Read on Twitter
@SydneeCaldwell and I have a new version of our paper “Outside Options in the Labor Market” (my 2018 jmp).
We develop a method to estimate workers’ outside options and show that differences in options explain a large portion of the gender gap. 1/6 https://www.dropbox.com/s/eo6qau6d5fqu9jk/Caldwell_Danieli.pdf?dl=0
We develop a method to estimate workers’ outside options and show that differences in options explain a large portion of the gender gap. 1/6 https://www.dropbox.com/s/eo6qau6d5fqu9jk/Caldwell_Danieli.pdf?dl=0
Key idea: you can estimate workers’ options from the allocation of similar workers into jobs
In our model, the concentration of similar workers across all observable dimensions (occs, industries, locations) is a sufficient statistic for the effect of outside options on wages 2/6
In our model, the concentration of similar workers across all observable dimensions (occs, industries, locations) is a sufficient statistic for the effect of outside options on wages 2/6
We call this statistic the Outside Option Index (OOI). We estimate the OOI for 1% of the German population and find that women have 25% fewer options than men and that options are better for more educated workers and in big cities. 3/6
Workers in high-paying occupations often have fewer options. That’s because they tend to have industry-specific skills (e.g. physicians, pilots).
**
Using a shift-share instrument, we estimate that 10% more options yield 2% higher wages. 4/6
**
Using a shift-share instrument, we estimate that 10% more options yield 2% higher wages. 4/6
Taken together, we estimate men earn 5% (0.25*0.2) higher wages just because they have more options (¼ of the gender gap in Germany). Using counterfactuals we show this is entirely driven by different willingness to commute/move. 5/6
We show the OOI can be used for various exercises such as predicting who will recover faster from a mass-layoff, or analyzing who benefits from a new train. To use it - download our R package, and you can calculate it with just one line of code 6/6 https://www.orendanieli.com/r-packages
