Dueling with the Visible Hand (GST is regressive!)

Over at TVHE (guess which blogs I've been reading lately) Matt argues that:

The main criticism I often hear about this is that GST is regressive. Now I used to spout that line as well, after all poor people have a lower marginal propensity to save then wealthy people, as a result they spend more of their income, and so more is taxed.

However, then I was told to think about it a different way. Over our lifecycle we should spend all our money, so that we are on the boundary of our budget constraint. As everyone spends all their income over their lifetime, GST must be a flat tax.

I read TVHE not only because Matt and his co-bloggers are smart but also because they have a near monopoly on interesting counter-arguments to my own beliefs. I enjoy what I read even when I disagree with it, because I'm always learning something.

Having said all this, I have to confess that I think Matt's got it plain wrong here. Surely the mere existence of inheritance taxes provides fairly sound proof of the fact that people don't spend all their income over time. And I'd imagine, although I'm open to being dissuaded by evidence to the contrary, that the wealthy tend to die with more than the poor do. What theory suggests about people's consumption patterns and what they actually do are two different things altogether.

Yet, even if the wealthy really did not save more than the poor, GST would still be a regressive tax. Why? In three words: diminishing marginal utility. Which, in this context, is a fancy way of saying that money matters more to those who have less of it. If you give a person who earns a dollar a day a pay rise of a dollar a day it will improve their wellbeing significantly more than if you give the same pay rise to me. The idea of diminishing marginal utility is so strongly intuitive that it ought to be axiomatic, I think. Moreover, most of the recent happiness research confirms its existence. It is, in other words, a pretty safe bet.

And, as good (albeit troubled) utilitarians, we ought to think about tax in terms of its impact on people's wellbeing (utility) rather than on money. Money, after all, is but a means to an ends.

And, in the same sense that a pay rise matters more to someone in poverty than it does to someone who is well off, a rise in costs will also harm someone who is poor more than it well their wealthy compatriot.

For example:

Imagine an economy where no one saves and with only three goods: rice, cheese and caviar.

Rice alone is sufficient to fulfill our needs, but cheese tastes better and caviar better still*. So, where they can, people will replace rice with cheese and cheese with caviar. Upgrading from rice to cheese or cheese to caviar will make people happier (and so better off) but the most important thing to their wellbeing is that they consume calories sufficient to stave off hunger and ill-health.

Now imagine two people - Karl and Milton.

Karl earns $5 a week, Milton earns $15. Both need to eat 1kg of rice, cheese or caviar, or some combination of the three to avoid hunger. Rice costs $5 a kg, cheese $10 and caviar $15/kg.

So Karl is full if unfulfilled eating a kilo of rice a week. Meanwhile, Milton's styling on pure caviar.

But one day the Nolanists sweep to power in a landslide election victory, and they decide to impose a GST of 20%**.

Rice now costs $6 a week and Karl's health suffers as does his happiness, eaten away by gnawing hunger. Milton on the other hand shifts some of his caviar consumption to cheese. He's worse off, but not by nearly as much as Karl.

And this is why GST is regressive: it disproportionately impacts on the wellbeing of the least well off, even if the proportion of income effect is neutral.

_________________________________________________________
*Obviously, I'm stretching the truth here: in the real world caviar is repulsive.
**I'm ignoring the corresponding income tax cut here because I'm too lazy to do the maths, and because it doesn't actually change the picture if distributed evenly and if revenue neutral.