| |
A technique for assessing the relative importance of different components in regional employment growth or decline. Regional employment growth may be due to the region having a high concentration of industries that are growing nationally (such as micro-electronics), or it may be due to locational shifts within industries or differential regional trends within an industry. Shift-share models try to disentangle and measure these effects using simple algebraic manipulation of growth rates.
Let Eij0 be the level of employment in industry i in region j in the initial time period 0, and let Eijt be the level in the next period t. Then the total employment in region j at time 0 is
{img src=show_image.php?name=bkhumgeofm24.gif }
total national employment in industry i is
{img src=show_image.php?name=bkhumgeofm25.gif }
and total national employment is
{img src=show_image.php?name=bkhumgeofm26.gif }
The figures for time t are then obtained in the same way. The total shift (TS) is then the difference between actual regional employment growth and that which would have occurred if the region had grown at the overall national rate:
{img src=show_image.php?name=bkhumgeofm27.gif }
This total shift can then be divided into a proportionality shift (or \'composition effect\') and a differential shift. The proportionality shift (PS) measures change due to regional concentration in slow- or fast-growing sectors, and is calculated by applying to each industry a growth factor that is the difference between the actual industrial growth rate and the overall national rate. For industry i, this factor is:
{img src=show_image.php?name=bkhumgeofm28.gif }
and the proportionality shift for region j is {img src=show_image.php?name=bkhumgeofm29.gif }. The differential shift (DS) is the shift due to locational changes within industries and is obtained as DS = TS — PS.
The technique is purely descriptive, and does not explain why certain sectors are growing or declining or why locational shifts are taking place, but the data are easily obtained, and it gives a useful starting point for further inquiry. The method is very dependent on the level of sector aggregation (more sectors produces lower differential shifts), and it takes no account of linkages or multipliers. Limitations of the technique, especially the way it overlooks the need to standardize industry growth rates for their regional mix effects, have led to the construction of \'multifactor partitioning\' by Ray (1990), which overcomes this problem and also identifies interaction effects and an allocation effect. These extensions complicate the formulae but can readily be programmed. (LWH)
Reference Ray, M. 1990: Standardising employment growth rates of foreign multinationals and domestic firms in Canada: from shift-share to multifactor partitioning. Multinational Enterprises Programme: Working Paper No. 62. Geneva: International Labour Office.
Suggested Reading Armstrong, A. and Taylor, J. 1978: Regional economic policy and its analysis, 2nd edn. Deddington, Oxford: Philip Allan, 300-8. |
|
