050311 miguet slides en

Voting about Immigration Policy :What does the Swiss Experience tell us ?: 

Voting about Immigration Policy : What does the Swiss Experience tell us ? Jaime de Melo, Florence Miguet and Tobias Müller University of Geneva March 2005

Overview of the immigration debate: 

Overview of the immigration debate Immigration policy occupies center-stage in the policy debate in many developed countries Before EU enlargement : (May 2004) Economic predictions : moderate migrations flows and limited impact on labor market But attitudes surveys give a different picture : a majority of EU citizens thinks that it will increase immigration (65%, Eurobarometer September 2002) After EU enlargement : 14 out of 15 old Member States imposed employment and/or welfare access restrictions on new Member States’ citizens (until May 2006) Possible transition until May 2011

Overview of the immigration debate (cont’d): 

Overview of the immigration debate (cont’d) EU “ Green Paper ” : (January 2005) Stimulate a debate on how to regulate the entry and residence of third-country nationals who are seeking employment in the EU Before NAFTA expectations : (January 1994) Reduce economically motivated Mexico-US immigration Post NAFTA outcomes : Increased legal and especially illegal Mexico-US immigration Legalization of workers mainly through guest-worker programs Unlimited number of skilled Canadians, Mexicans and Americans (BA or more) are allowed to cross borders (January 2004) In Switzerland, regular voting about immigration tells us that it is quite a “ sensitive ” subject to manage

Why be against immigration ?: 

Why be against immigration ? Nationals’ attitudes towards immigration are shaped by : (ESS 2002) cultural preferences / social capital EU15 : 5.71 (2.47) / CH : 6.27 (2.24) (up 10 = enrich cultural life) EU15 : 3.07 (2.05) / CH : 2.86 (1.80) (up 10 = less crime problems) welfare take-up EU15 : 4.13 (2.23) / CH : 4.11 (1.95) (up 10 = put in more) labor market effects EU15 : 2.93 (1.13) / CH : 2.87 (1.00) (up 5 = wages are not brought down)

Central issues considered: 

Central issues considered What are the determinants of attitudes toward immigration when voting ? Are these attitudes influenced by “ economic ” considerations ? Do labor market characteristics play an important role ? Do political economy predictions help explain citizens’ attitudes ? How does vote data improve upon opinion poll data ? In what sense may opinion polls outcomes differ from observed voting choices ? How has voting upon immigration shaped the Swiss immigration policy ?

Standard economic predictions: 

Standard economic predictions How do factors’ ownership play a role in the determination of migration policies ? Factor-Proportions Analysis model : “ unskilled workers nationwide should prefer policies to lower (raise) immigration inflows of low (high) skilled workers ” Heckscher-Ohlin model : (used mostly by international trade economists) “ if a country is small and immigration inflows are small, this country will absorb immigrants by changing its output mix and not its factors’ reward ”

Political economy predictions: 

Political economy predictions Median voter framework Benhabib (1996) : “ the immigration option will be opposed in a direct democracy if the median voter’s capital (physical and human) is lower than the country’s average capital level before immigration ” Grether et al. (2000) : (Ricardo-Viner model) “ if capital evenly distributed within groups of skilled and unskilled citizens, skilled natives always adopt a position which is systematically opposed by unskilled natives ”

Preliminary results from the ESS 2000: 

Preliminary results from the ESS 2000 “ Average wages and salaries are generally brought down by immigrants ? ” 1 strongly agree – 5 strongly disagree “ Immigration is generally good / bad for the economy ” 0 very bad – 10 very good

What do empirical studies tell us ?: 

What do empirical studies tell us ?

What do empirical studies tell us ? (cont’d): 

What do empirical studies tell us ? (cont’d) Increasing interest in the determinants of attitudes towards immigration since mid-90s : Increased legal and illegal migration pressure Increased diversity of immigration’s source countries Decrease in immigrants’ skill level Increased availability of survey data Virtually all surveys rely on opinion poll data (ISSP, WVS, ESS, Eurobarometer, NES, etc) “ Do you think that the level of immigrants / foreigners in your country should be : decreased, kept the same or increased ? ”

What do empirical studies tell us ? (cont’d): 

What do empirical studies tell us ? (cont’d) Generally, the skill level (years of education or wages) has a significant positive impact on the probability to favor immigration Age and living in a rural area have a rather negative impact / having foreign born parents or being foreign born has a rather positive impact / being unemployed or living in a area with a high percent of foreigners has an ambiguous impact “ Non-economic ” factors (measures of racial or cultural prejudice / of national pride) and political orientation also matter a lot in shaping attitudes towards immigration Bauer et al. (2000) country’s immigration policy has an important effect on attitudes towards immigration

General comments on opinion poll data : 

General comments on opinion poll data Advantages : Big sample size Many countries included Follow-up in time Shortcomings : “ Hypothetical bias ” “ Selection bias ” (participation choice)

Voting in Switzerland: 

Voting in Switzerland Broad voting rights (direct democracy) Mandatory / Facultative referendum (11 on immigration since 1960) Popular initiative (9 on immigration since 1960) Regular voting about immigration policy Foreigners have no voting rights Feld et al. (2000), “ political outcomes are closer to citizens’ preferences in direct democracy compared to representative democracy ” Swiss more “ satisfied with the way democracy works ” than other European countries : mean 6.57 (2.00) vs. 5.47 (2.40) out of 10 (ESS 2000)

Initiatives on immigration since 1960: 

Initiatives on immigration since 1960

Immigrants’ share in Switzerland (2002): 

Immigrants’ share in Switzerland (2002) One of the highest immigration share in Europe during the whole post-war period Total foreigners’ share Total foreigners’ share from EU countries

The shaping of the Swiss immigration policy: 

The shaping of the Swiss immigration policy Recession 70s’ Recession 90s’ Bilateral agreements CH - EU Popular initiative (<18% foreigners) 3 circle policy Introduction of quota system 1st popular initiative (<10% foreigners) Bilateral agreement CH - Italy

Main facts about Swiss immigration policy : 

Main facts about Swiss immigration policy Minorities’ opinions against more immigration have been taken into account by government Smooth transition from guest worker status to permanent resident status Bilateral agreements I with EU15 : (voted and approved May 2000) Free mobility of labor between EU15 and Switzerland with measures of accompaniment : June 2002 End of measures of accompaniment : June 2007 Possible safeguard clauses until 2014 Bilateral agreements II with EU25 : (vote September 2005) Enlarge free mobility of labor to the 10 New EU Members States Reinforce measures of accompaniment

Post-vote data versus opinion poll data : 

Post-vote data versus opinion poll data No “ hypothetical bias ” : outcome of vote is implemented Mayer (1984), “ political outcome can differ widely from underlying attitudes ” Able to handle the “ selection bias ” : not everyone vote No “ voting paradox ” in Switzerland : the magnitude of the opposition to a vote is taken into account

Who votes ? : 

Who votes ? In a direct democracy voting is not compulsory Participation is not random Participation choice decided by an individual cost-benefit analysis Abstention is a choice often connected to the vote

Post-vote data used : 

Post-vote data used VOX post-vote survey conduced by the Gfs institute (since 1983) Random / quota methodology Sample size : about 1’000 individuals each time Include : participation choice, vote choice, socio-economic characteristics, political orientation, etc

Two initiatives against immigration: 

Two initiatives against immigration Initiative “ for the regulation of immigration ” September 24 2000 : “ foreigners’ share should not exceed 18 % of the total Swiss population ” Rejected : 63.8 % and participation : 45 % Initiative “ for the limitation of immigration ” December 4 1988 : “ limit the total annual immigration flows to 2/3 of the total annual emigration flows + limit the number of seasonal workers and cross-border commuters ” Rejected : 67.3 % and participation : 52 %

Econometric framework: 

Econometric framework Bivariate probit model with censoring and sample selection 3 possible outcomes : do not participate : p = 0 ; vote not observed participate : p = 1 ; vote “ yes ” : v = 1 participate : p = 1 ; vote “ no ” : v = 0 Simultaneous choice of p and v

Econometric framework (cont’d): 

Econometric framework (cont’d) Set-up p* and v* are latent variables : Possible correlation between errors : If ρ = 0, then both equations could be estimated independently Estimation by log-likelihood

Variables description: 

Variables description

Variables description (cont’d): 

Variables description (cont’d) Is skill or human capital a significant determinant of attitudes towards immigration ? For the 2000 initiative : years of schooling “ potential earnings ” (Mincer) constructed : For the 1988 initiative : 3 level of education only (dummy)

Survey related “ bias ”: 

Survey related “ bias ” Voting anonymously raises the percentage of “ anti-immigrants ” : % rejection in real vote 64 % in 2000 and 67 % in 1988 % rejection in VOX survey 72 % in 2000 and 77 % in 1988 Survey participants are also those who vote more : % participation in real vote 45 % in 2000 and 58 % in 1988 % rejection in VOX survey 56 % in 2000 and 73 % in 1988

Initiative 2000 : vote analysis: 

Initiative 2000 : vote analysis

Initiative 2000 : vote analysis (cont’d): 

Initiative 2000 : vote analysis (cont’d) Negative effect = rejection of anti-immigrant initiative Higher education : negative effect Right-wing political orientation : negative effect Age : U-shaped with minimum around 56 : age-earning profile ? Unemployment share : rather positive effect Foreigners’ share : no significant effect Italian speaking : negative effect

Initiative 2000 : vote analysis (cont’d): 

Initiative 2000 : vote analysis (cont’d)

Initiative 2000 : vote analysis (cont’d): 

Initiative 2000 : vote analysis (cont’d) Higher education + higher wage : negative effect Retired : negative effect Right-wing political orientation : negative effect Age : U-shaped with minimum around 56 : age-earning profile ? But insignificant if combined with wage Unemployment share : rather positive effect Foreigners’ share : no significant effect Italian speaking : negative effect

Initiative 2000 : participation analysis: 

Initiative 2000 : participation analysis

Initiative 2000 : participation analysis (cont’d): 

Initiative 2000 : participation analysis (cont’d) Higher education : positive effect Right-wing political orientation : U-shaped with minimum around 4.6 (close to mean 4.9) Age : U-shaped with minimum around 75 Unemployment share : negative effect Foreigners’ share : positive effect French and Italian speaking : negative effect Owner : positive effect

Initiative 2000 : selection bias: 

Initiative 2000 : selection bias

Initiative 2000 : selection bias: 

Initiative 2000 : selection bias Selection bias exists :   0 highly significant  < 0 implies that an individual who has a high probability to participate has also a high probability to reject the initiative due to his unobserved characteristics

Initiative 2000 : selection bias: 

Initiative 2000 : selection bias Conditional probability, P(v=1| p=1), represents the model fit to the outcome observed : 28.3 % Marginal probability on the whole sample, P(v=1), indicates how people voted or would have voted had they participated or not (i.e. opinion poll) : 54.6 % Marginal probability on the sub-sample of voters, P(v=1), indicates what the outcome of the vote would have been if we had not considered the correlation between the vote choice and the participation choice : 51.8 % If everyone had voted, the initiative would have been accepted !!

Initiative 1988: 

Initiative 1988

Initiative 1988: 

Initiative 1988 Vote choice : only having a high education has a significant negative impact / political orientation has no effect Participation choice : close to 2000 No sample selection bias :  = 0 Marginal and conditional probability give the same outcome

Comparison of the two initiatives : 

Comparison of the two initiatives For both initiatives, probability to accept the initiative is significantly negative as education increases, there is an opposition between low and high-skilled citizen : Is that comforting for the political economy predictions ? Selection bias very strong for the 2000 vote but not for the 1988 vote : Why ? Differences in participation rates ? High explanatory power of the political variable (in vote and participation choice) for the 2000 initiative and no power for the 1988 initiative : Increased political polarization of the electorate ? Stronger extreme right ?

Main contributions : 

Main contributions Political economy prediction : “ skilled natives adopt a position which is systematically opposed to that of unskilled natives ” is verified “ Selection bias ” may induce big differences in outcomes of opinion polls and of real vote (opinion polls may be overly pessimistic) Specific to Switzerland : higher political polarization observed at the end of the 90s is noticeable in the data

Policy implications : 

Policy implications Insofar, as anti-immigrant hostility is motivated by economic fears (labor market), governments could “ compensate” those who loose as a result of immigration Transition period from temporary to permanent resident status for immigrants helps accommodate negative natives sentiments High skilled immigration easier to accept without restrictions Regular votes on immigration have a strong impact on policies implemented Opinion polls mostly overly pessimistic (due to bias)

Slide41: 

Tank you

Political economy of immigration (cont’d) : 

Political economy of immigration (cont’d) Grether et al. (2000) : “ attitudes toward immigration are evaluated by capital ownership of unskilled labor with respect to capital intensity of the sector that uses them intensively ”

Some statistics for Switzerland: 

Some statistics for Switzerland Immigrants’ country of origin : Education’s levels :

Initiative 2000 : selection bias: 

Initiative 2000 : selection bias

Initiative 2000 : selection bias: 

Initiative 2000 : selection bias Simple bivariate probit model on vote yields biased results Highly significant and large negative selection bias

Referenda on immigration since 1960: 

Referenda on immigration since 1960

Outline: 

Outline Overview of the immigration debate Political economy of immigration policy Determinants of attitudes toward immigration : a literature review Shaping of Switzerland’s immigration policy Voting on immigration policy Data and methodology Results Conclusions and policy implications