Firms as Revenue Safety Nets: Political Connections and Returns to the Chinese State

Abstract The political connection between the state and firms in the context of China's corporate restructuring has been little explored. Using the clientelist framework and unpacking the incentives of both firms and the state, we analyse political connections as repeated patron–client exchanges where the politically connected firms can help the state fulfil its revenue imperative, serving as a failsafe for local authorities to ensure that upper-level tax quotas are met. Leveraging original surveys of the same Chinese firms over an 11-year period and the variations in their post-restructuring board composition, we find that restructured state-owned enterprises (SOEs) with political connections pay more tax than their assessed amount, independent of profits, in exchange for more preferential access to key inputs and policy opportunities controlled by the state. Examining taxes rather than profits also offers a new interpretation for why China continues to favour its remaining SOEs even when they are less profitable.

most SOEs remain unprofitable. Such problems beg the question of why the state keeps giving these firms special treatment. Some earlier studies suggest that the state relies on SOEs to keep layoffs to a minimum to maintain social stability. While that may be the case, as the social security system becomes more developed and institutionalized over time, the argument seems inadequate. A question that needs asking is what economic returns does the state get from allowing selected firms to use political connections to get ahead?
Using new research drawn from China's corporate restructuring, we argue that the relationship between the state and its politically connected firms is a complex and nuanced patron-client reciprocal exchange, where political connections yield benefit but also entail costs for firms, including helping the state meet its revenue imperative. Existing research, including that on China, already shows that firms with political connections can maintain a patron-client relationship by providing individual officials rents, i.e. bribes, in exchange for preferential access to limited opportunities and resources. 4 We do not dispute that possibility; instead, we highlight an unexplored side of firm-state relations that involves extra tax payments to local government. The key difference is that the extra payment is going not into private pockets but into local coffers to meet the local tax collection quotas, which have more diffuse benefits for the populace at large. This type of payment is, therefore, distinct from bribes and other forms of corruption that directly financially benefit individual officials.
Our empirical evidence leverages a unique set of data drawn from original surveys conducted in China covering a continuous 11-year period (1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004). The scope and depth of our data provide three advantages. First, rather than using official or off-the-shelf datasets, we designed and administered the surveys ourselves, which allows us to examine restructuring from several new angles and construct variables that are usually unavailable. Second, the longitudinal nature of the surveys allows us to exploit variation within the same firm over time, thus accounting for unobserved firm-specific factors that confound cross-sectional analyses. Finally, our research team in China was able to ensure data integrity.
Taking advantage of the variation in political connections and the different timing of restructuring in the data in a two-way fixed-effects model, we show that restructured firms with political connections, measured as current or former government officials on the firm board, are more likely to pay more taxes, even when the profits of these firms remain flat. These findings support our argument that the "helping hand" relationship is part of a reciprocal exchange between the state and its politically connected firms, rather than just a one-sided provision of goods or opportunities to the firms by the state. Tax payments, especially overpayment above the amount set by the statutory rate, are the premiums these restructured SOEs pay to maintain their clientelistic relationship, thereby ensuring continued preferential access to benefits from the government, including loans from state banks and help with debt financing.
Our study is the first to empirically document the cost of political connection and the benefit for the state and contributes to a fuller understanding of the meaning of political connection between firms and the state in China. By showing how interests are aligned between firms and local authorities, we also provide an answer for why China, unlike Russia, has been able to collect more taxes from restructured firms and why politically connected firms are more likely to pay even more. Finally, our findings that the state pursues its interests through tax collection rather than increased firm profits offers an explanation as to why the Chinese leadership continues to embrace state-owned or state-controlled firms as a necessary and crucial part of the economy, despite their flagging profits. 5

Political Connection as a Patron-Client Tie
The use of patron-client ties by state agents has been documented in reform and pre-reform China. During the Mao period, agents of the state used their official positions to control access and allocate publicly owned goods as a basis of power. 6 In return, whether in the rural or urban areas, patron-client ties allowed state agents to discharge their official duties more easily and effectively, often acting as an additional set of eyes and ears to help fulfil policy goals and targets. In the early reform period, private firms used what David Wank calls symbiotic clientelism in littoral localities like Xiamen to do business in a context of insecure property rights, providing gifts to local officials in exchange for protection. 7 More recent literature has expanded the scope of patron-client ties to show how they resolve principal-agent problems in political hierarchies. Instead of seeing patron-client ties as synonymous to corruption and an impediment to governance, for example, Junyang Jiang highlights the enabling function of patron-client relations. 8 In the post-Mao period, local officials try to increase GDP to impress superiors and to score higher on the annual cadre evaluations. 9 Nonetheless, regardless of whether GDP increases, local officials must meet the annual tax quotas set by the upper levels, the fulfilment of which signals loyalty and competence to their principals. 10 Little attention has been paid to the fact that local authorities are dependent on firms to meet the tax quotas. When a locality's tax collection quotas are in danger of going unmet, politically connected firms have an opportunity to strengthen their clientelistic ties to local government by complying with requests to pay more taxes than the assessed amount.
One might question whether the relationship between firms and the state should be considered clientelistic because resources and opportunities are 5 Lardy 2019. 6 In the countryside within communes, see Oi 1985;1989;in factories, see Walder 1983;1986. 7 Wank 1999. 8 Jiang 2018. 9 Edin 2003Whiting 2004;Chen, Ye, Li andZhou 2005. 10 Lü andLandry 2014. controlled by different offices. Moreover, local officials are frequently rotated out of localities. If politically connected firms are engaged in a clientelistic relationship, who is their patron? While a firm may see one official as its patron, in practice the firm will work to ensure that the entire local state sees it as a client. Local government is a complex and multi-agent operation, but as argued in the idea of local state corporatism, the locality as a whole, through all of its different bureaus and agencies, works together to pursue economic development. To this end, top leaders, who are the key decision makers, have the ability to coordinate the different bureaucracies to help specific firms more than others. 11 Thus, we can conceptualize the entire government of the locality as the patron, with different bureaus providing different types of goods.
Firms also know that local officials, at the top of local government as well as within bureaus, get transferred. Thus, there is added uncertainty. However, personnel changes do not necessarily make paying additional tax ineffective and costly. For politically connected firms, there is continuing certainty that agreeing to pay higher taxes when needed will still be effective regardless of personnel changes. It matters not whether there is a new official, he or she will likely run into periods of economic difficulty and need help meeting the tax collection quotas. This is one more reason why firms continue to "help" the state, especially when new officials take office. Such payment is a reaffirmation by a client to the local state. The relationship that these politically connected firms have with the state is more than simply having "personal ties" (guanxi 关系) with a single individual. Rather, it is a long-term relationship between a firm and the local state based on trust as well as exchange, where payback is not predetermined in time or amount. 12 As we shall detail below, a key breakthrough that led us to this line of inquiry was discovering the little-known but often-used practice of "tax assignment" ( paishui 派税), which allows local governments to collect more or less tax revenue as needed, irrespective of the assessed amount, through negotiation with their firms. While earlier studies have assumed that China's clear and uniform tax codes and rates determine how much tax firms are assessed, interviews with local officials and firms reveal that, in practice, this tax assignment system can govern how much a firm will pay each year.
To preface our more detailed discussion below, we want to underscore that like any patron-client relationship, there is no precise or immediate accounting in the exchange. Instead, like a repeated game, the goal is to keep the exchange going so that both sides can continue to benefit. Things received and given may not be of equal worth, but what is given is needed and thus valued by the recipient, whether it be the patron or the client. In the case of firms and the local state, the loans received by the firm are not equal to the extra taxes local authorities request the firm to contribute when it calls in its chits. What is key is the timing and access, whether it is a loan for the firm or extra tax revenues for the local state, that allows for the continued success of each party to the relationship. Furthermore, because this is a long-term, repeated game of exchange, extra payment by a firm one year is likely to mean paying less in another, if overall circumstances allow. The amount of extra tax a firm will pay varies based on how much the local authority is short, how many other firms can be called upon to pay extra taxes, as well as how often a particular firm has already provided a helping hand in the past. There is a ledger, but an informal one based on a series of unequal exchanges, where both the firm and the state are mutually dependent for their future success.
Below, we first explicate why corporate restructuring made political connection consequential for firms. We then detail the political and economic incentives created by the state that make it both desirable and feasible for these firms to use tax payments to maintain a long-term, clientelistic relationship with local authorities who can provide preferential access to valuable goods and opportunities.

Corporate Restructuring Makes Political Connection Consequential
Prior to restructuring, the state extended its "helping hand" to all SOEs. During that period, all SOEs were owned, managed and protected by the state. Thus, the meaning of political connection as it is used in the political science literature yields little leverage in understanding which firms received preferential access: they all did. All that changed, however, starting in the 1990s when the Chinese Communist Party (CCP) instituted corporate restructuring to unburden the state, most importantly the central state, of supporting SOEs.
A key fact that is sometimes overlooked is that restructuring meant that the state was no longer obligated to help any of its former SOEs. Because restructuring ended the political connection enjoyed by SOEs during the Mao and early reform periods, the state's "helping hand" is no longer guaranteed. While increasing marketization of the economy limited the resources and opportunities the state controlled, access to these still allowed politically connected firms to get ahead. It is in this post-restructuring context that "political connection" in China took on the meaning found in the more general literature on political connections.
One way a restructured firm could still be politically connected is through its board of directors. Corporate restructuring in China took a variety of forms, depending on the shareholders of the restructured firm. 13 A key change in the organizational structure of these restructured SOEs was the establishment of a 13 In contrast to many transitional economies where privatization was the most common form of property rights reform (Green and Liu 2005), in China, the goal of corporate restructuring was to diversify ownership, not to privatize per se. An SOE can be restructured into a shareholding corporation, a private company or a foreign joint venture (Oi 2005). In all of these instances, the state has diversified ownership, allowing non-state actors including workers, managers and others to hold majority shares, and sometimes has even given up all state shares in these firms (Oi 2011). board of directors. 14 The 1994 Company Law includes detailed rules about the establishment and function of the board of directors in restructured firms, but nothing regarding its composition. In practice, board members could include the original or new owners and their relatives, officials from the government and various bureaucracies, representatives of other SOEs and non-SOEs, and employee and union representatives. In many instances, the state influenced the composition of the board, particularly in firms where the state still held shares. For example, firms that were able to restructure through a debt equity swap received a board member from the state asset management company that had taken on its bad debt in exchange for equity shares. While some firms had officials imposed on their boards, others sought to attract former and current officials to take seats on boards, 15 akin to the practice in Japan known as amakudari. 16 As is the case in market economies, firms are likely to want to have as many former or current officials on their boards as possible in order to build their networks of political connection. 17 But not every firm is able to do so and, in fact, only a small portion of them can.
The point we want to stress is that China's corporate restructuring allowed the state at each level to decide which firms to help. Firms, even those with political connections, were not assured of preferential access. Firms had to prove that they were worthy of state supportthey had to demonstrate themselves to be earnest and loyal clients. Put differently, the uncertainty of whether they would continue to receive the state's helping hand created incentives for politically connected firms to find ways to impress those responsible for doling out the help, primarily local government officials. 18 As will be detailed below, an effective way for firms to demonstrate loyalty is to pay extra tax to help local authorities achieve their revenue targets.

State Interests in Corporate Restructuring: Taxes versus Profits
Our focus on tax rather than profit is unorthodox. To understand why tax is more important than profit in the Chinese context, we need to unpack state interests in China's SOE reforms. The comparative literature assumes that the firm is the decision maker in corporate restructuring. 19 This assumption, which is based largely on studies of market economies, has shaped the research done on transitional economies, including China. Findings that firms have failed to become more profitable after restructuring have been interpreted as evidence that the 14 As part of the reform process, boards of directors were established even in firms that remained wholly owned by the state. 15 Nevertheless, our interviews suggest that restructured firms also enjoy a lot of latitude in choosing their own board members. 16 Colignon and Usui 2003. Only in the past few years has China put into play rules that prohibit officials from taking board positions within three years from the time they retire from government. 17 Agrawal and Knoeber 2001. 18 A recent study argues that Chinese firms now actively build political connections with both local and central government officials (Wang, Yuhua 2016). 19 Hall and Soskice 2001. reforms have failed. 20 Some attribute this to the half-hearted commitment of the CCP to reform. 21 The fact is that in China restructuring was a state, not a firm, decision. 22 When the state, not the firm, determines when, how and if restructuring will take place, its goals for restructuring may differ greatly from those of the firm. Increasing profits is certainly a valid assumption regarding firms, but can we assume that it is also the state's top priority?
That political actors have different time horizons is a well-accepted concept within political science, yet few studies have explicitly taken this temporal variable into account when assessing outcomes of corporate restructuring in transitional economies like China. Every government wants its firms to be efficient and profitable, but this is not always attainable, especially in the near term. Reform of SOEs is often dependent on other systemic changes outside of firm control that are slow to be funded and institutionalized. 23 Restructuring may eventually have a positive impact on performance. In the mid-2000s, some Chinese SOEs started reporting record-breaking profit increases; however, that scenario was years in the making, and for some that has yet to be the case. For many restructured firms in formerly centrally planned economies, including China, profits remain an elusive goal.
Unlike increasing firm profits, increasing taxes is a feasible goal of SOE reform in the short and medium term, and one that the state can directly control. Beginning from the end of the 1970s, China introduced a series of regulations that changed the assignments of rights and obligations between firms and the state and increasingly de-linked state finance from SOE profit and deficit. 24 The 1984 "tax-for-profit" (ligaishui 利改税) reform was implemented to incentivize firms to be more efficient and profitable, allowing SOEs to pay taxes in lieu of submitting all of their profits. 25 From that point onwards, profits belonged to the firm while taxes belonged to the state. 26 Subsequently, tax revenues began to increase significantly. 27 The state further ensured the primacy of its fiscal interest by creating taxes that were separate from profits. As Figure 1 shows, from 1991 to 2008, the amount of value-added tax (VAT), which is not linked to profit, increased 30.8 per cent per 20 See, e.g., Woo 1999;Sachs and Woo 2001. 21 Steinfeld 1998. 22 Oi 20052011. 23 Oi 2011. 24 Walder 1992 In practice, firms still handed in some of their profits directly to the state. In 1987, with the worsening of fiscal conditions and rampant inflation of the mid-1980s, the government adopted a contracting scheme, whereby firms were given a profit target to pay to the state. That profit target was determined through prior negotiation, which later became problematic owing to information asymmetry between the state and the firms. 26 After the Chinese state introduced a new fiscal system in 1994, firm profits disappeared as a category of state revenues. This change included a provision whereby the central government allowed SOEs to retain all of their post-tax profits. 27 Han (2003) shows that the period between 1984 and 1985 was a turning pointtotal tax revenue jumped from 94.7 billion yuan in 1983 to 204.1 billion yuan in 1985, while the fiscal revenue (firm profits) from SOEs fell from 27.7 billion yuan in 1984 to 4.4 billion yuan in 1985.
year from 1991 to 2000. In comparison, the corporate income tax, which is the tax on profits, increased only 3.5 per cent per year during the same period. Correspondingly, the share of income tax to total fiscal revenue decreased from 29.2 per cent in 1985 to 6.6 per cent in 2000. 28 Thus, one reason why the state could afford to pursue corporate reform policies was that the risks to state fiscal interests were limited and independent of firm profits. We know from the comparative literature that administrative and legislative actions are not sufficient to ensure that a government will be able to collect taxes. Daniel Treisman persuasively argues that Russia's dramatic fall in tax revenues in the 1990s was directly linked to perverse incentives created by the way revenues were shared between central, regional and local levels of government. 29 We accept this line of thought but argue that the exact reverse is also possible. That is, China's authoritarian single-party state created incentives for local officials and firms to pursue their interests through collecting and paying more taxes.
In democratic systems, politicians always have an eye on the next election. While state agents in one-party regimes like China do not have to concern themselves with electoral pressures from below, there are constant political pressures from above in the form of bureaucratic superiors who evaluate their performances and determine their political futures. In line with our argument that profits are seen as a long-term goal, interviews with local officials and factory managers suggest that improving firm profits carries little weight in the annual cadre performance evaluations. In contrast, fulfilling the tax quotas set by the upper level is a key criterion on which annual cadre evaluations are based. 30 It is therefore in local officials' interests to ensure that tax quotas are met.

A Negotiated Tax Payment System
Having examined the institutional incentives that drive local officials to prioritize tax collection, we turn to a practical question: how can politically connected firms impress the state with more tax if taxes are assessed based on the official tax code? The answer lies in the informal practice called tax assignment, which we identified earlier, where local governments collect more or less tax revenue as needed, irrespective of the assessed amount, through negotiation with their firms.
Earlier studies have reported that firms in China can negotiate lower taxes by convincing local authorities to give tax breaks 31 similar to the benefits of political connection found in other countries. 32 Our research reveals comparable cases where firms periodically pay less than their assessed rate of tax, a recognized and well-known practice. Less known is that firms may also pay more than the amount assessed at the statutory rates. Press reports have noted this phenomenon but provide few numbers and scant detail. 33 Our fieldwork sheds light on why, how and when this happens.
Through tax assignment, the local state can collect needed tax revenue above and beyond the assessed amount. This is not a formal policy, which may explain why it has received little attention in the literature. 34 Rather, this is an ad-hoc strategy used by local authorities to meet the tax collection quotas that are set by the upper levels. 35 Whether and how this process is deployed depends on the quota from above and a locality's actual tax collection in a given year, much like the variation in the size of fees charged to private entrepreneurs based on the economic conditions in different localities. 36 In some years, there may be no need to use this practice at all. It is when revenue fails to meet the planned targets that local governments encourage some of their enterprises to contribute more. The system works such that those firms may pay less in the following year or be compensated in other ways. Similarly, in good years, when the growth in real tax revenue exceeds the planned targets, the tax bureaus may ask firms to pay less tax than the actual assessed amount to ensure that the tax quota 30 Guo 2009. 31 See, e.g., Oi 1992. See also Chen, Ling, and Zhang 2021. 32 See, e.g., Lassoued and Attia 2014. 33 In 2002, it was reported that tax bureaus in Inner Mongolia, Jiangxi and Sichuan collected over 1 billion yuan in excess taxes. In the most extreme case, a firm in Deyang, Sichuan, paid 7.63 times more taxes than its assessed tax rate. "Guotoushui yu zhibiao chongbai" (Overtaxation and the cult of quotas does not increase the following year. 37 Local officials have freely admitted to this practice in interviews, careful to stress that how much more or less a firm contributes depends both on the quotas in a given year as well as whether the firm paid more or less in the past. As a result, the actual amount of tax collected from firms often is only loosely aligned with what would be assessed using the statutory tax rates and a firm's performance. Rather, it is adjusted according to the annual official fiscal revenue plan.
The anecdotal evidence for negotiable tax is borne out by our survey data. Figure 2 illustrates the difference between statutory rates and annual effective VAT rates in our surveyed firms. The vertical line is the statutory rate of 17 per cent. To get the effective rates, we divide the VAT by the value added reported by the firms for each year. The results reveal huge variation in the effective VAT rates: the mode is zero and the mean is about 22 per cent. Only about 5 per cent of the effective VAT rates are within plus or minus 1 per cent of the statutory rate. Also worth highlighting is the fact that there are more instances of overpayment than underpayment, suggesting that on average firms

Notes:
The figure plots the effective VAT rates calculated for all firm-year observations in our sample.
37 In one case, a local official spoke about how he insisted that a firm pay less because the locality had already collected more than enough taxes and did not want to exceed the quota too much, lest the quota the following year be too high.
are paying more than the amount assessed based on the tax code, and the difference may be compensated by other benefits. The above calculations indicate that fluctuations in tax payments owing to paishui can occur in all types of firms regardless of their ownership structure after restructuring. Because firms overpay taxes in exchange for help from the state, our argument is that the politically connected firms will more likely come to the rescue of the state. The benefits these firms obtain from the "helping hand" in return, such as loans and policy concessions, may even allow them to consistently pay more than the assessed amount and still manage to grow, which in turn could put their managers ahead when it comes to promotion. This fuller understanding of political connection and how taxes are collected in China leads to two hypotheses. First, politically connected firms after restructuring should turn in more tax even when their profits remain flat. Second, the same group of firms should receive more preferential treatment compared to restructured firms that are not politically connected.

Survey and Data
In assessing the impact of restructuring on business performance, most existing works compare firms with different ownership structures cross-sectionally. 38 A second, better approach is to examine the performance of the same firms over time. This avoids problems of sample selection bias because each firm is compared to itself; however, data for this type of analysis are difficult to come by.
In this study, we follow the second analytical approach with longitudinal data on Chinese enterprises from surveys we designed and implemented in collaboration with the Institute of Economics in the Chinese Academy of Social Sciences. Our data allow us to compare the impact of restructuring on the same firms covering 11 continuous years (1994 to 2004). Moreover, because ours is a specially designed survey, we can examine restructuring from several new angles, using variables that are not usually available. 39 Finally, our research team in China took extraordinary care to ensure data integrity, a problem common with officially released datasets in China. 40 They painstakingly compared the questionnaires with the firms' records for errors and inconsistencies. Whenever in doubt, the team also went back to check with firm managers and officials in the local bureaus.
Our full dataset contains a panel of 208 firms over 11 years in three cities (Wuxi 无锡, Zhengzhou 郑州 and Jiangmen 江门) and four industrial sectors 38 Frydman et al. 1999;Jefferson and Su 2006;Zhang, Zhang and Zhao 2001. 39 One example is the National Statistical Bureau's annual firm-level industrial survey, which covers all state-owned manufacturing firms and non-state-owned ones with sales over 5 million yuan. Despite this dataset's impressive coverage (over 50 million observations and over 100 variables in the 2011 survey), it has no information on the organizational structure of the firms that would be needed to construct standard measures of political connection. 40 Wallace 2016. (textiles, machinery, electronics and chemicals). 41 These cities are at the typical level of urban development and reform in China. They are not the most economically developed nor do they lag behind the national average. The choice of industrial sectors is based on the consideration that the sectors should be evenly distributed across the country with no obvious geographic concentration. This makes our analysis less susceptible to confounding factors such as economic geography and specific industrial structure. In each city, we first cluster the enterprises based on industry and size. We then draw our sample from the list of enterprises with a sampling interval based on the ratio between our target sample size and the total number of enterprises in each cluster (see the supplemental information for the distribution of industry and firm size in our sample).
In the surveys, we ask each firm to complete three questionnaires. The first is the "enterprise questionnaire," which is usually completed by the management department. Questions include general background information on the firm, such as registration type and basic organizational characteristics. The second is the "manager questionnaire," which is usually completed by the top manager. Questions are mostly associated with property rights restructuring and the manager's personal assessment of the firm. The third is the "statistics questionnaire," which is usually completed by the accounting or statistical department. This questionnaire collects the basic financial information and ownership structure of the firm.
In all the surveys, we ask the firms to provide us with information regarding: (1) whether the firm has experienced restructuring that resulted in significant changes in ownership structure; and (2) when these instances of restructuring took place. Our argument requires restricting the analysis to firms that have experienced restructuring at some point during the 11-year period. Among the 208 enterprises, 45 did not answer either of these questions and 18 firms reported that they experienced restructuring but that the restructuring took place before 1994. We removed these 63 firms from the sample, leaving our final sample with a total of 145 enterprises. 42 Figure 3 illustrates the trend of restructuring for our entire sample over time. 43 In 1994, only six firms were restructured. In the next four years, the number of restructured firms increased steadily and peaked in 1998 when nearly a quarter of the firms underwent restructuring. This momentum broke down in 1999, with only 15 firms being restructured. This level remained for the next few years. By 2003, all firms in our sample had been restructured.
The key independent variable is political connection. In the surveys, we asked each firm to provide detailed information about their board members, which we use to construct a dichotomous measure that distinguishes restructured firms into two groups. Following previous studies, we define a firm to be politically connected if its board of directors includes either former or current government officials. 44 In our sample, 23 of the 141 firms (16.3 per cent) that provided board information are politically connected according to this measure. Looking at the year of restructuring, the state held an average of 18 per cent of the shares in firms with political connection and 15 per cent in firms without political connection. 45 Contrary to common perception that political connection is equivalent to state control, these tabulations confirm insights from our interviews that board members can be assigned by the state, sought out by the firm on their own, or both.
Our primary measures for the outcome variable are tax and profit. We measure tax as the amount of VAT paid by the enterprise. As discussed above, the VAT reflects the government's fiscal revenue obtained from the firms and is not linked to profits. In our sampled firms, VAT ranges from -8.6 million yuan to 145 million yuan. The 113 observations with non-positive values are not coding errors. Rather, they suggest that the firm was either loss-making or purchasing more inputs, such as raw materials or semi-finished products, than it was producing, Authors' survey.
44 See, e.g., Li, Wenjing, and Zhang 2010;Chen, Ye, Li and Zhou 2005. 45 Nor is political connection correlated with how these SOEs were restructured or with their eventual ownership structure. See Li, Xiaojun, and Oi 2018. or both. 46 In other words, these negative amounts are not owed tax and do not result from tax assignment. We measure the total pre-tax profit as the sum of sales revenue (or loss) plus investment, non-operating revenue, and other incomes.

Empirical Strategy and Findings
We examine the effect of restructuring on tax and profits for the two groups of firms in a two-way fixed-effects model: There are two explanatory variables in equation (1). The first one is Restructure, which equals 1 for firms that are restructured and 0 otherwise. The terms λ t and θ i are year and firm fixed effects, respectively. Vector X it−1 includes two input factors (net fixed assets and number of employees) to capture the scale of the firm's business operations.
To examine how the effects of restructuring vary between politically connected firms and non-connected firms, we normally would examine the coefficient γ in equation (1), which is the interaction between the variable PC and Restructure. Because PC is time-invariant and will be soaked into the firm fixed effects θ i , in practice, we split the full sample into two subsamples according to whether or not the firm has political connections and estimate the following equation separately for each subsample: In all of the models, we use one-year lags of the independent variables to account for the potential for reverse causation. In the supporting information, we present additional results from logit and Cox proportional hazard models, which confirm that neither the board composition nor the likelihood of restructuring is correlated with the pre-restructuring VAT and profits of the firm. In a separate paper, we also show that the timing of restructuring is uncorrelated with a battery of pre-restructuring firm-level economic indicators such as sales, debt-to-equity ratio and labour productivity. 47 Figure 4 presents the main findings from a series of models with the two subsamples of firms using equation (2). In all the models, we employ the Driscoll Kraay standard errors with a Newey West adjustment of one order lag length to address heteroskedasticity and contemporaneous, serial and spatial correlation in the data. The full model estimations are available in the supplemental information.
Focusing on tax first, we can see that the post-restructure dummy (β) is statistically significant for VAT in the subsample of firms with political connections. This supports our hypothesis that the amount of VAT payment increased only in politically connected firms after restructuring. 48 The substantive effect is also large. Holding everything else constant, a politically connected firm would pay 1.87 million yuan more in VAT after restructuring, with the average amount of VAT in the sampled firms being about 4 million yuan. In the meantime, restructuring does not seem to improve a firm's profits regardless of political connection. The coefficient estimates of the post-restructure dummy has a positive but insignificant effect on profit in both subsamples. This is consistent with existing work that shows that corporate restructuring fails to make firms more profitable in China.
Do these same firms also enjoy more benefits? The answers to additional questions in our 2005 survey suggest that they do and offer some ideas about what these benefits are. Owing to the large number of non-responses (>50 per cent) to these questions, we use simple mean comparisons to explore whether the politically connected firms receive more benefits. The first benefit is loans from the four state banks in China. About 65 per cent of the firms responded to this question and the answers ranged from zero to 1,063 million yuan. Two firms, both politically connected, took out more loans than their total assets. For comparison, we also asked about commercial bank loans obtained by the firms. The results are telling.

Figure 4: Differential Effects of Restructuring on Tax and Profits for Firms with and without Political Connection
Notes: This figure plots coefficients (β) of VAT and profits (in 1,000 yuan) for politically connected (PC) and non-politically connected (NPC) firms. The other control variables are omitted.
48 In the supporting information, we provide evidence against an alternative explanation that the results may be driven by under-reporting of taxes by firms prior to restructuring.
As shown in Table 1, politically connected firms on average can obtain 57 million yuan more from the state banks. The difference between the two is statistically significant; a two-sample t-test rejects the null hypothesis of equal mean with a p-value less than 0.001. The pattern is reversed for commercial bank loans as firms without political connection on average secured 7.6 million yuan more than politically connected firms, although the difference is not statistically significant. The fact that politically connected firms can receive large cash injections from the state may also explain why they can consistently overpay their VAT, as suggested by our findings. Thus, it is only when we put together the costs as well as the benefits of political connections that we get a full picture of state-firm relations. The second benefit we examine is management fees, which taps into the direct personal benefits enjoyed by the firms' managers. Not surprisingly, politically connected firms paid over twice as much in management fees as firms without political connections.
The third benefit is the freedom to adjust the workforce. Politically connected firms were able to lay off many more workers. On the surface, this seems counterintuitive because the state limits the excessive dismissal of workers for political reasons. One likely answer comes from interviews with local officials and factory managers who have themselves faced this thorny issue. The Chinese state is not opposed to layoffs per se; what the state fears is potential unrest and protests by laid-off workers. If a firm can provide a severance package acceptable to those workers who will be made redundant, then there is no problem. Our findings suggest that these politically connected firms are being provided sufficient resources to buy off more workers who will be laid off.
All these benefits allowed politically connected firms to expand after restructuring even though profits remained flat. This result can be seen in Figure 5, which examines three output measures post-restructuring for the firms with and without political connection using the same model specification as in Figure 4. Here, we can see that politically connected firms generated more sales, value added and industrial output after restructuring. On the other hand, those without political connection failed to see increases in any of the three output measures. In other words, by paying the cost of political connection to maintain their preferential access to the state, these firms have a much better chance to grow and succeed in China's post-restructuring economy. p values are from two-sample t-tests. NPC denotes "non-politically connected;" PC denotes "politically connected."

Conclusion
This paper set out to examine an underexplored side of the relationship between the state and its politically connected firms. In doing so, it also addresses the puzzles of why China has been able to collect more tax from restructured firms and why politically connected firms are more likely to pay more. The answers, we argue, lie in the long-term clientelistic exchange relationship between state agents with revenue imperatives and politically connected firms that have benefited from their preferential access to scarce goods and opportunities. When required, these firms contribute more tax to maintain this exchange relationship, which in turn allows them to develop and get ahead even without increasing firm profits. Our empirical evidence lends support to our hypothesis that tax payments increased significantly from the restructured firms that are politically connected, independent of profits, suggesting that extra taxes are repayment from these firms for the "helping hand" of the state. Our finding that firms' taxes increase despite flat profits also provides a new perspective by which to assess China's corporate restructuring. Failure to increase firm profits need not be taken as evidence of policy failure or the CCP's halfhearted commitment to reform. An alternative explanation for why corporate restructuring policies have not produced increased profits is that it was never the goal of the Chinese party-state in pursuing that reform. While reformers may want their firms to be efficient and profitable, this is not always attainable, especially in the near term. Once the time horizon is considered and it is recognized that state interests come first, it should be no surprise that restructuring has yielded little or no change in firm profits.
Our study contributes to recent works that highlight the "enabling" aspect of political connections, whereby the reciprocal relations both within the government Author's survey Notes: This figure plots the coefficients (β) of sales, value added and industrial output (in 1,000 yuan) for the two types of firms. Full estimation results are available in the supplemental information. and between government and other groups, may perform important public functions. 49 Moreover, it goes a step further to show that the state has an immediate economic rationale for granting preferential access to firms with political connections. While earlier studies have advanced the argument that the state counts on SOEs to implement political tasks such as keeping unneeded workers and minimizing layoffs to maintain political stability, 50 we show that restructured SOEs with political connections also may act as a revenue safety net by paying extra taxes to help meet the all-important revenue imperative.
By revealing the symbiotic relationship between the state and politically connected firms, our study also may offer an economic logic for why it has been so difficult for the state to turn away from some of its SOEs to embrace the private economy more fully. By de-linking key taxes from profit and getting its politically connected firms to pay more than their assessed rate, the state has been able to receive stable and even increased tax revenues, irrespective of firm profits. We would expect such a symbiotic relationship to be even stronger between the state and the remaining SOEs in China today. As a result, the state can afford a go-slow approach, leaving the more difficult task of improving firm profits as a long-term goal. 51 Taken together, our study sheds new light on the most recent effort of Chinese leaders to pivot back to the state sector. In September 2018, during a visit to a state-owned petrochemical company in Liaoning province, President Xi Jinping 习近平 vowed to strengthen China's state-owned enterprises, making them "stronger, better and larger." 52 Why did China go through the complicated process of corporate restructuring only to once again make SOEs the fulcrum of the economy, despite overwhelming evidence that shows SOEs are less efficient at using state resources? 53 Our findings suggest that the state continues to favour its remaining SOEs who maintain strong political connections with the state because even if the "stronger, better, and larger" firms do not yield more profits, the state can still meet its revenue imperative by extracting tax revenues, including advance payment of taxes, irrespective of those profits. Having such a revenue safety net is even more important now with a slowing economy and an uncertain global environment in the aftermath of the global pandemic.