A nice fuzzy RD design! The interpretation of their results, though, is difficult because the counterfactual is not very clear to me.
They are using admission score cutoffs to oversubscribed vocational schools. The intuition here is that people who score right below and right above the admission cutoff did so because of randomness, therefore their setup approximates a random assignment experiment at the cutoff.
For the results to be interpretable, we need to think through the counterfactual. That is, what happens to the students who score below the cutoff and that are not admitted to those over-subscribed vocational schools? Do they complete a general education degree at another school? Do they attend another non-over-subscribed vocational school? Do they drop out of school altogether?
I am curios about reduced form estimates of the gap (i.e., without the 2-stage estimates). I wonder how much of the $2000 could be explained by more experience on the job for people who complete a vocational track rather than a general education track+university. In that, I wonder how much of a delay entry to the job market could explain the returns of vocational education.
Another thing that I am wondering about is how elastic to education/credentials is the Finnish labor market. That is, how responsive are wages to credentials/higher education degrees. I hope that the actual articles/publication includes a description of this because it will be important for external validity of these results beyond that specific national market. For example, in the United States, there is a big wage gap between professional degrees (e.g., medical doctor, ~$120.000 year) and vocational degrees (e.g., welder, ~$40.000 year). I wonder if the same large gap in wages is present in the Finland. If that is not the case, I wonder how much these results would apply to other non-Finnish job markets.
I'm also not sure if I buy the authors' claim that "The regression discontinuity design estimates come from the middle of the distribution of academic ability. However, the benefits may be even larger for people with low compulsory-school GPAs who only apply to the vocational track, while vocational education may be detrimental for people with high GPAs who apply only to the general track." This feels like a misrepresentation of the RD estimate. They only estimate the effect for people who score right around the cutoff. They cannot really extrapolate the effects beyond a small bandwidth around that cutoff. They are really overselling their results when they discuss people away from their cutoff and just guessing what would happen to people at other parts of the distribution.
They are using admission score cutoffs to oversubscribed vocational schools. The intuition here is that people who score right below and right above the admission cutoff did so because of randomness, therefore their setup approximates a random assignment experiment at the cutoff.
For the results to be interpretable, we need to think through the counterfactual. That is, what happens to the students who score below the cutoff and that are not admitted to those over-subscribed vocational schools? Do they complete a general education degree at another school? Do they attend another non-over-subscribed vocational school? Do they drop out of school altogether?
I am curios about reduced form estimates of the gap (i.e., without the 2-stage estimates). I wonder how much of the $2000 could be explained by more experience on the job for people who complete a vocational track rather than a general education track+university. In that, I wonder how much of a delay entry to the job market could explain the returns of vocational education.
Another thing that I am wondering about is how elastic to education/credentials is the Finnish labor market. That is, how responsive are wages to credentials/higher education degrees. I hope that the actual articles/publication includes a description of this because it will be important for external validity of these results beyond that specific national market. For example, in the United States, there is a big wage gap between professional degrees (e.g., medical doctor, ~$120.000 year) and vocational degrees (e.g., welder, ~$40.000 year). I wonder if the same large gap in wages is present in the Finland. If that is not the case, I wonder how much these results would apply to other non-Finnish job markets.
I'm also not sure if I buy the authors' claim that "The regression discontinuity design estimates come from the middle of the distribution of academic ability. However, the benefits may be even larger for people with low compulsory-school GPAs who only apply to the vocational track, while vocational education may be detrimental for people with high GPAs who apply only to the general track." This feels like a misrepresentation of the RD estimate. They only estimate the effect for people who score right around the cutoff. They cannot really extrapolate the effects beyond a small bandwidth around that cutoff. They are really overselling their results when they discuss people away from their cutoff and just guessing what would happen to people at other parts of the distribution.