I agree with this, lying and BS get you far in politics. In an econ talk today, Nicolas Longuet-Marx was arguing that Democrats used to do a fair bit of this; said Pelosi gave House candidates a lot of leeway to say what they wanted but then imposed strict voting discipline on them once elected.

I think this is true, and putting more effort towards, say, keeping TAA alive would have ben good. Even if economists would have underestimated the adjustment costs 20 years ago, though, and not thought they could be 1/4 the gains, it still looks like their advice to open was right.

Most inexplicable timeout on the Bengals' second down too

I think (if I’m understanding you right) it depends on whether you hope the Xs explain the trend. This way says “look, the X’s are growing, but with a constant effect of X, that can’t explain the trend.” You could also do it detrending everything: how much of detrended Y_1 does detrended X_1 explain

What are you after? If the time periods are groups, you’re getting how much of Y_1 is explained by X_1? And then the time trend is leading X_1 predictions to be too high and X_10 predictions to be too low, and it looks like there’s a lot the Xs can’t explain?

Reminds me of this graph from here on how Samsung and LG were able to quickly relocate production to China and then Thailand in response to US tariffs (and once they'd relocated production, sold more from China to third countries that didn't have tariffs too).
www.aeaweb.org/articles?id=...

Yeah, here it is
drive.google.com/file/d/1IDP_...

Was it this?
x.com/LeonardBocqu...

Part of being educated though is being guided through books, building up to things. I think math is hugely valuable, but if you gave me an algebraic topology book and paid me to spend time with it my life satisfaction might fall.

Agree that if everything's uncorrelated with everything else you should just be able to go down in order.

Then the portfolio with those 4 pays 1.55 with certainty, but the pairwise correlation would only tell you that the first third of the 4 has a correlation of -0.33 or something with the first asset, and might miss the value of holding all 4 together.

Would be satisfying if you could only look at k and k+1, but I think those examples show you can't. And I'm not sure you can just look at pairwise correlations: suppose you have 4 assets, and with certainty 2 pay off (and pay 2, 1.9, 1.2, and 1.1).

Seems like you're doing this in the proof sketch, but you have to look ahead to all of the f_{im}s to see if any of them has, say, a correlation with X_1 of -1 (which could allow purely riskless arbitrage), and back at f_{1,k+1} to see if k+1 has a correlation of 1 with X_1, making it worthless.

Even if it all got replaced by gas that'd help the climate and have huge benefits for people's lungs. Are there meaningful permitting hurdles for gas power plants?

Important to impute (e.g. with presidential results) in races that were uncontested/D vs. D/R vs. R, but I don't think that shifts this

I wonder if this would be possible for quasi-experimental papers

Another way of putting it, making it harder to drive/easier to bike benefits all the people who are taking public transit either way and walking to/from their stop.

There’s good evidence that protected bike lanes increase biking in general!

Biking creates big positive externalities (space saved, air pollution from car brakes, carbon reduction, makes it more pleasant for inframarginal bikers, internalities for bikers themselves) that it’s worth paying high costs even if bike mode share isn’t that high.

In the first Trump administration, a bunch of companies just received straight up exemptions from the tariffs (Section 301 exemptions). Just for asking! Heard there's a trade paper in progress about the consequences of this too.
www2.lehigh.edu/news/politic...