And if you need code, or clarifications, feel free to email me!

-> But, it also discusses other methods for detecting change-points, such as information criteria - equally computationally efficient but under-explored - and also lasso and mixed-integer programming.

And, it also highlights new ideas for research. Check it out!

-> It highlights bootstrap methods for detecting change-points, which work under more general assumptions that are relevant to macroeconomic applications.
-> It focuses, when methods are available, on multiple change-points and sequential testing, because of computational efficiency.

-> It is an empirical guide first, so focuses only on the most general assumptions - dynamic models, or autocorrelated errors, or both - including local projections and vector autoregressions.
-> Unlike previous reviews, it also carefully reviews models with endogeneity - estimated via IV or GMM.

The posts are so vicious, I can hardly see ppl happy there.

Please try for the sake of it, to see what others are thinking. See if there is any bridge to be mend. Very hard.

Some ppl here don't get what losing half a thumb means. It is clearly a humor piece. But based on real facts: it really hurts! I was stung 3 weeks ago by what looked liked an Asian hornet, twice, in my right thumb. It hurt for two days, and there was even a sepsis line forming, but it died out.

So messed up!

I find it strange that ppl are afraid to walk on the streets rather than sidewalks. In big cities they do!

But maybe we should think of this tariffs as an exogenous, independent copy (see stupid Bayes prior) of something related to a lag of endogenous variable, by accident having the same value in one period. Or another creative way. It is ridiculous though what is happening.

No worries, in macro we purge the instrument of the lagged endogenous variation via OLS to get at the dynamic causal effect. In this case it's a one time shock, and purging it based on 2024 would yield a perfect zero. Nice instrument 😂

I am also wondering. I am going to US next week but I am staying only 6 days, I hope it's gonna be fine.

Congrats!!!

Or bike if it's safe? Would be 10-15 mins!

Additionally, I wrote the model as well, and it's far from obvious how to, for example, group parameters to get some identification power out of the scant data. And which to calibrate and which not.

Not to mention all data computations, fittting 4 streams of data by age and risk weekly, with missing observations, while model is daily. I never felt so alone in this endeavour, especially coding in Matlab while all co-authors code in R, not feasible here due to high-dim filtering.