# A tibble: 8 × 7
term estimate std.error statistic p.value conf.low conf.high
<chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 Intercept -1.86e+3 1.35e+2 -13.8 1.13e- 42 -2.12e+3 -1.59e+3
2 Bachelor's share -1.35e+3 1.72e+2 -7.88 3.68e- 15 -1.69e+3 -1.02e+3
3 Labor Tightness 2.83e-2 2.40e-4 118. 0 2.78e-2 2.87e-2
4 Patents per employee 1.13e+5 6.75e+3 16.8 2.88e- 62 1.00e+5 1.27e+5
5 STEM share 1.41e+3 3.87e+2 3.64 2.75e- 4 6.50e+2 2.17e+3
6 Manufacturing inten… 1.03e+3 1.61e+2 6.43 1.37e- 10 7.18e+2 1.35e+3
7 ICT intensity 2.77e+4 1.07e+3 25.8 7.87e-141 2.56e+4 2.98e+4
8 Turnover rate 1.18e+4 1.20e+3 9.84 1.00e- 22 9.46e+3 1.42e+4Model
Data Generating Mechanism
This analysis examines the relationship between county characteristics and AI job intensity using the following specification:
\[AI\_{it} = \alpha_i + \gamma_t + \beta_1 STEM_{i,t-1} + \beta_2 Tightness_{i,t-1} + \beta_3 Manufacturing_{i,t-1} + \beta_4 X_{i,t-1} + \varepsilon_{it}\]
Where: - \(AI_{it}\) is the share of AI-related job postings in county \(i\) at time \(t\) - \(\alpha_i\) are county fixed effects - \(\gamma_t\) are year fixed effects
- \(STEM_{i,t-1}\) is the lagged share of STEM degrees - \(Tightness_{i,t-1}\) is lagged labor market tightness - \(Manufacturing_{i,t-1}\) is lagged manufacturing employment share - \(X_{i,t-1}\) represents other lagged county characteristics - \(\varepsilon_{it}\) is the error term
Key Modification: Population Weighting
The critical methodological contribution examines how results change when observations are weighted by log(population):
\[w_i = \log(Population_i)\]
This weighting scheme gives more influence to larger counties in the estimation, potentially changing the interpretation of results from a county-centric view to a population-centric view.