Macroeconomic effects of reallocation shocks: evidence from the U.S.

Abstract

This thesis reexamines the Sectoral Shift Hypothesis. The latter was introduced by Lillien (1982). After his seminar work was emerged a huge amount of empirical research, although there is no unifying theoretical framework that has obtained consensus. There are several empirical studies that support SSH and many studies that do not. This dispute exists because there is empirical difficulty in separating reallocation unemployment from unemployment generated by aggregate shocks. This problem is called observational equivalence problem, the contribution of the present thesis is on the univariate and the multivariate methodology which are employed in order to examine SSH. In univariate methodology, a GARCH in mean model is applied on unemployment rate equation, after the purging of Lilien’s dispersion index from aggregate shocks which are potentional cause the observational equivalence problem. In multivariate methodology, the existing literature treats sectoral shocks in a symmetric way, meaning that unemployment rate will have the same amount of response in both cases of negative or positive sectoral shocks, this is an assumption which hides from us important information about the shocks. Using Hatemi (2013) asymmetric approach it’s feasible to decompose sectoral shocks in positive and negative shocks and form the cumulative term of shocks, after that they can be examined if there is any difference in unemployment’s rate variance decomposition, which it can be attributed to the sign of sectoral shocks.