21 October 2016

Xenophobia Britannica? Anti-immigrant attitudes in the UK are among the strongest in Europe

First posted at LSE Brexit blog
The vote for Brexit took place. I had hoped that this would have been the end of the obsession with immigration. Brexit would allow the UK to ‘take back control’ of  its immigration policy, thereby nullifying the need for politicians to talk about it on an almost daily basis. However, in fact the reverse is happening. Having decided to leave the EU, the vote is increasingly interpreted as a call to end immigration almost entirely and, furthermore, it is discussed even more often and more negatively than before the referendum.

Instead of just imposing immigration controls if the government wishes to do so, we now hear a flurry of xenophobic policy proposals which are ironically unlikely to have any noticeable effect on immigration flows. There were suggestions that companies would be ‘named and shamed’ for hiring foreigners. Other ideas included preventing the government from seeking advice from non-UK citizens about Brexit related matters, presumably for ‘security’ reasons. Immigrants should not, we are told, “take jobs British people could do”. 

These proposals are economically illiterate as they implicitly accept the lump of labour fallacy that there is a fixed quantity of jobs in the economy. They are also clearly nationalist insofar as they posit work should be shared first and foremost among ‘deserving’ natives,  that we must identify and count foreigners, and that both companies and the government should not rely on foreigners to advise or work for them, at least in some instances. The question remains, however, whether these policy proposals and this obsession with immigration represent the preferences of the wider population? Or put differently, is it the case that the UK is noticeably more sceptical of immigration than other European countries?

To investigate this question, I turned to the survey data available from the European Social Survey; without a doubt one of the best cross-national academic surveys available. The 7th wave of the ESS was carried out in 2014 and covers 20 countries of western and eastern Europe (one is not an EU member: Switzerland). I analysed the responses to the available questions on immigration. For each question, I then ranked countries according to their responses as a rough indicator of where the UK sits in Europe. Obviously, a more complex analysis could be – and should be – carried out, but the following discussion gives us a first glance at the magnitude of the anti-immigration sentiment in the UK.

Ethno-nationalism, racism and multiculturalism

Concerns about immigration do not seem to be primarily driven by ethnonationalism, i.e. a conception of the nation being premised on a certain ethnicity and/or religion. Or put differently: these questions are not where the UK ranks worst.

Do people think “it is important for immigrants to be white”?

In the UK 1.4 percent of respondents thought it was ‘extremely important’ for someone to be white in deciding whether someone born, brought up and living outside the country should be able to come and live here. It is difficult to assess in and of itself whether this is high, but we can say that this is the 10th highest response (Hungary scores highest with 13.1 percent of respondents believing this is extremely important).

This only captures those respondents that selected 10 on a 10 points scale ranging from 0 ‘extremely unimportant’ to 10 ‘extremely important’. We can broaden the net by adding all those that selected 6 to 10: 7.7 percent in the UK did so, which again ranks them 10th, compared to more than 40 percent in Hungary and Lithuania (the two countries with the highest two percentages).

Do people “want to allow Muslims to come and live here”?

Overall, 17.3 percent of UK respondents answered ‘allow none’ to the question: how many Muslims should be allowed to come and live in one’s country? This is a very high percentage, but it still ranks the UK 13th with the top 2 countries (the Czech Republic and Hungary) having more than 55 percent of respondents choosing none. As much as 13 percent in the UK chose to ‘allow many to come and live here’ and 42 percent chose ‘allow some’ while 27 percent chose ‘allow a few’.

The importance of customs and traditions

Overall, 12.1 percent of respondents in the UK ‘strongly agree’ it is better for a country if almost everyone shares the customs and traditions, which ranks it 10th (the Czech Republic ranked first with 29.5 percent), while 30.7 percent ‘agree’, 25.5 percent ‘neither agree nor disagree’, 25.9 percent ‘disagree’, and 5.8 percent ‘strongly disagree’.

Close friends from a different race or ethnic group

The following question is clearly not about immigration but may tell us something about interaction between people: “Do you have any close friends who are of a different race or ethnic group from most [country] people? IF YES, is that several or a few?” In the UK, 21.7 percent responded ‘Yes, several’ which is the 3rd highest share after Sweden and Switzerland, and 36.9 percent said ‘Yes, a few’, but 41.5 percent said ‘none at all’.

Good contact with different race or ethnic group

The UK has ranked 13th worst in terms of the percentage of respondents who responded that ‘contact with a different race or ethnic group’ is ‘extremely bad’ (Hungary and the Czech Republic were the two highest). It does, however, rank high  in terms of the percentage that responded that it was ‘bad’ (as much as 6.3 percent).

Anti-immigration and negative perceptions of immigrants

Anti-immigration and negative perceptions of immigrants

As much as 17 percent of respondents in the UK want to ‘allow no’ immigration from poorer European countries (3rd after Hungary and Lithuania) and 31.9 percent say ‘allow a few’. Only 10.1 percent think we should allow many to come and live in the UK.

Unpacking immigration preferences towards different types of immigrants

We can unpack this further by analysing differences between high and low skill immigration. Responses are quite different when comparing whether to allow professionals from poorer countries and whether to allow unskilled labourers. Only 6 percent of respondents in the UK want to ‘allow no’ professionals from poorer European countries (11th highest) compared to 35.1 percent for labourers (3rd highest), and 19.9 percent want to ‘allow a few’ professionals compared to 28.7 percent for labourers.

Policy proposals that want to limit immigration into the NHS or highly skilled occupations may therefore not have as much popular support. Questions about the importance of good educational qualifications for immigration provide consistent answers: 75.5 percent believe the educational qualifications of immigrants is ‘important’ (which ranks it 5th highest in the sample, after Austria, Germany, Estonia and Lithuania – the answers are ranked from 0 ‘extremely unimportant’ to 10 ‘extremely important’ and I added the numbers who chose 6 to 10).

The importance of language and skills

Respondents in the UK attached a particularly high importance to language: 84 percent chose 6 to 10 in response to the question about the importance of speaking the country’s official language, where 0 is ‘extremely unimportant’ and 10 is ‘extremely important’. This placed the UK 2nd highest after Austria. As much as 38 percent responded that language was ‘extremely important’ by choosing 10.Equally, respondents emphasised the importance of skills: 83.3 percent chose 6 to 10 in response to the question of whether work skills needed in the country is an important aspect of immigration, where 0 is ‘extremely unimportant’ and 10 is ‘extremely important’. This ranked the UK 2nd highest among countries surveyed and 27.2 percent responded that it was ‘extremely important’ by choosing 10.

Concerns about immigration related to the ‘way of life’, jobs, public services and crime

Responses indicate that British citizens think it is important that immigrants should be committed to society’s ‘way of life’: 85.7 percent chose a number between 6 and 10, where 0 is ‘extremely unimportant’ and 10 is ‘extremely important’. This ranked the UK 6th highest among the countries considered in the ESS and 31.2 percent thought it was ‘extremely important’. 

But respondents were also concerned about whether immigrants take jobs away from the country. Respondents could choose any number between 0 ‘take jobs away’ and 10 ‘create new jobs’: 37.1 percent chose a number between 0 and 4 (those which on balance think immigrants take jobs away more than they create jobs), while 8.2 percent chose 0 (definitely think that they take jobs away). This placed the UK in the 6th highest position.

Moreover, many respondents were convinced that immigrants take out more than they put in in terms of taxes and public services – contrary to what the evidence suggests. Indeed, 42.8 percent chose a number between 0 and 4, where 0 is ‘generally put in less’ and 10 is ‘generally take out more’. This ranked the UK 6th worst in the countries under consideration in the ESS.  

The ESS also asked respondents: Compared to people like yourself who were born in [name of a country], how do you think the government treats those who have recently come to live here from other countries? 19.7 percent responded ‘much better’ which ranks the UK 1st highest (followed by Ireland and France) and 26.4% responded ‘a little better’.

Finally, the UK was 7th in terms of the percentage of people that believe immigrants make the country’s crime problems worse with 53.1% choosing a number between 0 and 4, where 0 is ‘crime problems made worse’ and 10 ‘crime problems made better’, and 8.7% chose 10, the most convinced expression of immigrants making crime worse.


Overall, there is a strong anti-immigration preference among a significant part of the population, which is rooted in associations between immigration and a number of problems including higher crime, way of life, and insufficient jobs and public services. These associations clearly helped the Brexit camp and play into the hands of the far right, which has long promoted these associations. Misperceptions seem to abound about the effects of immigration. 

But even the assessment of the number of foreigners in the country is widely off the mark for most people. Indeed, when asked, “out of 100 people how many were born outside the country?” 3 percent said 60, about 9 percent said 50, about 9 percent said 40, more than 10 percent said 30, and more than 10 percent said 20, just to give some examples. In reality, a recent House of Commons library briefing paper placed the percentage of people living in the UK who were born outside the UK at 13 percent (Hawkins, Oliver (2016) House of Commons, library, briefing paper, migration statistics, Number SN06077, 5 September 2016: page 18). More than 40% of respondents grossly over-estimate the number of immigrants in the country.
The survey data reveals that the UK has among the highest anti-immigrant survey responses, especially within Western Europe. These results suggest that recent policy proposals tap into widespread anti-immigration sentiment, which may, of course, have been created by the media and political discourse that has been framing the issue so negatively for a long time now.

At the same time, the problem seems to be not just about whether immigration should or should not be restricted, and if so how, but more importantly about how to promote a more facts-based discussion of immigration. The fact of the matter is that a large part of the electorate has now wholeheartedly embraced anti-immigration attitudes.

Note 1: Design weights are applied throughout.

Note 2: I use the exact language and wording of the ESS for simplicity.

05 October 2016

The crisis of legitimacy of the UK welfare state

The ongoing austerity agenda in Britain seems puzzling until we look at social attitudes and how they have evolved over time. This is precisely what the British Social Attitudes survey allows us to do by monitoring people's views of a variety of topics. It asks a random sample of 3000 people a series of questions since the early 1980s.

The analysis of their questions related to the welfare state reveals a striking crisis of legitimacy. The share of those that think unemployment benefits are too low fell from above 50% in early 1990s to less than 30% after 2010. This occured alongside drastic reductions in the unemployment benefit replacement rate (which measures the % of income that is replaced by unemployment benefits).

At the same time, the % of respondents that agree that the government should spend more on welfare has decreased from above 60% in 1990 to 30% in 2014. And this probably overestimates the % of people who vote that actually believe the government should spend more on welfare state.

This is surprising on two counts. First, assuming some degree of redistributing effects of welfare state policies, and given that the median income is under the average income, we would expect that it is beneficial for more than one third of people to favour redistribution, and by extension welfare spending.

Second, the crisis has - at least initially - increased unemployment and insecurity among workers, two issues that welfare state policies are supposed to address. One solution to this puzzle lies with the fact that people believe cutting the welfare state would incentivise unemployed people to find jobs. Indeed, in the early 1990s, less than 30% of respondents agreed that a less generous welfare state would encourage people to 'stand on their own two feet'. By 2010, this was above 50%.

At a time where fiscal pressures are seen - rightly or wrongly - to be significant, this low public support limits the electoral attractivness of policy proposals to expand the welfare state and strengthens the political viability of austerity.

04 September 2016

Regulations in 'rigid' labour markets are less likely to be enforced

How difficult is it to fire a worker in different countries?

To answer this question, most social scientists have created many indicators that capture the conditions, costs and uncertainty associated with firing an employee.

For instance the OECD Employment Protection Legislation (EPL) index measures "the procedures and costs involved in dismissing individuals or groups of workers and the procedures involved in hiring workers on fixed-term or temporary work agency contracts"

The determinants and consequences of EPL

This index is then used in statistical analysis to assess the impact of EPL on economic outcomes, such as unemployment, or to identify the determinants of EPL reforms, for instance partisanship.

The conventional wisdom is that EPL has adverse economic consequences (studies by IMF and OECD, Layard, Botero and others), but the stability of the findings to different specification has been contested (for instance see work by Baker, Avdagic and others).

Employment Protection Legislation: rules versus enforcement

For a long time a more obvious problem has been that legal restrictions on firing can only be expected to have any effects on labour market performance if the legislation is actually enforced on the ground.

The issue here is not so much that employment protection legislation might not enforced (which is likely) but more importantly that it might be enforced to varying degrees in different countries in ways that we cannot observed.

New database on enforcement

This shortcoming is now being partly addressed in recent research by Kanbur and Ronconi published in the Centre for Economic Policy Research.

In a shorter version of the paper they have published in VoX they explain how they created a new indicator of enforcement that combines both inspections and penalties.

Their results are interesting in at least two respects 

First, once they control for their measure of enforcement in a statistical analysis of the determinant of labour market performance, EPL no longer has a statistically significant adverse effect for most measures of performance. In other words, labour market regulations do not in fact seem to have a consistent effect on labour market outcomes.  

Second, they show that countries with more stringent EPL have lower enforcement levels, which then makes it difficult to know which country's labour market really is more 'rigid'. This is shown in the figure below that plots a de jure employment index on the horizontal axis and their enforcement index on the vertical axis.

Interpreting the figure: some country examples

The ranking should be interpreted as follows: countries which are lower on the scale have higher ranking and hence more protective institutions or more enforcement. 

Thus for instance, while Canada and Denmark have among the lowest ranking in de jure employment protection (they are not in the top 150 countries), they rank really high in terms of enforcement (in top 25). By contrast, France and Spain score high on EPL but do not rank well in terms of enforcement: their stringent regulations are not well-enforced compared to many other countries.

Figure: Enforcement and labour law


Brexit and non-UK born population

19 July 2016

Blog has passed 150,000 pageviews since its creation!

Being unemployed in the US after the Obama presidency

First posted on Tim Vlandas' website

Obama was elected in 2008 during the height of the crisis with much expectations that he would improve the conditions of the least well-off in American society. As his term nears its end, I use the latest OECD data to assess how unemployed individuals fare now compared to both other developed countries and to when he started his term.

In a nutshell, I show that in all plausible scenarios, individuals in the initial phase of unemployment are worse off in the US than in a median OECD or EU country and their relative situation has gotten worse since 2007, especially if they are in the middle class.

OECD data

The OECD has data on the net replacement rates for six family types in the initial phase of unemployment. The latest data available is 2014 so it is in principle plausible – though unlikely – that the situation has improved massively in the last two years and that this is not captured by my data. Note further the emphasis on the initial phase of unemployment: after a certain time in unemployment – which varies by country – unemployed lose eligibility to certain benefits (or experience falls in the replacement rate).

The OECD tax-benefit Model allows you to specify the marital situation of the family (single, one earner married couple, and two earners married couple) and whether they have children (in this case no children versus two children) for different levels of the average wage.

For simplicity I show the replacement rate for a case when the family does not qualify for cash housing assistance or social assistance in either the in-work or out-of-work situation. I also do not consider the case of earners with 150% of average wage.

The situation in 2014

Figure 1 shows the difference between the US and the EU/OECD average replacement rates in 2014. As an example consider the case of a single person with no children earning 67% of the average wage prior to becoming unemployed. In the US, the person would get 61% out of work income as a percentage of previous earnings equal to 67% of the average wage. The equivalent OECD median is 65% while the EU median is 68% so the gap between the US and the OECD median is 4 percentage points while it is 7 percentage points between the US and the EU.

Comparing different family and income situations reveals that the US is least generous compared to the OECD and EU median for ‘middle class’ (100% of average wages) families that are composed of lone parents with 2 children. The next biggest gap between the US and the OECD/EU median is for low income families with either one earner couple or a lone parent.

By contrast, low income families (67% of average wage) with two earners with or without two children would fare almost exactly the same in terms of replacement rate in the median OECD and EU country as in the US.

Figure 1: The difference between the US and the EU/OECD average replacement rates in 2014
EU US gap in 2014


Cross-national variation

In Figure 2, I show the 2014 cross-national variation in the replacement rate for a single earner with no children that does not qualify for cash housing assistance or social assistance and had previous earnings of 67% of average wage. This reveals that the US is not the worse country among developed countries (the worse is not surprisingly the UK – though note that the situation does not quite look as dire for the UK if the family qualifies for cash housing assistance). But it is located in the bottom half of the ranking.

Figure 2: 2014 replacement rate for a single earner with no children that does not qualify for cash housing assistance or social assistance and had previous earnings of 67% AW
Figure 3


Changes since 2007

In 2014, the average across all family types and both income situations for the US is 59% compared to 70% for the OECD median and 71% for the EU median. This represents a fall from 2007 where the average was 62% for the US, 60% for the OECD and 71% for the EU median.

Figure 3 shows how the gap between the US and the EU has evolved between 2007 and 2014 for different family-income scenarios. Positive values indicate that there has been an increase in the gap between what a person would get in a median EU country and what they would get in the US. Thus for instance, we see that the biggest increases in the gap has been for one  earner married couple with no children that earned 100% of the average wage prior to becoming unemployed and a single person with no children also earning 100% of the average wage prior to becoming unemployed. The next biggest increase has occurred for a lone parent with two children.

Thus, over the Obama presidency, the welfare of vulnerable middle class families in the US relative to their counterpart in the Europe has gotten worse. This is a striking result given the significant retrenchment of European welfare states that has taken place in Europe between 2008 and 2014 (the period under consideration here).

Figure 3: The difference between the EU-US gap in 2007 and in 2014

Change between 2014 and 2007.jpg 

Disappointing but not surprising

While disappointing, the falling welfare of the unemployed in the US is not entirely surprising from a theoretical perspective. The welfare state literature makes clear that liberal welfare regimes’ structure (e.g. targeted means tested benefits) limit the popular support for generous welfare state benefits for the unemployed that are seen as particularly undeserving.

As Rehm brillantly discusses in his latest book “Risk inequality and Welfare state“, countries with concentrated risks of unemployment among low income workers are less likely to exhibit pro-welfare state cross-class coalitions. At the same time, the US type of capitalism limits the power of the unions while making it unlikely that employers will consent – in the words of Korpi – to more generous social policies (see Varieties of Capitalism literature).

As a result, where there is no clear efficiency imperative (e.g. Obamacare in the context of objective inefficiencies in the health care sector in the US), it is therefore difficult even for a left leaning government to undertake an expansion of welfare state policies.