The first version of the A-К Model had exogenous labor supply and proportional taxes. In the period right after the conference we teamed up with Jon Skinner to add variable labor supply, endogenous retirement, progressive taxes, and a lump-sum redistribution authority that could help us distinguish efficiency gains from intergenerational redistribution. This collaboration resulted in Auerbach, Kotlikoff, and Skinner (1983). In the midi 980s, Alan and I added investment incentives and quadratic costs of adjusting the capital stock. As a result, we had a model that we could use to see how intergenerational redistributions could be achieved via policy-induced asset market revaluations and how the stock market’s value would adjust in response to policy changes. We also added a social security system and, in a simplified version of the model, demographics, including the presence of children whose consumption entered in their parents’ utility functions when they were children. All of these features were included in our book Dynamic Fiscal Policy that we published in 1987 with Cambridge University Press.
After teaming up with Robert Hagemann and Guiseppe Nicoletti (who were at the OECD at the time) to study demographic change in four OECD countries (Auerbach, Hagermann, Kotlikoff, and Nicoletti 1989), Alan and I took a break from dynamic fiscal simulations in the early 1990s to work on generational accounting and other issues. But we always knew that the simulation model had lots more potential. I started using the model again a couple of years ago to study the degree to which generational accounting was approximating true intergenerational fiscal incidence (Fehrs and Kotlikoff 1996) and the privatization of social security (Kotlikoff 1996).
At that point it became clear that the biggest shortcoming in the model was its failure to address intragenerational equity issues. But it also seemed that Fullerton and Rogers (1993) had already addressed that issue. Their book, which contained simulations based on a dynamic life-cycle simulation model, seemed to deal with both intragenerational and intergenerational issues. The Fullerton-Rogers model also featured lots of goods (our model had just one) and lots of industries producing these goods.
In considering whether we’d be simply reinventing Fullerton and Rogers, I realized that, for all its attractive features, their model lacked one essential element — the ability to solve for the economy’s perfect foresight transition path; i.e., the Fullerton-Rogers model used the same myopic expectation assumption adopted by Summers back in 1980.