On the Historical Association between National IQ and GDP per capita

A remarkable, unquestioned assumption in (1–3) and subsequent studies measuring the association between national average Intellectual Quotients (IQ) and Gross Domestic Products (GDP) per capita is that a supposedly immutable1 genetic2 factor (IQ) may be correlated with a markedly fluctuant one (the wealth of nations). This short paper questions this assumption and presents the following results:


Introduction and summary of results
A remarkable, unquestioned assumption in (1)(2)(3) and subsequent studies measuring the association between national average Intellectual Quotients (IQ) and Gross Domestic Products (GDP) per capita is that a supposedly immutable 1 genetic 2 factor (IQ) may be correlated with a markedly fluctuant one (the wealth of nations). This short paper questions this assumption and presents the following results: 1. Using historical GDP per capita data produced by the Maddison project (5,6), we find that, over history, the (Pearson product-moment) correlation coefficient (r) between average IQ and GDP per capita is highly variable and ranges from strong negative values to strong positive values. The correlation between national IQ and GDP per capita is a snapshot of the world order at some point in time, and historical data allow us to identify several other eras.
2. The reported positive correlation between national average IQ scores and GDP per capita thus only concerns "today's GDP". However, today's GDP was never difficult to explain and predict in the first place. We show that arbitrary ad-hoc scores based on a country's continental location present a more significant correlation with contemporary GDP per capita. As an economic variable, the predictive value of IQ is thus lesser than that of the common sense observation that North-America is, currently, richer than Europe which is in turn richer than Africa, etc.
3. We conclude this paper by questioning the purpose of IQ studies in Macroeconomics. If this purpose is explaining the wealth of nations then confounding variables such as literacy cannot be ignored, and the Pearson productmoment correlation cannot be considered as a sole criterion to draw causal conclusions. If, on the other hand, the purpose is predicting the wealth of nations then simply using the geographical location of countries, which is no less circular than the use of IQ due to the confounding role of literacy, would be a better predictor of GDP.

Related work and data sources
General knowledge regarding average national Intellectual Quotients (IQ) and their association with economic outcomes is largely based on two books by Richard Lynn and Tatu Vanhanen, "IQ and the Wealth of Nations" (1) and its followup "IQ and Global Inequality" (2), as well as a dataset (3) by the same authors. With these publications, IQ gained entry into macroeconomic research and started being considered a valid independent variable to explain and predict the Gross Domestic Product (GDP) of nations, because of the high reported correlation of .82 3 . Since then, the confounding role of literacy in the association between IQ and GDP has been thoroughly established. Indeed, Marks has shown that IQ variations across time and race are explained by literacy differences (7) and that literacy, not intelligence, is in fact the key predictive factor for economic development (8).
A recent (June 26, 2020) retraction of a publication by Clark et al. in Psychological Science (9), based on data from (3) notes that the above data are "plagued by lack of representativeness of the samples, questionable support for some of the measures, an excess of researcher degrees of freedom, and concern about the vulnerability of the data to bias".
In this work, we overlook these shortcomings, as well as inherent shortcomings of IQ tests as a measure of an individual's intelligence (10), and question the idea that a fairly static racial factor is associated with the historically variable variable that is GDP per capita.

D R A F T
The beginning of new cycles can be linked to important historical changes (industrial revolution, postmodernism starting after the second World War, decolonization).

Modern GDP per capita was never difficult to predict
We divide the world in 13 regions and assign an ad-hoc integer score from 1 to 10 reflecting the wealth of the region (1 for Subsaharan Africa, 10 for North America), according to table 2. Each country is assigned the score of the region it belongs to. Figure 3 compares the coefficient of correlation of this lazy ad-hoc score with that of national IQ.

Conclusion
The purpose of IQ research in Macroeconomics is unclear. If it is an attempt at explaining the wealth of nations, e.g. to predict the value of investment in increasing intelligence, then this type of analysis cannot avoid controlling for literacy rates and other confounding variables (nourishment, health, etc.). If on the other hand, it is an attempt at predicting the wealth of nations based on an independent variable (notwithstanding the poor test-retest correlation of IQ test), e.g. to inform immigration policies, with correlation as the only criterion, then assigning lazy 1 to 10 scores to different continents based on their current wealth would be a better model than national IQ. The wealth of nations (and of anyone for that matter) is best studied as a time-series. Any association with a static variable is bound to be uninformative.