The problem that Magee and Doces tackle is that it’s hard to figure out when regimes are being honest or dishonest about their rates of economic growth, since it’s the regimes themselves that are compiling the statistics. It’s hard to measure how honest or dishonest they are, if all you have to go on are their own numbers. This means that researchers need to find some kind of independent indicator of economic growth, which governments will either be less inclined or unable to manipulate. Magee and Doces argue that one such indicator is satellite images of nighttime lights. As the economy grows, you may expect to see more lights at night (e.g. as cities expand etc). And indeed, research suggests that there’s a very strong correlation between economic growth and nighttime lights, meaning that the latter is a good indicator of the former. Furthermore, it’s an indicator that is unlikely to be manipulated by governments.
Magee and Doces look at the relationship between reported growth and nights at light and find a very clear pattern. The graph below shows this relationship for different countries – autocracies are the big red dots. Most of the dots are above the regression line, which means that most autocracies report higher growth levels to the World Bank than you’d expect given the intensity of lights at night. This suggests that they’re exaggerating their growth numbers.
The two countries with the biggest difference between their reported growth and their actual growth (as best as you can tell from the intensity of nighttime lights) are China (although the discrepancy was considerably larger in the mid-1990s than now) and Myanmar. More broadly:
if democracies report their GDP growth rates truthfully, then dictatorships overstate their yearly growth rate by about 1.5 percentage points on average. If democracies also overstate their true growth rates, then dictatorships exaggerate their yearly growth statistics by about 1.5 percentage points more than do democracies.
The authors conclude:
the existing literature on economic growth overestimates the impact of dictatorships because it relies on statistics that are reported to international organizations, and as we show, dictatorships tend to exaggerate their growth. Accounting for the fact that authoritarian regimes overstate growth slightly diminishes the effect of these regimes on long-run economic growth. In light of this point, much of the evidence showing growth benefits associated with authoritarian regimes is less compelling and the case for democracy looks better than before.