Connections as Liabilities: The Cost of the Politics–Business Revolving Door in China

Abstract The established literature has recognized revolving-door hiring as a means for firms to obtain protection against political risks and advance their business interests. This article theorizes about the cost of the revolving door between politics and business, which transmits uncertainty during turbulent times from the political system to firms. I empirically estimate the cost of revolving-door recruitment for over 3,000 publicly listed firms in the early years of a major corruption crackdown in China. I show that firm-level returns on revolving-door recruitment became negative during this period. In contrast to conventional corporate governance explanations, the mechanism proposed in this article emphasizes the external perceptions and financing of firms. Text analyses of over 1 million equity reports and bank-loan record data show that political connections act as a negative signal to external financiers, thereby discouraging external financing.

Revolving doors between politics and business, a phenomenon whereby firms recruit former government officials, are prevalent around the world.Through revolving-door hiring, firms establish connections with political elites and, in return, obtain enormous preferential benefits (see, for example, Blanes i Vidal, Draca, and Fons-Rosen 2012;Draca 2014).However, building political connections through revolving-door channels can be a double-edged sword.A burgeoning body of literature shows that dalliances with powerful political leaders can introduce tremendous problems for firms, such as rent-seeking, corruption, and poor management, which eventually harm firm performance and the health of the financial market (see, for example, Bertrand et al. 2018;Fan, Wong, and Zhang 2007;Fisman and Wang 2015;Shleifer and Vishny 1993).
While most scholars have focused on how political connections weaken firm performance by jeopardizing firms' operational management, this study focuses on an alternative cost triggered by revolving-door connections.I theorize that the politics-business revolving door transmits enormous uncertainty from the political system to the financial market.When political leaders experience unexpected dismissals, political connections secured through revolving-door channels are no longer a signal of powerful political capital, but a liability that can drag firms into scandals.This political risk theory has been empirically examined by various studies that have focused on the stock market reaction to scandals implicating revolving-door firms. 1 This article challenges the conventional view that political risk affects only those specific firms that have ties with political leaders involved in scandals.Instead, I theorize that there is a spillover effect from political connections that results in collateral damage to all politically connected firms when the external political risks are extremely high.I argue that all firms that engage in revolving-door hiring, regardless of whether they have direct involvement in a political scandal, experience losses because of increased political risk.This negative effect of political connections does not just emerge in the stock market, as shown by the short-term decline in stock prices.Rather, the cost of political connections also appears on balance sheets as direct damage to firm profits through an external perception channel.Financiers who face rising risks are less willing to provide credit or investment to implicated, politically connected firms.A lack of external finance then creates a liquidity crisis that eventually harms firm profitability.
I empirically test this political risk theory by analyzing firm performance during a massive anticorruption campaign in China, in which approximately 1.3 million officials at various levels of government were demoted, investigated, and even dismissed from public office (Gan and Choi 2018).I characterize the campaign as an environment that generated considerable political risk for both public officials and firms.I document the negative effects of political connections for all firms listed on the Chinese stock market in the early years of the campaign, from 2012 through 2016. 2 In this article, I emphasize the employment connections between firms and political leaders, as public and formal relationships expose firms to enormous political risks during anticorruption efforts.I define a firm as politically connected if it hires at least one former government or Communist Party of China (CPC, or "the Party") official as a board director or senior executive. 3I construct this measure of political connections by building an original database of all revolving-door officials (or "revolvers") who held positions in publicly traded Chinese firms.I match the firm's revolving-door recruitment information with other financial indicators and stock return data to conduct the empirical analysis.
The main results show sizable losses triggered by the firms' recruitment of revolving-door officials.The average effect on firms that hire revolving-door officials is estimated to be approximately 0.6 per cent to 1.9 per cent on various profit indicators, including return on assets (ROA), return on equity, profit margins, and earnings per share.A back-of-the-envelope calculation shows that the estimated cost of revolving-door recruitment is equivalent to approximately RMB14 million (USD2.2 million) per year for a firm with average assets.The findings are robust to various tests, including a parallel trend analysis, alternative treatment measures, a placebo treatment, subsample analysis, and different model selections.
Beyond documenting the cost of revolving-door recruitment, I empirically examine the mechanism by which political risk reduces the performance of politically connected firms.In support of this external financing mechanism, I show that firms that hire revolvers suffer from decaying external investor perceptions in the financial market.A sentiment analysis of over 1 million equity research reports documents that after the campaign's onset, financiers evaluated connected firms, that is, those that hire revolvers, more negatively than unconnected firms.I also show evidence on how political risk further affects external financing.The borrowing records indicate that banks provided 24.5 per cent fewer loans to firms hiring revolving-door officials than to unconnected firms.In addition to providing evidence on the external perception mechanism, I rule out several alternative explanations for the cost of revolving-door recruitment, including the decline in preferential benefits, such as government subsidies, favorable taxation, and land purchases.The evidence suggests that none of these explanations serve as the main channel by which political risk increases the cost of revolving-door recruitment.
This study sheds further light on political risk theory by considering two extensions.The first extension focuses on the heterogeneous effect.As revolving-door recruitment connects politicians and firms, local political and market environments can marginally determine its cost.The marginal effect analyses show an amplifying effect of local political purges and a mitigating effect of market development.In addition to the heterogeneous effect analysis, I further demonstrate the role of political risk arising from corruption dismissals.Specifically, I compare the effect of revolving-door recruitment before and during the campaign.I show that the cost of revolvingdoor recruitment appears only after the start of the campaign in late 2012, not in the precampaign period, when political risks are low.
By documenting the sizable cost of revolving-door recruitment in a major authoritarian country, this article contributes to the comparative analysis of politics-business revolving doors.The existing literature, which primarily concentrates on advanced democracies, presents abundant evidence on returns on revolving-door recruitment (Bertrand, Bombardini, and Trebbi 2014;Blanes i Vidal, Draca, and Fons-Rosen 2012;Luechinger and Moser 2014).A recent article, however, shows the declining value of revolving-door connections in the US because of increased political turnover (Strickland 2020).As a step forward, this study reveals the sizable cost of the politics-business revolving door during a turbulent time in a country with weak institutions.I show that revolving-door channels transmitted the uncertainty related to the connected officials' status during purges in a massive anticorruption campaign, discouraged investment and borrowing, and ultimately led to profit loss.
More broadly, this study contributes to the research on corporate political connections (Szakonyi 2018;Truex 2014;Wang 2017).The cost of political connections echoes the idea of the transitional gains trap introduced by Tullock (1975), who argues that rent seekers always fail to obtain above-normal returns on their original investment.Recently, scholars have provided empirical evidence on the cost of the political connections arising from rent-seeking activities and the poor managerial decisions of firm executives with government backgrounds (see, for example, Bertrand et al. 2018;Fan, Wong, and Zhang 2007).In contrast to these corporate governance explanations, the mechanism proposed in this article emphasizes the perception of firms by external parties.I argue that political connections act as a negative signal to investors and lenders, thereby discouraging external financing.

The Cost and Benefits of Political Connections
Research shows that revolving-door officials offer two types of assets to corporations.First, firms hire revolving-door officials for their expertise to influence policies that advance firms' business interests.The informational role of revolving-door hiring is widely recognized in advanced democracies, particularly in the US lobbying industry.In the US, revolving-door lobbyists convey their clients' demands to legislators, voice interest group preferences, and advance the quality of policy making (Austen-Smith and Wright 1992;Bennedsen and Feldmann 2006).In addition to seeking expertise, firms or interest groups hire former officials for their connections in the incumbent government.By building ties with politicians, firms: reduce the negative impacts of policy uncertainty; obtain preferential treatment in government procurement, subsidies, and taxation; and consequently gain considerable financial returns (Eggers and Hainmueller 2009;Luechinger and Moser 2014;Simon 2022).
However, building political connections with public officials is by no means cost-free.A dominant explanation to account for the cost of political connections is that it weakens corporate governance.As suggested by Shleifer and Vishny (1993), politicians extract rents through these connections, and the marginal costs of rent seeking may offset the benefits of the connections.Mounting empirical evidence supports this rent-seeking argument.Analyzing 20,000 listed firms in forty-seven countries, Faccio (2006) shows that hiring former politicians as board directors does not increase corporate value.Similarly, despite generating extra revenue through informal ties with government officials in the short run, political connections fail to improve the long-term profitability of firms in Italy (Cingano and Pinotti 2013).
As Tullock (1975) notes, rent seekers can hardly change inefficient institutions, even when profits decline.Tullock's idea of the transitional gain trap manifests in a way that the costs of political connections can even outweigh the benefits.A growing empirical literature documents the cost of political connections in various contexts.For example, firms suffer from efficiency losses caused by the electoral concerns of their connected public officials in France (Bertrand et al. 2018).Similarly, firms led by chief executive officers (CEOs) who formerly worked in government underperform their unconnected counterparts because these CEOs make poor personnel decisions in terms of hiring more former government colleagues than professionals (Fan, Wong, and Zhang 2007).

Political Risk in the Revolving-Door Channel
This study shifts the research focus on the cost of political connections from an efficiency loss mechanism to the external political risk channel.I argue that political connections convey political risks that can harm firm profitability.The defining feature of political risk is an uncertain political environment.Political uncertainty manifests in various forms, including abnormal political turnover, massive protests, and arbitrary policy shifts.In the extant literature, scholars have focused on macro-level political risks, arguing that they affect foreign direct investment (FDI), asset prices, international capital flows, employment growth, and business cycles in a crossnational setting (for example, Belo, Gala, and Li 2013;Besley and Mueller 2018;Handley and Limão 2015;Jensen 2008;Siemer, Verdelhan, and Gourio 2015).Political risks also vary across different institutional contexts.There is growing consensus that political risk is higher under authoritarian regimes, where the institutions are weak, than in advanced democracies (Busse and Hefeker 2007;Jensen 2003).
Beyond the cross-country variation, political risks affect individual firms differently through their political ties with the government.Specifically, political connections incur a cost to firms when external political risks are high.This political risk mechanism was first introduced by finance scholars, who focus on how investors make decisions based on nonmarket information (Acemoglu et al. 2016;Ferguson and Voth 2008).Research shows that financiers are informed about the political background of firms (revolving-door hiring, relatives in government and so on) through various means, including information disclosure files, media coverage, and consulting services (Chen and Kung 2019;Fisman 2001;Li 2021).While viewing a firm's political connections as a signal of government protection and preferential treatment in normal times, financiers perceive political ties as a risk that transmits the uncertainty from the political connections when political turmoil emerges.They often manage the risk by reducing these politically connected firms' positions in their portfolios.As a result, the stock price of these connected firms declines.Considerable empirical evidence demonstrates these short-run negative returns on political connections triggered by political turmoil.For example, the groundbreaking work of Fisman (2001) shows that politically connected firms in Indonesia suffered from the declining health of their political patron, Suharto.Similarly, Acemoglu, Hassan, and Tahoun (2018) show that all firms with ties to Mubarak's government experienced profit losses because of the tremendous political uncertainty triggered by the ongoing street protests in Egypt.
While this immediate cost of political risk to the stock market is widely documented in the literature (as illustrated in the upper part of the diagram in Figure 1), I further theorize that the political risks identified in the stock market have a spillover effect that can harm the profitability of politically connected firms through an external perception mechanism (see the lower part of the diagram in Figure 1).Recent literature has shown that political risk impacts the perceptions of firms held by external financiers who provide credit and investments (Hassan et al. 2019).Once financiers observe the profit losses triggered by multiple incidents related to a firm's connected politicians, their future valuation of political connections changes.Ties with public officials whose status is unstable are no longer a signal of strong political capital, but a possible liability that drags firms into future political turmoil.The reputational cost means that politically connected firms face difficulties in obtaining external financial resources, including equity and debt financing.In countries with poor legal institutions, lenders and investors rely heavily on nonmarket information, such as the political background of a firm, to make decisions (Boubakri et al. 2012;Malesky and Taussig 2009).When the market perception of politically connected firms deteriorates, investors are less likely to buy additional shares or bonds from these firms.In the credit market, risk-averse lenders are less willing to approve loans to firms with poor reputations to avoid future default.Lacking sufficient credit and investment, firms encounter a shortage of external financing, so that they are unable to expand their business or invest in profitable projects, thereby leading to a decline in firm profits.4I formulate the key hypothesis of political risk as follows Political risk hypothesis: Politically connected firms earn lower profits than firms with no connections to revolving-door officials in a high-risk environment.

Context
This study takes China's ongoing anticorruption campaign as a case to examine political risk theory.China serves as a quintessential example of a weak institutional context, as its transition of political power is not regularized and the government has the power to confiscate private property.Specifically, the recent anticorruption campaign in China created a political shock, whereby millions of public officials were dismissed from public office, investigated, and prosecuted at a scale of political turnover that is as sizable as that of revolutions and great purges in other countries.This section describes how the anticorruption campaign serves as a major political shock to politicians and firms, and how the value of revolving-door connections has declined in the campaign.Since it began in late 2012, the ongoing anticorruption campaign in China has posed considerable risks to public officials due to its scale and intensity (Manion 2016;Wang 2022) Moreover, the dismissals of public officials were spread across the country spatially, occurring in 323 of the 330 cities in China.The draconian campaign created a tense atmosphere among public officials given the recognition that no one below Xi Jinping was completely safe.A distinct feature of this campaign relative to previous corruption crackdowns is that "immunity" was no longer granted to senior political leaders.From 2012 to 2016, over one hundred officials at or above the level of the deputy minister were purged.In the most high-profile case, Zhou Yongkang, formerly the second-most influential figure in the Party, was investigated and sentenced to life in prison in the first year of the campaign.
The massive purges during the ongoing campaign have significantly affected the value of patronage connections in the political system.Belying the conventional understanding that anticorruption campaigns target only political foes, personal ties to top Chinese leaders have provided little protection in the ongoing corruption crackdown (Lu and Lorentzen 2016).More strikingly, local political leaders now show a bias against their connected subordinates when making appointment decisions to reduce career risk due to perceptions of corruption (Li and Manion 2021).
Beyond changing the political landscape, the political risk ignited by the campaign goes even deeper into the boardrooms of publicly traded firms.From November 2012 to December 2016, eighty-two firm directors or senior executives were investigated for corruption.Most of these corruption investigations in firms were severe.According to the available sentencing data on twentynine cases, the corruption charges against firm directors or senior members ranged from six years of imprisonment to a suspended death penalty combined with the confiscation of private property.Investigations led to future risks of expropriation for nonstate-owned enterprises.For example, Jinlu Group, a mining company with political ties with Zhou Yongkang's protégé, was divided and acquired cheaply by the Sichuan State-owned Assets Supervision and Administration Commission.In state-owned enterprises (SOEs), the investigations of revolvingdoor officials were followed by further waves of Central Disciplinary Inspection Commission (CDIC) inspections, investigations, and detention.These negative shocks create a looming prospect of building political connections with power politicians in China.In the next section, I specifically discuss the rise and fall of the returns to political connections before and during the anticorruption campaign.

Rise and Decline of the Value of Political Connections
In recent decades, political connections established through the revolving-door channel have become widespread among publicly traded firms (Wang 2016).Chinese financiers are also aware of the revolving-door employment of firms.Due to the mandatory disclosure rules, information on board composition and the career experience of the top firm personnel are publicly available.Large financial data corporations have developed databases on firm personnel for financiers and analysts. 6While not every financier can obtain complete information on the dynamics of revolving-door hires, public media constantly report the recruitment and departure of revolving-door officials.For example, Investor Journal, a business-focused newspaper, reveals that thousands of retired officials served in publicly traded firms before the start of the campaign (Zhou 2010).Financiers not only have access to information on political connections, but also view political ties as a reputational signal for political background in China.Truex (2014) presents both qualitative and quantitative evidence about the reputation effect of political connections.Investors perceive enhancements in political connections, specifically the board director obtaining National People's Congress membership, as a reputational boost for firms.Such reputational advancement consequently drives up firm profits.
While bringing reputational benefits to firms in normal times, political connections also convey the status uncertainty of politicians connected to the market.After the start of the campaign, state media raised corruption concerns in the revolving-door channel.For example, People's Daily (Wu 2013), the mouthpiece of the Party, has covered the slow decline in the revolving-door recruitment of publicly traded firms and criticized some hires violating cadre personnel regulations.Investors are also concerned about the rise of political risk under the campaign.The scholarly literature shows that investors perceive political connections as serving as a liability to firms during the anticorruption campaign because of the growing concern of corruption involvement.Since stock shorting is not allowed in China, a risk-averse choice for investors is to reduce the investment in politically connected firms, thereby lowering the returns of these firms (Pan and Tian 2017;Wang et al. 2018).A reputation decline of politically connected firms also emerges in the financial market.On the two major stock forums in China, Guba and Tonghuashun, investors have extensive discussions about the negative impact of the anticorruption campaign in thousands of posts.Figure A1 in the Online Appendix shows several examples of these forum discussions.Investors have called these firms involved in anticorruption campaigns fan fu gai nian gu ("anticorruption concept stock") and discussed how to adjust their portfolio to avoid future losses.This study further finds that this exacerbation of firm reputation reduces external financing and consequently jeopardizes firm profits.Before presenting the empirical evidence to show the cost of political connections, the next section describes the data and stylized facts about the revolving-door connections and other firm-level financial indicators.

Data and Stylized Facts
To empirically examine political risk theory, I began by compiling a database that contains information on revolving-door recruitment and financial indicators for 3,466 companies listed publicly on the Chinese stock market between 2012 and 2016.The sample includes all A-share firms that use RMB as their transaction currency and are listed publicly on the Shanghai or Shenzhen Stock Exchanges.I gather data from three sources.
The data on firms' revolving-door connections are from the Chinese Revolving-Door Officials Database (CRDOD), an original database that contains extensive demographic and career information for all former government officials who have been firm executives in any publicly traded Chinese firm since 1994.I first obtain the raw text of the resumes of all firm executives and directors from cninfo.com, the official information disclosure agency of the Chinese Securities Regulatory Commissions.I transform the raw text data into a standardized career experience database using both deep-learning algorithms that detect named entities and a manual check.For each revolving-door official, the database provides their rank, location, government branch, and tenure in each work position on their resume.I aggregate the information on revolving-door officials to the firm-year level to analyze the overall impact of revolving-door hires on firm performance. 7 In this article, I define revolving-door officials as senior firm executives or board directors who have previous work experience in the two executive branches: the Party and the government. 8My data show that revolving-door recruitment has become increasingly popular among publicly traded Chinese firms in recent decades (see Figure 3).The number of revolving-door officials increased rapidly from 1994 to 2014 but displayed a slight decrease in 2015 and 2016.This decline is partly due to the discouraging effect of the anticorruption campaign and the increased enforcement during cooling-off periods.9Nevertheless, the hiring of former officials is still a major corporate political strategy.Since the start of the campaign, 2,481 of the 3,466 firms (71 per cent) have hired at least one revolving-door official.The preference for revolving-door recruitment varies across localities.In Online Appendix C, I show that firms located in poorer areas hire more revolvers than firms in more developed regions, as there is a strong correlation between gross domestic product (GDP) and revolving-door recruitment.The distribution of revolving-door officials hired by firms is left-skewed, in that 81 per cent of listed politically connected firms have hired fewer than three former officials, while extreme cases, such as Shanghai Pudong Development Bank (SHE: 600000), have recruited as many as twenty revolving-door officials.
Table A1 in the Online Appendix shows the demographic and career information for revolving-door officials compared with that of other firm executives or board directors; the data indicate that about 9 per cent of firm executives or board directors were previously public officials.Moreover, these revolving-door officials are more commonly male, older, and better educated than other firm executives.They also have a stronger political background than other firm executives, as shown by the share of CPC members, People's Congress members, and Chinese People's Political Consultative Conference (CPPCC) members within this group.In addition to the data on revolving-door officials, I obtain the daily stock price data for the event study using the China Stock Market & Accounting Research (CSMAR) database, a widely used financial database.I gather information on firms' financial indicators and registration using the RSSET financial database, which is developed by a data services company based in Beijing.I also double-check the data with CSMAR.I obtain data on the officials purged in the ongoing anticorruption campaign from the CCID developed by Wang and Dickson (2021).

Results
I examine the political risk theory illustrated in Figure 1 by taking several steps.The first step is to verify a premise for the spillover effect of political risk.Specifically, I document how dismissals of revolving-door officials due to corruption lead to collateral damage for other firms that hire revolvers by conducting an event study analysis.After verifying this premise, I document the cost of revolving-door connections during anticorruption efforts by using firm-year panel data and show that the results are robust to various model specifications, alternative measurements, and the parallel trend assumption.I then examine the external perception and financing mechanism.Using a novel measure of market perception and bank-loan data, I show that revolving-door recruitment imposes a reputational risk for politically connected firms and results in a shortage of external financing.Last, I rule out alternative channels by which the cost of revolving-door recruitment is driven by the decline in favorable treatment by the government.Evidence on taxation, subsidies, and land shows that firms that hire revolvers do not receive less preferential treatment than unconnected firms.

A Premise Check
A crucial premise for the cost of revolving-door connections is that political risks have a spillover effect: firms that hire revolvers, even those without direct involvement in a corruption scandal, suffer from collateral damage due to the political risk.Before documenting the cost of revolvingdoor hires, I empirically verify this premise by using an event study analysis with a multiple-group comparison (see, for example, Wang 2015).Specifically, I define three types of firms to examine the collateral damage due to political risk through the stock market.The first group, Losers, includes firms with revolving-door officials who were involved in a corruption scandal.The second group, Survivors, includes firms that hire former officials but do not directly experience a corruption crackdown.The last group, Unconnected, includes firms that do not recruit any revolvers.The empirical implication of political risk having a spillover effect is that not only Losers, but also Survivors, suffer from the political risks triggered by the anticorruption campaign.
In total, I manually collect a dataset on thirty-eight events in which listed firms' revolvingdoor officials were investigated for corruption from 2012 to 2016.I obtain these event data from multiple sources, including the CCID, the corrupt officials database constructed by China Files, and the media coverage of corruption investigations in China's major finance news media, such as Caixin and Diyicaijin.To create a valid counterfactual, I use one-to-two matching and obtain fifty-seven Survivors and twenty-three Unconnected firms. 11Specifically, I compare the firm-level daily returns for the three types of firms using daily stock return data (CSMAR database).
I use CARs as a proxy for firm returns in response to the investigation event.Cumulative Abnormal Returns (CARs) are widely used in finance and economics research to capture abnormal changes in the stock market returns of firms and are calculated by summing the cumulative differences between the expected and actual returns of a security.Details on computing CARs are available in Online Appendix B. I present the time trend in the CARs from the event day (t = 0) to twenty trading days after the event (t = 20) in Figure 4. 12 It shows that Losers experience profit losses because of their direct involvement in corruption scandals, as shown by the downward trend in the CARs during the twenty days after the investigation date.In contrast, Unconnected firms do not suffer from these scandals, as their CARs do not change drastically within the event window.More importantly, Survivors also suffer from the collateral damage of corruption scandals, as their CARs decline after the start of events.While the collateral losses of Survivors are not as large as the losses of Losers, their size is salient when compared to unconnected firms that do not suffer from corruption scandal shocks.
In addition to the visualization, I quantify the direct and collateral losses caused by corruption dismissals.I do so by estimating the differential effects of corruption investigations on the CARs of the three types of firms. 13Table B1 in the Online Appendix presents the results, showing a robust pattern in line with the visualization.Not surprisingly, firms that hire revolvers who are investigated for corruption (Losers) experience consistent profit losses and underperform by 2.8 per cent to 6.5 per cent relative to Unconnected firms in event windows ranging from three days to twenty days.The collateral damage to Survivors is also salient.Even without direct involvement in a corruption scandal, Survivors experience profit losses of approximately 1.1 per cent to 3.3 per cent in various event windows (three to twenty days), suggesting that the market loses confidence in other politically connected firms after continual corruption investigations.Overall, the results show that politically connected firms, with (Losers) or without (Survivors) direct involvement in corruption, experience declining returns following corruption investigations.The result 11 Using one-to-two matching enables me to observe a sufficient number of firms in each group.
12 The event study window varies across different studies, and an event window of twenty trading days is common in finance studies because it covers almost a calendar month (see, for example, Corhay and Rad 1996;Lease, Masulis, and Page 1991).

13
The model is specified as follows:Y ikt = β 1 FirmType i + λ k + γ t , ϵ it , where Y ikt are the CARs of firm i on dayt and FirmType i is a categorical firm-type variable coded as "0" if firm i does not have any political connections (Unconnected), as "1" if firm i hires revolvers but does not experience a purge (Survivors), and as "2" if the firm is politically connected and one of the firm's board directors is investigated for corruption.I include sector and day fixed effects (λ k , γ t ) in the model.provides empirical support for the crucial premise for the cost of revolving-door connections, as illustrated in the theoretical framework in Figure 1.

The Cost of Revolving-Door Recruitment
After verifying the premise that all politically connected firms suffer from political risks, I estimate the cost of revolving-door recruitment on firm profitability by using a fixed-effects model.The model is specified as follows: where Y it is a profitability metric for firm i in year t.The dependent variable in the baseline specification is ROA, which indicates how profitable a firm is relative to its overall resources.Not only is ROA a key indicator that managers, analysts, and investors use to evaluate a firm's performance (Birken and Curry 2021), but it has also been widely used in political-economy and finance research (Bertrand et al. 2018;Li et al. 2008;Truex 2014). 14It is calculated by dividing profit margins by total assets.Larger ROA values suggest a greater ability of firms to generate profits.Figure A2 in the Online Appendix shows the time trend in ROA, suggesting that there is considerable cross-time variation, which allows me to conduct within-firm estimation.In addition to ROA, I use three alternative measures of firm performance, including return on equity, profit margins, and earnings per share.I present the corresponding results in the Online Appendix.
RevolvingDoor it is my independent variable of interest.In this article, I define revolving-door officials as retired officials who used to work in one of the two executive branches of the Chinese government: the government or the Party.In the main analysis, I choose to use a binary measure of a firm's revolving-door recruitment status, coding it as "1" if at least one of the board directors or senior executives of firm i in year t was previously a public official and as "0" otherwise. 15 Treatment status is displayed in Figure A3 in the Online Appendix.
X it is a set of time-variant firm-level covariates.First, firms that hire revolving-door officials may also have other types of political connections that can affect firm profitability.The research shows that firm directors or executives who are legislators in the People's Congress or CPPCC (Hou 2019;Truex 2014), or are CPC members (Li et al. 2008), can affect firms' enormous benefits.To exclude the effect of these alternative political connections, I constructed binary measures to indicate whether a firm has at least one executive and board director who is (1) a CPC member, (2) a People's Congress deputy, or (3) a CPPCC member.Beyond controls for government-business relations, I include economic covariates that can affect firm profitability.First, large firms are more likely to establish political connections.Following the literature in economics, I control for firm size by taking the logarithms of total assets.Moreover, firms with larger revenue have more financial resources to hire retired officials.I also control for firm revenue (logged). 16Summary statistics of the dependent variables, independent variables, and covariates are presented in Table A2 in the Online Appendix.λ i and γ t capture the firm and year fixed effects, respectively, while ϵ it is the error term.Standard errors are clustered at the firm level to account for the serial correlation in the observations of each firm.The results are also robust to alternative clustering strategies (Table A3 in the Online Appendix).
The main results are presented in Table 1.I begin by regressing ROA on RevolvingDoor and the year and firm fixed effects.I find that the estimated coefficient on RevolvingDoor is −0.63 and statistically significant (see Column 1).In Column 2, I add covariates into the model.The standard error for RevolvingDoor decreases slightly, and the estimate remains significant at the 1 per cent level.From Columns 3 to 5, I adopt a stricter estimation strategy than that in the baseline Estimating the causal effect with a continuous treatment is challenging in designed-based causal inference because its assignment mechanism is complicated (Xu 2021).As it is easier for interpretation purposes, I use the binary measure in the baseline and check the robustness by using a continuous measure of revolving-door recruitment.specification by including sector-and province-specific time trends.Including these trends ensures that the effects of all unobserved sector-and province-specific differences (for example, local and sectoral trends in firm profitability, and the business and regulatory environment) that vary smoothly over time are removed from the parameter of interest,β 1 .With the inclusion of both province-and sector-specific time trends, the full model in Column 5 yields a significant coefficient on revolving-door recruitment.In addition to its statistical significance, the magnitude of the effect of the politics-business revolving door is sizable: given that the average ROA of firms analyzed in this article is 3.8 per cent, the most conservative estimate shows that the cost of revolving-door employment is equivalent to approximately 14 per cent of the average ROA.A back-of-the-envelope calculation suggests that the profit loss caused by the revolving-door channel is approximately RMB14 million (USD2.2 million) a year for a firm with average assets.
Although the baseline specification yields significant estimates, we need to be concerned about the parallel trends assumption, a key premise underlying the validity of the baseline specification.Specifically, in this study, there must be no divergence in the trends of the profitability measures of firms with and without revolvers before the firms with revolvers recruited former officials.To show whether this assumption holds, I use a dynamic specification, adding three leads and lags of the revolving-door dummy to capture the potential effects during the three years before and the three years after actual treatment.The model is specified as follows: where the dependent variables, set of covariates, and two-way fixed effects are the same as those in the baseline specification.Instead of using one treatment variable, I incorporate a set of dummy variables, t+3 t 1 =t−3 b t 1 RevolvingDoor it 1 , to obtain flexible estimates of the impact of revolving-door recruitment on firm profitability.
Figure 5 shows the dynamic effects of revolving-door employment by presenting both the point estimates and 95 per cent confidence intervals for the lags and leads (for the corresponding regression results, see Table A5 in the Online Appendix).The estimates for revolving-door recruitment are negative and statistically significant for the year in which the former official was hired.More importantly, firms with revolvers do not experience profit loss until they actually hire the officials, and the cost quickly disappears when firms terminate their revolving-door recruitment, as shown by the insignificant estimates for all lead and lagged terms.I also examine the dynamic effects using the alternative estimation strategy used by De Boef and Keele (2008).Consistent with the findings in Figure 5, I find a significant result for the short-term effect of revolving-door recruitment but mixed findings regarding its long-term effect using autoregressive distributed lag (ADL) and error correction model (ECM) specifications (see Table A6 in the Online Appendix).
In addition to analyzing the parallel trend assumption, I conduct five sets of robustness checks, which support the validity of my baseline findings.First, I show that the cost of revolving-door recruitment is still found when I use three alternative measures of firm profit: (1) profit margin, (2) return on equity, and (3) earnings per share (see Table A7 in the Online Appendix).Secondly, I show that the main finding is robust to the measurement of revolving-door recruitment in levels and shares (see Tables A8 and A9 in the Online Appendix).Thirdly, I conduct a placebo test using other forms of political connection as a fake treatment (see Table A10 in the Online Appendix).Fourthly, I show the cost of revolving-door recruitment by excluding all SOEs, as officials may be transferred to these firms by the upper-level Party Organization Department (see Table A11 in the Online Appendix).Lastly, I address concerns of potential confounders using a propensity score model and obtain results consistent with the baseline findings (see Table A12 in the Online Appendix).In summary, none of these robustness tests challenge the main finding that revolving-door recruitment imposes a significant cost on firms in China.

External Perception and Financing
Beyond documenting the cost of revolving-door connections, I empirically examine the external perception mechanism shown in the diagram (see Figure 1).The mechanism suggests that revolving-door connections serve as a negative signal to financiers in a high-risk environment and result in a shortage of external financing.The existing literature offers some suggestive evidence on this market perception mechanism (Ferguson and Voth 2008;Truex 2014), but direct evidence on sentimental and behavioral changes is limited.In this study, I document how financiers perceive political risks by developing a measure of financier sentiment.I construct this measure using the text of equity research reports published by major financial institutions (CSMAR database).In these reports, analysts of major financial institutions show their prospects for specific firms and make portfolio recommendations.To construct the firm-level measures, I first extract 570,000 sentences that are directly related to specific listed firms from the raw text of over 1.2 million research reports.17I then use a supervised learning method to classify investor sentiment into two categories: positive and negative (for the details, see Online Appendix D).The upper two panels of Figure 6 show the top one hundred words that appear in the most positive and negative reports. 18Several optimistic keywords, such as "recommend" (推荐), "develop" (发展), and "optimistic" (乐观), are widely mentioned in the positive reports.In contrast, in the negative reports, pessimistic phrases, including "fall" (下跌), "shock" (冲击), and "decline" (降低), are widespread.The bottom two panels show the number of positive and negative reports for each firm.On average, firms receive more positive (5.1 reports per year) than negative (2.2 reports per year) evaluations.Notably, analysts provide more forecasts for large firms, as shown by the right-skewed distribution of firm reports.To address this issue, I measure market perceptions by computing the ratio of positive sentiments to the total number of forecasts for each specific firm in each year.
In addition to documenting the sentimental changes, I examine how political risks affect the external financing of politically connected firms.Following Khwaja and Mian (2005) and Li et al. (2008), I first document the changes in external financing by focusing on corporate lending.Similar to the chilling effect of political risk on market perception, lenders may provide fewer loans to firms that hire revolving-door officials than to unconnected firms.I measure corporate lending by using the total amount of approved bank loans for each firm in a certain budget year (CSMAR database).Moreover, firms may obtain less financial capital from other sources.Following the research of Szakonyi (2018), I gauge the ability of firms to access other financial resources by using liabilities (RSSET database).
Table 2 presents the results that support the external perception mechanism. 19The regression analysis yields negative and significant estimates in Columns 1 to 3, suggesting that firms that hire former officials have more negative ratings than do firms that do not hire former government officials.Moreover, connected firms obtain fewer loans than unconnected firms, with the difference amounting to approximately RMB220 million (USD31 million) according to the estimation; they are also less able to obtain financial resources than unconnected firms, as shown by the 6.8 per cent gap in firm liabilities.

Note:
The upper two panels show the word clouds for reports with positive and negative sentiments.The size of each word is proportional to its frequency.The bottom two panels show the frequency with which firms receive positive and negative reports.
To summarize, I find that the recruitment of former officials, aside from its direct costs on stock prices, imposes a significant cost on politically connected firms in turbulent times by lowering the ability to obtain external financing.These findings again confirm that external perceptions are the channel through which the politics-business revolving door reduces firm profitability.

Alternative Explanation: Preferential Treatment
In addition to the external perception mechanism, there is an alternative explanation for the profit losses of politically connected firms: the decline in preferential treatment.Mounting research shows that the key asset provided by former officials is their ability to obtain favorable treatment (Chen and Kung 2019; Goldman, Rocholl, and So 2013;Luechinger and Moser 2014).In the context of this article, massive corruption crackdowns may increase the cost of exchanging favors between firms and politicians.Facing increased costs of corruption, revolving-door officials grant less-favorable treatment to firms.In China, firms rely heavily on offers of favorable government treatment in relevant matters (Li et al. 2008).Losing these profitable benefits may result in cash-flow shortages and thereby affect firm performance.
Of the various forms of preferential treatment, I focus on three important benefits that retired officials provide to firms in China: access to subsidies, access to land, and favorable tax treatment. 20The research has shown that government officials who anticipate revolving-door job opportunities provide favorable government subsidies to their future employers (Li 2021).To measure the size of these subsidies, I aggregate the government subsidy data from the CSMAR database to the firm-year level.In addition to government subsidies, local governments use tax holidays as an incentive to attract foreign and domestic investment (Adhikari, Derashid, and Zhang 2006;Jensen and Malesky 2018).Specifically, scholars have found that tax benefits are a common type of favorable policy treatment granted to Chinese firms with close ties to the government (Li et al. 2008;Wu et al. 2012).I measure tax benefits (Tax) using the income tax paid by firms in the data from the CSMAR database.Land is another valuable resource controlled by local governments, which have the discretion to sell it to different types of buyers.The recent literature has also shown that political connectedness is a key determinant of a firm's ability to obtain price discounts in land transactions (Chen and Kung 2019).To address the effect of revolving-door connections on land purchases, I use real estate investment (RSSET dataset) as a proxy for preferential treatment in land purchases (Land).
Table A13 in the Online Appendix presents the results for favorable treatment.I regress measures of preferential benefits on revolving-door recruitment.The results show that compared with Due to limited data availability, information on other forms of preferential treatment, such as the granting of government procurement contracts, is not available.their unconnected counterparts, firms that hire revolving-door officials do not obtain fewer government subsidies or less favorable tax rates, and do not invest less in real estate.Several explanations can account for this insignificant difference in preferential treatment.An intuitive reason is that some favorable treatment was offered by officials before 2012 but was implemented during the campaign.In this case, firms could continue to receive a favorable tax rate or receive subsidies that were agreed upon with the government before the beginning of the campaign.However, a more convincing explanation is that firms and revolving-door officials were uncertain about their future when the gloom of the anticorruption campaign suddenly appeared in 2012.According to several interviews with firm managers and revolving-door officials, both groups held wait-and-see attitudes toward the anticorruption campaign because it was possible that the campaign would not last long. 21Maintaining a quid pro quo relationship can still be profitable in the long run.The decline in the number of revolving-door officials began in 2015, approximately three years after the start of the campaign, providing suggestive evidence that firms adopted a wait-and-see attitude toward revolving-door employment (see Figure 3).Taken together, the evidence does not support the preferential treatment explanation.

Effect Heterogeneity
This section shows several extensional analyses that further support political risk theory. 22The first set of analyses focuses on effect heterogeneity.As political connections serve as the nexus between government and market, both the local political and economic environment can marginally determine the returns on political connections.Given that the anticorruption campaign is the key driver for firm-level political risk, an empirical support for political risk theory is that the cost of revolving-door recruitment is higher in areas with intensive purges of political leaders.To test this conjecture, I construct the purge intensity measure by counting officials who (1) work in a firm's headquarters' city, (2) hold a rank at least at the division level (处级), and (3) were dismissed for corruption.Examples of officials with such a rank are the department heads of city governments, city mayors, or CPC secretaries for prefectural cities.The purges of these officials create enormous political uncertainty because they are the key decision makers for local economic policies.I estimate the effect heterogeneity across areas with different purge intensities (logged number of purged officials) by performing a marginal effect analysis.The upper panel of Figure H1 in the Online Appendix shows that the marginal cost of revolving-door recruitment increases with purge intensity.To obtain a better sense of the magnitude, I conduct interaction analyses, reported in Table A14 in the Online Appendix.The coefficients on RevolvingDoor*Purge range from −0.27 to −0.30 and are all statistically significant.
In addition to the political environment, the market environment is another important moderator that affects corporate political strategy and its outcomes in the context of weak institutions (Gehlbach, Sonin, and Zhuravskaya 2010;Markus 2012).Although China has a unitary legal system to regulate the market, the enforcement of law and the progress of market development in China vary significantly across regions.According to the World Bank's (2008) Doing Business in China report, establishing a business in the north-western provinces costs entrepreneurs over 100 per cent more in expenses and takes 48 per cent more time than in the south-eastern provinces.As numerous studies have shown, a benign market environment and political connections are substitutes (Kung and Ma 2018;Li et al. 2008).In places with an advanced market environment, firms can access such resources as land and labor without seeking connections with officials.In contrast, in a poor business environment, firms are more likely to establish political connections to secure their property rights (Hou 2019).To document this substitution effect, I use the province-level market development index developed by Wang, Gang and Lipeng (2019) as a proxy for the strength of market development.Accounting to Fan, Wang, and Ma (2012), this index reflects the ongoing development of the market in different Chinese provinces along various dimensions, including property rights protection, the tax burden, and government interference. 23The market development index ranges from 0 to 10, with a larger value indicating overall stronger market development progress.The lower panel of Figure H1 in the Online Appendix shows the results of the marginal effect analysis (for regression results, see Table A14 in the Online Appendix).The estimation suggests that the cost of revolving-door hires diminishes with greater market development.Firms located in provinces with an advanced market environment (for example, Zhejiang or Shanghai) have approximately 0.3 per cent higher ROA than their peers in areas with a weak market environment (for example, Hebei or Xinjiang).The marginal effect analysis shows that market development acts as a salient moderator that reduces the cost of political uncertainty. 24  Comparison of Losses before and during the Campaign Finally, I demonstrate that political risks are the key driver of the profit losses incurred by politically connected firms by conducting an analysis using firm data from both before and during the anticorruption campaign.As political risk theory suggests, the scale and intensity of the anticorruption crackdown created a distinctive high-risk environment that imposes costs on firms that exchange favors with revolving-door officials.Therefore, I expect that firms with revolvers do not experience profit losses in periods of low purge intensity.In other words, the cost of revolving-door recruitment is expected to appear only during the campaign period, not during the precampaign period.
To empirically examine this key conjecture, I first estimate the effect of revolving-door employment on firm performance before the start of the ongoing anticorruption campaign (2008)(2009)(2010)(2011).The results are presented in Table G1 in the Online Appendix.Across the different specifications, the estimates of revolving-door recruitment are positive but not statistically significant in the precampaign period when political risk was low.I also adopt a flexible estimation strategy to obtain the marginal effect of revolving-door recruitment in the years before and during the campaign.To do so, I employ a kernel smoothing estimator and visualize the local marginal effects in Figure H2 in the Online Appendix.The marginal effect of revolving-door recruitment, which is closer to zero and statistically insignificant before 2012, declines sharply from 2012 onward and is significant at the 5 per cent level for most of the years during the campaign.I also use a model that interacts the key independent variable RevolvingDoor with the campaign period dummy (year ≥ 2012) (for the detailed specification, see Online Appendix G).Table G2 in the Online Appendix shows the results, which are consistent with those of the flexible estimation, suggesting that political risk increased after the campaign started.Taken together, the findings support the hypothesis that political risks have imposed a cost on firms that recruit revolving-door officials during the anticorruption campaign.

Conclusion
This article shows that the politics-business revolving door transmits political risk to firms, discourages investors, and reduces firm performance when institutions are weak.Using China's ongoing anticorruption campaign as a case study, I find that returns on the revolving-door channel have been negative since 2012, when the campaign began to sweep thousands of Chinese

23
The details of this market index and its validity checks are provided in Appendix E in the Online Appendix.24 I also examine cross-sector heterogeneity, showing that the cost of revolving-door recruitment is greater for firms in sectors with greater reliance on the government and higher entry barriers, including the natural resource, construction, finance, and real estate sectors (see Table A15 in the Online Appendix).

Figure 1 .
Figure 1.Theoretical framework.Note: PC stands for politically connected firms.
Figure 2 displays the time trend in, and spatial distribution of, corruption crackdowns; it shows a sharp increase in the dismissal of public officials since the beginning of the campaign.According to China's Corruption Investigations Database (CCID) compiled by Wang and Dickson (2021), the Disciplinary Inspection Commission of the CPC toppled over 17,000 public officials of various ranks from the state level (D m) to the sector level (m) in the first five years of the campaign (2012-16). 5

Figure 2 .
Figure 2. Time trend in the dismissal of officials for corruption.Source: China's Corruption Investigations Database. 10

7
The CRDOD (Version 3.0) has several advantages when compared to the existing firm personnel database constructed by China Stock Market & Accounting Research (CSMAR) and Wind, including a clear definition of political connections and thorough standardized coding of career experience.

Figure 3 .
Figure 3. Number of revolving-door officials in publicly traded Chinese firms.Source: CRDOD.

Figure 4 .
Figure 4. Time trends in CARs.

16
While scholars use different specifications, firm size and revenue are two common controls used in political-economy research when modeling firm profitability (see, for example, Ahamed, Almsafir, and Al-Smadi 2014;Khwaja and Mian 2005;Truex 2014).

Figure 5 .
Figure 5. Dynamic effect of revolving-door recruitment.Note: Dots and vertical lines denote the point estimates and 95 per cent confidence intervals for the effect of revolving-door recruitment on ROA (%).Controls are firm size, revenue, and CPC, People's Congress, and CPPCC membership.

Table 1 .
Effect of revolving-door recruitment on firm performance Notes:The dependent variable is ROA (%).Standard errors are clustered at the firm level and reported in parentheses.*p < 0.1; ** p < 0.05; *** p < 0.01.15

Table 2 .
External perceptions The dependent variables are positive sentiment, bank loans, and firm liabilities.The controls are operating revenue (in logs), firm size, People's Congress membership, CPPCC membership, and CPC membership.Standard errors are clustered at the firm level and reported in parentheses.* p < 0.1; ** p < 0.05; *** p < 0.01. Notes: