Moment estimation in Auerbach-Kotlikoff models: How well do they match the data?

Despite their widespread use for the analysis of economic questions, a formal and systematic calibration methodology has not yet been developed for Auerbach-Kotlikoff (Auerbach and Kotlikoff 1987) overlapping generations (AK-OLG) models. Calibration as estimation in macroeconomics involves choosing free parameters by matching moments of simulated models with those of the data. This paper maps this approach into the framework of AK-OLG models. The paper further evaluates the back-fitting properties of three different versions of a prototype AK-OLG model along a number of dimensions of mostly US data for the time period 1960-2003.