The BLS occupational employment projections developed for the 1988-2000 period were reasonably accurate, correctly capturing most general occupational trends. As with previous evaluations, however, the inaccuracies that surfaced reflected a conservative tilt to the projections. The primary source of error was the projection of changes in the utilization of occupations by industry, or staffing patterns, rather than the projections of industry employment themselves.
In the study presented in this article, several different measures were used to assess the accuracy of the projections for both major occupational groups and detailed occupations. Among the various measures, the most traditional involved comparing actual with projected employment in terms of percent change, numerical growth, and share of employment growth between 1988 and 2000. An absolute percent error--the absolute value of the numerical error divided by actual employment in the target year of the projection--was calculated for all major groups and detailed occupations. The actual and projected directions of change also were compared, to see whether employment in occupations that were projected to grow or decline actually did so. (1) Finally, because the 1988-2000 occupational employment projections were the basis for job outlook information presented in the 1990-91 edition of the Occupational Outlook Handbook, the accuracy of the projections was assessed in terms of the assumptions made about the factors affecting employment growth or decline.
Major occupational groups
Total employment grew by 21.7 percent between 1988 and 2000, slightly faster than the 15.3 percent that had been projected. The difference is largely the result of an underprojection of total employment by about 7.6 million. The direction of the employment change was anticipated correctly for all but one of the nine major groups. Employment in eight of the nine groups was underestimated. (See table 1.)
All but three of the major groups had absolute percent errors of less than 10 percent. The category of agriculture, forestry, fishing, and related occupations had the highest error of any major group, 16.6 percent. Professional specialty occupations and operators, fabricators, and laborers also had relatively high errors, 11.2 percent and 11.0 percent, respectively. The group with the lowest absolute error, 0.1 percent, was executive, administrative, and managerial occupations.
Five of the nine major groups had absolute errors below 5 percent. Of the five, the category of executive, administrative, and managerial occupations not only was the most accurately projected major group, but also was the lone group for which employment was overprojected. Technicians and related support occupations and precision production, craft, and repair occupations also had very low errors, 1.1 percent and 1.0 percent, respectively. The absolute percent error was 2.2 percent for service occupations and 3.4 percent for administrative support occupations.
In addition to making reasonably accurate employment projections at the aggregate major group level, the Bureau projected the share of total job growth of each group fairly accurately. For example, professional specialty occupations had the largest numerical error, off by more than 2 million workers, but still, the category's share of total job growth was underprojected by only 3.2 percent. The largest difference in share of job growth was 7.9 percent, for operators, fabricators, and laborers; the group's growth was projected to be 1 percent, but actually was 8.9 percent. The gap was due mainly to an overestimation of the effects of automation on the demand for workers.
Although agriculture, forestry, fishing, and related occupations had the highest employment projection error, at 16.6 percent, the group's share of total job growth was underprojected by only 2.8 percent. …