Via the Peterson Institute for International Economics, a look at North Korea’s economic growth:

Every so often we get reports that attempt to estimate how fast (or if) the North Korean economy is growing. Such exercises require a bit of ingenuity, creatively, guesstimates, and, regrettably often, the suspension of disbelief. A study(link is external) released by Hyundai Research Institute released last September estimated that per capita North Korean GDP rose to $1,013—the first time in HRI’s estimation that it surpassed the thousand-dollar mark.

The HRI model, first covered on this blog in 2012, differs from other North Korea GDP estimation models in that it utilizes infant mortality rates as an indicator for income levels. And because UN data on infant mortality is only released every five years, HRI incorporates grain outputs on top of the five-year infant mortality figures. HRI does this to compensate for some of the price structure issues that the Bank of Korea (BOK)’s North Korea GDP prediction model has, which Marc Noland has written about in some detail here. It goes without saying that neither the infant mortality estimates nor the grain production figures that HRI employs are estimated with the precision of say, the US Department of Commerce’s data on last month’s housing starts.

Given these basic data and estimation issues, it’s no surprise that the growth rate figures are wildly divergent across sources: Hyundai predicts GDP per capita growth in 2015 at 9 percent while BOK estimates a contraction of -1.1% (one of the first things I learned under Marc’s tutelage is to never trust a North Korea datum with a decimal point). Obviously both figures cannot both be correct.

The challenges of accurately computing North Korea’s GDP are many and are derived principally from a paucity of credible macroeconomic data. One of the areas where the BOK likely underestimates output is in services which are much more difficult to predict than a commodity like minerals. Thus, in 2015 even as mineral output decreased, a continuing shift to a service-based economy—which anecdotal evidence suggests is occurring—could make up for this gap.

The HRI model does not factor in trade but its impact could be partially picked up in the comparable country variable that the model includes. To generate this variable the model averages the characteristics of other underdeveloped countries with GDP per capita under $3,000. But North Korea is so unique that using an average comparable country variable may be more of a stretch than with other developing countries.

So we can’t be sure if North Korea’s economy grew in 2015; it probably did, but probably not by 9 percent. Although growth rates provide us with an understanding of how productivity is changing, if the initial GDP estimates in the model are widely divergent these growth rates won’t tell us much about the size of the economy as a whole. In fact, computing the HRI figures for the whole economy yields a GDP estimate of $25 billion while the BOK figures predict a real GDP of $27 billion. In a few years these estimates may be virtually identical. Each model overestimates and underestimates different things and the truth is probably somewhere in the middle.