Journal Article
No. 2009-26 | June 10, 2009
Håkon Sælen, Simon Dietz, Cameron Hepburn, Jennifer Helgeson and Giles Atkinson
Siblings, Not Triplets: Social Preferences for Risk, Inequality and Time in Discounting Climate Change
(Published in Discounting the Long-Run Future and Sustainable Development)


Arguments about the appropriate discount rate often start by assuming a Utilitarian social welfare function with isoelastic utility, in which the consumption discount rate is a function of the (constant) elasticity of marginal utility along with the (much discussed) utility discount rate. In this model, the elasticity of marginal utility simultaneously reflects preferences for intertemporal substitution, aversion to risk, and aversion to (spatial) inequality. While these three concepts are necessarily identical in the standard model, this need not be so: well-known models already enable risk to be separated from intertemporal substitution. Separating the three concepts might have important implications for the appropriate discount rate, and hence also for long-term policy. This paper investigates these issues in the context of climate-change economics, by surveying the attitudes of over 3000 people to risk, income inequality over space and income inequality over time. The results suggest that individuals do not see the three concepts as identical, and indeed that preferences over risk, inequality and time are only weakly correlated. As such, relying on empirical evidence of risk or inequality preferences may not necessarily be an appropriate guide to specifying the elasticity of intertemporal substitution.

Data Set

JEL Classification:

C90, D01, D63, Q51


Cite As

Håkon Sælen, Simon Dietz, Cameron Hepburn, Jennifer Helgeson, and Giles Atkinson (2009). Siblings, Not Triplets: Social Preferences for Risk, Inequality and Time in Discounting Climate Change. Economics: The Open-Access, Open-Assessment E-Journal, 3 (2009-26): 1–28.

Comments and Questions

Anonymous - miscellanous comments
August 06, 2009 - 16:26 | Author's Homepage
I profited very much from this article and I think it is extremely important, both in terms of an eye-opener regarding the theoretical assumptions behind the standard model and in terms of its practical applications in climate economics. Here are some miscellaneous thoughts... ----- In terms of further research, the one thing that I'd be interested the most is to see whether people treat the trade-off between a unit of utility now and a unit of utility later differently depending on whether "later" means "within my own lifetime" or whether "later" means "after my lifetime". Intuitively, I think the trade-off might be approached in a very different way (for example, within my own lifetime I might maximize a sum of discounted utility whereas this might not be the way I treat the utility of my descendants. As far as their utility is concerned, I might be more interested in guaranteeing a certain level of utility - say, equally much as I have or sufficiently much - rather than weighing it up with my own utility).Also, in case people should make a categorical difference between their own lifetime and their descendants' lifetime, I'd be very interested to know HOW they conceive of this difference. And how could we model the difference (if there actually is a difference)? And how can we model the difference for society as a WHOLE because for society as a whole (in contrast to the individual perspective) there is a GRADUAL transformation from "one's one lifetime" (i.e. "the present generation") into the lifetime of our descendants (i.e. "the future generations")? Do bequest models represent the preferences for how society as a whole bequeathes something to its descendants? ----- Regarding the question of whether eta is an ethical parameter, I am critical of the first paragraph of section 3.1 where an "intermediate" approach is suggested. I personally think it is more correct to say, that for some purposes, determining the value of eta is an empirical question and for other purposes, determining the value of eta is an ethical question (my own view can be found on p. 21-22 of this paper:; it's about the discount rate but it can be analogously applied to eta (written for an audience of moral philosophers)) ----- I think it is problematic to tell respondents to the survey that some of the policies can be thought of in a way similar to how they make decisions about how much to spend and how much to save (p. 9) and also to tell them that there is no welfare state and to exclude the poorest and richest from the questions about inequality aversion (p. 9). I am not claiming that this is problematic in the general sense that is always problematic to a certain degree to make simplifications. I think it is problematic because it introduces a bias towards making people wear their "expected utility hat" which is not the hat they always wear. ----- Your conclusion from the paper is that people operate according to alternative mental models (especiall in section 5.2). This is the most important message I personally draw from this paper: Atttitudes are guided by a preference structure which cannot be captured within a simple and standard expected utility model. Choices are framed differently than standardly assumed and this is the reason why eta diverges.I would go even further and claim that it is not only impossible to capture risk aversion and spatial and temporal inequality aversion within one parameter, but that it is even impossible to capture risk aversion within one parameter, spatial inequality aversion within one parameter and temporal inequality aversion within one parameter. I am not complaining that summarizing our attitudes towards, say, spatial inequality within one parameter is a simplification (one ALWAYS has to simplify) but that it is TOO much of a simplification.Of course, there is a lot of research done on how well real-world behaviour can be captured by an expected utility maximization framework. But I just want to cite some observations from everyday-life which make it plausible that not only is it impossible to summarize the three aversions within one eta parameter but that it is also questionable to summarize any one of the three aversions taken for themselves within one parameter:--- Many people think that they have no duty to equalize incomes through giving away their own income to poorer people. These same people, however, think it is very important that the state does exactly this through the tax system. So, it's difficult to say what their eta is. It depends on whether equality is furthered through collective or individual action.--- Many people who think equality is something important favour MORE unequal states of affairs once these states of affairs are characterized in more detail. Let's assume we have two persons, R and P (i.e. rich and poor). If R is rich because he worked while P was lazy or if R is rich because his mother was very attached to him and bequeathed her fortune to him, then many people who value equality would prefer an unequal distribution of goods between R and P above an equal distribution of income. So, desert and similar considerations are important: people evaluate distributions very differently depending on how they came about.Contrary to what is said in section 5.2.3, I think it is very questionable to capture an emphasis on desert via eta (in the sense that an emphasis on deserts implies a low eta). Many people think that people in sweat shops DESERVE a higher wage which would look like a display of a high eta in a survey. Neither does a concern with needs necessarily imply a high eta: many people are very concerned with erasing poverty/need but are hardly concerned with creating equality. So questions of desert, need, partiality towards the near&dear are sometimes very independent of an underlying preference for equality. Also, it seems very questionable whether the logic of Rawls' difference principle makes it possible to be interpreted as an infinite eta. (It should also be noted that considerations of desert etc. seem more important in spatial inequality than in temporal inequality).-- Many people might be more risk-averse when IMPOSING a risk than when TAKING a risk. If I had to manage the finances of a friend who fell into a coma and who will possibly wake up again, I'd follow a more conservative investment strategy than when managing my own finances. Similarly with future generations: we might be more risk-averse with regard to their consumption than we are regarding our own consumption.Also, ambiguity aversion messes up measuring our attitudes towards risk in such cases as climate policy where many of the risks are very hard to ascertain.-- From discussing the discount rate, I know how some people consider it strange to frame climate policy as a savings/investment decision. To them, it is a new (and sometimes repulsive) perspective to consider mitigation as a kind of investment whose returns can be compared to the returns of other investments. Especially, environmentally-minded people often separate protecting the environment from other costly measures which benefit the future. So, measuring their temporal inequality aversion with one parameter might be messy.-- The expected utility model assumes that people decide on consuming or saving depending on what yields higher overall utility. But people might make this decision within a completely different framework: They might place a very large value on guaranteeing a certain level of utility to posterity (regardless of whether marginal utility of consumption at that level is equalized across time) and they might place an extremely low value on exceeding this level of utility for posterity (even if the marginal loss of utility from further savings would be very small today and the marginal gain of utility to future generations from further savings would be very large).in any case, I think we are all familiar with such examples. The bottom line is: Expected utility theory traces risk aversion, spatial and temporal inequality aversion COMPLETELY down to diminishing marginal utility. A more realistic behavioral model would argue that people have 9 reasons for being risk-averse in 4 different senses, 11 reasons for being spatially inequality-averse in 5 different senses and 7 reasons for being temporally inequality averse in 3 different senses. If we try to press these 27 reasons within one number - or even if we try to press them into THREE different numbers - we will obviously run into problems.I think we all agree that the Expected Utility Model is good for many things. I think we also all agree that it is a simplification. The big question is: is it TOO much of a simplification for a given purpose? For purposes of capturing inequality and risk aversion in climate policy, my answer is clearly "yes" (for other purposes, it is "no"). I think this is the most important conclusion from this great paper. And so I think the really important research question is not what kind of eta (and delta) people want to apply WITHIN an expected utility framework but the important question is what kind of FRAMEWORK people want to apply to questions of climate policy. ----- I wonder what we would find out if we elicited preferences within a framework where people first have to set a given a goal for environmental and other future-oriented policies (where this goal is then to be achieved in the most economically efficient way in a second step). How people set this goal represents their attitude towards exposing future generations to risk. In my view, separating the preferences about risk and time in setting such a goal from preferences about risk and time for effects appearing within one's own lifetime, might make the attitudes of people look more consistent.Note also that such a framework (a cost-effectiveness framework) is a very easy and tractable alternative to the utilitarian idea of maximizing the sum of utilities through time and states of nature. ----- Small comment regarding the second paragraph of 5.1: You say that the survey might not represent people's true attitudes because some questions were cognitively too demanding and people therefore relied on simplifying heuristics. An obvious question is whether these simplifying heuristics aren't what people apply in reality, too, to such complex choices as climate policy. ----- BTW, a small other issue I would find interesting: Comparing the average climate economists' estimate of how much "consumption" (or also utility, in case utility is implicitly or explicitly given a cardinal interpretation in terms of well-being) grows over the next century to the average citizens' guess of how much consumption or well-being grows over the next century.That might also reveal something important about people's attitude towards climate policy. In my experience, most of the outsiders are surprised to hear about the economists' growth assumptions. Also, in my view damage estimates from climate change are estimated to be higher by the man on the Clapham Omnibus than by the economist. This might also account for some of the surprising attitudes of people.