EQUIVALENCE OF THE STANDARD AND THE MODIFIED SWITCHING REGRESSION-MODELS

Ophem (1993) employed a switching regression model with earnings enter ing the choice equation to investigate earnings differentials between the public and private sectors. Ophem also described a ''modified swit ching regression model'' in which only the equation for unknown earnin gs is substituted into the choice equation. He argued the coefficients in the choice equation can be identified in the modified switching re gression model without exclusion restrictions and estimates from the m odified switching regression model are more efficient than estimates f rom the standard switching regression model. We show that there are fl aws in Ophem's analysis which invalidate his claims.