Sunday, 28 June 2009

How much is the future worth?

As promised, a more detailed to Chris Dillow's posting Is behavioural economics wrong? Analysed in comparison with Age and time discounting from the Geary behavioural blog.

The key factor to note in the research Chris posts (by Laurie Pounder) is that it is based on a very different age cohort from most behavioural experiments. The subjects were aged between 53 and 73 because the author wanted to better understand the behaviour of those who contribute the majority of savings in the US economy - which happens to be that age group.

As Chris mentions, the results show that people over-save compared to the theoretical prediction of the permanent income hypothesis (PIH). This hypothesis predicts that rational people will smooth their consumption over time according to their expected lifetime income - borrowing if their income is lower than the lifetime average, or saving if it's higher. The results, however, indicate that people save more and consume less than the theory predicts, and therefore end up with more savings than they need.

Most previous behavioural research, on the other hand, indicates the reverse. It shows that people save less on average than theory predicts. Chris wonders therefore whether the basic data behind behavioural economics may be wrong.

My responses to this are:
  1. Behavioural economics covers a lot more than savings behaviour, but I'm sure Chris was just being a little bit provocative with the title of his article. I will focus in this response only on the savings-related aspects of the field.
  2. It's possible for people to behave irrationally in different ways at different times. My interpretation of the results of this research is that even if they have saved too little early in life, people in their 50s and 60s not only save enough to make up for it, but actually over-compensate. This is "irrational" according to the PIH, but plausible in a psychological model where people are driven by availability heuristics and not just by analytic calculation. If I know I don't have enough saved up, then in conditions of uncertainty I may well save more than I really need to, in some kind of emotional compensation mechanism.
  3. One of the key insights of (say) Thaler and Benartzi's work on saving is not that people do not know how much to save, but that they have a self-control problem in getting started. A lot of behavioural research is in areas like this. In this view, people can be thought of as having a rational 'core' - their underlying utility functions do look like those of rational agents - but their short-run behaviour systematically departs from what is predicted by those utility functions.

    Possibly the major debate between behavioural and traditional economists is about how much this matters. Classical economists tend to say that the errors* will average out or correct themselves, and that they don't matter much for economic theory. Behavioural economists tend to believe that the errors do make a real difference - for example in creating asset bubbles, or reducing average saving over a lifetime to below the utility-maximising level - or in the case of these results, increasing average saving beyond the optimal level. Of course, over a whole lifetime, it can't be both! But it is entirely plausible for savings behaviour to depart from optimality in two different directions at different times, reducing total utility in both cases.
Interesting to compare this study with Time discounting over the lifespan (Daniel Read and N.L. Read - public version available at the link given) via Kevin Denny at the Geary blog, which examines time discounting at different ages. In other words, how much do people privilege present gains over future gains? The results find that:
older people discount more than younger ones, and that middle aged people discount less than either group
This would support the idea that young people do not save enough, the middle aged save more, and the old least of all. Unfortunately the age groups (mean ages 25, 44 and 75) are not compatible with the 53-73 range tested by Pounder, so it's hard to compare the results. But it is intuitively reasonable to imagine a model of under-saving up to 40, over-saving from 40 to 65, and drawing down of savings after retirement - which would fit both results.

Denny suggests a reasonable explanation for greater discounting by old people - that they are less likely to be around later and therefore it is rational to spend more money now. This doesn't however explain the difference between young and middle-aged people - though Read and Read's paper does suggest some possible rationales, among which are:
  • evolutionary fitness creates a preference to consume resources in youth when fertility is highest
  • it takes time to learn that the future exists and to accurately estimate the likelihood of being able to draw on saved resources, therefore young people are more biased towards immediate gratification
Read & Read's paper is an exemplar of good behavioural economics - it does not just throw out random departures from classical rationality, but measures those departures and proposes different explanations for them, allowing room for those be extended into a richer decision making model. One could certainly argue that it is in fact rational to make decisions in these richer ways, and I think ultimately the gap between behavioural and "normal" economists will be closed in this way. The idea of plain "irrationality" makes good journalism but bad economics.

Most of the popular books on behavioural economics do not take this step of rationalising the behaviour they describe - which may be why they are popular, of course. Looking at the world with an economic mindset can be a harsh experience, and offering an escape valve from this worldview is probably a good way to make people feel better. But it is not very helpful in understanding society better.

* By errors I refer to errors in the theory's prediction of human behaviour; I do not mean to suggest that individuals are making errors when they depart from the rules given by economists.

1 comment:

PunditusMaximus said...

Part of the problem also is that we count purchase of education and investment in childrearing as exclusively consumption, rather than a combination of consumption and savings.

How can we possibly understand the issue when we don't bother to measure the variables correctly? When I started grad school, I thought economics was a science. Now I know it's the battleground between scientists and rich men.