search

neilalex / ad-replication

Paper Replication: The Growth of Low-Skill Service Jobs and the Polarization of the U.S. Labor Market (AER 2013)
https://neilalex.com/paper-replication-the-growth-of-low-skill-services-jobs-and-the-polarization-of-the-us-labor-market/
MIT License
1 stars 3 forks source link

Achieving consistency with Dorn's occ panel #4

Open IoannisPapadakis opened 6 years ago

IoannisPapadakis commented 6 years ago

In etlIpumsCzOcc.do @neilalex in line 31 divides by 10. It is not clear to me why you did that.

For 2005 the OCC (provided by IPUMS USA: ACS ) is a 3-digit code?! Dorn's panel of occupations allocates occupations (OCC) to his occ1990dd codes. So why you changed OCC dividing by 10?

neilalex commented 6 years ago

When Dorn prepared his occupation panel, IPUMS had been using 3-digit OCC codes. However, more-recent pulls of the ACS part of the data come with 4-digit OCC codes.

Fortunately, I believe with the 2005 ACS data at least, these new codes are simply the original codes with an appended zero--and we can get back the original Dorn-compatible codes by just dividing by ten (here's one reference on it e.g.).

You bring up a good point in general though--in the post-2005 ACS datasets, I believe these codes do more than append a zero. And so for, say, any potential future extensions of the paper, it unfortunately may not be possible to take Dorn's ACS occupation panel and simply apply it to the given ACS data year; I think we'd need to either create a new panel, or hope that Dorn chooses to refresh his.