Hunger and Human Rights:
The Politics of Famine

After nearly a decade of relief efforts, North Korean practices still fall well below international norms with respect to transparency and non-discrimination in the distribution of humanitarian relief.

During the postwar period, the public humanitarian relief system centered on the UN agencies, particularly on the WFP, which has developed a well-articulated set of norms governing the implementation of relief operations. At the core of these norms are principles of non-discrimination and distribution based on need: "priority in food aid should be given to the most vulnerable populations" and "such aid should be based on the needs of the intended beneficiaries." The notion of non-discrimination is defined with respect to age, gender, social status, ethnicity, and political beliefs (Ziegler 2002).

These basic norms, as well as basic principles of accountability within donor countries, drive the related insistence on thorough monitoring of aid, which is codified in the WFP handbook in a standard operating procedure embodying reciprocal obligations on the parts of donors and recipients. There have been parallel attempts to codify norms within the NGO community, including prior understanding of basic conditions; evaluation of effectiveness; participation by recipients in the design, management, and monitoring of programs; distribution of aid through a transparent system that can be monitored and adequately audited; and impartiality or the distribution of aid in a fair and equitable manner (Sphere 2004).

The desire to articulate clear norms among the humanitarian community is not simply an exercise in idealism; it is also designed to solve a particular set of incentive problems that can emerge in any humanitarian operation. In the absence of normative constraints, the differences among donors and competition among them can lead to a "race to the bottom": a willingness to turn a blind eye toward diversion; a tendency to exaggerate aid effectiveness; and even the empowerment of groups who bear responsibility for causing the humanitarian crisis in the first place.

In this context, it should be observed that the practices of two of the major donors, China and South Korea, ignore international humanitarian norms. The terms governing China's shipment of grain and other major foodstuffs is largely opaque, but the country makes no pretense of targeting vulnerable populations or monitoring. Indeed, the apparent Chinese practice of providing food directly to the North Korean military is reputedly undertaken so that the North Koreans can claim that multilateral aid does not go to the military. South Korea's practices are only marginally better: donations go directly into the PDS and monitoring is minimal. A number of South Korean NGOs have complained repeatedly that food is channeled to undeserving groups, in ratios that may be as high as 50 percent (Good Friends 2005).

In this context, it is also important to be clear on the meaning of diversion, a term that is used casually and has multiple meanings. The most common image is of the military seizing grain to feed the army and party cadre. But the political and military elite has a variety of channels for accessing food, including "first draw" on the domestic harvest, access to imports from China and South Korea that are weakly monitored or not monitored at all, and access to grain and other foods via the market through privileged access to foreign exchange.

This type of diversion is no doubt real, but almost certainly exaggerated. Much less attention has been given to the effect of the huge difference between controlled and market prices on the incentive to divert food for economic reasons: to sell it in the market. These incentives operate with respect to farmers-who can earn more by selling to the market than surrendering grain to the state-and they almost certainly operate with respect to aid as well. Local military and political officials and those involved in the transport of grain have strong incentives to divert aid to non-deserving groups or to the market.

Since its early operations in the country, the WFP has sought to address this problem through two means: (1) devising lists of target groups, and (2) selective monitoring of the institutions and programs-such as food-for-work programs-through which aid is delivered to recipients. Public distribution centers (PDCs) are the main channel for the delivery of food to the general, non-targeted population. These centers can be thought of as final "retail" outlets, where households purchase prescribed amounts of food using ration cards. The primary channel for delivering food to targeted groups is via more than 40,000 institutions such as schools, orphanages, and hospitals.

However, there is no separate channel in North Korea for distributing food to these institutions; food passes through the same county-level PDS warehouse before it is distributed to the final units. These county-level warehouses are controlled by People's Committees made up of mid-level government and party officials. These groups confront multiple demands on the food they control, from central authorities wishing to reallocate the food regionally, to local military and work units, to outright corruption.

Addressing the problem of diversion is particularly difficult in North Korea because even the most basic international norms are not observed. In essence, there exists a fundamental lack of trust between the government of North Korea and the international donor community. Indicative of this stance is the fact that Jean Ziegler, UN Special Rapporteur on the Right to Food, has been denied entry to North Korea five times, despite the fact that UN programs have been feeding nearly one-third of the population on an ongoing basis.

For nearly a decade, the North Korean government has consistently violated its fundamental obligation to facilitate WFP activities within the country. NGOs have been subject to similar obstacles, though given the sheer diversity of NGOs operating in North Korea some have been more successful in handling these constraints than others. There have been marginal improvements in access over time, but these have been grudging and have had a two-steps-forward, one-step-back character. What follows is meant to outline the basic constraints on project design, implementation, and monitoring. The characterization below should neither be interpreted as holding universally (some of the smaller NGOs may have effectively negotiated better access terms), nor necessarily holding at every moment in time since 1995. Where there have been recent improvements in practice, these are noted. However, it characterizes broadly the de facto monitoring regime for the single major channel of food to North Korea, the WFP.

With respect to basic issues of program design, the North Korean government imposes restrictions on the activities of the WFP and other groups that fundamentally inhibit the implementation of rational famine-relief programs:

• With respect to the NGOs, the North Korean government routinely denies visas in the absence of a pre-commitment to a target aid level, making pre-delivery assessment or program design difficult, if not impossible.
• The WFP and other groups have been denied access to parts of the country¡ªparts of the country believed to be particularly vulnerable. When the aid effort first began, aid was denied to whole provinces on the East Coast. After 1997, these restrictions were eased, but since 2000 there have been only marginal improvements in the number of counties to which the WFP has access, and 42 of 203 counties in the country remain off limits. As a result, the WFP has no information on the food situation in these restricted areas.
• The WFP and others generally do not have access to markets where food is sold, even though information on market prices is indispensable to understanding the distribution of food.

Once a program has been agreed upon, the donor must recruit staff to implement it. With the exception of the South Korean government effort, which has only a minimal monitoring regime, official relief agencies are not permitted to use Korean speakers or ethnic Korean staff. In the case of NGOs, the North Korean government has been reluctant in granting visas to Korean speakers, though this is not uniformly the case. In 2004, the North Korean government allowed resident WFP staff to begin Korean-language lessons.

• Without Korean speakers, the WFP and other organizations are reliant on government-supplied interpreters who owe their primary allegiance to the Flood Damage Rehabilitation Committee (FDRC), not the relief agency that pays their salary.
• In the case of the NGOs, the North Korean government has shown reluctance to issue long-term visas. For example, NGOs planning to deliver aid for five months were given two-month visas and the government has denied issuance of multiple-entry visas.
• The WFP has faced constant North Korean opposition to its desired number of monitors.
• Agency staffing levels are contingent on the dollar value of aid. Operational protocols specifically reflect this and if donations drop, staff members are asked to leave.
• The government restricts the use of vehicles and requires the use of FDRC-seconded drivers. Travel outside Pyongyang requires pre-approval (typically a week or more in advance) and accompaniment by official escorts or gminders.h Not until April 2002 could WFP sub-office employees outside Pyongyang walk outside their hotels without being accompanied.

The implications of the inability to recruit, train, and maintain authority over local staff should not be underestimated. As one observer put it, "their reporting loyalties are almost always toward the government" (Bennett 1999). One aid worker described a situation that would be amusing if not for the stakes. Unable to speak Korean or read Korean language signs, the aid workers had no idea if they were being shown the institutions that they had requested to visit, and in one instance suspected that they had been taken to the same institution twice. Even if humanitarian organizations are not worried about their North Korean counterparts "cheating" them, North Korea's insistence on staffing these positions with people trained as translators undermines effectiveness. As Dammers, Fox, and Jimenez (2005) observe, of the UN Children's Fund's (UNICEF) ten North Korean counterparts, none have specific technical or sectoral skills.

Once a program is initiated, North Korean practices make it impossible for the WFP or non-resident NGOs working through its Food Aid Liaison Unit (FALU) to track relief shipments from port to recipient. The opportunities for leakage in such a system are multiple. Typically the WFP and NGOs rely on a paper trail of transport waybills and transaction receipts at local PDCs to track supplies. Major diversion at the port is unlikely, but much food does not go from port to truck but rather to trains and barges before it is transferred to trucks; these shipments are not tracked. Relief officials have at times expressed the view that these records were fabricated, though whether this was done to hide diversion or as the result of simple lack of administrative capacity is unclear. There have also been a number of eyewitness accounts by foreigners, as well as refugee and defector testimonies to outright diversion by military units, though whether this was part of a centrally directed conspiracy or simply opportunistic behavior by local commanders is also unknown.

Once food reaches the county warehouse, the only check on delivery to the final institutional destinations-whether PDCs or targeted institutions-and on the use made of the food by those final destinations is through spot checks by WFP sub-offices. Although large-scale diversion at higher stages in the distribution chain is possible, it is at this lower level that monitoring is necessarily the weakest and diversion thus most likely to occur. The magnitude of the task of tracking supplies across tens of thousands of end-user institutions under abysmal working conditions should not be underestimated. Ironically, some NGOs, operating on a vastly smaller scale may actually have a more accurate grasp of where their contributions are ending up, despite the fact that they undertake less rigorous monitoring.

The behavior of these county-level institutions is difficult to characterize, but numerous accounts provide revealing information on what are certainly larger patterns. Dammers, Fox, and Jimenez (2005) reported on an EU-funded UNICEF program that distributes therapeutic milk, a product that can be fatal if administered incorrectly. According to the 2003/2004 agreement, the milk was to be provided to three provincial hospitals with properly-trained staff. However during a monitoring visit in November 2003, the EU's technical assistant discovered that the supplies were being distributed to baby-homes in the cities of Hyesan and Chongjin. The North Korean government then proposed for the 2004/2005 cycle that the product be distributed to 157 rehabilitation centers of various sorts, an alteration in terms of reference that Dammers, Fox, and Jimenez describe as without justification, costineffective, and potentially dangerous. These diverted supplies did not disappear into the ether: they were consumed, but not by the intended beneficiaries. This example is a small-scale one, but there are suggestions that much larger-scale diversion from intended purpose is occurring. Good Friends, a South Korean organization with long involvement with the issue, estimates that as much as 50 percent of Korean aid is going to non-deserving groups, including the military. In a particularly interesting development in 2004, the WFP reported that county authorities were buying and selling grain among themselves (WFP 2004). In all of these cases, humanitarian assistance is being consumed; the issue is not one of a deadweight or total loss. But donors have very little control, and aid is undoubtedly going to less deserving groups and therefore bypassing or only indirectly affecting targeted, vulnerable populations.

Given norms of accountability among donors and the limits on staffing resources even under the best of circumstances, the integrity of relief efforts are typically maintained through random, unannounced inspections. The North Korean government imposes restrictions on operations that make satisfactory monitoring of implementation and effectiveness through this means almost impossible:

• The North Korean government still has not provided a comprehensive list of institutions that benefit from WFP support despite repeated requests over a period of years.
• Pre-notification is required and visits to specific sites may be denied. The standard procedure is for the WFP to make weekly requests to visit facilities in particular regions that North Korean authorities review. In 2002, about eight percent of requests were denied. By 2003, this had fallen to one percent. WFP officials claim that they can increase the effective "randomness" embodied in this procedure, for example, by proposing to visit an orphanage in a particular county (of which there are, say, seven) and on visitation day demand to be taken to a particular orphanage of the seven possible.
• Interviewees at any given site cannot be chosen at random and the WFP is not allowed to interview households that are not already receiving aid --thus undercutting their ability to assess whether aid is going to the most needy or allocated on a politically discriminatory basis.
• When making monitoring visits, WFP staff are accompanied not only by local officials --who may be quite sympathetic to WFP concerns-- but by FDRC staff as well. Given the rigidly authoritarian nature of the political system, the presence of representatives of the central government stifles the creation of alliances and networks with local officials or the revelation of any information that may be unflattering to the government.

In December 1996, North Korean leader Kim Jong Il allegedly gave a speech commemorating the founding of Kim Il Sung University that was transcribed and smuggled out of the country. The speech is a wide-ranging review of the problems the country faced at the time, including the food problem. In it, Kim Jong Il admits that lower-level party and administrative officials had sought to mislead him about actual conditions on the ground. If Kim Jong Il himself has been unable to solve this fundamental information problem, is there reason to believe that the WFP will be more successful?

What appears to be at work is not centrally directed conspiracies (though these may exist), but rather local politics. The county-level administrators, who have enormous influence over the disposition of supplies, have a number of conflicting motivations, ranging from genuine desire to reach targeted groups and distribute food equally, to sincere differences with donors over priorities, to the universal phenomenon of local political "back-scratching," to personal pecuniary gain from diversion. The latter is undoubtedly important when considering that these mid-level government and party officials are themselves living on rapidly eroding won-denominated salaries.

How large is this diversion, and what effect does it have? No one knows for sure, but it is likely to be substantial. Good Friends recently estimated that as much as half of aid is diverted in the sense of going to non-targeted, privileged groups (Good Friends 2005). 2 Extensive interviews with a number of people affiliated with official and non-governmental organizations involved with humanitarian relief operations in North Korea who have intimate knowledge of the operations of their respective organizations, both public and private, yielded estimates of diversion that ranged from 10 percent to 30 percent. These estimates are significant, though well below the Good Friends figure. If one accepts the notion that there have been improvements in monitoring in recent years, then it could well be the case that losses were even higher in the past.

Some indirect evidence can be adduced from refugee surveys. One of the most astonishing things to come out of one recent survey of nearly 1,000 refugees is the relative absence of self-reported receipt of aid (Chang, forthcoming). Only 63 percent of the respondents in this survey reported even knowing of the existence of foreign humanitarian assistance. Ten years into the humanitarian effort, nearly 40 percent of the population remains unaware of the aid effort, despite the fact that it purports to target virtually all of the school-aged children in the country. Of those who knew of the program, only seven percent reported having received aid (or less than five percent of the total sample including those who were unaware of aid deliveries). These numbers do not imply that only seven percent of the population received aid, nor do they constitute proof of diversion. They do, however, testify to the extraordinary power of the government to control information. When asked who respondents thought were the primary recipients of aid, fully 98 percent responded "the military." Again, these responses do not prove that the military has been the primary recipient of food aid, but they are powerful testimony to North Korean views of how food is distributed in the country.

To get a rough sense of the magnitude of the estimates of diversion, the humanitarian effort organized by the WFP has aspired to provide at least a minimum ration to approximately 30 percent of the North Korean populace in recent years. If the estimate cited above is correct, it implies that the diversion of aid is sufficient to feed a significant share of the North Korean people. In light of the high real price of food in North Korea, and the astronomical rents that could be reaped through diversion, those who manage to get control of these supplies have a strong pecuniary incentive to both maintain the aid program, from where they derive their profits, and escape detection.

In the presence of markets, the welfare effects of diversion are ambiguous, however. Diversion directly moves food away from intended beneficiaries. But food is fungible to an important degree. To the extent that the recipients of diverted aid substitute it for food that they would have otherwise purchased, diversion tends to depress prices in the market where many of the beneficiaries or their families are, in reality, obtaining most of their food. Again, we know this because neither North Korean nor WFP estimates of daily PDS rations are sufficient to meet even the minimum caloric intake, even if corrections are made for the presence of other types of foods besides grains.

• This analysis leads to an important policy conclusion. In addition to gaining better access to the PDS, outside monitors should be tracking developments in markets, where signs of food distress often appear first as wildly fluctuating grain prices.

Is further progress likely? The record over the last year has been mixed. On one hand, more of the WFP's requests for monitoring visits have been denied than in the past. The total number of visits by WFP monitors has been reduced by roughly 40 percent, and North Korean authorities have restricted the nature of questions that the WFP can pose in their focus groups. Although a handful of previously closed counties were reopened, there is still 17 percent of the population of the country living in counties that are closed to inspectors.

On the other hand, the focus groups and more detailed questionnaires have provided an important window on household behavior. The WFP also appears to have reached an agreement in principle with the North Korean government to introduce a number of changes in the monitoring regime, including baseline surveys, closer monitoring of distribution centers and food-for-work projects, and the issue of new ration cards. The WFP is also discussing the introduction of modern inventory-management systems that would allow the WFP to track individual bags of grain electronically. If implemented, these changes would improve the monitoring climate, perhaps even substantially. However, they would probably still leave it short of standard humanitarian principles in some important respects, including most notably the small number of expatriate staff allowed in-country.