Abstract: | Traditional univariate shift-share studies of employment provide an unreliable indicator of the relative performance of a region or an industry for they fail to separate the effects of output and productivity change on the demand for labor. An extended shift-share model is proposed that overcomes this weakness and permits identification of different processes of regional development. This model is used to investigate annual employment change in twenty (two-digit SIC) manufacturing industries in nine census regions of the U.S. between 1950 and 1986. The timing and depth of the exodus of manufacturing jobs from the snowbelt to the sunbelt is illustrated along with the business cycle performance of industries and regions. Productivity growth in the sunbelt is positively associated with rapid output expansion, whereas in the snowbelt it is associated with the loss of market share and economic rationalization. |