top of page
  • Lucas Rivero

Gross Data Garbage



The recording of national data has only been an expectation for a few decades. The thinking goes that if the state could better understand the needs of the people, the state would better be able to service the needs of its people. What a shame it was, then, that the entirety of the 1937 census report was, by the understanding of Stalin, ‘unsatisfactory’ and for the betterment of the state all those civil servants would be executed and the numbers ‘corrected’. Numbers matter. They represent to us what is true and evident and help us make better decisions in all aspects of our lives. What happens, then, when the numbers are manipulated? What happens when what once helped us to understand the health of a nation is as arbitrary as a thermometer that always shows 98 degrees even when one has a fever? Luis Martinez, in his paperHow Much Should We Trust the Dictator’s GDP Estimate,” sought to discover the truth about the GDP numbers published by autocratic countries.


China’s GDP figures are “man made.” These are Li Keqiang, the Premier of the People’s Republic of China from 2013 to 2023, own words in reference to the credibility of the official Chinese numbers. But aren’t all numbers ‘man-made’? Why is there some sort of cynical clarification? The numbers we use today are tied to the history of the second quarter of the twentieth century. With the end of WWII, the allies wanted to form a new standard to be able to better compare countries, for greater cooperation, understanding and coordination. They wanted a more free, more democratic world order and believed that math, as a practice of truth, could help in this endeavor. “Freedom is the freedom to say that two plus two make four. If that is granted, all else follows” This eventually culminated as the Nations’ System of National Accounts, or the SNA, in 1951. The SNA became the new standard and basis of measurement for the economies of countries. People could now point to new recorded measurements like GDP, inflation, unemployment and other numbers to gauge the economic health of their country.


This system worked for democracies because of checks and balances; incumbent politicians have limited access to manipulation of statistics because opposition, media and the other branches impose much higher costs for lying. The better the integrity of a nation’s democratic norms, the more useful the data becomes as a tool of accountability. However, in the case of autocracies, this logic is inverted. If there are no checks and balances, then there are limited costs to manipulating statistics and lying because no one can hold leaders accountable. The numbers then become a new set of tools for autocrats to manipulate the public. This is the argument of Martinez's paper: that the more autocratic the country, the more GDP will be exaggerated.


Martinez discovered this relation to be true by measuring the yearly growth of night lights in all types of regimes. He examined satellite images of countries at night and compared the growth rate towards the growth rate of GDP. The more growth, the more lights, and if countries are honest there should be a pattern different than that of dishonest countries. The results found a relatively similar average increase in lights of about 5% year by year for all countries. Yet for the same observed growth rate of night lights, the average growth rate of GDP varied between democratic and autocratic regimes with the most authoritarian regimes reporting between 15% and 30% higher GDP growth than democratic countries. For China specifically, Martinez discovered that the growth rate was approximately 30% more exaggerated from 1992 to 2008. If we take Luis’ model, an 8.73% reported growth rate becomes a growth rate of 5.32%. If we run the model to 2021 that puts Luis’ estimate to about 38.4% of the 2021 figures. That means that overall Chinese GDP would be about 60% smaller than reported in 2021. The second largest economy in the world is lying so much about its size that the entire economy of Japan, the third largest in the world, could approximately fit in this 60% range twice, or the economy of two Frances and two Italy could also approximately fit in this 60% range.


This sounds like a bombshell report. How can China be faking two whole Japans? Why should anyone trust Martinez’s numbers instead of the official numbers of the Chinese government? There is some truth to this question, because the numbers are a rough estimate based on night lights, but they are still more reliable and likely to be true than the Chinese government’s numbers. The paper demonstrates a clear relationship between a physical and easily observable thing such as night lights, which satellites already record, and that of reported GDP growth, which governments already record. Martinez double checked his results against geography, the economic structure of autocratic systems, and their development in general, and for all of these tests he got the same evident GDP growth. The only possible explanation for the discrepancy was that autocratic regimes do in fact overstate their GDP numbers. His paper also has been used at various seminars at the World Bank, the Central Bank of Chile and the University of Illinois, and published in the Journal of Political Economy, meaning it went through a lot of hurdles to be accepted. More than this, the recorded number of night lights can be corroborated by many other papers, and the reported GDP is public information. All Martinez did was test for a relationship and then test all of the potential things that could break the relationship and none of them did. It’s not Martinez’s word against the world, but the world’s own words against itself.


Comments


bottom of page