Learning to save in a voluntary pension system: toward an agent-based model

Mandatory pension systems partially replace old-age income, therefore the government matches additional life-cycle savings in a voluntary pension system. Though the individual saving decisions are apparently independent, the earmarked taxes (paid to finance the matching) connect them. Previous models either neglected the endogenous tax expenditures (e.g. Choi et al., in: Wise (ed) Perspectives in the economics of aging, University of Chicago Press, Chicago, pp 81–121, 2004) or assumed very sophisticated saving strategies (e.g. Fehr et al. in FinanzArchiv Pub Finance Anal 64:171–198, 2008). We create twin models: myopic workers learn (i) from farsighted workers using public information (analytic model) and (ii) also from each other (agent-based model). These models provide more realistic results on saving behavior and the impact of matching on the income redistribution than the earlier models.