Kronecker

I think what chris_tyman is saying is that after 10-20 years into your career in academia, it doesn't really matter which school you got your PhD from. Sure, the connection is going to help, but that's only initial condition; convergence is all that matters. The connection is going to help you get good job placement, but then you have to prove your worth. At the end of the day it's going to be how well you are at producing good-quality research paper-- and dissertation (and LoRs, again) is a good indicator of that. Regardless, the correlation kudo pointed out is definitely there.

I agree with asquare that it's really dependent on a lot more factors than just working at the Fed. Like the OP, I was debating between going straight to grad school (I got decent offers from a few top 20 schools) or work at a regional Fed and I decided to do the latter. The main thing that adcom look for is LOR. If you get to work for an economist who is well-known and is willing to write you a fabulous LOR, I say it's really worth it. In my case, I ended up getting good LoRs from my economists, will get a few co-author papers by the time I leave, and I'm going to a top 5 school next year, so clearly it's not a waste of time, ex-post. I'm not saying that working at a Fed is the ticket to top programs. As 2010app9357 said, conditional on getting good quality work and getting great LoR from a well-known economist, I say it is a value-added for your application. Plus you get lots of $$$ and learn tons of programming, and in my case, some economic theory. So before accepting an RA offer from the Fed (or any RA offer for that matter), be sure to know who you will be working for and be willing to work more than what an average RA might put in.:)