Introduction

Between the 1970s and the 1990s, a highly unexpected set of outcomes took place in the People’s Republic of China. The Chinese Communist Party enacted “market-preserving federalism” by devolving fiscal and regulatory authorities from the central to local governments, thus limiting the interventionist role of the central government (Qian et al. 1995; Qian and Roland 1998; Qian and Weingast 1997). In some regions of China, decentralization allowed for the flourishing of private businesses, which fueled over two decades of spectacular growth in these regions. The Chinese-style “market-preserving federalism” is puzzling because central predatory behavior is expected to be particularly severe in a one-party dictatorship with a strict party hierarchy and a top-down appointment system like the Chinese Communist Party (CCP) (Cai and Treisman 2006: 506). Yet, scholars have uncovered ample evidence that fiscal and administrative decentralization served as a “pillar” for high growth in China from the 1970s to 1990s (e.g., Qian 1999; Xiang 1999; Oi 1992; Xu 2011). In order for decentralization to be a key component of the China growth miracle, it needed to be credible, but how was it possible under a one-party regime?

This piece contributes to the discussion on China’s credible decentralization in several ways. First, we show that the historical foundation of decentralization from the 1970s to the 1990s was not simply top-down decentralization enacted in the late 1970s. Rather, the Cultural Revolution, which began in 1966, eradicated most of the central officials from the political elite, thus paving the way for the enactment and persistence of decentralization policies through the 1980s and into the 1990s. We also show the reversal of this logic as central officials began to make up a larger share of the political elite in the late 1980s and in the 1990s. This paved the way for centralizing tendencies in China’s growth model today. Also, because the Cultural Revolution was a highly unusual political shock for one-party authoritarian regimes, sustained economic decentralization, which provided great impetus for reform in the 1970s and the 1980s, is unlikely to take place again in China. Another path to reform would have to be found.

This piece also contributes to the literature on policy-making in one-party regimes by highlighting the role of elite composition in making policies credible. Political shocks that radically alter the composition of the elite can reorient elite policy preferences, thus sustaining a particular policy direction such as economic decentralization in the medium term. This finding resonates with the second-generation fiscal federalism literature that examines political institutions with an emphasis on political parties (e.g., Filippov et al. 2004; Riker 1964). Finally, a growing literature on authoritarianism provides insights into the function of authoritarian institutions in preserving elite unity, but pays insufficient attention to the genesis of institutions (Pepinsky 2014). In the case of market-preserving federalism in China, we offer a novel and consistent account of its origin, persistence, and retreat, thereby contributing to the discussion on the political origins of authoritarian institutions.

Theoretical Framework

In the traditional literature on fiscal federalism, federalism is realized through proper institutional design, which, once implemented, is assumed to be difficult to change (Oates 1999; Bednar 2003). Yet, scholars argue that the realization of federalism is much more than a problem of institutional design. Riker (1964) and Weingast (2009) note that a well-functioning federalism requires a set of political institutions, especially political parties, which prevents both national level predatory behavior and local level free-riding. Also, integrated party systems are able to sustain federal systems by institutionalizing elite competition (Filippov et al. 2004).

Because much of the vast literature on federalism assumes a democratic setting with a credible constitution, the literature has neglected the other problem identified by Riker—central authority’s predatory behavior to make claims on local revenue or to intervene in local decision-making against federalist arrangements (Garman et al. 2001; Bednar et al. 2001). Without credible constitutions or institutionalized electoral and party systems, the center can roll back the regions’ privileges and thus render federal concessions temporary (Bahry 2005). Among weak democracies and authoritarian regimes in particular, political institutions lack perpetuity such that political leaders can alter the rules at will and renege on the federalist arrangements (Weingast 2014).

The case of China, which had a well-documented period of economic decentralization, seems to defy the theoretical literature on federalism. The extensive literature on Chinese federalism provides ample theoretical insights and evidence on the linkages between economic decentralization and China’s stunning economic performance in the reform era (Qian 1999; Xiang 1999; Oi 1992; Xu 2011). However, it largely left unanswered the questions of how economic decentralization materialized in the first place and why fiscal and economic decentralization was sustained for a relatively long period given the weak force of the constitution and the hierarchical nature of the ruling Chinese Communist Party (CCP).

Our elite composition account seeks to answer these two questions. Our account begins with the key role of nominally representative institutions like the Central Committee in committing dictators to a given set of policies by imposing costs on reneging. According to a robust body of theoretical literature on authoritarian regimes, regimes with unelected representative bodies composed of segments of the elite credibly guarantee some stream of payoffs to the elite (Boix and Svolik 2013; Gandhi 2008; Gehlbach and Keefer 2012). Dictators set up such representative institutions in order to aggregate heterogeneous preferences of supporters, thus reducing the costs of governance, as well as to credibly signal to supporters future streams of payoffs (Boix and Svolik 2013; Gandhi 2008). Because representative institutions like the Central Committee reduce the transaction costs of staging a challenge or even a coup against the dictator, they credibly commit the dictator to the promised stream of payments to the elite (Boix and Svolik 2013; Gehlbach and Keefer 2012). These insights add further force to Shirk (1993) and Bueno de Mesquita et al.’s (2003) observation that dictators, although not elected, still rely on a small selectorate to stay in power.

In this view of authoritarian regimes, the dictator would be wise to pursue policies favored by a winning coalition within the selectorate, especially if she sits in a representative institution like the Central Committee, else faces higher costs of staying in power. This links the composition of the selectorate with policies likely to be pursued by the top leaders. While members of the selectorate have common preferences like more pay and better housing, their varied institutional affiliations within the regime generate heterogeneous preferences. Thus, shifting the composition of the selectorate alters the policy that the dictator would pursue to minimize the cost of staying in power, as long as the policy change does not endanger some other core interest of the dictator.

To be sure, the composition of the selectorate likely evolves endogenously on the margin, affected by various policy outcomes. However, the composition of the selectorate in one-party regimes tends to be sticky because of the high transaction costs of major leadership reshuffling, thus making it an exogenous variable in the short run (Pepinsky 2009; Nathan 1973). In rare instances, the benefits of a major reshuffling outweigh the costs, which introduce exogenous political shocks dramatically altering the composition of the selectorate. These shocks in turn alter the incentive of dictators to pursue one policy over another. Even though institutions and norms remain unchanged, shifting incentives for top leaders can make de facto federalism a reality even in an authoritarian regime (Bednar et al. 2001).

In China’s case, the domination of local officials in the CCP’s Central Committee was a necessary condition of persistent economic decentralization, although top-down command or some other shock put it in motion also. This theoretical argument leads to four deductive implications illustrated in Table 1: (1) if local officials dominated the selectorate, economic decentralization would be implemented and would persist (the left-top cell); (2) if local officials did not dominate the selectorate, economic decentralization would not be implemented or would fail to persist (the right-top cell); (3) if local officials dominated, economic centralization would fail (the left-bottom cell); and (4) if local officials did not dominate, economic centralization would succeed (the right-bottom cell).

Table 1 Deductive implications of our theory
Full size table

The Cultural Revolution in China (1966–1976) dramatically altered the composition of the selectorate by purging most of the senior level central officials from their government and party positions. Mao’s power plays were not motivated by economic goals but gave rise to a Central Committee (CC) dominated by local officials at the 1969 9th Party Congress and the 1973 10th Party Congress. Our data on Central Committee composition reveal that the supremacy of local officials established at the outset of the CR persisted through subsequent party congresses due to relatively lower turnovers in subsequent congresses. When Deng Xiaoping took power in 1978, he sensibly continued decentralization policies to appease the main constituency in the political elite at the time. In contrast, decentralization during the 1958 Great Leap Forward was not accompanied by elite reshuffling, which resulted in reversal soon after. The leadership’s strong preference for decentralization did not change until the 1990s after the central bureaucracy experienced a decade of rebuilding. Throughout these periods, the composition of the CCP elite in the CC, instead of institutional changes, provided an important foundation for top leaders to either pursue centralization or decentralization policies.

To illustrate the generalizability of our findings, we investigate two “shadow cases”—Taiwan under Kuomintang(KMT)’s rule and in the Soviet Union—and show that similar patterns linking party elite composition and policy orientation can also be found. Taiwan’s case suggests that Taiwan’s economic liberalization especially the shift from import substitution industrialization policy to export-led growth strategy in the late 1950s was associated with the dramatic decline of the Executive Yuan representatives (central technocrats)’s political clout in KMT’s Central Committee. In the case of the Soviet Union, although changes in economic policy were correlated with the changing composition of the party elite, episodes of economic decentralization were only temporary in part because elite composition did not experience a fundamental shift. It bears emphasis that the three cases in our study are not compared directly in a cross-case comparison analysis. Our research design gains inferential leverage from process tracing within-case studies (Collier 2011; Mahoney 2012).

Our analysis is organized in chronological order and systematically traces the decision-making process around every historical turning point. We first discuss the reversal of decentralization policies after the Great Leap Forward (1958–1961) due to the dominant clout of central officials in the Central Committee (top right cell of Table 1). The Cultural Revolution, a purge of the party caused by Mao’s growing suspicion of his colleagues, fundamentally altered the elite composition of the Central Committee which rendered economic decentralization policies sustainable (top left cell of Table 1). Because a large share of the followers in these factions, as well as other “capitalist elements” targeted by Mao, were central officials, a large number of high-level central bureaucrats were removed from the party’s upper echelon. This rendered the central bureaucrats a relatively weak political force for the next two decades. After this significant reorientation of elite power dynamic, it made good political sense for Mao to pursue decentralization regardless of his policy preference.

When Deng took power in the late 1970s, provincial dominance in the Central Committee was an established fact that Deng could not have changed immediately without considerable costs. A much less costly and more sensible strategy was to give major concessions to provincial interests in order to win local support for his political struggle against his competitors. Among other things, such concessions took the form of continued fiscal and firm decentralization. Policy outcomes in the 1980s were consistent with the implication in the left-bottom cell of Table 1.

Over the 1980s, both endogenous evolution and exogenous pressure gradually raised the share of central technocrats in the Central Committee. Endogenously, decentralization policies gave rise to an increasing number of economic problems which required a larger and more competent central state to handle (Wedeman 2003; Shih 2008). At the same time, the top leaders’ desire to reduce the role of the military in elite politics and the passing of the veteran revolutionary generation gave rise to a generation of young, educated technocrats who took up leadership positions in a growing number of central organs. Because these positions came with guaranteed admission into the Central Committee, the share of central officials in the CC also rose steadily throughout the second half of the 1980s and the first half of the 1990s.

When the Tiananmen unrests and the collapse of the Soviet Union shocked the top leadership into action, pursuing centralization policies was no longer politically difficult. When the leadership implemented centralization policies such as “grasping the big and letting go of the small” and fiscal centralization in the mid-1990s, they had the support of a powerful coalition of central technocrats and party officials in the Central Committee. This scenario fits neatly into the bottom right cell of Table 1.

An important element of our argument is that large-scale elite reshuffling in the CCP was rare after the establishment of the People’s Republic. Figure 1 displays the incumbency ratio of the Central Committee (Shih et al. 2012). At the 1956 8th Party Congress, nearly 60% of full Central Committee members had been full or alternate CC members at the 7th Party Congress. Mao’s Cultural Revolution radically reshuffled the elite such that by the 1969 9th Party Congress, only 30% of the full CC members had served in the previous CC as full or alternate members. No other leader in China, including Mao himself, was able to replicate this degree of elite reshuffling. For the remainder of the Cultural Revolution, incumbency ratio of full CC members was quite high at nearly 80%. Although the 13th Party Congress saw an overall incumbency ratio of less than 50% among full CC members due the retirement policy, it still was 15% higher than the incumbency ratio at the 9th Party Congress. Into the 1990s, the incumbency ratio of full Central Committee members has hovered around 70%, which means it took 10 years—two party congresses—to turn over the majority of the Central Committee. Thus, after the Cultural Revolution, the change in elite composition became a slow-moving, decade-long process which guaranteed the medium-term stability of the degree of decentralization.

Fig. 1
figure 1

Overall, Central Committee (CC), and Alternate Central Committee (ACC) incumbency ratios: 8th–16th Central Committees

Full size image

The relatively high incumbency ratio and the sizable representation by central technocrats and SOE managers in recent years mean that policy centralization will remain a predominant mode of economic management in the foreseeable future. Because sustainable decentralization will not be a path to reform, leaders in China will need to derive new avenues of fashioning reform coalition and credible reform policies.

Regarding alternative explanations, scholars focus either on Mao’s grand vision for communism or on the destruction of central planning capacities during the Cultural Revolution as explanations of decentralization policies. The economics literature accurately focuses on the Maoist legacy of decentralization. However, these accounts do not explain the persistence of Maoist policies into the 1980s despite massive reshuffling in the top leadership. Maskin et al. (2000), Sachs et al. (2000), Qian (1999), and Xu (2011) observe that China’s economy, even during Mao’s reign, operated on a geographical basis instead of on a functional basis, as the Soviet Union did.

Although top leaders’ policy preference can have a profound influence on outcomes in authoritarian regimes (Jones and Olken 2005), we are not convinced it tells the whole story here. Even if Mao genuinely had preferred decentralization, we show clear evidence that many of the decentralization policies were rolled back after the end of the Great Leap Forward in 1961, suggesting that economic policies and institutions lacked stickiness (Lardy 1975). The malleability of economic institutions further adds to the puzzle of economic decentralization in the Deng era. When Deng ascended to the top position in the CCP, he reoriented policies away from autarky and collective agriculture (Fewsmith 1994; Shirk 1993). Given Deng’s clear wishes to distance himself from Cultural Revolution excesses, why did he follow Mao’s footstep to pursue decentralization?

To be sure, the answer may be that Deng, like Mao, preferred decentralization. However, examining the policies pursued by Deng prior to the reform, he did not consistently pursue decentralization. Like most senior leaders, Deng supported Mao’s radical decentralization policies during the Great Leap Forward, but he also lent his help in the Great Leap recovery period, which saw re-centralization (MacFarquhar 1983; Donnithorne 1980). Finally, given that Deng still had considerable power after the Southern Tour, why did he allow fiscal centralization to take place in 1994?

Beyond the “great man” hypothesis, scholars also argue that the Cultural Revolution decimated the central bureaucracy so that it no longer had the capacity to fully re-impose a centralized planned economy on China after 1976 (Shirk 1993: 13). Indeed, the State Planning Commission, the heart of the centralized planned economy, lost nearly 90% of its cadres by 1970 (Oksenberg and Tong 1991). The Ministry of Finance was so disrupted by Red Guard activities that it was occupied by the military between 1967 and 1975 (Xiang 1999). However, the central bureaucracy had been on the mend since the Lin Biao Incident in 1971 (Swaine 1986; Chen 2005). By the early 1980s, the State Planning Commission was making important decisions on production, prices, and investment (Shih 2008).

More so than previous explanations, our account explains the rise of decentralization policies in 1970, the continuation of these policies through leadership changes in the late 1970s and into the 1980s, as well as the sudden reversal of decentralization policies in the early 1960s. Furthermore, our account predicts the re-emergence of fiscal and firm re-centralization in the 1990s, as well as the persistence of a centralized mode of economic policy into the foreseeable future.

Data and Measurement

Unlike the extant literature, which focuses primarily on fiscal decentralization, we see economic decentralization as a broader concept. Xu (2011) posits that economic decentralization is featured by “the governance of the national economy is delegated to subnational governments. Regional economies (provinces, municipalities, and counties) are relatively self-contained, and subnational governments have overall responsibility for initiating and coordinating reforms, providing public services, and making and enforcing laws within their jurisdictions.” Accordingly, we conceptualize economic decentralization as the devolution of power of controlling and allocating economic resources—including fiscal revenues, investment funds and state-own enterprises (SOEs)—from the central government to subnational and local governments.

Based on the share of local revenues in national revenues, the share of the non-state budgetary investment in capital construction in total capital construction investment, and non-SOE share of industrial output, we use dynamic factor analysis to construct a composite indicator to measure economic decentralization. As shown in Fig. 2, the composite indicator of economic decentralization increased temporarily during the Great Leap Forward, declined again during 1962–1966, increased secularly after 1969 (the 9th Party Congress) but stalled after 1994. The change of this composite indicator corresponds well with the economic decentralization and centralization cycles discussed in the subsequent section.

Fig. 2
figure 2

Composite indicator of economic decentralization

Full size image

The Composition of the Chinese Communist Party Elite

In terms of our main explanatory variable, elite composition, we follow Shirk (1993) by assuming that the Central Committee contains the bulk of China’s political elite. Full CC members typically occupy crucial positions in central ministries, Party organs, military institutions and local authorities, and CC plenums, they vote on important decisions. We make use of a quantitative database of every CC member, which traces their entire career trajectories, to construct time-series data on annual percentage of CC members made up of central, provincial, and military officials from 1956 to 2006 (Shih et al. 2012).

We adopt three definitions of “central CC members.” The first only counts those who served in a central party or State Council organ. The second, broader definition also includes CC members who held positions in the National People’s Congress (NPC), the Chinese People’s Political Consultative Committee (CPPCC), and the Supreme Court or Procuracy. As seen in Fig. 3, the two definitions of “central CC members” exhibit similar trends. In Fig. 4, we only consider State Council and central SOE officials as “central CC members” since they had a high stake in maintaining central economic authorities (Shih 2008). Second, we define “provincial CC members” as CC members who worked in party or government posts at the provincial or regional level in a given year.Footnote 1 Finally, “military CC members” refer to those who served in the People’s Liberation Army (PLA), including the Central Military Commission (CMC), PLA headquarters, and the military regions.

Fig. 3
figure 3

The share of central, military, and provincial officials in the CC. Source: COD & PHRC 2004

Full size image
Fig. 4
figure 4

Share of central and provincial officials in the CC. Source: COD & PHRC 2004

Full size image

Analysis of the China Case

Rapid Reversal of Decentralization During the Great Leap Forward

We begin our analysis with the composition of the Central Committee at the 1956 8th Party Congress, which elected the CC elite for the next 13 years. A large body of scholarship suggests that both fiscal and enterprise decentralization had their origin during the Great Leap Forward, which began in 1958 (Xu 2011; Donnithorne and Lardy 1976; Oksenberg and Tong 1991; Naughton 1988). However, this temporary bout of decentralization was quickly followed by re-centralization in the early 1960s (Figs. 2 and 5). Examining the composition of the Central Committee elite at the 1956 8th Party Congress, the ease of re-centralization was likely helped by the fact that the vast majority of the political elite at the time served as central officials. Figure 3 presents a dynamic picture of the share of central, military, and provincial cadres in the CC. From the 1956 8th Party Congress to the outbreak of the Cultural Revolution in 1966, central officials constituted the overwhelming majority in the Central Committee. According to the first definition, the central share of CC members hovered around 75% during this period. The broader definition produced a central share of the CC in the 80% range in most years of this period. In contrast, local political elites were marginalized in the 8th Party Congress, and their disadvantage continued until the Cultural Revolution.Footnote 2

Fig. 5
figure 5

Ratio of local expenditure, local revenue, and extrabudgetary revenue (%). Source: SSB (various years)

Full size image

Given the dominance of central officials in this period, it was not surprising that Great Leap decentralization policies were quickly reversed after 1960. In Fig. 5, local share of total revenue and expenditure rocketed upward during the Great Leap Forward (GLF) in 1958 and maintained at that level until the late 1970s, when it began to fall. Examining fiscal policies closely, China first began fiscal decentralization in 1957, which devolved control of nearly all state-owned enterprises (SOEs) to the local levels, shifted economic planning from the central to the local levels, and implemented revenue sharing agreements between the center and provinces which were fixed for 5 years. The results were breath-taking: the share of industrial output produced by enterprises subordinated to the central government shrank from 40 to 14%. The share of central revenue likewise decreased from 75% to below 50% (Fig. 5).

The decentralization coincided with the GLF and was an integral part of Mao’s strategy to spur local enthusiasm for increasing production (Donnithorne and Lardy 1976). Despite impressive official figures showing considerable fiscal decentralization, the reality was much more complex. First, as Lardy shows, the central authorities still tightly controlled expenditure so that provinces which collected more revenue did not spend more than before the decentralization (Lardy 1975). Furthermore, although the center initially promised to set revenue sharing rates with provinces for 5 years, it withdrew the promise in 1959 and set new rates with provinces in accordance with the budgetary outcomes in 1958 (Lardy 1975).

In the aftermath of the Great Leap Forward, the central government also reclaimed much of its power even as the provinces continued to collect the bulk of the revenue. Thus, the provinces enjoyed little de facto fiscal autonomy between 1961 and 1970. Figure 5 shows local share of expenditure sharply declining at the end of the Great Leap. Figure 6 shows that the boom in collective enterprises financed with local revenue quickly came to the end in the early 1960s as spending authority re-centralized. Furthermore, the Ministry of Finance froze the extrabudgetary bank accounts of all governmental units and enterprises to prevent inflation, which also had the effect of depriving local units of fiscal autonomy (Xiang 1999). Figure 5 reveals that local extrabudgetary revenue relative to the national budget froze at the same level at the end of the Great Leap and did not grow until the early 1970s. Finally, the launching of the Third Front construction project in 1964 concentrated enormous budgetary funds at the hands of central authorities to focus on key construction projects (Naughton 1988). Without a credible constitution and with a political elite dominated by central officials, the rapid re-centralization seen in the post-Great Leap period was fully expected.

Fig. 6
figure 6

SOE and collective share of industrial output. Source: SSB (various years)

Full size image

Credible Decentralization: Removal of Central Officials from the CC

The second, more permanent wave of decentralization began in 1970. Although this wave of decentralization shared many features of the 1958 decentralization, the level of decentralization was deeper and longer lasting. Most large-scale SOEs were again devolved to provincial and municipal control. In the fiscal realm, a “lump sum transfer” system was adopted, which devolved the bulk of fiscal resources to the local level (Oksenberg and Tong 1991). Unlike the GLF, the pursuit of decentralization after 1969 naturally drew strong support from an elite selectorate dominated by local officials, whereas centralization policies elicited opposition and foot-dragging.

As seen in Figs. 3 and 4, Cultural Revolution was a watershed in the composition of the CC. When the Cultural Revolution descended on the CCP, a substantial number of CC members working in central Party organs and the State Council were deprived of their CC membership. This political shock also was exogenous to economic policies and had to do with the political struggle between Mao and his designated successor Liu Shaoqi. The tension between Mao and Liu Shaoqi shifted into high gear after the 1962 7000 cadre conference, when both Liu and his followers directly criticized Mao for committing an error during the disastrous GLF (Gao 2003: 90). Mao finally mobilized contingents of the People’s Liberation Army (PLA) under Lin Biao and radical propaganda officials in Shanghai to unseat Liu and other potential rivals in 1966 (MacFarquhar 1997; MacFarquhar and Schoenhals 2006). On the eve of the Cultural Revolution, Liu’s followers occupied over 40% of the Central Committee seats (Shih et al. 2010).

Thus, Mao’s decision to eradicate his potential enemies inadvertently led to the decimation of the central bureaucracy. As seen on Table 2, scores of ministers and vice ministers, including many who had previous ties with Liu like Cheng Zihua, Bo Yibo, Yang Yichen, Yang Xiufeng, and Zhang Linzhi, among others, were removed from the State Council at the on-set of the Cultural Revolution. Naturally, they were not re-elected into the CC at the 1969 9th Party Congress. Of the 22 ministers who had either been full or alternate Central Committee members in 1965, 13 had been purged during the CR, while another two were rotated to provincial administration. Only six ministers in 1965 remained in power by the 1969 9th PC.Footnote 3

Table 2 Ministers in 1965 and their fate during the CR
Full size table

Why did not Mao replace the ministers with his trusted followers who were high-ranking CC members? In brief, doing so would have served no political purpose for Mao. Mao launched the Cultural Revolution in part to eradicate increasingly powerful central state organs dominated by factions that challenged him. He was not interested in rebuilding them only to face the same threat again. Mao loyalists instead took control over provinces and military units.

Thus, when political power was redistributed among different groups at the 9th Party Congress in 1969, a fundamental shift in the power structure favoring decentralization had occurred. As seen on Fig. 3, the central share of CC membership, calculated by the more restricted definition, plummeted from around 75% to below 30%. Figure 4 reveals that the share of State Council and central SOE CC members dropped considerably from 48% in 1965 to only 16% in 1969.

Meanwhile, the Cultural Revolution saw the rise of the local political elite: the provincial share of CC membership soared from approximately 16% in 1968 to 49% in 1969 and reached its peak in 1973 (Figs. 3 and 4). The rise of provincial representation stemmed from three major factors. First, because Mao only needed to destroy the most threatening factions rather than all factions, provincial administrators, who had come from a much more diverse array of factions than central ministers, were not purged to the same extent as the central ministerial purge. Table 3 reveals that only slightly above half of the provincial leaders who had been CC or alternate CC members in 1965 were purged during the CR, a lower ratio than for ministers who had been CC or ACC members.

Table 3 Provincial officials in 1965 and their fate during the CR
Full size table

In addition, because the military played a large part in pacifying the chaos in the provinces, military officers took over numerous important positions in the provinces when the revolutionary committees were formed to replace party committees (MacFarquhar and Schoenhals 2006). Unlike central ministries, the new provincial military rulers tended to be high prestige Mao loyalists who comfortably entered the CC. The decision to place his trusted followers from the military in provincial administration led to a rising presence of the military in the CC from 26% in 1968 to 52% in 1969. Only Lin Biao’s purge in 1971 led to a decline in PLA representation back to around 40%. The rise of provincial CC representation was also caused by Mao’s decision to enlarge the CC by introducing scores of provincial “mass representatives” into the 9th Central Committee, presumably to balance against the military (MacFarquhar and Schoenhals 2006).

The radical reorientation of the political elite during the Cultural Revolution had a substantial impact on the incentives of the top leaders. With the CC dominated by local officials, senior leaders who wanted to form a winning coalition must ally with a sizable portion of provincial leaders.Footnote 4 Naturally, their concerns and economic interests also became important considerations in top-level economic decision-making. Although Mao may have preferred decentralization, his successors were committed to a path of decentralization because of an elite selectorate dominated by local interests.

After the 9th Party Congress, the provinces began to pursue drastic decentralization policies. In the fiscal realm, a federal “lump sum transfer” system was adopted. Under this arrangement, some revenue sources, including customs duties and central SOE revenue, were designated to the central government, while other sources, like local SOE revenue, were defined as local income (Oksenberg and Tong 1991). Most important, fiscal surplus was divided between the center and the provinces according to pre-arranged agreements, and if there was remainder after the agreed-upon remittance to the center, local governments could keep all of it. More so than previous revenue sharing arrangements, this system created strong incentives for local authorities to maximize revenue (Oksenberg and Tong 1991). Figure 5 reveals that localities obtained the highest share of revenue in the first half of the 1970s, not the 1980s or the 1990s. Figure 5 also shows that local extrabudgetary revenue relative to the national budget began to climb after the 9th Party Congress. Local fiscal autonomy was not reversed until the implementation of the tax-sharing system in 1994.

Enterprise decentralization also underwent a similar experience as fiscal decentralization in the early 1970s. Most of the state-owned enterprises were delegated to local governments at the provincial, municipal, and county levels. The share of industrial output produced by central SOEs declined from about 50% in 1965 to only about 8% after 1970. This clear trend of enterprise decentralization persisted for two decades. By 1985, central SOEs only accounted for 27% of total SOE industrial output, while provincial and municipal SOEs accounted for 45% (Table 4) (Qian and Xu 1993).

Table 4 Industrial output produced by central SOEs as a percent of total SOE production (%)
Full size table

An often discussed ingredient of China’s rapid growth, township and village enterprises (TVEs), also saw rapid expansion during the Cultural Revolution (Oi 1992; Huang 2008). Figure 6 shows that collectives, predecessors of TVEs, began to produce a larger and larger share of industrial output after 1970. Between 1970 and 1978, total collective enterprise output relative to SOE’s output more than doubled.

In terms of investment, decentralization began in earnest in the early 1970s. Aside from the Great Leap interlude, Fig. 7 reveals that 90% of capital construction investment before 1970 was directly allocated from the central budget. After 1970, however, firms began to draw from increasingly diverse channels for financing. In essence, when SOE ownership was transferred from the central to the local level, local governments obtained the authorities to set up new firms and to make investment with “self-raised funds” which existed outside of the formal budgetary channels (Wong et al. 1995). Extrabudgetary revenue grew by at least 10% per year for the first half of the 1970s (Fig. 5). Throughout much of the 1970s, local government construction investment exceeded central investment.

Fig. 7
figure 7

Share of fixed asset, capital construction, and technical update investment financed by budgetary appropriation. Source: FAISDSSB & IINDRC (various years)

Full size image

When they took supreme power in the late 1970s, Deng Xiaoping and Chen Yun were faced with a CC dominated by local officials, many of them military figures. This situation did not change until the 1982 12th Party Congress, when the share of State Council officials in the CC spiked to above 30%, still a shadow of its former representation. Unlike Mao, Deng and Chen ruled by coalition and could not thoroughly reshuffle the elite selectorate. As Fig. 1 shows, even at the height of Deng’s campaign to force his colleagues into retirement at the 13th Party Congress, the Central Committee incumbency ratio stayed above 45%, some 15% higher than was the case at the 9th Party Congress. Into the 1990s, CC incumbency ratio hovered around 70% (Fig. 1). This shows that the elite selectorate had a stickiness that was not easily altered within a short time. Short of a thorough purge like the Cultural Revolution, it took a decade or more to fundamentally shift the composition of the elite selectorate, thus providing policies preferred by the elite some medium-term credibility.

The heavy weight of local leaders in the political elite had an influence on decision-making even after the Cultural Revolution. For example, at an April 1979 meeting on economic planning, the provinces lobbied for the universal adoption of fiscal contracting that Jiangsu had enjoyed since 1977. Although Deng had centralized the fiscal system in the early 1960s, as the supreme leader of China, he could ill afford to directly offend a majority in the CC. Faced with the same political constituency, even a stalwart of central planning like Chen Yun also supported provincial fiscal contracts over the Ministry of Finance’s objection (Deng 2005: 144). The provincial majority in the CC at the time had the muscle to “force a showdown” (tanpai) with Deng and Chen (Shirk 1993: 165). Even in 1986, the State Council’s plan to replace the tax-contracting system with the centralized “tax-sharing system” was stymied by opposition from local officials (Zhang 2007). Thus, with the exception of some retrenchment in fiscal revenue, most of our indicators show that decentralization policies continued throughout much of the 1980s, giving rise to the general observation that there was considerable local fiscal autonomy in the 1980s (Qian et al. 1995; Wong et al. 1995; Oi 1992). Unlike in the early 1960s, rapid re-centralization was not a politically feasible option.

Regaining the Commanding Heights: the Rise of Centralizers in the 1990s and Beyond

During the era of reform and opening, central officials increased their presence in the CC gradually and reasserted their dominance in the CC in the 1990s. After the Cultural Revolution, Chen Yun, Deng’s ally in the struggle against Hua Guofeng, scrambled to rebuild ministries and offices in the State Council, which had been his traditional power base prior to the Cultural Revolution (Shih 2008; Cui 2003).

Furthermore, as the economy grew and became more complex in the 1980s, the central government also required additional agencies to handle new problems, such as information asymmetry, inflation, corruption, and trade friction (Fan et al. 1996; Fewsmith 1994). In reaction to these problems, the Export-Import Commission, the Special Economic Zone Office, the State Economic Commission, and the State Statistical Bureau, as well as the four specialized banks, were formed as ministerial or vice-ministerial organs in the central government. Given the bureaucratic ranks of these organs, their leaders were inducted into the CC either as full or alternate members. The rise of the technocrats also dovetailed nicely with Deng’s rejuvenation effort (Cui 2003). A key component of Deng’s rejuvenation campaign was the systematic transfer of military CC members, many of whom with provincial positions, to the Central Advisory Committee or to full retirement, while younger, more educated officials were placed in new State Council positions (Manion 1993). At the same time, the Organization Department also began to consider CEOs of large state-owned enterprises as potential candidates for the Central Committee (Cui 2003: 94). These various forces led to a steady increase in the share of State Council officials in the CC throughout the 1980s and the early 1990s such that State Council CC members exceeded provincial members significantly by the 1992 14th Party Congress (Fig. 3).

Because of the steady accumulation of power by central officials in the 1980s, top leaders who enacted centralization policies in the 1990s had the support of a sizable block of central officials in the Central Committee. To be sure, the switch from the tax-contracting to the tax-sharing system, a milestone in fiscal centralization, necessitated tremendous political will from the central leadership and the dismissal of a few recalcitrant provincial leaders (Fewsmith 2001; Brahm 2002; Yang 1997, 2004). However, State Council and central SOE officials made up over 30% of the CC by that time, while central officials as a whole made up nearly the majority after the 1992 14th Party Congress (Figs. 3 and 4).

As for provincial officials, their dominating presence in the CC ended in the late 1980s, especially after the Party Congresses. Their presence further slipped at the 14th Party Congress in 1992. By the early 1990s, local officials were no longer in the position to “force a showdown” with central leaders.

Given this new political alignment, Huang’s (2008) excellent study and our indicators reveal a pattern of centralization of the Chinese economy in the 1990s and beyond. Figure 5 shows the well-known collapse of local share of revenue after the tax-sharing reform in 1994, as well as the dramatic fall in extrabudgetary revenue relative to the national budget.Footnote 5 Meanwhile, central SOEs regained ground as they consolidated so that their output as a share of all SOE production rose from 28% in 1993 to 75% in 1997 (Table 4). Although SOE output as a share of total industrial output continued to fall in the 1990s, the statistics do not fully reflect the strong presence of state-controlled joint-stock enterprises. In 2009, central SOEs and state-controlled joint-stock conglomerates generated 12.6 trillion RMB in revenue, over one third of China’s 2009 GDP (State Asset Supervision and Administration Commission 2010).

More importantly for China’s current economic model, which relies heavily on investment, the central government has dramatically centralized the sources of funding for investment, including budgetary funds, loans from major state banks, and bond and stock issuance approved by central agencies. Figure 7 shows that for a while, central budgetary allocation was replaced by local extrabudgetary income and increasingly bank loans. However, the 1990s saw loans from centrally controlled state banks becoming a major source of funding for investment (Shih 2008). In combination with the consolidation of SOEs and rising share of central officials in the Central Committee, investment share undertaken either by the central government or central SOEs bottomed out in the early 2010s and actually began to rise in 2015 to around 40% in 2017 (Fig. 8). This is rather shocking considering the torrent of FDI and private investment since the 1980s. As Fig. 8 shows, however, domestic private sector investment share has stagnated just above 30% after a decade of robust growth in the 2000s. The accumulation of central financial power, made possible by the accumulation of central political power, has led to the continual dominance of the central government in China’s investment-led growth model.

Fig. 8
figure 8

Central government and domestic private sector’s share of monthly investment (2005–2009/2017). Source: CEIC, NBS

Full size image

“Shadow Cases”: Taiwan and the Soviet Union

The association between the party elite composition and policy biases is by no means unique to mainland China. Similar patterns emerged in other Leninist regimes, particularly Taiwan under Kuomintang (KMT)’s rule and the Soviet Union. Although the causality between changing elite composition and policy biases in these cases may not be as stark as the case of China, they suggest that elite composition contributed to at least the ability of new policies to take hold without dramatic reversals.

To a large extent, Taiwan’s economic rise after World War II can be attributed to a shift to outward-oriented economic polices during the 1958–1963 period (Haggard and Pang 1994). This shift to export-led growth strategy was by no means inevitable. The conservative forces composed of Executive Yuan ministries, SOEs, and the military benefited from import substitution industrialization (ISI) model (Kuo and Myers 2011; Haggard 1990). Only when the conservative camp was undermined by the reformers did Taiwan embark on the path toward export-led growth (Haggard 1990).

In the midst of this policy shift, the share of the Executive Yuan representatives (central technocrats) in KMT’s Central Committee (CC) dropped from 30% in 1952 to 19% in 1963 because of Chiang Kai-shek’s desire to decrease the power base of a potential rival, Chen Cheng (Taylor 2009). In the meantime, the share of local representatives in the KMT CC soared from 9% in 1952 to 21% in 1963, credibly signaling the rising influence of local, pro-private sector representation (Li 1993). Unlike Communist China, local elite in Taiwan included local officials, as well as native Taiwanese businessmen and notables, who had strong incentives to promote reform and decentralization. From 1963 to 1972, Taiwan’s average annual growth reached a breakneck 10.9%, fueled by private sector-dominated export growth (Cheng 2001).

In the 1980s, Taiwan witnessed further economic liberalization, including tariff reductions, industrial upgrading, financial liberalization, SOE privatization, and the launch of the Hsinchu Science-Based Industrial Park (HSBIP) (Hsueh et al. 2001: 279–285; Pao et al. 2008). During this period, the share of local elites in the KMT CC expanded to 26%, suggesting their increasing clout in the political arena (Li 1993).Footnote 6 Moreover, our measurement underestimates the political clout of local elites because a sizable number of them took positions in supervisory branches like the National Assembly and the Legislative Yuan. Taking this into account, we find that the share of local elites constituted one third of the KMT CC in 1988. Although other factors undoubtedly contributed to Taiwan’s liberalization policies, the steady representation of local interests in the KMT CC provided reformist leaders and technocrats in the regime natural allies in the policy debates.

In the case of the Soviet Union, there was a high correlation between elite composition and policy changes. In 1921, wartime communism marked by economic centralization was replaced by the New Economic Policy (NEP) aimed at taking a step back from central planning. The NEP granted more rights to individual peasants, privatized small-scale enterprises, and decentralized decision-marking throughout industries (Davies 1998:24; Gregory and Stuart 1990: 60). Similar to China during the Cultural Revolution, the share of the central government officials in the Communist Party of the Soviet Union Central Committee (CPSU CC) dropped from 30% in 1922 to 21% in 1924. However, in the 1930s, Stalin instituted an increasingly powerful central bureaucracy, which exerted direct control over agricultural and industrial production, as well as on investment (Davies 1994: 14–23). Not surprisingly, the share of central technocrats in the CC surged to 37% of the CPSU CC in 1934.

After Stalin’s death, Khrushchev consolidated his power and launched the Sovnarkhoz Reform in 1957, which devolved economic management from Moscow to the regions. Accordingly, 105 regional economic councils were established, and all but a few ministries were abolished (Gregory and Stuart 1990: 143). Khrushchev’s reform intended to build reform coalitions by strengthening the territorial apparatus and party control (Tompson 1993). Consequently, the influence of central bureaucrats was diminished such that their share in the CC decreased from 25% in 1956 to 17% in 1961. Yet, representatives from the Republics and the oblasts remained relatively unchanged at around 50% through this period.

After Khrushchev was ousted in 1964, his successor, Brezhnev, quickly abolished regional economic councils and re-centralized the economic bureaucracy (Gregory and Stuart 1990: 148). Brezhnev’s policy agendas enhanced the economic involvement of the ministries whose representatives in the CC increased to 24% in 1976. Gorbachev assumed the leadership in 1985 and was faced with economic stagnation, prompting him to launch perestroika, part of which aimed at decentralizing decision-making to local level (Treisman 2012: 16). As it turned out, the reforms of Gorbachev era radically changed the composition of the CC, leading to the sharp decline of central bureaucrats’ share in the CC from 24% in 1986 to only 6% in 1989.Footnote 7

In short, the perpetuation of economic decentralization policies in Taiwan was accompanied by the steady expansion of local elites’ seats in the CC. In contrast, the reform policies enacted by Lenin and Khrushchev were subverted by their successors. The elite composition of the CC in the Soviet Union did not shift fundamentally and was unable to credibly signal decentralization to local governments and economic agents.

Future Reform Path for China

The analysis above presents a political mechanism—elite reshuffling—that helps even authoritarian regimes commit credibly to decentralization. In China, both exogenous shocks and endogenous processes led to changing composition of the CC elite, which changed the top leadership’s incentive to either pursue decentralization to appease provincial interests or centralization policies supported by the central bureaucracy. The composition of the political elite compelled the top leadership to pursue either centralization or decentralization policies in similar ways that federalist institutions committed democratic leaders. The stickiness of elite composition in most periods provided further credibility to the stability of elite policy preference.

Despite the chaos and suffering of the Cultural Revolution, the wholesale removal of State Council officials and the resulting domination by provincial officials in the Central Committee committed the post-Mao leadership to pursue decentralization policies throughout much of the 1980s. Credible decentralization in the medium term paved the way for a series of reform in the 1970s and the 1980s which promoted growth, including fiscal decentralization and the rapid proliferation of township and village enterprises. We show that the domination of local officials in the Central Committee likely made these policies credible in the medium term. The flip side of the same coin was that the two decades of decentralization also rendered central fiscal and financial resources relatively weak. This prevented the central government from dictating investment patterns.

However, in the 1980s, senior State Council leaders aggressively rebuilt the central bureaucracy, and increasingly complex economic demands made their arguments of a larger central bureaucracy more plausible. After several rounds of turnovers in the 1980s and the early 1990s, central bureaucrats’ presence in the CC finally rivaled that of local officials, reversing the dramatic reorientation of the Cultural Revolution. By the early 1990s, pursuing centrist policies was no longer such a difficult political proposition. In the 1990s and 2000s, the capacity of the central state multiplied with fiscal and banking centralization successively, which endowed the state banks with tremendous firepower with which to finance central investment and industrial policies (Shih 2008).

Given the status quo of central dominance both politically and economically, credible decentralization spurring local policy and reform initiatives is an unlikely path for reform in the foreseeable future. In fact, the opposite may be happening. Knowing that the central government will continue to control the lion’s share of financial resources in the near future, local governments may forego local initiatives that foster growth in favor of lobbying activities that earn them a larger slice of central resources. For example, to achieve a given growth or investment target, local officials likely find it easier to draw up a multi-billion RMB urban development plan, the approval of which would entitle local authorities to billions in bank loans and bond issuance, than to entice private and foreign investors with tax breaks and market access.

In this environment, genuine reform will require the central government to overcome vested interests and to self-impose limits on its own roles in allocating financial resources, which was 251% of GDP at the end of 2017 (IMF 2017). If the market is to play a “decisive role in allocating resources,” as stated in the 19th Party Congress Work Report (Xi 2017), a strong leadership will need to enact credible and durable policies that provide private actors control over a large share of financial resources. Instead of keeping private capital in a small corner of the financial system, the Chinese government would need to allow private financial institutions to grow much bigger or to privatize major financial institutions.

For the literature on credible decentralization in authoritarian regimes or weakly institutionalized democracies, the case of China suggests that in addition to a focus on institutional changes, which often were outcomes of political processes, scholars would do well to focus also on the impact of political calculus and shocks on economic outcomes such as decentralization and redistribution. To be sure, a focus on political shocks lies behind much of the path dependence literature (e.g., Thelen 2004). Unlike in established democracies, where political coalitions in the early modern period or during the Great Depression shaped relatively stable institutions, institutions and policy tendencies were much more malleable in authoritarian regimes and weak democracies. Thus, even in the medium term, short-term political calculus and struggles can shape federalist or redistribution regimes for decades (e.g., Albertus 2015). The next wave of comparative scholarship hopefully will fine tune our understanding of the types of political conflicts and calculus that may produce longer lasting changes in economic policies.