| I revisit the security externality hypothesis of trade, and then attempt to reconfirm it in an entirely different way. One of the most famous theoretical explanations of the relationship between alliance and trade is that "trade follows the flag" because of security externalities. Several empirical studies have confirmed that stronger alliance commitments would have larger positive effects on trade. But, those studies have two serious problems.;First, according to the gravity model of trade in international economics, trade flows may be a result of trading partners' economic size, geographic proximity, and factor endowments. Put differently, since trade flows are a very indirect measurement of a country's general trade policy, they cannot capture policy makers' intentional, political, or strategic decisions to increase positive security externalities of trade.;Second, the positive relationship between alliance and trade is inconclusive. For example, one group found that alliances generally increase trade, whereas another group found that defense pacts are positively associated with trade among members, but that trade between members of neutrality, nonaggression, or consultation pacts is statistically indistinguishable from trade between non-allies. The other group reported that alliances generally do not increase trade.;I propose two ways to resolve these puzzles. First, regional integration is employed as a dependent variable. Regional integration is a direct measurement of political trade preferences to lower tariffs and non-tariff trade barriers. Moreover, unlike other trade agreements lacking a contiguous institutional framework, it can capture a developmental institutional process moving through numerous phases in orderly fashion.;Second, given the uncertainties of international politics, alliance pacts alone may not guarantee positive security externalities. Even defense pacts may be incredible due to a high level of policy flexibility, which can reverse past commitments. In this sense, the interaction among (1) the level of alliance commitments (2) the number of domestic veto-points associated with allies' policy flexibility and (3) whether or not alliance pacts have been ratified are expected to have much stronger positive effects on regional integration than alliance itself. This expectation is supported very strongly by both case studies and empirical analyses based on two alliance data (ATOP and COW), two samples (Total and Total-EU), and four statistical methods (OLS regression with PCSEs, interval regression, generalized ordered logit regression, and binary logit regression).;Especially, these results are found to hold even when the possibility of endogeneity between alliance (reliability) and regional integration is controlled. Besides, I construct the full model of regional integration. It includes seventeen independent variables discussed long in international politics and international economics since the 1940s. As expected, alliance reliability is found to be the best predictor of regional integration. In contrast, other variables' signs and statistical significance are very weak or whimsical.;In addition, an in-depth analysis of outliers unexplained by the full model identifies seven idiosyncratic causes, four general causes to be added back into the statistical model, and one systemic measurement error. Lastly, when evaluated from a viewpoint of Lakatos's methodology of scientific research programs, this dissertation project succeeds in adding novel facts to the IR literature (especially, security externality research program of trade and alliance reliability research program) in progressive ways. |
