Essay I: What differences in public policy can be made by an establishment of a legislature through an election in authoritarian regime? In particular, how high tax rates can be formed in authoritarian regimes with a legislature? Paying attention to a legislature's role of a signal which conveys information about dictator's economic policy preferences to capital owners, this essay revisits Escriba Folch's simple signaling game. By relaxing some restrictive assumptions about dictator types, and considering the dictator's capability of achieving his tax rate through the legislative process, the model gives answers to some empirical puzzles: 1) why some dictators are willing to maintain the legislature formed through elections; and 2) why capital owners would move their mobile assets abroad in some conditions even when a dictator allows a legislature to be established through a competitive election. The model claims that a dictator who is not tax-benevolent has an incentive to misrepresent his actual tax rate, and the probability is inversely associated with his capability in the legislature. After observing the

legislature not dissolved, capital owners who believe the dictator is not tax-benevolent are not willing to move their mobile assets away. On the other hand, after observing the legislature maintained by a dictator who is less capable of achieving his tax rate, capital owners consider moving their mobile assets away only when they believe that the dictator is tax-benevolent. Thus, it is more likely to see relatively low tax rates under capable dictatorships with legislatures. It is ironical that dictators who inherently have low tax rates to invigorate the economy by inducing a higher level of investment cannot enjoy the benefits of maintaining (or creating) a legislature. Case studies of two military regimes in Korea (1961-1987) demonstrate that the predictions made by the theoretical model are empirically supported in the Korean cases.

Essay II: Assuming that electoral incentives of three political actors (individual legislators, a ruling party, and a president) may not be aligned with each other in common institutional settings, this essay attempts to construct an integrated theory about the relationship between the allocation of intergovernmental grants and the political actors. It empirically tests three hypotheses derived from the theory by examining the case of the

Special Local Allocation Grants in Korea (2005-2006). To properly capture regional variation in the allocation, this essay employs the multilevel linear regression model in the Bayesian framework. First, the individual legislator's membership of the specific committee to monitor the execution of the intergovernmental grants is positively associated with the amount of the grants, which supports the Legislators' capability hypothesis, as in the classical regression models. Second, vote margin at the district level is positively associated with the amount of the grants, which provides strong evidence against the Unstable Electoral Districts Hypothesis, however. The result from Bayesian multilevel linear model supports the Unstable provinces hypothesis which states a significant positive association between the amount of intergovernmental grants delivered at the province level and being an electorally unstable province within a broader region in which voters are motivated by their regional identities even after controlling for the need-based criteria. It implies that it is more efficient to target an electorally unstable province even within a supporter region because voters affiliated with a regional (or ethnic) identity in the electorally stable province may not resent the allocation of grants even if they are not the main beneficiaries. Consequently, the distribution of the grants at the higher level can be decided by the efficient targeting strategy, whereas the grants tend to be delivered to strong supporters at the district level. These statistical results are consistent with the ideas that this essay adopts. During the period under investigation the liberal president who could not be reelected by the constitution intended to secure his key policies after his retirement. To help his successor from his faction to earn more votes in the future presidential election, he was willing to allocate considerable amount of government resources to some opposition districts, which may not be completely aligned with the target strategy of the ruling party who sought to maximize the number of seats in the National Assembly under Single-Member District Plurality rule. Regardless of their respective party lines the specific committee members who wanted to be reelected attempted to deliver more grants to their own districts.

Essay III: Why do some rising powers unilaterally declare an ideal ordering principle against regional powers deeply attached to existing principles although it is not likely to be realized in the near future? Under what circumstances can such an ideal principle be successfully implemented in the region or escalate into an armed conflict? This paper aims to form a game-theoretic model of bargaining between two rival states with potentially incompatible ordering principles, and provide solid evidence from U.S.-Japanese conflicts in Northeast Asia in the early twentieth century. Constructed on two-sided uncertainty, the model coherently explains why the United States, a new external force in Northeast Asia, was deeply attached to the Open-Door principle as a practically efficient option in bargaining with Japan, a regional power, which adhered to the partition principle, but the potential conflict escalated into a war in the end. Fifteen historical cases from 1899 to 1941 strongly support the predictions made by the model.

This thesis aims to properly test three hypotheses derived from

existing political theories about distributive politics by employing

Bayesian multilevel modeling. The specific case of

intergovernmental grants, the Special Local Allocation Grants,

in Korea at two nested levels (districts and provinces) from 2005 to

2006, verifies that unlike classical regression models, the Bayesian

multilevel regression model can capture regional variations in the

allocation and utilize substantive knowledge from previous

literature. In particular, the model finds that a significant positive

association between the amount of intergovernmental grants and being

an electorally unstable province in a broad region affected by

regional voting behavior (i.e., Electorally Unstable Provinces Hypothesis) even

after controlling for the need-based criteria. It justifies the chief

executive's strategy to target an electorally unstable [swing]

province even within a supporter region because people in the

electorally stable province are strongly affiliated with a regional

(or ethnic) identity so that they may be satisfied with the allocation

of grants even if they are not the main beneficiaries. Thus, while the

allocation is concentrated on core supporters that are well known

quantities at the district level, the allocation at the higher level

can be decided by the efficient targeting strategy. This finding provides a

strong implication for decentralized democratic governments under

circumstances where significant regionally (or ethnically)

affiliated-voting is observed. In Korea, the disproportional

allocations of central government grants to electorally unstable

province within its supporter region (Jeolla) from 2005 to 2008 helped

the government party to increase its vote share in the unstable

province (Jeonam) by 20% in 2008. The vote share in the electorally stable province (Jeonbuk) reduced only by 2.9\% in comparison with that in the previous election. It was a remarkable outcome, considering the the government party was defeated by the wide margin 13% nationwide and experienced a swing against it by 20.3% across the country in the election.

Objectives

Background

Methods

Results

Conclusions