Bell, D.N.F. and Branchflower, D.G. (2018), ‘The Lack of Wage Growth and the Falling NAIRU’, National Institute Economic Review, August, pp. R40-R55. Doi: 10.1177/002795011824500114.

From page R48, 2nd column, final paragraph, to page R50 1st column following the italic text the following replacement paragraphs apply:

The second piece of evidence in Vlieghe's speech was an econometric analysis estimating a wage growth equation – a Phillips curve – using 256 observations from 1997 through 2017 disaggregated by sector. He includes a lagged unemployment rate, a lagged dependent variable and a set of sector dummies. His findings suggest a significant impact of unemployment on wages – “one percentage point increase in the unemployment rate in a sector lowers wage growth in that sector by half a per cent in the following year”. This suggests a considerable sensitivity of wages to unemployment: given that the unemployment rate has fallen from a peak value of 8.5 per cent in September–November 2011 to 4.2 per cent currently (March–May 2018), wage growth should have increased by around 2 per cent. However, the increase in wage growth over this period has only been around 0.7 per cent. Nevertheless, Vlieghe points out that Phillips himself did not expect the relationship to hold rigidly. He also acknowledges that the pattern of time dummies in his model, which capture factors affecting all industrial sectors simultaneously, has shown an increasingly negative pattern over time, implying that other unobserved factors have been depressing wage growth. These factors may include underemployment.

He also estimates a Phillips curve which includes lagged unemployment and wage growth interacted with a post-crisis dummy. Neither turn out to be significant, though how these variables interact with what is being captured by the year dummy effects would have been worth exploring. His interpretation is that the Phillips curve is in robust health, but that it moved downwards for a period of time during the crisis and its immediate aftermath. Given that the underemployment rate rose faster than the unemployment rate during the recession and has subsequently returned to close to the unemployment rate, his finding is also not necessarily inconsistent with a model where underemployment is the main driver of wage growth, rather than unemployment. Indeed in his discussion, Vleighe acknowledges that underemployment may have played a role in depressing wage growth over and above the unemployment rate.