Local government innovation diffusion in China: an event history analysis of a performance-based reform programme

Abstract

The performance-based reform programme launched by Fujian province in 2000 has been adopted by many other Chinese provinces, including Zhejiang, Hebei, Anhui and Sichuan, over the past 12 years. This article aims to explore the antecedents of the adoption of this programme, in particular, the effects of senior figures' political promotion incentives and diffusion mechanisms. Specifically, event history analysis based on probit regression is used to examine data from 31 Chinese provinces for the 2000–2012 period. The results show that leaders' relative age and chances of being appointed to the Politburo, and distance to the general election, are significantly negatively correlated with the reform programme's adoption, but top-down diffusion is significantly positively correlated with it.

Points for practitioners

This study confirms that the nomenklatura system in China shapes the diffusion of innovations through the mechanisms of political promotion incentives and intergovernmental interactions. Thus, the dynamics of innovation diffusion are, to some extent, rooted in particular political institutions and shaped by political contexts. Furthermore, the desire for political promotions may figure as a general deep reason for decisions about whether to adopt innovations; therefore, strengthening these incentives for adopting reforms becomes a key strategy.

Keywords

diffusion mechanisms performance management political promotion incentives reform in China

Introduction

Recently, many countries have adopted innovative reforms to improve government performance (Borins, 2001; Brinkerhoff and Wetterberg, 2013; Walker, 2007; Wu et al., 2013). These include the well-known New Public Management practices used in Western countries and China's performance-based reform programme, which was launched with a comprehensive framework of performance-improving measures in 2000. Wejnert (2002) noted that reforms and innovations only become fully effective, leading to social change, when they are widely adopted by local governments and departments. Hence, understanding innovation diffusion has begun to attract the attention of many scholars (Berry and Berry, 2007; Damanpour and Schneider, 2006; Kim, 2013; Shipan and Volden, 2008; Walker, 2007). Although the Chinese reform programme is part of a global public management movement, the determinants of its diffusion in China have not been investigated.

Previous studies have argued that innovation diffusion is driven by two clusters of factors: socioeconomic factors that are internal to particular jurisdictions as parts of local policy environments (Damanpour, 1991; Damanpour and Schneider, 2006; Lee et al., 2011; Ma, 2013; Moon and deLeon, 2001; Walker, 2007); and diffusion mechanisms through which innovation ideas cut across jurisdictional borders (Berry and Berry, 1990, 2007; Kim, 2013; Mooney, 2001; Shipan and Volden, 2008; Welch and Thompson, 1980). However, both strands of the literature conceptualise government organisations as the ‘actors’ that adopt policies; they do not examine how individual leaders make decisions about policy innovations in their organisations. This omission is an important gap in the understanding of innovation diffusion in China, which is a centralised country with a monopolistic nomenklatura system (Burns, 1987, 1994; Chan, 2004; Kung and Chen, 2011). This public personnel system divides public positions into different ranks (Edin, 2003; Manion, 1985) and provides different degrees of political promotion incentives for different provincial leaders at different times (Burns, 1989). Our knowledge is still very limited about how innovations spread in the state administrative system and what role potential political promotion incentives play.

Furthermore, previous studies of the mechanisms of diffusion have focused on the federal system in the US and have emphasised horizontal interactions, especially between neighbouring states (Berry, 1994; Berry and Berry, 1990; Mooney, 2001; Oakley, 2009; Rincke, 2007). However, the drivers of innovation diffusion may be very different in centralised countries, and especially in China, due to different and significant vertical influences. For example, Heilmann (2008) found that an experimentation-based policy process, with a pattern of central–local interaction, was effective in Chinese reforms. Ma (2013) concluded that regional diffusion effects and upper-tier pressure were positively and significantly associated with the adoption of police microblogging, with upper-tier pressure having a greater contribution. Zhu (2014) used a case study to define ‘mandatory policy diffusion’, which is diffusion facilitated by administrative commands. However, vertical intergovernmental interactions are still under-researched in China and vertical diffusion effects (bottom-up and top-down) need to be further tested and explored empirically.

In this study, we focus on a particular performance-based reform in China and explore how political promotion incentives and diffusion mechanisms shaped its adoption by taking Chinese provincial governments as the basic unit of analysis. Specifically, the study addresses two key questions: ‘Does the Chinese nomenklatura system shape innovation diffusion through political promotion incentives?’ and ‘How have vertical and horizontal diffusion mechanisms collectively affected innovation adoptions?’ We first present an overview of the performance-based reform programme, its diffusion and the nomenklatura system; then, relevant hypotheses are formulated and the data sources, variables and models are described. Finally, we provide a statistical analysis and then discuss our major findings and implications.

The performance-based reform programme and its diffusion

In the 1990s, the Chinese public had limited access to public services, and officials treated requests from the public in a dilatory manner. In April 1999, the government of Zhangzhou city addressed this issue by initiating a comprehensive performance-based reform programme, named Xiaoneng Jianshe (XNJS), which involved all of the city's township authorities and institutions with administrative functions. The programme attracted the attention of the Fujian Provincial Commission for Discipline Inspection and Supervision Department, which after evaluating the programme, recommended its comprehensive adoption by provincial government organisations. In March 2000, the Chinese Communist Party (CCP) Fujian Committee led by Chen Mingyi and the Fujian Provincial Government lead by Xi Jinping officially issued the document Decision to Carry Out XNJS, marking the formal initiation of the reform at the provincial level.

Following this decision, Fujian introduced a series of measures to implement XNJS. First, Fujian established a group to lead the reform and an office for XNJS, with Xi serving as the team leader. Moreover, a centre for efficiency complaints was established in Fujian; the Centre issued The Service Regulations of this Centre and opened a 24-hour complaint hotline. In addition, Fujian developed eight guidelines for government personnel, such as: having clearly defined responsibilities for each job; explaining all the procedures and required materials to the customers one-time; having the first consulted official take full responsibility for answering questions and guiding the customers; processing applications within a specified time; having another official in a similar position perform the responsibilities of an absent official rather than waiting for him to return; and so on.

Fujian also became the first province to streamline its administrative procedures and eliminate red tape. Nearly two thousand items needing examination and approval were cancelled and hundreds of administrative examination procedures were simplified. An online platform was set up where citizens can submit applications and officials can examine them online. Furthermore, Fujian issued the Interim Measurements to Open Public Affairs, which described in detail principles of information disclosure, issues and items to be disclosed, and forms of disclosures. Since then, policies, regulations, procedures and standards have been disclosed on the portal websites of the Fujian government and some other departments. Finally, Fujian has emphasised democratic supervision and performance management. It has appointed 9457 citizen supervisors and completed more than 3000 open and private investigations (ming cha an fang) since 2000. Fujian also issued the Interim Measures of Efficiency Warning of Government Personnel in Fujian and Appraisal Plan of Government Units' Efficiency, which tied the individual performance of officials to their career promotion. In recent years, a democratic appraisal website has been established where citizens can appraise the performance of governments through the website or text messages.

Overall, the XNJS performance-based reform programme aims to improve government performance by using results-based management, transparency and citizen participation, and reducing red tape. By the end of 2012, after 10 years of development, it had spread to 22 provinces (see Figure 1) and had attracted much attention from the national media, scholars and practitioners. In 2013, the standing committee of the People's Congress of Fujian province approved the first legal regulation for XNJS, which means that it is now institutionalised.

Figure 1.

Diffusion of XNJS among Chinese provinces.

The nomenklatura system in mainland China

China is a party-state regime with five administrative tiers (central, provincial, municipal, county and township) and characterised by a hierarchical control-and-command chain. The CCP committee and its standing committees in all tiers are the de facto centre of authority, whereas the government operates as the administrative organ (Ma, 2013). The provincial tier is comprised of four municipalities (Beijing, Shanghai, Chongqing and Tianjin), five autonomous regions (Inner Mongolia, Ningxia, Tibet, Guangxi and Xinjiang), two special administrative regions (Hong Kong and Macau), and 23 provinces (including Taiwan). Nearly 3000 counties and municipal districts fall within the jurisdiction of more than 280 middle-ranking prefecture-level cities. The country's township governments are its lowest administrative level, in charge of villages and communities, which have a high degree of autonomy.

China's centralised authoritarian regime is characterised by a monopolistic personnel management mechanism, the nomenklatura (Burns, 1987, 1994; Chan, 2004; Kung and Chen, 2011), which originated in the former Soviet Union and other Eastern Bloc communist countries (Manion, 1985; Liang and Langbein, 2015). The Department of Organisation of the CCP established and institutionalised China's contemporary cadre management system in 1979 (Chow, 1988, Liang and Langbein, 2015). The term ‘nomenklatura’ means ‘a list of positions, arranged in order of seniority, including a description of the duties of each office’ (Manion, 1985). It generally refers to the lists of offices (zhiwu mingchengbiao) controlled by various party committees, that is, those offices over which the party committees have authority (Chan, 2004).

Under the Chinese nomenklatura system, cadre positions are divided into different ranks; members of the CCP Politburo have the highest ranking. The appointment, promotion, transfer or removal of any important party or state leader requires the approval of the party committee controlling the nomenklatura under which that leader's office is listed (Manion, 1985). Subordinate leading bureaucrats (i.e. party secretaries and government chiefs) are hierarchically accountable to the level of government immediately above, which exercises nomenklatura authority and makes personnel decisions about leadership positions (Edin, 2003; Liang and Langbein, 2015). Officials' personal performance records are used by their superiors to make decisions about appointments, transfer, removal, promotions, demotions and merit-based pay of bureaucrats, especially those in leadership positions (Liang and Langbein, 2015).

Political ambitions are the paramount motivations of individuals in the bureaucracy governed by the nomenklatura system in China, and career-oriented leading cadres desire to advance up the ladder of power and authority in the system (Liang and Langbein, 2015). It follows that the administrative behaviour of lower-tier bureaucrats who wish to advance will reflect the policy preferences of their hierarchical superiors making nomenklatura personnel decisions (Liang and Langbein, 2015). Burns (1989) has illustrated this clearly:

In general, wealth, power and status in China go to those who can climb the ladder of official position. Because there is no real alternative to employment outside of the ‘mono-organisational’ cadre hierarchy, and because officials are almost completely dependent on their official sinecures for their livelihood, they need patrons at higher levels of the bureaucracy, particularly those who are well connected to party committees and core groups, to further their careers.

Hypotheses

Based on the Chinese nomenklatura system, hypotheses are therefore proposed from the perspectives of political promotion incentives and policy diffusion, respectively.

Political promotion incentives perspective

The effect of age on innovation diffusion is usually regarded as negative, as the cognitive resources needed to make decisions, such as learning, attribution and memory, decline with age (Bantel and Jackson, 1989) and younger managers are more adventurous (Hambrick and Mason, 1984; Huber et al., 1993). However, we argue that given the nature of political promotion in China, this may not be true in this system. Chinese provincial secretaries are generally more than 50 years old and 65 is the mandatory retirement age for top provincial leadership.¹ Age thus becomes important political capital, and differences in psycho-physiological competence (such as attribution, memory, etc.) appear to be smaller and less important. In fact, Li and Zhou (2005) found that older provincial secretaries are less likely to be promoted due to the disadvantage of their age. In addition to personal features, many other factors determine cadres’ promotion, especially performance (Li and Zhou, 2005). XNJS is a performance-based reform aimed at improving work efficiency, management effectiveness and social effects. Hence, we expect that older secretaries who are eager to increase their chances of promotion are more inclined to adopt performance-enhancing reforms during their limited career ‘life’, and thus the effect of the variable of relative age at the time of the next National Congress of the Communist Party of China (NCCCP)² will be significantly positive.

In addition, bureaucrats in different positions have different incentives under China's nomenklatura mechanism. Top-level bureaucrats, who are already at the pinnacle of power and authority, have little incentive to pursue actions that are tied to career prospects; the marginal rewards they received for political and economic achievement diminish and the time value of additional rewards is lower due to their greater age (Liang and Langbein, 2015). Research has found that lower-level, alternate members of the Central Committee of the CCP have much stronger incentives to pursue political radicalism than full members (full members have the second-highest political rank and Politburo members have the highest) (Kung and Chen, 2011). Provincial leaders who are Politburo members are expected to be less likely to adopt XNJS because they are at a higher level than those who are not.

The political environment is a key factor affecting government officials' desire to innovate, and the most important factor in the political environment is the election cycle (Berry and Berry, 1990), during which government officials generally have greater incentives to adopt innovative activities that have positive political effects (Kiewiet and McCubbins, 1985). Unlike the direct elections that take place in the US, local government officials in China are nominated by a higher level of government and appointed at the National People's Congress. It is thus crucial that they attract the attention of their superiors through their work (Chan, 2004). Ambitious officials are thus more likely to adopt XNJS, a performance-based reform programme, to improve government performance and impress their superiors in the years nearer to a general election.

Not only did cadres' differences in the position of authority lead to different estimates of political career prospects which can influence cadres' enthusiasm for pursuing policy agendas (Kung and Chen, 2011, Liang and Langbein, 2015), but leaders working in different provinces can also have distinct prospects of political promotion despite their similar positions. Under the Chinese nomenklatura system, the Politburo lies on the top (Burns, 1987, 1994; Chan, 2004; Kung and Chen, 2011). Though regional representation in the Central Committee of the CCP is becoming increasingly balanced, politicians from more powerful provinces such as Beijing and Shanghai are more likely to become Politburo members than those from other provinces (Liang and Langbein, 2015). Leaders with less chance of appointment to the Politburo may need to innovate more to increase their career prospects. Thus:

H₁: The relative age of a provincial secretary is negatively correlated with the adoption of XNJS.

H₂: A provincial secretary's position as a Politburo incumbent is negatively correlated with the adoption of XNJS.

H₃: Distance to the general election is negatively correlated with the adoption of XNJS.

H₄: Chance of appointment to the Politburo (political rank of provinces) is negatively correlated with the adoption of XNJS.

Policy diffusion perspective

Governmental interactions are an important focus of policy research (Berry and Berry, 1990, 2007). First, it suggests that there are strong and continuous interactions between governments at the same level, such as learning and competition (Berry and Berry, 2007; Christensen et al., 2008). Many studies in the US have shown that innovative activities of state governments and their departments will drive neighbouring states to adopt these activities (Berry, 1994; Berry and Berry, 1990; Mooney, 2001; Oakley, 2009; Rincke, 2007). Using these studies as a starting point, this study examines the effects of the interactions between adjacent provinces.

Furthermore, government units at a higher level can influence subordinate units in various direct ways, such as vertical mandates (Berry and Berry, 2007) and financial incentives for innovation adoption (Welch and Thompson, 1980). In addition, other more indirect signals may encourage innovation diffusion. For example, discussion and deliberation over a bill in Congress can influence policy adoption by state governments through information transfer (McCann et al., 2010), and a shift in national policy direction can send policy signals that drive policy diffusion (Kim, 2013). In the case of the XNJS reform, its sponsor, Xi, became a member of the Politburo standing committee of the CCP and was expected to be promoted to top leader of China in 2007. Figure 1 shows the marked increase in XNJS's adoption between 2007 and 2008, even though the State Council did not issue any mandatory documents or provide any incentive for its adoption. Xi's promotion may be considered a political signal under the Chinese nomenklatura system, and we argue that since 2007, provincial leaders have tended to advocate his innovation to enhance their own career prospects.³

Some politicians have called on scholars to pay more attention to bottom-up diffusion (Shipan and Volden, 2006). In China's centralised system, successful innovation practices implemented by local governments are usually considered to be a type of ‘policy experiment’ that is then adopted by governments at a higher level (Lee and Warner, 2004; Xue and Zhong, 2012). In terms of the performance-based reform programme, Fujian province adopted this practice after the positive experience of Zhangzhou city. Therefore;

H₅: The number of adjacent provinces that have adopted XNJS is positively correlated with its adoption.

H₆: Xi Jinping being a member of the Politburo standing committee of the CCP is positively correlated with the adoption of XNJS.

H₇: The number of governed prefecture-level cities that have adopted XNJS is positively correlated with its adoption.

Methods

Data sources and measurements

Dependent variable

The data for the dependent variable are obtained from Internet searches for related government documents, meeting information and news reports. The keywords in the search are ‘Xiaoneng Jianshe’. This method has been used to measure the adoption of microblogging (Ma, 2013). Consistent with previous studies, a dichotomous (0 or 1) variable is used to measure the adoption of XNJS (Berry, 1994; Berry and Berry, 1990; Mintrom, 1997; Nicholson-Crotty, 2009). If a province adopts XNJS in the given year, the dependent variable takes the value of 1, and 0 otherwise. In this study, three criteria are used to determine whether a province has adopted XNJS. First, did the province publish XNJS regulatory documents in the given year? Second, did the province convene a meeting on the deployment of XNJS or hold an annual summary meeting? Third, are there any news reports about XNJS in that province in the study year? If the Internet searches for these three criteria yield no hits, the provinces are classified as not adopting XNJS in that year. If any of the criteria are met, the provinces are identified as adopting it. Given the dichotomous dependent variable, we use this method to avoid data collection bias to the greatest extent possible.

Independent variables

Relative age, Politburo incumbency, distance to the general election and chance of appointment to the Politburo are all coded according to provincial secretaries' archives or the Chinese political cycle. A leader's relative age is the difference between the year of the next NCCCP and the year of his birth as age in the year of the next general election is an important factor affecting a politician's promotion (A similar index was developed in Liang and Langbein, 2015) Politburo incumbency takes the value of 1 if the secretary concurrently has Politburo membership, and 0 otherwise. Distance to the general election is the difference between the year of the next NCCCP and the year under investigation; we measure it with this method because we expect cadres to have more incentives to adopt innovation in years nearer to the next NCCCP. Furthermore, we develop an index to estimate the historical likelihood of promotion to the Politburo for each province which is calculated using the Politburo membership representation from 1997 (the 15th NCCCP) to 2007 (the 17th NCCCP). For each NCCCP, one province gets one point for an incumbent Politburo member who worked or was working as the party secretary, governor or both. We measure this variable with the average score per NCCCP for each province. (A similar index was developed in Liang and Langbein, 2015). The score of the 15th NCCCP is used in the 2000–2002 period, that of the 15th to 16th NCCCP is used in the 2003–2007 period and that of the 15th to 17th NCCCP is used in the 2008–2012 period.

We use the number of neighbouring provinces that have adopted XNJS as a measure of the horizontal diffusion effect (Berry and Berry, 1990). In October 2007, Xi Jinping, the sponsor of this reform at the provincial level, was elected to the Politburo standing committee and was expected to be promoted to the top leadership position in China. To explore the effect of his promotion, we set an ordinal variable to measure the top-down diffusion effect. We expect the effect of this political signal to weaken over time as provincial leaders would adopt it as soon as possible to impress Xi in a timely manner. Hence, its value is set to 0 for the years before 2008, and to 5 for 2008, 4 for 2009, 3 for 2010, 2 for 2011 and 1 for 2012. Following Kim (2013), we also use a dichotomous variable, set to 1 for the years after 2007 and 0 for other years, in order to measure the top-down diffusion effect in case the empirical result is caused by our coding strategy. We use the number of governed prefecture-level cities that have adopted XNJS to measure the bottom-up diffusion effect.

Control variables

First, previous studies have shown that highly educated officials tend to adopt innovations (Damanpour and Schneider, 2006; Huber et al., 1993) and that the professional experience of leaders influences their innovative behaviour (Kearney et al., 2000). We use two dichotomous variables to control for the effects of education and experience. Second, organisational and jurisdictional factors are also regarded as crucial for organisational innovation, especially organisational size (Walker, 2007), scarcity of resources (Damanpour, 1991), deprivation (Damanpour and Schneider, 2006; Moon and deLeon, 2001) and the number of Internet users (Lee et al., 2011). Hence, organisation size measured by the number of government employees (Walker, 2007), fiscal health calculated as the difference between fiscal revenue and fiscal expenditure divided by fiscal expenditure (Berry and Berry, 1990), the unemployment rate (Moon and deLeon, 2001), and the proportion of Internet users (Ma, 2013) are all controlled. Third, autonomous regions and municipalities that are directly administered by the central government enjoy more rights than general provinces; we create two dummy variables as controls for this. Data for the control variables are mainly obtained from the China Statistical Yearbooks of the National Bureau of Statistics of China and from provincial secretaries' archives (see Table 1). The control variable data are collected for the same time point as the independent variables but lagged by one year from the dependent data, allowing us to avoid a reverse causal relationship.

Table 1.

Description of variables, summary statistics and sources.

Variables	Description	Mean	St. Dev.
Adoption^a	Adoption of the performance-based reform programme	0.0738	0.2619
Relative age^b	The age of party secretary at the next NCCCP	61.1510	4.2364
Politburo incumbency^b	Dummy variable; 1 for provinces whose party secretary is a member of the Politburo, and 0 otherwise	0.1879	0.3913
Distance to the general election^b	Difference between the year of the next NCCCP and the year under investigation	2.8356	1.3694
Chance of appointment to Politburo^b	An index measured by the average number of Politburo members who served or were serving as a secretary or governor in the investigated province for the 15th to 17th NCCCP. The index of the 15th NCCCP is used for the 2000–2002 period, that of the 15th to 16th NCCCP for the 2003–2007 period and that of the 15th to 17th for the 2008–2010 period, respectively	0.6795	0.9034
Top-down (ordinal)^a	0 for the years before 2008, 5 for 2008, 4 for 2009, 3 for 2010, 2 for 2011 and 1 for 2012	0.7651	1.5651
Top-down (dichotomous)^a	0 for the years before 2008, 1 for 2008, 2009, 2010, 2011 and 2012	0.2282	0.0244
Horizontal^a	Number of neighbouring provinces that have adopted XNJS	0.7416	1.2963
Bottom-up^a	Number of governed prefecture-level cities that have adopted XNJS	1.1678	2.1317
Education^b	The highest degree of education attained	0.3624	0.4815
Experience^b	Participated in the reform when previously working in other provinces	0.0268	0.1619
Deprivation^c	Unemployment rate	3.6190	0.8134
Netizen^c	Citizens using the Internet divided by the total population	0.1198	0.1319
Size^c	Number of government employees (log)	1.4752	0.3304
Resource^c	Difference between fiscal revenue and expenditure divided by fiscal expenditure	–0.4945	0.2011
Municipality^a	Dummy variable; 1 for municipalities	0.1577	0.3651
Autonomous^a	Dummy variable; 1 for autonomous regions	0.1409	0.3485

Sources: ^a Constructed by authors, mainly from online searches. ^b Constructed by authors, based on party secretaries' archives. ^c China Statistical Yearbooks of the National Bureau of Statistics of China.

Model specification

We use the standard approach for discrete-time EHA (Berry and Berry, 2007) to empirically examine the antecedents of the diffusion of XNJS. This approach divides time into equal discrete units using positive integer values (t₁, t₂, t₃ …), and some independent individuals are observed from t₁ to t_i, at which point either the investigated event occurs or the observation is censored (Allison, 1982). The approach aims to model the conditional probability (hazard rate) that an individual will experience the event during an investigated time period, given that he or she has never experienced it before (Berry, 1994; Berry and Berry, 1990; Miao, 2015). Generally, the hazard rate is not directly observable and the observed dependent variable is dichotomous: a unit either experiences the event during the investigated period of time (taking the value of 1), or does not (taking the value of 0). The used data set is pooled cross-sectional time-series and composed of an observation for each unit for each time period when the unit is at risk of experiencing the event (Berry, 1994), for which the probit regression with maximum likelihood estimation techniques is suitable. In this work, the units are 31 provinces in mainland China (Hong Kong, Macau and Taiwan are excluded), the time periods are calendar years, and the event is the adoption of XNJS.

Specifically, the EHA model is as follows:

Adoptio n_{ij} = Φ (α + β_{1} X + β_{2} Y + β_{3} a (t) + e_{ij}),

(1)

where Adoption_ij is the dependent variable that identifies whether province i adopted XNJS in year t, given that the province has never adopted it prior to year t, and Φ denotes the cumulative normal distribution function in the probit regression. X and Y are a vector of the independent variables and a vector of the control variables, respectively. a(t) is a function of survival time, and e_ij represents the stochastic error term. It is worth noting that we first need to specify a(t) in order to make the baseline hazard rate change over time. Some scholars have noted that there are the non-parametric approach (including discrete-time dummy variables) and parametric approach (including variables taking the functional form of the time) to modelling duration dependence (Buckley and Westerland, 2004; Miao, 2015). Here, we specify a(t) using a natural log of survival time t rather than the non-parametric approach according to Akaike information criterion.

Results and discussion

The variables are described in Table 1: 22 of the 31 provinces in China had adopted XNJS by 2012. As the reform originated approximately 10 years ago, this demonstrates a high incidence ratio for the performance-based reform programme (70%), and EHA can therefore be used effectively to explain the variance in the dependent variable. We use five models to explore the factors that affect its adoption (see Table 2). We report the main parameters of the regression models, including −2 log likelihood, the chi-square values and their significance, Nagelkerke's R², and the estimates of the regression coefficients and their significance levels. Table 2 shows that the chi-square values of these five models are significant at the 1% level, indicating a good fit. More importantly, the R² value of the baseline model (model 1, just with control variables) is 0.1831 and it increases dramatically when the independent variables are substituted group by group into models 2, 4, 3 and 5. That means our independent variables can explain the variance of the dependent variable to a great extent. To test for multicollinearity, we use a linear regression based on the least squares method to estimate our models, and find that the variance inflation factor of each variable is much less than the cut-off value of 10 (Neter et al., 1985). There is thus no serious problem of multicollinearity between the independent variables.

Table 2.

Results of the probit regression analyses with the robust option.

		Top-down is ordinal		Top-down is dichotomous
	Model 1	Model 2	Model 3	Model 4	Model 5
Relative age			0.115(0.053)**		0.111(0.053)**
Politburo incumbency			−0.184(0.561)		−0.074(0.544)
Distance to the general election			−0.236(0.092)***		−0.182(0.092)**
Chance of appointment to Politburo			−0.675(0.329)**		−0.659(0.334)**
Top-down		0.147(0.077)*	0.234(0.096)**	0.873(0.449)*	1.087(0.564)*
Horizontal		0.094(0.129)	0.111(0.124)	0.072(0.127)	0.097(0.122)
Bottom-up		0.096(0.092)	0.108(0.096)	0.094(0.091)	0.109(0.095)
Education	0.378(0.272)	0.370(0.273)	0.946(0.408)**	0.377(0.276)	0.937(0.404)**
Experience	0.776(0.489)	0.766(0.526)	1.097(0.538)**	0.669(0.526)	0.920(0.540)*
Size	−0.029(0.389)	−0.456(0.523)	−0.618(0.558)	−0.546(0.528)	−0.703(0.559)
Resources	1.476(1.003)	1.646(1.026)	2.851(1.185)**	2.123(1.050)**	3.284(1.224)***
Deprivation	0.032(0.145)	0.031(0.171)	0.073(0.201)	0.039(0.178)	0.065(0.201)
Netizen	−2.747(2.100)	−5.010(2.266)**	−6.259(2.081)***	−6.668(2.442)***	−8.211(2.550)***
Municipality	−0.683(0.726)	−0.156(0.712)	0.083(0.720)	−0.092(0.680)	0.126(0.701)
Autonomous	0.537(0.349)	0.632(0.368)*	0.463(0.406)	0.683(0.374)*	0.495(0.411)
Ln(time)	1.095(0.652)*	0.698(0.540)	0.695(0.492)	0.750(0.529)	0.785(0.498)
−2 Log likelihood	128.2529	121.0647	111.6302	120.8221	112.4228
Chi²	22.01***	31.41***	37.73***	29.74***	35.08***
Pseudo R²	0.1831	0.2289	0.2890	0.2304	0.2839

Note: ***p < 0.01; **p < 0.05; *p < 0.1; N = 298.

Hypotheses H₁–H₄ examine the effects of political promotion incentives on the adoption of the reform programme. In models 3 and 5, the relative age of leaders is significantly positively correlated with the adoption of XNJS, which supports hypothesis H₁. The effects of a leader's relative age are related to the nomenklatura system, in which older provincial secretaries are under greater pressure to get promotions. Disadvantaged leaders urgently need to adopt innovations to improve their career prospects. Our particular age variable is based on a different functional logic than that in previous studies (Damanpour and Schneider, 2006), which assumed that accumulated professional experience is a more important factor than career ambitions. Hypothesis H₃ is also supported as distance to the general election is significantly negatively correlated with the adoption of XNJS, meaning that innovation diffusion is affected by the political cycle. As the NCCCP approaches, leaders have more incentives to adopt innovations to increase their chances of promotion. Likelihood of being promoted to the Politburo is significantly negatively correlated with the adoption of XNJS, supporting H₄. Ambitious cadres with lower levels of power and reputation or from provinces with less representation on the Politburo must devote more effort to attracting their superiors' attention. Hypothesis H₂ is rejected because the negative correlation between Politburo incumbency and its adoption is non-significant in models 3 and 5. Although no significant correlation is found between Politburo incumbency and the adoption of XNJS, the direction is always negative, which suggests that its effect needs to be explored further.

Hypotheses H₅, H₆ and H₇ examine three diffusion mechanisms predicted by theories of policy diffusion. Top-down intergovernmental interaction, as measured by our ordinal variable, is significantly positively correlated with the dependent variable at the 10% or 5% level in models 2 and 3, fully supporting hypothesis H₅. When we measure it with the second method in models 4 and 5, top-down intergovernmental interaction is still positively correlated with the adoption of reforms, proving that this effect is stable. The horizontal diffusion effect and bottom-up diffusion effect are not significantly correlated with the dependent variable (models 2, 3, 4 and 5), so hypotheses H₆ and H₇ are rejected. This suggests that governmental top-down interaction plays a more important role than horizontal interaction in China's centralised regime. The observation that in China, the status of the innovation sponsor is positively related to the adoption of XNJS represents a significant contribution to current innovation diffusion research. In China's authoritarian party-state regime, the top leader is the most powerful and influential individual. The 17th NCCCP established the promising core position of the sponsor of XNJS, Xi, in the next generation of Chinese leaders. In this situation, provincial leaders tend to promote Xi's reforms. This is inconsistent with the results of innovation diffusion research conducted in Western countries, which show that horizontal regional effects, measured as the number of adjacent provinces that have adopted the innovation, do not significantly affect diffusion (Berry and Berry, 1990; Gray, 1994; Mooney, 2001). These differences may be due to the effects of vertical diffusion in China. We do not find evidence that bottom-up diffusion drives the spread of XNJS. Although previous studies have identified the importance of Chinese ‘policy experiments’ at the local government level (Lee and Warner, 2004), they may only play an important role in the early stages of innovations, rather than throughout the process of innovation diffusion.

Conclusions

Over the past decade, the performance-based reform programme XNJS has spread rapidly at the provincial level of government in China and the belief that performance-based reform can improve the efficiency and effectiveness of public services has led to the introduction of many new issues and research agendas. Among these emerging issues, one relatively unexplored topic is the determinants of the adoption of performance-based reform programmes. What factors drive its diffusion in the centralised Chinese political system, which does not have a federal structure? As our literature review demonstrates, most prior studies of innovation diffusion are based on data collected in Western democratic societies rather than in countries with centralised governments (Berry, 1994; Berry and Berry, 1990; Damanpour and Schneider, 2006; McCann et al., 2010; Mooney, 2001); in such cases, innovation diffusion may not be connected to well-established centralised political institutions. This study addresses these missing gaps in our understanding of innovation diffusion in non-Western contexts.

First, we find that in China, the diffusion of performance-based reform programmes is affected by the political promotion incentives that characterise the Chinese nomenklatura system. Political promotion incentives are shaped by a leader's relative age, current position and chance of appointment to the Politburo. Leaders of older age, in lower-ranked positions and who work in less powerful provinces have limited opportunities for promotion (Li and Zhou, 2005). However, a performance-based reform programme, as an innovation expected to improve government performance, can allow them to improve government performance or impress central leaders and thus improve their career prospects. Previous innovation diffusion research has conceptualised government organisations as the ‘actors’ in policy adoption and have not examined how individual leaders make decisions about policy innovation. By highlighting the role of political promotion incentives, this study contributes to the current literature and confirms the view that innovation adoption is, to some extent, a reflection of top leaders’ political career prospects.

Second, we find that innovation diffusion is related to the political cycle. Provinces are likelier to adopt a reform in years nearer to the next general election, suggesting that local leaders in China have the incentive and capacity to manipulate government behaviours at crucial points during their tenure in order to improve their prospects of political advancement. Unlike Western officials in a democracy, Chinese officials are seen as catering to superiors rather than to the public. Additionally, the relationship between the political cycle and innovation diffusion varies with different kinds of innovation. As Berry and Berry (1990) have noted, new taxes are most likely to be enacted in a year immediately following elections and least likely to be adopted in election years as such an action gives the public the maximum amount of time to forget the government's unpopular action before the next election. However, our results suggest that performance-based reform is likelier to be adopted in years before the next election because it is aimed at improving government performance and is welcomed by both top leaders and the public.

Third, we find a new top-down diffusion mechanism driving innovation diffusion in China: the potential influence of political transition signals. In previous research, vertical mandates or coercion – ‘hard’ top-down mechanisms – were found to be important factors in the diffusion of innovation policy and practice (Berry and Berry, 2007; Dobbin et al., 2007). In recent years, researchers have begun to examine ‘soft’ top-down mechanisms. For example, studies have examined how discussion and deliberation over a bill in Congress can influence policy adoption by state governments through information transfer (McCann et al., 2010), and a shift in national policy direction can send policy signals that drive policy diffusion (Kim, 2013). However, this study shows that the political promotion of an innovation's sponsor as a political leader can drive its diffusion. This is a ‘soft’ top-down diffusion mechanism that works through sending political transition signals.

Our results have important implications. They reveal the strong effect of political promotion incentives on policy leaders' decisions to adopt an innovation, especially in the Chinese nomenklatura personnel management system. They also suggest that future research on innovation diffusion should be nested in particular institutional contexts. Furthermore, although the effects of horizontal diffusion have been the focus of studies of Western federal countries, top-down diffusion plays a more significant role in China. Thus, future studies must give more consideration to top-down interactions, especially in centralised countries. Our results show that officials are expected to adopt innovations to improve their government's performance or to cater to their superiors. In both democratic and centralised systems, government officials can manipulate their behaviours to increases their popularity among those to whom they are accountable. This can mean adopting a popular innovation or rejecting an unpopular innovation at a crucial time.

Finally, this assessment of the province-wide adoption of a performance-based reform programme provides the first step in understanding government performance management behaviour. Although the adoption of reforms is a necessary precondition of improved performance, it does not necessarily guarantee concrete performance improvement. How these reforms are actually used to guide changes in government behaviours and their effect in improving government performance needs to be investigated. In addition, it also has some measurement and data source limitations. The dependent variable is a simple binary variable as it is difficult to collect data about the extent to which individual provinces implemented XNJS in a given year. However, using previous studies of the US government reinvention movement (Moon and deLeon, 2001) as a reference, future studies can develop a scale to measure the key components and explore its antecedents with cross-sectional data. Another limitation may be that some of the independent variables are binary or counting variables coded according to objective archives. Although this method has been used in some studies (Berry, 1994; Berry and Berry, 1990; Ma, 2013), the results must be interpreted with caution and more robust evidence based on survey data should be examined in the future.

Footnotes

Acknowledgements

We are grateful to Chunping Hu, Li He, two anonymous reviewers and the editor of International Review of Administrative Sciences, who gave invaluable comments and suggestions on earlier versions of the article.

Funding

This work was supported by the National Natural Science Foundation of China (grant numbers 71173167, 71103140) and the Ministry of Education in China (grant number 13JZD015).

Notes

Jiannan Wu, PhD, is a Distinguished Professor and Executive Vice Dean of the School of International and Public Affairs at Shanghai Jiao Tong University, China. He is also a Professor at the School of Public Policy and Administration at Xi'an Jiaotong University, China. His research interests include innovation and performance management in the public sectors.

Pan Zhang is a PhD candidate at the School of Public Policy and Administration at Xi'an Jiaotong University, China. His research interests include government innovation and performance management.

References

Allison

(1982) Discrete-time methods for the analysis of event histories. Sociological Methodology 13: 61–98.

Bantel

Jackson

(1989) Top management and innovations in banking: Does the composition of the top team make a difference? Strategic Management Journal 10: 107–124.

Berry

(1994) Innovation in public management: The adoption of strategic planning. Public Administration Review 54(4): 322–330.

Berry

(1990) State lottery adoptions as policy innovations: An event history analysis. The American Political Science Review 84(2): 395–415.

Berry

(2007) Innovation and diffusion models in policy research. In: Sabatier

(ed.) Theories of the Policy Process, (2nd edn). Colorado, CO: Westview, pp. 223–260.

Borins

(2001) Public management innovation: Toward a global perspective. The American Review of Public Administration 31(1): 5–21.

Brinkerhoff

Wetterberg

(2013) Performance-based public management reforms: Experience and emerging lessons from service delivery improvement in Indonesia. International Review of Administrative Sciences 79(3): 433–457.

Buckley

Westerland

(2004) Duration dependence, functional form, and corrected standard errors: Improving EHA models of state policy diffusion. State Politics and Policy Quarterly 4(1): 94–113.

Burns

(1987) China's nomenklatura system. Problems of Communism 36(5): 36–51.

10.

Burns

(1989) The Chinese Communist Party's Nomenklatura System: A Documentary Study of Party Control of Leadership Selection, 1979–1984, New York, NY: M.E. Sharpe.

11.

Burns

(1994) Strengthening central CCP control of leadership selection: The 1990 nomenklatura. The China Quarterly 138(1): 458–491.

12.

Chan

(2004) Cadre personnel management in China: The nomenklatura system, 1990–1998. The China Quarterly 179: 703–734.

13.

Chow

(1988) The management of Chinese cadre resources: The politics of performance appraisal (1949–84). International Review of Administrative Sciences 54(3): 359–377.

14.

Christensen

Dong

Painter

(2008) Administrative reform in China's central government – How much ‘learning from the West’? International Review of Administrative Sciences 74(3): 351–371.

15.

Damanpour

(1991) Organizational innovation: A meta-analysis of effects of determinants and moderators. The Academy of Management Journal 34(3): 555–590.

16.

Damanpour

Schneider

(2006) Phases of the adoption of innovation in organizations: Effects of environment, organization and top managers. British Journal of Management 17(3): 215–236.

17.

Edin

(2003) Remaking the communist party-state: The cadre responsibility system at the local level in China. China: An International Journal 1(1): 1–15.

18.

Gray

(1994) Competition, emulation and policy innovation. In: Dodd

Jillson

(eds) New Perspectives in American Politics, Washington, DC: Congressional Quarterly Press, pp. 230–248.

19.

Hambrick

Mason

(1984) Upper echelons: The organization as a reflection of its top managers. The Academy of Management Review 9(2): 193–206.

20.

Heilmann

(2008) Policy experimentation in China's economic rise. Studies in Comparative International Development 43(1): 1–26.

21.

Huber

Sutcliffe

Miller

et al. (1993) Understanding and predicting organizational change. In: Huber

Glick

(eds) Organizational Change and Redesign, New York: Oxford University Press, pp. 215–265.

22.

Kearney

Feldman

Scavo

CPF

(2000) Reinventing government: City manager attitudes and actions. Public Administration Review 60(6): 535–548.

23.

Kiewiet

McCubbins

(1985) Congressional appropriations and the electoral connection. Journal of Politics 47: 59–82.

24.

Kim

D-R

(2013) Local government policy diffusion in a decentralised system: Childbirth support policy in South Korea. Local Government Studies 39(4): 582–599.

25.

Kung

Chen

(2011) The tragedy of the nomenklatura: Career incentives and political radicalism during China's Great Leap famine. American Political Science Review 105(1): 27–45.

26.

Lee

C-p

Chang

Berry

(2011) Testing the development and diffusion of e-government and e-democracy: A global perspective. Public Administration Review 73(1): 444–454.

27.

Lee

GOM

Warner

(2004) The Shanghai re-employment model: From local experiment to nationwide labour market policy. The China Quarterly 177: 174–189.

28.

Zhou

(2005) Political turnover and economic performance: The incentive role of personnel control in China. Journal of Public Economics 89(9/10): 1743–1762.

29.

Liang J and Langbein L (2015) Performance management, high powered incentives, and environmental policies in China. International Public Management Journal 18(3): 346–385.

30.

(2013) The diffusion of government microblogging: Evidence from Chinese municipal police bureaus. Public Management Review 15(2): 288–309.

31.

McCann PC, Shipan CR and Volden C (2010) Intergovernmental policy diffusion: National influence on state policy adoptions. Paper presented at the Annual Midwest Political Science Association Meeting, Chicago, IL, 22–25 April.

32.

Manion

(1985) The cadre management system, post-Mao: The appointment, promotion, transfer and removal of party and state leaders. China Quarterly 102: 203–233.

33.

Miao Q (2015) What drives climate preparedness: An assessment of state climate adaptation planning in the United States. Available at SSRN: http://dx.doi.org/10.2139/ssrn.2555129.

34.

Mintrom

(1997) Policy entrepreneurs and the diffusion of innovation. American Journal of Political Science 41(3): 738–770.

35.

Moon

deLeon

(2001) Municipal reinvention: Managerial values and diffusion among municipalities. Journal of Public Administration Research and Theory 11(3): 327–351.

36.

Mooney

(2001) Modeling regional effects on state policy diffusion. Political Research Quarterly 54(1): 103–124.

37.

Neter

Wasserman

Kutner

(1985) Applied Linear Statistical Models, Homewood, IL: Irwin.

38.

Nicholson-Crotty

(2009) The politics of diffusion: Public policy in the American states. The Journal of Politics 71(1): 192.

39.

Oakley

(2009) Agenda setting and state policy diffusion: The effects of media attention, state court decisions, and policy learning on fetal killing policy. Social Science Quarterly 90(1): 164–178.

40.

Rincke

(2007) Policy diffusion in space and time: The case of charter schools in California school districts. Regional Science and Urban Economics 37(5): 526–541.

41.

Shipan

Volden

(2006) Bottom-up federalism: The diffusion of antismoking policies from U.S. cities to states. American Journal of Political Science 50(4): 825–843.

42.

Shipan

Volden

(2008) The mechanisms of policy diffusion. American Journal of Political Science 52(4): 840–857.

43.

Walker

(2007) An empirical evaluation of innovation types and organizational and environmental characteristics: Towards a configuration framework. Journal of Public Administration Research and Theory 18(4): 591–615.

44.

Wejnert

(2002) Integrating models of diffusion of innovations: A conceptual framework. Annual Review of Sociology 28: 297–326.

45.

Welch

Thompson

(1980) The impact of federal incentives on state policy innovation. American Journal of Political Science 24(4): 715–729.

46.

Yang

(2013) Innovation in the Chinese public sector: Typology and distribution. Public Administration 91(2): 347–365.

47.

Xue

Zhong

(2012) Domestic reform and global integration: Public administration reform in China over the last 30 years. International Review of Administrative Sciences 78(2): 284–304.

48.

Zhu

(2014) Mandate versus championship: Vertical government intervention and diffusion of innovation in public services in authoritarian China. Public Management Review 16(1): 117–139.