Research Statement Research Statement-Lee.pdf

Working Papers

(1) The Decision to Arm: Domestic Political Concerns and Armament

While my dissertation focuses primarily on alliance formation, the decision to ally is not an isolated one. Instead, leaders who choose to ally do so over other alternatives such as armament. In this context, Chapter 1 of my dissertation examines alliance formation in relation to buying weapons or doing nothing. According to the analysis, the decision to arm is also shaped by domestic politics. Leaders of different regime types have different incentive structures. If the benefits of arming, in particular the increase in the state’s own military capabilities, outweigh the advantages of allying, democratic targets will be more likely to buy weapons, whereas autocratic targets are less likely to choose to arm if forming an alliance is feasible. The analysis also shows, moreover, that even if forming an alliance is better than arming, democracies will sometimes still prefer to buy weapons. Autocracies, on the other hand, are more likely to choose alliances, regardless of the potential advantages that buying arms may provide.

(2) Armament, Alignment and Alliance in Anticipation of Potential Conflict: The Impact of Domestic Politics in Extended Deterrence (With Alastair Smith)

We propose an incomplete information model of extended deterrence in which nation A can attack nation B. If attacked, B decides whether to resist and then nation C decides whether to intervene. Nations B and C have several options to enhance their fighting capability and potentially deter nation A. For instance, nation B can invest in additional weapons, an option that is more attractive to democratic targets than autocratic ones. Alternatively, B can shift its policy position to align with C's policy goals. Such policy alignments increase the chances that C will intervene on B's behalf, but this approach to building security is relatively unattractive when B has inclusive political institutions, such as democracy. Nation C can deter A from attacking by forming an alliance with B, an option that is relatively popular for powerful democratic nations.

(3) Improving Dichotomous Foreign Policy Similarity.

One of the advantages of dichotomous measure is its tractability for improving. To improve dichotomous foreign policy similarity measurements, it is important how to deal with non-allied states.  As Signorino and Ritter (1999) pointed out, there are three instances in the “no alliance” category: no alliance because of hostility, because of irrelevancy to each other's security, and because of an implicit alignment. For not-allied states due to irrelevancy to each other’s security, we can take advantage of statistical reasoning. That is, without any additional information, and considering the high threshold of alliance formation and huge number of dyads, we may be able to say that before a security-threatening or war-prone situation emerge, prior probability is ½. That is, in time of peace, we should weight not-allied states by ½ as a baseline both on capability-weighted and unweighted alliance portfolios. In addition, it may be more critical than irrelevancy-weighting, there are cases that we can determine whether non-allied states represent hostility/alignment: warring parties in disputes/wars. That is, if non-allied states fight together with one of the states we want to measure foreign policy similarity, the non-allied states can be considered “allied” states. On the other hand, if non-allied states fight against with one of the states of the similarity measurement, the non-allied states would be “non-allied” because of hostility for sure. So, we can code “the states fighting together” as “allied” and “the states fighting against” as “non-allied” with weighting 1.




(4) Leader Specific Punishment – Evaluating Domestic Political Concerns

I extend McGillvray and Smith (2000)’s model to incorporate domestic political concerns. In the original paper, the agents’ accountabilities depend on the impatience and the costs of replacement. The payoffs are all public goods. In this paper, however, the agents’ accountabilities depend primarily on the regime type. And the payoffs consist of both public goods and private goods. In particular, the extended model specifies which strategy we should employ to induce the cooperation of a non-democratic state. That is, instead of the public goods aspect of punishment, we have to focus on the punishment of private goods in the non-democratic target country. Since public goods are all the same inside and outside the winning coalition of smaller  states, for the punishment to be effective, punishing private goods provisions is far more effective. On the other hand, in a large winning coalition system, democracy, it is more efficient for a state to punish the public goods aspect. Hence, in developing an agreement, the contents must differ depending on what kind of regime type the partner state has. If it is a democratic state, the contract should focus on the public goods of the partner state. Then, if the partner state defects, the audience in that state, which is a large winning coalition, will react promptly.  If, conversely, the partner state is autocratic, the contract should focus on the private goods aspect of the partner. Then, if the leader of the partner defects, it immediately affects the private goods provision of the state, and the smaller winning coalition will react in a desirable direction. Note that this model does not need the condition that “if both agents are accountable” in terms of public goods provision for the agent specific grim trigger (ASGT) strategy to be a subgame perfect Nash equilibrium.

Future Research

(1) Selectorate economic sanctions.

I will propose a model of selectorate economic sanctions. A few studies have applied Selectorate theory to economic sanctions, but they do not present a formal model and have some issues in testing the hypotheses, largely because they focus on comprehensive sanctions. The Selectorate sanction theory is basically a theory of targeted sanction (smart sanctions) because according to the Selectorate theory, economic sanctions will be effective only if they are aimed at the winning coalition members’ welfare. Smart sanctions have been applied to target states and non-state actors such as terrorist organizations by the UN since the mid-1990s. However, little in the way of theoretical foundations have been developed for the empirical evaluations of the effectiveness of targeted sanctions. As noted, the selectorate sanctions theory suggests that for economic sanctions to be effective, they must aim at the welfare of the winning coalitions of the targeted states. Thus, for autocratic regimes, sanctions must aim at the private goods of the small winning coalition or at the leaders’ private resources used to reward their small winning coalition members, whereas for democratic targets, sanctions must be made against public goods as this will directly affect the large coalition members’ (or voters’) welfare.

Therefore, per Selectorate theory, true regime effects can be evaluated if we consider the winning coalitions, not just the nominal regime types. Sanction effects differ between targeted sanctions and comprehensive sanctions depending on the regime type: targeted sanctions are more effective against autocracies and comprehensive sanctions are more effective against democracies. For a targeted sanction to be effective, in addition, it must decrease the prospects of political survival for the leaders of the target state. To do that, the sanction must decrease the rewards that the incumbents can provide to their winning coalition to a point far lower than the rewards that the new leaders can provide after the sanction ends. If so, the leaders of target states will surrender before the actual sanctions are implemented. This is consistent with Smith (1995).

Therefore, the more the effective the sanctions, the less likely that implemented sanctions will be successful. This explains why targeted (actually implemented) sanctions have not been so successful and in some cases have even led to worse outcomes. Because of the selection effect, to assess the success of economic sanctions, we have to consider both the threat to sanction as well as the actually implemented sanctions. Therefore, I will use the Threat and Imposition of Sanctions (TIES) data 4.0 (Clifton, Bapat, and Kobayashi 2014) and the Targeted Sanctions Consortium dataset (TSC), instead of the widely used Hufbauer, Schott, and Elliott (HSE) dataset.

(2) Keeping alliances: the guns vs butter tradeoff

As in my dissertation, alliances are formed because there are military (security) and policy (autonomy) gains from the alliance. Once those gains are embedded in domestic political resource distributions, the leaders of alliances can consistently save resources for their own good as long as they maintain the alliances. In particular, if an alliance is terminated, a defender has to pay the costs of the policies that the alliance target previously paid by concessions under the ex-alliance.

A target, on the other hand, has to make up the decreased military capabilities due to the disappearance of the defender or risk being exposed to potential threats (if power decreases, security decreases). Usually, there is a division of labor in force structures under an alliance, so a target may have developed an asymmetric force structure that is balanced by the defender. Therefore, even though the target may not need to increase her military capabilities, she may need to spend a considerable amount of resources to fill in the gap in force structure after the termination of the alliance. Furthermore, considering synergy effects of the alliance, the cost to make up the vacancy of the defender can even be higher. Therefore, an alliance termination may decrease the leaders’ discretionary resources – for both defenders and targets – that previously had been used for their own political survival, which is ultimately disadvantageous for their ongoing political survival.

In case of democratic defenders, if they terminate alliances, they have to pay the policy costs that were previously taken up by targets as well as reputational costs. Since autocratic defenders, on the other hand, are less sensitive to public goods related policy issues, they have less incentive to maintain alliances compared to democratic defenders, holding other factors constant. Democratic targets must weigh the military costs due to the disappearance of a defender against the savings due to discontinuing the policy concessions if they terminate an alliance. If the savings are greater than the costs, they have an incentive to terminate the alliance. To autocratic targets, on the other hand, the savings from stopping policy concessions may not be large because from the beginning, they would have been less concerned about the policy issues that did not pay much for them. Therefore, relative to their democratic counterparts, autocratic targets tend to be influenced more by the military costs if they terminate an alliance, and they have less incentive to terminate the alliances compared to democratic targets. In general, holding other factors constant, the alliance durability of democratic dyads depends more on the relative size of democratic targets’ savings from policy gains to military costs caused by the military void due to an alliance termination. The relative size, however, depends on each target’s policy issue salience such that the durability can be either longer or shorter than other pairs of regime types. In the mean time, autocratic defenders and democratic targets are most likely to terminate alliances if the threat is gone.

(3) Civil war and selectorate theory

The selecotorate theory is a theory that connects domestic politics to international politics. Its core insights, however, can also be carried over to the study of intrastate war. Unlike the states as actors that selectorate theory typically addresses, civil wars can happen within states that are still in the course of institutionalizing the loyalty norms (W/S) or after the existing institutions collapse. To apply selectorate theory to civil wars, we should consider the difference between institutionalized regimes and uninstitutionalized regimes.

If governments have dominant force, it is difficult for civil wars to occur; this is why civil wars typically occur in not-fully-institutionalized or collapsed states. The abrupt change in balance of power between governments and rebel groups can be achieved through the collapse of the central government, power vacancies when the former dominant nation retires due to colonial independence, foreign government support for anti-government forces, and acquisitions of new resources.

When domestic political arrangements are not fully institutionalized, the affinity of the winning coalition members cannot be taken as the same and, essentially, some winning coalition members may be leaders of their own regions within a state with military forces. In those regions, the leaders may also have their own winning coalitions to remain as leaders. According to selectorate theory, to expel an incumbent leader, at least one winning coalition member has to be obtained by a challenger. To survive, the incumbent has to provide the rewards that equal or surpass the expected rewards that the challenger is expected to provide. If we apply the framework of selectorate theory to civil wars, civil wars can be understood as the process of trying to obtain the national or the challenger’s regional resources exclusively, and to redefine and purge the selectorate and the winning coalition by instigating and mobilizing people using ethnicity, religion and/or ideology. 

Redefining and purging the selectorate and the winning coalition have two effects. First, it affects the loyalty norm, or the probability that the current winning coalition members will be included in the new regime. By making it clear who will be purged through the redefining selectorate, the challenger may be able to induce the betrayal of some of the current winning coalition members including the leader herself. Second, it affects the size of rewards that the winning coalition members will receive. By reducing the selectorate size and the winning coalition size, the challenger can increase the expected size of rewards for winning coalition members, which also contributes to seducing some of the current winning coalition members. 

Redefining and purging the selectorate and the winning coalition may appear in the form of genocide or politicide.

Whatever it is, if political leaders can mobilize people and redefine the selectorate and the winning coalition, this can be an excuse for civil wars. Representative examples for excuses for such mobilization tend to be ideological and identity-based, because they are the most accessible and appealing pretexts. In other words, ethnicity, religion, and ideology can send a signal of the expected reward to some of the current and potential winning coalition members by informing them how the selectorate and winning coalitions will be redefined in the new regime. The decline of civil war from 1995 may not be due to international arbitrations or pressures, but rather to the settlement of national or local level loyalty norms through purges and redefinition.

In this context, a civil war has a considerably higher mortality and duration than a general international war, in that it seeks to "replace the leader" with a challenger. This occurs because, following the redefinition of the selectorate, it is necessary to kill or expel people who do not fit the redefined selectorate, and often to “replace the leader”, which result in the strongest motivations both to the incumbent and the challenger. Therefore, we might need data on the factions and the winning coalitions and selectorates of the factions in the states in civil wars or potential civil wars. However, I think that the rebel groups can be assumed autocratic in general, whereas the governments can either be democratic or autocratic. Fortunately, we have data on governments’ regime types. So, based on the assumption that the rebel groups are autocratic, we may apply the selectorate theory to civil wars. Data on the military capabilities of each faction are also needed.

As mentioned earlier, regarding the onset of civil wars, civil wars may occur if the existing balance of power between the government side and the anti-government side breaks. Opportunity or power increases can even create intentions that were not even present before.

Since revolutionary civil wars, if successful, secure exclusive access to resources for the entire country, the expected rewards are usually great. Therefore, civil wars can occur even with relatively low probability of winning, as rebel groups may wage civil wars even under relatively poor power conditions. On the other hand, in a secessionist civil war, the secessionist’s expected resources are limited to the region, so the utility is relatively small after independence. Therefore, seccesionist civil wars are likely to occur when the power is relatively larger than in the revolutionary case.