The New York Times made waves this week with another piece on inequality, saying that it has not risen since 2007. The article was based on this paper by GWU’s Stephen Rose.
The article also suggests that expansions are not a good way of looking at trends in inequality (as I have done in the past, also covered by the NYT). Instead, one needs to look at the business cycle. It also concludes that, thankfully, because of government tax and transfer policies, inequality has not been “that bad” over the last few years and governments can clearly do something about it.
So what’s wrong with this picture?
Here is the graph that appeared in the NYT (I’ve reproduced it below showing only the bottom 90% and top 10% of families using the same Saez data).
Now let’s reproduce the exact same graph, using the same data but excluding capital gains. The trends reverse. The bottom 90% of families have lost proportionately more than the top 10% since 2007.
“Now, I am not fond of excluding capital gains (I am in favor of annuitizing them), because they are very important to income dynamics, but still, without capital gains, the bottom 90% lose proportionately more (relative to the top 10%) than with them.
In any case, if we include the top 1% and the 0.01% in the above two charts, one would find that they do lose proportionately more including or excluding capital gains.”
In any case, if we include the top 1% and the 0.01% in the above two charts, one would find that they do lose proportionately more including or excluding capital gains.
However, the bottom line is this: this exercise gives an extremely narrow look at income distribution trends, based on a very incomplete picture. As Nick Bunker from the Washington Center for Equitable Growth put it:
“Reasonable people can disagree about the best benchmark. But what isn’t reasonable is using a peak as a benchmark to claim inequality hasn’t increased over an incomplete business cycle.”
So let’s look at complete business cycle data. The following chart shows how the distribution of income growth has evolved from one peak to another.
There is a clear shift in trend after the 80s. During 3 out of the last 4 complete business cycles, the wealthy 10% have gotten a proportionately greater share of the growth. And in the last full business cycle (2000-2007), they got all of the growth, while incomes of the bottom 90% fell. Yes, since 2007, both groups shared the losses about equally, but why should we be surprised that the top 10% shouldered 45% of the decline? (Again, this is not a complete business cycle yet!)
We live in a casino economy driven by serial asset bubbles, where the incomes of the wealthy (and not just their capital gains) are increasingly tied to stock market performance.
So when the biggest bubble in human history popped, the wealthy families lost a ton of income. At the same time middle class households fell into poverty, lost their decent jobs and pay, and got unemployment insurance or food stamps from the government. Can one really conclude from this that inequality is not “that bad”?
As an example, inequality will not be “that bad” if one person in the US earned 100% of all the national income, and then the ‘evil government’ (or ‘benevolent dictator’; take your pick) decided to tax most it and then gave transfers to the rest. But is this the kind of ‘better’ income distribution that we are aiming for? Aren’t we all talking about an economy where most people have decent jobs, decent wages, decent salary growth prospects, and decent chance to participate and share in that growth?
There are many ways to slice and dice this data.
I have looked at expansions because they answer a very specific question: Once the economy returns to some normalcy and promises to deliver prosperity, to whom does it keep its promise? And the answer is, increasingly to the top 10%.
The problem with the NYT article is not the inequality chart, even though it shows an incomplete and thus misleading business cycle picture. The problem is the conclusion: that ‘taxes’ and ‘transfers’ are the solution to the deep structural economic problems that are causing the generation and distribution of incomes to be so inequitable from the very outset.
So the next time someone tells you that “a rising tide lifts all boats”, you can respond “no, increasingly it sinks most”.
*updated (2/20/2015 14:00 CT)
Dr. Tcherneva, thank you so much. I replied to Leonhardt’s Tweet of his article with yours. I hope he’ll come here and comment.
“At first these results may seem counterintuitive. Without capital gains, the incomes of the bottom 90% fall faster than with them.”
I think you’re misreading the top two charts. By 2009 , the bottom 90% had lost 11-12% of income in each case , and capital gains had little impact on the extent of their losses throughout the time period shown. This is as you would expect , since they hold relatively few financial assets. In contrast , the top 10% , who hold significant financial assets , lost ~23% by 2009 if cap gains are included vs 11% if they’re excluded.
Most of the top 10% financial assets are owned at the top of the top 10% , as you noted , so if you compared , say , the bottom 8 of the top 10% to the 90% , the curves wouldn’t be so different.
Btw , your excellent chart showing income distribution during expansions would probably hold true for many other advanced economies based on a chart from today’s Economic Report to the President. It’s shown as fig 4 in this WSJ piece :
http://blogs.wsj.com/economics/2015/02/19/the-u-s-economy-according-to-the-white-house-in-10-charts/?mod=WSJBlog&utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+wsj%2Feconomics%2Ffeed+%28WSJ.com%3A+Real+Time+Economics+Blog%29
Marko, thank you for pointing that out! Part of the text actually doesn’t correspond to that chart. I copied and pasted from another piece I’m writing for Levy that looks at growth shares, not % changes for the 99% vs. 1%. It should be corrected soon!
What the Levy piece shows (coming up) is that when we compare the shares of income growth during the last expansion (2009-2013) (including and excluding capital gains, respectively, for the 99 percent and 1 percent), one notices that the distribution of income growth appears less unequal when capital gains are included. At first, this may seem like a counterintuitive result. What the data illustrates is that, even though capital gains are a very small share of income for the bottom 99 percent of families (only about 2 percent of their income comes from capital gains), they are the difference between a shrinking income and a marginally growing income for those families during that period.
The point in this blog still stands: We can slice and dice the data many ways, but whichever way we look at it, it better be a sensible comparison. The first 2 charts don’t tell us much, as they are considering an incomplete business cycle.
Thank you for the link. I did look at other countries, using the same database. Though the data does not show the smooth and clear trend as my US expansions chart, there is evidence of worsening inequality across the advanced economies.
The final graph is not for US business cycle peaks. There is one, I don’t know why you didn’t use it. Also, since it ends at 2012, it is not accurate to say that it is measuring peak-to-peak, including the current expansion. It is pretty clear from Sentier Research’s monthly updeates of real median household income (a slightly different measure) that the trough in real incomes was in 2011 or 2012.
Additionally, since income falls dramatically upon retirement, and massive amounts of Boomers have been retiring, unless it adjusts for demographics, it is very vulnerable to that criticism.
I don’t dispute the overall point of this article, but the correct graph should be used, noting its limitations.
New Deal democrat,
Thanks. This last chart is a business cycle chart. If takes literal peak years in average incomes from the Saez spreadsheet and examines what happens during that time. The objection to my expansions chart has been that it ignores declines (the idea being that the rich lose a lot more proportionately during down years). But they also recover more than what they’ve lost for most of the cycles since the 70s. So the above chart looks exactly at those periods that correspond to a decline in incomes from peak to the recovery until the next peak. The data here uses the earlier Saez spreadsheet and is not updated to 2013, which my upcoming Levy note does, but the trends do not change.