Uncertainty is pervasive in the business world. It exists not only in the firm’s market environment (e.g. R&D uncertainty, contract uncertainty) but also in its nonmarket environment (e.g. regulatory uncertainty). Research on corporate political strategy has provided rich insights into how firms use political strategy to mitigate undesirable consequences of regulatory uncertainty (e.g. Haveman et al., 2017; Bonardi, Holburn and Vanden Bergh, 2006). Nevertheless, less attention has been paid to uncertainty in political markets per se.
In this study, we analyze the effect of uncertainty in political markets on corporate political activity. Political markets are markets where firms exchange valuable political resources (votes, money, information, etc.) with policy makers/regulators for favorable policy/regulatory treatment or other resources valuable to the firm (information, loans, etc.) (Buchanan and Tullock 1962; North, 1990). The political markets framework is widely in studies on corporate political strategy. Scholars use the framework to analyze firms’ tendencies to participate in political markets (e.g. Hillman and Keim, 1995; Bonardi, Hillman and Keim, 2005; Kingsley, Vanden Bough and Bonardi, 2012) and the effectiveness of firms’ political strategies (Bonardi et al., 2006; Holburn and Vanden Bergh, 2013). The analysis has so far been primarily focused on the supply- and demand-side rivalries. Other market characteristics such as uncertainty are largely ignored.
This study aims to fill the void by adding a piece of evidence on the effect of political market uncertainty on corporate political activity and synthesizing a theoretical perspective to explain why. We examine political market uncertainty in an authoritarian regime – China, where a recent anticorruption campaign has generated considerable turbulences in regional political markets. Initiated in December 2012 by the then newly-elected President Xi Jinping, the campaign is considered the most influential anticorruption movement in modern China. As of May, 2016, more than 150 high-level government officials were cracked down. Even two members of the Central Committee of CCP were removed, which had never happened before in history. Tens of thousands of lower-level officials were investigated and convicted. The campaign not only affected the political world of China but also the corporate world. State owned enterprises (SOEs) were not immune to anticorruption investigations. It is estimated that over 100 CEOs of SOEs were forced to resign due to corruption investigation. Regardless of the actual effect of the campaign on corruption reduction, it has certainly disturbed the political market of China and generated considerable uncertainty.
We investigate changes in the political activities of listed firms in China during the anticorruption campaign. Since political markets in authoritarian regimes are quite different from those in democratic regimes (there is no lobbying expenditure or campaign contribution), we extend the political markets framework to include authoritarian regimes in the discussion and show the commonalities and differences of political markets in democracies and authoritarian regimes. We emphasize that in authoritarian regimes, rivalries on the supply- and demand-sides are quite different from those in democracies in nature. We then introduce political market uncertainty and analyze its impact on corporate political strategy. We suggest that when faced with an uncontrollable jump in political market uncertainty, firms are likely to postpone their political investments.
To test our prediction, we use an unconventional proxy for corporate political investment: corporate charitable donations. Politically-motivated corporate philanthropy is not unique to firms operating in authoritarian regimes. For instance, Bertrand et al. (2018) estimate that a sizable amount of corporate charitable donations in the U.S. were politically motivated. Nevertheless, in authoritarian regimes such as China, politically-motivated corporate philanthropy is particularly salient (Zhang, Marquis and Qiao, 2016). Firms use charitable donations as tool for developing goodwill and trust with local governments, who have critical stakes in the firms’ local operations. Although we cannot tease apart politically-motivate corporate philanthropy and corporate philanthropy motivated by other reasons such as activist pressure (Luo, Marquis and Zhang, 2016) or pure altruism (Baron, 2001), we are confident that the variations in corporate philanthropy caused by political reasons are large enough to survive the noise.
Cross-regional variations in political market uncertainty are utilized to identify the effect of political market uncertainty in a difference-in-difference (DD) model. There are 31 province-level administrative areas in China. The number of groups (31) is large enough for a DD model to generate reliable causal inferences (Bertrand, Duflo and Mullainathan, 2002). For robustness checks, we also try alternative specifications and alternative measures of the main variables.
This study contributes to the corporate political activity literature and the political markets literature. We stress the role of uncertainty, which is overlooked in the previous literature. Future studies can investigate the determinants of a firm’s capability in navigating political market uncertainty, or the heterogenous responses among firms when faced with such uncertainty.
THEORY AND HYPOTHESES
The concept of “political markets” originates from political economy. Key to this is the work by Nobel Laureate Buchanan and Tullock (1962), who propose that the policy making process in democracies is a market process. Rather than being determined by ideologies of government elites or the “public interest,” policy outcomes are driven by competition between interest groups. The nature of the process is self-interested exchanges between the actors (voters, political parties, social groups, firms, etc.). As Buchanan notes, “the relevant difference between (economic) markets and politics does not lie in the kinds of values/interests that persons pursue, but in the conditions under which they pursue their various interests. In the absence of individual interest, there is no interest” (1987: 246). One important contribution of the political markets perspective is the notion of exchange, which becomes the foundation of corporate political strategy analysis.
Below we discuss several important characteristics of political markets: supply- and demand-side rivalries, incomplete information, transaction cost and uncertainty.
Supply- and demand-side rivalries
According to the political markets framework, political markets consist of suppliers of public policy such as the executive, legislators, regulators and courts. It also consists of demanders of public policy – voters, firms, interest groups, etc. Suppliers and demanders interact with each other to exchange valuable resources such as money, information and policy outcomes. Different interest groups often have conflicting interests. Firms often compete with other policy demanders for policy outcomes favorable to themselves. For instance, in many cities, the ridesharing company Uber fought to gain regulatory approval of operating in the cities. In the meantime, Uber’s competitors—the traditional taxi firms—fought to persuade policy makers to ban ridesharing services.
Like how equilibrium market price is determined in the classical supply and demand model, the attractiveness of political markets to a firm is determined by the intensity of rivalries on the supply- and the demand-sides (Hillman and Keim, 1995; Bornardi et al., 2005; Kingsley et al., 2012). On the supply side, when there is greater rivalry among politicians, politicians are more likely to trade with interest groups for valuable resources to boost their electoral perspective. Empirical evidence are largely corroborative. For instance, Paik, Kang and Seamans (2018) investigate the determinants of Uber being banned in U.S. cities and find that the less political competition (measured by mayor tenure) there is in a city, the greater the likelihood of Uber being banned in the city. In a similar vein, researchers find utility firms are more likely to increase their political investment (campaign contributions) when there is greater rivalry between democrats and republicans (Holburn and Vanden Bergh, 2013). Those evidence suggest that firms invest in political markets calculatedly, and supplier-side rivalry is certainly one dimension they carefully consider.
On the demand side, firms compete with other firms or interest groups for favorable policy outcomes. In a study on utility rates regulation, researchers find that the more salient the power of consumers and environmental groups in the political market, the less effective the utility firms’ political strategies are (Bonardi et al., 2006). Unlike product markets, however, rivalry on the demand side of political markets tend to make the supply of policies less efficient (Bonardi et al., 2005). In sum, a political market is in general more attractive to a firm when there is greater rivalry on the supply side and less rivalry on the demand side.
Incomplete information, transaction cost and uncertainty
North (1990) points out that political markets are very imperfect. They are characterized with costly information, subjective models on the part of the actors, and imperfect enforcement of agreements. In addition, participants are boundedly rational and opportunistic. Those factors combined explain the market failure of political markets.
First, consider imperfect information and bounded rationality. The first question North (1990) asks is: how do interest groups know their interests? That is not to say that they are not sensitive to their welfare, but that they are often not fully aware of the consequences of candidate policies on their future welfare. The reason behind is information complexity, information cost and bounded rationality. Sometimes it takes an expert to fully understand the consequences of a policy. Sometimes the information needed to evaluate a policy is not equally available to everyone. Sometimes people just do not bother to think about it carefully because they satisfice (Simon, 1947).
The second limitation of political markets is high transaction cost and uncertainty. This problem arises from difficulties in measuring the rights being traded and difficulties in enforcing the exchange of rights. Exchange in political markets are essentially trading for future commitments. How can a buyer ensure that the seller will fulfill his commitment in the future? Both formal and informal institutions may help mitigate the problem, but the problem cannot be entirely avoided. In addition, the policy intentions and policy outcomes are often misaligned. Unanticipated policy outcomes often happen, even if the ex-ante intentions are there.
Political Markets in Authoritarian Regimes
Governments in authoritarian regimes often play a more important role in the economy and have greater influences on firms. In authoritarian regimes, governments control a great amount of resources. From the resource dependency perspective (Pfeffer & Salancik, 1978), it is critical for firms to develop political access to gain access to valuable resources. In addition, due to weak rule of law, policy making and regulatory execution in authoritarian regimes can sometimes be quite arbitrary. Wiggling room exists for government officials to selectively impose regulatory burdens on firms, or even abuse the entrusted power to extort firms. In authoritarian regimes going through economic reforms, firms rely on connections to officials to navigate policy uncertainty (Haveman et al., 2017). Political embeddedness is therefore critical for firms in authoritarian regimes (Marquis and Qian, 2013; Dickson, 2003). Empirical evidence shows that firms with better political access have better access to bank loans (Haveman et al., 2017), better innovation performance (Li, Xia and Zajac, 2017) and better financial performance (Wang and Qian, 2011).
Surprisingly, marketization progress and creation of formal institutions do not make political embeddedness any less important for firms in authoritarian regimes. Haveman et al. (2017) use panel data to see how marketization progress affected the benefits of political embeddedness of firms in China. Contrary to the intuition that marketization reduces the importance of ties and make transactions increasingly contractual (Peng, 2003), they find that as marketization level increase, the effect of political ties becomes even greater. Ang and Jia (2014) investigate how political connectedness affect the use of courts to resolve disputes for firms in China. They find a “perverse complementary” relationship between political connectedness and the use of courts: better connected firms are more likely to use courts, everything else equal.
Following the analytical framework of the above section, we discuss the supply- and demand-side rivalries and the factors that lead to imperfect political markets in authoritarian regimes. Additionally, we discuss the rights exchanged and the types of political strategies, which are quite different from those used in democracies.
Supply- and demand-side rivalries
Unlike political markets in democracies, the political markets in authoritarian regimes consists of a quite different set of participants. On the supply side, there is no between-party rivalry. Policy makers are appointed rather than elected. Rivalry on the supply side, therefore, is at most factional fights between groups within the party. On the demand side, the number of interest groups who have power affecting policies is too small. There are no voters. No legitimate activists. Hence the power structure is highly asymmetric. Government officials almost have absolute power over firms. What promotes collaboration between them, however, is economic incentive (Firms can only win by collaborating with governments, not by fighting governments).
The economic incentive that aligns the interests of firms and governments is GDP growth (Haveman et al., 2017). Government officials rely on GDP growth for promotions. Firms contribute to local GDP growth. Government officials therefore have the incentive to help firms that contribute greatly to local GDP. Besides, firms sometimes voluntarily contribute financial resources to local social projects. In exchange, better policy treatments or greater accesses to resource are offered.
The rights exchanged
Firms are not buying policy, but access to resources, information, lighter regulatory burden and immunity from political misappropriation. Government officials are buying things that polish their resumes – GDP growth, social welfare improvement, etc. The exchange is much more relational than contractual (i.e. more incomplete contracts), and there is a greater focus on long-run rather than short-run benefits.
Incomplete information, transaction cost and uncertainty
Despite the many differences between democracies and authoritarian regimes, factors that make political markets in democracies imperfect apply to political markets in authoritarian regimes, too. First, information is imperfect, costly and asymmetric. Second, transaction cost is high. Third, there is uncertainty in the enforcement of “contracts,” even if they are just relational.
Political strategies in authoritarian regimes
Since there is no election or voting, firms in authoritarian regimes cannot gain political access by making campaign contributions or lobbying. Instead, firms develop political ties and goodwill with governments. The political ties approach has been well documented by the literature. For instance, it is common for listed Chinese firms to have government officials on their boards of directors. Firms may also develop goodwill with the government by actively responding to government signals (Luo, Wang and Zhang 2017), participating in socially responsible practices (CSR) and making charitable donations (Zhang et al., 2016; Wang and Qian 2011).
Impacts of Political Market Uncertainty
Since Adam Smith (1776), economists have recognized that trade is limited by the extent of market. Although uncertainty has always been an element of political markets, actors are able to gradually develop informal and formal institutions to deal with it (North 1990). Or they can simply form the right expectations, being aware of the imperfectness of the market. That happens when uncertainty stays at a relatively stable level.
When uncertainty suddenly increases, however, there is no time for new institutions to emerge to counteract its effect. A sudden, unexpected jump in uncertainty is likely to restrict trade and cause the market to shrink.
Consider the effect of a major political reshuffle (i.e. a sudden, grand, and rapid change in the composition of men holding offices) and its possible impact on trade in political markets. A political reshuffle generates uncertainty on the supply side. The uncertainty arising from the situation consists of two parts: the objective part and the subjective part. The objective part, or the Knightian uncertainty (Knight, 1921), is that whether the actors on the supply side of political markets (i.e. the government officials whom the firms make deals with) will be in office or not at a future time point is less knowable to everyone, including the agencies that initiate the reshuffle and the government officials at-stake. The subjective part of the uncertainty, on the other hand, is caused by incomplete information or information asymmetry between the less informed entities and the more informative entities.
Sudden increases in uncertainty lead to substantial consequences in firms’ strategic decisions. From the resource-based view perspective (Barney, 1986, 1991), uncertainty makes it difficult for firms to form correct expectations about the future value of a strategic resource. Corporate political investments how firms obtain strategic political resources (goodwill, trust, access to valuable resources). A sudden jump in political market uncertainty makes the return of the investments less certain.
Strategy is essentially about investing for the future. When faced with uncertainty, firms have to make tradeoffs between acting early and acting later, and between focusing and diversifying (Wernerfelt & Karnani, 1987). If the uncertainty is uncontrollable (i.e. exogenous) and there is little first-mover advantage, waiting is often a more desirable strategy. Alternatively, firms may choose to spread out its resources and mitigate risk through strategy diversification.
This discussion leads to our main hypotheses:
H1. When faced with a sudden increase in uncertainty on the supply-side of political markets, firms postpone reduce political investment.
Our second hypothesis concerns the purity in the purpose of the political investments. While certain political investments (e.g. lobbying, campaign contributions, ties with officials) serve solely political purposes, other political investments such as CSR and corporate philanthropy often serve multiple purposes and affect multiple stakeholders (Werner, 2015). Compared to the investments that are purely aimed to generate political influence, investments that serve multiple purposes are likely to be affected less by sudden changes in conditions of political markets. This leads to our second hypothesis:
H2. The purer the purpose of a political investment in serving political purposes, the more affected is the investment by a sudden increase in political market uncertainty.
The empirical context of this study is an authoritarian regime that has recently been through a major political reshuffle—China. In December 2012, a massive anticorruption campaign was initiated in China, affecting tens of thousands of government officials in the following years and changing the political landscape of the country dramatically. We utilize this exogenous shock to identify the effects of increased political uncertainty on corporate political investments.
It has been a challenge to study political investments in China, since there is no public information on how much firms spend on corporate political activities. In addition, most ways of making political investments in China are fuzzy and ambiguous. There is no clear-cut measure. Research on corporate political strategy in China often study observable outcomes such as government official board membership (e.g. Haveman et al. 2017) or use survey data to study more refined questions. Survey data have great advantage and can measure unobservable variables (e.g. Ang and Jia, 2014). However, surveys are subject to the limitation in data structure: most survey data are cross-sectional data. It is costly to conduct repeated surveys through multiple years to generate panel data.
Change is the theme of this study. Panel data for multiple years are needed. Lacking optimal measurements and data sources, we use an unconventional proxy for corporate political investments: corporate charitable donations.
Corporate Philanthropy in China
Research on corporate philanthropy suggest that there are strong political motivations behind corporate charitable donations in China (Zhang et al. 2016; Wang and Qian, 2011; Dickson, 2003; Jia et al., 2018). Although over the past forty years the country has achieved impressive marketization progress and economic development, the political structure remained the same. The government remains the most influential player in the economy and a powerful stakeholder of firms. It not only allocates critical resources (e.g. land, loans, infrastructure, tax incentives, subsidies) but also decides how the games are played (government agencies decide the allocation of licenses, permissions, regulatory burdens). The prosperity of the market economy is accompanied by the by-products of increased income inequality, which threatens social stability. As a result, the government seeks to maintain tight social control by co-opting with business leaders. For instance, the government requires businesses to conduct socially responsible practices (Marquis and Qian, 2014). The government also enacts successful business leaders to the provincial or national People’s Congress (PC) or to the Chinese People’s Political Consultative Conference (CPPCC). The political appointments creates channels for business leaders to “negotiate” terms with the government. In exchange, the business leaders provide financial resources to social welfare projects and information to governments. Corporate philanthropic contributions are especially appreciated when local governments have limited resources of their own (Dickson, 2003). Business contributions to social welfare projects help alleviate governments’ financial burdens.
Another reason why corporate philanthropy in China is especially political is that almost all charities in China are organized by the government (Zhang et al., 2016). The government also closely control the nonprofit sector and it is almost impossible for private charities to operate without government backing.
Studies find firms with greater need for government resources make greater amounts of charitable donations (Zhang et al., 2016) and benefit more from corporate philanthropy (Wang and Qian 2011). Firms that are not state-owned benefit more from corporate philanthropy (Jia, Shi and Wang, 2018), suggesting that the nature of charitable donations are primarily political investments.
Note that corporate charitable donations in China are relationship-specific investments. Different government leaders have different policy agendas and resource demands. In terms of social welfare projects, different leaders often have different preferences for which projects to pursue. Firms’ political strategies are tailored to suit the specific leaders they try to access. For example, some leaders aspire to reduce poverty, others aspire to reduce environmental pollution. Since the outcomes of exchange in authoritarian regimes are relationship based and person-specific, how firms donate is influenced by who are in office.
The 2012 Anticorruption Campaign
The 2012 Chinese anticorruption campaign provides a unique laboratory for this study. It creates a sudden exogenous shock to the supply side of political markets, enabling us to observe the impacts of increased uncertainty on corporate political investment.
The campaign was launched soon after President Xi took office on November 15, 2012. It targeted both “tigers and flies” – that is, both high-level “corrupt” officials and low-level “corrupt” officials. Judging from its impacts, the campaign has been the largest organized anticorruption movement in modern China. Several top national leaders were targeted, including former military leaders Xu Caihou and Guo Boxiong and former Politburo Standing Committee (PSC) member Zhou Yongkang. The indication of PSC members broke the unspoken rule of “PSC immunity.” Many other national and provincial high-ranking officials were affected. As of May 2018, over 150 officials provincial and above have been cracked down. Much more (tens of thousands) of lower level officials were indicated. The Central Commission for Discipline Inspection (CCDI) keeps a record of indicted high-level officials on its website. The list of people investigated is still growing when this draft is written (June 2, 2018).
The campaign has generated a political earthquake in several provinces. Notably, Shanxi became the worst “disaster zone” affected. A total number of nine provincial rank officials were removed for corruption. Among the 13 seats on the provincial standing committee, only three remained two years after the initiation of the campaign. Another “disaster area” is Guangdong, where several of the top leaders were dismissed for corruption. In contrast, some provinces experienced relatively moderate impact of the campaign. Some suspect that the campaign could in part be factional fight in nature. Regardless of the nature of the campaign, it has certainty generated an exogenous shock to regional political markets, with some regions being impacted more and some regions being impacted less.
Data and Sample
Our sample consists of all public firms listed on Shanghai and Shenzhen stock exchanges from 2008 to 2017. We obtain most of the firm-level data from China Stock Market and Accounting Research (CSMAR) database, the Chinese Research Data Services (CNRDS) database, company annual reports, and company websites.
Corruption investigation data comes from the official website of the Commission of Discipline Inspection (CDI) (http://www.ccdi.gov.cn). The website provides information on both investigated government officials and investigated managers. Since some studies point out that some CEO investigations are not published on the website, we follow those studies to manually search in CSMAR for CEO dismissal events. After identifying the events, we search the reason why the CEOs were dismissed. If the reason of a CEO dismissal was related to anticorruption, we code the firm as investigated firm.
Regional characteristics are obtained from the National Bureau of Statistics of China (NBSC).
Our main dependent variable is Corporate charitable donations. We log-transform the amount of charitable donation for cases with positive donations. For cases with no donation, we code the variable as zero.
political market uncertainty (PMU)
Our first main independent variable is political market uncertainty (PMU). How PMU is operationalized is critical to the credibility of the findings of this study. Unfortunately, there is no direct measure for uncertainty. The best we can do is to use observable proxies to observe uncertainty indirectly. This practice is consistent with prior studies in other contexts. For example, in the biotechnology R&D context, researchers often use the success rate of new drug invention in the drug category as a proxy for R&D uncertainty. Researchers may also use the novelty of an invention as a proxy for its marketization uncertainty. The measure of uncertainty largely depends on the context. In our context, we use the number of convicted government officials in each province as a proxy for political market uncertainty. Intuitively, the greater the number of officials who are affected by the anticorruption campaign, the greater political turbulence the campaign has induced in the province. Consequently, we conjecture that the political market uncertainty level of provinces with a large number of convicted officials is higher than that of provinces with a small number of convicted officials.
We consider three candidates for the count approach. The first is the total number of convicted government officials. While this measure captures the scope of the political turbulence, it does not capture its depth. Some investigated government official are top national/provincial leaders in the Party. Some are just officials of bottom-level administrative units (counties and towns). The uncertainty associated with the crackdown of a top provincial official (e.g. governor or mayor) is certainty different from that associated with the crackdown of an administer in a town. Therefore, we introduce a second measure – number of convicted high-level officials. We consider the convicted official’s rank to be high if his rank is provincial-level or above. The CDI has clear classification of the rank of the convicted officials. We will just use information from the CDI. Lastly, we consider integrating the first and the second measure to create an integrated measure- the average rank of convicted officials. To calculate average rank, we code the rank of central government officials and the “first-hands” of provincial and municipal government officials as 5, other provincial and municipal government officials as 3, and the rest as 1. Translating the measures into PMU, we indicate PMU with “high,” “moderate” or “low” dummies based on the quantiles (top 25% – high; bottom 25% – low; 25% – 75% – moderate) the values are in.
Using the count/average rank of convicted officials as proxies for political market uncertainty may risk significant measurement problems, however. For one thing, there might be considerable noise in the actual corruption crackdown process. Some important officials who were at high risk of being investigated might have escaped the investigation, while some with minor problems were investigated for reasons such as they were easily “exposed.” The measurement error arising from the noise in the anticorruption process creates attenuation bias. That is, it makes it less likely for us to observe an actual effect or the observed effect to be smaller than its actual magnitude. The second measurement problem arises from the inaccuracy of the information from the CDI website. It is possible that not all of the investigated officials were revealed to the public. That, again, creates attenuation bias to our results.
Nevertheless, the attenuation bias arising from measurement problems makes us less likely to make the Type I error (false positive). If we actually observe a significant effect, the measurement problem only makes the true effect greater than what we estimated (assume there is no endogeneity problem, which we address with the choice of control variables and estimators).
Political investment purity (PIP)
Our second hypothesis concerns the moderating effect of political investment purity. We use achieved political affiliations (APA) as a proxy for this variable. “Achieved political connections indicate more of an ongoing exchange relationship between business elites (and their organizations) and the government” (Zhang et al., 2016:1311). The politically-motivated proportion of charitable donations of firms with achieved political affiliations is likely to be higher than firms without such affiliations. Following Zhang et al. (2016), we code this variable as one if the chairman of the firm serves as a delegate to the national- or provincial-level People’s Congress (PC) or the national- or provincial-level Chinese People’s Political Consultative Conference (CPPCC) and has no other government affiliations.
We include control variables to account for possible selection into treatment. Variables that affect both the treatment status and the outcome can interfere our causal inference. The treatments in this study are (1) being “assigned” high/low PMU and (2) had achieved political affiliations. The outcome is the change in corporate charitable donations. We include region-level and firm-level variables that we suspect to affect both treatment status and outcome. We also include some variables that do not interfere causal inference to soak up residual variance.
Regional characteristics (GDP per capita, number of NGOs, marketization level, population). One potential problem associated with using variations in PMU across regions to identify the effect is selection bias. For instance, provinces with higher PMU might have lower GDP per capita. If GDP per capita was positively related to corporate philanthropy growth, we would overestimate the magnitude of the coefficient on PMU, or even have a false positive inference. To avoid the problem, we control for the GDP per capita. In a similar vein, our estimation would be biased if provinces with higher PMU were characterized with underdeveloped philanthropic institutions, which led to less corporate philanthropic donations. We deal with the problem by controlling for the number of NGOs in each province. Aside from GDP per capita and NGO development, the marketization level of a province may also have a confounding effect on corporate philanthropy. In general, less marketized regions tend to have higher levels of corruption, therefore may have higher PMU in the event of an anticorruption campaign. We therefore control for marketization level using the widely-used (e.g. Chang and Wu, 2013; Jia, Huang and Zhang, 2018; Wang & Qian, 2011) marketization index developed by the National Economic Research Institute (NERI) (Fan, Wang and Zhu. 2011) as a proxy. We also control for population. For simplicity, we follow prior research (Fisman and Miguel, 2007) and treat all time-varying control variables as if they were constant. We use the 2012 values to proxy for their actual values throughout the observation period.
Several firm-level control variables are worthy of our special attention.The first is Investigated firm. The anticorruption campaign not only shocked the public realm, but also reached into the corporate world. Some corporate leaders were investigated and convicted to be corrupt. Griffin, Liu and Shu (2018) identified 150 investigated and convicted CEOs of listed firms. They find that both corrupt practices and political connection affect the likelihood of a firm being investigated. Their R-squared indicates that corrupt practices and political connections explain similar proportions (14% and 19% respectively) of the variation in the likelihood of a firm being investigated. We code the variable as “1” if any of the firm’s top leaders was investigated.
Government official board member. Another important firm-level control variable is whether the firm had government officials on the board of directors. In October, 2013, the CPC passed “Rule 18” which forbids government officials to serve as board members of listed firms. Over 800 government-official board directors resigned after the passing of the policy, affecting 30 percent of listed firms (Hope, Yue & Zhong, 2017). The coerced departure of these government official board members is likely to have impacted corporate philanthropy. Prior studies find that firms with government officials as board members are more likely to engage in reciprocal exchanges with the government and are more likely to make charitable donations. The coerced leaving of government official board members may therefore have a negative impact on corporate philanthropy. We code the variable as “1” if the firm had any government official board members in 2012.
Other firm characteristics (size, ROA, slack resource, industry, central government affiliation). Other firm characteristics may also affect both the assignment of PMU, achieved political affiliation and the change in corporate philanthropic donations. Political investment is more important to larger and more visible firms. We control for firm size and measure it as the logarithm of revenue. Firm profitability and slack resources may affect the firm’s ability to invest in politics. We measure profitability as return on assets (ROA) and slack resources as free cashflow over total assets (Wang and Qian 2011). The industry of the firm may also affect the firm’s tendency to engage in philanthropy. Some industries are geographically agglomerated, which in turn affects the assignment of treatments. To soak up residual variations, we control for central government affiliation, as firms affiliated with the central government might not be affected as much by the local political market as other firms.
Table 1. Summary of Variables, Measures and Data Sources
|corporate charitable donations||log(donations) if donations != 0; otherwise 0||CNRDS|
|political market uncertainty (PMU)||No. of convicted government officials||CDI website|
|No. of convicted high-level government officials||CDI website|
|average rank of convicted government officials||CDI website|
|political investment purity (PIP)||achieved political affiliation – 1 if the chairperson of the firm is a delegate to the national or provincial PC or CPPCC and has no other political affiliation; otherwise 0||CSMAR
|GDP per capita (province)||GDP per capita||NBSC|
|number of NGOs (province)||No. of NGOs||CNRDS|
|marketization level (province)||NERI marketization index||NERI|
|ROA||return on assets||CSMAR|
|slack resource||free cashflow/ assets||CSMAR|
|central government affiliation||1 if is affiliated with the central government||CSMAR|
We adopt a difference-in-difference (DD) method as our main identification strategy. Technically, if there is no other unobserved factor that leads to abnormal changes in corporate charitable donations during our observation window, we can use an OLS model to identify the effect of PMU (controlling for natural time trend). Nevertheless, the assumption that no other unobserved factors have affected changes in corporate philanthropic donations during the ten-year observation period can be too strong, given the pace and scope of change in China. The DD method allows us to filter the effect of other unobserved changes. We also considered using a regression discontinuity (RD) model. The problem with using RD is that treatment was assigned gradually over time. There is no sharp cut-off point. We conclude that the best statistical approach for our study is DD.
In the DD model, treatment PMU is a continuous variable. That is, observations (the firms) receive different dosages of treatment. We handle time information in two ways. One way is to keep the information of each year. The other is to collapse time into two periods: before 2012 (including 2012) and after 2012. The reason why we do not use firm-year as the unit of observation only is that there is likely to be strong cross-time spillover. The treatment dosage assigned to a particular province in a specific year may not reflect the actual PMU perceived by firms, because the investigations take time and government officials at risk are likely to start taking measures since the onset of the campaign. Such information may spillover to firms. Consequently, the number or the average rank of convicted officials in a year may not indicate the actual PMU in that year. Collapsing time information into “before” and “after” alleviates the problem of cross-time spillover.
Now let’s turn to another potential spillover problem: cross-region spillover. It could be the case that high-risk government officials started taking measures to avoid investigation once they observed the conviction of government officials in other provinces. Again this information could be obtained by local firms. This cross-region spillover would make the observed treatment dosage to be smaller than what they should be, creating another attenuation bias. If the coefficient on PMU turned out to be significant, we could be optimistic that the actual effect would be even greater or more significant.
Standard errors. Bertrand, Duflo and Mullainathan (2002) point out that incorrect ways of dealing with standard errors in DD models often lead to false positive inferences and overestimations. It is particularly troubling when the treatment is serially correlated over time (in the worst case, it is “once on, always on”). Collapsing time information into two periods helps us avoid the problem. For the regression using annual data, we cluster standard errors by region, hence allowing for serial correlation in treatments over time. Problems with standard errors can also arise from assigning group-level treatment to individual observations (Moulton, 1986). In our model, treatment PMU is a group-level variable – firms in the same province share the same PMU. The standard error would be biased if we treat PMU as if it was randomly assigned to firms within the province. To get the standard error right, we cluster them at the province level.
Our first model takes the following form:
Djrt= αr+ γt + φ∙Xjrt+ β1∙PMUrt +β2∙PIPjt+β3∙PMUrt∙PIPjt+ εjrt
represents firm j’s philanthropic donations to province r in year t
is the province fixed effect
γtis the year fixed effect
are time-varying firm- and province-level controls
The coefficients of interest are
β3. β1is the difference-in-difference estimate of the effect of political market uncertainty on corporate political investment.
β3is the moderating effect of achieved political ties on the main effect.
In another model, we include firm-fixed effects (
Djrt= αr+ γt +ωj+ φ∙Xjrt+ β1∙PMUrt +β2∙PIPjt+β3∙PMUrt∙PIPjt+ εjrt
Alternative measures of uncertainty
As we mentioned earlier, one major challenge of this study is measuring PMU with something we can observe. We discussed earlier the potential bias of using count information as a proxy for PMU (not all high-risk governmental officials were convicted). In the robustness check section, we try an alternative measure of PMU: first investigated provinces. Since the onset of the anticorruption campaign, the Central Leading Group for Inspection Work (the anticorruption execution arm of the Central Committee of the Communist Party of China) has dispatched inspection teams to provinces, central government agencies and state-owned enterprises. The inspections happened in several rounds: some provinces were investigated in the earlier rounds. If early arrival of an inspection team signals the region being a main target of the campaign, the first-investigated regions are likely to have high PMU. We therefore use early inspection as an alternative measure for PMU and code this alternative measure as “1” if the province was investigated during the first three rounds and “0” otherwise.
Alternative starting time and observation windows
Next, we consider alternative starting time and observation window. In the main model, the cutoff time point we use is 2012, at the end of which the anticorruption campaign was officially announced. However, it is possible that firms and government officials anticipated the campaign in 2011, when President Xi was elected. It is also possible that firms and government officials didn’t take the announcement seriously when the campaign first started, and only came to realize its seriousness after a year. Therefore, we try 2011 and 2013 as alternative cutoff time points.
Additionally, our main model has an observation period of ten years (five years before and five years after). The 10-year observation period is relatively long, if the actual influence of the campaign may have attenuated after a few years, or the uncertainty perceived by firms was partially resolved over the first few years. We therefore try alternative time windows of eight years, six years and four years for robustness check.
The too-many-control-variables problem
Finally, we address the potential problem of having too many control variables. As Clarke (2005) shows, the omitted variable bias may or may not be mitigated by including a subset of relevant control variables. Including more control variables in the model may increase or reduce the omitted variable bias. Models with larger specifications are not necessarily less biased than models with smaller specifications. In response, we include models of different sizes of control variables (from zero to full) in our analysis.
Ang, Y. Y., & Jia, N. (2014). Perverse Complementarity: Political Connections and the Use of Courts among Private Firms in China. The Journal of Politics, 76(2), 318–332. https://doi.org/10.1017/S0022381613001400
Barney, J. (1991). Firm Resources and Sustained Competitive Advantage. Journal of Management, 17(1), 99–120. https://doi.org/10.1177/014920639101700108
Barney, J. B. (1986). Strategic Factor Markets: Expectations, Luck, and Business Strategy. Management Science, 32(10), 1231–1241. https://doi.org/10.1287/mnsc.32.10.1231
Baron David P. (2004). Private Politics, Corporate Social Responsibility, and Integrated Strategy. Journal of Economics & Management Strategy, 10(1), 7–45. https://doi.org/10.1111/j.1430-9134.2001.00007.x
Bertrand, M., Bombardini, M., Fisman, R. J., & Trebbi, F. (2018). Tax-Exempt Lobbying: Corporate Philanthropy as a Tool for Political Influence (SSRN Scholarly Paper No. ID 3095686). Rochester, NY: Social Science Research Network. Retrieved from https://papers.ssrn.com/abstract=3095686
Bertrand, M., Duflo, E., & Mullainathan, S. (2002). How Much Should We Trust Differences-in-Differences Estimates? (Working Paper No. 8841). National Bureau of Economic Research. https://doi.org/10.3386/w8841
Bonardi, J.-P., Hillman, A. J., & Keim, G. D. (2005). The Attractiveness of Political Markets: Implications for Firm Strategy. The Academy of Management Review, 30(2), 397–413. https://doi.org/10.2307/20159126
Bonardi, J.-P., Holburn, G. L. F., & Bergh, R. G. V. (2006). Nonmarket Strategy Performance: Evidence from U.S. Electric Utilities. Academy of Management Journal, 49(6), 1209–1228. https://doi.org/10.5465/AMJ.2006.23478676
Buchanan, J., & Tullock, G. (1999). The Calculus of Consent (Volume 3 ed. edition). Indianapolis: Liberty Fund, Inc.
Chang Sea‐Jin, & Wu Brian. (2013). Institutional barriers and industry dynamics. Strategic Management Journal, 35(8), 1103–1123. https://doi.org/10.1002/smj.2152
Clarke, K. A. (2005). The Phantom Menace: Omitted Variable Bias in Econometric Research. Conflict Management and Peace Science, 22(4), 341–352. https://doi.org/10.1080/07388940500339183
Dickson, B. J. (2003). Red Capitalists in China: The Party, Private Entrepreneurs, and Prospects for Political Change. Cambridge, UK ; New York: Cambridge University Press.
Fan, G., Wang, X., & Zhu, H. 2011. NERI Index of Marketization of China’s Provinces 2011
Report. Beijing, China: Economics Science Press.
Fisman, R., & Miguel, E. (2007). Corruption, Norms, and Legal Enforcement: Evidence from Diplomatic Parking Tickets. Journal of Political Economy, 115(6), 1020–1048. https://doi.org/10.1086/527495
Griffin, J. M., Liu, C., & Shu, T. (2018). Is the Chinese Corporate Anti-Corruption Campaign Authentic? (SSRN Scholarly Paper No. ID 2779429). Rochester, NY: Social Science Research Network. Retrieved from https://papers.ssrn.com/abstract=2779429
Haveman, H. A., Jia, N., Shi, J., & Wang, Y. (2017). The Dynamics of Political Embeddedness in China. Administrative Science Quarterly, 62(1), 67–104. https://doi.org/10.1177/0001839216657311
Hillman, A., & Keim, G. (1995). International Variation in the Business-Government Interface: Institutional and Organizational Considerations. The Academy of Management Review, 20(1), 193–214. https://doi.org/10.2307/258892
Holburn Guy L. F., & Vanden Bergh Richard G. (2013). Integrated market and nonmarket strategies: Political campaign contributions around merger and acquisition events in the energy sector. Strategic Management Journal, 35(3), 450–460. https://doi.org/10.1002/smj.2096
Hope, O.-K., Yue, H., & Zhong, Q. (2017). China’s Anti-Corruption Campaign and Firm-Level Transparency (SSRN Scholarly Paper No. ID 2899403). Rochester, NY: Social Science Research Network. Retrieved from https://papers.ssrn.com/abstract=2899403
Jia, N., Huang, K. G.-L., & Zhang, C. M. (2018). Public Governance, Corporate Governance, and Firm Innovation: An Examination of State-Owned Enterprises. Academy of Management Journal. https://doi.org/10.5465/amj.2016.0543
Jia, N., Shi, J., & Wang, Y. (2018). The Interdependence of Public and Private Stakeholder Influence: A Study of Political Patronage and Corporate Philanthropy in China (SSRN Scholarly Paper No. ID 3092504). Rochester, NY: Social Science Research Network. Retrieved from https://papers.ssrn.com/abstract=3092504
Kingsley, A. F., Vanden Bergh, R. G., & Bonardi, J.-P. (2012). Political Markets and Regulatory Uncertainty: Insights and Implications for Integrated Strategy. Academy of Management Perspectives, 26(3), 52–67. https://doi.org/10.5465/amp.2012.0042
Knight, F. H. (2014). Risk Uncertainty and Profit. Martino Fine Books.
Li Jing, Xia Jun, & Zajac Edward J. (2017). On the duality of political and economic stakeholder influence on firm innovation performance: Theory and evidence from Chinese firms. Strategic Management Journal, 39(1), 193–216. https://doi.org/10.1002/smj.2697
Luo, X. R., Wang, D., & Zhang, J. (2017). Whose Call to Answer: Institutional Complexity and Firms’ CSR Reporting. Academy of Management Journal, 60(1), 321–344. https://doi.org/10.5465/amj.2014.0847
Luo, X. R., Zhang, J., & Marquis, C. (2016). Mobilization in the Internet Age: Internet Activism and Corporate Response. Academy of Management Journal, 59(6), 2045–2068. https://doi.org/10.5465/amj.2015.0693
Marquis, C., & Qian, C. (2013). Corporate Social Responsibility Reporting in China: Symbol or Substance? Organization Science, 25(1), 127–148. https://doi.org/10.1287/orsc.2013.0837
Moulton, B. R. (1986). Random group effects and the precision of regression estimates. Journal of Econometrics, 32(3), 385–397.
North, D. C. (1990a). A Transaction Cost Theory of Politics. Journal of Theoretical Politics, 2(4), 355–367. https://doi.org/10.1177/0951692890002004001
North, D. C. (1990b). Institutions, institutional change, and economic performance. Cambridge; New York: Cambridge University Press.
Paik, Y., Kang, S., & Seamans, R. (2018). Entrepreneurship, Innovation, and Political Competition: How the Public Sector Helps the Sharing Economy Create Value (SSRN Scholarly Paper No. ID 2925077). Rochester, NY: Social Science Research Network. Retrieved from https://papers.ssrn.com/abstract=2925077
Peng, M. W. (2003). Institutional Transitions and Strategic Choices. Academy of Management Review, 28(2), 275–296. https://doi.org/10.5465/AMR.2003.9416341
Pfeffer, J., & Salancik, G. R. (1978). The external control of organizations: a resource dependence perspective. New York: Harper & Row.
Simon, H. A. (1947). Administrative behavior; a study of decision-making processes in administrative organization. Oxford, England: Macmillan.
Smith, A. (177). The Wealth of Nations. London: Methuen & Co., Ltd.
Wang, H., & Qian, C. (2011). Corporate Philanthropy and Corporate Financial Performance: The Roles of Stakeholder Response and Political Access. Academy of Management Journal, 54(6), 1159–1181. https://doi.org/10.5465/amj.2009.0548
Werner, T. (2015). Gaining Access by Doing Good: The Effect of Sociopolitical Reputation on Firm Participation in Public Policy Making. Management Science, 61(8), 1989–2011. https://doi.org/10.1287/mnsc.2014.2092
Wernerfelt, B., & Karnani, A. (1987). Competitive strategy under uncertainty. Strategic Management Journal, 8(2), 187–194. https://doi.org/10.1002/smj.4250080209
Zhang, J., Marquis, C., & Qiao, K. (2016). Do Political Connections Buffer Firms from or Bind Firms to the Government? A Study of Corporate Charitable Donations of Chinese Firms. Organization Science, 27(5), 1307–1324. https://doi.org/10.1287/orsc.2016.1084