140 THE REVIEW OF ECONOMICS AND STATISTICS TABLE 2.-DETERMINANTS OF IMMIGRATION-POLICY PREFERENCES:TESTING THE HECKSCHER-OHLIN AND FACTOR-PROPORTIONS ANALYSIS MODELS 1992 1994 1996 Regressor Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 Occupation Wage -0.349 -0.811 -0.541 (0.130) (0.135) (0.133) Education Years -0.044 -0.074 -0.059 (0.010) (0.011) (0.012) Gender -0.022 -0.008 0.022 0.083 -0.020 0.024 (0.048) (0.046) (0.056) (0.054) (0.060) (0.057) Age -0.000 -0.002 0.000 -0.002 0.004 0.002 (0.001) (0.001) (0.002) (0.002) (0.002) (0.002) Black -0.207 -0.225 -0.222 -0.211 -0.238 -0.241 (0.080) (0.080) (0.091) (0.092 (0.096) (0.097) Hispanic -0.064 -0.122 -0.306 -0.360 -0.124 -0.172 (0.111) (0.110) (0.136 (0.137) (0.120) (0.121) Immigrant -0.158 -0.150 -0.213 -0.193 -0.220 -0.207 (0.066) (0.066) (0.076 (0.076 (0.087) (0.087) Party ID 0.003 0.008 -0.006 -0.002 -0.023 -0.016 (0.013) (0.013) (0.016 (0.016) (0.016) (0.016) Ideology 0.057 0.050 0.054 0.041 0.080 0.072 (0.020) (0.020) (0.028) (0.029) 0.025) (0.025) Number of observations 2485 2485 1795 1795 1714 1714 These results are multiple-imputation estimates of ordered-probit coefficients based on the ten imputed data sets for each year.Each cell reports the coefficient estimate and (in parenthesis)its standard error.In both models.the dependent variable is individual opinions regarding whether U.S.poliey should increase.decrease.or keep the same the annual number of legal immigrants.This variable is defined such that higher (lower)values indicate more estrictive fless-restrictive) policy prefe nces.For bre ed C.Econometric Model correlation is strongest in high-immigration labor markets. as hypothesized in the area-analysis model.To allow for any Our empirical work aims to test how skills and other factors affect the probability that an individual supports a differences across our three survey years,we estimate each cross section separately certain level of legal immigration.The level of immigration preferred by a respondent could theoretically take on any value,but we do not observe this level.We observe only V.Empirical Results whether or not the respondent chose one of five ordered categories.Because we have no strong reason to think,ex A. Testing How Skills Affect Immigration-Policy ante,that these five ordered categories are separated by Preferences equal intervals,a linear-regression model might produce Our initial specifications allow us to test the HO and biased estimates.The more appropriate model for this situ- factor-proportions analysis models.Table 2 presents the ation is an ordered probit which estimates not only a set of results for each year's full sample,where in model 1 we effect parameters but also an additional set of parameters measure skills with Occupation Wage and in model 2 we use representing the unobserved thresholds between categories. Education Years.The key message of table 2 is that,by In all our specifications,we estimate an ordered-probit either measure,skill levels are significantly correlated with model in which the expected mean of the unobserved Immigration Opinion atat least the 99%level.Less-skilled preferred immigration level is hypothesized to be a linear (more-skilled)individuals prefer more-restrictionist (less- function of the respondent's skills,a vector of demographic restrictionist)immigration policy.This skills-preferences identifiers,political orientation,and(perhaps)the immigra- link holds conditional on a large set of plausible non- tion concentration in the respondent's community.The key economic determinants of Immigration Opinion.Among hypothesis we want to evaluate is whether more-skilled these other regressors,Gender;Age,Hispanic,and Party individuals are less likely to support restrictionist immigra- Identification are mostly insignificantly different from zero. tion policies as predicted in the HO trade model and in the Black and Immigrant are mostly significantly negative: factor-proportions analysis model.Accordingly,in our base- blacks,and the group of immigrants plus children of immi- line specifications,we regress stated immigration-policy grants,prefer less-restrictionist immigration policy.Ideol- preferences on skills,demographic identifiers,and political ogy is significantly positive:more-conservative individuals orientation.In a second set of specifications,we also include prefer more-restrictionist immigration policy.Our nonskill a dummy variable indicating whether or not the respondent estimates are similar to those in Citrin et al.(1997)and lives in a high-immigration area and an interaction term Espenshade and Hempstead(1996).14 between this indicator and the respondent's skills.These second specifications test whether the skills-immigration 14 Appendix A table Al reports results for the table 2 specifications estimated on the listwise-deletion data sets for each year.The qualitative 52.2%of the sample was female,11.5%was black,and the average age results are similar to those discussed in the paper using multiple imputa- was43.3. tion.However,using conventional rules for inference,the statistical This content downloaded from 202.120.14.193 on Mon,15 Feb 2016 10:04:26 UTC All use subject to JSTOR Terms and Conditions140 THE REVIEW OF ECONOMICS AND STATISTICS TABLE 2.-DETERMINANTS OF IMMIGRATION-POLICY PREFERENCES: TESTING THE HECKSCHER-OHLIN AND FACTOR-PROPORTIONS ANALYSIS MODELS 1992 1994 1996 Regressor Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 Occupation Wage -0.349 -0.811 -0.541 (0.130) (0.135) (0.133) Education Years -0.044 -0.074 -0.059 (0.010) (0.011) (0.012) Gender -0.022 -0.008 0.022 0.083 -0.020 0.024 (0.048) (0.046) (0.056) (0.054) (0.060) (0.057) Age -0.000 -0.002 0.000 -0.002 0.004 0.002 (0.001) (0.001) (0.002) (0.002) (0.002) (0.002) Black -0.207 -0.225 -0.222 -0.211 -0.238 -0.241 (0.080) (0.080) (0.091) (0.092) (0.096) (0.097) Hispanic -0.064 -0.122 -0.306 -0.360 -0.124 -0.172 (0.111) (0.110) (0.136) (0.137) (0.120) (0.121) Immigrant -0.158 -0.150 -0.213 -0.193 -0.220 -0.207 (0.066) (0.066) (0.076) (0.076) (0.087) (0.087) Party ID 0.003 0.008 -0.006 -0.002 -0.023 -0.016 (0.013) (0.013) (0.016) (0.016) (0.016) (0.016) Ideology 0.057 0.050 0.054 0.041 0.080 0.072 (0.020) (0.020) (0.028) (0.029) (0.025) (0.025) Number of observations 2485 2485 1795 1795 1714 1714 These results are multiple-imputation estimates of ordered-probit coefficients based on the ten imputed data sets for each year. Each cell reports the coefficient estimate and (in parenthesis) its standard error. In both models, the dependent variable is individual opinions regarding whether U.S. policy should increase, decrease, or keep the same the annual number of legal immigrants. This variable is defined such that higher (lower) values indicate more-restrictive (less-restrictive) policy preferences. For brevity, estimated cut points are not reported. C. Econometric Model Our empirical work aims to test how skills and other factors affect the probability that an individual supports a certain level of legal immigration. The level of immigration preferred by a respondent could theoretically take on any value, but we do not observe this level. We observe only whether or not the respondent chose one of five ordered categories. Because we have no strong reason to think, ex ante, that these five ordered categories are separated by equal intervals, a linear-regression model might produce biased estimates. The more appropriate model for this situ￾ation is an ordered probit which estimates not only a set of effect parameters but also an additional set of parameters representing the unobserved thresholds between categories. In all our specifications, we estimate an ordered-probit model in which the expected mean of the unobserved preferred immigration level is hypothesized to be a linear function of the respondent's skills, a vector of demographic identifiers, political orientation, and (perhaps) the immigra￾tion concentration in the respondent's community. The key hypothesis we want to evaluate is whether more-skilled individuals are less likely to support restrictionist immigra￾tion policies as predicted in the HO trade model and in the factor-proportions analysis model. Accordingly, in our base￾line specifications, we regress stated immigration-policy preferences on skills, demographic identifiers, and political orientation. In a second set of specifications, we also include a dummy variable indicating whether or not the respondent lives in a high-immigration area and an interaction term between this indicator and the respondent's skills. These second specifications test whether the skills-immigration correlation is strongest in high-immigration labor markets, as hypothesized in the area-analysis model. To allow for any differences across our three survey years, we estimate each cross section separately. V. Empirical Results A. Testing How Skills Affect Immigration-Policy Preferences Our initial specifications allow us to test the HO and factor-proportions analysis models. Table 2 presents the results for each year's full sample, where in model 1 we measure skills with Occupation Wage and in model 2 we use Education Years. The key message of table 2 is that, by either measure, skill levels are significantly correlated with Immigration Opinion at at least the 99% level. Less-skilled (more-skilled) individuals prefer more-restrictionist (less￾restrictionist) immigration policy. This skills-preferences link holds conditional on a large set of plausible non￾economic determinants of Immigration Opinion. Among these other regressors, Gender, Age, Hispanic, and Party Identification are mostly insignificantly different from zero. Black and Immigrant are mostly significantly negative: blacks, and the group of immigrants plus children of immi￾grants, prefer less-restrictionist immigration policy. Ideol￾ogy is significantly positive: more-conservative individuals prefer more-restrictionist immigration policy. Our nonskill estimates are similar to those in Citrin et al. (1997) and Espenshade and Hempstead (1996).14 52.2% of the sample was female, 11.5% was black, and the average age was 43.3. 14 Appendix A table Al reports results for the table 2 specifications estimated on the listwise-deletion data sets for each year. The qualitative results are similar to those discussed in the paper using multiple imputa￾tion. However, using conventional rules for inference, the statistical This content downloaded from 202.120.14.193 on Mon, 15 Feb 2016 10:04:26 UTC All use subject to JSTOR Terms and Conditions