Revisiting aggregate social welfare spending of the United States and the social democracies

Here, I revisit a previous post, or specifically, this graph by the OECD:

This graph shows the total (public + private) social welfare spending of OECD countries as a percentage of GDP. The take-home is clear: the United States does not actually have a less expensive social welfare system than the social democracies. Once we account for private spending on social welfare, it is clear that the American social welfare system costs just as much as the social democratic model.

Two things prompted me to return to this graph. First, the OECD updated the web page the graph was on (my original post linked to the page in 2014), and the graph is no longer at that link. Thus, it is no longer clear where the data from the graph comes from. Second, I didn't initially realize how important this graph is. This graph deserves its own post.

The page where I got the above graph is actually the home of OECD's social expenditure (SOCX) dataset. SOCX compares national expenditures on social welfare and--according to this dataset--the social democracies spend about 30% of GDP on social welfare, the United States about 20%. This only includes spending on government benefits.

However, in 2011, a team of OECD researchers realized that these data were extremely misleading and detailed their findings in a working paper, Is the European Welfare State Really More Expensive? (Adema, Fron & Ladaique). I'll focus on their findings for the United States and the social democracies since this is my primary concern on this blog. There was also a 2014 update, which we'll take a look at after reviewing SOCX methodology and the refinements of Adema, Fron & Ladaique.

First, the SOCX dataset greatly overestimates the the social spending in the social democracies. The biggest issue is cash benefits. In the United States, cash benefits (like Social Security payments) are not subject to income taxes. But that's not so in the social democracies, and the result is that a huge amount of social welfare benefits are actually returned to the welfare system through income taxes. Adema, Fron & Ladaique refer to this as "claw-back." Clearly, it makes no sense to count every dollar in social welfare benefits as a social expenditure, since much of those benefits will be taxed and thus returned to the welfare state. In Denmark, the amount of public social welfare benefits that are returned to the welfare state by direct taxation amounts to a whopping 4% of GDP. Clearly, SOCX is wrong to disregard this.


A parallel concern is sales taxes (technically VAT), which are far higher than in the United States. Social welfare benefits are obviously subject to sales tax as beneficiaries spend their benefits, and the social democracies use sales taxes to claw back even more social welfare spending and return it to the welfare state. All told, 6-7% of social welfare expenditures are returned to the welfare state by some manner of taxation in Sweden and Denmark--meaning that when the SOCX simply reports welfare state expenditures in the social democracies without accounting for claw-back, it overestimates social welfare spending by 6-7%. That 6-7% might be paid out in benefits, but it isn't available for beneficiaries because it is recycled back into the welfare state.

For the United States, the SOCX is similarly misleading. The SOCX dataset only counts direct spending as a social expenditure. It does not count tax expenditures as social welfare spending, which is preposterous; tax expenditures are equivalent to direct spending. Tax expenditures for social purposes (subsidies for employer-sponsored health insurance, subsidies for 401(k)'s and IRA's, etc) must be taken into account in order to understand how much the United States spends on social welfare. Again, this is not negligible; the United States' tax expenditures for social welfare account for a mammoth 2% of GDP but are basically zero in the social democracies:

In sum, taking into account claw-back and tax expenditures radically shakes up the rankings by social expenditure:

Moving from gross public to net total social expenditure not only leads to greater similarity in spending levels across countries it also changes the ranking among countries. Denmark, Finland, Norway, Luxembourg, and Spain drop more than five places in the rankings (Chart I.11) and all these countries tax benefits and associated consumption above the OECD average (Table I.4)...By contrast, Chart I.11 shows that Australia, Canada, Germany, Iceland, Japan, the Netherlands, the United Kingdom and the United States move up the rankings by 4 or more places...As private social spending is so much larger in the United States compared with other countries its inclusion moves the United States to 5 th place when comparing net total social spending across countries.

Here is a graph I made of the social expenditure data from Adema, Fron & Ladaique comparing the United States and the social democracies:

Clearly, there is no appreciable difference in aggregate social welfare spending between the United States and the social democracies when we do a full accounting (public and private expenditures on social welfare, claw-back, tax expenditures for social welfare, etc). The United States spends about as much on social welfare as a percentage of GDP as the social democracies, clustering around 25% of GDP.

In 2014, the OECD repeated this analysis for 2011 data and found that the United States ranked even higher in private plus public social welfare spending. Again, I've highlighted the United States and the social democracies by adding red boxes. The orange diamonds indicate social welfare spending as a percentage of GDP. The orange diamonds is the net spending: the dark blue is government direct and tax expenditure on social welfare; the light blue is private expenditure on social welfare; and the gray is claw-back. Dark blue + light blue - gray = orange diamond.

For 2011 data, the United States had higher social welfare spending as a percentage of GDP than all of the social democracies. Remember, the Great Recession technically ended in June 2009, but even if we consider these data an aberration of the Great Recession, the above 2007 data predate the Great Recession and shows a firmly consistent conclusion: the United States has about the same aggregate social welfare spending as the social democracies.

There are a few other points worth mentioning.

First, these data look only at expenditures, not at contributions. In other words, these data count social security benefits but not taxes; they count private pension payouts but not contributions. Thus, the data have nothing to say about, for example, a pension with a funding shortfall. Since we're interested in comparing total cost of the welfare state (ie, welfare spending), this is exactly the dataset we want.

Second, the authors also note an important limitation in the data for the United States (emphasis added):

...employer-provided health benefits to their workers, dependents and retirees were estimated to be USD 760 billion in 2007 or 5.5% of GDP (these expenditures do not include payments by individuals for health services).

In other words, even after the corrections of Adema, Fron & Ladaique, SOCX still underestimates the social spending of the United States because they cannot account for out-of-pocket spending on health care. The authors only have data for spending on health insurance premiums. Clearly, by ignoring out-of-pocket costs, the authors have greatly underestimated the cost of the American welfare state. By contrast, the social democracies' health care systems have very little out-of-pocket spending.

Finally, the two largest areas of social spending in all countries are for health care and pensions. Since much health care and nearly all pension spending* is explained by age structure, it's worth considering age structure:

With the smallest share of people of retirement age and the largest share of working age adults, the United States should have an easier time supporting a large welfare state.



*I say "nearly all" because survivors' benefits paid to spouses and children of deceased workers count as pension spending.



Note: The graph I used for the original 2014 post (the graph at the top of this page) does not match the data of Adema, Fron & Ladaique (2011) or the 2014 update. For Adema, Fron & Ladaique, the United States and the social democracies cluster around 25%; in the graph at the top of the page, they culster around 30%. And for the 2014 update, the United States has higher aggregate social welfare spending than the entire social democratic group. Though the graph was definitely made by the OECD, I can't figure out what data they used because of the broken link. I considered that it might have included tax expenditures but not included claw-back, but the numbers didn't match. I did a tineye search for the mystery graph but got no useful results. If anyone has any ideas, please leave them in the comments; I'm perplexed as to the origins of this mystery graph.