Rethinking Redistribution

One of the oft cited purposes of progressive income tax systems, like the Canadian tax system, is to redistribute wealth. Assuming the wealth person doesn’t enjoy giving up their money generally means making value judgements about the relative importance of different people’s happiness. The judgement implicit in the re-distribution is that the wealthy person matters less to us than the poor person. They are made less happy than they were and the poor person is made happier than they were. At this level we can only call this an improvement by saying the happiness of the poor person is more important to us. If the wealthy person received sufficient wellbeing from equality to make this worth their while, wouldn’t they have made the transfer voluntarily?

Personally, I find making these sorts of judgement uncomfortable. What, short of a divine right, would qualify a person for making those kinds of judgements? What is the basis for deciding between people? Doesn’t this on some level dehumanize those paying taxes and somehow “uberhumanize” those receiving government payments? Isn’t such dehumanization the basis for all kinds of evil behaviours?

One response to this line of questioning that is often presented in economics is an appeal to diminishing marginal utility of income. The argument is that the extra happiness generated by another dollar of income falls as a person has more money. Therefore, if we take a dollar away from a person with high income, the loss in happiness is likely to be small. If we give that same dollar to someone with a low income, the happiness gained is likely to be large (or at least larger). The result is an increase in net happiness as the gains exceed the losses. While this ignores all the problems with trying to compare utility between individuals at least it is has the virtue of treating the two human beings involved as equally important.

As far as this goes, it isn’t a bad argument. A fly appears in the ointment when we start to think carefully about how people respond to gains and losses, however. Way back in 1979 two psychology researchers (Kahneman and Tversky) noticed that people reacted differently to hypothetical gains and hypothetical losses, a difference that has been replicated in a variety of economic experiments using real incentives. I’ve seen a difference in my own research. Most estimates estimate the relative cost of a loss to be twice the benefit of a gain. The sensitivity of a subject to a loss of $5 is the same as the sensitivity to a gain of $10.

This puts a slightly different spin on income redistribution. Taking one dollar from someone and giving it to an identical person can actually reduce the simple sum of welfare. In order for redistribution to be welfare improving on the basis of a simple sum of utilities, the marginal benefit of an extra dollar of a tax payer must be less than half the marginal benefit of an extra dollar of someone receiving transfers. While this might be the case, it isn’t a certainty.

We might want to rethink income redistribution – it might actually be welfare reducing