Costly signaling in autocracy

ABSTRACT

Those who would revolt against an autocrat often face a dilemma caused by uncertainty: they would like to revolt if the ruler would respond with democratization, but they would prefer to concede if the ruler would choose instead to violently suppress the revolution. Consequently, the autocrat must decide how to best signal his willingness to use violence in hope of deterring revolt. Using a simple signaling model, we find that rulers cannot meaningfully convey their type by transferring wealth to the citizenry. However, they can convey their type through shows of force, as long as the strong type of autocrat – who would use violent repression in the case of revolution – has a competitive advantage in displaying his strength. We additionally demonstrate that rulers favor shows of force when their willingness to suppress revolution is questioned and that citizens at times prefer to pay the direct cost of shows of force to learn about the ruler’s type, rather than to remain uninformed. The results illustrate a more general result in costly signaling models: information transmission is only possible when the cost of the signal is smaller for the type that wants to distinguish himself.

Quienes se rebelarían contra un autócrata a menudo se enfrentan a un dilema causado por la incertidumbre: les gustaría rebelarse si el gobernante respondiera con la democratización, pero preferirían ceder si el gobernante optara por reprimir violentamente la revolución. En consecuencia, el autócrata debe decidir cuál es la mejor manera de señalar su predisposición a utilizar la violencia con la esperanza de disuadir la rebelión. Mediante la utilización de un modelo de señalización simple, descubrimos que los gobernantes no pueden transmitir de manera significativa su tipo al transferir riqueza a la ciudadanía. Sin embargo, pueden transmitir su tipo mediante demostraciones de fuerza, siempre y cuando el tipo fuerte de autócrata, que utilizaría la represión violenta en caso de una revolución, tenga una ventaja competitiva para demostrar su fuerza. Además, demostramos que los gobernantes favorecen las demostraciones de fuerza cuando se cuestiona su voluntad de reprimir la revolución y que los ciudadanos a veces prefieren pagar el costo directo de las demostraciones de fuerza para conocer el tipo de gobernante, en lugar de permanecer desinformados. Los resultados demuestran un resultado más general en modelos de señalización costosos: la transmisión de información solo es posible cuando el costo de la señal es menor para el tipo que quiere distinguirse.

Les personnes souhaitant se révolter contre un autocrate sont souvent confrontées à un dilemme causé par l’incertitude: elles voudraient se révolter si le dirigeant réagissait par la démocratisation, mais elles préféreraient céder si le dirigeant décidait plutôt de réprimer la révolution par la violence. Par conséquent, l’autocrate doit décider du meilleur moyen de signaler sa volonté d’avoir recours à la violence dans l’espoir de dissuader la révolte. Nous avons utilisé un modèle de signalization simple et constaté que les dirigeants ne pouvaient pas communiquer leur type de manière significative en transférant des richesses aux citoyens. Ils peuvent toutefois le communiquer par des démonstrations de force tout le temps qu’il est puissant – qu’il indique qu’ils auraient recours à une répression violente en cas de révolution – et qu’ils ont un avantage compétitif à afficher leur force. De plus, nous démontrons que les dirigeants favorisent les démonstrations de force lorsque leur volonté de réprimer la révolution est remise en question et que les citoyens préfèrent parfois payer le coût direct des démonstrations de force afin d’en savoir plus sur le type du dirigeant plutôt que de rester non informés. Nos conclusions illustrent un résultat plus général pour les modèles de signaux coûteux: la transmission d’informations n’est possible que lorsque le coût du signal est moindre pour le type d’autocrate qui souhaite se distinguer.

Notes

¹ See Gehlbach and Svolik (2016) for an overview of the literature on formal models of autocracy.

² As is conventional, the policies also have direct effects in the model with transfers making loyalty to the autocrat more attractive and shows of force making revolution more costly.

³ They are different than cooptation, which could entail institutional design and may be related to democratization (Acemoglu and Robinson 2006) or could make future repression more efficient (Taylor 2011). Both transfers and shows of force also require state capacity; we here hold state capacity constant and assess the signaling ability of each of these actions.

⁴ Though we follow Machiavelli's logic and focus on one-sided incomplete information, it should be noted that informational problems can be two-sided (Wintrobe 1998). For example, the stronger an autocrat is, the less the citizenry is willing to say in political discourse, meaning the less the autocrat knows about the citizens’ beliefs and preferences.

⁵ Similar to Crescenzi (1999), the Ruler’s type can be thought of as whether he is hard- or soft-line.

⁶ It is plausible that the marginal cost of transfers is lower for the Weak Ruler. Perhaps he is better equipped to provide transfers. This lower cost would enable the Weak Ruler to signal its weakness, but no government would want to communicate that information (reinforcing the results below). In equilibrium, larger transfers are incentive compatible for the Weak Ruler, because the cost of revolution is higher for him. This behavior is derived from the model.

⁷ None of the results depend on show of force having a direct cost, but because it sometimes does and this could go against our findings, we wanted to allow the cost to enter the model.

⁸ If we instead assumed that the leader makes the transfer regardless of the citizens’ decision, the citizens’ decision would be independent of the transfer, as they would receive it whether they revolt or not. This objective function abstracts from commitment problems by the government.

⁹ Although $\sigma$ has a direct effect, the reduction in utility occurs whether the Citizens revolt or back down, so it does not affect the citizens’ decision.

¹⁰ Recall the leader’s utility function: $YB-a_{\theta }\sigma -\tau$. The marginal cost of transfers is $1$, which does not depend on shows of force. The marginal cost of showing force is $a_{\theta }$, which does not depend on transfers. In equilibrium, the ruler will select the policy, transfers, or shows of force, that yields the smaller total cost. The decision is a bit more complex than just comparing the marginal cost, $1\stackrel{>}{<\phantom{{ }_x}}{a}_{\theta }$, because the size of the transfer and shows of force needed to prevent revolution differ.

¹¹ There is also no semi-separating or semi-pooling equilibrium, as these require the ruler to mix between the strategies played in the separating and the pooling equilibrium (which means he must be indifferent between them). Since the separating equilibrium is never supportable in pure strategies, it also cannot be played as part of a mixed strategy equilibrium – the ruler would never play a dominated strategy with positive probability in a mixed strategy equilibrium.

¹² The Citizens’ expected utility in the separating equilibrium is $\alpha [Y(1-B)-\sigma ]+(1-\alpha )[Y(1-B)+YBt$, which is always greater than their payoff from the pooling with revolt equilibrium $\alpha [Y(1-B)-c_C]+(1-\alpha )[Y(1-B)+YBt]$, as $\sigma <c_C$.

¹³ Police shot and killed at least two protestors on March 10, 2011, but violence was still limited (Black March 11, 2011).