Picking Places and People: Centralizing Provincial Governance in China

Abstract China's political system has been characterized by two institutions since the 1980s: an explicit “layer-by-layer administrative hierarchy” and the “appointment of cadres one level down.” There have, however, been two departures from these administrative practices. First, some provinces have “empowered prosperous counties” by placing them in a dual-reporting relationship with both prefecture-level cities and provinces. Second, some provinces have restored personnel control going “two levels down” by appointing key officials at the county and urban district levels of government. These deviations evolved as responses to China's GDP-centric policy environment during the early reform era. Based on field interviews and nationwide analysis of city-level personnel data, this article argues that such adaptations have generated unintended conflicts between provinces and prefecture-level cities. While prior studies of evolutionary change in China highlight the relationship between state and non-state actors, this study demonstrates how interactions among state actors themselves may fundamentally transform the dynamics of administrative governance.

to lower levels of administration. For over three decades, China's political system was similarly characterized by two organizing principles: the structure of an "explicit layer-by-layer administrative hierarchy" and "one-level-down personnel control." 2 The vertically layered administrative apparatus was stipulated by Articles 108 and 110 in the 1982 Constitution of the People's Republic of China (PRC), which granted local administrative entities downward authority at or above the county level. 3 Two years later, the principle of appointing cadres "one level down" was enforced by the central government and appeared to be institutionalized as an administrative practice (Figure 1). 4 In 2016, however, this bureaucratic rule was compromised by two provinces when Zhejiang and Guangdong's provincial organization departments publicly announced the appointment of cadres at the county and urban district level, which are two levels below the provincial government. In effect, prefecture-level cities in these two provinces have lost appointment authority over their immediate subordinate governments. How can this deviation from the 1982 constitutional mandate be explained? 5 What are the implications of this reconfiguration of administrative authority among China's sub-national units?
Based on original field interviews and city-level data, this article observes that evolutionary shifts in administrative governance have generated unintended conflicts in China's inter-governmental relations. Specifically, in the early reform era, a popular provincial strategy for boosting GDP figures took the form of enhancing the authority of economically strong counties. This entailed placing counties in a dual-reporting relationship to both prefecture-level and provincial governments. Despite the possibility that increasing the autonomy of counties might weaken prefecture-level governments, leaders of prefecture-level cities accepted this arrangement because the GDP of empowered counties was still counted in official statistics reporting the overall economic performance of prefectures. 6 As the growing autonomy of counties appreciably contributed to China's economic development during the 1990s, 7 the central government reinforced linkages between counties and provinces by recommending the "empowerment of economically strong counties" in a series of policy documents during the 2000s. Various provincial governments subsequently extended personnel control to county and urban districts, thereby reviving the pre-1984 practice of appointing cadres "two levels down." 8 We contend that this creeping evolution in the reach of personnel appointments has yielded unintended consequences. In an era of more moderate economic growth, the central government has shifted to prioritizing multidimensional developmental targets, 9 and advocates coordinated investment across an entire jurisdiction. Hence, prefecture-level cities no longer prioritize the statistical contribution of empowered counties; instead, they seek the capacity to distribute fiscal resources throughout their territorial jurisdictions. 10 This new expectation for prefecture-level cities has exacerbated tensions with provincial governments: prefecture-level governments seek greater fiscal autonomy just as provincial authority has expanded. However, a complicating phenomenon is that provincial capitals rely less on the economic development of counties than non-capital cities owing to the legacy of weaker counties surrounding capitals. As a result, provinces prefer to support their capital cities over regular prefecturelevel cities at a similar economic scale.
This article proceeds as follows. The first section conceptualizes vertical intergovernmental relations within the framework of evolutionary governance in China, and proposes that the progressive circumvention of reporting lines by * includes four autonomous regions; "n" indicates the number of this administrative entity. Administrative entities placed at the same horizontal level reflect the same administrative rank. 9 Ma, Zhigang 2015. 10 Interviews with prefecture-level officials, Ningbo, 5 February 2016; interviews with academic experts, Ningbo, 11 July 2017; Hangzhou, 7 April 2019; Chung 2016, 52. specific levels of government has changed China's formal administrative hierarchy over the past three decades. The second part traces the emergence of "empowering economically strong counties" and why provincial governments selected this as an adaptive strategy to centralize provincial authority. The third section elaborates on the extension of the province's personnel reach down to the county/urban-district level. As shown in the fourth section, the rise of direct province-county relations has intensified fiscal conflicts between provinces and prefecture-level cities. The conclusion reflects on the implications of these dynamics given the present leadership's centralizing efforts.

China's Vertical Intergovernmental Dynamics
Evolutionary governance among state actors Shocks such as wars, revolutions and external occupation are familiar critical junctures during which political institutions may undergo major transformations. By contrast, explaining gradual institutional change during "normal" times has inspired analysis of mechanisms endogenous to a political system. For example, James Mahoney and Kathleen Thelen observe that because institutions are based on negotiations conducted by particular political coalitions, even the most apparently stable institutions may be subject to renegotiation at the margins. 11 Over time, apparently minor or incremental adjustments may yield significant changes. In this regard, the logic of evolutionary theories provides insights for exploring the mechanisms of endogenous institutional change. As Orion Lewis and Sven Steimo observe, "Imperfect copies of earlier compromises help to ensure imperfections in the replication of institutions, which is often a primary driver of further change." 12 From a relational approach, Yuen Yuen Ang highlights mutual adaptation as the key distinction between "evolution" and "gradual institutional change" and advises tracing the process of institutional adaptation by identifying "two (or more) domains of significance." 13 Ang's study maps the co-evolution between the local state and market forces in China. 14 Along similar lines, various scholars have examined co-evolutionary dynamics between China's NGOs and relevant state bureaucracies. 15 Overall, social scientists who engage with the concept of evolutionary governance in China focus on the interaction between state and non-state actors. 16 By contrast, this study demonstrates how interactions among state actors themselves may fundamentally re-define the dynamics of administrative governance. Even in an authoritarian regime with a unitary political system and one-party rule, we observe opportunistic adaptations in the evolution of vertical intergovernmental dynamics, as lower layers of government attempt to safeguard their autonomy from higher levels.

Mechanism of institutional adaptation: directed improvisation
Various studies note that China's multitiered administrative structure breeds institutional adaptation at the local level. Yongshun Cai explains that "a political system with multiple levels of authority can reduce uncertainty by granting conditional autonomy to lower-level governments." 17 Relatedly, Xiaoxiao Shen and Kellee Tsai identify variation in the institutional adaptability of local states following the global financial crisis. 18 But why and how does institutional adaptation occur? Our study builds on Ang's concept of directed improvisation to elucidate the incentives driving institutional adaptation. Directed improvisation allows for creativity at the local level within the parameters of a particular institutional context. "A skilled director," Ang observes, "is not the one who dictates to actors what exactly they should do, but empowers them in the creative process." 19 Most studies of China's reform era are premised on a longstanding administrative structure of governance with an explicit "layer-by-layer administrative hierarchy" and "one-level-down personnel control." 20 Operationally, decentralization is expected to occur based on vertical layers of administration, 21 and the relative autonomy of each administrative unit is secured by its control over the appointment of personnel in the immediate subordinates.
However, when the central government started strengthening provincial authorities in the 1980s, in a parallel process it also empowered some prefecture-level cities by designating them as "separately-planned cities" ( jihua danlie shi 计划单列市). 22 These cities were granted access to negotiate economic affairs directly with the central government, bypassing their provincial authorities. Following central mandates to encourage decentralization, provincial governments started to empower "economically strong counties" (qiangxian kuoquan 强县扩权), sidestepping some prefecture-level cities. 23 Consequently, the earlier arrangement of directed improvisation within a particular level of government was destabilized by a lower layer of empowerment between their superiors and subordinates.
Our research delineates how state actors have adapted to the disruption of directed improvisation in certain localities and contends that the conventional 17 Cai 2008, 411. 18 Shen, Xiaoxiao, and Tsai 2016. 19 Ang 2016, 69. 20 See, e.g., Burns 1994Huang 1999;Oi 1995. 21 Xu 2011. 22 Landry 2008Chung 2010. 23 Lu andTsai 2019. depiction of China as possessing a symmetrical chain of power delegation is merely a snapshot of the co-evolutionary process of interaction among state actors at multiple administrative layers. Over the past three decades, some provinces have lost authority over separately-planned cities while gaining access to empowered counties. Meanwhile, some prefecture-level cities have been elevated as separately-planned cities, while ceding fiscal authority over empowered counties. On balance, this gradual reconfiguration of inter-governmental relations has eroded the capacity of certain prefecture-level units to realize urbanization goals set by the central government.

Contemporary implications of institutional adaptation
When directed improvisation is the underlying mechanism for institutional adaptation, the resulting equilibrium may be unstable. This is because actors who were incentivized at the outset of institutional distortions might change their minds, ask for more benefits or refuse to cooperate later on. Yet this process is not readily observable. As Mahoney and Thelen point out, "Gradual or piecemeal institutional changes often only register as change over a longer time frame." 24 For example, the US budget reforms of 1974 resulted from an extended process of institutional layering, meaning the superimposition of new arrangements on top of pre-existing structures designed to serve different purposes. 25 Eric Schickler argues that institutional layering reflected battles in various periods, starting from 1890, and finds that the last dynamic emerged nearly a century later, between 1970 and 1989. 26 In another context, Dan Slater contends that the political balancing that allowed Suharto to concentrate autocratic power in Indonesia between the 1940s and 1990s eventually paved the way for regime change. 27 Over time, the coercive capacity of co-opted institutions and organizations was undermined. 28 Joy Langston similarly attributes the effects of dedazo resistance in Mexico during 1952 to 1987 to its collapse in the 1990s. 29 Such historically grounded studies have contributed to concept formation in comparative institutional analysis and can be harnessed for explaining contemporary politics. Indeed, the concept of layering is especially well suited for analysing political dynamics in China's complex bureaucratic and administrative context. 30 Although scholars often discuss central-local relations in dichotomous terms, China's administrative structure comprises five official tiers, ranging from the central government to townships. 31 Lower-ranked governments are expected to accept institutional adjustments introduced by higher-level ones, yet their 24 Mahoney and Thelen 2010, 2. 25 Schickler 2001, 191. 26 Ibid., 246, 252-53. 27 Slater 2010. 28 Mahoney and Thelen 2010, 9. 29 Langston 2006. 30 Tsai 2006. 31 Chung 2010 compliance may be reluctant and fragile. Local governmental incentives may change, leading to vertical inter-governmental bargaining and even overt competition. As Lewis and Steinmo explain, "From an ontological standpoint, evolutionary systems are never at equilibrium." 32 The current study demonstrates how institutional layering has gradually altered China's multitiered administrative hierarchy since the 1990s and created developmental constraints for prefecture-level cities.

Institutional Evolution in China's Early Reform Era
Privileging regions by granting dual-reporting linkages China's reform process has been marked by a drastic downward delegation of central authority and the active promotion of local initiatives. 33 Decentralization of fiscal and administrative powers to lower levels of government constituted a defining feature of post-Mao economic development. 34 Concurrently, however, the cadre management system remained centralized, which enabled upper levels of the party-state to implement high priority policies. 35 China's political economy has been described as "regionally decentralized authoritarianism," whereby the party-state exercises layer-by-layer personnel control, with key positions at every level of government appointed by the level immediately above it. 36 In this hierarchical arrangement, subordinate governments and their appointed officials are directly accountable to superiors at the next level up (Figure 1). 37 Building on Ang's notion of directed improvisation, such a structure allows for what we refer to as the "improvisational autonomy" of each administrative entity. When a city government exercises full authority over personnel in its jurisdiction, there is scope for local creativity and the distribution of resources to achieve developmental goals. 38 This administrative design was intended to tighten the accountability of all sub-national governments one level up; however, it also enabled higher-level governments to pillage the resources of their immediate subordinate governments. Some studies have observed distrust between prefecture-level cities and provinces, including that between Hubei and Wuhan during the 1980s. 39 Others have pointed out how prefecture-level cities sacrificed counties' benefits, or diverted their revenues to urban areas, leaving counties fiscally strapped. 40 Despite 32 Lewis and Steinmo 2012, 320. 33 Chung 2010. 34 Oi 1999. 35 Edin 2003Huang 1999, 92. 36 Xu 2011. 37 Tsui and Wang 2004 We acknowledge that local administrative autonomy could also provide cover for rent-seeking activities.
As Ang (2020) points out, however, certain forms of corruption (in the form of "access money") can be growth promoting rather than growth inhibiting. 39 Solinger 1996;Yu 2014. 40 Schroeder 1992Chien 2010;Zhang 2009. these tensions, it was not realistic to eliminate prefecture-level units, which have existed since the 1950s, let alone provincial governments, which have been part of China's administrative hierarchy for over 2,000 years. 41 This was the context in which a new pattern that enabled certain administrative entities to directly interact with units two levels higher emerged.
During the 1980s, the central government initiated dual-reporting relations to alleviate the exploitation of select administrative units and to promote their development. 42 As shown in Figure 2, separately-planned cities were granted direct economic linkages with the central government, bypassing their corresponding provincial governments. 43 More specifically, separatelyplanned cities were empowered with provincial-level authority in certain areas of economic management, and its administrative rank rose above that of regular prefecture-level cities. Building on indicators proposed by Kyle Jaros to measure the policymaking power of China's local governments, Table 1 summarizes the privileges of separately-planned cities compared with those of regular prefecture-level cities. 44 Mimicking this strategy, during the 1990s provinces started empowering economically strong counties by authorizing them to negotiate directly with provincial governments, sidestepping their superior prefecture-level cities. 45 Taken together, the two sequential policies of creating separately-planned cities and empowering economically strong counties circumvented the original structure of level-by-level administrative authority. Moreover, both initiatives introduced similar upgrades in administrative privileges. In terms of rank, the Party secretaries of empowered counties are listed within the prefectural standing committee, thereby signalling a higher status than that of ordinary counties. Fiscally, empowered counties bypass prefecture-level cities by negotiating and splitting revenues directly with provinces. These reconfigurations of inter-governmental relations constitute institutional layering such that new institutions were added to pre-existing structures because the latter were not readily dismantled. 46 As a result, the original vertical structure of directed improvisation has been compromised. The improvisational autonomy of both provinces and prefecture-level cities has become unsettled by the addition of administrative relations between the central government and separately-planned cities, and between provincial governments and empowered counties, respectively. 47 The designation of "separately-planned cities" Logically, the creation of dual-reporting relations would not be welcomed by governments whose authority over subordinate entities is diminished. During the 1980s, friction emerged between provincial capital cities and provinces, especially when provincial capitals were granted separately-planned city status, thereby enabling capital cities to circumvent their provincial superiors. 48 For example, rumours circulated that Wuhan would be granted separately-planned or even centrally-administered status, which would have cut its fiscal, if not administrative, ties with Hubei province. In response, Hubei province prioritized other cities and was slow to approve Wuhan's proposed infrastructure projects. 49 The rationale for this reluctance is straightforward: provincial authorities sought to rein in separately-planned cities under their jurisdiction, while separately-planned cities tried to operate more independently from provincial governments. 50

Notes:
The thicker arrow lines pointing up denote the emergence of extra administrative reporting relations; centrally-administered municipalities are excluded because they do not have dual-reporting relations; administrative entities at the same horizontal level are at the same administrative rank. * includes four autonomous regions; number of counties includes both empowered and regular counties because official statistics do not indicate the number of "empowered counties." 48 Solinger 1996;Chung 2010;Yu 2014. 49 Schroeder 1992Solinger 1996. 50 Chung 20102016, 41.  Such strains extended beyond Hubei and Wuhan. Tensions similarly arose between other separately-planned cities/provincial capitals and their provincial governments. This strain in intergovernmental relations led the State Council in 1989 to order that provinces improve their relations with provincial capitals that had been elevated to separately-planned cities. 51 Provinces then initiated various negotiations with the central government to salvage their own standing with these separately-planned cities. 52 As a result, even though Wuhan gained separately-planned city status in 1984, the city, along with seven other provincial capitals, lost this status in 1993. By 1997, five non-provincial capital cities had managed to retain their separately-planned city status because their provinces regarded them as causing fewer negative effects. The number of separatelyplanned cities remains unchanged to date. 53 Nonetheless, conflict continues between these cities and provinces. 54 The National Academy of Governance even recommended the removal of Ningbo's 宁波 separately planned status in order to foster smoother relations among the central, provincial and city governments. 55 With regards to establishing separately-planned cities, institutional adaptation emerged among the central government, provincial authorities and separatelyplanned cities. At the outset of this process, Beijing introduced the designation of separately-planned cities and other select cities to serve as economic magnets. In so doing, the centre built regional economies that leveraged their areas of comparative advantage and promoted national economic growth during the 1980s. 56 Over time, the institutional adaptation of creating separately-planned cities undermined the improvisational autonomy of provincial authorities, especially since the Central Organizational Department controls the appointment of Party secretaries and mayors in separately-planned cities. 57 Shortly thereafter, provincial anxiety that their capitals would become "centrally-administered municipalities" proliferated, resulting in several rounds of central-provincial bargaining. Provincial governments tried to retain direct control over their provincial capitals (with separately-planned status), while the central government attempted to maintain the privileged status of these provincial capitals. 58

Adaptive empowerment of counties during the 1990s and the 2000s
In a parallel logic, the empowerment of strong counties allowed these counties to build dual-reporting relationships with both prefecture-level cities and provinces 51 Yu 2014. 52 Schroeder 1992Solinger 1996. 53 Yu 2014 Interviews with prefecture-level officials, academic experts and local intellectuals, Ningbo, 19 September 2017. 55 National Academy of Governance 2015. 56 Schroeder 1992;Landry 2008, 63;Chung 2010. 57 Ibid. 58 Chung 2010 ( Figure 2). In this instance, provincial governments initiated institutional adaptation, whereas prefecture-level city governments tried to preserve the original administrative hierarchy. Following the objections of provinces to separatelyplanned cities, one might assume that prefecture-level governments would in turn oppose the empowerment of counties. This has not been the case, however. Unlike the stagnation in creating separately-planned cities since 1997, a growing number of counties have been granted direct linkages with provinces since 1992. 59 Moreover, since the practice of empowering economically strong counties commenced in Zhejiang, this pattern of bolstering the authority of economically strong counties has spread to other provinces. 60 Indeed, beginning in the 1990s, relations between prefecture-level governments and empowered counties became less contentious. 61 During the 1990s, prioritization of economic growth stimulated horizontal interregional competition among cadres. 62 Leaders of prefecture-level cities cared more about aggregate GDP figures, irrespective of whether growth was generated from counties or urban areas. 63 Suzhou's 苏州 developmental profile exemplifies this situation. Even though Suzhou has been noted for having a "weaker urban core compared with surrounding counties," its overall prosperity has made it a "cradle for provincial governors" (shengzhang yaolan 省长摇篮), a reference to the fasttrack promotion record of Suzhou's officials. 64 Prefecture-level city leaders accepted their diminished authority over counties because their economic performance improved the statistical indicators of prefecture-level cities. 65 Although GDP incentives could plausibly motivate provincial leaders to accept separately-planned cities, provinces worry that such cities may be further elevated and re-titled as centrally-administered municipalities. 66 Unlike the circumstance of empowering economically strong counties, where prefecture-level cities still enjoy the statistical contribution of counties, the promotion of separatelyplanned cities into centrally-administered municipalities eliminates their statistical contribution to provinces. As noted above, after the administrative concept of separately-planned cities was proposed in the 1980s, Hubei feared that Wuhan might become a centrally-administered municipality. 67 In 1997, fuelling provincial unease about the potential fate of separately-planned cities, the central government upgraded Chongqing from a separately-planned city to a centrally-administered municipality. 59  Building on "Dual-reporting Relations": Growing Provincial Authority Following Zhejiang's empowerment of counties in 1992, the expansion of powers to select economically strong counties has become common practice in various provinces. 68 The nationwide adaptation of empowering economically strong counties was built upon the realization that such linkages facilitated China's economic development. 69 Intensified competition among counties and substantial decision-making power on the part of county-level governments were the chief drivers of rapid growth during the 1990s. 70 Subsequently, the policy of empowering economically strong counties appeared in various central-level official documents in the 2000s, marking a growing shift in authority over counties from prefecture-level cities to provinces (Table 2).
Unlike the situation for separately-planned cities, whose leaders are directly appointed by the central government, the process of empowering economically strong counties did not affect the appointment of county leaders by prefecturelevel cities. However, in light of central guidelines to build provincial-county relations, exceptions in the appointment system have started to occur. 71 As the pioneer of empowering economically strong counties, Zhejiang was the first province to directly manage personnel within its counties, 72 about one decade before the central government proposed that counties be administered directly by provinces. 73 Our research reveals that the pattern of direct prefectural appointment of county/county-level city leaders is likely limited to the early reform era. Continuing expansion of the province's personnel reach is evident. In 2015, Guangdong's provincial organization department announced the appointments of Party secretaries in Shenzhen's 深圳 urban districts, indicating the provincial government's direct control over the appointment of urban district-level officials. 74 The next year, Zhejiang's organization department announced the appointment of 115 officials, including Party secretaries and mayors of counties, county-level cities, and even urban districts under prefecture-level cities. 75 A 68 Chien 2010;Cartier 2016. 69 Interviews with officials in Zhejiang DRC and organizational departments of Zhejiang and Ningbo, 9 May 2018; 29 October 2018; 8 April 2019. 70 Cheung 2008, 61, 68;Oi 1999. 71 As Bulman and Jaros (2019) note, "concurrent leadership appointment," whereby a local leader is simultaneously assigned to a leadership position in a higher-level administrative entity, also mediates China's hierarchical authority structure. However, such practices have never been regularized and rarely change the formal personnel hierarchy. For example, Party secretaries in Suzhou have had concurrent appointments as provincial standing committee members since 1994, yet Suzhou Party secretaries are still appointed by the provincial organization department rather than the central counterpart. In other words, concurrent leadership appointment did not change the provincial authority over the appointment of Suzhou's Party secretaries. Field interviews confirm this also applies to the concurrent appointment of county leaders. similar deepening in provincial authority over personnel has occurred in Jiangsu. In 2016, the Taizhou 泰州 prefecture-level city government was still appointing leaders in its urban districts, although recent prefectural documents reveal that Jiangsu province now controls those same positions. 76 Zhejiang, Guangdong and Jiangsu's provincial activism demonstrates their personnel control over all administrative units that are two levels below the provincial level, including urban districts. Although counties/county-level cities and urban districts are apparently at the same administrative level, they should be distinguished from one another because empowering strong counties only builds direct linkages between counties and provinces, while urban districts are fully subordinate to prefecture-level cities according to formal administrative rules. 77 Therefore, extending personnel reach to counties makes procedural sense for provinces when there are dual-reporting relations between counties and provinces; however, provincial reach to urban districts bypasses another level of government because districts only report to prefecture-level cities. 78 All provinces maintain a document entitled, "Publicity before official appointments" (guanyuan renqian gongshi mingdan 官员任前公示名单), which indicates the personnel authority of appointed officials. 79 If the appointment of specific officials is publicized by the provincial or prefecture-level organizational department, then the provincial or prefecture-level city government exerts personnel authority over these officials. We collected public announcements of official appointments from all the counties, county-level cities and urban districts of prefecture-level cities in each province and compiled a database indicating the personnel reach of provincial governments. 80 Table 3 shows major departures from the formal institutional mandate of one-level-down appointment. Although the central government still appoints provincial officials, the reach of provinces in personnel appointments varies. Only four provinces maintain the one-level-down appointment system, allowing prefectural leaders to appoint county and district leaders. Ten provinces, including Zhejiang, have taken over the authority for appointing both Party secretaries and mayors of districts. Our finding that 14 out of 27 provinces have already extended their reach to the appointment of urban district cadres demonstrates the expanding authority of provincial governments. Figure 3 maps the growth in the number of provinces appointing cadres two levels down.

The Emergence of Unintended Institutional Effects
The previous two sections delineated how institutional layering and directed improvisation have incrementally changed China's multilayered administrative hierarchy ( Figure 4). Ostensibly, growing provincial authority strengthens the autonomy of counties while compromising prefecture-level cities' authority over their entire jurisdiction. 81 Given the increasing prevalence of direct provincial-county relations, since 2005 domestic scholars have deliberated about eliminating the prefecture level for better development of county-level units. 82 The erosion of prefecture-level cities' authority is detailed in this section.

Why coordinating resource allocation at the prefecture level matters
Although counties were key drivers of China's economic prosperity during the 1990s, 83 prefecture-level cities have become increasingly important since the 79 Interview with academic experts; interview with prefecture-level officials, Ningbo, 24 March 2017; 14 July 2017; 11 May 2018; 2 April 2019. 80 Centrally-administered municipalities are excluded. Owing to data limitations, it is unrealistic to access the public announcements of counties/county-level cities/districts for all prefecture-level cities. Our sample covers all the cities with special administrative status and significance within each province (e.g. deputy provincial-level cities, including separately-planned cities and provincial capital cities without deputy-provincial status). When it comes to regular prefecture-level cities, we chose at least one regular prefecture-level city within each province with the assumption that provincial governments treat regular prefecture-level cities fairly from a personnel perspective. Interviews with officials from the organizational departments of Zhejiang and Ningbo, 5 February 2016;24 March 2017;29 October 2018. 81 Landry 2008Chung 2016, 52. 82 Lu and Tsai 2019. 83 Cheung 2008 2010s. In 2012, Premier Li Keqiang 李克强 underscored the developmental significance of prefecture-level cities for enhancing urbanization in China's future. 84 The centre's shift in urbanization metrics indicates the growing strategic position of prefecture-level cities, as seen in the 2010 "Regional plan of the Yangtze River Delta city cluster," which highlighted regional core cities. 85 After the 19th Party Congress in 2017, the central government again called for granting prefecturelevel governments more economic management authority. 86 We recognize that smaller urban centres also appear in the central government's urban strategy. To resonate with the State Council's urbanization plan (2014-2020), the National Development and Reform Commission's (NDRC hereafter) 2020 policy document expressed support for small-and medium-sized cities to play a greater role in China's new urbanization strategy. But the development of large prefecture-level cities is equally if not more critical from the centre's perspective. Indeed, the NDRC specifies "enhancing the resource allocation capacity of central prefecture-level cities" as a fundamental requirement (zongti yaoqiu 总体要求) for China's urbanization process. 87 Rather than targeting all prefecture-level cities, this is aimed at the central prefecture-level cities that, according to Beijing's intentions, were expected to manage resources throughout their entire jurisdictions but failed to do so.
To amplify the point, the NDRC's 2020 document includes nine guidelines for optimizing the spatial structure of urbanization. Only one of these nine guidelines mentions counties in China's new urbanization construction, while five specify  Drafted by authors based on Table 3.

Notes:
Provinces that reached the personnel appointment of urban districts are shaded. Field interviews indicate that Zhejiang was the only province exercising two-levels-down appointment as of 2008.

Source:
Drafted by authors. central support for prefecture-level cities. These measures include promoting key urban clusters and urban metropolitan areas; promoting central prefecture-level cities such as provincial capitals, separately-planned cities as well as other major prefecture-level cities; supporting prefecture-level cities to better allocate resources by converting counties into urban districts, and so on. 88 This repeated focus on prefecture-level cities indicates their perceived centrality to China's urbanization strategy. From a policy perspective, one could question the need for prefecture-level cities to coordinate resource allocation at the city level. Perhaps it would be more economically rational for prefecture-level cities to grant more autonomy to counties in the spirit of directed improvisation. However, improvisational autonomy requires that prefecture-level governments, rather than provincial authorities two layers removed, control the personnel appointment of county leaders. If county officials are still appointed provincially, then granting county-level governments more authority would exacerbate prefecture-county tensions. As we have pointed out elsewhere, maintaining administrative barriers leads prefecturelevel cities to compete with their subordinate counties. Such competition impedes the urbanization of both and, by extension, the developmental goal of urbanrural integration. 89 In short, dual-reporting represents a major obstacle in China's urbanization plan because it impedes the ability of prefecture-level cities to manage resources under its territorial jurisdiction. 90 Optimizing China's administrative structure by eliminating dual-reporting relations thus constituted a key message in President Hu Jintao's 胡锦涛 report, "Deepening reform and opening up in a comprehensive manner," presented at the 18th Party Congress in 2012. 91 This goal has remained under Xi Jinping's 习近平 leadership, as evidenced by the establishment of a leading small group devoted to deepening administrative reforms.
Given that two generations of leadership have flagged the developmental constraints and "irrationality" posed by dual-reporting, why does it persist? The challenge, ironically, lies in the centre's countervailing measures aimed at strengthening provincial authority over localities. 92 For example, in 2017 the NDRC assigned 90 per cent of the approval rights over publicly financed transportation projects to provinces. 93 Further empowerment occurred in 2020 when the central government granted provincial governments one of the most critical areas of approval authority: the power to approve transfer of agricultural land to construction use within the entire province. 94  counties with dual-reporting to prefecture-level cities and provinces reflects an internal inconsistency between the publicly stated importance of rationalizing China's administrative system, and the central government's desire to strengthen provincial authority in a centralizing spirit. Dual-reporting cannot be eliminated by central fiat if the centre itself is simultaneously empowering provinces in other ways. The twin objectives of administrative clarity and centralization of authority to the provincial level are working at cross purposes.

The source of fiscal conflicts
As discussed, the central government's initial emphasis on aggregate GDP figures motivated prefecture-level city leaders to accept their diminishing authority over counties. By the mid-2000s, however, China's shift to adopting a "scientific outlook on development" (kexue fazhan guan 科学发展观) led to recommendations that reckless local competition over GDP figures should be complemented by additional developmental priorities. The emphasis on prioritizing multidimensional targets in the "new normal" era encourages coordinated investment of resources, rather than merely focusing on GDP. 95 As interjurisdictional competition among counties/districts hampers the prefecture-level cities' capacity to provide public services, 96 they seek new growth engines to transition towards an urban economy which coordinates resources throughout their entire jurisdiction. The rationale is to reduce duplication in local services and promote implementation of development projects proposed by higher levels of government. Prefecture-level officials explained: Based on current economic conditions, multi-dimensional development targets call for resource coordination and allocation. What differentiates first-tier cities from others is not solely their economic scale. The fundamental difference is their capacity to allocate resources, especially fiscal resources, throughout the entire jurisdiction, which is the ultimate target of prefecture-level cities. 97 It can be misleading to look at aggregate levels of fiscal revenues without distinguishing their source and whether the prefecture-level city exerts authority in mobilizing them. As an example, Suzhou's fiscal revenue ( yiban yusuan shouru 一般预算收入) reached 173 billion yuan in 2016, which exceeded that of Guangzhou at 139.4 billion. 98 Yet it would not be accurate to conclude that Suzhou's prefecture-level city government is in a stronger fiscal position to execute its developmental plans. Guangzhou administers urban districts under its prefecture-level government. By contrast, Suzhou's urban districts generated only 92 billion in revenuesand nearly half of its fiscal revenue came from four strong counties that are fiscally linked with Jiangsu province rather than 95 Ma 2015. 96 Li, Huiping, Wang andZheng 2017. 97 Interview with prefecture-level officials, Ningbo, 12 February 2018. 98 This is counterintuitive given that Suzhou's population is smaller than Guangzhou's population (10.7 versus 13.5 million).
Suzhou. 99 Operationally, Guangzhou's prefecture-level government has access to more fiscal resources than Suzhou's prefecture-level government. 100 This is why prefecture-level cities such as Suzhou are no longer motivated solely by the statistical contribution of counties but instead seek to restore control over them. Provinces and prefecture-level cities now have competing interests over counties. 101

How provincial governments pick places
Given prefecture-level cities' interest in allocating resources throughout their entire jurisdictions, we argue that provincial governments derive more benefits from supporting provincial capitals than regular prefecture-level cities at a similar economic scale owing to the legacy of weaker counties surrounding provincial capitals. Indeed, provinces have been urging their provincial capitals to apply for the designation of "national-level core city" (guojia zhongxin chengshi 国家中心城 市), which was introduced to denote urban areas with the greatest strategic significance in China. In 2010, the Ministry of Housing and Urban-Rural Development proposed five cities -Beijing, Shanghai, Guangzhou, Tianjin and Chongqingas national-level core cities. 102 In 2016, Chengdu was supported by the State Council to be the sixth national-level core city. 103 Chengdu's success in raising its strategic importance has inspired competition among cities at similar levels to join the group of national-level core cities. 104 Provincial capitals are not necessarily cities with the highest political status or largest economies in a particular province, but with weaker counties, they pose less competition to provincial governments than non-capitals.
This provincial preference for promoting their capital cities is a relatively recent phenomenon. Earlier on, conflict surfaced between provincial capitals and their provinces when capitals were granted direct fiscal linkages with the central government and sidestepped provincial governments. Since 1997, the situation has stabilized such that only five non-provincial capitals have pdf. Although being a national-level core city does not immediately entail specific preferential policies, the designation provides prefecture-level cities with the prospect of being treated equally as centrally-administered municipalities in the future. 104 Lin 2018;Liu 2017; "Gudu Nanjing yao zhengzuo 'guojia zhongxin chengshi' guihua yi xianxing, quanguo zhiyou 10 ge ming'e" (Ancient city of Nanjing is applying for "national-level core city" separately-planned status. The more recent lure of elevating provincial capitals into "national-level core city" status would not affect provincial access to the fiscal revenues of its capital city. The overall logic has remained consistent: provincial governments opportunistically support administrative strategies that enhance their fiscal position. The unintended consequences of growing provincial authority on a subset of prefecture-level cities can be seen in the following comparison of Hangzhou with Ningbo.

Comparing Hangzhou and Ningbo
Hangzhou and Ningbo are at the same administrative rank and the central government has named both as prefecture-level cities that should enhance their capacity to allocate resources at the city level. 105 Their main difference is that Ningbo is a non-provincial capital affected by the institutional adaptation of empowering economically strong counties. Out of China's five separatelyplanned cities, Ningbo relies most heavily on the economic development of its counties. 106 The following comparison shows how dual-reporting relations became problematic for Ningbo's developmental efforts. 107 During the early reform era, the central government regarded Ningbo as having better natural endowments. As a coastal city with a world-class port, in 1984 it was designated one of the 14 "coastal open cities" ( yanhai kaifang chengshi 沿 海开放城市). Ningbo's geographical advantage supported the rise of its export processing industry, which was concentrated in rural counties with cheaper land. The proportion of Ningbo's GDP as a percentage of Hangzhou's climbed from 68.3 per cent in 1981 to 76.5 per cent in 1984 ( Figure 5). Ningbo's county economy further thrived after Deng Xiaoping's 邓小平 1992 southern tour. Ningbo's GDP reached over 80 per cent of Hangzhou's GDP in 1996 and maintained that proportion for the next 18 years. During the early 2000s, economic success earned Ningbo the reputation of being one of Zhejiang's "twin cities," along with the provincial capital of Hangzhou. By the end of 2010, Ningbo's GDP had reached 87 per cent of Hangzhou's GDP. 108 A senior provincial official in Zhejiang revealed that 80 per cent has long served as a benchmark indicator for measuring Ningbo's relative economic performance vis-à-vis Hangzhou: "If the share surpasses 80 per cent in a particular year, then Ningbo has fulfilled its growth target, and vice versa." 109 However, by 105 NDRC 2020. 106 Interview with prefecture-level officials, Ningbo, 25 January 2017 There are four other separately-planned cities that are potentially comparable with relevant provincial capitals. However, our focus is on how empowering economically strong counties became developmental barriers. Specifically, as the provincial capital, Hangzhou's urban core is stronger than that of Ningbo. Provincial capital cities' urban area has been supported by provincial institutions and enterprises, and heavy reliance on county economies typically occurs in non-provincial capital cities such as Ningbo, Wuxi 无锡, Foshan 佛山 and Suzhou. To be sure, as the site of numerous universities including Zhejiang University, Hangzhou attracts more talent than Ningbo, which has enabled the IT industry to thrive in Hangzhou. Yet Ningbo hosts a major port that supports manufacturing in counties. 112 Hangzhou and Ningbo Statistical Yearbooks 2019.

Note:
The dash line represents the 80% benchmark set by Ningbo's government to measure Ningbo's relative economic performance vis-à-vis Hangzhou.
Given the personnel arrangements and political authority granted by provincial governments, counties normally have bargaining power vis-à-vis prefecturelevel city governments. Hangzhou's strong urban core clearly contributes to the prefecture-level city's ability to mobilize resources throughout its entire jurisdiction. 113 The heavy reliance of non-provincial capital cities on counties can be seen by the geographical location of China's top 50 "strongest counties." 114 In 2020, 46 out of China's top 50 "strongest counties" were located in non-provincial capital cities. 115 The clustering of strong counties in non-provincial capitals reveals two distinct development patterns such that provincial capitals derive economic strength from their urban cores while non-provincial capitals rely on strong counties. Table 4 shows the sharp contrast between Ningbo and Hangzhou's relative reliance on counties. Following major administrative redistricting in both cities in 2002, Hangzhou's urban area accounted for 76.7 per cent of its total GDP compared with 57.2 per cent in Ningbo. The-then economically strongest county Fuyang 富阳 accounted for only 7 per cent of Hangzhou's GDP, while Cixi 慈溪 and Yuyao 余姚 accounted for 12.4 per cent and 13.8 per cent of Ningbo's GDP, respectively. By 2017, Hangzhou managed to expand its portion of urban GDP to 92 per cent, compared with Ningbo's 63.8 per cent. Meanwhile, Ningbo's strongest counties, Cixi and Yuyao, still maintain their status, accounting for 15.6 per cent and 10.2 per cent of Ningbo's GDP. Officials from the Ningbo Statistical Bureau summarized the essence of the dilemma: Ningbo's economic figures have benefited from empowered counties, but at the same time, strong counties constrain Ningbo prefecture-level cities' ability to coordinate the resources throughout the entire jurisdiction. Without coordination, strong counties only contribute to Ningbo's economy from a statistical perspective. 116 The legacy of reliance on counties also limits the ability of prefecture-level cities to mobilize land revenue. Counties in Huzhou 湖州 and Ningbo, for example, remit a much smaller share of land revenue to prefecture-level governments than counties in Hangzhou. 117 This trajectory reflects how the relative bargaining power of counties vis-à-vis prefecture-level cities limits the latter's fiscal capacity. Ang argues that a hierarchical chain of personnel appointment secures the relative autonomy of each administrative unit, as a specific unit's immediate superior is always able to "direct" its resource allocation. 118 In other words, personnel control constitutes the primary source of leverage that a higher-level authority possesses over its immediate subordinate in the one-level-down system. The revival of provincial personnel reach to counties/urban districts thus significantly  undermines the ability of prefecture-level cities to mobilize and redeploy the resources of territorially subordinate areas. Comparing Ningbo and Hangzhou illustrates the logic of how the institutional adaptation of creating province-county linkages inadvertently evolved into a hindrance for further development. If the one-level-down principle were maintained, then prefecture-level cities could allocate resources in urban districts as well as counties. However, the revival of personnel appointments two levels down impedes prefecture-level city governments from exercising such authority over counties. With weaker counties, provincial capitals tend to be selected to represent the highest development level of provinces, as seen by the rapid urban expansion of provincial capitals such as Wuhan, Nanjing, Guangzhou and Hangzhou. This finding affirms the argument that China's metropolitan-oriented development has been driven in large part by provincial governments. 119

Conclusion
Privileging certain administrative units by introducing dual-reporting relations was a popular yet underexplored expression of an adaptive institution during China's early reform era. Our study identifies the circumvention of "directed improvisation" as an intermediate mechanism in this process, which reflects institutional adaptation among state actors. Following the central initiative of creating separately-planned cities in the 1980s, the province-county linkage was motivated by the GDP incentives facing prefectural leaders in the 1990s and adopted by the central government in the 2000s. This process has generated profound changes on China's intergovernmental relations, marked by the revival of personnel control two levels down. Although provincial-county linkages performed well economically during the early reform era, this structure became a developmental impediment when prefecture-level cities shifted to seeking new growth engines by coordinating resources throughout their entire jurisdictions rather than focusing merely on GDP figures. It is more challenging for prefecture-level cities to secure the compliance of counties when they report to more than one administrative level. Compared with provincial capitals that are moving towards better resource allocation, non-provincial capital cities with a legacy of stronger counties are constrained by provincial governments, even when they have achieved a similar or even higher economic scale than capitals. These conflicts are obscured if one views China's vertical administration from a static perspective.
Considerable work on China's vertical inter-governmental relations has depicted its formal administrative structures as operating in an explicit layer-by-layer manner. Yet we observe that this structure is merely a snapshot taken during part of a longer sequence of institutional evolution. Building on recent research on China's mounting provincial authority and provincial 119 Jaros 2016;2019. influence in forging developmental patterns of prefecture-level cities, 120 this article attributes China's inter-governmental dynamics to institutional layering and the ambiguity of implementing dual-reporting relations. In so doing, our study focuses on the continuous process of institutional development rather than those processes associated with leadership succession and retirement norms. The precise scope of dual-reporting relations in China's multitiered governing structure has yet to be specified constitutionally, as vertical administrative units compete for resources. Decades long in the making and sustained by incremental momentum, these institutional adaptations are not readily re-calibrated. In the absence of major reforms, provincial governments are likely to continue strengthening their authority, 121 and prefecture-level cities will encounter additional administrative hurdles as they seek to grow their economies.
Going forward, how the central government manages centrifugal forces under growing provincial authority merits further exploration. Andrew Mertha has observed the advent of "soft centralization," whereby a growing number of regulatory bureaucracies are being consolidated up to the provincial level. 122 This has eroded the authority of sub-provincial governments over the economy within their jurisdictions, as provinces are becoming a centralizing node in multiple spheres of regulatory governance. Finally, it is worth reflecting on why the central government allowed provincial governments to strengthen their authority by extending their downward reach over personnel appointments. One possibility is that the centre limits provincial authority over the structure of territorial governance by retaining ultimate authority for approving administrative reorganization. The extent to which this serves the intentions of the present leadership's centralizing efforts in practice warrants additional research.