Estimating life-cycle income processes including means-tested transfers

This paper undertakes a Monte Carlo experiment to evaluate whether the canonical “persistent-transitory” model of income can effectively measure means-tested government transfer policies. First, I show that means-tested transfers (modelled as an income floor) are not well represented by a stationary stochastic process for income with the persistent-transitory structure, since means testing generates age dependence in the autocovariance structure of income. Second, I quantify how this affects inference in a standard life-cycle model with incomplete financial markets. The structural model reveals how including transfers in an exogenous (persistent-transitory) income process affects both consumption and savings decisions, and the estimation of key structural parameters (compared to a model with an income floor). This exercise demonstrates that including transfers implicitly in the income process generally leads to biased estimates of key structural parameters (i.e., the discount factor and coefficient of risk aversion), relative to explicitly accounting for transfers in the model.