We argue that if benefit-cost analysts are charged with assessing whether aggregate “wealth” is increased (as implied by the Kaldor-Hicks potential compensation test) it would require discounting individual citizen’s future benefits and costs by the citizen’s own time preference discount factor.  Yet, the social discount rate used in standard benefit-cost analysis ignores heterogeneity in time preferences.  Given vast heterogeneity in time discounting across citizens then produces substantially different discounting of projects with future benefits and costs than found using a standard social discount rate. We note that such heterogeneity produces discounting that declines as the time horizon for the project lengthens (similar to Weitzman (2001), but for very different reasons).  We then discuss the practical and theoretical challenges that follow from incorporating individual discount factors, including the prospect for ex post policy regret.

Using individual discounting brings on substantial practical challenges of identifying the joint distribution of project benefits/costs and time discount factors, particularly if standing is given to citizens who have not yet been born (or who are too young to elicit such valuations).  Moreover, as we demonstrate in this paper, incorporating such individual discount rates potentially puts the benefit-cost analyst in the undesirable position of recommending a policy be adopted that the analyst knows will be regretted by society in the future if either (a) citizens have hyperbolic discounting of future values, (b) if some citizens benefit in a particular year while other citizens face costs and actual compensation is not included in the policy design in the future year, or (c) if the effect of individual discounting is to result in a social discount rate (which we term the “Equity Rate”) which is below the social opportunity cost of capital.  In this case the use of the Equity Rate would be inappropriate. The last of these problems can be solved by never using a social discount rate that is less than the opportunity cost of capital, regardless of individual time preferences.  

Even if it is not feasible or desirable to fully implement individual discounting, we argue that analysts should take note of the results of our social discount rate (which we term the Equity Rate) that asymptotes towards the time discount rate of the citizen with the lowest time preference rate as t goes to infinity.  While Weitzman (2001) recommended a declining social discount rate as a result of uncertainty amongst expert analysts regarding the “correct” discount rate, our Equity Rate declines due to heterogeneity across citizens.  The combination of our analysis and Weitzman’s results should give more substantiation for the argument that social discount rates should be declining as the time horizon lengthens.  Combining these insights, we recommend a social discount rate that is declining in time, and is equal to the maximum of the declining Equity Rate and the social opportunity cost of capital (which may be declining in time due to uncertainty across experts in the correct value of this opportunity cost).