Nightlight data and fuzzy math

This is almost a little too pat to be entirely convincing… but it makes sense that authoritarian regimes would have more ability, if not more incentive, to manipulate, exaggerate and outright make up statistics than democracies:

China, Russia and other authoritarian countries inflate their official GDP figures by anywhere from 15 to 30 percent in a given year, according to a new analysis of a quarter-century of satellite data. […]

“The key question that the paper tries to tackle is whether the checks and balances provided by democracy are able to constrain governments’ desire to manipulate information or, more specifically, their desire to exaggerate how well the economy is doing,” Martinez said via email. “The way I try to answer the question above is by comparing GDP (a self-reported indicator, prone to manipulation) and nighttime lights (recorded by satellites from outer space and much harder to manipulate) as measures of economic activity.” […]

For the world’s freest democracies — places such as the United States, Canada and Western Europe — a 10 percent increase in the average intensity of nighttime lighting in a given year correlated with, on average, a 2.4 percent increase in year-over-year GDP. Less free and open countries, however, reported larger GDP gains for the same percent change in nighttime lighting. And the least-free countries of all showed huge increases in annual GDP relative to the freest countries, working out to between a half and a full percentage point of extra GDP for the same light increase.

One question that comes to mind: Is it possible that countries where personal consumption accounts for a relatively small share of GDP (e.g. China) are able to grow more “darkly” than countries where personal consumption plays a larger role (e.g. the US)? The idea being that factories might give off less light per capita at night than, say, driving to a restaurant or staying up late watching Netflix in a brightly lit house. (I have absolutely no idea whether this is the case.) Likewise for economies where the services sector plays a greater role (think of all those computer screens and florescent-lit offices). Glancing at the paper, it appears the author has already accounted for just those possibilities, and that the autocracy bias is still strong even when controlling for differences in economic structure.

Of course, it’s not exactly surprising that China manipulates economic data rather liberally; this is a country where the official population figure could be off by 90 million.

Leave a Reply

Your email address will not be published. Required fields are marked *