American Political Science Review Vol.98,No.1 TABLE 2.Summary of Model for Policy Liberalization Result Variable Predict Japan EU Operationalization(Measure) Linkage × + Strength of cross-sector linkage (1-4) Threat × Specificity of threat (1-4) Trade balance × Bilateral trade balance in goods(U.S.billion $ Product value + Product share in total U.S.agr.exports(%) Politics + × NA LDP share of Lower House seats(%) NA French presidency of council(1-3) Budget Annual agriculture budget growth(%) Growth Annual growth(per capita GDP)(%) Multi + Multiple-country demand(0-1) Export NA Case about export subsidy(0-1) Count + Case number in series of repeated negotiations(1-6) Note:Variables represent those used in the regression analysis.The second column gives the expected direction of the variable's influence on liberalization outcomes.The Result columns summarize the direction of influence for statistically significant regression estimates(base models in Table 3),which are discussed in text.NA,not applicable. a moderate linkage.Cases in multisector negotiations area where GATT trade law has been weakened by with a weak linkage in the agenda are coded two,and exemptions and vague language(Hudec 1993,147-54). negotiations that only addressed agricultural topics are The combination of stronger interests and weaker rules coded one. makes these among the toughest negotiations with Additional variables represent factors that other the- the EU. ories have highlighted as important for trade negotia- tions.The coercive influence of U.S.threats is measured EMPIRICAL ANALYSIS by a four-level ordinal scale,ranging from absence of threat to a specific threat or the initiation of retaliatory I evaluate the effect of issue linkage by analyzing data sanctions.Two variables,the bilateral trade balance from U.S.-Japan negotiations and U.S.-EU negotia- and the value of the specific product in the profile of tions.I choose not to combine the Japan and EU U.S.agricultural exports,control for possible sources datasets since separate analysis allows me to examine of variation in the pressure from the U.S.demand for whether the same variables have different effects in liberalization. each political context.Given that the dependent vari- The idea that domestic political constraints may add able measures policy liberalization with three ordered bargaining leverage is represented in the Japan model categories,I use the ordinal probit model. by the strength of the LDP majority.A proxy of tied Table 3 presents the regression analysis.The results hands in the EU analysis is the power of France over the show that in Japan and Europe,cross-sector linkage negotiation,which is assumed to be relatively greater significantly increases the likelihood of liberalization when it holds the presidency of the EU Council of Min- There is also evidence that threats and LDP strength isters during the initial or concluding year of the nego- promote liberalization by Japan and that the role of tiation.Domestic political constraints are expected to France along with budget conditions influences liber- be higher when the LDP is weak or when France holds alization by the EU.The impact of these other factors, the EU presidency. however,appears to be smaller compared with the im- A variable for agriculture budget growth tests the pact of cross-sector linkage. argument that the need to reduce spending motivates The findings were robust to different model specifica- liberalization.Periods of budget growth apply less pres- tions.The results were consistent when a probit model sure for liberalization than when there are budget cuts. is estimated using a recoded liberalization variable that A variable for economic growth controls for other eco- collapsed the top two categories of major and minor lib- nomic factors that may influence decisions about trade eralization to form a simple binary outcome.Recoding policy. linkage into a binary variable for whether the agenda Finally,I include an indicator variable for multiple included only agricultural issues or multiple sectors countries being involved in the negotiation.This allows also produces a significant positive effect.Results were me to evaluate the effect of linkage,while controlling also consistent when I introduced a time period vari- for the fact that most negotiations with linkage also able to account for interconnection between cases not involve many participants. captured by the other variables.15 As a test for omit- For the EU,another variable is necessary to account ted variables related to the different agricultural prod- for the special nature of negotiations dealing with pro- ucts,I ran a fixed effects model including commodity tection oriented toward export promotion.Since Japan has minimal agricultural exports and no export subsi- dies,this variable is not applicable to the Japan model. 15 The time period variable divided the 30 years into five categories In the EU policy context,however,export subsidies related to the two major trade rounds:1969-72(pre-Tokyo Round). are a core policy of CAP.Moreover,they represent an 1973-79 (Tokyo Round),1980-85 (pre-Uruguay Round),1986-94 (Uruguay Round),and 1995-99(WTO). 161 Reproduced with permission of the copyright owner.Further reproduction prohibited without permission.Reproduced with permission of the copyright owner. Further reproduction prohibited without permission