Body
In earlier posts we reviewed analyses by Dan Goodkind, Loraine West, and Peter Johnson and by the South Korean government on famine mortality. These studies used the North Korean censuses of 1993 and 2008 to generate new estimates of deaths during the 1990s, and concluded that famine deaths were lower than previously thought (including in earlier work by Goodkind and West themselves). We expressed skepticism about these analyses not because they are technically inadequate but because there are good reasons to doubt the veracity of the North Korean censuses on which they are based.
Now Thomas Spoorenberg and Daniel Schwekendiek have thrown their hats into the ring with a paper in the Population Development Review again using the 1993 and 2008 censuses to derive estimates of famine deaths. What they do is essentially a calibration exercise, and it goes as follows. We have observations of the population 15 years apart that shows a net increase in population size. The estimation of famine deaths comes down to a series of assumptions: what the pattern of births and deaths were during the entire intercensal period; what one assumes about the incidence of births and deaths during the famine period, which they specify as 1993-2000; and what occurred after the famine period.
Spoorenberg and Schwekendiek generate several scenarios based on North Korean official data but calibrating mortality patterns from China’s Great Leap Forward famine. They then generate five counterfactual projections which in turn incorporate counterfactual assumptions about what North Korean life expectancies might have been under the following scenarios:
(1) assuming no change in life expectancies during the famine period using parameters derived from North Korean official data;
(2) assuming life expectancy had changed at the same rate as it did in South Korea during 1993-2000 (i.e. during the famine period);
(3) assuming life expectancy during the famine period had changed at the same rate as it did in China 1993-2000;
(4) assuming life expectancy had changed at the same rate as it did in South Korea during the entire intercensal period 1993-2008;
and finally (5) if life expectancy had changed at the same rate as it did in China during the entire intercensal period 1993-2008.
As the authors observe, counterfactual (1) essentially establishes a floor for famine deaths, since it pushes the incidence of the overall shortfall in projected population into the post-famine period. Conversely, however, if one believes that life expectancy should have grown more rapidly in North Korea than it did in South Korea or China during this period then Spoorenberg and Schwekendiek are producing low-ball estimates. We might plausibly expect life expectancy in North Korea to grow more rapidly than in either South Korea or China because North Korea was an extremely poor country at the outset of the period, and as a result had much more room for improvement than its richer neighbors where life expectancy was already high.
So what do these calculations yield? In the first case, they get an estimate of 237,000 famine deaths. This would be in line with the 1999 public statement of a North Korean official that 220,000 famine deaths had occurred between 1995 and 1998. The second and third counterfactuals yield estimates of roughly 320,000 and 420,000 famine deaths, respectively. The final two counterfactuals are interesting since they effectively account for excess deaths due to continuing economic and social deterioration during the post-famine period, and generate estimates of between 600,000 and 850,000 excess deaths for the whole intercensal period.
An extremely interesting conclusion follows. If these latter estimates are accepted, they imply a greater human toll during the period following the famine than during the famine itself! This result comes from some combination of the aftermath of the famine--the long-lived results of chronic malnutrition--and the fact that the economy failed to grow over this period, resulting in a stagnation in life expectancy.
Needless to say, this whole exercise is fraught with arbitrary assumptions and we are still recovering from the howler repeated several times in the paper that Nick Eberstadt believes that North Korea’s official population data are “generally reliable” (sorry, guys, he has practically built a career arguing that the data are systemically falsified).
Fourteen years after the fact the mortality estimation game basically amounts to an intellectual exercise of interest to a handful of people like us with a professional stake in it. But while we are skeptical of the specific conclusions, Spoorenberg and Schwekendiek’s paper does a useful service in reminding us that while the famine may have ended in 1998, its effects continue to reverberate through North Korean society. Just last week we ran a post on continuing research into the long-term effects of famine exposure on infants. If one accepts their bottom line result, it implies that the last 12 years have been a bigger catastrophe for North Korea than the famine itself.
We are not ready to sign on to that conclusion but it does serve as a useful reminder, especially in a world in which there is a tendency to focus on the day-to-day diplomatic maneuvering of "high politics." The North Korean regime imposes terrible costs on its own population. And those profound harms are exacted day in and day out, missile launch or not.
