This paper demonstrates how, through the capital reallocation channel, increased automation in routine occupations has reduced employment and wages in non-routine occupations. Automation in routine occupations absorbs capital from non-routine occupations, reducing employment and wages in the latter. This mechanism is referred to as automation cross occupation spillovers. Between 1980 and 2010, automation reduced average labor income by 27%. Cross-occupation spillover is responsible for 62% of this drop. For example, the increase in automation in the ten percent most routine intensive occupations between 1980 and 2010 reduced average labor income in the 90% least routine intensive occupations by 2.04%. Furthermore, I find that automation has contributed to the rise of inequality in the United States. Indeed, between 1980 and 2010, automation was accountable for 30.3% of the increase in occupational labor income inequality.
This paper reconciles two stylized facts that characterize the modern economic growth, balanced growth and structural change, in a context where the labor share of the goods sector is greater than the labor share of the services sector. I extend the neoclassical growth model to two sectors, goods and services, where services are more labor intensive than goods. I demonstrate that balanced growth is consistent with structural change, as evidenced by the fact that goods TFP increases faster than services TFP. Along the balanced growth path, the output share of services rises while the output share of goods falls.