Economists define institutions as the basic rules, customs and practices of society, that determine and constrain the way that people interact. Institutions are the foundations of the economic and political systems of each country, and have been implicated in the explaining the differences in economic development between countries (see this article by Dietrich Vollrath, and my post on that topic as well). Indeed, a range of excellent work by Daron Acemoglu and James Robinson and their collaborators focuses on the role of institutions in explaining development, as detailed in their book Why Nations Fail (which I will review sometime this year - it is near the top of my large pile of books-to-be-read).
But what causes countries to have different political and economic institutions? Acemoglu and Robinson have focused attention on colonial relationships, as well as the roles of health and disease. So, I was interested to read this 2018 article by James Ang (Nanyang Technological University), Per Fredriksson (University of Louisville), Aqil bin Nurhakim, and Emerlyn Tay (both Nanyang Technological University), published in the journal Kyklos (ungated version here). Ang et al. focus on what, to me, seems a surprising angle - the role of sunlight (specifically, UV radiation) on institutional quality across countries. Their argument is that:
...populations facing a permanent threat of developing eye disease have historically had a lower incentive to invest in cooperation via institution building. Moreover, specialized activities such as institution building necessitate a food surplus, which requires prior investments in skills and technologies. Such activities and investments are hindered by a higher likelihood of blindness and hence fewer individuals specialize in law creation activities.
If that argument holds, then countries that (in the past) experienced high levels of UV radiation, where eye disease (specifically, cataracts) were more common, would have lower quality institutions in modern times. They use data on UV radiation that is derived from NASA, along with the World Bank Worldwide Governane Indicators (WGI) as their measure of institutional quality. The data do seem to support a relationship, as shown in Figure 1 from the paper:
And their regression analysis supports this. After controlling for a range of geographical variables (mean elevation, distance to nearest coast, terrain ruggedness, precipitation, whether the country is landlocked, a small island, and which continent the country is in):
...a one standard deviation (SD) increase in the intensity of UV-R (1 SD=0.778) leads to a 0.628 SD decrease in institutional quality (1 SD=0.262), all else equal.
The effect seems quite large:
For example, Papua New Guinea has a high UV-R intensity equal to 2.653 (max=3.285), and a poor institutional quality score of 0.272 (max=0.98). Based on the estimates... if Papua New Guinea instead experienced UV-R intensity similar to Estonia (0.514), the estimated institutional quality score would equal 0.725, a substantial rise. This is only slightly lower than Estonia’s current score (0.817), and is similar to those of neighboring Lithuania (0.736) and Latvia (0.710), with UV-R intensities similar to Estonia.
Indeed, possibly implausibly large. Could UV radiation really explain 83 percent of the difference in institutional quality between Papua New Guinea and Estonia? [*] It seems unlikely to me. However, in their robustness checks, Ang et al. go on to rule out a range of other things, including distance from the equator (absolute latitude), land area, agricultural suitability, the proportion of the population living in the tropics, and the number of frost days. They then show that cataract prevalence, unlike other disease prevalence, is negatively correlated with institutional quality, even when UV radiation is also in the econometric model. They conclude that:
...UV-R and institutional quality should be negatively related. Our results provide considerable support for this notion.
However, there are two problems remaining with this study. First, I don't really like their approach to the measurement of institutional quality:
Percentile rank data ranging from 0 to 100 for each country are used, where the country with the lowest ranked institutions is assigned a value of 0. The ranking scores for six abovementioned indicators are first averaged over the period 2006 to 2015 and are then combined into a composite index by taking their average. The data of this variable are then divided by 100 in order to give a measure that varies between 0 and 1, where a larger value signifies greater institutional quality.
I'm not sure what the average percentile ranking actually tells us. Country X can improve its ranking by improving its own institutional quality, or because the institutional quality of other countries improves, but slower than that of Country X. Although Ang et al. show that their results are robust to using other measures of institutional quality, they never show that the results are robust to using the underlying WGI scores rather than the ranking. That in itself should make us a little skeptical.
Second, Ang et al. never show that their results are robust to the inclusion of controls for colonial ties. Institutional quality has been shown to be associated with which colonial power (if any) historically controlled a country, along with the type of control (extractive or inclusive) that the colonial power exercised. This is the big omission, and is the basis of some of Acemoglu and Robinson's prior work. I suspect that the reason that UV radiation looks like it explains such a large proportion of differences in institutional quality is that UV radiation is correlated with colonial ties variables, which themselves are highly explanatory of the differences (economists refer to this as omitted variable bias). Ang et al. try to argue that omitted variable bias is not an issue, but it very much obviously is, when the existing literature suggests that there are important variables that have not been included.
So, I think we can file this study under interesting, but not convincing.
*****
[*] The difference in institutional quality between Estonia (0.817) and Papua New Guinea (0.272) is 0.545. If Papua New Guinea's institutional quality were raised to 0.725, the difference with Estonia would only be 0.092, an 83 percent decrease.
No comments:
Post a Comment