In Jacobin, Virgilio Urbina Lazardi has a piece where he argues that European countries have lower inequality than the United States because they distribute market income much more equally, not because of their welfare states.
A paper published a few years ago in the American Economic Journal has raised eyebrows within and outside of the profession. Combining national accounts data with household surveys, its authors found that the United States redistributed a greater share of its gross domestic product through taxes and transfers to its poor than any of its wealthy, mostly Western European, peers.
As it turns out, the “inequality gap” between the United States and Europe is not explicable by the comparative generosity of the latter’s welfare states, which are in fact funded by more regressive systems of indirect taxation. Rather, the key to Europe’s relatively higher levels of equality is a more egalitarian distribution of pretax market incomes.
This is not true. Lazardi is simply the latest victim of the most deceptive paper ever written on this topic.
When trying to figure out what is most responsible for low inequality in developed countries, the welfare state always beats market-income compression for one simple reason: half of the population does not work. The nonworking half of the population drives up market inequality because they add a ton of zeroes to the bottom of the market income distribution. Providing welfare benefits to this nonworking half replaces those zeroes with positive numbers in direct proportion to how generous the welfare state is. The effect of moving these zeroes to nonzeroes completely overwhelms any other inequality-reduction mechanism.
To reach a contrary conclusion, one has to either remove nonworking people from the analysis or count certain welfare incomes as market incomes. The paper Lazardi is referencing — Blanchet et al. (2022) — uses both tricks.
Blanchet uses a pretax-income concept that counts unemployment, disability, and old-age pension benefits as pretax income rather than as welfare benefits. Blanchet’s pretax-income concept also subtracts from each unit any “social contributions” (taxes) paid towards these benefits. So the major way that welfare states reduce inequality — taxing workers and paying out benefits to unemployed, disabled, and elderly people — is actually baked into “pretax income” under Blanchet’s specification. Put differently, all of this welfare state action is considered by Blanchet to be “predistribution” not “redistribution.”
In the US, this means that the Social Security program (both taxes paid and benefits received) and the Unemployment Insurance program (both taxes paid and benefits received) are not counted as ways that the welfare state reduces inequality. Amusingly enough, in countries where they lack hypothecated “social contributions” for these benefits, e.g. Denmark, Blanchet even goes as far as to grab some of the nation’s income tax and pretend that it is a hypothecated “social contribution” that does not count as redistribution.
When assigning this income, so defined, to household units, Blanchet looks only at the personal income of adults aged 20 or above. Unmarried individuals are assigned their entire personal income. For married individuals, he takes their combined personal income and then assigns half of it to each individual. Children, which is the largest nonworking population (making up 44 percent of nonworkers in America) and a major welfare beneficiary population (receiving child care, school, and child allowance benefits in most countries), are completely excluded from this analysis.
Ignoring the existence of children in this way generates some bizarre outcomes. A single mother with three kids who has $30,000 of income is counted as being equal to a married coupled with no kids who have $60,000 of income. Yet, the single-mother family has $7,500 of income per person while the married couple has $30,000 of income per person. If we were to use square-root equivalization — a popular way of comparing families of different sizes — we would find that the married couple has 2.8x as much income as the single-mother family. Blanchet says they are equal.
In fact, Blanchet’s choice to exclude children makes it so that child-related welfare benefits — one of the few benefits he actually counts as redistribution — are potentially disequalizing. Imagine two adults, each receiving $30,000 in posttax income so defined. At this point, these adults are equal to one another. One of the adults has a child and, in their society, this entitles them to $3,000 per year in a child allowance. Under normal specifications, this $3,000 child allowance reduces inequality because having a child increases your household size, having a bigger household size reduces your household’s equivalized income, and then the child allowance bumps that equivalized income back up closer to the income of the childless adult. But in Blanchet’s specification, the $3,000 child allowance causes the adult with the child to have $33,000 of income while the adult without the child has just $30,000 of income. Thus, the child allowance no longer reduces the inequality between these two adults. It actually increases it!
The authors only consider three things to be “redistribution” in their specification, which they describe this way:
Social assistance transfers (5 percent of national income) are observed in survey data, so they can be added directly to individual incomes. We distribute other public spending proportionally to posttax disposable income (17 percent of national income), with the exception of public health expenditure (8 percent), which we distribute in a lump-sum way, considering that the insurance value provided by health systems is similar for everyone
“Social assistance transfers” includes child benefits, which are neutralized (or even made disequalizing) by excluding children from the analysis. “Other public spending,” which is the majority of what they consider redistribution, is neutralized by counting it as being distributed “proportionally to posttax disposable income.” Put differently, the author’s simply assume that this public spending does not affect inequality. All that is really left is public health expenditure, which, because they allocate it as a lump sum, does have a redistributive effect.
So once you put it all together, what the authors have done is shown that, if you count unemployment, old-age, and disability benefits and the taxes used to fund them as not being part of the welfare state, exclude children from your analysis, and assume the majority of all other public spending is not redistributive, then market-income compression is the major driver of lower inequality in Europe. This is a joke. Of course market-income compression drives the lower income inequality in this analysis. Virtually all of the welfare state mechanisms have been defined out of existence!