Shift and share projection models: A reformulation: Fid-Bau Portal

Shift and share projection models: A reformulation

Floyd, Charles F.

Conclusions Tests of the simple projection models conducted in this study showed that a naive shift and share employment projection was superior to a model (Regional Rate) that depended only upon historical regional growth rates. The data confirmed, however, that there is enough instability in the regional share component to make a shift and share projection based upon a simple extrapolation of the component somewhat inferior to models (Ingrow and Super Ingrow) that utilize national historical and projected industry growth rates and omit any form of regional industry growth rate adjustment. The slight inferiority of the simple shift and share model tested does not mean that the technique holds no promise as a projection tool, but rather that the assumption of absolute stability of the regional share component over time requires modification. Even the very simple modification allowing for some convergence in regional industry growth rates (Modified Shift and Share) improved its predictive value to the point of superiority over the Ingrow model and to near equality with the Super Ingrow model. Further modification to substitute a multiplier projection assumption in local service industries and to take in account an inability to adequately project the regional share component for certain industries by setting the component equal to zero (Super Shift and Share) produced a model that was clearly superior.