What is a really feasible discount rate? The issue of choosing a discount rate, for example in models predicting costs of climate change, is not easy. There is also a risk that one may choose a discount rate that is too low, which would cause us to instead overestimate the costs. Pondering these questions easily makes one think of an accessible paper by William Nordhaus written in 2007 in which he argues that the Stern Review on the Economics of Climate Change uses a discount rate that is too low. Nordhaus claims that the Stern Review’s estimate of overall costs of climate change being equal to the world losing 5% of global GDP each year from now on is dependent on a near-zero discount rate and a very specific utility function.
In this debate there are several relevant perspectives. One of them is economic. In some cases we might actually underestimate the costs, for example when it comes to permanent losses of biodiversity. On the other hand, in other cases we risk overestimating costs, often due to undervalue the rate of technological and economic progress.
The issue of discounting does also have a clear philosophical dimensions. How much are future generations worth? As long as there is an existing risk of extinction of our species there should probably be some form of discount rate due to this uncertainty. But how large should this discount rate be? What is appropriate? How much is the generation of our children worth? And what about their children?
One could also raise the inconvenient question of whether it would be easier for future generations to pay for the harm we cause the environment simply because they likely will be much wealthier than people living today. This argument is one that was discussed by The Economist in a summary of the “Stern-Nordhaus debate.”
A little more recently, Larry Karp has done some interesting research on discounting, and he finds that discounting over time is not constant. This is usually called “hyperbolic discounting,” which implies that our discounting preferences are dependent on the time frame, and generally we are willing to spend almost as much on our great great grandchildren’s generation as on our great grandchildren’s generation because it is so far into the future. This would imply a non-linear relationship over time, as opposed to the linear and time-consistent relationship that is usually assumed. This is an example of how experimental evidence and psychological insights, pioneered by Daniel Kahneman and others, can contribute to economic theory. Hyperbolic discounting is likely to be a topic paid much attention over the coming years.
And finally, two other recommended readings on the choice of discount rate are a blog post by Gary Becker, and a more recent working paper by Lawrence H. Goulder and Roberton C. Williams III.