INNOVATION OR IMITATION – A COMPARISON OF PERFORMANCE EVALUATION MODELS IN CHINA

Abstract Innovation of public sector has been greatly advocated and supported in China, and numerous so-called local innovations have emerged. Are they really innovations or just imitations from others? Through a quantitative comparison analysis with 66 so-called innovations in performance management of China’s local governments, we fi nd that these innovations show more similarities than differences to each other, and they are imitations rather than innovations. Further, we argue that the ‘de facto federalism’ of China’s power structure, the incentive of China’s bureaucratic system, and the organizational culture of the public sector may result in such an outcome that rhetoric innovation is much more than substantive innovation.


Introduction
Since the rise of the New Public Management (NPM), the importance of government innovation has been recognized for it can drive the performance of public sector and increase their excellence (Pott s, 2009; Golden, 1990;Borins, 2000;Borins, 2014;Hartley, 2005;Walker, 2006;Hood, 1991;Osborne and Gaebler, 1993).Numerous policymakers and scholars around the world try to promote government and/or policy innovations, and China is no exception.Nowadays in China, the public sector, from central to local, all regard innovation as a critical approach to realize bett er governance and higher management effi ciency (Yang, 2007;Chan and Chow, 2007;Wu et al., 2013;Lan and Galaskiewicz, 2012;Su et al., 2013;Liu and Li, 2013).
However, innovation in China contains deeper implications.Usually scholars defi ne innovation in a more lenient way as 'an idea, practice, or project that may not have been invented by the polity adopting it but that is new to that polity' (Berry and Berry, 1999;Gray, 1973;Walker, 1969), while innovation in the Chinese context has been much more narrowly-defi ned.Innovation means something which is new not only to the adopter but to others as well, at least to other jurisdictions in China.Policy innovation in China then is equivalent to what scholars defi ned as 'policy invention', which is original, unique, creative, non-traditional, and departing from the existing practice (Berry and Berry, 1999;Dewar and Dutt on, 1986).The polity who designed and adopted this policy innovation can be called a pioneer while the other adopters, even though this policy is new to them, can only be regarded as followers since they followed, either by learning or by imitating, a policy which has been already adopted by others. 1ven though quite a few innovation scholars deny the concept of 'policy invention', and insist that 'there is no new thing under the sun' (Dewar and Dutt on, 1986; Dempster and Wildavsky, 1979; Kingdon, 1995;Majone, 1991), there are still some scholars believing that entirely new policy could be possible (Berry and Berry, 1999;Jordan and Huitema, 2014).Without joining this scholarly debate, we, based upon the opinions from both sides, can understand that invention, even if possible, would be highly diffi cult.But the diffi culties of coming up with entirely new ideas do not block the enthusiasm of innovation in China's public sector.In the past decades, diff erent forms of 'innovation' emerge at all levels of the public sector, and an 'innovation' movement is witnessed in China.To echo this innovation movement, several agencies with either political authority or academic authority, including the Central Compilation and Translation Bureau, the Party School of the Central Committ ee of the China Commu-nist Party (CCP), and the Center of China Government Innovations at Peking University have jointly set up a program of 'Innovations and Excellence in Chinese Local Governance' (IECLG) since the year 2000.This program, similar to the Innovations in the American Government Awards Program in the United States and Beacon Scheme in the United Kingdom, aimed at encouraging, communicating, and disseminating the 'best practice' of innovations in China's local governments (Wu et al., 2013).
To ensure the justice and legitimacy of this program, IECLG publicized clear eligibilities, rules, standards, and procedures on its website, and all materials for fi nalists, including the fi nal oral defense, were open to the public.Based on the rules of the program, all local governments and public organizations are eligible to apply for the program.The fi nal decision, which was made through several steps by an expert committ ee, was made on the basis of six criteria -signifi cance, eff ectiveness, innovativeness, economic promotion, public participation, and potential for replication and diff usion.The IECLG provides detailed explanation to each criterion.For 'innovativeness', it just coincides with what scholars defi ned as 'policy invention' (Berry and Berry, 1999), meaning that the innovation should be originally designed by the developer, and cannot be a replication of any others.
Even with such strict eligibilities and requirements, IECLG never stops local governments' pursuit for innovation.By the end of 2014, the program has been held for seven rounds, and received more than 1,500 applications from all over China.More than 140 creative policies or program initiatives, abiding by the demanding criteria and passing through the competitive selection, are entitled winners or fi nalists.Over the years, the IECLG program att racted numerous att ention from both practical and academic fi elds (Wu et al., 2013) and further motivated the local governments to keep reforming and innovating.Along this innovation movement as well as under the motivation of IECLG and higher authorities in China, nobody would like to lag behind and a competition, not just movement, for innovation also appears (Zhu, 2014a;Liu and Li, 2016).
To have more innovations in the public sector is positive but not easy.Then here come our concerns.Among all the innovations in various levels of governments in China, are they really innovative as the IECLG required?Can they really be called 'innovation' or is this another 'Great Leap Forward' of innovation?In this article, we try to make such a judgment by evaluating China's innovation movement.
To answer our research question, we conduct a quantitative comparison analysis with 66 so-called innovations in the area of performance management of China's local governments.We fi nd that these innovations show more similarities than diff erences to each other.Therefore, most of these self-claimed innovations have few uniqueness or originality.Rather than innovation, they are merely imitation from each other.Lastly, we also proposed our tentative explanations to this phenomena.
The rest of this article is deployed as follows.In the second section, we take a holistic view to discuss the situation of government/policy innovation in contemporary China.We also examine the existing literature on China's government innovation briefl y.The quantitative comparison, including the sampling, methodology, results, and analysis, is presented in the third section.In this section, the results of data analysis show that the innovations in our sampled models are actually mere imitation.They did not show enough innovativeness and uniqueness as the IECLG program required.In the fourth section, we, from the perspectives of structure, process, and culture, provide tentative understandings for this phenomenon.We argue that the 'de facto federalism' of China's power structure, the incentive of China's bureaucratic system, and the organizational culture of public sector all result in such an outcome that rhetoric innovation is much more than substantive innovations.The last section provides policy implications and discusses the limits of this research.

Innovation of China's public sector -practice and theory
Innovation drives development (Romer, 1994).This statement has been widely believed in China, and several generations of leaders all emphasize the importance of innovation.Innovation in the public sector was att ended, encouraged, and advocated by China's highest authority, and fi nally became the national strategy.
After China's market economy has evolved for decades, innovation in the public sector has been att ended.In the 16 th National Congress of the CCP, the highest leadership level confi rmed that 'innovation is the soul for a nation's progress, the driver for a country's prosperity, and a source for a Party's vitality'.China should focus on theoretical innovation and promote innovation in institutions, technology, culture, and other aspects.
In the 17 th National Congress of CCP, innovation has been set up as a national strategy.It is believed that 'improving independent innovative capability and building an innovation-driven country is the core of the national development strategy, and also the key to improve comprehensively national strength'.To insist on China's independent innovation path, (we should) advocate innovation in all areas of modernization building.This concept of 'innovation' was soon brought to the fi eld of public management and administration.When disseminating the main administrative tasks of the State Council, the then Premier Wen stated that promoting government innovation is an important content of economic and political reform.The second plenary session of the 17 th National Congress of CCP issued an offi cial statement, 'opinions on deepening the reform of the administrative system', sett ing up the goal of the CCP toward year 2020 and further encouraging public sector, under the leadership of the central government, to actively devote itself to local innovation.The then General Secretary of CCP Hu proposed that 'innovation should be encouraged and protected.All innovative ideas should be respected; all innovative behaviors should be supported; all innovative intelligence should be inspired; and all innovative achievements should be rewarded'.
In the 18 th National Congress of CCP, the Party set up the goal of reform during the current stage, which was to innovate the governance approach, to build-up governments' credibility, and to implement governmental performance management.Under Xi's leadership, China implemented the strategy of 'innovation-driven development', and depended on innovation to speed up the transformation of the economic development, to solve deep-seated contractions, and to enhance endogenous vitality of the overall development.
With these advocacy and mobilization from the highest leadership level in China, all levels of local governments and public sector were dedicated to innovation.Along with this practice in China's public sector, quite a large amount of research about China's innovation in local governments has been presented.Scholars have examined the various fi elds of innovation in the public sector, including the adoption of new technology (Hartford, 2005;Holliday and Yep, 2005;Ma, 2013), the reconstruction of the organizational structures and management processes (Dong et al., 2010;Tsai and Dean, 2014;Pieke, 2009;Gao, 2009;Foster, 2006), the delivery of new social services (Guo et al., 2008;Simon and Teets, 2012), and the collaboration between government and various social actors (Hsu, 2010;Hsu and Hasmath, 2014;Hildebrandt, 2011).Various policy areas such as housing (Zhu, 2014b), healthcare (Wang, 2009), urbanization (Liu and Qin, 2016), and public budgeting (Wu and Wang, 2011;He, 2011), are thoroughly discussed.In addition to these studies on specifi c issue areas, the more comprehensive examinations are also presented to explore the overall typology and distribution of China's local innovation (Wu et al., 2013).
However, most of this research adopted the method of case study, examining one case thoroughly but individually.In other words, these cases are not compared with each other.Within their restricted jurisdiction, they can all be called innovation, which, per se, is a 'context-specifi c' concept (Rogers, 2003).But, put in a larger area, are they still innovation?The extant literature obviously ignored this question, and this is the key focus of our research.
This question, although largely neglected by innovation scholars, has special importance.Firstly, to examine innovation across one single geographical location may provide us a comprehensive picture of the current 'innovation movement' in China, for it happens everywhere, not in one place.Only taking the self-claimed innovation out of its original area, can we fi nally know whether it is a true innovation or just an imitation of others; whether it is a substantive innovation or merely a rhetoric innovation?Secondly, this question may provide implication to further understanding China's central-local relations, as well as relationships of various polities.Agreeing with the previous scholars, we admit that those abovementioned cases are innovation within their own context or jurisdiction.But to understand the mechanism of an authoritarian system with a strict vertical and hierarchical governing structure, it's not always enough to narrow the innovations within one area or jurisdiction.The infl uence from above and neighboring governments should always be included (Liu and Li, 2016;Zhu, 2014a).Due to this consideration, we should go beyond the original context and put local innovations in a larger context to conduct a comparative analysis.Thirdly, with the cases of China's innovation in the public sector, we also want to join in the theoretical argument of innovation.With this research and its following analysis, we try to further understand the drivers and impediments of local innovation.In the following section, we conducted a quantitative comparison analysis of local innovations using the cases of performance evaluation innovations in China's local government.

Comparative analysis of performance evaluation models
So many self-claimed innovations emerged in local governments.Are they really innovation, or just imitation of others?Zhu (2014a) fi nds that local governments, for the sake of competition, innovate.Especially for neighboring local governments, a 'championship policy diff usion' mechanism forms among them, and it leads to divergent policy (Zhu, 2014a).Thus innovation is expected.Admitt ing this mechanism, Liu and Li (2016) extend Zhu's research and fi nd that local governments innovate fi rst but imitate later on.Although Liu and Li provide a new possibility, their conclusion has not been tested through large N data.In this section, we, by conducting a large N quantitative analysis, test the hypotheses of the above authors.
We choose our data from the fi eld of performance evaluation of the public sector.Since the end of 2000s, performance management has burgeoned in China.The central government did not issue any specifi c plan or model but encouraged the local governments to start trial and experiment by themselves.Correspondingly, numerous local innovations emerged in this fi eld.With large available and measurable data, we are able to focus our research on the fi eld of performance evaluation of the public sector.

Sampling
We rely on two databases -China's Local Government Innovation Case Database and China's Government Performance Evaluation Report, to come up with our data.Both databases are developed by the Central Compilation and Translation Bureau, one of the hosts of IECLG, and include 'best practices' of China's local governments.We collected all cases of performance evaluation from those databases, and came up with 66 models of performance evaluation as our sample.
Without exception, all 66 models claimed that they were an original innovation.Most of them were even titled with the name of the polity where it was developed, such as 'Sichuan Model' developed by Sichuan Province, 'Qingdao Model' initiated in the city of Qingdao, etc.This special way of naming further declares uniqueness of these local innovations, emphasizing that they were exclusively developed and adopted, distinguishing them from others.

Variables
Surely all the samples are plans or models for public sector' performance evaluation.As we argued earlier, if one model is an innovation, it should be unique, original, and diff erent from others (Berry and Berry, 1999; Dewar and Dutt on, 1986); otherwise, one model could easily fi nd a similar one.With 66 models in hand, we can conduct a comparison to see whether these models are diff erent from, or similar to, each other.If these models are quite diff erent from each other, they might be innovation by themselves.On the contrary, similar models will indicate the imitation of each other.The unit of our analysis is the model of performance evaluation.
To compare these models, we compared the indicators that they used because indicators are the essential part of the evaluation, and the most powerful management tool (Adam and Gunning, 2002;Peterson et al., 2010).Therefore, we used the indicators and their relative weights to defi ne the uniqueness of each individual model.
Among our sampled models, we found that the indicators, without exception, are all separated into three levels -dimensions, basic indicators, and operational indicators.Among these levels, only the third one is of real signifi cance since only operational indicators contain the information of performance, while the other two are used as categories for result-reporting and publishing.Therefore the operational indicators and their weights in each model were what we measured.
Through a pilot coding, we found that almost all indicators could be grouped into eight categories: political orientation (PO), economy achievement (EA), society development (SD), culture harmony (CH), people's satisfaction (PS), resource and environment protection (REP), one-vote veto (OVV), and bonus or negative bonus items (BNBI).Then the specifi c operational indicators of each model were coded into those eight categories.
With the procedure of data-cleaning, we set up four variables to compare the sampled models.In other words, we used these four variables to judge whether those models were distinguishable from each other or similar to each other.The variables included indicators and their weights, plus the eccentricities of them.The 'eccentricity' may show whether those measurements were close to each other or not.High value of eccentricity represented more similarity.Specifi cally, our variables included the number of indicators in one category (NIOC), eccentricity of numbers of indicators in one category (ENIOC), categorical weight (CW), eccentricity of categorical weight (ECW).

Measurement and methodology
Test variable 1: Number of Indicators in One Category (NIOC) We follow Adam and Gunning (2002)'s approach to measure NIOC, which represents the number of a model's operational indicators in a certain category.Taking the Liaoning Model as an example, there are six operational level indicators in the area of 'political orientation' (PO), including 'government capacity building', 'lawbased administration', 'construction of government eff ectiveness', 'the Party developing', 'government integrity', and 'people's involvement'; the NIOC of 'politics indicators' in this mode is therefore coded as 6.
Testing variable 2: Eccentricity of Number of Indicators in One Category (ENIOC) ENIOC measures the diff erence between the NIOC in a certain category of a certain model, and the mean of the NIOC in all models in the same area calculated by available data.The calculation formula is: It shows the deviation of an individual model in a certain category from the average of all models in the same category.The greater the deviation is, the more distinctive the model is.
Testing variable 3: Categorical Weight (CW) CW stands for the weight in a certain category of a certain model (Rossi et al., 2003).Still take the Liaoning model as an example, the raw weight of PO is '95'.In our calculating process, if the original weight values are percentiles, the CW values will be the original values; otherwise, we will transform them into percentile forms.For example, CW of the Liaoning model in the category of PO is 95/(95+425+285+50+70+4 5)*100%=9.8%,while that of Guangdong is 5.9%.
Testing variable 4: Eccentricity of Categorical Weight (ECW) ECW means the diff erence between the CW in a certain category of a certain model and the average value of CW of all models in the same category calculated by available data (Rossi et al., 2003).The calculation formula is: It shows the deviation of a single model in a certain category from the average of all models in the same category.The greater the deviation is the more distinctive the model is.

Control variables
With all the indicators of these 66 models cleaned and scanned, we found that there are some indicators with universal signifi cance.These indicators show the basic requirements of the general performance evaluation, and carry no local features.They are 'non-innovative' indicators in nature.For example, all models of performance evaluation will measure the degree to which an organization loyally implements public policy.Thus the indicator of 'policy implementation' is a required indicator for all models, and should be treated as a control variable to avoid misjudgment of innovation.
To screen our control variable, we examined the descriptions of each indicator of all sampled models.When the content of indicator descriptions were largely coincident, we took that indicator out and regarded it as controlled.In total, we picked 18 indicators from our 66 models.

Results
We conducted a clustering analysis with our tested variables.Figures 1 and 2 show the results, which intuitively refl ect the phenomenon of 'model colonizing'.Judging from the cluster of NIOC, no obvious diff erences can be found among different models.It can be told from Figure 1 that almost all indicators are surrounding the origin with small fl uctuations.The centrality (indicated in red) of ENIOC, which represents the similarity between the indicators and the averages, is especially notable.Most indicators are within the interval of (-5, 5).In other words, the centrality of both NIOC and ENIOC are of high degree.This shows that the 66 models, judging from their operational level, have no signifi cant diff erence.
Judging through the cluster of CW and ECW in Figure 2, we get the same conclusion.Although the centralities of CWs and ECWs are slightly less notable than those of NIOCs and ENIOCs, the results show large degree of centrality.Telling from Figure 2, we fi nd that most ECWs are within the intervals of (-10%, 0) and (0, 10%), showing that these models have slight diff erences from each other, and the degree of their diff erences, measured through CW, is less than 10 percent.
Through a quantitative comparison of the 66 models of performance management, we fi nd that these models do not show many diff erences from each other.On the contrary, they are, judged through the most important elements of the evaluation models -the number of operational indicators and their weights, quite similar to each other.Even though these innovations are actually imitations, they are not exact copies of one another.There are still slight diff erences among our sampled models.One explanation to these minor diff erences is that the developers of these 66 models borrowed the existing experiences from other places, revised others' models slightly, and named them uniquely as if they were their original design.Fairly speaking, local governments may not be intentional cheaters of innovation, but they overly exaggerate their revisions, and glorify themselves as innovators.
Therefore, we reveal this phenomenon in China.In the fi eld of performance management of public sector, while many local governments rhetorically claimed that they designed an innovative model and named the model uniquely, these selfclaimed innovative models essentially had no signifi cant diff erence from each other, and were just imitations of each other.

Tentative explanation
With the quantitative comparison of the 66 self-claimed innovations in performance management, we fi nd that they show more similarities than diff erences.Based on this observation, we conclude that these 'innovations', essentially, are not innovations at all.They are mere imitations from each other 2 .Then here comes a dilemma, at least in the fi eld of performance management for the public sector: the local governments rhetorically claim that they are innovating, while essentially they are imitating from others.Why could this happen?What elements lead to the inconsistence be-tween local governments' behavior and utt erance?In this section we try to tentatively explain this phenomenon from three perspectives of structure, process, and culture (Lieberthal, 1992;Chung, 1995).
Structurally, this phenomenon can be understood under China's political power structure.China's 'de facto' federalism has been widely acknowledged, where the central government uses various methods to control local governments while the localities have their own spaces and enjoy a certain degree of autonomy (Zheng, 2006).Under such a power structure between the central and local governments, policies can not only be selectively implemented (O'Brien and Li, 1999) but also be partially or superfi cially implemented.In the fi eld of government innovation, the central government presented huge enthusiasm to promote innovations of local governments.
Lying at the lower level of the power hierarchy, local governments are under the infl uence of the central government and have to implement its policy.Thus when the central government set 'innovation' as a national strategy, various local governments, at least orally, expressed their support to innovation and claimed they have worked hard to develop local innovations.But the autonomy provides local governments a chance to behave diff erently.Without direct coercion and supervision, the local governments had much space to manipulate and deviated from the policy of the central government.They behave diff erently, imitating rather than innovating.Therefore, China's special central-local relations resulted in such a phenomenon that everybody claims innovation while they just imitate each other.
Taking a process perspective and examining the behavior of the local governments may also provide clues to understand this phenomenon.To implement a policy is not a one-time deal.The policy includes multiple elements and the implementation process includes various steps.It is not necessary for local governments to totally resist an unfavorable policy (Gobel, 2011).Following O'Brien and Li's logic (1999), we believe that the local governments implement the easier parts of the policy while muddling through the essential and diffi cult parts of that policy.Once the central government called for innovation, the local governments would immediately follow to admit the advantages and necessities of innovation.For local governments, this is an easy and manipulative way to show their political obedience and loyalty.But to put expression into action will be another story.Innovation in action is not easy and may involve high costs and risks (Pott s, 2009).To imitate others then becomes a secure way for it avoids the costs of error-trial.Especially, if the model which is about to be imitated is testifi ed as successful in other places, the local governments would have stronger incentive to copy.Thus the local governments, on the one hand, claim that they are innovators to show their obedience and political loyalty to the central government; on the other hand, they imitate others to avoid extra costs and risks.
At last but not at least, this phenomenon can also be understood from the organizational culture of China's public sector.China's public sector is strictly organized under bureaucratic principles, with focuses on upward accountability, effi ciency, formal rules, hierarchy and control (Chan and Rosenbloom, 2010).With the deepening of China's performance management policy, this trend has been further strengthened.Judged upon Cameron and Quinn's classical model (apud Wu et al., 2013), the organizational culture of China's public sector belongs to hierarchy culture, valuing stability, predictability, and effi ciency in organizational operations.Numerous research has found that hierarchy culture is associated with imitation strategy, rather than innovation (Naranjo-Valencia et al., 2011;Burns and Stalker, 1994).This may naturally refl ect on the behavior of China's local governments.To echo the advocacy of innovation from the central government, local governments borrow the existing experiences from other places, revise them slightly, and then claim that they have developed an innovation.This phenomenon therefore partly results from the organizational culture of China's public sector.

Conclusion and discussion
Through large N comparative analysis, we fi nd that most self-claimed innovations in the fi eld of performance evaluation are mere imitation because these models show more similarities than diff erences from each other.A question still exists.Among all these 66 models, is there any outlier showing uniqueness?In this last section, we will provide the answer to this question.In addition, we will also present the policy prescription as well as the limitation of this research.
Although the majority of the sampled models show large degree of similarities, diff erences exist.As shown in both Figures 1 and 2, several branches are produced by leaps of ENIOC and ECW, and this means that there are several models deviating from the average and forming singular points.With carefully data cleaning, we detect eight 'innovative models' which are quite diff erent from the majority 3 .Among these outliers, there is still another interesting fi nding.The eight truly innovative models all present their uniqueness in the category of PO.They either set less indicators in the category of PO or give less weight of it.Therefore, another conclusion that can be tentatively drawn is that 'too much emphasis on political orientation leads to the lack of innovation'.This may not be diffi cult to understand.To stick to the 'political nature' of public administration has become a stale doctrine in China.All governments, ranging from central to local, must be governed by the shadow or real 'political committ ee' (Guo, 2007).Although China committ ed to developing an 'administrative state' in the reform of 1998, the reality shows that China is still a 'political state'.Even though much progress has been made in almost every fi eld, litt le has been witnessed in the fi eld of political issues ever since the era of the reform.
During China's reforming decades, a parallel trend taking place in Western countries' public administration is managerialism, focusing on the effi ciency, eff ectiveness and economy of the public sector (Pollitt and Bouckaert, 2004).The management tool, which is performance management, was learned by the Chinese government quickly.They found that using the performance indicators as ruling and controlling tools to reach political goals was an eff ective way.The political control plus operational tools of managerialism can also help the Chinese government to avoid criticism of dictatorship (Guo, 2007).As a result, most local governments pay much att ention to the category of PO, and their so-called innovations of performance evaluation thus show large degree of homogeneity.Several performance evaluation models pay less att ention to the category of PO and make breakthrough innovation.
Following this analysis, the fi rst policy implication of our research is that to release the political control will result in innovation.Although the totalitarian model no longer prevails, strict political control still exists and may sometimes restrict the vitality of innovation.To encourage innovation, the government should further mobilize intelligence among the society and further emancipate the mind.
Secondly, the Chinese government, from central to local, should overcome the myth of innovation, and seek down-to-the-earth policies to solve social problems and realize public goals.As we know, innovation is important, and diffi cult.Then in China's public sector, innovation is defi ned based on its external format of being original and unique; but the internal content is largely ignored.Due to this situation, the most so-called innovations in public sector are just old wine in new bott les.They are innovations in format or rhetorically.Actually, the myth of innovation should be reexamined.Regardless of innovation or imitation, as far as the policy really solves problems, it could be regarded as a success.We believe that some authorities in China have started to be aware of this.The IECLG, after lasting for fourteen years, changed its title to 'Best Practice Award' in 2015.The format features, like originality and uniqueness, are not emphasized anymore; but the eff ectiveness, effi ciency and economy are key criteria.Innovation is a good thing.To learn from successful cases, as some Chinese local governments have done, is not bad either.On the contrary, to innovate for the sake of innovation is not welcomed.
In this article, through a quantitative comparison of 66 self-claimed local innovations, we fi nd that these innovations contain litt le originality, and they are mere imitations of each other.However, this conclusion is not precise enough.When the performance evaluation models show large degree of similarities, this fact cannot deny the existence of innovation.Even though there is only one innovator, the others can still learn or imitate from it.When the innovation diff uses, the convergence of policy will prevail as an outcome.Thus, the complete process of China's performance evaluation for public sector should go as follows.
Firstly, one or two key local governments innovated and developed a model.Secondly, this model was informed to other local governments and they adopted this other-developed model in their own jurisdiction; at the same time, these followers denied the process of learning or imitation by claiming that they developed an original model.Finally, we witness a diff usion of certain performance evaluation model.This process echoes the previous empirical research (Liu and Li, 2016).In the whole process, several local governments innovate and the majority imitate.Then there is a major limit of this research.We do not distinguish innovators from followers.To do so, we need to examine the timing of adoption.Earlier adoption represents innovation, while late adoption of a similar model means imitation.This may need a 'time-model' data format, and it will be left for further study.

Figure 1 :
Figure 1: The cluster of NIOC and ENIOC of all sampled models

Figure 2 :
Figure 2: The cluster of CW and ECW of all sampled models