOEs, presidents, and deans of public universities, and leaders of national industry associations.⁹ We collected the biographies of independent directors from the China Stock Market and Accounting Research (CSMAR) and WIND databases and cross-checked them with company disclosures and online sources.

Apart from the resignation cases triggered by File 18, independent directors may resign for administrative reasons, such as when a director reaches the end of a six-year term or the mandatory retirement age (70 years), a director becomes a full-time employee of the firm, or the director dies or is under investigation for a crime. We count all resignations of independent directors following the enactment of File 18 (October 30, 2013) using December 31, 2019, as the end of our sample. The number of resignations due to File 18 sharply dropped after 2016.¹⁰ Between October 30, 2013, and December 31, 2019, 437 politically connected directors in private firms and 491 in SOEs resigned because of File 18, while 32 politically connected independent directors resigned for regulatory or administrative reasons.¹¹

When File 18 was announced, investors could easily find out from publicly available reports which listed firms employed politically connected independent directors. Thus, we expect that the news that politically connected independent directors were required to resign led to immediate changes in these firms’ stock prices around the announcement of File 18. However, politically connected directors who retired from their political positions before the enactment of File 18 could delay their resignation pending a review of whether their connections were no longer an issue. As a result, when a politically connected independent director actually resigned from a firm should also matter for its value.

3. Sample Construction

The sample comprises data from several sources. We obtain firm-level financial variables and stock prices from the CSMAR and WIND databases. Our baseline sample of firms that were in operation when File 18 was announced consists of A-share stocks listed on the Shanghai and Shenzhen exchanges. We exclude firms labeled as special treatment (ST) before the announcement of the regulation. We confine our baseline sample to private firms and drop firms that are ultimately controlled by the state government through its agencies or entities.¹² Appendix C provides a detailed summary of the sample construction. The panel sample comprises 12,944 firm–year observations from the 2007 to 2019 period and covers 1187 distinct private firms. Of these, 345 are “treated” because their boards had at least one politically connected independent director when File 18 was announced and at least one of these politically connected independent directors had resigned by the end of 2019. The control group comprises 842 firms; of these, 736 did not employ a politically connected independent director when File 18 was announced, while 106 employed directors who were perceived to be politically connected and still board members as of December 31, 2019.

The details of the politically connected directors in private firms who resigned because of File 18 between October 30, 2013 and December 31, 2019 are presented in Figure 1. The figure shows a sharp uptick in resignations roughly three months after the announcement, with spikes at the end of 2014 and 2015; there is then a sharp drop, with a negligible number of resignations in 2016 and almost none in 2017–19.

Figure 1

Resignations because of File 18.

Notes: This figure shows the number of resignations because of File 18 of politically connected independent directors who served on corporate boards of private firms as of October 30, 2013 when File 18 was issued. Whether or not a director is politically connected is coded using firm’s annual reports and online biographies. We drop all resignations that are not related to File 18, for example, because a director died, was under criminal investigation, reached the mandatory retirement age, etc. Details about the classification of reasons for a retirement can be found in Appendix B.

Open in new tabDownload slide

Table 1 describes the construction of the study variables. Table 2 describes the 1187 private firms used in the cross-sectional analysis of the impact of File 18 on firm value. Consistent with the coding in Section 2, PC is a dummy variable that indicates whether a firm has at least one independent director who is perceived to be politically connected and to be affected by File 18. Treat is set to one for firms that had at least one politically connected independent director when File 18 was announced who resigned because of File 18 as of December 31, 2019, and zero otherwise. Two variables, CAR[−2,+2] and BHAR[−1,+3yr], are used to capture the value effect of File 18; they are defined when they are used in the following sections. We also consider a set of standard control variables including firm size (Size), total liability scaled by assets (Leverage), net operating income scaled by assets (ROA), and ownership structure (Wedge). We include two further variables measuring board characteristics: the percentage of independent directors (Ind. Dir. Per.), and whether the CEO and board chairman positions are held by the same individual (Duality).

Table 3 Panel A compares the control variables in 2012 (the year before the File 18 announcement year) for treated versus control firms in the full sample of private firms (N = 1,187). The treated and control firms are balanced according to their leverage, net operating income, percentage of independent directors, duality, and ownership structure. However, the treated firms are significantly larger than the control firms. Thus, to mitigate potential concerns about selection, we construct a matched sample based on a propensity matching approach. We first run a probit model with all of the control variables and the industry and province dummies and obtain the propensity score for each firm. Next, each treated firm is matched without replacement with a control firm that has the closest propensity score. Panel B reports the balance tests for the matched sample, which comprises 345 treated and 345 control firms. Panel B shows that the treated and control firms have similar summary statistics across all covariates considered in the matched sample. Panel C presents the summary statistics of the matched sample for estimating the impact of File 18 on corporate restructuring in treated versus control firms post-treatment. The firm-level outcome variables include cost of debt, R&D expenditure (R&D Ex.), machine imports, RPT, and TFP. These outcomes are defined when they are used.

4. Value Effects of File 18

We first estimate the impact of File 18 on firm value by examining the changes in firms’ stock prices around the announcement of File 18. Although Xi Jinping initiated the anticorruption campaign soon after he became President, the Central Organization Department, the top-level authority in charge of Communist Party personnel arrangements, drafted File 18 in secret and made no mention of it until its announcement on October 30, 2013. Previous studies (e.g., Fisman 2001) document that political connections can create value for firms. We thus expect that the news that politically connected directors would have to retire in the near future triggered short-term drops in stock returns for treated firms. However, if firms used the removal of their politically connected directors as an opportunity to develop a more competitive and market-based corporate strategy, they might have rebounded from short-term value losses over the longer term.

We construct CAR around the announcement date of File 18 (October 30, 2013) based on the market model with value-weighted stock returns as the benchmark to measure the short-term market reactions. Following the standard practice in the literature (e.g., MacKinlay 1997), we estimate the firm-specific market model parameters from day −210 to day −11 relative to the event day (zero) and calculate the abnormal returns for each firm in the five-day window. CAR(−2,+2) is defined as the sum of the abnormal returns in the five-day [−2,+2] window relative to the event day.

In the BHAR formula, s indexes the date of the return relative to the event date; $ris$ is firm i’s return on day s relative to the event date; $mrs$ denotes the benchmark return, which is the value-weighted stock returns; and T marks the window’s length. In our analysis, we consider three-year buy-and-hold returns.

Table 4 contains estimates of the impact of File 18 in the short and longer run. As shown in columns 1 and 2, consistent with our expectations, firms that were most exposed to File 18 either because they employed politically connected independent directors when File 18 was enacted (PC_i = 1), or because their politically directors retired from their positions by the end of 2019, lost stock market value in the short term. This finding is consistent with other studies that document the short-term impact of File 18 (Tang et al. 2016; Xu 2017). However, columns 3 and 4 do not show a significant difference between the buy-and-hold returns of treated and control firms. The difference between short-term and long-term reactions suggests that firms may have implemented policies that were not fully anticipated by the investors when the regulation was announced. We posit that firms adapted to the new business environment by adjusting their business strategies. The next section provides evidence consistent with this conjecture.

5. The Real Effects of File 18

To understand why politically connected firms initially lost value but then recovered, this section documents that two countervailing forces were at work. Firm value would be reduced via the cost channel if politician–directors could no longer deliver benefits such as cheap finance and bailouts. However, the dynamic adjustment channel could be a force for raising value if politician–directors, who have noneconomic objectives, had less power to interfere in firms. Firms could then adjust, for example, by investing in innovation and operating more transparently. We speculate that the cost channel was more prominent following the announcement of the law; although the dynamic adjustment channel was initially weaker as firms needed time to adjust to doing business with fewer political connections, it eventually became stronger than the cost channel. Thus, while connected firms had a lower value early on, they eventually bounced back. In the next two sections, we provide evidence that both channels were at work.

5.1 The Cost Channel

The model in equation (3) is estimated using the matched sample illustrated in Panel C of Table 3. The control variables $Xi,t$ are the time-varying versions of the one-period control variables used in the cross-sectional event study. The standard errors are estimated using two-way clustering at the firm and year levels to adjust for the cross-sectional and time-series dependence of residuals in the financial data so that reliable inferences can be made (Petersen 2009).

Because bank loans are an essential source of external finance for firms, and politically connected firms around the world enjoy lower borrowing costs from banks (e.g., Claessens et al. 2008; Firth et al. 2009), Table 5 uses a firm’s cost of debt as a measure of its preferential access to finance. We measure the cost of debt as interest expenses divided by average total loans, including short- and long-term loans. Table 5 reports the estimates of four variants of the model in equation (3): column 1 includes only firm and year fixed effects, column 2 adds the control variables, column 3 also includes industry-by-year fixed effects, and column 4 includes province–year fixed effects. In all four models, the estimated coefficient for the impact of treatment in the first treatment year, Treat_Year(1), is statistically significant and positive, indicating that File 18 imposes temporary costs on the firms’ preferential access to finance. Treated firms suffer from an increase of approximately 1.1 percentage points in the cost of debt, accounting for 9.5% of the sample standard deviation (SD). However, the effect diminishes and becomes insignificant, starting in the second year of treatment.

Figure 2 Panel A illustrates the estimated coefficients and 95% confidence intervals for pretrends and post-treatment effects. The graph shows no evidence of a pretrend. It also shows that treated firms, which may have relied on their politically connected independent directors for access to bank finance, had a higher cost of debt than control firms the year after their politically connected directors left. The graph also illustrates that, starting in the second year of treatment, the effect of treatment becomes insignificant, which is consistent with the view that treated firms made adjustments over time that lowered the cost of capital and increased their ability to compete.

Figure 2

Dynamic effects.

Notes: These figures show the dynamic effects of File 18. The last year when a political connected director worked, t = 0, is set to zero and omitted. In all figures except panel C (machine imports), we consider a nine-year window spanning four years before and after period t = 0 the last year that a politically connected independent director was employed. Because machine import data are only available through 2016, panel C has an eight-year time window containing four years before and three years after period t = 0.

Open in new tabDownload slide

5.2 The Dynamic Adjustment Channel

Political connections provide firms with access to finance and other benefits. However, they also distort firms’ incentives in ways that can hinder efficiency because connected firms do not necessarily need to improve their efficiency to compete for resources and sustain their business (Johnson and Mitton 2003). Thus, when politician–directors are removed, we would expect firms to respond to the pain of intensified financial constraints and loss of government concessions by making adjustments to operate more competitively.

An obvious adjustment is that firms invest in innovation. It is well understood that innovation is a major driving force for economic growth (Solow 1957), yet it is also well known that motivating innovation activity can be difficult (Manso 2011; Tian and Wang 2011).¹⁴ Previous studies document the importance of financial and political factors in fostering innovation (Hsu et al. 2014; Moshirian et al. 2019).¹⁵ In a connection-based economy where rent-seeking is rampant, investing in innovation is even less rewarding (Murphy et al. 1993).

Removing politicians from boards could stimulate innovation through several channels. First, politicians are usually under pressure to get promoted or even re-elected (Nordhaus 1975), and many good innovative projects do not promote politically connected directors’ agendas. In the context of China, Li and Zhou (2005) document that good economic performance, measured by short-term indicators such as GDP output and social benefits, increases the chances that a local leader will be promoted. Therefore, politicians tend to push the firms that they can control to invest in their pet projects or projects that will generate short-term economic growth or short-term social benefits. Thus, removing politician–directors should reduce a firm’s dependence on the government and reallocate resources back to innovation-related investment. Second, in an economy where competitiveness is more important than political connections, improving innovation is critical because it establishes competitive advantages (Porter 1991). A functional capital market also stimulates firms to innovate and rewards innovative firms in the long run (Schumpeter 1911). Thus, we expect that connected firms became more innovative than unconnected firms following the enactment of File 18.

We consider a firm’s in-house innovation investment and its efforts to acquire innovative technologies. First, we use research and development (R&D) to proxy for innovation input. Following Png (2017), we define R&D Expenditures as the log of 1 plus total R&D expenditure in a fiscal year. As shown in Table 6, the increase in R&D Exp. becomes statistically significant starting from the second treatment year. The baseline estimation in column 4 indicates that File 18 leads to more R&D investment and the magnitude increases over the years, accounting for 3.7–9.2% relative to the sample SD. Figure 2 Panel B shows no pretrends in the baseline estimate in column 4.

Second, we measure firms’ activities in acquiring innovative technologies using detailed information on international trade. Firms can upgrade their product lines and improve their technological proficiency by importing advanced equipment (Eaton and Kortum 2001; Caselli and Wilson 2004). To obtain these data, we merge listed firms and their subsidiaries in the CSMAR database with the China Customs Database, which contains an annual summary of a firm’s import and export transactions at the product and counterpart-country level. Specifically, we manually standardize the firm names in the two sources and create a linking table. We first identify machinery imports using the Harmonized System two-digit product codes 84 and 85. We construct the variable “Machine Imports” as the log value of 1 plus the total value of imported machines from abroad. As shown in columns 1 and 2 of Table 7, File 18 leads to significant increases in machinery imports, starting from the first treatment year. The baseline estimation in column 2 indicates that the magnitudes account for 8.8–11.5% of the sample SD. Panel C of Figure 2 suggests that there are no pretrends.

Next, we explore whether these machinery imports are from countries with advanced technology. The variable Machine Imports (Advanced Countries) is the logarithm of 1 plus the value of machine imports from countries ranked above China in innovation level according to “The Global Innovation Index 2013” issued by the World Intellectual Property Organization. In our sample, 34 countries are ranked above China in terms of innovation level. Column 4 of Table 7 indicates that treated firms imported more machinery equipment from countries with more advanced technology.

Improvement in operational transparency is another indicator of treated firms’ adaption to the loss of politician–directors. Corporate transparency enables firms to enjoy lower transaction costs and greater liquidity, and thus more access to external finance (Lang et al. 2012). However, political connections enable firms that do not operate transparently to gain access to finance (Houston et al. 2014). In fact, connected firms may prefer to operate opaquely because the scrutiny that comes with greater transparency makes it more difficult to engage in rent-seeking activities (Leuz and Oberholzer-Gee 2006). As most firms need access to external finance, we expect that File 18 encouraged treated versus control firms to operate more transparently.¹⁶ To test this expectation, we use RPTs as an activity-based measure of corporate transparency.¹⁷

In China, the definition of a related party is quite broad and can include controlling shareholders, parent companies, subsidiaries, shareholders holding more than 5% of a firm’s shares, insiders (directors and managers), and relatives of insiders. RPTs include transactions such as providing collateral or guarantees, buying and selling, lending, investing, and leasing. RPTs are regulated; for example, transactions worth more than 300 thousand RMB between related parties must be disclosed, and transactions exceeding 30 million RMB must be discussed in a shareholder meeting. The pricing of RPTs is also regulated.

Nevertheless, there is substantial evidence that large shareholders in China frequently use RPTs as a scheme for extracting rents (Cheung et al. 2006; Jia et al. 2013; Fisman and Wang 2014; Liao et al. 2014). For instance, because the pricing of most RPTs is not, in reality, market-based, managers can sell their firms’ assets to controlling shareholders at a discount, or purchase products from related parties at a large premium. Such self-serving behavior intensifies corporate opacity and leads to higher borrowing costs (Lin et al. 2011). In competitive markets, firms tend to use RPTs parsimoniously because they may raise concerns about potential conflicts of interest (see Leuz and Oberholzer-Gee 2006). Thus, effective enactment of File 18 would mean that firms that previously relied on political ties would have to compete with other firms for capital based on their economic merit. Thus, we would expect depoliticized private firms to reduce their engagement in RPTs compared to the control firms to improve their corporate transparency and access to finance.

Consistent with the literature (e.g., Liao et al. 2014), we focus on major types of RPTs, including guaranteeing loans for related parties, asset sales, and goods and service transactions with related parties. RPT is the total value of RPTs scaled by assets. As shown in Table 8, after the enactment of File 18, there was a significant decrease in RPTs in the range of 1.4–3.3 percentage points in the years following the enactment, accounting for approximately 6.7–15.7% of the sample SD. Figure 2 Panel D shows that there are no pretrends for the baseline model, and the pretrend coefficients in columns (1)–(3) indicate that there are no pretrends in the other models.

Finally, we expect that removing politicians from boards gave firms incentives to improve their operational efficiency for two reasons. First, in a connections-based system, a firm may invest for political rather than economic reasons. Second, and more importantly, market competition in a merit-based system has a disciplining effect on managerial effects (Hart 1983), which is considered a highly effective mechanism for bolstering firm efficiency (Shleifer and Vishny 1997). Thus, following the enactment of the law, it is likely that managers in connected firms improved the operational efficiency of their firms.

A potential concern with our findings is that the resignation of politically connected independent directors after the enactment of File 18, which is the main exposure in our study, may be endogenous. To address this concern, in our Online Appendix, we use “PC,” which measures the political connectedness of the board generally perceived by the investors when File 18 was announced, as an instrument.¹⁸ Specifically, in the first stage of the instrumental variable regression, we retrieve the predicted exposure, “Treat”; then, in the second stage, we perform a “generalized difference-in-differences” estimation of the key outcomes, including R&D expenditures, machine imports, RPTs, and TFP. The results reported in Online Appendix I are consistent with our main findings.

An additional concern is the File 18 reform was effective primarily in regions where the anticorruption campaign was strictly enforced, suggesting that the File 18 reform and the broad anticorruption campaign are confounded. To address this issue, in Online Appendix, we use the number of anticorruption cases in a province in a given year as a measure of the intensity of the anticorruption campaign. We then estimate a set of models that interact the treatment variable for File 18 with the number of corruption cases in a province in a given year.¹⁹ The findings contained in Online Appendix II indicate that the realized anticorruption results do not affect most of the outcome variables considered in this study.

5.3 State-Owned Enterprises

The anticorruption reforms in China have primarily targeted government officials and managers of SOEs carrying significant political rank.²⁰ File 18 was unusual because it was perceived to be the anticorruption reform that primarily targeted private firms. While de jure File 18 also applied to SOEs, SOEs have so many key employees with political connections that we would not expect the removal of politician–directors to significantly influence their access to external finance and government concessions. Moreover, File 18 was just one of a large and extensive set of anticorruption reforms that SOEs were required to implement. The mixed set of antireform policies and projected weak impact of File 18 make it very difficult to identify the impact of File 18 on SOEs. In Appendix D, we tabulate the results based on a sample of SOEs matched using the same matching approach as in the baseline sample construction. The findings in Appendix D are unstable and difficult to interpret.

6. Conclusion

It is well documented that political connections have value for firms because they provide government concessions such as favorable regulations, government contracts, and cheap finance and bailouts. However, whether the weakening and removal of these connections encourage and enables firms to adapt and operate more competitively has not been carefully studied. In this article, we use China’s File 18 as a source of exogenous variation in the removal of government-related personnel from the boards of private companies. Consistent with the view that political connections have value, firms with politician–directors lost value around the time that File 18 was announced, and their cost of debt increased a year after its enactment.

However, we find evidence that File 18 incentivized connected firms to transform their corporate strategies and operate more competitively. One year after losing their politician–directors, these firms had boosted their productivity, imported more machinery, and operated more transparently: within two years, their cost of debt had recovered; and within three years, they were spending more on R&D. Consistent with the view that these firms were learning to operate more competitively, after three years, their market valuation had bounced back from losses around the time that File 18 was enacted.

Stretching somewhat further, our study sheds light on the relationship between anticorruption reforms and economic growth. Clearly, fighting the misuse of political connections is a central component of anticorruption reforms. Our findings suggest that when a reform is part of a broad and credible government commitment to reduce corruption, it is possible to eliminate political connections within private firms and thus stimulate them to rapidly adjust in ways that promote economic growth.

This is a substantially overhauled version of our working paper entitled “De-Politicization and Corporate Transformation: Evidence from China’s Anti-Corruption Reforms.” We are grateful to comments from the editor (Rocco Macchiavello) and three anonymous referees. We thank Raymond Fisman, Xavier Giroud, Daniel Jones, Ross Levine, Maria Petrova, James Robinson, Yogita Shamdasani, Andrei Shleifer, Zheng Michael Song, Michael Weisbach, and participants of the Harvard economics department China seminar, the development retreat at the Harvard Business School, the PRC New Economic Normal conference, and the University of Pittsburgh labor-development brownbag for their comments. Berkowitz conducted some of this research while he was a visiting scholar at the NBER. C.L. would like to thank the financial support from the Research Grants Council of Hong Kong (Project No. 17504117). S.L. would like to thank the financial support from the Research Grants Council of Hong Kong (Project No. 23601319).

Footnotes

Appendix A

File 18

“The Directive on further Disciplining Party and Government Officials Holding Full-time (and Part-time) Position in Corporations” (Central Organization Department [2013] File No.18)

(Translated from Chinese)

Issuance Date: October 19, 2013.

To strictly monitor Party and government officials, to build a sound cadre team, and to implement anticorruption campaign, basing on “Civil Servant Law of the People's Republic of China,” “Guidelines for Leading Cadres and Members of the Communist Party of China,” and other relevant documents, this directive is issued regarding further disciplining party and government officials holding full-time (and part-time) position in corporations.

1. Current Party and government officials shall not hold position in corporations, including those who are not in the office but do not officially retire.

2. Former Party and government officials who officially retired shall be strictly examined for the reasons and necessity of holding positions in corporations according to the relevant regulations. Within three years after the retirement, former officials shall not hold positions in corporations or engage in any for-profit activities in their former administrative areas or related business. With three years after the retirement, former officials who wish to hold positions in corporations or engage in for-profit activities outside their former administrative areas or related business shall apply to the Party committee of their direct government/government entity. The related corporation shall provide documents illustrating the reasons. The Party committee will review the application according to the provisions and guidance and forward it to the higher authority for approval. Former officials shall not take the position until being approved. After three years following the retirement, former officials who wish to hold positions in corporations shall apply to the Party committee of their direct government/government entity. The related corporation shall provide documents illustrating the reasons. The direct party committee will review the application and decide whether it is approved according to the related provisions and guidance and forward it to higher authority for the record.

3. Former officials who are approved to hold part-time positions in the corporation shall not receive salaries, bonuses, allowances, and other rewards, including equity and other extra benefits. Former officials are not allowed to take more than one part-time position. After the expiration of the first term of the position in the corporation, former officials shall apply to the party committee again for the approval and the record. And former officials shall not be in the position for more than two terms. Former officials shall not take part-time position after 70 years old.

4. Former officials who are approved to hold full-time positions in the corporation shall give up any of their administrative duties, only receive the salary from the corporation, no longer keep their civil servant status, and no longer retain various treatment from the Party and governments.

5. Former officials who are approved to hold part-time (or full-time) positions in the corporation shall strictly discipline themselves and be forbidden to use the political power to influence the enterprises or individuals for seeking improper interests. Former officials shall report their salary and all related reimburse in written form to the direct Party committee by the end of each year.

6. Regional governments and authorities shall investigate the existing violations according to this directive and to make correction within a specified time period. Officials who are not in conformity with this directive shall resign or be forced to leave the position in the corporation within three months following the issuance of this directive. Officials who wish to remain in the position of corporations shall be reviewed. Issues regarding the compensation received in the given position shall also be investigated.

7. After the “cleaning-up” investigations, if any violation including holding the position in the corporation or hiding the fact of receiving compensation be found and confirmed, the related parties shall be seriously punished according to the relevant regulations. Any violating behaviors in reviewing and approving the applications for holding positions in corporations shall also be investigated.

8. Party and government officials holding positions (part-time or full-time) in other types of for-profit organizations, “people’s organizations” and institutions shall also follow this directive.

9. Authorities in all levels of government shall formulate their specific rules in the spirit of this directive to discipline the behavior of Party and government officials in holding full-time (or part-time) position in corporations.

10. This directive is enforced on the date of issuance. This directive shall be followed in cases of any inconsistency between prior issued regulations and this one.

Appendix B

Reasons for Independent Director Resignations

Notes: This figure shows the number of resignations over time reason for the resignation which fits into one of the following three categories:

1. Due to File 18: the resignation announcement explicitly mentioned File 18.

2. Regulatory and administrative reasons unrelated to File 18

   • Term limit (six-year)

   • Age limit (70 years) and retirement

   • The director is hired by the firm or the director’s employer becomes the owner or subsidiaries of the listed company

   • Passed away

   • Reasons regarding the corporate operation

   • Crime related investigations

   • Specific personal issues (e.g., emigration)

3. Due to File 18: The resignation announcement referred to personal reasons or work-related reasons, or in the scenario, no reason is provided. Because there are no specific regulatory or administrative reasons and, these resignations occur after the enactment of File 18, it is highly likely that they are because of File 18.

Appendix C

Sample Construction

1. All A-share listed companies in Shanghai exchange and Shenzhen exchange;

2. Exclude firms labeled as “special treatment (ST)” before the announcement of File 18;

3. Exclude firms that have missing data for the main variables, including firm size, leverage, ROA, percentage of independent directors, and board duality indicator, and ownership structure measure (wedge). We restrict the sample to firms with nonmissing financials in 2012, the last fiscal year when File 18 was announced. To facilitate panel analysis using observations before and after the regulation, we exclude firms listed in 2012 and after 2012.

4. Include only firms whose ultimate controller is not the government or a government agency as defined by in the CSMAR database.

5. The baseline panel consists of 12,944 firm–year observations with 1187 private firms between 2007 and 2019, spanning six years before and after 2013, the year that File 18 was enacted. The treatment group contains 345 firms that have politically connected independent directors who served on a firm’s board when File 18 was announced on October 30, 2013 and who resigned because of File 18 by the end of 2019. The control group consists of 842 firms that either did not hire a politically connected director when File 18 was announced on October 30, 2013 or, who employed a politically connected independent director when File 18 was announced who did not resign by the end of December 2019. The matched panel contains 7564 observations with 345 treated firms and 345 control firms.

Appendix D

State-Owned Enterprises