The Steady-State Assumption and the Estimation of Distributional and Related Models

There is considerable interest in the estimation of linear structural equation models (including path analysis) of the evolution across generations of the distribution of income, of socioeconomic achievement, and of related variables. Given insufficient data, the estimation of the models has commonly proceeded by making assumptions which reduce the data requirements. In particular, it has been assumed (often implicitly) that the distribution of the variables being studied is in steady-state. In this paper we examine the role of this assumption and argue the need for care in its use.