On one side of the border fence, in Santa Cruz County, Arizona, the median household income is $30,000. A few feet away, it's $10,000….The key difference is that those on the north side of the border enjoy law and order and dependable government services — they can go about their daily activities and jobs without fear for their life or safety or property rights. On the other side, the inhabitants have institutions that perpetuate crime, graft, and insecurity.
With apologies to Douglass North, I am rarely happy with this kind of explanation. First, are the bad institutions cause or effect? Most likely we need a framework which allows them to be both.
Second, I want the theory to also explain the (quite large) difference between the truly poor Chiapas and the relatively wealthy northern Mexico. By many metrics northern Mexico is more corrupt than Chiapas (there is more to be corrupt over, for one thing, plus drug routes play a role) and it very likely has higher rates of violent crime. In general I prefer theories which explain three data points to theories which explain two. Chiapas, of course, isn't some weird outlier which I pulled out of a hat; it's in the same country as northern Mexico and many people from that region have populated both northern Mexico and Arizona for that matter. I could have picked many other parts of Mexico as well.
One factor is positive selection into northern Mexico, on grounds of ambition and desire for higher wages. Another factor is that northern Mexican norms are (partially) geared to support American multinationals and these norms have spread more generally, including to Mexican enterprises in the region.
On another point, as I get older, I tend to view "family structure which encourages an obsession with education" as an increasingly important variable for explaining levels in per capita income, if not always growth rates in the immediate moment. It's not a truly independent variable — when it comes to growth what is? — but it's one good place to start. It helps explain why the Soviet Union, after decades of state fascism/communism, slid into a living standard higher than that of much of Latin America. It explains quite a bit of Arizona vs. Mexico but less of northern Mexico vs. Chiapas. Acemoglu mentions education in his article, but he seems to view it as resulting from instiutions rather than causing them.
I don't buy into the genetic explanations but still I view "family structure which encourages an obsession with education" as very hard to replicate through policy. Emmanuel Todd's The Causes of Progress has many problems, but it is an under-mined book when it comes to the causes of both liberty and economic growth.