The Effect of Sterling on Regional Unemployment and Earnings

It is interesting that unemployment in the West Midlands was about the same as the GB average in 1995 and 1996 before the rise in sterling and that West Midlands unemployment is now approx 0.5 percentage points above GB average (as of the end of the first quarter of 2000).

The key finding of our earlier research ('Earnings, Unemployment and Housing: Evidence from a panel of British Regions') is that regions with more production workers than average tend to be harder hit by rises in the value of sterling.

The first two columns of the table show (for the Government Office Regions of Great Britain) the proportion of workers in the production sector at the end of 1999 and the long-run effect of a 10 per cent increase in the real exchange rate on the relative unemployment rate in each region.

For example, the proportion of production workers in the workforce of the West Midlands is 0.252, which is 0.0765 higher than the GB average. The long-run effect of a 10 per cent rise in the real exchange rate on the West Midlands is to raise relative unemployment by 0.644 of a percentage point (this is equal to 10*0.0765*0.8417, where 0.8417 is the long-run coefficient from our unemployment equation). Relative unemployment is simply the rate in the region minus the rate in the GB as a whole.

In contrast, given the very low proportion of production workers in Greater London (about 0.078), a 10 per cent rise in the real exchange rate will reduce relative unemployment by 0.824 of a percentage point.

The third column of the table shows the effect of a rise in the real exchange rate on mens' full-time wages. Given that a rise in the real exchange rate leads to lower relative wages for regions with lots of production workers, the lower wages will tend to offset the rise in unemployment (as seen in column four).

For the West Midlands, the table shows that a 10% real exchange rate rise leads to a 0.644 percentage point rise in relative unemployment, a 0.396% fall in relative mens' full-time wages, and a subsequent 0.056 percentage point fall in relative unemployment due to the fall in earnings. Clearly the effect of falling wages on unemployment is not enough to offset the direct effect of higher sterling on unemployment.

Here are four points to bear in mind:

first, since the overall effect of a rise in the real exchange rate on a region with lots of production workers will be to both raise relative unemployment and to reduce relative wages, the residents of the region suffer from the 'double-whammy' of lower wages and worse unemployment.

second, since the effect of a real exchange rate rise is estimated from an econometric model, it comes with an associated standard error. The coefficient on the real exchange rate interacted with the relative proportion of production workers is 0.8417 and has a standard error of 0.1851. This means that there is a 95% chance that the coefficient lies between 0.479 and 1.205.

third, there will also be an effect on womens' earnings of a similar size to the effect on male earnings.

fourth, the effect of a rise in the real exchange rate is seen more quickly on earnings than on unemployment. For earnings, most of the effect will be seen within one to two years, while for unemployment the effect will take three to five years to be seen.
Back