The facts of the European (EU27) income convergence

How is Europe becoming more equal

The issue of income convergence within the European Union has recently been much discussed. There seem to be two reasons for this. First, it is undeniable that using either macro indicators (like GDP per capita) or data from household surveys like SILC there has been a reduction of between-country and inter-personal inequality within the 27 members of the Union. Secondly, this convergence takes place while there is no convergence between the EU as a whole (or between the largest EU member-states) and the United States. The latter issue has been raised especially now because of the tensions in the relations between the two Western poles. The lack of convergence was brought up originally by Trump to show persistent American superiority, and then by Trump’s detractors who argued that the absence of European convergence was due simply to the fact that Europeans, by choice, work less and that output per hour of work may be the same or even greater in some European countries than in the US.

I will not discuss this second issue (I wrote about that here). But I will look empirically at the first. Recently my good friend Michael Dauderstädt from the Friedrich Ebert Foundation who has been studying income distribution in the European Union for a number of years has written a nice short paper on convergence and cohesion policies, published by “Social Europe” here) inspired me to look at my global micro (household per capita) income data and try to see what they can tell us about income convergence within the EU.

I take throughout the period studied, 1993 to 2023 (at five-year intervals), the current EU27 membership as given. I do not use the EU membership as it was then, but look at the twenty-seven current members as if they had all been members in all the years. The sample is thus in principle the same throughout except when for some years household survey data for some countries are missing. Luckily, this happens only in the early period (1993-2003) and the countries missing (principally Cyprus and Malta) are too small to affect the results.

Income data for each county come from nationally representative household surveys as harmonized by Luxembourg Income Study (LIS), and in the cases when LIS data for a country are not available, I use the data from World Bank’s Poverty and Income Platform (PIP; formerly POVCAL). The income concept is always disposable (after-tax) household per capita income (meaning that household income is divided by the number of household members) and income is adjusted for the difference in price levels between the countries by using international (or PPP) dollars. So, we measure real welfare of citizens from different countries by adjusting their nominal incomes by the estimated price level of the country. (Of course, we can expect that as countries’ incomes become more similar price levels would converge too, which indeed is the case for the EU27, but is not the subject of this Substack).

Figure below gives the main results. The height of the bar gives the overall Gini coefficient for EU27 as if it were one country. Gini has decreased from 0.415 in 1933 to 0.35 now. This is the headline result: a decrease of interpersonal inequality by 6.5 Gini points (41.5-35) or by some 15 percent of the original inequality. But what drove this decrease? Did inequalities within individual countries go down, and the gap between countries’ mean incomes remained the same, or the reverse? The latter: mean incomes of the countries (weighed by each county’s population) become more similar. This is called Concept 2 inequality: it assumes that everybody in a given country has the mean income of that country (so within-national inequality is zero), and calculates what would be EU27 inequality in that case. The blue part of the bar shows that the Concept 2 inequality was almost halved during the past thirty years: it went down from 0.25 to 0.14. It clearly illustrates the convergence of (population-weighted) average incomes between the countries. In other words, mean income in poorer counties like Romania and Bulgaria is now closer to the mean income of rich countries like Sweden and Netherlands than it was thirty years ago. The second Gini component (in red) which accounts for within-national inequalities and the so-called overlap term (which becomes positive when incomes of people from mean-poorer and mean-richer countries overlap) has increased.

Note: blue and red bars together yield EU27 interpersonal inequality. This can be compared with inequality in the US given by the green line.

The bottom line is that the significant reduction in EU27 inter-personal inequality was achieved thanks to the convergence of mean country incomes, not thanks to the reduction of within-national inequalities. For example, the average EU27 country Gini in 2023 was 0.31; it was 0.30 in 2008 and 0.29 in 1993. Within-national inequalities, on average, increased slightly.

How does inter-personal EU27 inequality compare with the United States? To see this, compare the height of the bar in the figure with the line which gives the US Gini for the same set of years (1993-2023). When our period begins, EU27 inequality was just slightly above the American inequality: Gini of 0.415 for EU27 vs. Gini of 0.395 for the United States. But over the next thirty years, EU27 inequality went down while US inequality increased. So, now EU27 has inequality that is more than 6 Gini points lower than the US. Broadly speaking, EU27 considered as a single country used to be slightly more unequal than the US, but is now significantly more equal. (One needs however to treat this result with caution because in the data for the United States, we do not adjust incomes for the differences in the cost of living/price level. In these data an income of $1,000 is treated the same when earned in New York and in Iowa. There is little doubt that if we were to adjust US state incomes by the state price levels US inequality would go down. I would expect that it may be reduced by 2-3 Gini points. Thus, short of doing this, we are not exactly comparing like with like in our EU27 vs. US comparison).

Finally, let’s ask this question: suppose that all EU27 countries achieve the same average income level while nothing changes in their internal income distributions; i.e., their within-country Ginis do not change at all, but the mean income gap between Luxembourg and Bulgaria becomes zero. What would be interpersonal inequality then? The answer is: Gini of about 0.32. That Gini is only 3 Gini points lower than the actual inter-personal inequality (recall that we found it to be 0.35) and this in turn sets the bound to what further mean-income convergence can accomplish in Europe. If inequality within the European Union is to be reduced, then most of the reduction in the next decades will have to come from within-national shrinking of income gaps, not so much any more from the convergence of countries’ mean incomes. This therefore contains an obvious message for European policy-makers.