A common focus in the public policy research on government bureaucracy is how agents sustain and expand asymmetric information to increase their autonomy from political principals on the one hand, and how principals such as lawmakers and political executives reclaim control through enhanced oversight and expanded information supply on the other (Brierley et al. Reference Brierley, Lowande, Potter and Toral2023). Debates about control are centered on the extent to which the principal is able to control bureaucrats despite the significant informational advantage the latter have compared to the former’s modest knowledge endowments (Miller and Moe Reference Miller and Moe1983) and what kind of strategies are more effective in either closing the information gap (McCubbins and Schwartz Reference McCubbins and Schwartz1984; Weingast Reference Weingast1984) or enabling control even as the gap remains (Bendor et al. Reference Bendor, Taylor and Van Gaalen1985). Against this backdrop, less attention is paid to the reverse, where principals possess information that agents do not have – such as information about policy preferences – and choose whether, when, and how to disclose it. Disclosure in this context is highly variable as well. It may be entirely withheld, selectively revealed, regularly signaled, or even oversupplied. This variation in disclosure pattern remains out of theoretical and analytical focus, perhaps because principals are assumed to have little strategic incentive to deny agents such information, making top-down disclosure less conflicted and so seemingly less urgent to theory development.
Several important gaps in the public policy literature have emerged as a result of this imbalance in scholarly focus. First, current models tend to view the principal’s behavior as a localized response to domain-specific activities, leaving factors that transcend multiple domains outside theoretical and analytical focus. Second, the traditional assumption of information scarcity has been challenged by public policy scholarship (Jones and Baumgartner Reference Jones and Baumgartner2005). Instead, information processing has opened a less explored causal pathway: instead of mitigating information asymmetry, increased disclosure may congest communication channels, heighten ambiguity, and hinder agents’ ability to act on signaled preferences (Jones et al. Reference Jones, Boushey, Workman, Guy Peters and Pierre2006). Third, major public policy theories on bureaucratic politics continue to have limited portability across regime types (Chan Reference Chan2024; Brierley et al. Reference Brierley, Lowande, Potter and Toral2023). In particular, without the “institutions of knowledge” to mandate “clarifying…. interests, imposing penalties for lying, [and] introducing the threat of verification” (Lupia and McCubbins Reference Lupia and McCubbins1998, p. 227), political regimes can alternate between frequent pronouncements and extended, calculated silence. As this variation makes preference disclosure a relevant driver of agent behavior outside of the liberal democratic setting, we draw upon the public policy theory of information processing to clarify the politics of disclosure and to trace its downstream impact on regulatory behavior and outcomes.
This study draws on two decades of regulatory work in China’s film industry to bridge these research gaps. This timeframe involves two successive governments and can be divided into two distinct eras (Shambaugh Reference Shambaugh2021; Shirk Reference Shirk2018). We refer to the first era (2002–2012) as the Contested Period and the second era (2012–2022) as the Established Period, in accordance with the prevailing typology in comparative politics that differentiates between modes of regime power sharing (e.g., Boix and Svolik Reference Boix and Svolik2013; Wilson and Woldense Reference Wilson and Woldense2019). The former maps to a contested regime where power is shared among factions in the ruling elite, while the latter represents an established regime where power sharing is circumscribed under regime consolidation (Svolik Reference Svolik2009). During the earlier period, governance emphasized broad coalition-building within the ruling structure. In the later period, governance became more centralized around a unified leadership core, reflecting a natural evolution of administrative coordination.
We took steps to develop an empirical strategy that acknowledges both the relatively closed and institutionally insulated nature of policy process and the role of subjective preferences in cultural governance in particular. First, the internal workings of China’s bureaucracy are difficult to observe due to limited transparency and institutional insulation (Reny Reference Reny2016). Given these constraints, we sample interviewees selectively so that the insights into the regulatory process encompass distinct institutional and sectoral vantage points. Second, the tension among leadership preferences, disclosure patterns, and regulatory discretion identified in the propositions becomes particularly pronounced in the field of film regulation, underscoring the case’s value for theory development. Subjective preferences carry greater weight in cultural domains than in areas governed by biophysical or professional constraints, such as public health or natural disaster management, thus reducing the case’s “resonance” with areas besides the arts (Steinberg Reference Steinberg2015).
A set of propositions follows, specifying first how regime politics reshapes preference disclosure, and second, drawing on information processing theory, how regulators respond to the pattern of disclosure and adjust behavior in ways not anticipated by principal–agent models. Original and secondary data from field interviews and other sources are then reviewed against each proposition. We consider possible confounders and alternative explanations before providing the conclusion.
Information processing in preference disclosure and regulatory behavior
Telling the agent what the principal wants may seem essential for effective control, but public policy research on this topic has shown that keeping the agent less informed about the principal’s true preferences sometimes improves control. Not only do principals lack the information to specify preferences against ground-level conditions, but their ability to signal what they want can also be constrained by strategic reasons. Information gaps enable agents to exploit their “superior knowledge… regarding the [principal’s] demand for its services” in deceptive representations that “underestimate the benefits, and overestimate the costs” of those services, to which the principal responds by “[keeping] the [agent] in the dark about its own preferences” (Bendor et al. Reference Bendor, Taylor and Van Gaalen1985; also Miller and Moe Reference Miller and Moe1983). The principal may also find direct preference disclosure suboptimal when confronted with “wicked problems” where “the relationships among variables are not linear and small shifts may produce large differences in the outcomes,” rendering “more open and procedural instruments… more appropriate… than more definitive interventions” (Peters Reference Peters2017, pp. 386–394; also Howlett Reference Howlett2000).
Our research extends this public policy debate by challenging the view that preference disclosure is a largely direct response to noncompliance and best examined through the lens of optimization (e.g., Bendor et al. Reference Bendor, Taylor and Van Gaalen1985). This frame offers limited explanatory power in contexts where disclosure decisions respond to factors external to principal–agent interactions. These factors, while also strategic in nature, do not primarily concern what the agents do, nor do they condition subsequent disclosure decisions made by the principal. This decoupling is crucial; patterns of disclosure that diverge from expectations under the optimization frame can emerge. As the propositions in the next section suggest, more frequent signaling of preferences by political leaders may heighten uncertainty for bureaucrats and, through mechanisms not fully spelled out in the current literature, lead to administrative inaction and inefficiencies.
We draw on China studies to shed light on these mechanisms, especially the literature on the bureaucracy. Existing research depicts a leadership struggling to rein in the administrative apparatus even though it is composed of what Wu (Reference Wu2012) calls “upwardly accountable” officials. In some policy domains, exchanges are often infrequent, making information transmissions difficult to verify (Tanner and Green Reference Tanner and Green2007). In others, information distortions persist despite efforts to step up oversight (Manion Reference Manion2004; Landry Reference Landry2008), undermining policy enforcement (Mertha Reference Mertha2005) and weakening the impact of feedback (Heilmann Reference Heilmann2008). A common theme across this set of work is the interplay between principals and agents over processes and outcomes with a specific domain in analytical focus – such as Tanner and Green (Reference Tanner and Green2007) on public security, Mertha (Reference Mertha2005) on intellectual property, and Manion (Reference Manion2004) on anti-corruption. The domain‑centric approach in China studies excludes in‑depth consideration of regime dynamics as a factor cross‑cutting individual domains, replicating the broader problem in the public policy literature.
Regime dynamics and preference disclosure
Regime survival in contested regimes is contingent on practices that sustain trust in the power-sharing arrangements. A critical feature of collective leadership is that the ruling elites can credibly threaten to overturn the leader’s decisions and even contest their hold on power. At the same time, the inherent weakness of the arrangement may motivate leaders to exploit any opportunities to sideline their one-time partners. The stability of the co-optative arrangement between the two parties thus depends on the leader’s ability to avoid actions that may weaken the collective trust required for effective and stable power sharing among opportunistic allies. Because frequent, extended discussions of policy preferences could cause concerns for possible overreach from the top, contested leaders tend to be reticent and withdrawn, only to break silence when more explicit forms of intervention from the top can no longer be delayed.
The calculations surrounding what and when to communicate regulatory preferences can be vastly different for leaders who are sufficiently established to act independently and to expect general subservience from the ruling elites (Svolik Reference Svolik2012, p. 78; also Svolik Reference Svolik2009). The established leaders often takes the very actions to show attitute clearly in both political and operational terms. Even if leaders operationally “depend on their administrators as a collective,” actions that underline their capability and willingness to intervene directly and personally – part of which includes frequent discussions of their policy goals, interests, and priorities – help prevent the “public perception that any individual administrator is indispensable” (Svolik Reference Svolik2012, p. 80). Here, pertinent to our argument is the observation that preference disclosure is arguably divorced from agency control as one would infer from standard models. Rather, it is anchored to the logic of regime survival (Timoneda Reference Timoneda2020).
Our first proposition links disclosure to leadership styles described in the literature on regimes:
Proposition 1: Established regimes signal preferences more frequently and extensively than contested regimes.
Apart from keeping bureaucrats informed, signals from the top provide the political backing they need to coordinate complex actions on the ground. In contested regimes, such signals can come from sources other than the central leadership, such as senior administrators and other key political figures who form part of the collective leadership. In established regimes where power sharing is largely circumscribed, the bureaucratic apparatus is selectively responsive, prioritizing signals that come directly from the highest level of the political hierarchy. Frequent and direct preference disclosure is an informational response to the surge in administrative dependence under the conditions of power consolidation.
Regulatory decision-making and outcomes
Our next set of propositions lays out what bureaucrats do in response to regime-driven divergence in the pattern of preference disclosure, starting with top-down signals containing out-of-domain concerns. In principle, all policy issues are interconnected, but in practice, they are grouped into specialized subsystems to make administrative routines more efficient and to bring transaction costs down (Simon Reference Simon1962). While bureaucrats normally prioritize issues in their own subsystem over developments in other policy domains, they might have to pay closer attention to developments in other related areas and adjust their own regulatory practices when the interconnectedness between what they specialize in and issues in other fields is highlighted, which occurs from time to time due to arbitrary reasons (Burstein Reference Burstein1991; Kingdon Reference Kingdon1984; Baumgartner and Jones Reference Baumgartner and Jones1991). For example, prior to the 9/11 attacks, the US bureaucracy addressing security threats was highly fragmented. Cybersecurity fell under the Department of Commerce, while natural disasters were managed by Federal Emergency Management Agency. The attacks heightened the interconnectedness of the security threats that these agencies had to deal with, driving the US government to respond with the creation of a single agency to oversee a more integrated and coordinated approach (May et al. Reference May, Sapotichne and Workman2009). Short of drastic reorganization, bureaucrats are also capable of significant changes in policy coordination in response to signaled changes in central priorities (Zhan and Zhu Reference Zhan and Zhu2022).
Frequent preference disclosure is hypothesized to trigger a similar surge in attention toward issue interconnectedness. Unlike bureaucrats who operate within specific subsystems, top leaders are generalists who consider issues in the context of broader policy objectives. When leaders discuss issues outside of their specialty, bureaucrats must handle their regulatory work with greater caution lest their actions should trigger unintended consequences for other domains and contradict the signaled preferences. The more leaders discuss their preferences, the more time and effort it takes for bureaucrats to come up with regulatory actions that would complement the overall vision.
Proposition 2.1: In established regimes, bureaucrats pay closer attention to generalist issues that lie out-of-domain and to the extent within-domain choices are aligned with them.
Proposition 2.2 links preference disclosure to preference formation (Druckman and Lupia Reference Druckman and Lupia2000). In earlier research, preferences are seen as highly stable because they are formed cumulatively through “an exhaustive memory search and weighting of evidence” against “each like and dislike equally” until a preferred option that “has the greatest net number of favorable attitudes” can be found (Kelley and Mirer Reference Kelley and Mirer1974, p. 574). But recent studies reveal important sources of volatility in people’s preferences. The so-called accessibility models argue that preferences are formed following “considerations that are immediately salient or accessible” (Zaller Reference Zaller1992, p. 49), which in turn depend on just how recently different considerations pertinent to a given preference have arisen following exposure to external events (Iyengar and Kinder Reference Iyengar and Kinder1987). A separate line of research shows that sudden and radical preference reversals can take place through the manipulation of framing and in the absence of material change in utility (Tversky and Kahneman Reference Tversky and Kahneman1991). In other words, preferences can change even if the objects over which they are formed stay effectively the same, making at least some instances of preference formation inherently unstable.
Some of the mechanisms leading to perceived instability in preference signals are already anticipated in the related public policy research on information overload (e.g., Walgrave and Dejaeghere Reference Walgrave and Dejaeghere2016). If preferences are prone to shifts and reversals as suggested, the inherent volatility of policy preferences and the perception of inconsistent signals from the central leader can intensify risk aversion among bureaucrats. Frequent communication of preferences increases the salience of whatever regulatory considerations are most accessible to the top leader at a given moment, leaving established regimes more susceptible to instability in regulatory preferences. In contrast, leaders who govern through elite co-optation are accountable to a broader range of stakeholders whose support they must retain, providing them with strategic incentives to avoid frequent disclosure and subsequent reversal of signaled preferences. By sending fewer signals, contested regimes are less exposed to perceived preference inconsistency.
Anticipating that leaders in established regimes may articulate divergent or wholly new regulatory preferences in subsequent communications, bureaucrats tend to adopt a more cautious, wait‑and‑see stance, postponing enforcement until the durability of leaders’ preferences becomes apparent. Bureaucrats are also aware of the risks of committing to measures that leaders may suddenly deem inadequate. They may see greater strategic value in incremental and measured changes over more drastic movements, allowing them to accommodate different preferences while leaving room for adaptive reversals when these preferences are revised again. Overall, increased disclosure of preferences could lead to important changes in the pace and scope of regulatory activities under established leaders, characterized by smaller but more frequent adjustments.
Proposition 2.2: Bureaucrats exhibit more hesitancy toward regulatory changes and take longer to assess options under established regimes as preferences are perceived to be more unstable.
Our third proposition suggests that frequent communication of preferences can prevent bureaucrats from making efficient trade-offs, without which the regulatory actions become less focused. Because policy regimes are increasingly complex, policy changes in one aspect can potentially send “waves of repercussions that ripple through such systematic networks” (Rittel and Webber Reference Rittel and Webber1973). For instance, a development policy that places increasing emphasis on pollution control and sustainable practices will improve environmental conditions and urban livability if implemented effectively, but at the same time the new effort may cause some streams of investment to fall, particularly in industries that generate high levels of pollution, making the regulatory turn toward sustainability potentially at odds with other developmental goals such as job creation and industrial competitiveness. Where the interdependencies are difficult to accommodate simultaneously, even compliant and disinterested administrators can fail to sufficiently align their actions with the leader’s general policy preferences.
Our focus departs from analyses that attribute regulatory misalignment to variants of the agency problem (Chan Reference Chan1994; Lorentzen et al. Reference Lorentzen, Landry and Yasuda2014). Instead, the information processing perspective concerns situations where regulators hesitate to prioritize among competing goals due to increased uncertainty about the policy mandate. Traditionally, the agency framework applies when the expected benefits of noncompliance outweigh the risk of detection and sanction; noncompliant agents can be dissuaded once oversight or penalties close this gap. In comparison, preference disclosure shapes regulatory activities by reshaping information supply. When preferences are revealed with restraint, the limited agenda space forces marginal goals to be excluded, which sharpens the overall message. However, when this restraint is lifted, a broader range of goals moves to the forefront, making it more difficult to discern the relative significance of competing objectives. Under the latter pattern, an agent’s ability to make regulatory trade-offs is constrained, even if the agent intends to comply out of loyalty or fear.
This problem is expected to become more acute in established regimes. As preference disclosure is less restrained under established leaders, more goals will be brought up, preventing bureaucrats from making critical trade-offs. When comparing their experiences, we expect officials to link the style of disclosure to how they assess and make policy trade-offs on the ground. In particular, we expect officials to link the less restrained forms of preference disclosure in established regimes to a more cautious approach to potential trade-offs, and show greater hesitancy and blame avoidance if the choice involves more critically opposed values or interests.
Proposition 2.3: Bureaucrats under established regimes avoid concentrating on specific goals and hedge their risks by pursuing a broad set of goals simultaneously.
Data
Our qualitative analysis draws on eight elite interviews conducted in the early 2020s with individuals familiar with China’s film industry. Two interviewees with regulatory experience talked about the administrative processes involved in film regulation (Interviewees #2 and #5). To broaden the scope beyond government sources, we interviewed a film director (Interviewee #4) and a professional from a film policy think tank (Interviewee #1). We also consulted two research academics whose work involves understanding the broader political and regulatory themes industry practitioners tend to overlook: a professor at a major film academy (Interviewee #3) and a film professor based in Beijing (Interviewee #6). Two interviews were later added to better represent regional views: one expert specializing in co-produced films (Interviewee #7) and another expert from a regional filmmaking hub (Interviewee #8).
Our sampling strategy employed purposive selection, stratified to represent three elite groups occupying different institutional and sectoral roles: officials involved in policy design, regulators responsible for enforcement, and industry professionals engaged in production and training (e.g., film directors, film school professors). This spread enabled triangulation across distinct vantage points and allowed potential inconsistencies in event recall to be detected. We utilized a semi-structured format that began with general questions about film censorship in the two periods. We prompted our subjects to describe the decision-making processes of bureaucrats in regulatory enforcement, the reasons for changes over time, and whether the switch from Contested to Established leadership was a factor, as stated in our central propositions. Their answers would help us empirically assess how much the general incentives and mechanisms in their descriptions are aligned with our propositions. In addition to the central propositions, we explored alternative explanations with reference to other contextual drivers such as China’s entry into the World Trade Organization and invited additional input on the role of other contingencies.
We used indirect questioning to support a nuanced discussion of complex institutional dynamics. For example, some questions were phrased in the third person, allowing the interviewees to analyze incentives and actions in hypothetical terms. This approach provided the necessary distance for subjects to respond without revealing details about specific individuals or events they might not wish to share. We removed details on specific individuals, organizations, and events brought up in the interviews. We follow the example of other qualitative studies and offer case descriptions without identifying information (e.g., Yan Reference Yan2016; Tsai Reference Tsai2007).
Film regulation in China
The film industry in China has traditionally faced a higher level of regulation than other countries, but the way it is managed has transformed over time. In the early decades, film production was directly managed in accordance with the Measures for Screening Scripts and Films. Film studios were required to submit plot outlines and scripts for review before production. Once the films were shot, they had to be reviewed once more before getting approval for distribution. In such an arrangement, what the studios could produce would be tightly connected with the prevailing policy priorities, and whether a film was approved at the end would depend on informal lobbying targeting very senior policy actors in the administration. Before the 1970s, industry regulation was directly enforced by these actors, who would specify changes to the plot and artistic elements as part of the review process (#4).
Following the introduction of the Film Regulation Ordinance of 1996 and Film Review Measures of 1997, China’s film industry regulation has gradually moved toward a rules-based model. On top of the separate committees created to screen and review films, the Ministry of Film, Radio, and Television came to take charge of the whole review process and implement licensing according to the formal standards and procedures specified by the Film Regulation Ordinance. Another period of regulatory transition came in 2001 when China joined the World Trade Organization, during which the rules-based process was strengthened to accommodate new market entrants and foreign investors. Further regulatory change during the 2000s gave provincial regulators more discretion, in an effort to localize the review process (shudi shencha) and to simplify the review procedure for films that do not touch upon topics of national interest. In the mid-2010s, the National People’s Congress adopted the Film Industry Promotion Law, formalizing China’s long-term transition toward more rules-based practices for film censorship (#3).
While these developments have made film industry regulation more institutionalized, top leaders continued to set the policy priorities for the industry. In the Contested Period, the central leadership saw the sector as an integral part of its strategy for cultivating “cultural soft power” (wenhua ruan shili). The emphasis on meeting international competition in cultural arenas led to increased policy protection and government spending for domestic players in the film industry, grooming them for the Chinese and overseas markets (#5). In the Established Period, the central leadership pursued a vision for China’s film industry as part of the broader national security agenda. Defined in terms of heritage protection and cultural self-confidence (wenhua zixin), “cultural security” became one of the principles onto which the leadership’s approach was anchored.
At the same time, the regulatory bureaucracy handling films became far more exposed to preference signals in the Established Period. Referring to the Contested Period, Interviewee #4 identified specific policy directives that came from decision-makers. While leaders then would talk about policy visions, few direct instructions were made on a day-to-day basis (#5). In the Established Period, decision-makers showed more interest in bringing the film regulator closer (#3). Originally overseen by a general government department, film regulation was absorbed into the unit specialized in political publicity, thus placing it under the direct charge of the leadership. In this new setting, communication with regulators included “not only public but also private policy instructions” and occurred at a frequency that can be seen by some to be “high” (#5).
The pattern of preference disclosure shifted in two important ways. First, both the scope and the frequency of disclosure expanded noticeably during the Established Period. Second, communications in this later period often appealed to a broader vision of security that cut across domain boundaries, whereas disclosures in the Contested Period tended to anchor preferences in the issues specific to films or the cultural sector.
Empirical support for Proposition 1: Film regulators became exposed to more frequent and detailed information about policy preferences in the established period
Widely perceived as a “consensus builder” (Brown Reference Brown2010), the central leadership during the Contested Period represented “the reign of always nice people” comprising “largely inexperienced and weakly networked technocrats” (Shih Reference Shih2022: 175). The top leadership at the time governed as “first among equals” (Shirk Reference Shirk2018: 32) and strictly followed the principles of collective leadership to preserve the negotiated stability among elite coalitions (Bo Reference Bo2007; Lam Reference Lam2006). Aside from gestures, analysts cited several structural changes of the leadership framework in that period as concrete steps toward a co-optative model of governance (Miller Reference Miller2008; Shirk Reference Shirk2018).
The Established Period reversed earlier trends through reforms aimed at rooting out corruption and reinforcing Party discipline, which substantially weakened elite factions and their patronage networks (Shirk Reference Shirk2018, p. 24; Dickson et al. Reference Dickson, Landry, Shen and Yan2016: 34). Policy coordination became more concentrated across the Leading Groups (Shambaugh Reference Shambaugh2018; Chan and Fan Reference Chan and Fan2021), enabling policy campaigns such as the “Chinese Dream,” the Belt and Road Initiative, and “Made in China 2025.” This coordinated posture departs from the decentralized approach that characterized the Contested Period (Naughton Reference Naughton2013). Key leaders became more directly involved in administrative decisions, issuing frequent, detailed instructions and seeking to align disparate sectors with a unified policy vision (Miller Reference Miller2017; Baranovitch Reference Baranovitch2021; Shirk Reference Shirk2018).
The divergence in disclosure patterns suggested in Proposition 1 carries observable implications. In the Contested Period, the central leadership did not “spend that much time managing [films]” (#5), favoring “laissez-faire rule” and “going with the flow” (#3). Regulators had considerable discretion insofar as the “predetermined pace” and “yesterday’s direction” set out by existing practices were followed (#3). Interviewees highlighted a marked shift during the Established Period, noting that communication became more engaged (#1, #3). Films under consideration for distribution licenses were, at times followed by updates to the guidelines (#2). An accurate reading of one’s mandate was critical to regulatory work, particularly in the Established Period. Both Interviewees #7 and #8 described related changes among regulators: Whereas “they mainly relied on laws, regulations, and formal procedural requirements to assess whether projects were qualified [in the Contested Period], now they must also take into account executive directives, the spirit of internal meetings, and leaders’ speeches” in approving applications, reflecting a shift from purely rule-based implementation to a more integrated approach that also incorporates leadership guidance and internal policy signals (#8). Clarifying the “red lines” or “bottom lines” became critical to the job, even when the information was communicated only internally (#7).
Empirical support for Proposition 2.1: As the information regulators received was expanded to include discussions about broader policy goals, they prioritized out-of-domain goals over artistic value, market potential, and other filmmaking considerations
The concept of “holistic management perspective” (quanjuxing kaolu) became prevalent in the Established Period. The trend toward a more generalist mindset was driven by, first, an increased use of “expressed opinions and rewards” by the central leadership to direct the film industry and, second, the inclusion of a wider range of “provincial authorities across multiple stages of review” (#1). Regulators now had to base their decisions on “the thoughts of the leadership” because they revealed the wider policy and regulatory patterns beyond film regulation (#1). “When the government agenda is dominated by milestone events,” Interviewee #3 noted, “the effective delivery of the government’s message is to be ensured against any competing or even contradictory themes in films.” Another interviewee similarly observed that “the intricacies and uncertainty involved in the cultural industry are positively correlated with the growing complexity of international environment,” as a result of which the leadership in the Established Period sought a more coherent, unified response straddling different domains (#1).
With the transition, regulators no longer viewed themselves as “administrative managers… whose core work is industry development.” Instead, their role was now of the “publicity cadres… in charge of political gatekeeping (zhengzhi baguan)” and safeguarding the “overall environment” (Interviewee #8). That was accompanied by a surge in risk aversion. According to Interviewee #3, “when sensitive issues outside the film domain are involved,” some regulators responded by “deferring to the judgment of the central government officials.” More generally, “avoiding mistakes in every conceivable way… has emerged as the fundamental principle of action” – or inaction – for many regulators. Our discussion with Interviewee #6 reveals that film productions increasingly converged on a number of key policy themes, such as inclusive growth (also #3). Regulators faced heightened uncertainty following the new emphasis on out-of-domain considerations, even though the surge in “expressed opinions and rewards” associated with established regimes (Proposition 1) should have reduced it.
At the same time, complications arose when out-of-domain considerations came into conflict with merits that domain experts traditionally focused on. On this point, Interviewee #5 brought our attention to a film that chronicles the events in the early decades of the twentieth century. Initially approved by regulators for both its award-winning cast and cultural significance, the film subsequently became embroiled in controversy. The license was not rescinded, but the film performed poorly at the box office as its publicity was subsequently contained. It was also suggested that even though similar criticisms about particular films were rarely made public, they would be informally shared and then worked into industry guidelines following internal discussions.
Consistent with Proposition 2.1, Chinese regulators in the Established Period spoke of an unusually intensive effort to infer central priorities, even on issues only tangentially related to film. They sifted through an expanding stream of directives, mandates, and speeches, and also probed the informal views that circulate through professional networks. Market reception, artistic value, and other filmmaking considerations played a greater role in the regulatory process in the Contested Period.
Empirical support for Proposition 2.2: An increase in information supply about loosely coordinated policy preferences intensified speculation among regulators as to what to promote or censor
Our interviews revealed a shared sense of increased uncertainty about the parameters for film regulation in the Established Period. As comments on regulatory activities were made more directly, “the guidelines evolve more dynamically, reflecting the leadership’s effort to respond to changing cultural and security priorities” (#3). Interviewee #5 made the observation that “a considerable number of decisions instituted could undergo drastic revisions, followed by an entirely new set of rules.” In sum, the supply of “abundant signals, rather than reticence, is the main contributing factor” to increased uncertainty (#5).
Pertinent cases were discussed in the interviews. Films about major international conflicts are generally restricted due to the sensitivity of the issue to both domestic and overseas audiences. Regulators sometimes adjust the restrictions in response to geopolitical contingencies, although such special accommodations can be withdrawn, even during post‑production, in short notice (#3). Following the release of a dramatized account of the lives of imperial consorts in the early 2010s, historical dramas on palace intrigues became a big hit with viewers. However, once the trend drew attention in state media, “stringent quotas on screening and production were imposed,” and “the genre largely fell off from the radar” (#1). Interviewee #4 described a similar abrupt shift when depictions of adultery, a common theme in soap operas, were suddenly banned. These examples highlight how regulators transmitted volatile signals and how industry actors struggled to interpret them.
The high level of dynamism revealed in preference disclosure fueled speculation even among regulators (#5). To manage the impact, regulators offered more guidance as well as incentives to projects aligned with “core” themes, while signaling to filmmakers that “sharp criticism and artistic exploration” could lead to extensions of the review cycle (#7). Industry professionals, however, experienced this surge in information as a constraint: their ability to balance compliance with commercial success was narrowed by the sheer volume and volatility of signals (#1). In response, filmmakers began to adopt information‑seeking practices, such as requesting “preliminary feedback (modi)” on draft screenplays. This was not an attempt to secure “special approval (tepi)” but rather to obtain some kind of “recognition (renzheng)” for the chosen themes and storylines (#8). Apart from a few exceptions, our interviewees found the scope and direction of preferences disclosed during the Contested Period to be relatively consistent. This consistency had enabled middle‑tier regulators to act more decisively and filmmakers to rely on information available locally from immediate regulatory and industry contacts, rather than searching across extended networks (#3).
These developments lend support to Proposition 2.2. In the Contested Period, administrators came to expect infrequent preference disclosure. That expectation meant film regulators could operate within parameters that they consider stable, even when particularly controversial cases were concerned. As disclosure grew in frequency in the Established Period, these parameters were seen as less stable. Regulators became less inclined to exercise discretion as they once did, instead focusing on general measures they believed could help industry partners navigate the perceived increase in uncertainty.
Empirical support for Proposition 2.3: Following the surge in information supply, policy preferences appeared less clearly ordered by priority, leading to decisions being deferred upward or responsibilities being passed around
The last proposition concerns how regulatory goals are ordered in priority. The current leadership states that films should “lead the way in promoting culture and values by means of artistic expression” and, more specifically, “represent China’s narrative on significant historical eras accurately.” This implies that films should deal with topics and issues approved by the regime. In the same speeches, filmmakers were urged to “create mainstream works embraced by the masses” and “address the constantly evolving demands of spiritual and cultural life.” With political objectives and audience preferences equally emphasized, how regulators should enforce content favored by the regime while also attending to unavoidable tensions with market-driven and artistic considerations has become less clear (#5).
When asked for specific examples of how the conflicting goals play out, several interviewees brought up another war drama in the early 2020s as an example. While regulators acknowledged the film’s artistic merit and commercial viability in internal discussions, other concerns were raised. The film depicted a contested period in the country’s modern history and featured elements that could be seen diverging from the mainstream narrative. Regulators found themselves caught between artistic considerations and macro policy objectives. According to Interviewee #3, relevant officials attributed their caution to the challenge of balancing an expanding array of priorities. The film underwent an unusually long review process lasting over a year before approval was granted (#5).
This case is representative of a broader trend of cascading deferral and circumvention as ambiguity increased. Regulators “report upward and wait for directives” (xiangshang qingshi, dengdai zhiling), offloading responsibility up the chain of command; those in higher levels defer decisions further upward (#8). The same interviewee spoke of the “voluntary over‑compliance” (ziwo jiama) phenomenon, by which he meant the “lowest-cost” and “most optimal” strategy for local regulators to meet expectations through “excessively refining (jizhi xihua) central mandates and eliminating potential issues by pre-emptively lengthening lead time [for the review cycle] (tiqianliang paichu).” As a result, although applications are formally required to be processed within 20 days, delays stretching beyond a year were commonplace in the Established Period (#3). Regulators respond to the absence of explicit and unequivocal directives with tactical delays in the hopes that the approval can be made by the higher-ups (#3).
In the Contested Period, regulators operated within an informational context that reduced cognitive burden because commercial viability and artistic acclaim were clearly prioritized, creating a hierarchy that simplified regulatory decision-making. This clarity meant that “administrators at that time… could align their personal goals with the preferences of higher authorities.” However, in more recent years, the tendency to “procrastinate and pass the buck” (#3) became more common following expanded disclosure that emphasized a wider array of goals but also made it more difficult to rank them by priority. When updates were relatively infrequent, regulators operated with a more localized, focused view on goals with which they identified, a dynamic that the current pattern of disclosure seems to have fundamentally altered.
Our interviews corroborate Proposition 2.3. Infrequent preference disclosure in the Contested Period left the hierarchy of goals relatively well-defined. Regulators had a more workable sense of which goals are prioritized and which ones are secondary, and that certainty gave them more confidence in handling the applications even if their decisions did not always sit well with some important but lower-priority goals. Responding to expanded disclosure in the Established Period, film regulators increasingly escalated decisions to manage compliance risks.
Confounders and alternative explanations
Our discussions covered additional factors that the interviewees regarded as complementing, rather than competing with, the information‑processing mechanisms linking behavior to preference disclosure. They are reviewed as follows.
Credibility of noncompliance sanctions
In the context of contested regimes, the central leadership prefers not to “rock the boat” because threats of sanction would undermine confidence in its commitment to power sharing. With less top-down intervention, bureaucrats can get away with decisions that do not fully align with the leadership’s preferences. In contrast, the scope of agreement would be much narrower in established regimes. Minor misinterpretation of the preference signals, even if entirely attributable to uncertainty, would invite sanctions from a powerful principal. Speaking from a historical vantage point, an interviewee pointed out that to understand film regulation is to appreciate the bureaucracy as “an executive body in perpetual alignment with the higher authorities (xiang shangmian kaolong).” Therefore, working under a decentralized power structure, regulators “could shield industry from top-down demands selectively; following centralization, everyone looks upward, and as soon as there is a task from above, it is given priority over all others” (#1).
The subjective nature of films
Risk-aversion was also amplified by the subjective nature of films. In cultural administration, policy judgments tend to be highly subjective and, according to Interviewee #1, often “involve very complex power discourses – very complex.” As a strand of art, cinema “inherently encompasses numerous gray areas for which rigid standards do not exist” and so its regulation can be “a highly subjective thing; when views clash, you may expect strong reactions [from the higher-ups]” (#3). Although our interviewees could not recall specific film regulators being disciplined for enforcement shortfalls (#5), dismissals in other sensitive areas, underlined the credibility of sanctions and inspired more stringent practices in film regulation (#1).
Policy tightening and shifting regulatory priorities
Regulators had a greater degree of freedom to deviate from top-level instructions in the Contested Period. The administration at that time understood the subjective nature of film regulation (#3), with multiple interviewees suggesting that domain-specific interests would occasionally be prioritized over central goals (#3, #5). These patterns were upended during the Established Period by the sharp concentration of power interacting with the subjective nature of films. Under these conditions, the film sector was increasingly treated not only as a cultural industry but also as a strategically important area of public communication (#8). Regulators adapted to this environment by acting more cautiously, for example, by withholding preliminary approval and avoiding definitive assurances of eventual approval when requesting revisions (#8).
Conclusion
Why are bureaucrats less certain about how to act when more information is made available? Why do principals increase information supply even as regulatory inefficiencies follow? Not only is the phenomenon counterintuitive, but it also runs counter to existing frameworks where the principal’s decision to disclose or withhold preferences is analyzed as a strategy to motivate compliance among agents and ultimately improve regulatory enforcement. Put differently, existing principal–agent models share the view that preference disclosure by leaders can be largely understood as part of the bargaining game with bureaucrats. Within this frame, what truly matters is how effective the agent is in maintaining asymmetric information over the principal, while information concerning the principal’s preferences should be relatively stable and less consequential to outcomes.
Yet our work reveals that preference disclosure can vary drastically and lead to distinct regulatory outcomes through mechanisms suggested by information processing theory. Following the surge in information supply under regime consolidation (Proposition 1), regulatory actions do not necessarily become more efficient. Instead, regulators may divert more attention to out-of-domain factors (Proposition 2.1), grapple with a greater number of diverging goals (Proposition 2.2), and contend with fewer clues about priority as the scope of disclosure expands (Proposition 2.3). Often considered a proxy of greater top-down control in public policy, regime consolidation at the top may also contribute to more processing delays, inertia, and other downstream inefficiencies in regulatory activities on the ground (e.g., Scoggins Reference Scoggins2021). To our knowledge, this study offers the first systematic account of the informational pathways that generate these counterintuitive effects.
These findings advance recent theorization of the public policy process across regime types (Li Reference Li2023; Mattingly Reference Mattingly2020; Yasuda Reference Yasuda2021) and align with work on policy advocacy (Li and Weible Reference Li and Weible2021; Grömping and Teets Reference Grömping and Teets2023), agenda setting (Jones and Baumgartner Reference Jones and Baumgartner2005), and policy networks (Teets Reference Teets2018), though important limitations warrant further research. In film regulation, as in other areas of the cultural sector, preferences regarding casting, themes, plot lines, and other artistic choices are highly subjective and are not constrained by biophysical limits or professional norms. Future research should examine whether similar patterns also emerge in fields that share fewer of these “resonance group” characteristics (Steinberg Reference Steinberg2015). Research should also examine how behavioral changes triggered by preference disclosure aggregate into broader patterns identified in mainstream public policy scholarship, such as agenda change (Jones and Baumgartner Reference Jones and Baumgartner2005), attention allocation (Boydstun et al. 2014; Hoerner and Hobolt Reference Hoerner and Hobolt2020), and policy responsiveness (West Reference West2004; Yackee Reference Yackee2006). In terms of practical implications, potential solutions include establishing norms that optimize the pace of internal communication and reconcile contradictions across disclosures, and future research can assess the extent to which such efforts might help.
Data availability statement
The study is based on interviews. The transcripts are not publicly available.
Acknowledgments
The project was funded by National Natural Science Foundation of China (72274104,72522004), Tsinghua University Initiative Scientific Research Program and the Research Grants Council of Hong Kong (17616217). The authors gratefully acknowledge the anonymous reviewers for their valuable feedback.
Competing interests
The authors declare no competing interests.
Open access
