Wednesday, October 1, 2014

Cyclical Misdirection in Income Distribution

A common complaint in the public discussion of income trends is the tendency to treat income bins as stationary.  If the top 10% of incomes increases by 50% over a decade, but the lowest 10% remains level, then someone might say, "All the gains went to the top 10%." or "The rich got richer while the poor got poorer."  But, the complaint is that these aren't the same people.  There is significant movement between bins, because of lifecycle effects, economic mobility, etc.  At long enough horizons, most of the original occupants of a given bin will have moved to other bins.

But, in some ways, especially with the popular treatment lately of describing the top 5% or 1% or 0.1%, there can be very strong errors from this linguistic statistical short cut due to cyclical changes, even in relatively short time frames.

I am going to describe this in a very broad brush.  Short of doing very involved panel studies, I'm not sure if empirical data is possible.  Please let me know in the comments if anyone knows of research along these lines.

Imagine two kinds of income earners in an economy - wage earners and capital owners.  In a deep recession like the one we have just experienced, they will have very different experiences.

The distribution of wage incomes will shift slightly to the left, mostly as the result of a small percentage of wage earners losing their entire income through unemployment.

The distribution of capital income might be fairly tight entering the recession, since in the mature portion of the previous recovery, bonds would be paying well, and equities would be growing at rates similar to broad economic growth.  During the recession, especially when equities lose half their value as they did in this recession, capital income will be negative for many owners, with more variance.  And, as the economy snaps back, capital income might be higher than normal, but again with more variance, as equities produce above average returns, and extra gains depend on how much exposure each owner has to equity.

In an economy peopled by, say, 90% wage earners, the distribution would be overwhelmingly peopled by wage earners, but the cyclical shifts would be almost entirely the product of capital income earners shifting in the distribution.

When we are talking about the extreme ends of the distribution (the top or bottom few percent), the error will always make the changes in distribution look less equitable than they really are, because the population shifts from the extremes can only go in one direction.  Households that move out of the top bin can only move into lower bins.

So, unless we are using panel data or somehow adjusting for these changes, in the pre-recession period the top 1% will be a rough mixture of capital and wage income.  Then, during the recession, the top 1% will consist almost entirely of wage income.  And, after the recession, because of the added variance in capital outcomes and because of the catch-up growth that tends to come after severe downturns, most of the top 1% will consist of capital income.

At the bottom of the distribution, the large number of capital earners with negative returns would increase the population of very low income units so that the lowest bins would decline and the median would decline.  Also, if the data source has a $0 lower bound and is based on tax reporting with carry-forwards, then the variance of post-recession capital income could be especially high, with many units claiming very low incomes until they use up their tax assets, then suddenly jumping back into the highest income bins if returns continue to recover.

So, we would want to say that the highest incomes suffered the least decline in the recession and then snapped back the most, while the lowest incomes and the median incomes suffered.  But, ironically, the suffering we would be measuring would be mostly the suffering of the capital owners.  Of course, capital suffers the brunt of the loss in recessions.  That's why it earns a premium.  That's why we don't recommend working class families and pension funds to take highly leveraged positions on the stock market.  As with so many things, most people understand this intuitively in their personal dealings.  But, when we approach the bigger picture, we can lose our moorings.


  1. Thanks, Travis. It seems like panel data would be the only way to get a good handle on these issues.

  2. Why did General MacArthur institute land reform in both Japan and the Philippines?

    1. We do have relatively widespread real estate ownership. Unfortunately, risk aversion has pushed real estate prices up so high that real estate ownership is nearly the same as long-duration bond ownership.

  3. All in all, that's a pretty good response. My largest disagreement would be with this:

    "And it is true that if you ignore the economy's tilt in favor of owners and against workers, and also ignore 15 years of persistent labor market weakness, then the overall economic picture looks a lot brighter."

    This is what is so difficult about public discourse. Tiny little rhetorical tools can overwhelm a lot of goodwill. We are all susceptible to this. It probably undergirds most disagreement. After a long response with lots of sensible ideas, this line is simply not rooted in reality. No capitalist, given the choice, would have chosen the last 15 years as a favored period. Yglesias must not have a significant amount of self-managed savings.
    And, the idea that labor markets have been weak is also backwards. I assume this comes out of the concern with falling labor force participation. But, while some of the decline in the past 5 years has been cyclical, over 15 years, this is clearly demographic. He simply hasn't dealt with the statistical effects of the current demographic profile in the labor market. In fact, how much of the mediocre growth in capital has gone to Silicon Valley entrepreneurs (Google, Facebook, E-Bay, etc.)? The conflation of this sort of disruption with oligarchy is ignorant.