|Date||July 10, 2015|
|Speakers||Giovanni FACCHINI (Professor, University of Nottingham) / Yotam MARGALIT (Associate Professor, Columbia University / Tel Aviv University)|
|Moderator||NAKATA Hiroyuki (Senior Fellow (Adjunct), RIETI / Senior Lecturer, School of Management, University of Leicester)|
This paper is intended to improve understanding of the U.S. immigration policy debate and how the preferences of the electorate have been translated into actual policymaking in Washington.
This analysis attempts to link the amount of media coverage that a politician receives to his/her behavior with regard to immigration and trade policy. This is important to understanding globalization patterns.
Immigration and trade differences
Standard economic theory suggests that the labor market effects of trade and immigration should be similar. The Heckscher-Ohlin model, which is frequently used in international trade, treats immigration and trade as substitutes with regard to the effect on the local labor market. This has been argued theoretically, and empirical work has shown that these two factors affect the behavior of politicians in the same ways. Still, we know that there are important differences in the drivers of the liberalization of trade and immigration policies, respectively.
Many studies have shown that immigration is much more salient politically than trade from an electoral point of view. For example, immigration was one of the most important issues in the 2010 elections in the United Kingdom, where often politicians won or lost on either a pro-immigration or anti-immigration stance. It is much more difficult to argue that trade plays such an important role in shaping elections at present.
Effect of media exposure on immigration and trade positions
The goal of this paper is to see whether media exposure, as a disciplining device, has a differential effect on immigration or trade.
The first issue is how to measure media coverage, and the main challenge is to find a way to describe how much attention a politician receives in the media in a way that is exogenous with respect to the variable of interest.
One approach from the literature is the idea that media markets and congressional districts in the United States may not be perfectly overlapped. Snyder and Stromberg use a measure called "congruence" between congressional districts and media markets to capture how much attention a politician elected in a specific congressional district receives in the local media. If a high degree of congruence exists between a media market and a congressional district, the politician from that district will receive more attention from the local media and his/her actions in Washington will be subject to greater scrutiny therein. More information on the politician will be available to the local electorate, making the politician more accountable to the electorate than in districts where congruence is low.
We chose the United States because very detailed information exists on how a politician has voted on specific policy measures, enabling us to match the voting behavior of an elected representative in the U.S. Congress to the preferences of the electorate in the politician's district. We used a biennial survey which includes questions on immigration and trade to construct a pro-immigration or pro-trade dummy variable and compute an average for the opinions expressed. We then applied the congruence measure for media coverage and additional controls.
To understand the main logic behind our analysis, we focus on two examples and investigate the link between individual opinions, media congruence, and voting behavior. The first concerns the state of Florida and a bill aimed at liberalizing skilled immigration and the state of Texas and a bill to extend fast-track trade negotiation authority for the U.S. president. Both bills were voted upon in 1998.
On the immigration bill, we focus on two districts in Florida which have similar opinions on immigration. In both districts, the average citizen is pro-immigration, and about 60% support increasing the number of skilled workers who come to the United States. On the other hand, the level of congruence between the districts and their media markets differs: one has high congruence and the other low. Elected officials from the first district supported the immigration bill, but the second district's representative did not. As the media scrutiny increases, a politician becomes more answerable to his/her district, and vice versa.
On the trade bill, we consider two districts in Texas which have a median voter in favor of liberalizing trade. The extent of congruence is again different; one district is highly congruent and the other is not. A politician from the first district voted against trade liberalization while the politician in the second district voted for it. The same accountability pattern therefore is not apparent.
To determine whether the patterns uncovered in our example hold more generally, we ran a series of linear probability models with several additional controls. Our key variables of interest were the opinions of the district on immigration and trade, the extent of congruence between the congressional district and the media market, and an interaction term between the two, which tell us whether a more congruent district is represented by a politician whose voting patterns reflect the district's favorable views of immigration or trade liberalization.
Our initial specification, in which we account only for average opinion at the district level and time varying state fixed effects, indicates that on immigration, we see a positive and significant link between opinions at the district level and politicians' behavior, whereas we see no effect with regard to trade.
When we allow the effect of opinion to differ across districts depending on the extent of congruence, we find that the more pro-immigration the district and the greater the media congruence are, the more likely it is that the representative will support immigration liberalization. With regard to trade, we don't find any significant effect.
What is behind these results? It is most likely due to an effect related to the electoral mechanism. We tried to distinguish districts in which a politician was elected with a large margin of victory, meaning that the seat is relatively safe, from districts which elected politicians with small margins who have shakier seats. Splitting the earlier results across such low- and high-margin districts shows that congruence is completely driven by districts characterized by a low-margin of victory. Politicians are more responsive to the information available to the electorate in tough elections than in easy elections. Performing the same split on trade results, we don't find any effect. Our interpretation is that trade is not an important issue in deciding elections.
Another mechanism to account for the electoral competition is by taking into account voter turnout. Very competitive elections tend to have much higher voter participation. Performing a similar split for voter turnout, the results point in the same direction. We don't find anything like this on trade, but we do find an effect on immigration in districts with above-average voter turnout, confirming the electoral mechanism.
One important caveat in interpreting our findings is that they might be affected by reverse causality. The very fact that the media covers a politician's actions can make the politician more accountable, but high congruence can also enable the politician in power to influence the local media and thereby his/her constituents. To address this concern, we implement an instrumental variable strategy, in which we explain individual opinions using exogenous demographic statistics. The results of these regressions confirm the robustness of our findings.
The second concern was the years included in the sample. Hence, we restricted analysis to years in which both immigration and trade issues were on the floor. The third concern was the role of public opinion. One shortcoming of the survey is that it is not very large, covering about 2,000 individuals in each cross-section. We carried out robustness checks, averaged out opinions over time, played with controls, etc., and eliminated districts with very few observations. Again, the results were basically the same.
Why does the media influence a politician's accountability on immigration but not on trade? The survey we have used suggests one possible channel. Between 0%-3% had no opinions on immigration, but between 25%-47% had no opinions on trade, so the electorate has much less clearly defined opinions on trade. The media can make a politician accountable but only when it is an important enough issue for opinions on it to drive an election.
My presentation will be of something much more preliminary: initial results from two large data collection efforts. A lot of the big-picture questions about immigration are still very much unresolved. We have a sense that the immigration debate has become more salient in some way. But how has it evolved? Do we see patterns and responses that are similar across left and right? Can we try to think of specific explanations for the evolution of the political debate that would actually stem from the specific characteristics of immigration? One of the major reasons we think that these questions have not been addressed is that they are very difficult to address empirically.
My coauthor Rafaela Dancygier and I have undertaken two major data collection efforts, partly in the hope that these data sets will become an open resource enabling the many researchers interested in the politics of immigration to use for their own work. By opening it up for broad use, we also hope to expand and improve the data set over time.
Data collection on immigration debate and immigrant characteristics
We first wanted to know what the political debate is centered on when it talks about immigration, so the first thing we did was try to collect information on what parties are actually saying about immigration when they are appealing to voters. Existing data were unsatisfactory for answering these questions.
We wrote a code book to convert language used by politicians, in the existing data sourced from manifestos, referring to immigration in 30 categories over a long, integrative process. The idea being that the coding of these translations into 30 categories in a systematic fashion will allow us to track changes and patterns in the immigration debate over time.
The second effort was to collect and consolidate all information available on the objective characteristics of immigrants coming into the different countries: their education level, skills, the reasons for their migration (categorized by work-related, family-reunification-related), and many more. We then tried to create harmonized measures for all of the relevant countries over time.
I will present the initial results from the collection efforts and some links between the two data sets.
The basic description of the data set is as follows: we focused initially on 12 western European countries. Japan is a country for which data exist to which we hope to add in the next round. There were over 300 manifestos since the 1960s; around 11,400 immigration-related sentences; and demographic data for almost 30 years until 2012. We focused on the major left, major right, and far-right parties.
Preliminary findings and insights
The evolution of the immigration issue in terms of salience--over time it involves a comparison between the three types of parties. The share of the overall manifestos is used so that we can see how much of it is being dedicated to the immigration issue over time. As the graph nicely indicates, immigration has clearly become a much bigger issue and consistently and systematically so.
Despite increased discussion of immigration, can anything be said about the way the issue is addressed? Parties, over time, started talking about very different aspects of immigration. In the 1960s and 1970s, a manifesto would address from two to four immigration-related issues--the impact on wages, housing, or the education system, for example. Over time, the discussion became much more complex. We can now see the issue in much of the public discussion, from its economic and cultural dimensions to many different angles. A consistent trend is that it is becoming a more multi-faceted discussion.
Given that politicians are talking about immigration in a more complex way that takes into account the many different dimensions of the phenomenon, is the discussion more positive or negative? We could see two things: a significant decline in the share of positive claims, resulting in a more negative discussion over time, and an increase in the share of negative, but this is when we include all parties, which is slightly misleading because of the disproportionate share of right-wing parties. Looking specifically at center-left and center-right parties, the pattern is slightly more complicated. In the 1980s, it was very positive overall for the left. But for the right, it was much more negative even from the 1970s. But we also see that, over time, the left (center-left) is looking increasingly more like the right. They look like how the right was in the 1970s. The big shift ultimately in becoming more negative is essentially in the center-left. The center- and far-right always were more skeptical.
What are the immigration issues that are most often highlighted? As the table shows, no matter the party, issues related to restrictions on immigration were always very salient: 22% of discussion on the left. Refugees and asylum were also discussed a fair amount by both left and right. Interestingly, the issue of jobs for natives is clearly not the central issue. The impact of immigration on wages for natives is actually very low. Sometimes the aspects we are more interested in don't necessarily translate to what the politicians are talking about and what the public is thinking about on a certain issue.
Clustering the different issues over time, the patterns are quite strong and interesting. The economic issue/dimension of immigration that was the most overwhelmingly central in the 1960s and 1970s has decreased somewhat over time. It has increased from the 1990s but is nowhere near where it was in the 1960s. What we see in tandem is a cultural turn in the immigration debate over time. Discussion of the cultural implications has grown consistently over time. We also see the issue of integrating immigrants already in the country becoming much more central. More research needs to be done on the factors that affect the degree of integration and how that bleeds into the political discussion. Lastly, we also see that asylum and refugees, which were non-existent issues in the 1960s and 1970s, are much more prominent in the ways politicians talk about immigration.
How has the debate changed from an economic perspective before and after 1990, left vs. right (excluding far-right)? The most important issue is the economics of immigration. On the left, the welfare state has become more of an issue but less so for the right. The issue of jobs is present but is not the central issue. Among the far-right parties, the issue of jobs used to be the central issue for attacking immigration, but much less so today. The welfare state has become more central.
We also need to study whether the types of immigrants who are coming in affect the way the parties discuss immigration. How much do they talk about the welfare state, jobs, and wages? Center-right parties don't differ from the center-left.
We really see some very clear trends over time in the way the immigration debate has evolved. Salience has increased clearly, the cultural dimension has become more prominent than the economic dimension over time, negativity has increased, and immigrant characteristics can influence party appeals, although further work is required to determine which specific characteristics are influential and how influential they are. We see the left and right becoming more similar in terms of the immigration-related topics they discuss as well as the negative sentiments expressed.
Q1. Professor Facchini, how should we interpret the policy implications? Also, people with stronger opinions vote, so I think there may be a sample selection bias. Finally, you used a fixed-effect model, but I am not sure whether this is effective for politicians.
In terms of the estimation framework, we used fixed effects, but we controlled for political party affiliation and many other characteristics of the politicians that we studied.
With regard to strategic participation (i.e., who votes), we have a specification to control the share of people showing up at the polls in a given district. That's the best we can do.
In terms of the policy implications, the basic question we asked in the paper is whether the media can make politicians accountable on different facets of globalization. Only if the policy issue is sufficiently important can you make a politician accountable, which is a policy issue itself.
Q2. Would it be right to predict that in the West, by combining your research, popular opinion on immigration is becoming more negative and politicians are expected to respond more negatively? To what extent will negative immigration policies prevail?
I think you are right. I think the debate is becoming more negative and constraining politicians. From a policy perspective, this is the important implication of which to be aware. This is something very consistent with what Giovanni Facchini has shown. Going forward, the trend has been quite clear in recent years. I think the big challenge will be the fact that a growing need for immigration exists in many countries. Here in Japan, the issue is even more extreme. The demand is strong from an economic perspective, but political resistance is equally strong. How this will ultimately be resolved is an open question. I think we would have seen much more open borders without the political constraints.
Looking at history, immigrants have been coming into the United States for a long time. The borders were fairly open basically until 1917, when the literacy test provision was introduced. The tradition in the United States has been that immigrants are more pro-open-border. The Latino voters are being targeted both by the Democratic and Republican parties as a key swing block. U.S. President Barack Obama managed to win because he carried 80% of the Latino population. This type of political calculus should also be taken into account.
Q3. Professor Facchini, did you control for political contributions? Have you linked your paper with shirking by legislatures?
We accounted for political action committee (PAC) contributions. They don't play a significant role on voting migration, while they do affect voting on trade. PAC contributions can be linked to individual politicians, but we cannot know the purpose of the lobbying activity. In a different paper, we looked instead at contributions directly aimed at affecting migration policy, even though we could not use them for our analysis in this paper, as they did not allow us to identify the politician who received them. In that paper, we also found that the more a company was spending on immigration policy, the larger the number of visas the sector the company was in was receiving. We haven't really looked into shirking.
Q4. Professor Facchini, the media market has been changing along with the Internet and social media. Doesn't this make measurement increasingly difficult?
That is a very good point. We used older data when local newspapers had greater importance. In the 1980s and 1990s, the local newspapers played an important role in transferring information--this is what the literature argues. This is the whole story of Snyder and Stromberg's paper. We hope to be able to use other media sources in the future.
Q5. I know that all of you are now planning to conduct a similar survey on the immigration policy in Japan. Do you anticipate different results compared with U.S. or European studies?
In our study, we want to provide individuals with different amounts of information on what immigration can do for a country. Japan has a rapidly aging population. Immigration could be an answer to this. I would be very happy to discuss your views on whether immigration could be an answer to this big demographic change. One of the issues in many countries is that popular opinion opposes immigration, maybe because the population isn't educated enough on the pros and cons of immigration. Our idea in our ongoing project on Japan is to determine whether access to more information can affect the way people perceive foreigners.
Q6. I have two questions. Attitudes have changed over the years. One reason could be an increasing number of immigrants. For both professors, is there a threshold at which the attitude changed from positive to negative? One such threshold might be the immigrants' ratio among laborers. Does a big difference exist between Europe and the United States?
More work is being done now on the threshold, but two issues seem to be quite relevant: 1) differences in electoral systems, and 2) the key state for the last several decades has been Texas. Republicans can't win any election without Texas, which is now becoming more Hispanic, making some believe that they have to take a more pro-immigration stance in the future. They haven't reached that threshold yet, but we could soon see that threshold, where they become more pro-immigration or lose out on Texas and become less politically competitive.
Many surveys have been conducted across countries that amassed information on immigration attitudes. You see quite a bit of heterogeneity across countries. One thing worth remembering is that in the European Union (EU), nothing really seems to have changed in people's opinions on immigration over a long period of time. That is a bit puzzling. With the many economic changes over the years, we would have expected much more dramatic shifts.
Q7. Professor Facchini, thinking about another implication of your results, politicians need to have ways to learn public opinion.
Politicians have access to election studies which tell them what is the average opinion of their electorate. We are claiming that the electorate learns what the politician does through the media, which affects the way how the electorate thinks about the politician.
Q8. Professor Facchini, I'd like to hear your views about the Greek issue. Does your study have any implications for the Greek issue?
There are two facets. On one hand, there is the moral hazard issue. People in other EU countries don't want to pay Greek pensions. But if you kick one country out, the question becomes which country comes next, and this would become a currency credibility issue. Overall, I think it would be less costly to keep Greece in.
Q9. For European countries, the Greek crisis has a big impact on the movement of people. But immigrants coming from Greece can also have a very positive impact. What do you think about the movement of Greek people to other EU countries?
It is a positive reaction. The flexibility of labor markets is definitely a plus. Of course, some countries like the United Kingdom are now experiencing massive inflows, with public opinion becoming a bit worried, but overall I think it has been positive.
The numbers are still small. It's true that, for Greece, it's a tragedy losing a whole generation of people, but from an EU economic perspective, these are fairly small numbers. On the previous point of whether it is more costly to keep Greece in or kick it out, politics is going to be driving a lot of what's going on. Politicians are unwilling to hear about debt relief. It is perhaps economically necessary but, at the moment, not politically viable.
*This summary was compiled by RIETI Editorial staff.