Thread by @CEBKCEBKCEBK

Brief thread on how & why to deal with terror. This paper uses synthetic controls to show that the 1996 start of organized terror in Xinjiang caused it to steadily fall further below its likely counterfactual trend—with income losses reaching 38% by 2018. Passes placebo test well

& this paper uses synthetic controls to show that the 2016 start of organized crackdowns in Xinjiang barely affected its GDP while hugely increasing its arrest rate & security spending. There are *dozens* of pages of robustness tests, & in all of them the takeaway looks the same.

You can also just eyeball big events to see an intuitive impact. For example, maybe the dot com bubble was already popping, but it really looks like 9/11 made this worse—while also inducing more stimulus, via interest rate cuts feeding the housing bubble & stupid wartime spending

Anyway plenty of papers find an unintuitively large negative impact from what are honestly small shares of total deaths. Obviously some of this is due to publication bias, but I’m convinced that glaring violations of basic norms actually succeed at eroding trust in outsized ways.

If common knowledge matters to eg public order & private commerce—if load-bearing norms in civilized civs are largely enforced by expectations rather than police—then hijackers of common knowledge about security & comity can derail our civ, even if we all know it’s “irrational”…

This was itself common knowledge: eg “Wealth of Nations” & Smith’s other works are filled with quotes like this; many contemporaries also recognized that commerce & civility go hand in hand; thus if you hack “expectations about others’ expectations” etc, things can spiral badly…