Sunday, July 29, 2018

A la recherche of the roots of US inequality “exceptionalism”


It has been long argued that American income inequality was, in the past 40 years or so, exceptionally high compared to other OECD countries. The latest results available by Luxembourg Income Study that harmonizes income concepts across countries show inequality in disposable (per capita) income in the US to be 41 Gini points, that is, higher than in any other similarly rich country (Germany’s Gini is 32, British 35, Italian 35, Dutch 28). So, this part is not controversial.

What is more controversial is technical (as opposed to substantive) explanation for this “exceptionalism”. Some people have argued that US market income inequality (that is, inequality before government redistribution through social transfers and direct taxes) is not much higher than elsewhere and that the entire explanation has to do with an insufficiently redistributive state. In simple terms, the argument is that the market generates same inequality in the US and Sweden, but Sweden redistributes much more through pensions, unemployment benefits, social assistance etc., and also taxes the rich more, so in the end disposable (after transfers and taxes) income inequality in Sweden is less than in the United States.

Janet Gornick, Nathaniel Johnson and I have recently looked at this more carefully. Without going through all explanations (which can be found in the paper here), we conclude that this is not entirely true: US market income inequality is generally greater than in other rich countries and  the American state  redistributes less. So, we argue, both the underlying (market) inequality is high and redistribution is relatively weak.

But one can go further than that, and ask the following question: what part of redistribution is “weak”: is it that US transfers are small and not sufficiently pro-poor, or is it that US direct taxes are not sufficiently progressive?

Now, I look at that issue in the following way. I define as “poor” the bottom 40% of individuals when people are ranked by their market income inclusive of government-paid pensions (social security in the US) which can be regarded as deferred wages.  I then look at how the income share of these very same people varies as we include other social transfers and finally as we deduct direct taxes. (Note that this calculation can be done only if you have access to micro data, as is the case with LIS data, because you need to “fix” these people and look at their income and income share as they go through the process of redistribution.)

We expect that the share of the “poor” increases as the state moves in to redistribute income. Indeed, in 2016 (the latest year for which we have US data), the “poor” received 11.7% of overall market income, but their share went up to 13.4% of income when we include all social transfers, and increased further to 15.8% when we include taxes too. (Note again that these are the same people throughout). The gain for the “poor” is thus 1.7 percentage points from social transfers (13.4-11.7) and an additional 2.4 percentage points from taxes (15.8-13.4).  

We can write it out:
In the US, the “poor” gain 1.7 points thanks to social transfers and 2.4 points thanks to taxes.

So, the government really “works” in the United States: it improves the position of the poorest people through government transfers and direct taxes. But the question is, does it work well enough?

One good comparator is Germany. We control for different age distributions in the two countries and the  fact that people retire earlier in Germany by treating government pensions as deferred wages. But that still leaves (as mentioned  above) other social transfers like unemployment benefits, family benefits (if any), welfare etc. So, in Germany in 2015, the “poor” (defined the same way as in the US) earned 15.3% of all market income. Their share went up to 18.3% when all social transfers are included, and further to 21.3% when we include direct taxes as well. Thus the “poor” in Germany gained 3 percentage points from social transfers (18.3-15.3) and 3 percentage points from taxes (21.3-18.3).

For Germany, we write:
The “poor” gain 3 points thanks to transfers and then an additional 3 points thanks to taxes.

Thus, not only is the starting point of the “poor” in Germany more favorable than in the United States (15.3% of market income vs. only 11.7%) but they gain more from both social transfers and direct taxes.

The results over time are shown in two graphs below. The “poor” always gain from redistribution but US gains are always smaller than German gains. What is noticeable is that the gains from social transfers were about the same in the US and Germany until 1995, then increased in both countries. In the US they were at their peak in 2010 when unemployment benefits were extended by Obama and afterwards, since US welfare is very modest, they rapidly went down.

Even more interesting is the evolution of the gains from direct taxes. Here we see that the American “poor” gain throughout less than the “poor” in Germany and that the level of gains does not seem to change much in the US. 
 
 

In conclusion, when we try to find the roots of lower pro-poor redistribution in the US we can find them both in more modest social transfers and in less progressive direct taxation. Combined with our earlier finding of relatively high market income inequality in the US, this means that American income inequality is “exceptional” because (a) underlying market income inequality is high, (b) social transfers are modest, and (c) direct taxes are not sufficiently progressive.

The policy implication is that reduction in US income inequality is unlikely to be achieved through one of these three channels alone but through a combination of  “improvements” in each of them. For example, through more accessible education and higher minimum wage to reduce the underlying market income inequality; through introduction of family benefits or more generous welfare; and finally through higher tax rates for the rich and higher taxation of capital incomes. Although this might seem like an extremely ambitious policy agenda, I think it is more reasonable to think that incremental changes in all three channels are easier to pass legislatively than a much more substantial change in any one of them alone. But it also means that if one wants to seriously grapple with high inequality in the United States, only a combination of different policies will do.