In accompaniment of a elegant opinion post for the New York Times, David Leonhardt produced what became an instantly celebrated graphic. He showed not only that economic gains in the 34 years before 2014 were tightly concentrated among the highest earners, but that this contrasts sharply with the 34 years before 1980.
The underlying data comes from a document by Thomas Piketty, Emmanuel Saez and Gabriel Zucman. Like the previous work of members of this team, the data-processing effort is as ingenious as the conclusions are. Leonhardt, for example, sees it as proof that something is seriously wrong with the way members of our society are rewarded. He writes:
In recent decades, by contrast, only very well-off families – those in roughly the top 1 / 40th of the income distribution – have received such large increases. Yes, the upper middle class has done better than the middle class or the poor, but the huge gaps are between the super rich and everyone else.
The fundamental problem is that most families used to receive something that approached their fair share of economic growth, and this is no longer the case.
Matt Bruenig is broken down the same data on capital income compared to labor income
For him, the conclusion was clear:
This data underscores the point made in Monday’s featured post: if we care about inequalities, we have to face capital.
Capital income is different
This last graph got me thinking.
Income from capital is a very different beast from income from labor. In many cases, it is not clear that he goes to anyone, at least not in the same way that a weekly paycheck goes to most workers. For example, the three peaks after 1995 in the Breunig chart bear a striking resemblance to the performance of the S&P 500 stock index over the same period. Yet high stock prices should be more properly viewed as a fall in the return on capital rather than an increase.
I also had chicane with Piketty in the past on how to think and analyze capital and the distribution of capital income seemed to play a key role in generating these provocative conclusions. I needed to have a clearer picture of what was going on below the surface.
Fortunately, Zucman is personal website has a host of resources on the work he’s done with Piketty and Saez. After going through two working papers, four paper-length data appendices, several huge Excel files, and a PowerPoint presentation, the picture became a bit clearer.
Measuring the distribution of income, especially over time, is trickier than you might think. Basically, you need to know exactly how much income each individual in the economy has received each year, for decades. In practice, you want enough information about enough individuals over time to statistically model the distribution with a high level of confidence. The richest source of income information that individuals receive comes from IRS tax return records.
Indeed, Piketty and Saez based their pioneering work on income inequality, on estimates they obtained from a sample of personal income tax returns. The problem is that a significant portion of capital income does not show up on personal income tax returns. Capital income is the return generated by investments in the economy. Yet most people do not invest directly. Instead, investments are made on their behalf by corporations, insurance companies, pension plans, and not-for-profit foundations. The biggest direct investment most people have is their family home, which does not provide them with direct income, but with housing.
The following graph by Picketty, Saez and Zucman (PSZ) shows the magnitude of the gap between capital income reported on personal income tax returns and total capital income in the economy.
Of all capital income in the United States, only the portion represented by the darkest series at the bottom of the graph appears on tax returns. In recent years, it has accounted for just under half of total capital income. As you can also see, it has also been relatively flat in recent years. If anything, there is a slight secular downtrend. So what is going on?
The rise of defined contribution plans
Well, what has increased sharply since 1980 is the income from capital that goes to pensions. This is the result of both the expansion of defined contribution pension plans such as 401 (k) s and the rapid rise in stock prices over the past 30 years. The question is how to account for capital income earmarked for pensions in the income distribution. This is not a conceptually or empirically straightforward task and in order to do so, Piketty, Saez, and Zucman have to make a series of heroic assumptions. I don’t blame them for it, but it is perhaps the central point of correctness and at the minimum of our interpretation of the growth of equality.
In particular, the authors first assume that 60 percent of defined contribution pension plans richness belongs to current retirees and 40 percent is owned by current workers in proportion to their income. It is this last part that is potentially important.
What makes defined contribution pensions attractive, and what has largely contributed to their growing importance in capital income, is the composition of returns. The savings you make early in your life will increase exponentially over time. Thus, one of the most important factors in determining the amount of a person’s defined benefit pension is the length of time they have contributed. However, the authors’ methods ignore this. Their analysis would assume that a corporate lawyer in the first year of law school would have a retirement wealth 3 or 4 times that of an office administrator who had contributed for 30 years.
This assumption may not be so bad. If you assume, for example, that some corporate lawyers have been contributing for a long time and others for a short time, and some office administrators are doing the same, it could all go away.
The problem is that wage inequalities have widened over time. Each year, the difference between the income of the office administrator and the corporate officer is higher than the previous year. And so, each year, the authors’ methods weighted the percentage of holdings ever to the lawyer, even though the actual holdings are determined by the historical investment.
This is made worse by the fact that the size of the total size of defined benefit pensions is increasing. The authors automatically attribute a growing share of an ever-growing share of national income to current high incomes, even if actual holdings are determined by historical contributions.
Rising stock prices are not the same as rising income
It all has to do with the way the authors calculate retirement wealth. Another aggravating problem is generated by the assumptions they make about income. They record capital gains retained by pension plans as income. In itself, this assumption is reasonable and accords with the way economists usually view income.
Yet when combined with the wealth distribution assumptions and current stock price trends, it has a perverse effect. Since the 1980s, stock prices have risen rapidly from historical standards. This means that the reinvested capital gains of pension funds have been particularly high. The growth rate is unlikely to be sustained and indeed it already appears to be slowing down.
This rapid appreciation in fact reduces the increase in inequality observed over the same period. This means that when high incomes have seen their wages increase, the effective price of saving for their retirement also increases. Yet the method of Piketty, Saez, and Zucman causes rising stock prices to worsen their measure of inequality.
These three effects, the weighting of the wealth of defined contribution pensions by current salaries, the secular growth in the size of these pensions and the temporary acceleration of reinvestment from 1980 to 2000 interact with the increase in salary inequalities over the same period. to worsen Measuring the growth of inequalities by Piketty, Saez and Zucman.
These interactions, I believe, generate a lot of the hockey stick phenomenon. As wage inequalities increased, the top earners were automatically allocated an ever larger share of a growing share of national income, and this increase in the allocation itself was counted as an increase in their current income.
Remove these effects and, if I’m correct, the distribution of income growth would be more evenly distributed among the richest 10 percent. It’s a concern, but it’s a different kind of concern, maybe even more insidious. It is more of a creeping rent-seeking of the type that Richard Reeves talks about in his book Dream Hoarders, than the madness of capitalism.
Karl Smith is director of economic research at the Niskanen Center