Earlier this month, Virgilio Urbina Lazardi had a piece in Jacobin Magazine where he argued that so-called “predistribution” is a bigger factor in Europe’s lower level of inequality than “redistribution.” As I noted in my response, Lazardi’s argument was based on a study that excluded children, counted most of the welfare state as pretax income, and then assumed virtually all other public spending had no impact on inequality. In other words, it managed to prove that the welfare state does not drive the low-inequality outcomes by defining the welfare state out of existence.
I felt bad for Lazardi for falling prey to a badly written paper (Blanchet et al. (2022)). I’ve been there before myself. But in my piece, I also explained what researchers do to generate this offbeat conclusion in order to help people avoid falling into this trap in the future:
To reach a contrary conclusion, one has to either remove nonworking people from the analysis or count certain welfare incomes as market incomes.
In his response to my piece, Lazardi points to a new paper that he says shows that “predistribution” really is the biggest factor:
For what it’s worth, there are multiple other studies that come to similar conclusions, including one published last month showing that the equality of the Nordic countries’ economies is explained to a greater extent by their two-tier wage bargaining systems than by the breadth and generosity of their taxes and transfers.
I read this new paper (Mogstad et al. (2025)), and found that it used the same tricks I warned about in my first piece. Specifically, the authors:
- Restricted the analysis to individuals between the ages of 18 and 64, which removes children and the elderly, i.e. the vast majority of nonworkers.
- Counts “employment-related social insurance transfers” as market income rather than welfare income. There’s no further elaboration, but this term is typically used to refer to unemployment, disability, old-age, and leave benefits that use earnings-related benefit formulas.
No worries though as Mogstad finds comfort in the fact that:
This finding is consistent with previous studies that compare the inequality in disposable income and market income between countries … [f]or example, Blanchet et al. (2022) …
It’s one thing to get tricked by one badly-done study. It’s another to have just been told how the trick works and then cite another study that uses the same trick. At some point, Lazardi has to actually read what he’s citing!
Nope
The rest of Lazardi’s piece, where he attempts a partial defense of Blanchet, is also full of weird nonsense. For example, he begins the defense by saying:
Bruenig has two principal lines of attack on the Blanchet et al. article. [a] The first is that the authors use a definition of pretax income that is net of the payroll taxes that fund certain social contributions, including public pensions as well as unemployment and disability insurance. He repeatedly casts aspersions on their use of this definition, as if it were an accounting “trick” peculiar to them, when in fact this is the way most states around the world measure pretax income. [b] In the United States as in Europe, income taxes are assessed net of payroll taxes, largely to avoid placing individuals in a higher tax bracket than is justified.
Regarding (a), the problem is not just that Blanchet counts social insurance contributions (which Lazardi calls “payroll taxes”) as reducing pretax income rather than as redistributive tax. It’s that Blanchet also counts social insurance benefits as pretax income. This means that unemployment, disability, old-age, and leave benefits are not counted as part of the redistributive welfare state.
To put this in concrete terms, imagine a country with a huge amount of poverty and inequality owing to the fact that a large share of unemployed, elderly, and disabled people receive little to no income. Then that country creates unemployment, disability, and old-age benefit schemes funded by taxes on labor. As a result of these new schemes, poverty and inequality plummet. Was this inequality reduction achieved by welfare state redistribution or by pretax predistribution? Based on normal understandings, clearly the former. Yet Blanchet, Mogstad and, by endorsement, Lazardi use odd accounting specifications to say it’s the latter.
Regarding (b), this is just factually false. In the United States at least, state and federal income taxes are assessed net of employer-side payroll taxes but not net of employee-side payroll taxes. Blanchet’s decision is also clearly not a quirk of payroll tax administration. As I noted in my piece, in Denmark, which has no payroll tax, Blanchet actually takes some of Denmark’s income tax and declares it to be a social insurance contribution as part of his accounting exercise.
Not counting employer-side payroll taxes as redistribution is also obviously indefensible. In Sweden, those taxes are 31.42 percent of gross pay, which is effectively a 23.9 percent flat tax, on all labor income with no cap. In the United States, the same taxes are 7.65 percent of gross pay, which is effectively a 7.1 percent tax, on all labor income up to $176,100 of income. For labor income beyond $176,100, the tax falls to 1.45 percent of gross pay, which is effectively 1.43 percent. Baking these taxes into your pretax/market/predistribution baseline, as Blanchet does, makes it so that the inequality reduction achieved by Sweden’s much more aggressive tax approach is not counted as redistribution at all.
Put differently, if the United States tripled the employer-side payroll tax while eliminating the cap so the higher tax applied to all income, Blanchet’s specification would score the resulting inequality reduction as “predistribution.”
Contrary to Lazardi’s theory, what I suspect is actually going on with Blanchet’s approach is that he is operating off the idea that “social insurance” (old-age, disability, and unemployment benefits) is non-redistributive because it involves workers “paying” for their “own” benefits or something to that effect. It’s the same sort of folk logic that drives a lot of weird distinctions in the US between things like Social Security (not “welfare”) and Food Stamps (“welfare”) and gives us classical political moments like when that guy shouted that he wanted politicians to “keep your government hands off my Medicare.”
Say What You Want But Be Clear
Ultimately, I don’t think Lazardi and I even diverge in our understanding of how to deal with inequality. When you look at the basic math involved in these questions, it’s clear what you have to do:
- Compress the wage scale, such as through unionization and collective bargaining.
- Redistribute capital’s share, either socially (as in Alaska) or through a higher labor share.
- Provide income to nonworkers via the welfare state.
What often happens in this debate is just that some people have various ideological hang-ups that make them think that (1) and (2) are real hardcore anti-capitalism and (3) is not. And downstream of that, they go about trying to prove that (1) and (2) matter more than (3). But, at least when it comes to overall inequality, this just isn’t true.
What is true is that:
- Wage compression is the most important lever for reducing inequality among workers.
- The welfare state is the most important lever for reducing inequality among the society as whole, which includes nonworkers.
If you really want to, you can generate the conclusion that wage compression is the most important way to reduce inequality among the society as a whole by excluding nonworkers and counting a bunch of the tax-and-transfer system as if it is predistribution. But what’s the point really? You get to publish a sentence saying “the welfare state actually is not the biggest driver of inequality reduction,” but only because you dropped a footnote 20 pages in that explains that you’ve essentially defined the welfare state out of existence. That’s as silly as it is confusing.
With that all said, if someone is so dead set on making sure that the word p-r-e-d-i-s-t-r-i-b-u-t-i-o-n has primacy over the word r-e-d-i-s-t-r-i-b-u-t-i-o-n in these debates that they are willing to grab huge chunks of the welfare state and move it out of the redistribution bucket and into the predistribution bucket, then I guess that’s fine too. In that case, I must of course agree “predistribution* is the biggest driver of inequality reduction.”
* “Predistribution” includes the welfare state.