Lousy and Lovely Jobs: The Rising Polarization of Work in Britain

Lousy and Lovely Jobs: The Rising Polarization of Work in Britain
By Maarten Goos and Alan Manning

This is one of several articles referenced by the Oxford University study that analyzes the effect of technology on employment. While the article has several graphs, charts and tables, it is the bar graph showing how the distribution of jobs has changed in Britain over the period from 1979 to 1999 that most impressed me and is reproduced on an approximate basis above this summary. The starting point for the construction of this graph was to categorize all jobs in the study into their decile based on the mean wage in 1979. Thus, the base number of jobs for each decile would be 10% of the total number of jobs in the study. With time, the number of workers in each job changes. A second tally was based on the number of workers in 1999 in each of the occupations. Such second tally was compared with the base tally. The bar graph is the result of this comparison and shows the following:

• a substantial increase in the number of jobs at the highest end of the wage spectrum
• a more moderate increase in the number of jobs at the lowest end of the wage spectrum
• a decrease in the number of jobs in the middle of the wage spectrum

A common theory of economists is that technology affects labor markets in a way that is skill biased. One way to apply the “skill biased” theory would be to contend that skilled workers are better able to use technology to increase their productivity which would lead to higher wages. This would explain the percentage increase in the nine and ten deciles in the above graph.

However, this theory would not explain the percentage increase in the number of workers at the lowest end of the wage spectrum.

For this explanation the authors introduced the routine / non-routine characteristics of jobs and theorized that it is the computerization of routine jobs in the middle of the wage spectrum (“middling” jobs) that led to the increase in the number of workers in the service jobs that typically are both non routine and lower paying jobs. The assumption is that routine jobs are more susceptible to being computerized than are non-routine jobs.

What I find most interesting is that this article appeared in the February, 2007 issue of The Review of Economics and Statistics. It was only three years later, in October, 2010, that Google announced that it had fully automated several Toyota Priuses. Such announcement demonstrated that while routine jobs may be the first to be lost to computerization, non-routine jobs may not be far behind.