He wrote (in pdf) that:
Empirically, I demonstrate economic distress stimulates religious intensity by exploiting the fact that rapid inflation caused relative prices to favor growers of staples, namely rice, and hurt sticky wage-earners, particularly government employees, whose salaries are set by federal law. I find that households experiencing a $1 decline in monthly per-capita nonfood expenditures are 2% more likely to increase Koran study and 1% more likely to switch a child to Islamic school. Moreover, participation in other social activities did not increase while labor supply increased, which seems inconsistent with an opportunity cost view of religious intensity.And here is (in another pdf) the catch-22.
Instead, using a difference-in-differences strategy, I show that households that increased Pengajian during the crisis have less unmet demand for alms and credit for meeting basic daily needs in the following time period. The effect of economic distress on religious intensity essentially disappears in places where credit is available. And religious institutions appear to facilitate consumption smoothing among villagers, suggesting religious intensity functions as ex-post social insurance.
I present an analysis of data from the Hundred Villages Survey and data from the Database on Social Violence in Indonesia 1990-2001. OLS estimates show a large positive relationship between religious intensity and social violence. Because most religious intensity measures are relatively time-invariant and are pre-crisis measures and because villages are unlikely toThese two papers are worth to read completely (skip the math, if you want). And by the way, back to my initial question, on why it is more urban than rural extremism that we observed lately, the first sentence of the first quote partly tells it: economics distress hit the sticky wage earners, the urban feature, more severely.
build schools, seminaries, or religious buildings in anticipation of social violence, reverse causality is unlikely to explain this association. In fact, a strong relationship between pre-crisis measures of religious intensity and social violence begins after the crisis. In addition, stronger forms of religious intensity are more strongly associated with violence.
p/s: you should see the syllabus of his teaching, too