Structural and frictional unemployment are usually considered among the unpleasant and exogenous facts of economic life about which little can be done. As technology advances and the composition of demand changes, employment must also shift. In the process of adjusting to a new equilibrium, some people will endure spells of unemployment. Usually, this is considered part of a healthy re-equilibration process, and the resulting unemployment is seen as part of the underlying “natural” rate of unemployment. Recent oil and trade shocks that have reduced manufacturing employment focus attention on how the economy adjusts to structural changes. This paper analyzes the nature of this adjustment process and shows the magnitude of gross flows of employment across industries and establishments. It dissects the flow of job creation and destruction, and develops a clearer empirical view of the dynamics of establishment size in relation to employment and unemployment. Only when the ongoing rate of job turnover in the economy is established, can we begin to judge how flexible the economy is and how great technological change would have to be to strain the economy’s ability to adapt.
At least since the Luddites forcefully expressed their opinions, many people have believed that technological change contributes to unemployment. With the advent of the computer age, fears have increased either that there will be little productive work left for people to do, or that as the pace of technological change quickens, the volatility of employment will increase as industries go in and out of technological style. The latter hypothesis depends on change generating unemployment. In a flexible economy, this need not necessarily be the outcome. With good information and low adjustment costs, workers may shift across industries without experiencing frequent or long unemployment. The goal of this paper is to show the normal level of turnover of jobs in the U.S. economy. I will show that in a normal year a substantial fraction of all jobs are destroyed and created. Most of this flux is within, not across, industry lines.
If technological change were a driving force in employment growth and decline, and if the degree of technological change or the impact of technological change on employment varied substantially across industries, then we would expect to see sharp differences in employment patterns among establishments in different industries. In fact, there is greater employment variation within industries than across industries. Part of the employment variation within industries may be due to temporary cost advantages reaped from technological advances, but this is not the sort of technological change that has a pervasive effect across an entire industry. The cross-industry shifts, which have attracted the most attention, and where one might expect technological effects to dominate, ignore the source of most job flux.
In a sense, the technological unemployment cup is both half empty and half full. If all the employment variation observed here is fundamentally caused by technological change, then technological change, as measured here, may account for roughly 2.2 percentage points of unemployment in an average year between 1977 and 1982. At the same time, an economy that loses one in nine jobs and creates one in eight jobs in an average year already has experience with great job volatility, which suggests considerable flexibility to respond to additional technological change.
The population of establishments analyzed here is described in the next section. Section III provides an overview of the economy of the state studied here, and of the growth and decline of employment. New evidence on the instability of jobs is presented in Section IV. The transient nature of demand shocks at the establishment level is demonstrated in Section V. Section VI tests for the existence of industry, area, or year effects on establishment growth rates. Section VII presents the conclusions.