By Yongsung Chang, Sun-Bin Kim and Frank Schorfheide
This paper assesses biases in policy predictions due to the lack of invariance of “structural” parameters in representative-agent models. We simulate data under various fiscal policy regimes from a heterogeneous-agents economy with incomplete asset markets and indivisible labor supply. Imperfect aggregation manifests itself through preference shocks in the estimated representative-agent model. Preference and technology parameter estimates are not invariant with respect to policy changes. As a result, the bias in the representative-agent model’s policy predictions is large compared to the length of predictive intervals that reflect parameter uncertainty.
When structural models are estimated, it is assumed that those structural parameters are policy invariant, or the Lucas Critique would be applicable. That is going to be a problem especially if aggregate data is used to estimate the model, as composition effects may bias estimates. This paper simulates data under various policies and finds a measurable bias is present. Should we thus stop using aggregate data to estimate these models?