Why Do Autocrats Disclose? Economic Transparency and Inter-elite Politics in the Shadow of Mass Unrest

Abstract

Autocratic governments hold a preference for opacity. Autocracies are less transparent than democracies and a closed informational environment preserves autocratic regimes from mass unrest. Yet, autocracies vary widely in the extent to which they disclose economic information. In this article, we offer an explanation for why some autocrats choose to disclose. We contend that, paradoxically, some autocratic leaders may benefit from increasing the capacity of the populace to mobilize. In so doing, autocratic leaders threaten rival members of the elite, reducing the risk of elite challenges and increasing their freedom of maneuver. We contend that transparency acts as one mechanism toward these ends. We formalize these intuitions and demonstrate empirically that leaders in transparent autocracies enjoy a reduced hazard of removal via coup relative to their opaque counterparts. Personalistic dictators and entrenched autocrats—who suffer the smallest risk of sanctioning by their elites—are particularly unlikely to disclose information.

Keywords

transparency autocracy coup protest political economy political survival

Governments have a taste for opacity. Limited transparency facilitates rent seeking and increases leaders’ freedom of maneuver (Adserà, Boix, and Payne 2003; Brunetti and Weder 2003; Ferraz and Finan 2008; Reinikka and Svensson 2003).¹ Moreover, collection and dissemination of economic data entails substantial financial costs. Presumably, leaders who face little pressure from the public to disclose would eschew such costs and provide less information about their actions or aggregate economic outcomes.

Transparency’s role in facilitating mass mobilization under autocratic rule stacks the deck still further against disclosure. Government disclosures—particularly of economic information—help members of the populace form shared beliefs regarding their leaders’ performance, making mass mobilization against the regime more likely. Hollyer, Rosendorff, and Vreeland (2015) demonstrate that autocratic regimes that disclose large volumes of information are more prone to collapse, due to mass protest or processes leading to democratization, than regimes that fail to disclose.

Yet autocratic governments do disclose information to their publics. While democracies, on average, disclose substantially more than autocracies, the latter vary considerably in the extent of their opacity.

We provide an explanation for why autocrats choose to disclose. We contend that, even as disclosure renders autocratic regimes more vulnerable to mass protest, it insulates autocratic leaders from opposition that emerges from within the regime. Those leaders more vulnerable to rival elites use the threat of mass insurrection, enhanced by transparency, to keep those rivals in line; those leaders less susceptible to coups will be less transparent.

We construct a model in which autocratic leaders balance risks arising both from rival elites and from the population. Attempts by members of the elite to oust their leaders may act as a focal point for mobilization by the masses. Mass mobilization threatens elite survival, since it may result the sweeping away of many, or all, members of the incumbent regime. In acting against their leadership, therefore, members of the elite jeopardize their own privileged positions. Leaders may have an incentive to facilitate this collective action by the masses, insofar as this induces quiescence from rival elites. Yet, leaders face a risk in taking such a step: the masses may mobilize against the regime even without elite infighting. Autocratic leaders must walk a knife-edge in dealing with these competing risks.

Building from Hollyer, Rosendorff, and Vreeland (2015), we take as given that government disclosure is one mechanism through which autocratic leaders facilitate mass collective action. They demonstrate, both theoretically and empirically, that increased levels of disclosure are associated with an increased risk of mass mobilization under autocratic rule and with an increased risk of regime collapse resulting from mass collective action. They argue that economic transparency facilitates collective action because it helps coordinate citizen beliefs: citizens will be unwilling to take to the streets unless they are certain their beliefs about the need to remove an autocratic regime are widely shared.

Similar forces are likely to influence citizen responses to discord within the autocratic regime. Individual citizens are better able to coordinate on engaging in protest following such infighting when they recognize that their perceptions of the incumbent regime are widely shared. Rival members of the elite understand that transparency facilitates mobilization by the masses and are less likely to act against their leaders when these leaders choose to disclose. Transparency (and the associated threat to elite survival) is used by leaders to cow recalcitrant rival members within the regime.

Leaders, therefore, are more likely to disclose when they perceive threats as emerging from within their ruling coalition. We measure the level of threat to the leader from the inner-circle elites in various ways. B. Bueno de Mesquita et al. (2003, 37, 65) argue that the threat posed by elites is likely to be falling in a leader’s time in power. Analogously, Geddes, Wright, and Frantz (2014) find that personalistic autocrats—who rule over regimes that derive their legitimacy from the leader himself—have less to fear from elites than those autocrats who rule institutionalized regimes, with designated succession mechanisms and popular legitimacy distinct from the leader’s identity. Finally, following Gandhi’s (2008) approach to classifying autocracies, “hierarchical” regimes—such as military dictatorships and monarchies—have less to fear from elites than the more “precarious” civilian-ruled autocracies. In all instances, we would expect disclosure to be rising in the threat posed by the elite—that is, disclosure will be greatest (1) when leaders are new to power, (2) under nonpersonalist regimes, or (3) under civilian controlled autocracies.

Argument

Authoritarian leaders must navigate two threats to their rule. One emerges from within the regime itself. The other is the danger of mass mobilization on the part of the public. We share this framework with much of the recent literature on autocratic institutions (see particularly, Svolik 2012). Models of autocratic rule often find these threats to be strategic substitutes—for instance, policies aimed at alleviating the threat of mass revolt may increase the risk of a coup (Svolik 2013b). “Here, we contend that the reverse holds, policies that increase the mobilizational capacity of the populace may force members of the regime to toe the line set by the leadership”.

Members of the regime, when dealing with a leader who acts against their interest, face a choice of whether to remove this leader. Removing the leader, on the one hand, opens up the possibility that his replacement will prove more amenable to the elite. But, this is a costly gamble; removing the leader increases the risk of regime collapse, potentially costing these same elite individuals their privileged positions (Besley and Kudamatsu 2007; B. Bueno de Mesquita et al. 2003; Gehlbach and Malesky 2010; Padró i Miquel 2007).

We contend that transparency alters these elites’ decision calculus. Previous work (Hollyer, Rosendorff, and Vreeland 2015) establishes that transparency enhances the mobilizational capacity of the populace, boosting the probability that attempts to remove the leader cause the regime to collapse.² This pushes members of the elite toward inaction. As the danger of insurrection mounts, members of the elite grow more complacent in the face of an uncooperative leadership. Following Slater (2010), a common external threat produces internal cohesion for the regime.

Transparency and Protest in Autocracies

Transparency, under autocratic rule, tends to facilitate mass mobilization by the populace against the regime. This effect arises due to coordination problems inherent in mass protest. Protests that draw widespread participation are likely to succeed in forcing the regime to change policy or leadership, or even in bringing about regime change. Smaller protests, on the other hand, are likely to be put down, at considerable cost to participants.

Any individual’s decision to protest thus depends on her beliefs about the willingness of others to similarly turn out. In this context, public information is likely to play an important role (Morris and Shin 2002). Information that is witnessed by all citizens, and known by all citizens to be publicly observed, allows individuals not only to update their beliefs about the performance of government but also their higher-order beliefs about the expectations held by others. In the absence of such information, uncertainty about the willingness of others to mobilize may render mass protest impossible, whereas public disclosures by the government may render protest feasible. Hollyer, Rosendorff, and Vreeland (2015) formalize this argument and demonstrate, both theoretically and empirically, that this particualr form of transparency (disclosure by the government of aggregate economic information) renders mass protest more likely under autocratic rule. See also E. Bueno de Mesquita (2010), Edmond (2013), Little, LaGatta, and Tucker (2015), and Shadmehr and Bernhardt (2014).

Specifically, these authors find—using a definition of transparency identical to that we present in our empirical sections—that transparency is associated with more frequent protests and strikes in autocratic regimes and that the level of transparency predicts autocratic regime collapse as a result of protest or processes leading to democratization.³ This relationship is visible in the raw data: transparency scores in the five years leading up to democratization are, on average, $\frac{1}{3}$ of a standard deviation above the average in the full sample of autocracy years, a difference that is highly significant. In more rigorous specifications, they find that a one standard deviation increase in transparency is associated with an increase in the hazard of democratization of between 38 and 61 percent.⁴

Transparency, Coups, and Autocratic Failure

We also take as a building block of our argument that elite infighting and coups and “coup attempts act as a focal point for protest (on a related point, see Casper and Tyson 2014)”. Leadership struggles act as signals to the citizenry of (1) incipient policy changes and (2) intra-regime conflict and weakness. Since mass mobilization entails strategic complementarities, citizens are only likely to take to the streets following a usurpation when they recognize that others share their perception of the implications of this conflict. Are the current leadership’s policies widely understood to benefit the populace—as, for instance, might be the case after reports of strong economic growth are disclosed—such that an incipient change is undesirable? Is the regime widely understood to be despised—as, for instance, might be the case after the disclosure of declining living standards—such that any moment of weakness should be exploited to ensure its removal? Since transparency helps ensure that (1) these perceptions are widely shared and (2) known to be widely shared, citizens are more likely to respond to elite attempts to replace the leader with protest in a transparent than an opaque environment.

Autocratic leaders determine the level of disclosure with these effects in mind. Disclosure constitutes a risky gamble for such a leader. Rival elites are rendered more docile by virtue of the increased mobilizational capacity of the populace. Yet, empowering the populace in this manner is hazardous: citizens may depose the regime even as the elite toes the leadership’s line. Leaders are thus placed in the position of trading off the threat they face from the populace against the threat from the elite. When the latter is high, the leader will choose to increase the popular threat to reduce that posed by the elite. By contrast, when the internal threat is low, there is little incentive to disclose. Any gains in internal regime cohesion are more than offset by increases in the threat of popular mobilization.

We do not wish to suggest that transparency is the unique or the most important tool through which leaders may play the threats posed by the populace against those from the regime elite. Our contention is merely that, following existing work on protest and transparency (Hollyer, Rosendorff, and Vreeland 2015) and on the interrelated nature of mass mobilization and coups (Casper and Tyson 2014), transparency facilitates mass protest in the wake of a coup. Autocratic leaders’ decisions over whether or not to disclose are be driven, in part, by the threat posed by regime elites.

Similarly, we do not wish to contend that leaders’ attempts to balance the competing threats from the mass public and within the regime are the only forces driving variation in levels of autocratic transparency. Autocrats may disclose information to appeal to international investors or the international financial institutions, or as a commitment to restrain censorship (Shadmehr and Boleslavsky 2015) or to better monitor their political opponents (Egorov, Guriev, and Sonin 2009). Our explanation is intended to complement these existing accounts.

Related Literature

Our intuition that elite attempts to discipline autocratic leaders risk the stability of the autocratic regime draws on B. Bueno de Mesquita et al. (2003) and Besley and Kudamatsu (2007). Padró i Miquel (2007) contends that this “politics of fear” allows leaders to expropriate from their winning coalition.⁵

Slater (2010) advances a related argument that a powerful and sustained external threat may enable the formation of elite “protection pacts” in autocracies. Leaders are left relatively secure and elites contribute large amounts of resources to sustaining the regime. Hence, these regimes are particularly long-lived. Slater (2010), however, particularly focuses on moments immediately surrounding a regime’s founding, whereas our argument applies throughout its time in power. Moreover, in our formulation, leaders manipulate the level of external threat, whereas, in Slater (2010), this is exogenously given.

We also follow a recent literature that examines the strategic trade-offs between popular and elite threats to autocratic leaders. For instance, Svolik (2012, 2013b) also explores the manner in which increased capacity to repress a popular opposition may leave leaders at risk of a coup. B. Bueno de Mesquita and Smith (2009) similarly stress that autocratic leaders must meet popular and elite threats, and like this article, they consider the role of public goods that enhance the ability of the populace to coordinate. Casper and Tyson (2014) examine the relationship between protests and coups and find feedback mechanisms between these two threats. Specifically, they find that media freedoms increase the risk that one form of threat sparks the other, in line with our primitive assumption that leadership struggles may give rise to protest, particularly in transparent environments.

Our work also contributes to a growing literature on public information under autocratic rule. The mass media has been the focus of much of this work. Accounts by Egorov, Guriev, and Sonin (2009) and Lorentzen (2014) argue that autocratic regimes can effectively outsource the role of monitoring their lower-level agents to the media. The benefits of such monitoring must be traded off against the risk that a free media may promote mass public opposition. King, Pan, and Roberts (2013, 2014), in a series of innovative studies of the PRC’s “Great Firewall,” find supportive evidence for these accounts, online censors permit criticism of local government corruption and other forms of misgovernance, but delete calls for protest. In complementary theoretical explorations of media control under autocracy, Gehlbach and Sonin (2014) and Shadmehr and Bernhardt (2015) examine autocrats’ incentive to engage in media censorship, given that citizens rationally update to discount good news about the regime or interpret no news as bad news, while Guriev and Treisman (2015) examine the trade-offs dictators face between employing censorship, propaganda, and the co-optation of elites.⁶

While these papers share our focus on information dissemination under autocratic rule, the conception of transparency used in these papers differs fundamentally from the approach taken here. Rather than focusing on the media or on the public providing information to the government (say about sources of political opposition or discontent or the performance of the bureaucracy), we emphasize the role of government disclosures of information to the public. Increasing transparency does little to enhance monitoring of lower-level public officials in our formulation, since there is no outsourcing of information collection to private organizations.

Boix and Svolik (2013) also examine the role of (a different form of) transparency under autocratic rule, and—like this article—they conclude that higher levels of transparency are associated with a reduced risk of coup. In their model, transparency relates to the ability of members of the regime to observe efforts by autocratic leaders to amass greater authority and power. Transparency is thus, in their formulation, the clarity of the “rules of the game.” The critical concern for Boix and Svolik (2013) is thus the information available to the elite.

By contrast, our concern is with information made available to the wider public. If a given piece of information is revealed to the public, it must also be accessible to the members of the regime elite—so, these two notions of transparency must be at least somewhat correlated. However, there is no reason to expect that the reverse holds—considerable amounts of information are likely circulated among autocratic elites and not disclosed to the broader public.

Our definition and—in the empirical sections below—our measure of transparency is derived from Hollyer, Rosendorff, and Vreeland (2014), which focuses on the disclosure of economic information by the government to the broader public. We treat the findings of Hollyer, Rosendorff, and Vreeland (2015) that transparency of this type increases the risk of mass public protest under autocratic rule, as a theoretical primitive. These authors also find, in an Appendix, that transparency is associated with a reduced risk of autocratic regime collapse brought on by coups, a finding which is consistent with our argument. Our interest here is in autocratic leader removal, which may or may not entail regime collapse. Unlike these other pieces, this work explores the conditions under which autocrats disclose.

Model

We present a model of bargaining within an autocratic elite that takes place in the shadow of mass unrest. Steps taken by regime members to discipline their leaders threaten the stability of regime. In these assumptions, our model shares features of work by Besley and Kudamatsu (2007) and Padró i Miquel (2007). We, however, incorporate a leader’s decision to pursue policies that may facilitate mass mobilization into the model. We label this term “disclosure,” and we interpret the choice to disclose as an increase in transparency, which we know helps coordinate political action. The regime’s leaders, in order to forestall sanction by the elite, strategically choose a level of disclosure, threatening both the leader’s and the rival elite’s survival.

Model Primitives

Consider an interaction between an autocratic leader L and a group of regime elites R. There also exists a nonstrategic mass of citizens denoted M.⁷ L chooses a level of disclosure $d \in (0, 1)$ . R observes the choice of d and must determine whether to retain him in office. We denote this decision $v \in {0, 1}$ , where v = 1 denotes a decision to remove.

Following the choices of d and v, a contest for power takes place between members of the regime and the populace. If R choses to keep L in place (sets v = 0), the probability that the regime falls is given by $p (d)$ , where $p' (d) > 0$ . We thus make a primitive assumption (an empirical regularity well-established in the existing literature, as noted above) that greater disclosure is associated with a higher potential for political protest. To keep things simple, we assume that $p (0) = 0$ and $p (1) = 1$ .

If there is political infighting and L is removed, the entire regime is destabilized. The probability the regime falls to mass insurrection is now given by $ω p (d)$ where $ω > 1$ . The term ω reflects the degree to which internal strife destabilizes the regime. One can think of ω as reflecting the strength of L vis-à-vis R. If the rival is “stronger” in that a coup is more likely to lead to L’s removal and R’s replacing L, then ω is small; if instead the rival’s actions are more likely to bring the entire regime down, and hence R is unlikely to challenge L, then we might view L as stronger and more entrenched: ω is large.

To anticipate the empirical flavor of ω, following the approach of B. Bueno de Mesquita et al. (2003), ω may be a function of L’s time in power. During L’s tenure, he entrenches himself and cultivates an R with increasing levels of dependency and loyalty (see chapter 3 in Svolik 2012). Little (2017) advances an argument that such entrenchment may arise due to actors learning the strength of the leadership over time. Alternatively, one might think of ω as being higher in personalistic than institutionalized regimes (Geddes, Wright, and Frantz 2014) or as being higher in hierarchical as compared to precarious regimes (Gandhi 2008).

Let national income (or, equivalently, the rents accruing to the regime) be normalized to 1; if a leader survives in office, he derives utility from the share of national income he consumes λ. We assume that $λ \in (\frac{1}{2},1)$ , which captures a divergence of interest between the leader L and the elite R—a benefit to being the leader, and a motive for the elite to desire to remove the leader. R earns the residual $(1 - λ)$ , if the regime is not overthrown by the masses.

If mass insurrection takes down the regime, then L and R get nothing, regardless of the value of v. We allow for a further “congruence” of interest across L and R by another variable, r. One can think of r as the returns from regime cohesion. It represents the additional payoff earned by the leader and the elite in the instance that the elite chooses not to remove the leader and the regime (the leader and the elite together) survives potential mass insurrection. This variable r thus offers a reduced form method of capturing the extent of policy congruence within the regime. When r is high, L and R hold similar policy preferences, such that both enjoy substantial benefits from continued joint rule. If she removes the leader, R diminishes her ability to enact these jointly preferred policies, at least temporarily, and so suffers a substantial cost. If, by contrast, r is low, policy disagreements are rife, and the opportunity costs involved in displacing the leader are consequently slight. To avoid the trivial case, we require $r \in [1, \frac{λ}{1 - λ}]$ .⁸

So, if both L and R survive in power together, then both get the added benefit r. If R sets v = 1, she loses r, and (the now-ousted) L receives a payoff normalized to 0. If v = 0, L has a $(1 - p (d))$ chance of receiving λ, and a $p (d)$ chance that mass insurrection brings down the whole regime, and leaves L with $0$ .

In a complementary fashion, if v = 0, R has a $(1 - p (d))$ chance of receiving $(1 - λ)$ (and the additional r), and a $p (d)$ chance that mass insurrection brings down the whole regime, leaving R with $0$ . If the elite decides to oust L (setting v = 1), she gets the bigger piece of the pie—the λ share (but loses the r). The ouster is costly in another way as well—it sends a signal to the masses. Sensing regime weakness, the masses may take to the streets and bring down the entire regime. The probability of successful mass insurrection is thus augmented by $ω > 1$ . So, if v = 1, R has only a $1 - ω p (d)$ chance of getting λ—and she faces a $ω p (d)$ chance that mass insurrection brings down the whole regime, leaving R with $0$ .

Summarizing the expected payoffs of L and R, we have:

Leader $E U^{L} (d, v; ω) = {\begin{array}{l} (1 - p (d)) λ r if v = 0 \\ 0 if v = 1 \end{array}$

Elite $E U^{R} (d, v; ω) = {\begin{array}{l} (1 - p (d)) (1 - λ) r if v = 0 \\ (1 - ω p (d)) λ if v = 1 \end{array}$

The sequence of moves has the leader $(L)$ choosing d and then the members of the rival elite $(R)$ choose v. We offer the equilibrium and the key comparative static in what follows. The proofs are in the Appendix.⁹

Proposition 1: The equilibrium $(d^{*}, v^{*})$ is characterized by $d^{*} = inf {d : E U^{R} (d,1; ω) \leq E U^{R} (d,0; ω)}$ and $v = {\begin{array}{l} 0 i f d \geq d^{*} \\ 1 o t h e r w i s e \end{array} .$

Our key comparative static finding:

Proposition 2: In equilibrium, disclosure, $d^{*}$ falls with ω.

We reach two main conclusions that we can test empirically: (1) disclosure reduces the probability of regime disunity. We will proxy for such infighting by examining instances of autocratic leader removal via coups. (2) High-ω leaders have little to fear from elites, and thus disclose less. We now turn to our empirical tests of these two results.

Empirical Measures

Outcome Variables

Throughout, our measure of transparency is the HRV index (Hollyer, Rosendorff, and Vreeland 2014). We present two sets of empirical results: the first, which examines the risk autocratic leaders face of a coup, treats this index as the primary explanatory variable of theoretical interest. In the second, which examines autocrats’ decision to disclose, the HRV transparency index is our outcome term. Proposition 2 states that this term should be falling in ω.

This index is a continuous measure which reflects the public disclosure, by governments, of credible economic information. It is constructed by relying on the presence or absence of data from the World Bank’s World Development Indicators (WDI) data series.

To be more precise, the HRV index scales the presence/missingness of 240 variables from the WDI using an item response algorithm. The criteria for these 240 variables are such that (1) each variable is reported by at least one country in each year covered (1980–2010), to ensure reporting carries a consistent meaning; (2) data compiled only for a subset of countries or that clearly are constructed only by the Bank with minimal input from national statistical agencies are omitted; (3) alternative transformations of the same underlying information (e.g., reporting in multiple currencies) are omitted. The item-response algorithm adjusts for the fact that certain terms (e.g., population) are reported with far greater frequency than others and weights the importance of the reporting of any one observation based on its ability to predict disclosure of other variables.

We do not assume that the public accesses information directly from the World Bank. Rather, we expect disclosure to the World Bank to correlate with the disclosure of credible information to the domestic press and influential members of the public. These actors then relay more or less precise impressions of the state of the economy on to the broader public. Essentially, the degree to which information is conveyed and then accepted as valid, and subsequently published by the World Bank is a proxy for the informational environment within a given autocracy more generally. The greater is the willingness of the leadership to provide information to the World Bank (or international lenders or other entities), the more likely it is that the population more broadly has more and better information about economic aggregates. Moreover, the Bank’s screening of data acts as a check on the credibility of the information available to the public—overt lies by the government are likely to be treated as missing data by the World Bank and to be disregarded in equilibrium by the populace.¹⁰

In our initial set of regressions, the outcome of interest is the removal of an autocratic leader via a coup. We rely on Svolik’s (2012) data set on autocratic leaders, which defines both the duration of a given leader’s rule and the method of his removal, to measure this term. Coups are defined as instances in which the leader is ousted either by the military or an elite faction, where this group either applies force or uses the threat of force. Coups are the most extreme form of inter-elite struggles depicted in our theory. In relying on this measure, we likely undercount instances of regime infighting—particularly in more institutionalized regimes. In practice, this implies that the coefficients on our measures of institutionalization (described below) are biased downward.

In our regressions examining autocratic disclosure, Svolik’s data define our unit of observation: the autocratic regime year. They also define an explanatory term of theoretical interest: $New leader \in {0, 1}$ , which takes the value 1 if a given leader has been in power for five years or less.

Controlling for Confounds

One possible concern with our analysis lies in the likely correlation between levels of government disclosure and the level of state capacity. Governments choose whether to disclose economic information based not only on a political calculus but also on their ability to collect and process the relevant data (Stone 2008). Hollyer, Rosendorff, and Vreeland (2014) discuss this concern in detail and argue that economic transparency reflects both a state’s willingness and ability to disclose. For instance, less developed states of all regime types disclose at low levels, but, as capacity rises, democracies respond by increasing transparency far more than autocracies.

Nonetheless, the correlation between transparency and state capacity may bias our results with regard to the association between economic transparency and coups. Note, however, the direction of the bias is nonobvious: while regimes with low levels of capacity may be prone to collapse generally, highly capable bureaucracies may be more prone to staging coups (Egorov and Sonin 2011; Svolik 2013b). To adjust for these potential biases, we include a control for gross domestic product (GDP) per capita, measured in thousands of constant 2005 US dollars, from the Penn World Table version 7.1 in all specifications. In using GDP per capita as a proxy for capacity, we follow the influential study of Fearon and Laitin (2003). As described below, we also flexibly control for a past history of coups, which—if coups and state capacity are indeed correlated, and given the highly persistent nature of capacity—should help to adjust for confounding.

In our models that examine variation in the level of disclosure, we also include a lagged dependent variable in all specifications. Besides adjusting for dynamics in the data-generating process, the inclusion of this term should help control for highly persistent confounds—such as state capacity—that might affect both past and present levels of disclosure.

We also rely on the Penn World Table version 7.1 for several additional economic controls. In all specifications, we control for the rate of growth in real GDP per capita, as measured in constant 2005 US dollars. In our specifications examining the risk of coup, this control helps adjust for the potentially destabilizing effects of economic recessions (Alesina et al. 1996; Haggard and Kaufman 1995). In our specifications examining levels of disclosure, this control adjusts for the possibility that autocrats choose to publicize good news when the economy is performing well and attempt to disguise underperformance through opacity (Wallace 2016b). When examining levels of disclosure, we additionally include controls for GDP, economic openness $(\frac{Exports + Imports}{GDP})$ , and levels of government consumption.

We also include an indicator for fuel exporters in regressions examining disclosure, given the findings of Egorov, Guriev, and Sonin (2009), who find that resource-abundant dictators tend to place greater restrictions on the media. Finally, we include an indicator for regimes that are currently subject to IMF conditionality, given that the IMF often imposes disclosure requirements on borrower countries.

Measuring ω: the Entrenchment of Leaders

Our theoretical model’s central comparative static prediction holds that autocrats should disclose more readily when they have more to fear from their inner elite—that is, autocrats disclose at greater levels as ω falls. Operationalizing this theoretical parameter poses challenges. To borrow a term from Pepinsky (2014a), ω reflects a “logics” approach to the study of authoritarian regimes—it corresponds to a measure of the distribution of power between regime elites and dictators. This balance of power both gives rise to certain observable institutional configurations and is sustained by such configurations (on this point, see Pepinsky 2014b; Svolik 2013a). Moreover, the mapping from ω into empirical covariates is imprecise.

To help address these concerns, we rely on several different indicators to capture variation in ω, each of which borrows from the literature on autocratic regimes. Following the logic of B. Bueno de Mesquita et al. (2003), who argue that elite loyalty increases over the tenure of an autocratic leader, long-serving leaders have higher values of ω.¹¹ We also use the autocratic institutions data from Geddes, Wright, and Frantz (2012; henceforth GWF). The GWF definition of “personalistic” leaders maps onto our model parameter ω, with personalistic regimes corresponding to high values of ω. Finally, we approximate ω using Gandhi’s (2008) categorization of autocracies. In contrast to civilian rulers, who hold precarious control over elites, leaders who enjoy a hierarchical relationship with their elites (military dictators and monarchies) correspond to high-ω regimes. We code an indicator for hierarchical regimes based on data from Cheibub, Gandhi, and Vreeland (2010; henceforth DD). We make use of the GWF and DD data sets alternatively.¹²

While this catholic approach to operationalizing ω helps to address concerns that the idiosyncrasies of any one measure could produce spurious results, it does not resolve all issues that arise from adopting a logics approach to autocratic institutions. Specifically, these institutions both reflect and sustain the underlying balance of power within a given regime. We emphasize that our empirical results pertaining to autocratic institutions should not be interpreted causally—institutions may play a causal role over both leader’s survival and disclosure, but institutions also are affected by unobserved intra-regime dynamics that may also have a causal effect on these terms.

Empirical Models

Transparency and Coups

Proposition 1 forms the first basis of our empirical investigation. We empirically interpret the equilibrium behavior as, conditional on autocratic institutions, transparency insulates leaders from coups.

Note that this is not a comparative static claim. Both disclosure and leader removal (i.e., coups) are endogenous in our model. However, it is clear from Proposition 1 that leaders disclose only to forestall the threat of removal at the hands of the elite. This is an equilibrium causal claim.¹³

Evidence of the relationship between transparency and the frequency of coups is visible in the raw data. Figure 1 plots the average frequency of successful coups among autocratic leader years in the sample, by transparency quartile. The simple bivariate relationship suggests that the risk of coup is falling in transparency—in particular, coups are less frequent when transparency is above, as compared to when it is below, the sample median.

We estimate the relationship between the hazard of leader removal via coup and transparency using a series of Cox competing hazards specifications. Competing hazards models are a means of estimating the hazard (the probability that an event takes place in time t, given that it has not already taken place) of one of several mutually exclusive events. In our case, the event of interest is leader removal via coup, and the competing hazards are alternate forms of leader removal.¹⁴ The unit of observation is the autocratic leader year.

To adjust for the possibility that past successful coups predict future coups, we stratify the baseline hazard rate in our Cox estimates based on two measures of coup history.¹⁵ The first is a simple indicator variable ${0, 1}$ , equal to one if any past autocratic leader was removed via a coup. The second is an ordered variable ${0, 1, 2, 3}$ , which, if equal to any element in ${0, 1, 2}$ , denotes the number of past autocratic leaders removed via coup and, if equal to 3, denotes that more than two previous leaders have been removed via coup. We additionally fit models in which we simply include a control for the binary indicator of past coups. We thus fit specifications of the following form:

h_{l} (t) = h_{0} (t, c_{l}) e x p (γ t r a n s p a r e n c y_{l, t - 1} + X_{l, t - 1} β)

where l denotes autocratic leader, t denotes time, $c_{l}$ is an indicator for coup history, and $X_{l, t - 1} β$ is a vector of time-varying controls and associated coefficients. $h_{0} (t, c_{l})$ is estimated nonparametrically within each stratum, based on the fraction of observations that experience a coup at time t as compared to the number of observations at risk (leaders who have not yet been removed for any reason). Duration dependence is thus factored out of the likelihood function, while a history of coups may shift both the shape and level of the baseline hazard function (Beck, Katz, and Tucker 1998; Box-Steffensmeier and Zorn 2002). Our hypothesis holds that the coefficient on transparency, γ, should be negative.

Results from the model specified in equation (1) are presented in Tables 1 and 2. Table 1 reports results that use the GWF definitions of autocratic institutions, while Table 2 reports analogous results that employ the definition of hierarchical intra-regime relations (i.e., military or monarchical regimes) defined by Gandhi (2008).¹⁶ When employing the GWF controls, we include an interaction between the indicator for party-based regimes and time, to address violations of proportional hazards assumption (Box-Steffensmeier and Jones 2004). Results in which the baseline hazard is stratified based on whether there was a previous coup are presented in the leftmost column; those stratified based on the ordered indicator of coup history are presented in the center column; and results that do not stratify the baseline hazard, but simply control for an indicator of past coups, are presented in the rightmost column of both tables.

Table 1.

Hazard of Leader Removal Via Coup.

	Past Coup Strata	Coup Experience Strata	Past Coup Control
Transparency	−0.248	−0.282	−0.240
	[−0.480, −0.016]	[−0.531, −0.033]	[−0.461, −0.019]
Growth	−0.003	−0.005	−0.000
	[−0.031, 0.026]	[−0.042, 0.032]	[−0.029, 0.029]
Gross domestic product per capita	−0.110	−0.094	−0.117
Gross domestic product per capita	[−0.208, −0.012]	[−0.175, −0.013]	[−0.229, −0.005]
Party	−1.793	−1.709	−1.735
	[−2.595, −0.991]	[−2.451, −0.967]	[−2.661, −0.810]
Party × t	0.113	0.112	0.109
	[0.045, 0.181]	[0.049, 0.175]	[0.037, 0.182]
Personal	−0.807	−0.676	−0.809
	[−1.609, −0.004]	[−1.437, 0.084]	[−1.592, −0.025]
Ever past coup			−0.047
			[−0.908, 0.814]
Number of subjects	89	89	89
Number of failures	36	36	36

Note: Results from Cox competing hazards regressions of leader removal via coup on transparency and controls. Ninety-five percent confidence intervals are reported in brackets.

Table 2.

Hazard of Removal Via Coup (DD Controls).

	Past Coup Strata	Coup Experience Strata	Past Coup Control
Transparency	−0.202	−0.228	−0.217
	[−0.401, −0.002]	[−0.450, −0.006]	[−0.420, −0.014]
Growth	−0.008	−0.006	−0.006
	[−0.035, 0.019]	[−0.040, 0.028]	[−0.033, 0.021]
Gross domestic product per capita	−0.073	−0.071	−0.078
Gross domestic product per capita	[−0.145, −0.001]	[−0.138, −0.004]	[−0.154, −0.002]
Hierarchical	0.410	0.280	0.430
	[−0.196, 1.017]	[−0.311, 0.871]	[−0.182, 1.041]
Ever past coup			−0.126
			[−0.969, 0.717]
Number of subjects	94	94	94
Number of failures	37	37	37

Note: Results from Cox competing hazards regressions of leader removal via coup on transparency and controls. Ninety-five percent confidence intervals are reported in brackets.

In all specifications, the coefficient on the transparency term is negative, with 95 percent confidence intervals bounded away from zero. Point estimates indicate that a one standard deviation increase in the HRV index is associated with a 35 to 40 percent fall in the hazard of a coup. The associated 95 percent confidence interval runs from a decline of 3 percent to a decline of 58 percent in the hazard function when the GWF controls are employed. The analogous 95 percent confidence interval with the DD controls runs from approximately 0 to 49 percent.

To ease interpretation of these results, we present plots of smoothed estimates of the hazard function, using the GWF controls, in Figure 2. The solid line depicts the smoothed hazard when the HRV index is at its tenth percentile in the sample, while the dashed line presents the same when the HRV index is at its ninetieth percentile. In the former instance, the hazard that a newly seated leader is ousted via a coup during the first year of his reign is roughly 3.25 percent. When transparency is at its ninetieth percentile, this falls to a hazard of roughly 1.75 percent.

Figure 2 also reveals that the hazard rate declines over time, particularly over the first twenty years of a leader’s rule. This is consistent with our claim that entrenched autocrats face fewer risks from within the ruling elite—a point to which we return below.

As is also consistent with theoretical mechanisms, personalistic regimes (as defined by GWF) are less likely to experience a coup than military regimes—a result that is significant at the 95 percent level in two of three specifications. Contrary to our expectations, party-based regimes are less likely to experience coups than military regimes, though this effect diminishes with the time a leader has served in office and goes to zero as a leader’s tenure approaches fifteen years in power. We expect that this reflects the more hidden nature of regime infighting in more institutionalized regimes. In party-based regimes, regime elites can resort to mechanisms short of violence to discipline or remove the leader. The greater incidence of unobserved infighting in such regimes tends to bias the coefficient on this indicator toward zero. Hierarchical regimes (as coded by the DD data) show no significant correlation with the incidence of coups.

Transparency and economic performance

The HRV measure of transparency captures, to some degree, the willingness of leaders to make aggregate economic data generally available. It may be that this willingness is affected by aggregate economic conditions themselves. For instance, since information collection, aggregation, and dissemination are costly, states may only be able to collect and publish data in “good times.” Alternatively, an autocratic leader, perhaps concerned about the effects of revealing information, may only do so when there is “good news” to share. Hence we might be concerned that transparency threatens mass rising and keeps elite rivals in line only in good economic conditions.

This leads us to check whether the underlying results are robust to these “good news” and good times effects. We introduce to the estimation an interactive term, transparency × growth, where growth is GDP growth.

As is reported in Tables 3 and 4, the coefficient on the interaction term is consistently effectively equal to zero, while the coefficient on transparency is largely unchanged. It does not appear to be the case that coup risk falls by more when transparency is associated with good (or bad) economic conditions. For further discussion of the empirical regularities concerning transparency and protest in autocracies and democracies, see Hollyer, Rosendorff, and Vreeland (2015).

Table 3.

Hazard of Leader Removal via Coup (Interaction).

	Past Coup Strata	Coup Experience Strata	Past Coup Control
Transparency	−0.222	−0.249	−0.223
	[−0.473, 0.030]	[−0.524, 0.026]	[−0.459, 0.013]
Growth	−0.006	−0.009	−0.002
	[−0.038, 0.026]	[−0.050, 0.031]	[−0.034, 0.030]
Transparency × growth	−0.008	−0.009	−0.004
	[−0.022, 0.006]	[−0.023, 0.005]	[−0.018, 0.009]
Gross domestic product per capita	−0.108	−0.090	−0.113
Gross domestic product per capita	[−0.204, −0.011]	[−0.169, −0.012]	[−0.222, −0.003]
Party	−1.724	−1.635	−1.660
	[−2.496, −0.953]	[−2.366, −0.903]	[−2.558, −0.761]
Party × t	0.116	0.117	0.111
	[0.047, 0.185]	[0.052, 0.183]	[0.038, 0.185]
Personal	−0.606	−0.430	−0.605
	[−1.351, 0.140]	[−1.161, 0.300]	[−1.325, 0.116]
Ever past coup			−0.051
			[−0.909, 0.807]
Number of subjects	89	89	89
Number of failures	36	36	36

Note: Results from Cox competing hazards regressions of leader removal via coup on transparency and controls. Ninety-five percent confidence intervals are reported in brackets.

Table 4.

Hazard of Removal via Coup (Interaction, DD Controls).

	Past Coup Strata	Coup Experience Strata	Past Coup Control
Transparency	−0.201	−0.224	−0.217
	[−0.408, 0.005]	[−0.455, 0.007]	[−0.426, −0.008]
Growth	−0.008	−0.008	−0.005
	[−0.042, 0.025]	[−0.047, 0.031]	[−0.040, 0.029]
Transparency × growth	−0.000	−0.002	0.000
	[−0.016, 0.015]	[−0.016, 0.012]	[−0.013, 0.014]
Gross domestic product per capita	−0.073	−0.071	−0.078
Gross domestic product per capita	[−0.145, −0.001]	[−0.138, −0.004]	[−0.154, −0.002]
Hierarchical	0.410	0.280	0.430
	[−0.196, 1.017]	[−0.311, 0.870]	[−0.182, 1.041]
Ever past coup			−0.126
			[−0.969, 0.717]
Number of subjects	94	94	94
Number of failures	37	37	37

Note: Results from Cox competing hazards regressions of leader removal via coup on transparency and controls. Ninety-five percent confidence intervals are reported in brackets.

Autocracy and Transparency

In addition to advancing claims about the relationship between transparency and the frequency of coups, our theory offers predictions about which autocratic regimes are likely to disclose information. Proposition 2 contends that the minimum level of disclosure needed to forestall a coup rises as ω declines. In contrast to our claims regarding the effects of disclosure, this is a comparative static claim. We treat this proposition as advancing an empirical claim that (1) leaders disclose less once entrenched in office—and disclose more when new to office (B. Bueno de Mesquita et al. 2003), (2) GWF’s personalistic regimes disclose less than other autocracies, and (3) DD’s hierarchical leaders disclose less than civilian-ruled autocracies. Leaders grow more willing to disclose as the internal threat from elites rises precisely because transparency is posited to diminish the risk of removal by elites.

In this section, we assess these claims through a series of varying-intercepts hierarchical regression models of the HRV index on a series of institutional and time-varying characteristics. Specifically, we estimate models of the following form:

\begin{matrix} t r a n s p a r e n c y_{i, t} = ρ t r a n s p a r e n c y_{i, t - 1} + α_{i} + X_{i, t - 1} β + ∊_{i, t}, \\ α_{i} : N (Z_{i} ζ, σ_{α}^{2}) \end{matrix}

where i denotes a given autocratic regime, t denotes the year, $X_{i, t - 1}$ is a vector of time-varying controls, $α_{i}$ is a regime-specific intercept term, and $Z_{i}$ is a vector of regime-level (time-invariant) controls. The unit of observation is an autocratic regime year.

The GWF or DD regime classifications and the fuel exporter indicator constitute the time-invariant regime characteristics $Z_{i}$ , from equation (2). The $New leader$ indicator and economic covariates constitute the time-varying terms $X_{i, t - 1}$ . We additionally control for the lagged value of the HRV index, which also varies over time. This last term is included to capture model dynamics (Beck and Katz 2011).

The varying-intercepts model is preferred over a fixed-effects estimator, given that our institutional covariates of interest are time-invariant. However, a leader’s time in office does, naturally, vary over time. As a robustness check, we therefore fit alternative specifications with regime and year fixed effects and omitting the lagged dependent variable (a difference-in-differences estimator) in Appendix Table B1. We drop all institutional covariates in these specifications, as these variables are perfectly collinear with the regime fixed effects. Results are substantively similar to those presented below, though estimated somewhat less precisely.

We estimate the model described in equation (2) via Markov chain monte carlo estimation in JAGS 3.3.0. We place separate diffuse multivariate normal priors on, respectively, the coefficients on regime-level variables ζ and the regime-year–level variables β. We place an informative prior on $ρ \sim N (0, 1)$ . All variables that are not either binary indicators nor time counters have been standardized.

The results from models based on equation (2) are presented in Tables 5 and 6, which report specifications employing the GWF and DD regime definitions, respectively.

Table 5.

Models of Disclosure: GWF Data.

	Model 1	Model 2	Model 3
Regime predictors
Party	0.002	0.003	−0.002
	[−0.035, 0.042]	[−0.031, 0.037]	[−0.035, 0.033]
Personal	−0.038	−0.037	−0.043
	[−0.079, 0.008]	[−0.075, 0.003]	[−0.081, −0.006]
Fuel exporter	−0.037	−0.037	−0.034
	[−0.08, 0.006]	[−0.078, 0.006]	[−0.071, 0.009]
Regime-year predictors
Lag transparency	0.961	0.961	0.964
	[0.943, 0.979]	[0.943, 0.979]	[0.945, 0.979]
Gross domestic product per capita	0.004	0.005
	[−0.013, 0.021]	[−0.011, 0.022]
Gross domestic product	0.003	0.002
	[−0.011, 0.015]	[−0.011, 0.015]
Economic openness	0.002	−0.01
	[−0.013, 0.016]	[−0.023, 0.004]
Growth	0
	[−0.012, 0.01]
Government’s consumption	−0.01	−0.01
	[−0.023, 0.003]	[−0.023, 0.004]
New leader	0.027	0.027	0.027
	[0.003, 0.048]	[0.004, 0.051]	[0.003, 0.049]
Cubic time
Polynomial	✓	✓	✓
Number of observations	1,530	1,530	1,530
Number of regimes	119	119	119

Note: Results from a hierarchical varying-intercepts linear regression of HRV transparency index scores on listed covariates. Covariates that shift the intercept term are described as “regime predictors,” while those that directly shift predicted transparency values are listed as “regime-year predictors.” All covariates that are neither indicators terms nor time counters have been standardized by subtracting the mean and dividing by the standard deviation. Ninety-five percent credible intervals are presented in brackets. GWF = Geddes, Wright, and Frantz.

Table 6.

Models of Disclosure: DD Data.

	Model 1	Model 2	Model 3
Regime predictors
Single party	−0.02
	[−0.083, 0.043]
Multiparty	0.014
	[−0.04, 0.067]
Legislature	0.028
	[−0.015, 0.067]
Hierarchical	−0.027	−0.036	−0.032
	[−0.057, 0.005]	[−0.062, −0.01]	[−0.057, −0.007]
Communist	0.024
	[−0.035, 0.083]
Fuel exporter	−0.025	−0.034	−0.026
	[−0.069, 0.019]	[−0.073, 0.004]	[−0.063, 0.012]
Regime-year predictors
Lag transparency	0.958	0.964	0.968
	[0.941, 0.976]	[0.948, 0.979]	[0.952, 0.984]
GDP per capita	0.011	0.008
	[−0.004, 0.028]	[−0.008, 0.021]
GDP	0	0.002
	[−0.012, 0.013]	[−0.01, 0.013]
Economic openness	−0.004
	[−0.018, 0.011]
Growth	−0.004
	[−0.014, 0.006]
Government’s consumption	−0.015	−0.016
	[−0.028, −0.001]	[−0.029, −0.004]
New leader	0.03	0.03	0.03
	[0.007, 0.051]	[0.007, 0.05]	[0.01, 0.055]
Cubic time
Polynomial	✓	✓	✓
Number of observations	1,623	1,623	1,623
Number of regimes	135	135	135

In all specifications, the coefficient on the $New leader$ indicator is positive with 95 percent credible intervals bounded away from zero. Recall further that the presence of the lagged dependent variable renders these models dynamic. We simulate the estimated marginal effect of a new leader over time in Figure 3. Estimates are from model 2 in Table 5, to the left, and model 2 in Table 6 to the right.

The estimated marginal effect of a new leader is small in absolute terms. The introduction of a new leader in time t = 0 is anticipated to increase transparency scores by roughly .12 standard deviations by time t = 5. However, our transparency scores tend to vary little within autocratic regimes over time. This increase in transparency scores associated with a new leader is equivalent to $\frac{1}{3}$ to $\frac{1}{2}$ , the average within-regime standard deviation in this term.

The coefficient on the indicator for personalist regimes is consistently negative—again in line with theoretical expectations. 95 percent credible intervals are bounded away from zero in all specifications. The estimated contemporaneous marginal effect of changing from a nonpersonalistic to a personalistic regime is a decline of between .038 and .044 standard deviations in the HRV index. Given the dynamics of the model, however, the long-term equilibrium association between regime classification and transparency is considerably larger. The long-term equilibrium difference between personalistic and nonpersonalistic military regimes is estimated to be roughly .97 standard deviations.

We see similar effects when the DD regime definitions are used in place of GWF. In all specifications, hierarchical regimes disclose at lower levels than nonhierarchical. 95 percent credible intervals on this term are bounded away from zero when measures of other regime characteristics are excluded from the model. A change from a nonhierarchical to a hierarchical regime leads to an expected contemporaneous decline in transparency of roughly .04 standard deviations. Model dynamics, however, imply that—over the long term—such a change would lead to a one standard deviation decline in the level of transparency.

Conclusion

Autocratic leaders face twin threats—from the populace and the elite—to their survival (Besley and Kudamatsu 2007; Padró i Miquel 2007; Svolik 2012). We advance a novel claim: autocrats can use the disclosure of information to navigate these threats. Transparency (and the associated threat of mass uprising) is used by autocratic leaders to instill fear and loyalty in rival elites, forestalling the risk of coups.

Across a variety of statistical specifications, an increase of one standard deviation in HRV index scores is associated with a roughly 35 percent reduction in the hazard of coup. While such leaders are more insulated from threats that emerge from within their regime, existing results indicate that they suffer an increased risk from the masses. Autocrats are thus placed in a delicate position when deciding whether or not to disclose: they must trade off the threats they face from within their regime against those from without.

We introduce a theoretical argument prefaced on such a balancing act. Transparency insulates leaders from internal threats, we argue, precisely because it exposes them to increased external threats. The common threat to both the leader and the autocratic elite posed by mass unrest ensures greater cohesion within the regime (Padró i Miquel 2007; Di Lonardo and Tyson 2015; Slater 2010). Autocratic leaders, therefore, can manipulate this threat to ensure greater freedom of maneuver vis-à-vis their winning coalition. Transparency serves as one mechanism to gain this freedom.

There are no doubt other potential determinants for the observed variation in transparency across autocracies. Autocratic leaders may vary in the degree to which they access international credit markets or the international financial institutions and may adjust their patterns of disclosure, accordingly. Or they may permit the collection of some information in order to exercise control over the bureaucracy or to identify sources of opposition. There may be other reasons for transparency to be associated with lower coup risk—more competent or flexible autocrats can collect and disseminate data and keep potential rivals at bay. The argument presented here complements these other potential narratives. A theoretical foundation provides a basis for an observed and robust empirical regularity: autocrats are most prone to disclose information when the internal balance of power tends to favor rival members of the autocratic elite over the autocratic leader. Consistent with this argument, entrenched autocrats are, empirically, less inclined to transparency than their newly installed and institutionalized counterparts.

This finding adds an important caveat to conventional accounts of informational politics under autocracy. While more information may promote mobilization by the masses against autocratic rule; high levels of transparency may also entrench autocratic leaders against schisms within the elite.

Supplemental Material

Supplemental Material, JCR-17-0546.R1 - Why Do Autocrats Disclose? Economic Transparency and Inter-elite Politics in the Shadow of Mass Unrest

Supplemental Material, JCR-17-0546.R1 for Why Do Autocrats Disclose? Economic Transparency and Inter-elite Politics in the Shadow of Mass Unrest by James R. Hollyer, B. Peter Rosendorff and James Raymond Vreeland in Journal of Conflict Resolution

Footnotes

Appendix A

Appendix B

Authors’ Note

All remaining errors are our own. The HRV index and materials related to the HRV Transparency Project are available from .

Acknowledgments

Authors would like to thank Pamela Blackmon, Allan Defoe, Scott Gehlbach, Anita Gohdes, Charles Hankla, Andy Harris, Jacqueline Sievert, Milan Svolik as well as participants in the 2014 Annual EPSA Conference, Panel on the Elements of a General Theory of Peace; the 2014 APSA Conference, Panel on Accountability and Development; the 2015 MPSA Conference, Panel on Decision-making in Autocracies; the 2015 EPSA Conference, Panel on Dictators and the Media; the 2015 Annual APSA Conference, Panel on Transparency in Democracies and Dictatorships: Challenges and Enduring Questions; the 2016 Annual ISA Conference, Panel on Political Business Cycles, Policy Credibility, and Transparency; the 2017 Annual APSA Conference, Panel on Formal Models of Autocratic Politics, the Faculty International Relations Colloquium at Princeton University, and the International Political Economy Reading Group at New York University for helpful comments and suggestions. Special thanks go to Vanessa Hofman for valuable research assistance. James Hollyer would also like to thank the Benjamin Evans Lippincott Foundation and the Niehaus Center for Globalization and Governance for Research Support.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Supplemental Material

Supplemental material is available for this article online.

Notes

References

Adserà

Alícia

Boix

Carles

Payne

Mark

. 2003. “Are You Being Served? Political Accounatability and Quality of Government.” The Journal of Law, Economics & Organization 19 (2): 445–90.

Alesina

Alberto

Ozler

Sule

Roubini

Nouriel

Swagel

Phillip

. 1996. “Political Instability and Economic Growth.” Journal of Economic Growth 1 (2): 189–211.

Beck

Nathaniel

Katz

Jonathan N.

. 2011. “Modeling Dynamics in Time-series-cross-section Political Economy Data.” Annual Review of Political Science 14:331–52.

Beck

Nathaniel

Katz

Jonathan N.

Tucker

Richard

. 1998. “Taking Time Seriously: Time-series-cross-section Analysis with a Binary Dependent Variable.” American Journal of Political Science 42 (4): 1260–88.

Besley

Timothy

Kudamatsu

Masayuki

. 2007. “Making Autocracy Work.” Development Economics Discussion Paper Series 48. http://eprints.lse.ac.uk/3764/1/Making_Autocracy_Work.pdf (accessed January 8, 2018).

Boix

Carles

Svolik

Milan W.

. 2013. “The Foundations of Limited Authoritarian Government: Institutions, Commitment, and Power-sharing in Dictatorships.” Journal of Politics 75 (2): 300–16.

Boräng

Frida

Cornell

Agnes

Grimes

Marcia

Schuster

Christian

. 2018. “Cooking the Books: Bureaucratic Politicization and Policy Knowledge.” Governance 31 (1): 7–26. doi: 10.1111/gove.12283.

Box-Steffensmeier

Janet M.

Jones

Bradford S.

. 2004. Event History Modeling: A Guide for Social Scientists. Cambridge, MA: Cambridge University Press.

Box-Steffensmeier

Janet M.

Zorn

Christopher

. 2002. “Duration Models for Repeated Events.” Journal of Politics 64 (4): 1069–94.

10.

Brunetti

Aymo

Weder

Beatrice

. 2003. “A Free Press is Bad News for Corruption.” Journal of Public Economics 87:1801–24.

11.

Bueno de Mesquita

Bruce

Smith

Alastair

. 2009. “Political Survival and Endogenous Institutional Change.” Comparative Political Studies 42 (2): 167–97.

12.

Bueno de Mesquita

Bruce

Smith

Alastair

Siverson

Randolph M.

Morrow

James D.

. 2003. The Logic of Political Survival. Cambridge: The MIT Press.

13.

Bueno de Mesquita

Ethan

. 2010. “Regime Change and Revolutionary Entrepreneurs.” American Political Science Review 104 (3): 446–66.

14.

Casper

Brett Allen

Tyson

Scott A.

. 2014. “Popular Protest and Elite Coordination in a Coup d’état.” Journal of Politics 76 (2): 548–64.

15.

Cheibub

José Antonio

Gandhi

Jennifer

Vreeland

James Raymond

. 2010. “Democracy and Dictatorship Revisited.” Public Choice 143:67–101.

16.

Di Lonardo

Livio

Tyson

Scott A.

. 2015. “Deterrence and Counterdeterrence in the Fight against Global Terror.” Presented at the Research Program in Political Economy Speaker Series, Princeton University, Princeton. https://www.princeton.edu/rppe/speaker-series/TysonDET.pdf (June, 2016).

17.

Edmond

Chris

. 2013. “Information Manipulation, Coordination, and Regime Change.” Review of Economic Studies 80 (4): 1422–58.

18.

Egorov

Georgy

Guriev

Sergei

Sonin

Konstantin

. 2009. “Why Resource-poor Dictators Allow Freer Media: A Theory and Evidence from Panel Data.” The American Political Science Review 103 (4): 645–68.

19.

Egorov

Georgy

Sonin

Konstantin

. 2011. “Dictators and Their Viziers: Endogenizing the Loyalty-competence Trade-off.” Journal of the European Economic Association 9 (5): 903–30.

20.

Fearon

James D.

Laitin

David

. 2003. “Ethnicity, Insurgency, and Civil War.” The American Political Science Review 97 (1): 75–90.

21.

Ferraz

Claudio

Finan

Frederico

. 2008. “Exposing Corrupt Politicians: The Effects of Brazil’s Publically Released Audits on Electoral Outcomes.” Quarterly Journal of Economics 123 (2): 703–45.

22.

Gandhi

Jennifer

. 2008. Political Institutions under Dictatorship. Cambridge, MA: Cambridge University Press.

23.

Geddes

Barbara

Wright

Joseph

Frantz

Erica

. 2014. “Authoritarian Regimes: A New Data Set.”

24.

Geddes

Barbara

Wright

Joseph

Frantz

Erica

. 2014. “Autocratic Breakdown and Regime Transitions: A New Data Set.” Perspectives on Politics 12 (2): 313–31.

25.

Gehlbach

Scott

Malesky

Edmund

. 2010. “The Contribution of Veto Players to Economic Reform.” Journal of Politics 72 (4): 957–75.

26.

Gehlbach

Scott

Sonin

Konstantin

. 2014. “Government Control of the Media.” Journal of Public Economics 118:163–71.

27.

Goemans

Hein E.

2008. “Which Way Out? The Manner and Consequences of Losing Office.” Journal of Conflict Resolution 52 (6): 771–94.

28.

Gordon

Sanford C.

2002. “Stochastic Dependence in Competing Risks.” American Journal of Political Science 46 (1): 200–17.

29.

Guriev

Sergei

Treisman

Daniel

. 2015. “How Modern Dictators Survive: An Informational Theory of the New Authoritarianism.” NBER Working Paper 21136. http://www.nber.org/papers/w21136 (accessed January 8, 2018).

30.

Haggard

Stephan

Kaufman

Robert R.

. 1995. The Political Economy of Democratic Transitions. Princeton, NJ: Princeton University Press.

31.

Hollyer

James R.

Peter Rosendorff

Vreeland

James Raymond

. 2014. “Measuring Transparency.” Political Analysis 22 (4): 413–34.

32.

Hollyer

James R.

Peter Rosendorff

Vreeland

James Raymond

. 2015. “Transparency, Protest and Autocratic Instability.” American Political Science Review 109 (4): 764–84.

33.

King

Gary

Pan

Jennifer

Roberts

Margaret E.

. 2013. “How Censorship in China Allows Government Criticism But Silences Collective Expression.” American Political Science Review 107 (2): 326–43.

34.

King

Gary

Pan

Jennifer

Roberts

Margaret E.

. 2014. “Reverse-engineering Censorship in China: Randomized Experimentation and Participant Observation.” Science 345 (6199): 1–10.

35.

Linsi

Lukas

Mugge

Daniel K.

. 2017. “How Offshoring and Digitization Undermine the Quality of International Economic Statistics: A Cautionary Note.” Presented at 2017 International Political Economy Society Meetings https://www.internationalpoliticaleconomysociety.org/sites/default/files/paper-uploads/2017-11-01-09_18_38-lukas.linsi@gmail.com.pdf (accessed January 8, 2018).

36.

Little

Andrew T.

2016. “Communication Technology and Protest.” Journal of Politics 68 (1): 152–66.

37.

Little

Andrew T.

2017. “Coordination, Learning and Coups.” Journal of Conflict Resolution 61 (1): 204–34.

38.

Little

Andrew T.

LaGatta

Tom

Tucker

Joshua A.

. 2015. “Elections, Protest and Alternation of Power.” Journal of Politics 77 (4): 1142–56.

39.

Lorentzen

Peter

. 2014. “China’s Strategic Censorship.” American Journal of Political Science 58 (2): 402–14.

40.

Morris

Stephen

Shin

Hyun Song

. 2002. “Social Value of Public Information.” American Econmic Review 92 (5): 1521–34.

41.

Padró i Miquel

Gerard

. 2007. “The Control of Politicians in Divided Societies: The Politics of Fear.” Review of Economic Studies 74:1259–74.

42.

Pepinsky

Thomas

. 2014a. “Authoritarianism: Logics and Institutions.” APSA Comparative Politics Newsletter 24 (1): 3–4.

43.

Pepinsky

Thomas

. 2014b. “The Institutional Turn in Comparative Authoritarianism.” British Journal of Political Science 44 (3): 631–53.

44.

Reinikka

Ritva

Svensson

Jakob

. 2003. “The Power of Information: Evidence from a Newspaper Campaign to Reduce Capture.” Working Paper: Institute for International Economic Studies, Stockholm University, Stockholm, Sweden.

45.

Shadmehr

Mehdi

Bernhardt

Dan

. 2014. “Vanguards in Revolution: Sacrifice, Radicalism and Coercion.” https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3015658 (accessed January 8, 2018).

46.

Shadmehr

Mehdi

Bernhardt

Dan

. 2015. “State Censorship.” American Economic Journal: Microeconomics 7 (2): 280–307.

47.

Shadmehr

Mehdi

Bernhardt

Dan

. 2017. 2017. “When Can Citizen Communication Hinder Successful Revolution.” Quarterly Journal of Political Science 12(3): 301–323.

48.

Shadmehr

Mehdi

Boleslavsky

Raphael

. 2015. “Institutions, Repression and the Spread of Protest” https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2731749 (accessed January 8, 2018).

49.

Slater

Dan

. 2010. Ordering Power: Contentious Politics and Authoritarian Leviathans in Southeast Asia. Cambridge, MA: Cambridge University Press.

50.

Stone

Randall W.

2008. “The Scope of IMF Conditionality.” International Organization 62 (4): 589–620.

51.

Svolik

Milan W.

2012. The Politics of Authoritarian Rule. Cambridge, MA: Cambridge University Press.

52.

Svolik

Milan

. 2013a. “Incentives, Institutions, and the Challenges to Research on Authoritarian Politics.” APSA Comparative Democratization Newsletter 11 (2): 7–11.

53.

Svolik

Milan W

. 2013b. “Contracting on Violence: The Moral Hazard in Authoritarian Repression and Military Intervention in Politics.” Journal of Conflict Resolution 57 (5): 765–94.

54.

Wallace

Jeremy L.

2016a. “Information Politics in Dictatorships.” Emerging Trneds in the Social and Behaviorial Science: An Interdisciplinary, Searchable, and Linkable Resource 1–11. doi: 10.1002/9781118900772.etrds0181.

55.

Wallace

Jeremy L.

2016b. “Juking the Stats? Authoritarian Information Problems in China.” British Journal of Political Science 14 (1): 11–29.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

38.90 MB

0.00 MB