S1E7: Not "operational"
A consensus without the Ramsey formula
This post is part of a series on the history of how economists model the future with the Ramsey formula, based on joint work with Pedro Garcia Duarte. See episode 1 , episode 2, episode 3, episode 4, episode 5, episode 6. Full Paper here
The 1970s saw intense discussion on theoretical consistency, tractability and ethics -with growing references to an optimality condition related to Ramsey – particularly regarding discount rates for energy models. Cost-benefit practitioners also refined their micro approaches, trying to reconcile social cost of capital and social discount rate approaches (think Bradford). Some, like Stanford economist Robert Lind, felt it was high time to nail down a consensus on how to set these rates.
After attending the CONAES meeting where Arrow presented his framework, Lind was tapped by the Electric Power Research Institute and Resources for the Future to host a full conference on “discounting for time and risk in energy policy.” This gathering brought together US and Europe-based heavyweights like Arrow, Feldstein, Stiglitz, Wilson, Dasgupta, Dorfman, Koopmans, Sen, etc. with the goal of providing “recommendations.” If the resulting collective book, published in 1982, gained substantial fame and influence, it was largely due to Lind’s introduction and survey chapters. Here, he laid bare the faultlines preventing economists from achieving a consensus, and proposed a practical procedure for selecting a discount rate. The conference cast a long shadow - the next major conference on discounting wouldn’t happen for another 2 decades. When it did, organizers Paul Portney and John Weyant noted in the introduction of the resulting book:
“it is fair to say that Lind’s primer came to assume great importance. This was due in large part to a compromise he proposed…for fifteen years or so after the publication of the book, when the questions ‘what discount rate should be used to calculate the present value of benefits and costs?” was posed, the standard answer was a brief: “See Lind.”
Lind argued that part of the disagreements over discounting “the distant future” stem from a messy tangle of debates on the social rate of time preference, private capital displacement, and risk. He built on ideas from Eckstein, Feldstein and Bradford suggesting to look at how public investments ripple through private investment streams, then convert these to consumption equivalents using a "shadow price of capital" (aka “the present value of the future stream of consumption associated with $1 of private investment discounted at the social rate of time preference”). The rationale for not discounting at the return on private capital was that it would unfairly penalize long-term projects like energy investments. Lind also pushed back against the then common practice of adjusting project discount rates for risk, citing recent empirical studies showing that the risk of energy investments was generally not correlated with the risk of a portfolio representing the whole economy. His proposed rates were in the 2-4% range, much lower than those used by US government institutions.
Despite the book's various nods to the Ramsey formula, Lind steered clear of it and generally avoided diving into the normative foundations of discounting. Indeed, Arrow, Stiglitz, Sen and Dasgupta’s chapters proposed theoretical frameworks that allowed “to infer the social discount rate from optimal policy.” Arrow opened the book by deepening the now well-known optimal growth framework developed with Kurz. Dasgupta likewise extended his analysis of resource depletion in an optimal growth model. He was perhaps the most articulate in distinguishing two equations associated with Ramsey. First, he clarified that the determination of the consumption rate of discount as the pure rate of time preference plus the elasticity of marginal utility times the consumption growth rate sprung from choosing a Ramseyesque social welfare function, Second, he explained that the "Ramsey rule" - which states that this consumption rate of discount should equal the interest rate for private investment - is an optimality condition in the growth model. Both Stiglitz and Sen’s contributions explored discounting rates tied to Ramsey’s work through the maximization of a social welfare function in a second-best world -with Sen proposing to move beyond such framework.
Whether rooted in welfare analysis, optimal growth, or both, Lind rejected inferring the discount rate from solving theoretical models for the optimal policy. He lamented that in such analyses, “in order to make the analysis tractable, we have to deal with a highly simplified economy,” for instance through assuming that the economy is in steady state. The Arrow-Kurz framework,” in his view, failed to provide an “operational basis” for choosing the discount rate in public investment policy. “For practical reasons,” he concluded, he had chosen an approach “more in the tradition of partial-equilibrium analysis than of general-equilibrium analysis.”
The disconnect between growing references to the determination of a discount rate in an optimal growth framework and the practical guidelines for discounting was a recurring theme in the 1970s and 1980s. Before energy took center stage, it had already played out in British-led public project evaluation guidelines for developing countries. Two competing reports appeared at the turn of the 1970s: one by James Mirrlees and Ian Little for the OECD, another by Sen, Dasgupta and Marglin for the UN. All of them had then contributed to welfare economics and/or the optimal growth literature, sometimes explicitly outlining the Ramsey formula. Yet, when it came to real-world government investment evaluation, for instance in India, debates circles back to familiar questions: should the discount rate reflect the social discount rate of consumption or the opportunity cost of private capital?
A decade on, Lind’s consensus was very much one forged by and for practitioners. Those insisting on working within an optimal growth framework were still the exception. Take Nordhaus, for instance, who was pursuing his quest to endogenize Co2 abatement pathways (see this previous post). In a 1980 Cowles working paper, he proposed to integrate a carbon cycle into an optimal growth model so as to map out “the shape of the optimal carbon control strategy over time.” His goal was to integrate the economic consequences of various Co2 control strategies to the intertemporal choice between consumption paths. This required making “value judgments” over how Co2 emission would be spread across time and generations, he explained, leading him into a deep dive on discounting. Without fully solving his model, he detailed how the “goods discount rate” was the combination of a “time discount” factor (how we weight different generations) and “growth discounting” (reflecting future generations' increased wealth):
“It is hard to defend a social rate of time preference above zero,” he continued, so that the goods discount rate, which he set around 7%, was mostly driven by the “high aversion to inequality” baked into growth discounting. He used his definition of the discount rate to explore various theoretical and empirical scenarios and concluded that such discount rate was key in determining the ultimate level of Co2 control and the control strategy over time (along with carbon cycle parameters, the damage function and growth rate) Though an outlier at the time, Nordhaus’s work foreshadowed the rise and stabilization of the Ramsey formula a decade later. It was one due to the irresistible rise of climate modeling in the 1990s.
Next, S1E8. “By popular demand” : the Ramsey formula comes to IPCC