Are the views of anti-immigrant people shaped by reality, or are they merely opposed to something they don’t know anything about? I attempt to answer this question in this post on an international level. I look at how attitudes towards immigration vary by country.
If it is the countries with few immigrants that are against immigration, then we could reasonably conclude that fear of immigration is driven by speculation and ignorance. As people in low immigration countries have very little experience with immigration. If, however, it is countries with many immigrants that are more against immigration, then we could conclude that the fear of immigration is more evidence- or experience-based.
Data. To measure attitudes towards immigration I turn to the World Values Survey, waves 5 and 6. I’m looking at answers to three questions:
- Would you mind having immigrants/foreign workers as your neighbors?
- When jobs are scarce, should employers give priority to natives over immigrants?
- Regarding people coming here to work, should the government (1) let anyone come, (2) let people come as long as there are jobs, (3) place strict limits on the number of foreigners allowed in, or (4) prohibit people from coming?
The first two questions are available for a larger set of countries (approx. 80), the last one only for around 50. I interpret the first question as more of a measure of the cultural consequences of immigration (i.e. attitudes towards resulting multiculturalism), the second one as a measure of the economic consequences of immigration (e.g. jobs), the third one as a general measure of attitudes towards immigration (both cultural and economic).
Furthermore, I also create several (four) composite measures of attitudes towards immigration using the three questions above. These are basically just used to summarize all aspects of immigration into one concise score.
To measure the number of immigrants, I use the World Bank’s data on international migrants as a % of the population for the latest year (2010).
The source of anti-immigration views. If we look at the three questions on attitudes towards immigration, then what we can find is that people tend to base their general opinion of immigration (Q3) much more on fears of cultural influences/nuisances (Q1) than on fears of the potential economic consequences (Q2).
In particular a country’s general view on immigration is much more correlated with cultural (.61) than with economic (.30) “fears”. Thus fears of the cultural consequences of immigration seem more important in shaping one’s opinion on immigration than fears of potentially bad economic consequences.
This relationship, however, is interestingly reversed if we only look at OECD countries. Suggesting that these richer countries are more preoccupied with the economic effects of immigration than with its cultural effects.
These facts are illustrated below. We can see that the relationship between the perceived cultural consequences of immigration and general views on immigration (top left) is stronger for the worldwide sample than the relationship between the perceived economic consequences and general views (top right). In other words, the fitted line is steeper in the top left panel. We can also see in the lower panels that this relationship reverses for the OECD sample.
Ignorance or facts? Let us now see the most interesting part of this article: whether countries with lots of immigrants hold worse views of immigrants. Below is a world map that shows countries’ answers to Q1 above, whether they’d like an immigrant as their neighbor.
The more blue you see, the larger fraction of people in those countries would have no problems with immigrant neighbors. Reds are the opposite. Assuming for simplicity that it is the rich world that has lots of immigrants (North America, Europe, Australia), we can see that these regions are quite blue. But then again, so are many less developed regions with few immigrants (e.g. Latin America). This suggests that there is little connection between number of immigrants and attitudes towards immigration from a cultural point of view.
Let us now look at a similar map for Q2, whether employers should choose natives over immigrants in times of job scarcity.
Here the world is much redder in general, but the rich world (North America, Europe) seems on average lighter red or more blue than the rest of the world. This suggests that countries with more immigrants are less likely to oppose immigration on economic grounds. In other words, this indicates that fears of immigration for economic reasons are not grounded in reality.
So do the numbers bear out these predictions? Let us first confirm the seemingly obvious assumption made above: that it is indeed the rich world that has a high stock of immigrants. The map below shows that Europe, North America, Australia and New Zealand are generally bluer than the rest of the world. (Side note: I excluded a small number of countries with more than 25% migrants for better aesthetics, this does not alter the main message.)
Now, the numbers. In the whole sample, countries with more migrants tend to be more wary of the cultural consequences of immigration, i.e. they don’t want immigrant neighbors. This is interesting, and not really supportive of immigration. The correlation is not very strong though, only -.13.
If we just exclude countries whose population is more than 25% migrants, then this result disappears. And we have a very weak positive correlation (.09) indicating that countries with more migrants generally tend to have more favorable views on immigration.
All this says to me that very massive immigration can potentially have negative consequences on a country’s culture, which is not all that liked by locals. Note though that this very massive immigration is only happening in very small states. In particular, countries with more than 25% migrants included Andorra, Bahrain, Hong Kong, Jordan, Kuwait, Palestine, Qatar and Singapore.
Finally, if we look at the OECD only (which unfortunately decreases our sample size to 20), the relationship becomes very strong and positive (correlation .58). This means that within the OECD, it is people in countries with few immigrants that don’t want immigrant neighbors. This indicates that this is mostly due to ignorance or a lack of experience with immigrants. Countries with more immigrants generally have no problem with immigrant neighbors.
For the economic consequences of migration, we have a positive correlation in all subsamples. For the whole world, having a positive view towards immigrant workers and migrant stock are positively correlated with a coefficient of .16, for <25% migrant stock countries the coefficient is .43 (!), for OECD only it is .48.
This, again, indicates that massive immigration (as evidenced in the small states mentioned above) deteriorates one’s views on immigration a lot. But in general, fears of immigration on economic grounds are much more widespread in countries that have actually very few immigrants. So this concern with the economic consequences seems to be based once again on irrational fears.
Putting this altogether, it seems that there could be a nonlinear, hump-shaped relationship between the number of immigrants and views on immigration. There is only around 8 countries with more than 25% migrants, so data points are very scarce above 25%. This makes it hard to verify this hump-shape claim statistically. Still, the hump-shape does arise and is statistically significant in several cases. Two of these are illustrated below: the relationship between views on immigration’s economic consequences and migrant stock for the whole sample (left), and the relationship between one of my composite measures of immigration attitudes (higher = more favorable view) and migration stock for the OECD (right).
Let us conclude by looking at how the composite measures of attitudes towards immigration correlate with migrant stock. These measures go from about zero (.02) to moderate (.28) correlation in the whole sample, moderate to strong correlation (.31 to .44) in the <25% migrant stock sample, and quite strong correlation (.53 to .60) in the OECD sample.
So if we aggregate across all measures of attitudes to construct one composite measure of attitudes towards immigration, we see the main conclusion of the article repeated: that countries with more migrants have more favorable views on immigration. The relationship is stronger if we exclude countries with a massive stock of immigrants, and even stronger within the OECD.
Robustness. Very interestingly, the results do not change qualitatively at all, and only change quantitatively very sightly if we exclude all second-generation immigrants from the sample. These are people who have at least one immigrant parent.
The results change even less if we exclude all people who don’t consider themselves to be part of the local nationality. I.e. when asked about how proud they are to be French (if the survey was conducted in France), British (if in Britain), etc., these people said that they were not French/British/etc. This should cover most first-generation immigrants, and perhaps some disgruntled second-generation immigrants.
These robustness checks indicate that the results are not driven by the fact that high immigration countries have lots of immigrants in the sample who are more likely to be immigrant-friendly. In fact, it appears that the views of immigrants don’t really differ from that of the general population.
Conclusions and caveats. First, what must be noted is that this is a correlational analysis. I am making no claims of causality here. In other words, it could be that countries with higher migrant stocks have more immigrants because they were already more welcoming towards immigrants (for cultural or whatever reasons). And this can explain the correlations.
It could also be that all countries started out as being relatively unfriendly towards immigrants, but as immigrants went to certain receiving, rich countries, those countries realized that immigration ain’t so bad after all. While the rest of the world still maintains its irrational fear of immigration.
So what can we conclude? First and foremost, this study suggests that fear of immigration is more the product of ignorance than of experience, it’s a sort of fear of the unknown. Secondly, this seems to hold for both the cultural and the economic consequences of immigration. In other words, people/countries not exposed to immigration tend to overestimate the potential damage (and underestimate the potential benefits) that immigration could do to their culture and economy.
Thirdly – and this is what we established at the beginning of the post – people seem to care more about the cultural than the economic consequences of immigration. This is interesting because while economists have established over and over again that immigration is good for the economy (although low-skill workers might suffer, they can – at least in theory – be compensated), there is – in my experience – much less research on the cultural consequences, and whatever research we have does not in general come to a conclusion that can be construed as an unequivocal support for immigration.
Regarding culture though, another concern with this analysis comes to mind. If culture is very malleable, then countries with many immigrants may not be so preoccupied with immigration simply because immigration has changed their culture already. In essence, it is possible that – due to cultural plasticity – humans don’t really care so much about whether their country’s culture will change in the long run, because they’ll change with it as well. And if this is the case, then the results could simply be driven by this change in culture in high immigration countries, and not necessarily by the fact that immigration brought so many benefits to those countries.
Therefore, my aim with this analysis is not to take either side in the immigration debate. It is more just to expose a set of stylized facts. Facts that would have to be addressed by anyone who is arguing either for or against immigration. So I guess I’ll let you make your own conclusions as to what this analysis says about the immigration debate.
Code and data. The R code used to make this analysis can be downloaded here. This file also contains the data used in the analysis. Everything I did in this post is 100% reproducible.