Incorporating Heterogeneity in the 2TSF Model
摘要
This chapter introduces heterogeneity in the 2TSF model, in different ways. We develop a model where moments of a distribution depend on covariates and are estimated by maximum likelihood. We also exploit the “scaling property” to obtain a model without distributional assumptions that can be estimated by nonlinear least squares. The empirical application of the chapter compares the various approaches.