December 6, 2018

Three Approaches to the Social Discount Rate

A Mercatus Symposium

The three essays presented here are the product of a symposium hosted and edited by James Broughel. These essays illuminate the debate on the appropriate discount rate to be used in benefit-cost analysis of public programs and regulations. The selection of this rate has profound implications for policy, yet to date there is no consensus among economists as to what the proper social discount rate should be.

The two dominant approaches to estimating a social discount rate are known as the social rate of time preference method and the social opportunity cost of capital method. Mark Moore and Aidan Vining make the case for the former. Their clear presentation starts from the principle that public policy should maximize social welfare. To make this approach operational, they assume social welfare is a function of per capita consumption. The solution to the welfare maximization problem yields a social discount rate that corresponds with society’s marginal rate of time preference. This rate is the sum of two terms: the pure discount rate on future welfare and the product of the growth rate of per capita consumption adjusted by the elasticity of welfare to consumption. In this manner, they obtain an estimate in the vicinity of 3.5 percent. It should be noted that, instead of this flat rate, they recommend the use of a declining rate—the further into the future, the lower the discount rate used—to evaluate policies with intergenerational effects.

David Burgess makes the case for using the social opportunity cost approach to determine the social discount rate. The guiding principle of this approach is to produce a realistic approximation of the government’s opportunity cost of borrowing. The benefits a public program returns must be greater than the returns foregone from the private consumption and investment that government borrowing displaces. Thus, the social discount rate is estimated as a weighted average of three rates: the marginal rate of return on investment (the production rate of interest), that rate after corporate and personal taxes are applied (the consumption rate of interest), and in an open economy, the marginal rate of return of nondomestic investments. The weights here represent the proportions of incremental funding drawn from these different sources. Following this approach, Burgess arrives at a recommended rate of roughly 7 percent.

The last essay in this symposium, by James Broughel, offers a few genuinely original challenges to some of the assumptions in the two standard approaches, as elaborated in the preceding papers. He points, for instance, to the accounting problem in the social opportunity cost method with nonpecuniary benefits. Should those benefits be discounted at the same rate as benefits that can be reinvested, like cash? Likewise, Broughel notes that the social rate of time preference approach relies on a social welfare function that is unlikely to be representative of society’s actual preferences, and furthermore the approach imposes a contestable assumption: consumption is worth less the farther it occurs into the future. He proposes an alternative welfare function that weighs all periods into the future evenly. By adopting this sensible approach, he does away with the artifice of setting up a welfare function easily amenable to maximization. This shifts the attention of the analysis to other values, such as the rate at which benefits realize. Even those policy analysts disinclined to adopt Broughel’s zero-discount-rate proposal will benefit from giving it serious consideration as it forces a discussion of the assumptions built into the two standard methods.

This trio of papers is not only a useful primer of the social discount rate, but it also sets an agenda for further examining this important tool of policy analysis. Altogether, they are an invitation to reassess the assumptions in the standard discounting methods and perhaps even to reexamine the foundations of benefit-cost analysis itself. Some discounting assumptions are adopted in the interest of making an intractable problem soluble, some are inescapable value judgments, and yet some others seem to be a bridge too far. The analysts who are self-aware and forthright about the assumptions they make will deliver a more transparent, credible, and useful policy analysis.