The textile bill of 1985: The determinants of congressional voting patterns

The following tables show the results of the probit analysis for the Senate and the House. A table including results for different combinations of the independent variables is included in an unpublished appendix available from the authors upon request. The results of the probit analysis for the Senate are shown in Table 1. All the variables except EXP had the expected signs, however the results indicate that the two other employment variables and party affiliation are the only factors which we can say with confidence influenced a Senator's decision to vote for or against this bill. The coefficients reported and labeled ‘der of prob’ measure the derivative of the probability of a yes vote evaluated at the means of the independent variables in question. Thus, for example the coefficient for UNEMP in the top line of Table 1 should be interpreted as indicating that a one percentage point increase in a state's unemployment rate will cause a 5.89% increase in the probability that the Senator will vote for the bill. The result for DR, the party of the Senator, shows that in general Democrats are more apt to vote for protectionist policies than are Republicans. This is consistant with President Reagan's negative stand on the bill, and his threat to veto it should it pass. The amendment in question passed the Senate on 24 October 1985, by a vote of 54 for to 42 against, but the bill it was attached to has been postponed indefinitely. The House bill, which was passed on 10 October 1985 by a vote of 262 for to 159 against, was vetoed by Reagan on 17 December 1985. The results of the probit analysis for the House of Representatives are shown in Table 2. State employment data was used for the House as opposed to data for each voting district. This was done for three reasons. First, employment data is usually reported by state or by county. Since counties are often divided by congressional district boundaries, much personal judgement would be necessary to assign employment figures. Second, the most recent county industry employment data is from 1981, while the data used in this analysis is from 1984. Third, we feel that in general ‘logrolling’ is prevalent among a state's representatives. For this bill, the representatives from twenty six states voted as a block, and in all but six states one third or less of that state's representatives voted contrary to the majority. All of the probabilities are significant at the .05 confidence level. Representative's voting patterns are influenced by all of the included factors. When unionization is added to the equation, its coefficient is very small and negative although it is significant at the .05 confidence level. One would expect more significance here since there are more degrees of freedom for the House analysis than for that of the Senate. When the data from the House and Senate are pooled the results are similar to those of the House alone. The null hypothesis that the data cannot be pooled was tested using a χ 2 statistic with eight degrees of freedom. χ 2 =9.721, meaning that the null hypothesis is rejected. As in the Senate, the state unemployment rate and the percent employed in textiles are important influences on the Representative's vote, and Democrats in the House are more likely to vote for protectionist bills than Republicans. However, unlike the Senate, the percent of a Representative's campaign contributions that are donated by industry special interest groups has a significant effect on voting patterns, although the point estimates for the magnitudes of the influence are very similar in both cases. Seventy-four percent of the House received campaign contributions from these special interest groups, while only 54% of the Senate received contributions from these same sources. This is probably because only one third of the Senate was up for reelection during the time period considered, while the entire House was in the process of campaigning for reelection. Exports also play a role in influencing the probability of a ‘yes’ vote, reducing this probability as employment in export industries increases, but the magnitude of the coefficient is only 11% of that for textile employment. The significance of this variable in the House is probably due to fear of retaliation, expressed by both Democrats and Republicans. This variable was not significant in the Senate model, most likely because the Senate version excluded from the 1% import quota growth rate many countries who might have retaliated. Representatives appear to respond to a wider variety of constituent interests than do Senators. While Senators' voting patterns are only significantly influenced by their political affiliation, their state's unemployment rate and the importance of textiles to their state, the voting behavior of Representatives is influenced by these factors as well as the importance of exports to their state and campaign contributions by textile special interest groups. These results affirm the validity of the approach used by Baldwin and Coughlin. They differ from Baldwin who finds exports are not a significant influence on the House, and from Coughlin, who does not include export employment as an explanitory variable or analyze the differences in voting patterns between the House and the Senate. Copyright Martinus Nijhoff Publishers 1987