We do not yet have a systematic exploration of these mechanisms, nor an accounting
of their relative importance. But there is some reduced-form evidence that historical
institutions do affect growth in the manner described by Sokoloff and Engerman. The
problem in establishing an empirical assertion of this sort is fairly obvious: good institutions
and good economic outcomes may simply be correlated via variables we fail to
observe or measure, or any observed causality may simply run from outcomes to institutions.
Acemoglu, Johnson, and Robinson (2001) propose a novel instrument for (bad)
institutions: the mortality rate among European settlers (bishops, sailors and soldiers to
be exact). This is a clever idea that exploits the following theory: only areas that could
be settled by the Europeans developed egalitarian, broadbased institutions.