Discussion: On the Study of Population Structure of Democratic People’s Republic of Korea

Abstract

This excellent paper by Jeon et al. (2015) presents a demographic analysis of data from North Korea. Their key concerns are with population age structure as well as patterns of mortality between the censuses of 1993 and 2008. Their assemblage of clues from the two censuses, as well as other statistics and official estimates, help to improve our understanding of North Korea. The 1993 and 2008 census counts by age and sex provide two “bookend” estimates of population structure. The authors then use the demographic balancing equation - the fundamental accounting system of demographic analysis - which requires that changes between the two bookends be due to births, deaths, and net migration. Of course, annual statistics for these components of demographic change may be unavailable, and the data that is available (including the reported census counts) may not be accurate. Thus, the challenge for demographers is to decide how to fit together (or adjust) the available demographic evidence as required by the balancing equation. When there is a lack of information or uncertainty about data quality, there are multiple ways in which the evidence can be fit together. An important concern that anchors their analysis is the estimation of “excess deaths” caused by the Great Famine in North Korea during the late 1990s. Shortly after the famine, experts speculated that it caused anywhere from one to three million deaths that otherwise would not have occurred. Goodkind and West (2001) then proposed a narrower range of 600 thousand to 1 million excess deaths based on indirect inferences - the lower estimate based on child malnutrition survey data and the upper estimate based on Chinese mortality during a famine following the Great Leap Forward. Following the 2008 census, which allowed for revised estimates based on intercensal analysis, the proposed range was reduced to about 500–600 thousand (Goodkind et al., 2011). Some experts suggested that this revised range was too low and that the original range extending to one million was more accurate (Noland, 2011). Conversely, others claimed the revised range was too high. For instance, an alternate demographic analysis using a different set of assumptions concluded that the range should be 240–420 thousand (Spoorenberg and Schwekendiek, 2012). Yet these challenges to the revised estimates by Goodkind, West, and Johnson are themselves highly questionable. There is no reason to believe that North Korean mortality should have followed