The causality could well runs both ways, since plausibly education improves productivity and income, even as countries/people spend more on education as they get richer. To get around this problem of reverse causality, the Dieppe et al analysis we rely on basically regresses 1960-2018 growth rates on 1960 education levels and other candidate explanations like innovation or institutions (since obviously future growth rates cannot affect the past).
However, as I note in the report, this doesn’t eliminate the problem completely, since possibly 1960-2018 growth rates are autocorrelated with pre-1960 growth rates, which then influence 1960 education levels etc.
That said, my conclusion is that this is sufficiently unlikely given the theoretical and empirical case against consistently positive autocorrelation between past and future productivity growth:
In theory, copying technological/educational/managerial innovations etc and catch-up growth is just easier than inventing further new ideas.
Empirically, advanced economies underperform the global average of productivity growth while developing economies over-perform.
And if we don’t expect future growth rates to be positively correlated with past growth rates, then it wouldn’t be growth boosting education (through higher growth allowing more spending on education). And hence, we can be fairly confident that reverse causality is probably not driving the results. I discuss this all in greater detail in the report.
“To get around this problem of reverse causality, the Dieppe et al analysis we rely on basically regresses 1960-2018 growth rates on 1960 education levels and other candidate explanations like innovation or institutions (since obviously future growth rates cannot affect the past)”
“Endogeneity. The Bayesian approach used in the chapter can help to overcome ad hoc variable selection and the arbitrary omission of variables. Issues of interpretation remain, because many candidate explanatory variables—innovation, democracy, rule of law, trade, education, health, investment, and so on—are best seen as equilibrium outcomes. Because growth and the explanatory variables are jointly determined, it is hard to draw conclusions about causal effects, and persuasive instrumental variables are hard to find.”
I have made this question many times before, and unfortunately your answer is among the best I have received (at least you are aware and have an estimate, while non-causal).
Now, given that a hard answer is unavailable, I will state my impression: when you educate people beyond the absortion capabilities of an economy, you end up producing social unstability and intensify elite overpopulation. So, I believe educational intervention works, but only as long as it is not generic education that feeds inter elite competitions, but it is targeted to economic needs.
Currently, for developing countries, intenet access and education focused on complementarity with the needs of developed contries (to ride the likely wave of outsourcing) is probably the best option to create export opportunities, instead of frustrated graduates looking for a revolution.
The elite overproduction hypothesis is always interesting (not sure of robustness even in the context of the US and other advanced economies where the case for it would be strongest), but probably not a worry in the context of low income countries where it’s just primary/secondary education we’re looking at.
The causality could well runs both ways, since plausibly education improves productivity and income, even as countries/people spend more on education as they get richer. To get around this problem of reverse causality, the Dieppe et al analysis we rely on basically regresses 1960-2018 growth rates on 1960 education levels and other candidate explanations like innovation or institutions (since obviously future growth rates cannot affect the past).
However, as I note in the report, this doesn’t eliminate the problem completely, since possibly 1960-2018 growth rates are autocorrelated with pre-1960 growth rates, which then influence 1960 education levels etc.
That said, my conclusion is that this is sufficiently unlikely given the theoretical and empirical case against consistently positive autocorrelation between past and future productivity growth:
In theory, copying technological/educational/managerial innovations etc and catch-up growth is just easier than inventing further new ideas.
Empirically, advanced economies underperform the global average of productivity growth while developing economies over-perform.
And if we don’t expect future growth rates to be positively correlated with past growth rates, then it wouldn’t be growth boosting education (through higher growth allowing more spending on education). And hence, we can be fairly confident that reverse causality is probably not driving the results. I discuss this all in greater detail in the report.
“To get around this problem of reverse causality, the Dieppe et al analysis we rely on basically regresses 1960-2018 growth rates on 1960 education levels and other candidate explanations like innovation or institutions (since obviously future growth rates cannot affect the past)”
https://openknowledge.worldbank.org/handle/10986/34015
In page 124 they say:
“Endogeneity. The Bayesian approach used in the chapter can help to overcome ad hoc variable selection and the arbitrary omission of variables. Issues of interpretation remain, because many candidate explanatory variables—innovation, democracy, rule of law, trade, education, health, investment, and so on—are best seen as equilibrium outcomes. Because growth and the explanatory variables are jointly determined, it is hard to draw conclusions about causal effects, and persuasive instrumental variables are hard to find.”
I have made this question many times before, and unfortunately your answer is among the best I have received (at least you are aware and have an estimate, while non-causal).
Now, given that a hard answer is unavailable, I will state my impression: when you educate people beyond the absortion capabilities of an economy, you end up producing social unstability and intensify elite overpopulation. So, I believe educational intervention works, but only as long as it is not generic education that feeds inter elite competitions, but it is targeted to economic needs.
Currently, for developing countries, intenet access and education focused on complementarity with the needs of developed contries (to ride the likely wave of outsourcing) is probably the best option to create export opportunities, instead of frustrated graduates looking for a revolution.
The elite overproduction hypothesis is always interesting (not sure of robustness even in the context of the US and other advanced economies where the case for it would be strongest), but probably not a worry in the context of low income countries where it’s just primary/secondary education we’re looking at.