I always assume it's the unadjusted gap unless otherwise stated. And in this case it even says so in the graph notes.

Both are valid and useful, they just capture different things.

We find correlations around 0.8--0.9, implying that these measures move together across space and time.

We conclude that intergenerational mobility and sibling correlations can indeed be used to compare levels of equality of opportunity across time and place. 4/4

In contrast, EOp estimators are derived from a normative ethical theory, and thus easy to interpret --- but they require very rich data to estimate empirically.

We estimate all three measures across Swedish local labor markets and birth cohort, and examine how closely correlated they are. 3/4

The first two are relatively simple to estimate, but their normative implications are not clear (what is the "right" level of intergenerational mobility?). 2/4

I spend way too much time meticulously rejecting them.

Thank you, looking forward to it!

That's an unfair characterization. Inequality, discrimination, and pollution are all large fields in modern economics, to give just three examples.

Looking forward to it!

Other advantages: writing a function in R is trivial, while in Stata I always had to spend time with the "program define" and "syntax" manuals. Makes me modularize my code more, which makes it less error-prone.

And making graphs (with ggplot2) is so much simpler!

I switched when I was writing an estimator that requires reshaping and merging of population data, all repeated hundreds of times in a bootstrap. Wrote it up in Stata, and one iteration took 40 minutes. Wrote the same thing in R (with the data.table package), and one iteration took 40 seconds.