Ryan Avent in the NY Times: "why should productivity growth have fallen in recent decades? … economic growth has contributed very little to the inflation-adjusted wages of workers without a college education — and since the turn of the century, everyone but the top 1 percent of earners. And so companies have felt very little pressure to replace stockers with robots, cashiers with touch-screens and customer-service staff with chatbots. Neither has there been much reason to squeeze more productivity out of workers by investing in training or by finding ways to equip them with new, productivity-enhancing technologies. Today, despite an unemployment rate at just 4.1 percent and fat corporate profits, wage growth remains well below the peak rates of the 1990s and 2000s.”
The NBER Digest on new research by Haltiwanger et al: "Upward movement of workers on a "job ladder" from low-productivity to high-productivity firms is heavily dependent on the business cycle. During booms, net employment at high-productivity firms grows faster than at low-productivity firms, resulting in workers moving up the ladder. During busts, these upward job-to-job changes essentially stop. Net employment flows are instead driven by layoffs, with low-productivity firms losing comparatively more workers than their higher-productivity counterparts. …
The researchers find that younger workers are disproportionately likely to climb the ladder by moving to more productive firms. … Less-educated workers also are disproportionately likely to move up the job ladder during expansions. More-educated workers are less likely to enter employment at low-productivity firms in the first place, but once in such firms they are less likely to separate from them. The researchers hypothesize that more-educated workers may be more specialized, and thus less mobile across firms. … Economic slowdowns, while imposing costs throughout the labor market, are particularly harmful to the employment prospects of younger, less-educated workers.”
They use Census LEHD data on firm productivity from 2003–11.