The recent discussion about public accountability highlights the issue of how to bring the citizenry closer to public administrators (Blair 2000; Frankish, Kwan et al. 2002; 2002; Haque and Mudacumura 2007; Liddle 2007; Eckardt 2008; Khan and Chowdhury 2008; Sarker and Hassan 2010). Among the others, one stream of research is particularly interested in how to engage citizens in performance measurement systems (Ho and Coates 2002; Rowe and Shepherd 2002; Holzer and Yang 2004; Heikkila and Isett 2007; Halachmi and Holzer 2010; Hildebrand and McDavid 2011). The basic tenets of public accountability apparently support that engaging citizens is beneficial, as it facilitates expressing public needs and concerns (Vries 2007), promoting public trust (Fard and Rostamy 2007), reinforcing institutional learning, enhancing the integrity of public employees, and increasing bureaucratic responsiveness (Haque and Mudacumura 2007).
Although engaging citizens in performance measurement sounds promising, the realization of public accountability however, depends on many factors. On the side of citizens, holding governments accountable not only needs citizens to consume time in accountability practices, but also rests on citizens' capacity to demand answers from agents about their proposed or past behavior, to discern that behavior, and to impose sanctions upon the agents (Eckardt 2008). It is also important for citizens, who occasionally represent their individual interests as service users, to exert "democratic control" on behalf of the whole society (Aucoin and Heintzman 2000; Haque and Mudacumura 2007; Liddle 2007; Heikkila and Isett 2007). But literature shows that when citizens are engaged, their commentary can be highly "framed" to represent their immediate interests, rather the than the interests of the "general public"(Jewell and Bero 2006).
On the side of governments, they have to ensure that the general public as external stakeholders participates in beyond evaluating the outcomes, but in setting goals and informing policy making and implementation (Moynihan and Ingraham 2003). In particular, citizens are anticipated to counter the manipulative behaviors by the bureaucrats in performance measurement, such as exaggerating achievements, building image, and covering up failures (Haque 2007; Halachmi and Holzer 2010). Moreover,
governments face the challenge that the involvement of citizens in performance measurement causes the tension between "upward" accountability and "downward accountability" (Haque and Mudacumura 2007), especially when the interpretations of administrative accountability are contradictory with public accountability, thus pulling public servants in different directions (Poulsen 2009).
Overcoming these difficulties entails the states develop capacity at skills, knowledge, and collaboration layers, and to demonstrate the improved capacity so that citizens are willing to participate (Bovens, Schillemans et al. 2008). For developing countries, integrating citizen engagement and performance measurement is a more daunting task. In these countries, undesirable public accountability is intertwined with power abuse, ineffective corruption control, and lack of transparency (Sarker and Hassan 2010). It is found that local level accountability is constrained by continuing control and power exercised by the state and the ruling party, and the dominance of local bureaucracy over local institutions (Haque 2008). As Head (2011) notes, too tight control from reformist policy entrepreneurs on the reform agenda renders the civic engagement on major issues symbolic rather than substantial. Citizens might be involved in performance measurement by governments, but "the fact that accountability tools can be available does not necessarily imply that these tools are used in the actual relation between government and associated entity (Kruijf 2010)".
China, the largest developing country, has reaped great success in economic development over the past twenty years. However, concerning political advancement, the state is labeled as a "growth without governance and democracy" (Kim 2010), which means that the public administration reform has yet prioritized public participation and citizen empowerment but largely relied on bureaucratic approaches. Recent research from China shows that performance measurement can be developed and used quite effectively to hold local agents accountable to their superiors (Chan and Gao 2008; Chan and Gao 2009). But till this day, little is known about how performance measurement systems are developed to provide access for citizens to hold governments accountable.
Combining in-depth interviews, observations and documents analyses on a policy experiment to engage citizens in performance measurement in Shanghai, the largest municipality in China, this article will show how China's local authorities formulate the citizen engagement procedures, design performance measures, and hold local public sectors accountable to the general public. We will illustrate that while the public accountability mechanism is developed over years, ultimately it is largely "bureaucrats driven". Examining this process helps drawing lessons on what has to be done to make citizen engagement effective in boosting public accountability. The article shall be structured as follows. The first section reviews the literature on the role of citizen engagement in promoting accountability in China. The second section illustrates data and methods. The third section presents the features of the experiment. Its effect on public accountability is elucidated in section four. In the fifth section, we will compare the findings with international studies, and discuss which problems the state should pay special attention to in strengthening public accountability.
RELATIONS BETWEEN CITIZEN ENGAGEMENT AND ACCOUNTABILITY IN CHINA
The definition of accountability depends on its context and administrative structure (Erkkila 2007). In the Chinese language the term "accountability" (wenze) refers to the intergovernmental relationships between organizations at higher and lower layers of the hierarchy, reinforced by responsibility delegation and sanction. Recent studies found that for the accountability mechanisms already used, the common core is for the superiors to exert administrative controls and impose sanctions on their subordinates, with the influence from citizens quite constrained (Chan and Gao 2009; Hsu 2009; Ma 2009; Lu and Xue 2011). Up till now, reaching the targets assigned from above and ensuring that policy priorities announced by higher level be accommodated has been of undeniable primacy in performance measurement (Chan and Gao 2009). In a case study on the township level governments, Gao found a majority of their performance measures (174 out of 278) concerned the accomplishment of economic constructions, followed by public service provision tasks (Gao 2009). Finishing the targets brings the local agents with political recognition and future resource allocation, and failures to accomplish the tasks lead to sanctions.
Although the performance regime is effective in rendering local agents highly responsive to their leaders, research warns that it has several limitations. Gao (2010) finds that the local agents are burdened with many "Non-mission-based performance targets", such as family planning, corruption control, party building, environmental protection, ensuring safety at work, and the handling of mass complaints. She noticed that although these targets reflect social development and public interests, there is an "output/outcome" problem because "even the most rational and energetic efforts by administrators may not produce substantial results where targets are related to intractable problems that are more likely to be resolved by national policies and coordinated, centralized administration". The result is that under a high-pressure environment, local government has a noticeable moral hazard to inflate performance reports, which ultimately hurts public interests. This short-term orientation is found to endanger social equality and stir citizens' dissatisfaction (Saich 2007), which is aimed at governments' irresponsible activities of image building and misusing public funds that caused irreversible damage (Cai 2004). Research reveals that the disappointed citizens are mobilized more easily to confront the local authority (Chen, Lu et al. 2007). Knowing that local agents have the habit of dismissing the annoying voice locally, citizens also resort to their political bosses by lodging the complaints (Ho 2001; Shi and Cai 2006), which accentuates social unrest and loss of institutional legitimacy.
Noting these threatening dynamics, some local authorities in China have engaged citizens to assess the performance of local public sector. The basic rationale is to use citizens as new source of performance information. Information and communication technology (ICT) are widely adopted to realize this goal. In the analysis about the "leaders' mailbox" used by two Yangtze Delta cities, Hangzhou and Nanjing, Hartford (2005) suggests that the use of online system enables citizens to report performance failures and corruption deed, which contributes to making local agents accountable. A research on the 31 Chinese provincial governments in using portals to interact with local residents also shows that citizens' online participation plays a "dual role" of deflating social tension and reestablishing Party legitimacy while pushing the government toward more transparency and accountability (Jiang and Xu 2009). Open discussion with citizens in policy formulation and evaluation is also undertaken. An example is the democratic consultation assemblies (DCA) in Wenling, Zhejiang province. It is found that the implementation of DCA renders the local level cadres appointed by their superiors to have "a heavy personal stake in the...