In one part of a rather ambitious article in the Journal of Economic Literature’s September issue, Norwegian economists Magne Mogstad, Kjell Salvanes, and Gaute Torsvik attempt to identify the causes of the relative income equality enjoyed by Nordic countries compared to other developed economies. Sadly, over the couple of months since the article was published, this bit of their research has been taken up by advocates of predistribution over redistribution as proof welfare programs are less effective weapons in the fight against income inequality.
The authors aren’t totally blameless of course. One of their paper’s key findings, they wrote, is “that a more equal predistribution of earnings, rather than income redistribution, is the main reason for the lower income inequality in the Nordic countries compared to the United States and the United Kingdom.” This is wrong.
Without getting too into the weeds, Mogstad, Salvanes, and Torsvik tried to draw out the effects of redistribution by looking at OECD data on countries’ Gini coefficients—a common measure of income inequality—before and after accounting for taxes and transfers.
A Gini coefficient of 1 reflects perfect inequality and a Gini coefficient of 0 reflects perfect equality. If the Gini of disposable income (after taxes and transfers) is less than the Gini of market income (before taxes and transfers), then government “redistribution” so defined reduces inequality by that difference.
This is more or less fine: the primary issue with the approach taken by Mogstad, Salvanes, and Torsvik isn’t methodological, it’s the specific data they chose to use. They only look at the Gini coefficients for working-age individuals (18 to 65 years old). As Matt Bruenig noted, this essentially “prove[s] that the welfare state does not drive the low-inequality outcomes by defining the welfare state out of existence.”
After all, the primary purpose of the welfare state is to distribute income to nonworkers. If you exclude one of the single largest groups of nonworkers (people of retirement age) from the calculation of your measures of inequality, of course it will look like the welfare state isn’t meaningfully reducing inequality. Data about the people who benefit the most from social security and other transfer programs are being totally excluded from the calculation of your Gini coefficients.
To show why this is the case, here’s a simple bar chart of Ginis of market income for working-age and retirement age populations. Across every single country Mogstad, Salvanes, and Torsvik highlight in their analysis of Gini coefficients, income inequality before taxes and transfers is far greater for older households. Intuitively, this is because a higher share of people above 65 years old have zero labor income which mechanically increases inequality within that group.
So if we want to understand how much redistribution actually reduces inequality, we need to look at the Gini coefficients for the entire population of each country before and after transfers. Thankfully, these are also published by the OECD. Once we make that simple swap, the picture suddenly changes.
| Country | Market Income Gini | Disposable Income Gini | Reduction in Gini | Percentage Reduction |
| Nordic Population Weighted Average | 0.39 | 0.27 | -0.12 | -30.8% |
| Denmark | 0.40 | 0.27 | -0.13 | -32.5% |
| Finland | 0.43 | 0.28 | -0.15 | -34.9% |
| Norway | 0.39 | 0.27 | -0.11 | -28.2% |
| Sweden | 0.36 | 0.27 | -0.09 | -25.0% |
| United Kingdom | 0.45 | 0.36 | -0.09 | -20.0% |
| United States | 0.47 | 0.39 | -0.08 | -17.0% |
| Country | Market Income Gini | Disposable Income Gini | Reduction in Gini | Percentage Reduction |
| Nordic Average (Population Weighted) | 0.453 (0.449) | 0.270 (0.271) | -0.183 (-0.17 | -40.4% (-39.9%) |
| Denmark | 0.445 | 0.268 | -0.177 | -39.8% |
| Finland | 0.512 | 0.273 | -0.240 | -46.9% |
| Norway | 0.427 | 0.261 | -0.166 | -38.9% |
| Sweden | 0.429 | 0.277 | -0.152 | -35.4% |
| United Kingdom | 0.508 | 0.366 | -0.142 | -28.0% |
| United States | 0.505 | 0.395 | -0.110 | -21.8% |
Using this more well-suited measure of inequality, taxes and transfers appear to reduce the United States’ Gini coefficient by 11 points, and reduce the average Nordic country’s Gini coefficient by 18.3 points. The final gap between the average Nordic Gini of disposable income and the American Gini of disposable income is 12.5 points, of which 7.3 points appear plausibly attributable to differences in tax and transfer programs.
Just by not needlessly excluding millions of people from the calculation of our Gini coefficients, suddenly around three-fifths of the difference in inequality between the U.S. and Nordic countries can be attributed to differences in their respective systems of taxes and transfers, compared to the Mogstad, Salvanes, and Torsvik estimate of around one-third. And as Bruenig has pointed out, moving “employment-related social insurance transfers” out of the definition of market income and into transfer income would likely increase estimates of the welfare state’s role in reducing inequality even more.
Mogstad, Salvanes, and Torsvik didn’t manage to show in their paper that predistribution is more important than redistribution as a factor explaining Nordic countries’ relative income equality. They showed that having a more egalitarian distribution of wages reduces income inequality more than the welfare state for the populations most likely to have wages as a major share of their income.
If you really want to reduce inequality, you need to ensure nonworkers have money. That means you need a strong welfare state.