1. Technology, Information, and the Decentralization of the Firm
2. All specifications include Census division dummies; baseline demographic characteristics of commuting zones (population, the share of working-age population, the share of population with college and with completed high school, the share of Blacks, Hispanics Asians); the baseline shares of employment in manufacturing, durable manufacturing, construction, and the share of female employment in manufacturing; and the exposure to Chinese imports, Mexican imports, offshoring of intermediate by age, gender, education, race, birthplace and relationship to household head. In each specification we commuting zones (population, the share of working-age population, the share of population with college and with completed high school, the share of Blacks, Hispanics Asians), the baseline shares of employment in manufacturing, durable manufacturing, construction, and the share of female employment in manufacturing; cells are defined by age, gender, education, race, birthplace and relationship to household head. All specifications include Census division dummies; baseline demographic characteristics of commuting zones (population, the share of working-age population, the share of population with college and with completed high school, the share of Blacks, Hispanics Asians); the baseline shares of employment in manufacturing;Columns 2, 5, and 8 show unweighted regressions. Columns 3, 6, and 9 report Li's robust regressions, which downweight cells are defined by age, gender, education, race, birthplace and relationship to household head
3. Columns 4 to 6 and 10 to 12 present stacked-differences estimates of the impact of the exposure to robots on employment and wages. In columns 1 to 6, the exposure to robots is measured using the 1990 employment shares; In columns 7 to 12, the exposure to robots is measured using the 1980 employment shares. The dependent variable is the change in Census private employment to population ratio in Panel A, the change in employment to population ratio from the County Business Patterns in Panel B, the change in the log hourly wage in Panel C, and the change in the log weekly wage in Panel D. The specifications in Panels A and B are estimated at the commuting-zone level; while the specifications in Panels C and D are estimated at the demographic cell � commuting zone level. Demographic cells are defined by age, gender, education, race, birthplace and relationship to household head. All specifications include Census division dummies; baseline demographic characteristics of commuting zones (population, the share of working-age population, the share of population with college and with completed high school, the share of Blacks, Hispanics Asians); the baseline shares of employment in manufacturing, durable manufacturing, construction, and the share of female employment in manufacturing; and the exposure to Chinese imports, Mexican imports, offshoring of intermediate goods and the baseline share of employment in routine jobs (see text for the construction of the sources of these variables);Covariates & sample restrictions: Baseline covariates Unweighted Down-weights outliers Notes: Columns 1 to 3 and 7 to 9 present long-differences estimates of the impact of the exposure to robots on employment and wages