Peterson Institute publications
The Peterson Institute for International Economics is a private, nonprofit, nonpartisan
research institution devoted to the study of international economic policy. More › ›
RSS News Feed Search

Working Paper 99-1

Rigorous Speculation: The Collapse and Revival of the North Korean Economy

by Marcus Noland, Peterson Institute for International Economics
and Sherman Robinson, Peterson Institute for International Economics
and Tao Wang, Peterson Institute for International Economics

© Institute for International Economics.

We would like to thank seminar participants at Hitotsubashi University, the Institute for Global Economics, the Korea Institute for International Economic Policy, and Korea University for helpful comments on earlier presentations of this material. Mina Kim provided research assistance.



In this paper we use cross-entropy estimation techniques to construct the underlying data base for a computable general equilibrium model (CGE) of the North Korean economy, starting from incomplete data ridden with gross measurement errors. The cross-entropy estimation approach is powerful and flexible, allowing us to make full use of what information we have in whatever form. CGE modeling forces internal consistency. The end product is a model that incorporates fragmentary information in a rigorous way and allows us to examine the implications of a number of alternative scenarios including rehabilitation of flood-affected lands, liberalization of the international trade regime, and military demobilization.

North Korea is experiencing a famine. Its economy is characterized by systemic distortions and comparative disadvantage in the production of grains. As a consequence, the potential pay-offs to economy-wide reforms, even defined narrowly in terms of domestic food availability, dwarf more targeted attempts to raise agricultural productivity. Too many, this finding that a famine might be better addressed by the export of manufactures than the recovery of flood-damaged lands—is a striking and counter-intuitive result. Moreover, we find that if reforms were undertaken, the country could generate a significant additional "peace dividend" by partially demobilizing its enormous military.



As well as can be ascertained, North Korea is now into its eighth year of economic decline. It has been facing food shortages at least since the early 1990s and is well into a famine of unknown magnitude. Despite the desperate situation internally, the government maintains the most militarized society on earth, with more than one million men (and increasingly women) under arms and devoting an estimated 25 percent of GDP devoted to military expenditures (US ACDA 1997). Economic reform could have enormous benefits in this highly distorted economy, especially in light of the country's dire situation. Yet, at the same time, the effects of reform—a significant increase in exposure to international trade and investment (much of this with South Korea and Japan, two countries with which North Korea maintains problematic relations) and huge changes in the composition of output, involving literally millions of workers changing employment—could be expected to have enormous political implications, possibly presenting large, perhaps insurmountable, obstacles to reform under the current regime.

The paucity of reliable statistical information about North Korea has bedeviled researchers, and, as a consequence, studies of the North Korean economy have tended toward either uncritical recitations of official statistics or compendia of anecdotes. In this paper we use cross-entropy estimation techniques to construct the underlying data base for a computable general equilibrium model (CGE) of the North Korean economy, starting from incomplete data ridden with gross measurement errors. The cross-entropy estimation approach is powerful and flexible, allowing us to make full use of what information we have in whatever form. CGE modeling forces internal consistency. The end product is a model that incorporates fragmentary information in a rigorous way and allows us to examine the implications of a number of alternative scenarios including rehabilitation of flood-affected lands, liberalization of the international trade regime, and military demobilization. We do not consider the likelihood of the current regime undertaking any of these actions, or, indeed, any significant policy changes at all. Rather we simply examine the possible implications of various alternative actions.

Although we apply this approach to a reclusive Stalinist regime, in principle the same techniques can be applied to other situations in which economic data is fragmentary and/or of questionable reliability—a situation frequently encountered in developing countries.

To preview the results, we confirm that the North Korean economy is extraordinarily distorted. Due to the large, systemic nature of these distortions and North Korea's comparative disadvantage in the production of grains, the potential pay-offs to economy-wide reforms, even defined narrowly in terms of domestic food availability, dwarf more targeted attempts to raise agricultural productivity. Under reform domestic production of food declines, but human survival requirements are easily met through imports. In contrast, flood rehabilitation leads to an increase in domestic food production, but this increase falls short of human survival requirements. To many, this finding—that a famine might be better addressed by the export of manufactures than the recovery of flood-damaged lands—is a striking and counter-intuitive result. Moreover, we find that if reforms were undertaken, the country could generate a significant additional "peace dividend" by partially demobilizing its armed forces and redeploying its soldiers to more economically productive uses.


Organization of the North Korean Economy

Since North Korea's inception, property rights have resided primarily with the state and resource allocation has largely been carried out through central planning, not markets. The roles of money, prices, and other familiar institutional features of market economies are still severely circumscribed. The distinguishing feature of North Korea has been the extremes to which central planning has been taken.

The organizing principle of the North Korean economy is the ideology of juche or national self-reliance. The result has been the development of the world's most autarckic economy. The international trade share (exports plus imports) is around 12 percent of GDP, well below the 50-55 percent observed in South Korea, and a fraction of the trade share that North Korea would exhibit if it were a "normal" country of its size and structure. The emphasis on self-reliance has been so great that North Korea never joined its communist brethren in the Council for Mutual Economic Assistance (CMEA), and its own central plans have been timed to frustrate any linkages with those of the other planned economies (Hwang 1993).

In North Korea, economic assets are predominately in state hands. The Constitution establishes two basic categories of economic organization (state-owned enterprises and social cooperatives) and appears to establish the goal of nationalizing the co-ops. The state retains ultimate property rights while the co-ops are granted usage rights.1 A Constitional revision promulgated in August 1998 mentions "private property" (Article 24), "material incentives" (Article 32), and "cost, price, and profit" (Article 33) in an otherwise thoroughly orthodox elaboration of a planned, socialist, self-reliant, juche economy. Whether the mention of market concepts in the Constitutional revision signals any move toward reform is unclear.2

The central plan, specified in physical terms, is formed iteratively by the State Planning Commission (the central authority), the Office of State Affairs, and local committees from each plant and enterprise. Prices, set by the central planners to reflect the cost of labor inputs, play an accounting, not a resource allocation, role. There are three sets of prices, roughly corresponding to three types of markets—state-run, co-op, and farmers. These prices are applied depending on the type of transaction. For example, state-run markets, which handle mostly industrial goods, transact at "wholesale prices". A transaction between a state enterprise and a co-op would be at "industrial wholesale prices", which is the "wholesale price" plus a transactions revenue tax wedge.3 The third price category is the retail price charged at commercial outlets, where queuing is apparently the norm. In addition, there are limited farmers' and citizens' markets in which prices presumably reflect scarcity values, as well as a black market in consumer goods.4

All financial transactions through the plan are intermediated by designated state-owned banks as a method of control.5 In any event, money is basically only a means of exchange since prices are irrational and resource allocation is bureaucratic. Limited interest-bearing accounts exist, with rates set by the planners, but these play no real role in determining the rate of capital accumulation, which is determined by the planners. Likewise, interest rates play no allocation role, as capital is allocated by the central planners through the government-controlled financial institutions in the form of investment grants (Hwang 1993). Monetary policy has been erratic: the government has been through two currency reforms in the past decade, foreign exchange certificates have been circulating in parallel to domestic North Korean won notes, and recently the government has established a dual exchange rate system under which inside the Rajin-Sonbong free trade zone the won is worth approximately one percent of its official value outside the zone.

A second, parallel, military economy exists outside the central plan. Its origins lie in a military modernization program adopted at the fifth plenary session of the Fourth Central Committee of the Korean Workers' Party in 1966 following the intensification of the Sino-Soviet split (Koo 1992; Namkoong and Yoo 1994). This parallel economy is highly secretive, even internally. According to Eberstadt (1994), the State Planning Commission reportedly was deprived of access to information on it in the early 1970s, and meetings with outsiders are not held in the usual government ministries, but rather in unofficial locations such as hotels etc. (DKIH 1991). This parallel economy amounts to autarky-within-autarky: the military maintains a completely integrated economic system from farms and mines on up through facilities that manufacture uniforms and weapons, with separate administrative structures, including foreign trade firms. DKIH (1991) estimates that half of the army is engaged in what elsewhere would be civilian economic activities—an estimate that has also been supported by a former high-ranking US intelligence official.


Statistical Review

At the time of the division of the peninsula, North Korea had relatively more physical and human capital than South Korea, was relatively more industrialized, and had higher per capita income. Much of the physical capital stock was destroyed during the Korean War, which was accompanied by mass population movements as well. Nevertheless, the conventional wisdom is that per capita income in North Korea exceeded that of South Korea well into the 1970s.6

In all likelihood, the North Korean economy was already encountering significant difficulties in making the transition from extensive to intensive development by this time. In the aftermath of the first oil crisis, North Korea, among others, borrowed significantly from foreign banks recycling their petrodollar receipts.7 North Korea was unable (or unwilling) to pay off these loans and defaulted, effectively foreclosing its access to international capital markets.8 Eberstadt (N.d.) points out that North Korean capital goods imports never reattained their mid-1970s values.

Kim Il-Sung's decision to default on his Western creditors left North Korea highly reliant on the USSR as its sole significant source of economic support.9 But by the mid-1980s the Soviets had become disenchanted and began pressuring the North Koreans to begin repaying their accumulated debts. In 1985 the USSR and North Korea agreed on a repayment schedule, and, according to the US Central Intelligence Agency, by 1987 net flows had turned negative—that is, North Korean repayments exceeded the value of new Soviet assistance.

These events were followed in short-order by the break-up of the Soviet Union and the collapse of the Eastern Bloc. In 1990, Russia announced that it would no longer supply North Korea with subsidized oil, and the North responded by suspending repayments. Eberstadt (1995) estimates that in 1991 North Korea suffered a trade shock equivalent to 40 percent of total imports due to disengagement with the Newly Independent States, and by 1993 imports from Russia had fallen to less than a tenth of their earlier levels.

These external shocks were compounded by a series of natural disasters. Although as with everything else North Korean, controversy surrounds the precise timing and magnitude, the consensus is that agricultural production peaked around 1989 and has fallen significantly since.10

As early as 1991, publicly available evidence suggested a worsening food availability problem in North Korea, and by 1994 North Korean radio broadcasts had admitted the existence of hunger.11 In July and August 1995—after North Korea had asked for and had begun receiving international food assistance—the country suffered catastrophic floods.12 The floods of 1995 were followed by more, though less severe, floods in July 1996, and in the following year the country was hit with a drought. Although these natural disasters exacerbated the food crisis, agriculture had been in secular decline since the beginning of the decade. Even without the disasters, North Korea would have entered the mid-1990s with a substantial food deficit.

Historically, output has been measured on a material product basis, which is basically incompatible with the system of national accounts (SNA) used in market economies. Net material product covers value-added in the material product sectors (manufacturing, agriculture, construction, commodity transportation, productive communication, productive commerce, and a few others). The sum of sectoral material products multiplied by prices yields the gross output value of social production.13

The South Korean central bank data indicate that 1990 was the most recent year that the North Korean economy registered positive growth (figure 1).14 Recently the UNDP has been working with the government of North Korea to construct standard national accounts. Some data were released to an IMF mission that visited North Korea in October 1997 (IMF 1997). These data appear to reflect the economy under the control of the central planners (that is, they exclude the military and activities in the informal sector), and indicate that the fall in output began in 1994. Although the quantitative estimates contained in these two sets of figures differ markedly, they are qualitatively consistent—the North Korean economy is shrinking.

Table 1 presents the official North Korean figures on the composition of output. What the data suggest is that the economy has collapsed around agriculture—that is, the fall in agricultural output has actually been less dramatic than the output decline in other sectors.15 According to these figures, industrial output fell by nearly two-thirds between 1992 and 1996, and construction activity declined by almost as much. Taking construction as a proxy for investment, investment may well have fallen below replacement level and the capital stock may have been shrinking. This notion is supported by the government expenditure data reported in table 2. Taking the economic development category as a proxy, investment has fallen by more than half from the values the North Korean government announced in the early 1990s.16 This point is further reinforced if one believes that certain military or military-related expenditures are hidden in the economic development budget. Estimates of North Korean military manpower and equipment do not show anything like this decline over the relevant period. Indeed, US and South Korean defense ministry figures show a slight increase in North Korean military deployment during this period. These figures imply that the nonmilitary part of the economy is being severely squeezed.

Two other things stand out in table 2. First, for 1993, the government reports expenditures and receipts far larger than GDP. This would appear to violate the basic precepts of national income accounting. Second, it reports expenditures on defense that are far smaller than normally cited. Indeed, data on labor force participation provided by the North Korean authorities appear to omit the military entirely. All in all, these figures should probably be taken with very large grains of salt.

Data on government revenues are reported in table 3. The largest single source of revenue is turnover taxes, which is typical in centrally planned economies (CPEs).17 Profits from state enterprises is the next largest source of government revenue. Excluding the problematic data for 1993, it appears that the government sector accounts for roughly 90 percent of national income, an extraordinarily high figure even by CPE standards.


International Economic Relations

Although the government of North Korea has released budget figures in the past, external trade statistics have been considered a state secret. In principle, "mirror statistics" reported by North Korea's trade and investment partners can be used to deduce North Korea's external transactions. North Korea's largest export sectors are apparel and marine products, with other exports tending to be concentrated in light manufacturing, some natural resources, and iron and steel. The largest import categories are petroleum, cereals, and textile yarn. Broadly speaking, it appears that North Korea imports energy, food, and capital goods.18 The US Arms Control and Disarmament agency reports a small volume of arms trade, though it is unclear whether this trade is omitted from the GATT statistics or is misclassified therein.19

Aggregate trade figures derived from the IMF's Direction of Trade Statistics and other sources indicate that China is by far North Korea's main trade partner, and has allowed the North Koreans to run annual bilateral deficits of approximately one half billion dollars (figure 2). The prominence of China in North Korea's trade would be even larger if aid, smuggling, and barter transactions were included in the calculations, though quantity, value, and concessional components of Chinese trade are unclear.20 Following China, North Korea's other largest trade partners are Japan, South Korea, Russia, and Germany.21 Most of the trade between North and South Korea consists of apparel manufactured on consignment and then transshipped through China. Noticeably absent from the list of trade partners is the United States, which effectively maintains an embargo against North Korea.

Two points stand out from the aggregate trade figures. First, reported trade volumes have not been growing, reinforcing the impression of an economy in secular decline. Second, North Korea has been running chronic trade deficits. These deficits must be financed in some way, but since a series of loan defaults in the 1970s, North Korea has been effectively cut off from international capital markets. If North Korea's trade with South Korea and China (which has permitted North Korea to accumulate large arrears in its trade account) is interpreted as politically determined, then these two countries would appear to finance nearly two-thirds of North Korea's trade deficit. Much of the rest may be financed by illicit activities such as drug and ivory trafficking and counterfeiting, (North Korean diplomats have been expelled from host countries for all of these offenses) as well as private remittances, particularly from Japan.22

Recently, North Korea has obtained significant international assistance to deal with its food shortage: bilaterally from China and multilaterally through the World Food Program from other countries.23 Similarly, North Korea has been receiving energy assistance in the form of oil imports and assistance with nuclear reactor construction through the Korea Energy Development Organization (KEDO), a multinational consortium established as part of a 1994 nuclear agreement with the United States.

A question naturally arises: what would the North Korean economy look like if it were to undergo successful reform? It is plausible to anticipate that there would be a reallocation of factors according to comparative advantage and, were reform accompanied by a reduction of political hostilities with the South, there could be a significant demobilization of the military followed by redeployment of productive factors to alternative uses. Based on the experience of other transitional economies, one would expect a significant reorientation of international trade flows away from socialist allies, and toward natural trading partners. One way to get a sense of how North Korea might look as a "normal" country is to use a standard "gravity" model of bilateral trade to simulate its post-reform trade pattern. Frankel and Wei (1995) report such a model for 1990 which characterizes the volume of trade as a function of country size, income level, proximity, and other factors. North Korean values of these explanatory figures were then substituted into the gravity model regression to generate North Korea's "natural" pattern and volume of trade.24

According to the results reported in table 4, North Korea's natural trade partners would be South Korea, Japan, China, and the United States, in that order. South Korea and Japan alone would account for nearly two-thirds of North Korea's trade. Moreover, the share of international trade in national income would roughly quadruple.

Noland (1996) analyzed North Korea's prospective comparative advantage using disaggregated trade and investment data for North Korea and its two largest prospective trade partners, South Korea and Japan. First, using North and South Korean trade data, 465 sectors were put through a series of five filters.25 Twenty-five of the original 465 sectors, largely in primary products such as fish and minerals, as well as manufactured goods such as textiles and apparel, met all five criteria for prospective comparative advantage (Noland 1996, table 16).26 A second calculation, this time using North Korea and Japanese data and a similar set of criteria, yielded an additional twelve sectors, mostly natural resource-based (Noland 1996, table 17).27

Finally to address the issue of North Korea's current exports not providing a useful signal of its future exports, the data were re-analyzed focusing exclusively on the South Korean and Japanese data.28 An additional fourteen sectors, nearly all light manufacturing industries, met these criteria (Noland 1996, table 18). In summary, an analysis of the North Korean, South Korean, and Japanese trade and investment data suggest that North Korea's sectors of prospective comparative advantage would largely be in primary products where North Korea's natural resources convey a comparative advantage, and light manufacturing industries which have been declining in Japan and South Korea, but could be competitive in lower-wage North Korea.29


A General Equilibrium Perspective on Reform

As noted earlier, natural disasters have reduced arable land in North Korea, exacerbating the food shortage. Land recovery could be expected to lead to increased agricultural output and employment. From the results obtained with the gravity model, one would expect trade reform to result in a large increase in trade flows in North Korea and a significant shift in the composition of output away from heavy and toward light industry. This shift would imply major adjustments for many North Korean workers and enterprises. At the same time, expansion would be expected to take place mainly in labor intensive industries, which could facilitate the transition by absorbing displaced workers. This labor absorption would be of paramount importance were military demobilization to release large number of workers for nonmilitary activities.

The economy-wide repercussions of land recovery, trade liberalization, and military demoblization have been analyzed using a CGE model which has a standard neoclassical specification, except that the model incorporates severe quantity controls in exports and imports with concomitant distortions in domestic product and factor markets. The markets for goods, factors, and foreign exchange are assumed to respond to changing demand and supply conditions, which in turn are affected by government policies, the external environment, and other exogenous influences. The model can be considered medium-to-long run in that all factors are assumed to be intersectorally mobile. It is Walrasian in that only relative prices matter. Sectoral product prices, factor prices, and the exchange rate are determined relative to an aggregate consumer price index, which defines the numeraire.30

The model has eleven sectors: rice, maize, other agriculture/forest/fisheries, mining, light manufacturing, industrial intermediates, capital goods, construction, public administration, the military, and services. There are three "demanders": a single aggregate household that buys consumer goods, a government that spends on goods and public administration, and an aggregate capital account that purchases investment goods. The government is the sole, and completely price inelastic, demander of military services. All goods and services are traded internationally with the exceptions of construction, public administration, and the military. Domestically produced and traded goods are specified as imperfect substitutes, which provides for a realistic continuum of "tradability" and allows for two-way intersectoral trade.

Primary factors of production are three types of land, capital, agricultural labor, high-skill urban labor, and low-skill urban labor. Aggregate production functions were estimated for aggregate capital and labor using data reported in Hwang (1993) and Y. S. Lee (1994). The results are remarkably robust and plausible given the quality of the underlying data. Constant elasticity of substitution specifications yielded estimates of the substitutability between capital and labor of around unity. The hypothesis that the aggregate production function was Cobb-Douglas could not be rejected. In most specifications, North Korea exhibited slightly negative total factor productivity growth, which is typical of many pre-reform socialist economies. In the CGE model, sectoral production technology is represented by a set of Cobb-Douglas functions of the primary factors, with the exception of the mining and military sectors which use constat elasticity of substitution (CES) functions with lower substitution elasticities. Intermediate inputs are demanded according to Leontief, fixed input-output coefficients. Labor and capital are intersectorally mobile; land is specific to agriculture but mobile within the three agricultural sectors. Migration is permitted between rural and urban low-skill labor markets.

The Loss and Recovery of Arable Land

Total arable land in North Korea is 1.85 million hectares. Following FAO (1997) and UNDP (1998) we distinguish three types of land: high quality land (0.4 million hectares) that is permanently irrigated, medium quality land (0.65 million hectares) that is not permanently irrigated but capable of supporting rice production, and nonirrigated lower quality land (0.8 million hectares) suitable only for the production of other cereals. Since the late 1980's, rice and maize have been produced on 0.58 million and 0.6 million hectares of land, respectively. The remaining arable land is used in producing other agricultural produce. All rice production occurs on irrigated, high quality or medium quality land. Of the 0.6 million hectares allocated for maize, only one third is irrigated. Most of the rest is grown on medium quality or low quality land that does not have permanent irrigation.

By using the county data from UNDP (1998), we estimate land lost in the 1995-96 floods county by county and class by class. About 15 percent of arable land was destroyed, with the incidence of flooding being higher on high quality land (28 percent) than on medium and low quality land (13 percent). We use the post-flood land distribution as the base in our simulation, then costlessly restore land in the "land recovery" scenario. It should be noted that the land recovery is not evenly distributed across the three types, and, to the extent that different crops use the three types of land in differing intensities, the Rybczynski effects on the three types of crops will also be unequal.31

Quantity Controls in Trade

The major distortion in the economy is assumed to be quantitative controls on both imports and exports.32 Demanders are assumed to treat imports and domestically produced goods as imperfect substitutes (the Armington assumption). They have a sectoral import demand function that depends on the relative prices of imports and domestically produced goods on the domestic market. These demand functions are parameterized according to the "normal" levels of sectoral imports that one would expect North Korea to have without any rationing, given the results from the gravity model. Then, we assume that the difference between desired imports and observed imports is due to the imposition of quantity rationing by the government. That is:

equation (M sub i/D sub i)rationed=qr sub i of (M sub i/D sub i)desired

where M is imports, D is domestic supply, qr is the quantity rationing rate, and the subscript I refers to the sector.33

The model also specifies sectoral export supply functions, in which the export supply ratio depends on the ratio of the export price to the price on the domestic market. The supply functions are also parameterized such that the desired ratio is consistent with the results from the gravity model. Symmetrically with the treatment of imports, quantity controls are specified such that actual exports are less than desired.

The result is that demanders are forced off their import demand curves and producers are forced off their export supply curves.34 The distortions are quite large, indicating large potential gains from liberalizing trade and allowing markets to clear. The trade rationing contributes to major distortions in the domestic price system on top of explicit internal taxes.

Military Demobilization

North Korea is probably the most militarized society on earth, devoting an estimated 25 percent of GDP to military activities and maintaining more than one million men under arms. As noted earlier these stylized facts may be a bit misleading: probably half of North Korea's armed forces are engaged in non-warfighting activities that would be performed by the civilian sector in most other countries. In our model, we allocate half of the army to nonmilitary activities, leaving a sizable share of the economy's resources tied up in military activities.

Assuming that trade and other reforms were undertaken, military demobilization could generate additional welfare gains as resources previously devoted to the military were released into higher productivity activities. To illustrate the potential gains, we consider a scenario where we reduce the size of the military to the roughly 3.5 percent of GDP devoted to military activities in South Korea in 1995, and then measure the "peace dividend" increase in GDP associated with this reallocation of resources.



The model utilizes two main databases: a macroeconomic and a microeconomic Social Accounting Matrix (SAM) of North Korea for 1996, the most recent year for which data are available. The SAM, described in Appendix 1, is a consistent array of economic transactions among agents that reconciles the input-output and national accounts. In estimating the SAM, we had to draw on a variety of sources, including incomplete national accounts, sectoral production and trade data, and estimates of government accounts. These data are not only incomplete, but also probably replete with serious measurement errors.

For the modeling exercise, we need various share coefficients from the SAM, such as sectoral intermediate-input and value added shares (for production functions) and expenditure shares (for consumption functions). Our problem is to estimate these coefficients, which requires estimating a consistent SAM for the base-year of the model, using scarce data measured with unknown error. Using standard econometric methods, the problem is essentially hopeless—there are not enough data to provide the degrees of freedom necessary in estimating the parameters, even if we were willing to make very strong assumptions about the error generation process—which we are not. However, in contrast to the usual situation in econometrics, we have a great deal of prior information about the parameters to be estimated. The structure of the SAM imposes powerful adding-up constraints, and we have information about the likely values of the various coefficients from a variety of sources, including comparative data from past periods and other similar countries. The issue is how to use this information efficiently.

In this situation, we use an estimation approach, which Golan, Judge, and Miller (1996) call "maximum entropy econometrics", that draws on information theory. The estimation philosophy is to use all the information available, including information about the coefficients to be estimated, but not to assume any information that is unavailable. Our particular estimation approach applied to SAMs is described in Golan, Judge, and Robinson (1994) and Robinson, Cattaneo, and El-Said (1998) and incorporates assumptions of estimation error (errors in variables) and prior knowledge about parts of the SAM (such as various macro aggregates). We incorporate prior information about the structure of the SAM by specifying an initial SAM that reflects all the information we have (even if inconsistent). We then estimate a new SAM that is not only "close" to the old SAM—minimizing a "cross entropy" measure of the deviation between the two—but also: (1) satisfies all the adding-up constraints inherent in the definition of a SAM; (2) includes any other constraints such as knowledge about parts of the SAM (e.g., some of the national accounts or other aggregates); and (3) incorporates stochastic information about constraints involving measurement error. The method is both flexible and powerful when dealing with scattered and inconsistent data.

So, for example, with respect to the macroeconomic SAM, we retained the assumption that the North Korean government makes all investments but at the same time introduced a more elaborate revenue-generating system consistent with the North Korean data. However, the North Korean data itself were internally inconsistent and the macro SAM would not balance. Base GDP is calculated to be roughly 32 billion won, higher than the officially reported 23 billion won figure shown in table 1. The reason is two-fold. First, the officially reported figure appears to exclude the military. Second, the officially reported data do not appear to be internally consistent when entered into a consistent social accounting matrix (SAM). Even assuming that little investment has occurred and that the capital stock has actually shrunk, the 1996 figures would imply a tremendous decline in output relative to the 1990 SAM constructed by Noland, Robinson, and Scatasta. This result implies that there were very big reductions in factor supplies or that much of the economy was operating at 10 to 15 percent of capacity. The simplest way to generate a consistent SAM was to raise output. While it may well be the case that floods, famine, and the practice of scrapping capital and bartering it for food have reduced factor supplies, and utilization of remaining capacity is low, there are also reasons to believe that the actual output is higher than reported by the authorities.35

With respect to the microeconomic SAM, the inter-industry relations from the Noland, Robinson, and Scatasta 1990 micro SAM were used as the proto-SAM, updating it to reflect the apparent reduction in the capacity utilization rate (or, alternatively, decline in the value of the North Korean capital stock) and obtain consistency with the macro SAM.36

Urban workers are divided into high-skilled (professional, technical, and managerial) and low-skilled (the remainder). The initial starting point for industry employment structure was taken from pre-reform Chinese data. The wage premium was calculated on the basis of South Korean data. While one might expect a priori that wage dispersion in the North would be less than in the South, at this level of sectoral aggregation, the skilled wage premium obtained from the South Korean data was within the dispersion observed in fragmentary data on North Korean wages.

Land is allocated across the three agricultural activities as described in the previous section. The share of land in value-added was initially estimated from cross-country comparisons, yielding reasonable starting estimates of "rental rates" for different types of land.


Policy Experiments

The model was used to run five basic scenarios and the analysis is carried out in ten incremental steps or experiments.37 In the first scenario, North Korea costlessly recovers land damaged in the 1995-96 floods. In the second, quantity rationing of international trade is removed and the economy experiences the static gains from trade associated with specialization according to comparative advantage.

The next three scenarios are extensions of the second. Recent research suggests that the world is characterized by international technological spillovers. These are quite important in the case of developing countries which benefit from technological developments abroad transmitted through international trade. In the case of North Korea, the parameters estimated by Coe, Helpman, and Hoffmaister (1996) indicate that complete liberalization would result in a total factor productivity (TFP) gain of approximately 18 percent as a result of its economic opening and importation of capital equipment embodying new technologies from abroad. (Indeed, the results reported in Noland, Robinson, and Liu (1997) imply that this is a conservative estimate.)38 The third scenario therefore assumes that the static reallocation effects of trade liberalization are accompanied by an 18 percent sectorally uniform increase in TFP.

The fourth scenario concerns the value of the post-liberalization capital stock. There are two points to consider. First, due to the putty-clay nature of technology, the capital stock accumulated under one set of output and factor prices is likely to be sub-optimal for different relative prices. While this is true for all economies, the impact is particularly acute for transition economies, where the relative prices under central planning were wildly different from those observed in world markets. Second, economies sheltered from international trade may manufacture products that are essentially worthless in world markets. Think of televisions or radios without tuners—both of which are produced in North Korea. To the extent that capital is product-specific, this capital will be effectively worthless when the economy is opened up to trade.39 Sinn and Sinn (1992) report that one-half to two-thirds of East Germany's capital stock was worthless after unification. If lack of exposure to international trade is taken as a proxy for internal distortion, the North Korean economy is likely to be even more distorted than was the East German economy. On the basis of the East German experience this shock was calibrated as one half to two-thirds of the value of the 1990 pre-opening capital stock in Noland, Robinson, and Scatasta (1997). In the current model, calibrated to 1996, the capital stock has already shrunk, and the obsolescence shock is assumed to be smaller at 25 percent.40

In the fifth and final scenario we examine the issue of military demobilization and the "peace dividend". In this scenario, the North Korean military is demobilized by 70 percent, until the share of the military in national income approximates the 3.5 percent exhibited in South Korea.41

These five scenarios are implemented in 10 experiments or steps that indicate a range of possible results and also show the impact of partial reforms.

The simulations suggest that North Korea is indeed an extraordinarily distorted economy. In scenario 1, North Korea costlessly rehabilitates flood-affected lands in 10 successive steps. As can be seen in figure 3, the impact on GDP is minimal, increasing it by less than two percent. Domestic production of rice and other agricultural commodities increases by around four percent and corn production rises by 12 percent, leaving domestic consumption below the FAO's human survival target.42 At first glance this result might seem odd—increasing the arable land endowment of a famine-afflicted country should have a bigger macroeconomic impact. These results reflect two important underlying facts: first, North Korea was already experiencing a famine prior to the floods that commenced in July 1995, second, only around 15 percent of the arable land was affected by the floods. So, while natural disasters may have exacerbated the food availability problem, the famine is not a product of bad weather. Rather, systemic mismanagement and lack of intermediate inputs such as fuel and fertilizer are its proximate cause. The latter, in turn, are due to severe balance of payments constraints and policy decisions about the use of foreign exchange—which are fundamentally outside the agricultural sector. Nevertheless, this result underlines a critical issue: if only a relatively modest component of the decline in output can be attributed to flood-related declines in agriculture, then what explains the rest?

Scenario 2 addresses the main distortion in the model, the severe repression of international trade. The gravity model results reported in Noland (1996) suggest that, in 1990, the total share of imports and exports in North Korean GDP would have been roughly 70 percent if North Korea had exhibited the economic behavior of a "normal" country of its economic size. With the economy now smaller, the expected trade share should be even higher. In scenario 2, the quantitative restrictions on trade are relaxed in 10 steps, and the impact on GDP is shown in figure 3. As might be expected on the basis of microeconomic theory, the impact of the relaxation of the constraint is greatest at the beginning and declines thereafter. A complete freeing of this constraint increases GDP by approximately 60 percent due to static reallocation of factors alone. Domestic availability of rice and maize on commercial terms increases by more than 80 percent and 90 percent, respectively—far more than in the land recovery scenario—and unlike the land recovery scenario, in excess of the FAO's human survival needs target.

This static reallocation effect would not be the only impact of liberalizing trade. Increased importation of capital equipment from abroad could increase TFP substantially. As shown in figure 3, these gains are almost exactly offset by the negative impact of an assumed obsolescence shock to the capital stock of 25 percent. Obviously one should not attach too much weight to the exact figures derived from this modeling exercise. Rather, these results are probably best interpreted as an indication that, in an economy as distorted as North Korea's, an increase in the economy's openness to international trade could have enormous macroeconomic effects, dwarfing the impact of a flood.

The final scenario is a 70 percent military demobilization, which reduces North Korea's expenditure on the military to a share of GDP similar to the South's. Obviously there would have to be major diplomatic breakthroughs for this to occur, but the experiment is presented as a heuristic exercise to illustrate how large the potential pay-offs might be. This experiment is probably dependent on the trade liberalization experiment. Without liberalization, it is unclear where the demobilized resources would be redeployed; however, with liberalization, they could be redeployed to their highest efficiency uses.

As shown in figure 3, redeployment of resources on this scale could add another 18 percent to GDP, with the five scenarios undertaken together increasing real GDP by nearly 79 percent from the base. Domestic consumption of rice and maize more than triples. Holding non-defense government spending constant in nominal terms, the other components of GDP rise. Real consumption nearly triples, while investment nearly doubles. International trade expands enormously. Holding the trade balance constant, the real exchange rate experiences a modest appreciation; if the real exchange rate is held constant, the trade deficit expands somewhat.

The composition of output changes enormously (figure 4) as light manufacturing, mining, construction, and services expand, while industrial intermediates, capital goods, and the army contract in response to changes in the relative price structure and the decline in government demand for military services (figure 5).43 Within agriculture, maize output falls, then expands, as the highly inefficient production of rice is abandoned and land is increasingly devoted to the production of maize. Light manufacturing experiences an export explosion, with exports increasing 40-fold, and mining exports more than triple. Imports rise in all traded goods categories, led by a 15-fold growth in capital goods imports.

These changes in composition have profound effects on factor usage and returns (figure 5) . Two and one half million workers leave the agricultural sector and another 350,000 leave the army, the bulk of which are re-employed in the light manufacturing sector. Even in the agricultural sector, the employment and wage changes are greater under economic reform than under land recovery. Employment also increases in the mining, construction, and service sectors. The real wages of all three classes of labor more than triple, with the largest increases experienced by the highly skilled. The rate of return on capital also more than triples. The rate of return on land falls, however, as the increased availability of imported agricultural goods reduces food scarcity, and with it the implicit scarcity returns to land. The distribution of income shifts slightly in favor of the urban high-skilled.

These results are speculative and subject to a degree of spurious precision. Nevertheless, the modeling work conveys several important points. First, even when defined narrowly in terms of domestic food availability, the pay-offs to reform dwarf the impact of more narrow efforts to raise productivity in agriculture. Second, there are massive shifts in the composition of output under reform, and millions of workers switch jobs. Third, the importance of international trade increases tremendously with reform. The latter two findings could raise thorny issues for the current regime.



In this paper we have used cross entropy estimation to construct the data base for a CGE model of the North Korean economy. The modeling work extends the literature on the North Korean economy in a number of important ways. The model we construct is, to our knowledge, the first behavioral economic model to explicitly address the issue of North Korea's famine and the potential gains to the economy from military demobilization.

Because the North Korean economy is so distorted, we find that the returns to systemic reform dwarf those associated with more narrowly conceived policies focused on agricultural recovery. Even defined narrowly in terms of the domestic food availability, the pay-offs to systemic reform are multiples of even costless recovery of flood-affected lands and the replacement of the flood-damaged agricultural capital stock. Potential increases in GDP from reform are on the order of 60 percent, with an additional 18 percent obtainable as a "peace dividend" if North Korea were to substantially demobilize its enormous military. With reform, domestic output of food declines, but domestic food consumption easily exceeds human survival requirements because of the availability of imports. In contrast, flood rehabilitation leads to an increase in domestic food production, but this increase falls short of the human survival target.


Appendix 1: A SAM for North Korea

A Social Accounting Matrix (SAM) provides a tabular snapshot of the economy at one point in time. Each non-zero cell in the SAM represents the value of an economic transaction between actors. The accounts of the SAM define the transactions and income flows among five basic actors in the economy: producers/enterprises, households, government, capital account and the rest of the world. The input-output notion of inter-industry linkages is generalized to the idea that each actor's purchase is another actor's sale. Any flow of money from one actor to another is recorded in the SAM as a payment by some actor (the column) to some other actor (the row). The SAM also generalizes the national income accounting notion that income equals expenditure. The SAM must in fact be balanced: the sum of each column must equal the sum of each row, so that a budget constraint is imposed on each productive sector, labor category, household type, and so forth. This means that (1) costs (plus distributed earnings) exhaust revenues for products, (2) expenditure (plus taxes and savings) equals income for each agent, and (3) demand equals supply for each commodity.

The SAM is divided into a number of blocs. The Activities bloc describes the costs and revenues for domestic producers. In the columns, the producers buy intermediate inputs, make value-added payments to primary factors and transfer indirect, value-added, and export taxes to (or receive subsidies from) the government. In the rows they sell goods on domestic and foreign markets. The Commodities bloc describes markets for final products. The row describes sales on the domestic market, distinguishing between intermediate, consumption and investment demand. The column identifies absorption, which equals the value of domestic products sold on the domestic markets plus imports (valued at world prices), consumption taxes, value-added taxes, and tariffs. The Factors bloc describes value-added payments to primary factors (in the row) and their distribution to specific institutions (enterprises, households, and government) plus the payment of direct factor taxes (in the column). The remaining blocs describe transfers among institutions.

For a detailed treatment of Social Accounting Matrices see Pyatt and Round (1985), Stone (1986), or Devarajan, Lewis, and Robinson (1994).

View SAM (Please print out before viewing—you will need Adobe Acrobat Reader)

Acrobat Reader is available here:


Appendix 2: Model Description

Six equation "blocks" are distinguished.

Price Equations

On the import side the model incorporates the "small-country" assumption: equation 1 gives the domestic price of sectoral imports in terms of exogenous world prices, the exchange rate and tariff rates. North Korea is also assumed to be small on the export side; equation 2 shows that the domestic price of sectoral exports is determined by world prices, exchange rate, and any applicable export subsidy or tax, depending on the sign of parameter te. As previously noted, trade distortions are modeled as quantitative restrictions and affect the quantity equations (see next section). The major form of internal distortion is assumed to be indirect turnover taxes.

Consumers buy a composite consumption good and producers sell a composite production good. Their sectoral prices are the weighted averages of the domestic prices of their components, as shown in equations 3 and 4.

A value added (or net) price is defined in equation 5 as the unit value of output minus the cost of intermediate inputs. These are priced on the basis of PC, the consumption price of composite goods, defined in equation 6 as the adjustment of the composite consumption good price by final consumption tax rates.

Finally, in equation 7 a cost of living index, defined as a weighted average of sectoral commodity prices, is fixed as the numeraire.44

Quantity Equations

Sectoral production technology is represented in equation 8 by a set of Cobb-Douglas functions45 of the primary factors—the three types of land, capital and the three types of labor (rural, urban high skill and urban low skill). Equation 9 shows the sectoral demand for primary factors. Their remuneration is set equal to the value added price (net of both indirect taxes and intermediate input costs) times the partial of the production function with respect to each factor. Equation 9 therefore also represents the first-order derivative conditions for profit maximization. Factor demand equations assume that primary factors are paid the same average rental rate (WF), but the model allows for distortions in factor markets through the sector-specific parameter (wfdist) which measures the extent to which the sectoral marginal revenue product of the factor deviates from its average return. The treatment of this parameter in the closure section of the model is what determines whether factors are sector-specific or mobile. Equation 10 specifies the Leontief intermediate input demand function, based on fixed input-output coefficients.

Each sector is assumed to produce differentiated goods for the domestic and export markets. Equation 11 shows that producers decide how to subdivide their total sectoral output as a constant elasticity of transformation (CET) function of the amounts sold in the two markets. Subject to this transformation function, producers maximize revenue from sales. The first-order conditions yield equation 12, which expresses the ratio of exports to domestic sales in terms of the ratio of their prices. Similarly, imported and domestic products are differentiated at the sectoral level, equation 13 indicating the composite (consumption) good as a CES aggregate. Consumers minimize the cost of obtaining a given amount of composite good, resulting in a relationship between the ratio of the amounts imported and domestically produced and the ratio of their prices (equation 14). Such product differentiation permits two-way trade and provides some autonomy to the domestic price system not found in models that assume perfect substitutability between domestic products and imports in consumption and between domestic and export markets in production.

Initially, the economy's market equilibrium is computed with parameters which are benchmarked on the basis of the analysis presented earlier. The central planners impose a rationing solution which implicitly puts the economy on a lower transformation function than would be attainable in the absence of rationing.46 The gravity model results provide an indication of the extent to which rationing reduces aggregate trade. The commodity analysis of North Korea's competitiveness was used to parameterize the measure in which rationing was binding at the sectoral level. Finally, the curvature of the CET functions were parameterized along the lines of previous studies as to the degree of intrinsic differentiation of the commodity aggregates.

For the specific case of North Korea, equations 12 and 14 are modified to introduce restrictions to free trade in the form of export and import quantity rationing ratios. Such ratios distort the formation of the composite production and consumption goods, but does not produce direct spillovers in other markets through wedges in relative prices. Still, the rationing of exports and imports distorts producers' and consumers' decisions on both domestic and external markets. Therefore, this formulation enables us to model economy-wide distortions by introducing quantitative restriction only on the external side of the economy.

Income and Expenditure Equations

Equation 15 combines the incomes, across production sectors, accruing to each of the primary factors.

Equation 16 represents a compact formulation of three equations which simultaneously determine real government expenditure, nominal private consumption, and nominal investment demand. The three expenditure aggregates are derived from the distribution of total factor income to the economy's institutions (households and enterprises) and finally from its allocation to the three expenditure categories. This set of equations simply reproduces the flows which are represented in the macroeconomic Social Accounting Matrix (SAM).

In the closure section of the model—not entirely shown to avoid clutter—nonmilitary government expenditure is held constant at the base value, while investment and household consumption adjust in response to changes in income induced by the policy experiments. The trade balance is fixed and the real exchange rate is allowed to adjust endogenously.47

The next three equations specify the sectoral composition of the expenditure aggregates. Sectoral private consumption is based on a Cobb-Douglas utility function, with constant expenditure shares (equation 17). Government demand by nonmilitary sectors is given by constant shares of real government nonmilitary spending (equation 18). Real fixed investment by sector of origin is determined in equation 19. The CGE model is static. Investment is only a category of final demand, with no effect on the capital stock in the period, and so has no role in the supply side of the model.

Market-Clearing Conditions

The remaining equation block presents the market-clearing conditions. Equilibrium in product markets is represented in equation 20, equating supply of the composite good in each sector to the sum of product demands. Equation 21 specifies equilibrium in each factor market; total factor supply is fixed and the equilibrating variable is the average factor price (WFf). External balance is portrayed in equation 22, which specifies a fixed aggregate balance of trade.

These generic equations must hold for all product and factor markets. Altogether, the model has 275 endogenous variable and 276 equations, as equation 16 actually denotes three equations. However, the model satisfies Walras' Law, in that the number of independent equations is equal to the number of endogenous variables.

See Equations


Appendix 3: Parameterization

Sector Import Substitution Export Transformation

Rice 5.0 0.5
Maize 5.0 0.5
Other agricultural goods, forestry, and fisheries 5.0 0.5
Mining 0.5 5.0
Light manufacturing 0.5 5.0
Industrial intermediates 0.5 1.5
Capital goods 5.0 0.5
Services 1.5 1.5

  Rationing ratios: actual over desired
Sector Imports Exports

Rice 0.05 0.90
Maize 0.05 0.90
Other agricultural goods, forestry, and fisheries 0.05 0.90
Mining 0.20 0.05
Light manufacturing 0.20 0.05
Industrial intermediates 0.20 0.50
Capital goods 0.05 0.90
Services 0.20 0.50

Note: The smaller the ratio, the larger the distortion.



Coe, David T., Elhanan Helpman, and Alexander Hoffmaister. 1996. "North-South R&D Spillovers." Economic Journal 107, no. 440: 134-49.

Dervis, Kemal, Jaime de Melo, and Sherman Robinson. 1982. General Equilibrium Models for Development Policy. New York: Cambridge University Press.

Deutsch-Koreanische Industrie und Handelskammer. 1991. Nordkorea: Einblicke in Wirtschaft und Lebenweise. Seoul: Deutsch-Koreanische Industrie und Handelskammer.

Devarajan, Shanta, Jeffrey D. Lewis, and Sherman Robinson. 1994. Getting the Model Right: The General Equilibrium Approach to Adjustment Policy. Forthcoming.

Dyck, I. J. Alexander. 1997. "Privatization in Eastern Germany: Management Selection and Economic Transition." American Economic Review 87, no. 4: 565-97.

Easterly, William, and Stanley Fischer. 1994. The Soviet Economic Decline: Historical and Republican Data. NBER Working Paper No. 4735. Cambridge MA: National Bureau of Economic Research.

Eberstadt, Nicholas. 1994. "Reform, Muddling through, or Collapse?" In Thomas H. Henriksen and Kyongsoo Lho, editors, One Korea? Stanford, CA: Hoover Institution Press.

Eberstadt, Nicholas. 1995. "The Collapse of Soviet and Russian Trade with the DPRK, 1989-1993." The Korean Journal of National Unification: 87-104.

Eberstadt, Nicholas. 1996. "How Much Money Goes From Japan to the DPRK?" Asian Survey XXXVI, no. 5: 523-42.

Eberstadt, Nicholas. 1998a. "Development, Structure, and Performance of the DPRK Economy: Empirical Indications." Paper presented at the conference on Developing Social Infrastructure in North Korea for Economic Cooperation between the North and the South, Seoul, Korea, 9-10 November.

Eberstadt, Nicholas. 1998b. "The DPRK's International Trade in Capital Goods, 1970-1995: Indications from >Mirror Statistics'." Journal of East Asian Affairs XII, no. 1: 165-223.

Eberstadt, Nicholas, and Judith Bannister. 1992. The Population of North Korea. Berkeley, CA: Institute of East Asian Studies.

Food and Agriculture Organisation. 1997. "Special Report: FAO/WFP Crop and Food Supply Assessment Mission to the Democratic People's Republic of Korea" (25 November).

Frankel, Jeffrey A., and Shang-Jin Wei. 1995. "Is a Yen Bloc Emerging?" Joint U.S.-Korean Academic Studies 5: 145-75.

Golan, Amos, George Judge, and Douglas Miller. 1996. Maximum Entropy Econometrics: Robust Estimation with Limited Data. New York: Wiley.

Golan, Amos, George Judge, and Sherman Robinson. 1994. "Recovering Information from Incomplete or Partial Multisectoral Economic Data." The Review of Economics and Statistics LXXVI, no. 3: 541-49.

Hwang, Eui-Gak. 1993. The Korean Economies. Oxford: Clarendon Press.

International Monetary Fund. 1997. "Democratic People's Republic of Korea Fact Finding Report" (12 November).

Kim, Pyung Joo. 1995. Money and Banking System in North Korea. Seoul: Korea Institute of Finance.

Koo, Bon-Hak. 1992. Political Economy of Self-Reliance. Korean Unification Studies Series No. 14. Seoul: Research Center for Peace and Unification in Korea.

Lee, Chan-Woo, Tsutomu Niacin, and Makoto Nobukuni. 1995. "Estimate of the Supply and Demand for Grain in the Democratic People's Republic of Korea: 1995." Niigata, Japan: Economic Research Institute for Northeast Asia (ERINA), Research Division. Manuscript (July).

Lee, Hy-Sang. 1994. "Supply and Demand for Grains in North Korea." Korea and World Affairs XVIII, no. 3: 509-54.

Lee, Sang-Man. 1993. "A Study on Patterns of Economic Integration Between North and South Korea." East Asia Review 5, no. 3: 91-133.

Lee, Young Sun. 1994. "Economic Integration of the Korean Peninsula: A Scenario Approach to the Cost of Unification." In Sung Yeung Kwack, editor, The Korean Economy at a Crossroad. Westport, CT: Praeger.

Namkoong, Young, and Ho-Yeol Yoo. 1994. "North Korea's Economic System." In Tae Hwan Ok and Hong Yung Lee, editors, Prospects for Change in North Korea. Seoul: Research Institute for National Unification.

Noland, Marcus. 1990. "Prospective Changes in Japan's Trade Pattern." Japan and the World Economy 2: 211-38.

Noland, Marcus. 1991. "Prospective Changes in the Commodity Composition of U.S.-Korea Trade." Joint U.S.-Korea Academic Studies 1.

Noland, Marcus. 1996. "The North Korean Economy." Joint U.S.-Korea Academic Studies 6: 127-78.

Noland, Marcus. 1997. "Why North Korea Will Muddle Through." Foreign Affairs 76, no. 4: 105-18.

Noland, Marcus. 1998a. Stumbling Toward the Apocalypse. Washington: Institute for International Economics. Forthcoming.

Noland, Marcus. 1998b. "The External Economic Relations of the DPRK and Prospects for Reform." In Samuel S. Kim, editor, North Korean Foreign Relations. Hong Kong: Oxford University Press.

Noland, Marcus, Sherman Robinson, and Li-Gang Liu. 1997. Economics of Korean Unification. Working Paper Series No. 97-5. Washington: Institute for International Economics.

Noland, Marcus, Sherman Robinson, and Monica Scatasta. 1997. "Modeling North Korean Economic Reform." Journal of Asian Economics 8, no. 1: 15-38.

O, Wonchol. 1995. "A Response to North Korea's Food Crisis." Revised and edited English translation of "Ssalboda nonsabop wonjo p'ilyo." Sin Dong-A (September): 460-76. Canberra: Australian National University.

Pyatt, G., and J. I. Round, eds. 1985. Social Accounting Matrices: A Basis for Planning. Washington: The World Bank.

Robinson, Sherman, Andrea Cattaneo, and Moataz El-Said. 1998. Estimating a Social Accounting Matrix Using Cross-Entropy Methods. Trade and Macroeconomics Division Discussion Paper No. 33. Washington: International Food Policy Research Institute.

Sinn, Gerlinde, and Hans-Werner Sinn. 1992. Jumpstart. Cambridge: MIT Press.

Smith, Heather. 1998. "The Food Economy: Catalyst for Collapse?" In Marcus Noland, editor, Economic Integration of the Korean Peninsula. Special Report No. 10. Washington: Institute for International Economics.

Snyder. Scott. 1997. "North Korea's Decline and China's Strategic Dilemmas." Special Report (October). Washington: United States Institute for Peace.

Stone, R. 1986. "Nobel Memorial Lectures 1984: The Accounts of Society." Journal of Applied Econometrics 1:5-28.

Suh, Jae-Jean, and Byoung-Lo P. Kim. 1994. "Prospects for Change in the Kim Jong-Il Regime." Series No. 2 Policy Studies Report (December). Seoul: The Research Institute for National Reunification.

United Nations Development Program. 1998. "Thematic Roundtable on Agricultural Recovery and Environmental Protection in DPR Korea." Geneva, 28-29 May, processed.

United States Arms Control and Disarmament Agency. 1997. World Military Expenditures and Arms Transfers 1995-1996. Washington: Department of State.

World Bank, 1985. China: Economic Structure and International Perspective. Washington: International Bank for Reconstruction and Development.



1. Enterprises are classified as state-owned or worker cooperatives. They are further classified by size and, for administrative control purposes, as central or local. Larger plants are under the control of the central party and the Economic Bureau of the Office of State Affairs. Local plants are under the control and guidance of local party and economic guidance committees at the provincial level (Hwang 1993). A new incentive system for enterprises (the "Independent Accounting System") was introduced in 1984 and recently a greater degree of mutliplant integration under a common management structure with an associated domestic trading firm (similar to the East German Kombinat or South Korean chaebol) has been permitted. Worker remuneration is set on the basis of job classification and seniority among other factors, and there exists a material incentive system. Similarly agricultural operations are organized into state farms and peasant cooperatives, the County Farm Management Committee and the Central State Agriculture Committee providing economic management and party organs providing management guidance. The recent tendency has been to increase the importance of state farms (which are considered ideologically more advanced) and to integrate the agricultural sector more firmly into the central plan. On the state farms, peasants are paid fixed salaries. In the case of the cooperatives, which are theoretically owned by the members, the latter receive equal shares paid in cash and in kind, with bonuses going to work units overfulfilling targets. "In reality, the peasants are reduced to employees in either case" (H. S. Lee 1994, 511).

2. Some prominent "reformers" within the leadership appear to have been purged. Soon after the constitutional revision was announced the daily newspaper and theoretical magazine of the Korean Workers Party published a joint editorial excoriating economic reform and opening, describing it as "honey-coated poison," and reiterating the intention of maintaining a planned, self-reliant economy. The signals coming out of North Korea are mixed, to say the least.

3. One implication of taxing enterprises differentially by ownership type (in addition to violating horizontal equity) is that it makes calculating sectoral tax burdens effectively impossible.

4. For a description of the formal pricing structure and black markets, see Hwang (1993) and Suh and Kim (1994) respectively.

5. For more on the North Korean financial system, see P. J. Kim (1995).

6. On the conventional wisdom see Eberstadt (1994) and S. K. Kim (1994). The conventional view would be consistent with the econometric evidence of Easterly and Fischer (1994) which found that CPEs typically exhibit stronger than average performance in the initial stages of industrialization and then deteriorates into weaker than average performance as allocative efficiencies become more and more costly. According to Hwang (1993), South Korean per capita income did not exceed the North's until 1986.

7. See Noland (1996) for quantitative material on the debt (non-) burden. In this and other regards, comparisons between North Korea and Romania are striking (Noland 1997).

8. This debt trades in an (admittedly thin) secondary market at a considerable discount, with recent quotes in the range of 12-15 cents on the dollar (International Financing Review, 29 August 1998). The discount narrowed in late 1996 reaching roughly 50 percent and remained there through early 1998 when it began widening. Market participants indicate that the discount is an inverse indicator of regime viability: as conditions in North Korea worsen, the likelihood of collapse and absorption by the South increases and with it the prospect of debt holders being paid by the government of the unified peninsula. According to this logic, the market has been signaling an enhanced expectation of regime survivability.

9. Soviet assistance took a variety of forms including provision of key commodities at "friendship prices," technology transfer, and provision of significant trade credits and other forms of concessional economic exchange. See Noland (1996) for quantitative estimates and further discussion.

10. See H. S. Lee (1994), Lee, Niacin, and Nobukuni (1995), and O (1995) for discussions of North Korean agriculture.

11. See Noland (1998a) and Smith (1998) for more detailed assessments of the food situation and the diplomacy of food aid.

12. The most heavily affected areas were the corn growing provinces of Chagang, and North P'yongan in the northwest region of the country and the rice producing area of North Hwanghae, south of Pyongyang. We use this information on geographical incidence of the flooding in our subsequent modeling effort.

13. There are three major problems with this definition. First, produced intermediates are doubled-counted. Second, non-material sectors (which include housing, health and welfare expenditures, education, science, art, personal services, state administration, etc.) are ignored. Third, the prices, which are calculated to reflect the labor theory of value, do not reflect true scarcity values. As a consequence, even under the best of circumstances (such as having a cooperative government), it is difficult to reconcile material product accounts with the system of national accounts. Given the secrecy of the North Korean government, concordance is virtually impossible. Hwang (1993) makes a valiant effort, but the results are not fully convincing.

14. These data are of questionable accuracy. They are apparently calculated by taking measures of physical output and then calculating income using South Korean value-added weights. These figures are then subject to interagency bargaining within the South Korean government prior to public release. Given the collection methods, there is no way to verify the estimates of physical output. Furthermore, it is far from obvious that the South Korean value-added schema is the most appropriate. In other countries where such methods have been applied, the resulting estimates appear to overestimate the volatility of output, as industrial output exhibits greater variance than service activities.

15. The collapse of the industrial economy has adversely affected agriculture. The two primary fertilizers used in North Korea, urea and ammonium sulfate, are both petroleum-based, and shortages of petroleum feedstocks have adversely affected domestic production of fertilizer. Likewise, periodic blights have been worsened by a shortage of agricultural chemicals. Fuel shortages have impeded the use of agricultural machinery, forcing the reintroduction of draught animals. Problems in production have been compounded by difficulties in distribution and the use of output. Fuel shortages have hampered distribution. At the same time, some outside observers have questioned the uses to which output has been put: scarce cereals continue to be used to produce luxury products such as noodles, urban areas with high concentrations of KWP members and government officials have received preferential allocations, and it has been claimed that military stockpiling continues. The end result of these difficulties has been a secular deterioration in food production, and in the absence of additional imports, in the food balance.

16. Data on capital goods imports from an earlier period reported by Eberstadt (1998b) suggest that the decline in investment may be a secular phenomenon going back decades.

17. These taxes present special problems for analysis because they are levied at differential rates depending on the legal status of the transacting parties. (For example, the tax wedge imposed on an exchange between two state enterprises is different than the wedge imposed on a transaction between a state enterprise and a cooperative.)

18. These statements are based on partner country reports to the GATT, supplemented by data from KOTRA in the case of countries such as Iran which do not report the commodity structure of their trade with North Korea to the GATT. See Noland (1996) for details.

19. The ACDA figures for the North Korean arms trade are consistent with the statistics on DPRK-Russian trade reported by Eberstadt (1995).

20. See Snyder (1997) for a discussion of North Korea's food problem and the extent of Chinese assistance.

21. Even here one must be careful. Partner coverage my be incomplete. Additionally, there is a venerable history of trade ministry clerks around the world mistaking North and South Korea. Such errors have occurred in the past decade in Mexico, Austria, and most recently Lebanon. Moreover, the IMF reporting convention counts transport and insurance as part of imports; when one calculates the "mirror statistics" one has to adjust the constructed North Korean figures for this accounting convention. Lastly, these figures ignore barter trade. Although the existence of barter transactions would presumably not bias trade balance estimates in any particular way, barter could significantly distort analysis of trade volumes and composition. See Noland (1998b) for a more complete discussion of these issues.

22. The latter are sometimes reported to be in the billions of dollars, though recent research (e.g., Eberstadt 1996; Noland 1996, 1998b) indicates that these private aid flows are probably considerably smaller than claimed.

23. See Smith (1998) for a highly informative discussion of famine-related issues.

24. See Noland (1996) for details of this calculation.

25. The filters were: (1) sectors in which North Korea had a "revealed" comparative advantage, defined as the ratio of North Korea's share of world exports of a particular commodity category to its share of world exports as a whole; (2) sectors in which "revealed" comparative advantage had increased between 1980-92; (3) sectors in which South Korea has declining "revealed" comparative advantage; (4) sectors in which South Korea export shares were forecasted to fall in Noland (1991); and lastly (5) sectors in which the growth of South Korean outward foreign direct investment has exceeded that of the economy. The list is debatableCone might not expect North Korea's current exports, in essence the residual of the plan, necessarily to be the exports in a reformed, market economy. Nonetheless, this seems like a reasonable place to start.

26. Although South Korea had been the rice bowl in the past, if trade were opened today North Korea might end up being a net exporter of agricultural goods to the South, especially if capital markets were relatively unified and labor markets were not. O (1995) asserts that a reformed North Korea could have a comparative advantage in fruits, vegetables, dairy products, and meats (including beef, chicken, and pork). Both countries would undoubtedly remain global net importers of food.

27. The criteria were: (1) sectors in which North Korea had a "revealed" comparative advantage; (2) sectors in which North Korea's "revealed" comparative advantage had increased between 1980-92; (3) sectors in which Japan has declining "revealed" comparative advantage; 4) sectors in which Noland (1990) projected Japanese net exports to fall; and 5) sectors in which the growth rate of Japanese outward foreign direct investment has exceeded that of the economy.

28. A similar set of criteria were employed: (1) sectors in which both countries had "revealed" comparative advantage; (2) sectors in which this "revealed" comparative advantage has declined over time; (3) both countries were forecast to lose competitiveness in these sectors; and (4) the rate of outward foreign direct investment has been in both countries is higher than national income.

29. This analysis was done partly on the basis of historical experience. It is possible that with changes in real exchange rates (such as an appreciation of the yen) industries of greater capital intensity (auto parts, for example) than those reported in Noland (1996) could be future candidates for relocation to North Korea. GM, for example, has expressed interest in constructing an auto parts plant in the Rajin-Sonbong Special Economic Zone (Financial Times, 17 June 1995).

30. The exchange rate variable in the model can be seen as a price level deflated (PLD) real exchange rate, deflating by the numeraire cost of living index.

31. The Rybczynski effects refer to the impact of a change in a factor endowment on output quantities. In an "even" international trade model where the number of goods equals the numbers of factors, an increase in a factor endowment will lead to an increase in the production of at least one good and a decrease in the production of at least one other. This simple version of the theorem breaks down when the model is "uneven," includes nontradables, and differentiated products, as in the case at hand. Nevertheless, as an empirical matter, one may still obtain the counterintuitive result that an increase in an endowment can cause a decrease in an output quantity.

32. Data on aggregate revenue from turnover taxes are reported in table 3. But since these taxes are assessed on the basis of the legal status of the transacting enterprises, sectoral tax rates per se do not exist. In our modeling work, we have allocated these receipts more or less evenly across the industrial sector with a mild degree of escalation by degree of processing, as typically has been the case in other CPEs.

33. This approach to modeling import rationing was first used by Dervis, de Melo, and Robinson (1982). See their paper for a discussion of the properties of this approach, including questions of incentive compatibility.

34. The degrees of sectoral quantity rationing is given in Appendix 3.

35. First, they have an incentive to understate output to increase international aid flows. Second, as mentioned in the text, the official data appear to refer only to output or resources controlled by the central planners. Evidence indicates that the military economy and economic activity outside the plan have increased, or at least have not decreased as rapidly as formal activity under the plan. Finally, aid flows which account for a considerable share of food consumption, do not appear to be included in the official figures.

36. The input-output coefficients contained in the 1990 proto-SAM were in turn derived from a pre-reform (1979) Chinese input-output table compiled by the World Bank. This table was constructed to SNA standards, expanding on the material product accounts (World Bank 1985). The assumption is that a starting point (or prior) for the inter-industry input-output relations in North Korea is pre-reform China, reflecting their common links to 1970s vintage Soviet manufacturing technology. See Noland, Robinson, and Scatasta (1996) for sensitivity analyses.

37. In the figures these experiments are labeled EXP1 through EXP10. They are alternative static equilibriaCthey do not represent changes over time.

38. See Noland, Robinson, and Liu (1998) for further discussion as well as an analysis of sectorally non-uniform changes in TFP. In all likelihood, importation of capital equipment would not be the only avenue of TFP-enhancement associated with opening. Dyck (1997) argues that introduction of new management practices was critical in the East German case.

39. This treatment is obviously a stylized one. One way to think of it is that there are goods with positive prices in autarchy and a world price of zero. When the economy is opened up, product specific capital depreciates instantly.

40. There is a counterargument that the North Korean capital stock might not decline by as much as the East German case. Two reasons are given. First, it is argued that the decline in the value of the East German capital stock was partly a result of West German transfers that facilitated the shift in demand from formerly East German home goods to imports from the West. If the North Koreans received fewer transfers, they would be forced to continue buying home goods, maintaining the value of the North Korean capital stock. Second, the East Germans lost their major markets in other centrally planned economies, contributing to the decline in the capital stock. It has been asserted that China represents a viable market for cheap, low quality North Korean manufactured goods. If one accepts these arguments, then one should focus on the previously described scenarios in which the value of the North Korean capital stock is implicitly maintained.

41. As indicated earlier, we assume that half of the army is engaged in activities (construction, manufacturing etc.) that normally would be performed in the civilian sector of the economy, and that half of the army is engaged in "war fighting" activities. Demobilization is modeled as a reduction of resources devoted to strictly military activities.

42. One can think of at least two possible channels by which the impact of the floods could be underestimated. First, if one assumes high substitutability among primary inputs, labor and capital could simply substitute for land. Second, some of the capital stock in the agricultural sector was destroyed in the floods. As a mental experiment and check on robustness, a variant of the first experiment was run in which the agricultural capital stock was augmented as land was recovered. The addition of capital along with land did indeed increase the output response, but the impact on GDP was still only around two percent.

43. For the sake of brevity the remaining discussion refers to scenario 5 experiment 10.

44. The weights being equal to the share of each sector in total demand.

45. Constant elasticity of substitution (CES) functions are used for the military and mining sectors.

46. See Noland, Robinson, and Scatasta (1997) for more detail and a diagramatic exposition.

47. Alternatively one could fix the real exchange rate and allow the trade balance to adjust. In point of fact, the results of the policy experiments are quite similar under this alternative closure.


Book: Witness to Transformation: Refugee Insights into North Korea January 2011

Peterson Perspective: North Korea's Immunity to Outside Pressure: Part I December 12, 2012

Policy Brief 10-1: The Winter of Their Discontent: Pyongyang Attacks the Market January 2010

Working Paper 10-2: Economic Crime and Punishment in North Korea March 2010

Paper: FTAs and the Future of US-Korean Trade Relations November 2009

Paper: Implementing the KORUS FTA: Key Challenges and Policy Proposals February 2008

Policy Brief 08-6: North Korea on the Precipice of Famine May 2008

Policy Brief 07-7: The Korea-US Free Trade Agreement: A Summary Assessment August 2007

Policy Brief 06-4: Negotiating the Korea–United States Free Trade Agreement June 2006

Working Paper 07-7: North Korea’s External Economic Relations August 2007

Book: Avoiding the Apocalypse: The Future of the Two Koreas June 2000

Book: Free Trade Between Korea and the United States? April 2001

Working Paper 08-4: Migration Experiences of North Korean Refugees: Survey Evidence from China March 2008