Which inequalities matter and which taxes are appropriate?

Professor Piketty and his colleagues at the Top Income Distribution Study have put us all in great debt for the great increase in our knowledge of historical development of inequalities in income and in wealth in a number of leading countries.

Notice I have already mentioned two inequalities, income and wealth. There is one more leading inequality which does not receive much attention in Piketty’s work: consumption. Papers and books have already appeared which try to measure this inequality. Many more inequalities, e.g. health, educational achievement, race, and gender differences have been the subject of study, but these are more specialized and less central to economic analysis.

There is a strong argument for emphasizing consumption. Why, after all, do we consider inequality in wealth, income, or consumption to be undesirable? If we consider only economic arguments, it is because the poor are being deprived of goods that are valuable to their lives, exactly because they are more basic than the desires of the rich.

This has important implications for how we evaluate Piketty’s arguments about inequality. It suggests an alternative metric of inequality, one under which some of the problems that Piketty identifies are not, in fact, problematic.

Consider a world, like that envisioned by Piketty, in which the rich consume relatively little (compared with their property income). They accumulate wealth by investing in industry, thereby increasing output in the future. If they do not consume more in the future, but instead, simply continue to accumulate, then the additional future output is available for the consumption of the poor.

If, instead of being available to the poor, the additional output were somehow reinvested in the productive sector, we would find a world in which the ratio of investment to consumption is steadily rising. This is not the world we live in, and would produce visible results contrary to even casual observation.

In the neoclassical picture, consumption is the ultimate end of the economy. The rich accumulate for ultimate consumption, perhaps of generations in the far future, or, in some significant part, for philanthropy. Piketty seems instead to have a picture of the economy as a process of automatic accumulation, without regard to planned consumption. Estates grow at the market rate of return (100% saving out of property income). This is not a realistic account of how rich people – or indeed anybody – treats their income. It also leads us to ignore the politics of how this wealth is actually consumed.

Taking consumption seriously has important implications for measurement. If we are truly concerned with inequality, we should be most concerned with the distribution of consumption. The measurements we should look to are measurements of inequality in consumption, since it is differences in consumption that we really ought to care about.

This also has implications for policy: for example, if what we care about is differences in consumption, we might consider a progressive tax on total consumption of an individual. This would have to be done on an annual basis, like the current income tax, not at point of sale. Such a tax was long ago proposed by John Stuart Mill and later by Irving Fisher and Nicholas Kaldor. Piketty refers to Kaldor’s work but does little to refute it, saying only that no such tax exists. This is true, but of course the progressive wealth tax favored by Piketty is equally untried in practice.

We might be especially moved to consider a consumption tax if we consider that Piketty’s proposed wealth tax seems in any case to be much higher than it sounds. If we are to assume, say a 5% return on property, then a 2% per annum tax on wealth would amount to about 40% of property income. If investment is financed by property income, this implies a very considerable reduction in investment. Is this desirable? One might doubt it, especially since the effects on investment would be substantial, even apart from incentive effects, which might also be quite considerable.