Ancestry and Long-Run Growth Reading Club: Chanda, Cook, and Putterman
Welcome to the fourth and final installment of the EconLog Reading Club on Ancestry and Long-Run Growth. This week’s paper: Chanda, Areendam, C. Cook, and Louis Putterman. “Persistence of Fortune: Accounting for Population Movements, There Was No Post-Columbian Reversal.” American Economic Journal: Macroeconomics 6 (3): 1-28.
The authors data is here.
Summary
Acemoglu, Johnson, and Robinson (AJR) famously argued that the world’s former colonies have seen a great “reversal of fortune.” On average, the more advanced such countries were in 1500, the poorer they are today. In this week’s paper, Chanda, Cook, and Putterman (CCP) argue that AJR omitted a mighty confounding variable: ancestry. While nations‘ fortunes reversed to some extent, peoples‘ fortunes persisted. Countries inhabited by the descendants of relatively successful tribes in 1500 remain relatively successful today.
CCP begin by applying the Putterman-Weil World Migration Index to AJR’s measures of initial development: urbanization and population density. Ancestry adjustment easily re-reverses AJR’s reversal.
Then CCP replace AJR’s measures of early development with the “SAT scores” (state history, agricultural history, and technological history) we’ve seen in earlier installments of this reading club. Unadjusted for ancestry, these measures marginally confirm AJR’s reversal; the signs are right, but none are statistically significant. Adjusted for ancestry, all three measures predictably show strong persistence of fortune.
CCP then move on to a host of robustness checks. As usual, geographic controls dramatically cut the measured effect of ancestry. Results for urbanization and population density are fragile, but SAT scores are robust. I leave the multitude of other robustness checks to the reader; clearly the authors wanted to cover their bases.
“Persistence of Fortune” then argues AJR were too quick to claim victory for “institutions” over “human capital” stories of development. The big lesson:
We find no evidence of an important subset of national groups converting themselves from relatively “backward” to relatively “advanced” by adopting better institutions. The AJR (2002) reversal is instead associated with people from places hosting societies that were relatively socially and technologically sophisticated in 1500 migrating to places that had been relatively backward and that accordingly had relatively low population densities (which were further diminished by absence of resistance to Old World diseases). The most straightforward explanation of the reversal of fortune for territories, then, would be that the connecting of “old” (Eurasia plus Africa) with “new” (Americas, Oceania, and other islands) worlds that began in the fifteenth century led to population transfers in which (inter alia) the technological and social advantages of peoples from the most advanced civilizations sank new roots in previously backward lands.
Critical Comments
1. This is an extremely convincing piece. AJR’s “reversal” evidence was always pretty thin. While CCP are basically able to replicate the reversal, AJR’s results are sensitive to mild tweaks. Adjusting all the results for ancestry, in contrast, dramatically changes the picture.
2. Suppose you knew none of AJR or CCP’s results. You have five measures of early development: urbanization in 1500, population density in 1500, millennia of agriculture, state history, and technology in 1500. A priori, which measures most credibly capture “initial success” that might or might not be reversed? To my mind, technology in 1500 comes in first, followed by a two-way tie between urbanization and population density. State and agricultural history seem least relevant. If a society was barbarous in antiquity but visibly successful in 1500, poverty in 2000 shows reversal, not persistence.
3. If you buy my ranking, CCP’s results are somewhat puzzling. Persistence works well for the best measure (technology), but only slightly for the two runner-ups – urbanization and density. For the least plausible measures, it works best. While it’s tempting to paint whatever measures predict most strongly as the “truest” measures of ancestral development, that’s not good science.
4. While we’re on the subject, none of the papers we’ve examined try to measure early per-capita GDP or use it to predict modern economic performance. They have a standard theoretical excuse: Early economies were Malthusian, stuck at subsistence, so we shouldn’t expect early per-capita GDP to predict current per-capita GDP.
I’ve long been suspicious of this whole story. First, while no pre-modern societies were rich by our standards, their living standards certainly varied. Second, the very existence of slavery shows human beings earned more than their subsistence since antiquity. There’s little point owning a person who eats as much value as he makes. Third, if early per-capita GDP did predict current per-capita GDP, economists would clearly treat this fact as relevant – as they should. By basic Bayesian logic, the alleged failure of early per-capita GDP to predict current per-capita GDP should tip our mental scales in the opposite direction.
Coming up: A general assessment of the ancestry and long-run growth literature, especially as it relates to the case for immigration and open borders, followed by an Ask Me Anything.
The post appeared first on Econlib.