Thursday, March 15, 2007

Personal Income and Capital Gains

When one attempts to measure inequality, one faces the problem that much of the income of the very rich is disguised as capital gains and the deeper problem that the variable of interest is income plus capital gains and not income or even income plus realised capital gains. At least one faces this problem if one is Emmanuel Saez

A standard approach to this problem is to report the distribution of income excluding capital gains and the distribution of income plus realised capital gains. I have a suggestion for a third calculation.

The problem with using income excluding capital gains is illustrated by the fact that total personal income is not equal to national income. National income (net national product) is the sum of personal income and retained earnings, corporate income which is not distributed as dividends. In theory retained (reinvested) profits should show up as capital gains to shareholders. In (simplistic) theory realised capital gains should be a moving average of capital gains obtained year by year. In practice, the stock market bounces up and down insanely and realisations are, in the short run, extremely sensitive to changes in the tax code.

I think that there is a simple way to assign retained earnings to individuals which is crude but might be useful. There are aggregate data on dividends paid and retained earnings giving an aggregate dividend payout ratio. The inverse is the ratio of corporate income to dividends.
There are individual data on dividends received. Why not just divide dividends received by the dividend payout ratio. This will give dividend income plus the corresponding share of retained earnings owned as a shareholder. If there were not systematic differences in the dividend payout ratio of firms whose shares are owned by high and low income people, this will work fine. There are such systematic differences of course, but I doubt that they are huge (basically I guess that almost all dividends which are paid to individuals are paid to individuals whose income reaches the top bracket).

This is a way to immunise estimates of the upper tail of the income distributions to changes in dividend payout without infecting the estimates with vulnerability to irrational exuberance.

Now another huge difference between true income and reported income is due to tax shelters in which overstated depreciation of structures hides true income and then when the structure is sold the hidden income appears as a capital gain. The amount of this activity fluctuated enormously. It was large. When accelerated depreciation was introduced in 1981 (effective 1982 on) it became huge, when tax rates were cut and the practice was specifically banned in 1986 (I think) the amount of old fashion tax sheltering declined sharply.

The period after tax reform shows a huge spike in reported personal income of the richest 1%
This is hard to correct. First for an estimate of aggregate income hidden I would look at total depreciation minus a guess of true depreciation. I would assign it proportional to capital gains (which is just a total silly rough rule of thumb cause I can't think of anything better).

I would guess that true inequality increased even more than inequality of reported personal income in the period 1979-1986 as increasingly massive amounts were hidden from the IRS n this period. Similarly, I would guess that the huge increase in the share of the top 1% from 86 through 89 is largely the effect of a reduction in such sheltering.


Anonymous said...

Comments simply do not work, but I will try again. I like this and left thoughts on DeLong's copy of your post.


Mark Thoma said...

Thanks Robert.

I need to check in here more often - time to grab your RSS feed...