Routine-biased technical change, structure of employment, and cross-country income differences

We investigate links between routine-biased technical change, the structure of occupational employment, and cross-country income differences. To implement this, we combine several data sources, including national labour force surveys and the Penn World Tables. In our novel dataset spanning 92 countries, we document a negative relationship between the employment share of routine occupations and GDP per hour worked. We then conduct a development accounting exercise in which we differentiate labour inputs by occupation and allow for occupation-specific technologies. We find a systematic relationship between occupation-specific productivities and development. Labour in routine occupations tends to be much more productive in richer countries, with effective productivity in the top 25 percent being about 14 times higher than in the bottom 25 percent of countries ranked by GDP per hour worked. Cross-country gaps in routine technology alone account for about 17 percent of the 90–10 ratio of GDP per hour worked, whereas differences in abstract labour technology do not contribute to the observed GDP dispersion. Eliminating all occupations’ and capital’s technology differences would compress the GDP distribution by 32 to 40 percent.