Jobs for life aren't dead yet

Job insecurity has been much in the news in recent years. Commentators have heralded an end to "jobs for life", describing a much more uncertain working life for this and future generations. We are warned to expect a succession of dead-end jobs and a consequent increase in insecurity. The twin culprits are alleged to be trade and technology - to be precise, the expansion and intensification of competition (globalisation) and the pervasive presence of information technology.

Let's look at the facts. First, what do the terms "job insecurity" and "job tenure" mean? Job insecurity is a rather nebulous term, but presumably it indicates fear of joblessness. It is usually assumed that this insecurity about one particular job translates into concern about jobs in general; that is, higher unemployment.

But this is not so: the unemployment rate depends on both hiring and firing. A low level of unemployment can arise from more long-term jobs, but also from more short-term jobs and high rates of hiring.

Job tenure is a more easily measurable concept. In the large surveys often used by economists, job tenure means the length of time a person has spent with a firm so far. So this is not a task-based definition, but employer-based.

How long do jobs typically last in Britain? The answer is that there is a huge range. Some jobs last a matter of days, some literally a working lifetime. Using data for 1990, Hedley Rees and I have estimated that the average length of a job is about 12 years for women and 18 years for men. These are big numbers, and suggest that there is still a large element of stability in the UK labour market. These averages embrace a wide disparity: while 24 per cent of men's jobs ended in less than five years, more than 40 per cent lasted more than 20 years and 24 per cent more than 30 years. For women, the figures are 41 per cent, 18 per cent and 12 per cent respectively.

These figures show clearly that a substantial share of workers are in short-term jobs at any particular moment. A common argument is that this fraction has increased a lot in the past 20 years or so, and the proportion in long-term jobs has fallen correspondingly. In fact, the data show this is not the case. Elapsed job tenure was the same for women in the early 1990s as it was in 1975; for men it had fallen by about one year.

Of course, these averages might conceal offsetting changes for different groups in the economy. Our recent analysis has shown that sub-dividing the data to study particular groups defined by age, education level, occupation, industry and region does not overturn the broad conclusion. There is a more obvious decline in job tenure for low-earning men, mirroring their poor showing in earnings growth comparisons.

There are caveats and puzzles. First, jobs end for two types of reason: workers leave voluntarily, or they are forced out. Changes in job tenure therefore measure changes in both factors. It could be, then, that a much higher rate of involuntary job loss is balanced out by fewer resignations, leaving the average tenure figures unchanged. But other data argue against this line.

Second, it could be that changes in long-tenure groups are simply slow to accumulate and we would not expect to see anything in the data yet. This is probably true, but changes at the short end - jobs of one year or less- would be expected, and these have not materialised.

Why is there this discrepancy between the facts on job tenure and the widespread public concern over insecurity? One explanation is that media folk have, possibly justifiably, been feeling insecure and have transmitted that fear to the rest of us. Another possibility is that individuals on temporary contracts, or with less formal job protection, may feel insecure even if the contracts are generally renewed, as the data suggest. Finally, individuals feeling worried about their jobs may take action to offset the perceived threats. The insecurity may reveal itself in longer working hours, lower wage claims and the like, all of which may counteract the forces of "trade and technology" to produce stable job tenure outcomes.

The facts are that jobs last about as long now as they have done for the last 20 years or so. Can we say anything about what is the best level for average job tenure? The public debate suggests that from an individual's viewpoint, the longer a job lasts the better.

From the viewpoint of the economy as a whole, the answer is much less clear. What matters above all for a country in the medium and long term is aggregate productivity growth. Job tenure affects this in two ways. First, long-term job tenures are conducive to training and provide the right environment for appropriate work incentives. But second, modern economies are continually buffeted by shocks - changes in tastes, the arrival of new products and production techniques - and short-term job tenures facilitate the rapid reallocation of workers from suddenly less productive to more productive businesses.

It is not straightforward to compare job tenure patterns across countries, because the aggregate figures are affected by a variety of factors - industrial structure, age structure, education, the stage of the business cycle, unemployment and so on. These need to be controlled in order to isolate the contribution to the picture of a country's institutions. I have carried out such an investigation with Lia Pacelli and Hedley Rees, comparing the length of jobs in Britain and Italy. Despite different labour market regulations and institutions, we find little difference in job tenure; indeed, if anything, jobs appear to last slightly longer in the less regulated UK labour market.

The facts suggest fears of a big change in the nature of work and the emergence of an industrial peasantry are overdone. Reports of the death of "jobs for life" appear exaggerated.

q Simon Burgess is a Research Fellow in The Centre for Economic Policy Research's Human Resources programme and a Senior Lecturer in Economics at the University of Bristol. He develops these views further in a series of Discussion Papers published by CEPR. For further information, call 0171-878 2917.