How unpredictable is technology's impact on the job market?

Abstract

The U.S. Bureau of Labor Statistics (BLS) Employment Projections frequently identify the correct direction of occupational change but consistently underestimate its magnitude, historically missing total job mix dynamism by a factor of 2-3. While projections for sectors like healthcare show accuracy, forecasting politically sensitive jobs yields mixed results. Technological shifts pose the greatest challenge. For example, data scientist roles grew by 553% against a BLS projection of 23%, despite early indicators of demand. The predicted stability in retail sales employment was a major miss, as actual decline began post-2014, with Census data showing earlier downturns, possibly due to BLS survey limitations regarding small businesses and multi-job holders. Conversely, the long-standing decline in secretarial and administrative roles was conservatively projected to stabilize, contradicting historical patterns and the potential impact of AI. To improve, BLS forecasts should integrate historical visual trends, demand rigorous justification for deviations from established trends, incorporate ranges of uncertainty, and focus heightened attention on large occupations for broader impact. – AI-generated abstract.