How Should We Value Future Utility?

As a society, we have to make tradeoffs between future and present consumption. When we assess the costs and benefits of public infrastructure, action on climate change, or changes in the tax code, we discount future consumption by a compounding interest rate. This rate can be broken down into a few parts,

r = delta + ng

where r is the real interest rate, delta is the rate of time preference over utility, g is the growth rate, and n is the elasticity of marginal utility. People have declining marginal utility from consumption, and so we are less benefitted by positive shocks than we are hurt by negative shocks of the same size. N is better known as the coefficient of relative risk aversion. Delta refers specifically to how much more important having your utility now, rather than later, is.

What I will argue is that the rate of time of time preference over utility should be zero, and that there being a positive rate of time preference is either nonsensical, actually just due to risk, or due to externalities. It is nonsensical insofar as it is already adjusted into the thing we care about – utility. If it were consumption, then smoothing consumption makes sense because of declining marginal utility. But it is literally impossible to have declining marginal utility. It is, by definition, adjusted. Each unit adds the same amount.

Individuals care about having their utility now, rather than later, but this isn’t due to an idealized preference for now over later which would exist even in a riskless world with immortal agents. People are not able to receive utility. Rather, they can receive consumption which correlates to utility. People’s circumstances can change – they might even die. This should already be included, however, in the coefficient of relative risk-aversion. It isn’t actually a preference for delaying consumption, just a heuristic.

The preference for utility now rather than later is partly an externality. How would we feel if there were two countries, one of whom has financial markets and will cause climate change, and the other one which has no access to financial markets and will be harmed by climate change, and found the optimal level of climate action based only on the opinions of one country? It is not an enormous normative leap to say that this is bad, and we should take into account the preferences of everyone. Why should it matter if it takes place across time? As a moral matter, we should treat humans in the future the same as humans in a country which we cannot observe. As it would be obviously inappropriate for us to send fire missiles into a country which could not negotiate with us, so too would it be morally wrong for us to leave unpleasant presents for the future.

We do not take this fully into account. People do have some degree of care for their children, but there are many people who do not fully internalize the welfare of future descendants. Why should we expect people without children to be altruistic?

How we treat future welfare is incredibly relevant to our study of how much we should do about climate change. The optimal actions now are incredibly dependent on our choice of discount rate. Martin Weitzman illustrates this in his excellent review of the Stern Report. Conventional estimates use a real interest rate of 6%, with roughly 2% each for delta, n, and g. These were inferred from people’s actual behavior in financial markets. Meanwhile, the Stern report argued for delta equalling zero, took a lowball n at 1, and a low growth estimate of 1.4%. Comparing now and 100 years hence, the benefits of climate change action in the second case are 100 times higher than in the second.

My criticism is a sort of midpoint. I don’t think that market behavior gives us the correct estimate of the discount rate. Part of this is ethical, as I believe that we oughtn’t treat people differently over time for its own sake. Part of this, however, is that some of what we regard as time preference is actually risk-aversion, and this should be included in the second part of the equation. The observed selfishness of humans now is no reason for it to be right.

P.s. While reading in the process of writing this article, I came across some remarks from Deborah Lucas on finding the optimal discount rate from market behavior. What I found most striking was point 3.1, which argues that we cannot infer the cost of capital from government borrowing. The low cost of borrowing proceeds from an implicit guarantee that the government will always raise taxes, rather than default – thus, using the risk-free rate implicit in treasury bonds will systematically understate people’s preference for consumption now.