Analysis, Refugees & Migration

Refugees, Internal Displacement, and the Future of the Korean Peninsula

Refugees, and the fate of civilians in general, are critical to planning in case of dramatic political instability in North Korea. Most concerning strategically is that civilians attempting to flee to China would prompt the People’s Liberation Army (PLA) to intervene in North Korea, and could lead to further internationalization because the South and its allies would also intervene in response.1 Unfortunately, existing estimates of refugee flows are too vague to contribute to the careful contingency planning required.2 Still, we can arrive at better estimates of whether, when, and where North Koreans might seek refuge by examining civilian flight from current civil wars, and by customizing the projections with country-specific demographic, topographic, and infrastructural data.

The initial results of my ongoing research in this area challenge much of the conventional wisdom about potential North Korean refugees. First, evidence suggests that few refugees are likely to attempt to cross the DMZ. Most refugees that arrive in South Korea would have to arrive via third countries. Second, the scale and pace of refugee flows across China’s border crossings is likely to be substantially smaller and slower than others have predicted. Finally, internally displaced persons (IDPs) are likely to be far more numerous than refugees, presenting a serious—and so far underappreciated—humanitarian emergency and assistance delivery challenge.


Young North Korean Seeks Asylum in South Korea


North Korea is an “information poor” state. We simply lack accurate, reliable data to help us answer many of the most important political and social questions about the country. To date, the most widely circulated estimates of potential refugee flows into South Korea in the event of the North’s collapse rely on simple rubrics, for example assuming that one out of every four household members decides to flee to South Korea (and arriving at 3 million).3

Estimates of potential flows into China are even less transparent (often arriving at more than 3 million).4 However, we do know important things about North Korea that can help us to estimate refugee flows. Taken from the civil war literature on refugees, these include both “push” factors within the country, such as the immediate threat of violence or familial persecution, and “pull” factors, such as connections abroad with the funds to enable flight or easy access to a porous international border.5


North Korea’s population is concentrated in the south (approximately 9.4 million resided in Pyongyang and further south as of the 2008 census, including at least 8 cities with populations over 100,000).6 The majority of the country’s wealth and military power is concentrated there and it houses the central government.

The DMZ is one of the most militarized border zones in the world…

In the event of violent instability and civil war, Pyongyang and other important southern cities such as Kaesong and Nampo are likely to be heavily contested. All else being equal, this immediate threat of violence makes human flight much more likely and the DMZ is the most proximate means of fleeing the country. However, the following factors suggest that massive flight to the DMZ is unlikely:

  1. If conflict is concentrated in the south, populations will not flee toward the violence. Instead, they will head north or, if they are among the very few with access to a vessel, to the seacoast. There are not yet robust human-trafficking networks in coastal areas to ferry North Korean refugees abroad as there are in other ongoing civil wars in the Middle East and North Africa. Illicit trading networks could be reoriented, but this takes time—and most of the DPRK’s illicit traffic is associated with the Kim regime, so it may not be likely.7 The closest northern land border crossing is at Sinuiju/Dandong. From Pyongyang via roadways (the northernmost city in this area), Dandong is approximately 200 km away. From Kaesong (the southernmost city), the trip is over 485 km.

  3. The DMZ is one of the most militarized border zones in the world and is becoming more so as the DPRK continues to lay landmines.8 Unless and until South Korea secures a passage through the DMZ, and is able to communicate that opportunity to would-be refugees in the North, populations are unlikely to displace in that direction. The lack of access to technology, and especially cell phones, means that even in the rare chance that such passage is possible, knowledge of the opportunity will spread more slowly than it has in other recent conflicts.

  5. Populations in the areas closest to the DMZ are unlikely to have as much desire to displace. In most civil conflicts, the educated and socioeconomically advantaged are more likely to flee, and tend to flee earlier. These populations have “pull factors” including international social networks, foreign job opportunities, and outside money to gain passage out of the country and secure refuge elsewhere. While all of these pull factors are more compelling for the North Korean elite than they are for the general population, the intensity of the pull is weak. North Korea’s international isolation means few of its people have extensive international networks. Those most likely to have friends and family in the near abroad (with whom they are regularly in touch) are concentrated in distant Chinese border provinces such as North Hamyong (which contributed over 18,000 of 54,000 total documented defectors to South Korea as of June 2016) rather than the capital.9

Without very different conflict dynamics or South Korean contingency plans in place to secure and open the DMZ, it is difficult to imagine 25 percent of the regional population (as estimated) fleeing southward and successfully crossing into South Korea, even over the course of several years of conflict.


Human flight out of North Korea and into China is likely to be concentrated around three border cities. While individuals could theoretically cross at any point along the still-porous border, there are numerous obstacles to their doing so. Most pressing among them are the Yalu and Tumen rivers, which create a natural boundary between the two countries along most of the official border. The rivers are most passable during the winter, but of course, getting there is also more difficult in the cold. Man-made impediments to border crossings exist at various points, including fencing, walls, and military and police patrols. Additionally, even those who might veer off of well-traveled roadways in order to evade capture are likely to take roads on their way to the border and, therefore, are not likely to stray too far. The average North Korean does not have access to motorized transportation, so walking is the most likely form of transportation (some may bike as well).

There are three major border crossing city pairs, DPRK side (China side): Sinuiju (Dandong), Manpo (Ji’an), and Namyang (Tumen). Four North Korean counties are adjacent to China and two are adjacent (one province and one special city) to Russia.10 Those along the Chinese border have a total population of 7.1 million. Assuming a brisk walking pace of approximately 5 km per hour, unladen, and with no depreciation in speed over time, it would take between 2.34 and 80.1 hours to walk to the nearest border crossing from those provinces.11 Among those provinces that are not adjacent, the closest walk is over 24 hours and the farthest is nearly 195 hours, or nearly 20, 10-hour days of walking without rest or slowing. The true rate of travel for most refugees would be significantly slower, especially considering the age and health of the North Korean population.

Human flight out of North Korea and into China is likely to be concentrated around three border cities.

The conflict in Syria, a full-scale civil war since the end of 2011, has produced 4.9 million refugees according to the United Nations High Commissioner for Refugess (UNHCR).12 For various reasons, including its recency, population size, physical and socioeconomic characteristics, and the nature of the war (internationalized civil war including conventional and unconventional force), this case provides a more appropriate refugee analogy than the oft-referenced East German case. For our purposes, the significant differences between the two are access to technology, transportation, and the availability of border crossings. Syria’s population had widespread access to cell phones and mechanized transport. Its borders are also substantially longer and more passable. In sharp contrast to North Korea, there are at least 34 formal border crossings between Syria and its neighbors and only small portions have natural water boundaries like rivers or seacoasts. As with the travel time and distances presented above, the Syrian case should, therefore, err high with respect to North Korea.


The first refugees from Syria recorded by the UNHCR did not cross international borders until nearly four thousand wartime casualties had been recorded—nine months into the conflict. Nearly two years of fighting passed before 500,000 refugees were recorded. Afterward, at least 100,000, and sometimes over 200,000, refugees were recorded each month in the second and third years of the conflict. This was not due to dramatic increases in violence, though. Instead, it was due to the loss of basic services and sustainable livelihood coincident with the war.

Syria’s biggest prewar cities, Aleppo (45 km) and Damascus (56 km) are much closer to accessible international borders than are North Korea’s. Therefore, even given a similar scale of violence within its major cities, we should expect North Korean refugees to flee toward China over the course of several years of fighting, and even then, to arrive more slowly than Syrian refugees due to the distance and difficulty of travel. A critical “push” factor should also be the loss of sustainable civilian life in the major cities. Yet North Korea’s population has become inured to difficult living conditions (e.g., famine, drought, and economic sanctions) and may still not flee under conditions where other populations would.


Even where there is limited opportunity to cross an international border, as in North Korea, civilian flight from conflict zones is still likely to occur. In the event of major instability, internal displacement away from violence, short of emigration, is empirically the most common civilian response. 13 Even where the immediate threat of violence is small, internal migration is likely to occur when the population lacks food and other human necessities.14 There is also a strong disincentive for human flight to China due to its well-known tendency to deny asylum and send would-be North Korean refugees back.15 Civilians may therefore decide that being returned to the DPRK authorities makes internal displacement more attractive than attempting to become a refugee. Though the scale and location of human flight depend on a conflict’s characteristics, it is likely that internal displacement will significantly outpace any international flight in the event of North Korean instability. If this is true, populations stuck inside North Korea, without access to basic necessities, will also be exceedingly difficult for the international community and humanitarian agencies to reach. So long as the Kim regime has not collapsed entirely, outsiders will have to negotiate with the regime in order to provide help—a notoriously complicated effort even in the best of times.16

There is also a strong disincentive for human flight to China due to its well-known tendency to deny asylum and send would-be North Korean refugees back.

In sum, a refugee crisis in the event of dramatic North Korean instability is likely to be much smaller than most suggest, but the internal humanitarian crisis is likely to be substantial. In the event of major violence, civilians will flee, but there will be fewer opportunities for North Koreans to cross international borders, and the flight itself will be more arduous than what we are currently seeing from Syria. Insofar as international escalatory dynamics are concerned, this is good news because China would not face an immediate need to intervene to stem the tide of refugees across its border. However, this also means that internal humanitarian aid delivery should be the overwhelming priority in contingency planning, since most civilians will not displace across international borders.17


  1. Paul B. Stares and Joel S. Wit, “Preparing for Sudden Change in North Korea,” Council on Foreign Relations, Center for Preventive Action, Special Report No. 42, January 2009; Bruce W. Bennett, “Preparing for the Possibility of North Korean Collapse,” Santa Monica, CA: RAND Corporation, 2013.
  2. Interested governments including South Korea, the United States, and China may have produced detailed estimates; however, these studies have not been released to the public.
  3. A Bank of Korea (BOK) study predicts that 3 million refugees would try to cross into the South if the North were to collapse. Jeong-ju Na, “3 million NK Refugees Expected in Crisis: BOK,” Korea Times, January 26, 2007. On the lower end, Stares and Wit report that “South Korean planners have projected collapse scenarios in which up to one million refugees might flee the North with more than five hundred thousand entering China, up to three hundred thousand to South Korea, and the rest to Russia and Japan.” Stares and Wit, “Preparing for Sudden Change in North Korea,” 23. However, the authors do not discuss the political or security dynamics prompting that level of human flight, nor do they provide details as to the timing or route of those refugees.
  4. In addition to endorsing the Bank of Korea estimate of 3 million refugees to South Korea, Bennett argues that an “even larger number” would try to cross into China. Bennett, “Preparing for the Possibility of North Korean Collapse,” xviii; Na, “3 million NK Refugees Expected in Crisis: BOK.”
  5. Christian A. Davenport, Will H. Moore, and Steven C. Poe, “Sometimes You Just Have to Leave: Domestic Threats and Forced Migration, 1964–1989,” International Interactions 29, 2003, 27–55; Will H. Moore and Stephen M. Shellman, “Refugee or Internally Displaced Person?: To Where Should One Flee?,” Comparative Political Studies 39:5, 2006, 599–622; Will H. Moore and Stephen M. Shellman, “Whither Will They Go? A Global Study of Refugees’ Destinations, 1965–1995,” International Studies Quarterly 51: 4 (December 2007), 811–34; Jacqueline H. Rubin and Will H. Moore, “Risk Factors for Forced Migrant Flight,” Conflict Management and Peace Science 24, 2007, 85–104; Idean Salehyan, “Refugees and the Study of Civil War,” Civil Wars 9:2, 2009, 127–41; Prakash Adhikari. “Conflict Induced Displacement, Understanding the Causes of Flight,” American Journal of Political Science 57:1 (January 2013), 82–89; Stephanie Engel. “Displacement Due to Violence in Colombia: A Household-Level Analysis,” Economic Development and Cultural Change 55:2 (January 2007), 335–65; Mathias Czaika and Krisztina Kis-Katos, “Civil Conflict and Displacement: Village-Level Determinants of Forced Migration in Aceh,” Journal of Peace Research 46:3, 2009, 399–418; Vera Mironova, Loubna Mrie, and Sam Whitt, “Fight or Flight in Civil War? Evidence from Rebel-Controlled Syria,” Working Paper, 2014.
  6. All of the numbers presented here referring to North Korea’s population come from the 2008 census. North Korea’s population since the census is estimated to have grown at 0.85 percent annually, though it is not clear whether and how internal population growth and settlement patterns have been affected. Therefore, the 2008 numbers were not adjusted for growth. DPRK Central Bureau of Statistics, “DPR Korea 2008 Population Census, National Report,” Pyongyang, DPR Korea, 2009.
  7. Sheena E. Chestnut, “The ‘Sopranos State?,’ North Korean Involvement in Criminal Activity and Implications for International Security,” Nautilus Institute Special Report, 109, 2005.
  8. “North Korea plants land mines near border to prevent defection by soldiers: sources,” Korea Times, August 23, 2016,
  9. According to the South Korean government:
  10. Potential refugee flows to Russia are not included here.
  11. Walking pace estimates come from Annie A Butler, Jasmine C Menant, Anne C Tiedemann, and Stephen R Lord, “Age and gender differences in seven tests of functional mobility,” Journal of NeuroEngineering and Rehabilitation, 6:31, 2009, doi:10.1186/1743-0003-6-31.
  12. Syrian refugee data come from the UNHCR:
  13. According to Davenport, Moore, and Poe, “the transaction costs associated with relocation are substantial, and . . . movement across international borders will be the exception rather than the norm.” Davenport, Moore, and Poe, “Sometimes You Just Have to Leave: Domestic Threats and Forced Migration, 1964–1989,” 41.
  14. Victor D. Cha, David C. Kang and Leif-Eric Easley, Korea After Unification: Planning for the Inevitable (contracted Columbia University Press). Cha, Kang, and Easley provide a detailed discussion of the potential refugee and IDP problem in North Korea given a non-violent unification. They concur that many estimates of North Korean out-migration err high and that internal movement is more likely. Further, emigration in this circumstance is “more likely to resemble those after the collapse of communist regimes in Europe than the 2012-2014 refugee flows out of Syria.”
  15. Many, including the UNHCR, South Korea, and the United States, criticize Beijing’s policies toward North Korean migrants and lack of adherence to its international commitments to protect them (Lankov 2004 868). Drew Thompson and Carla Freeman, Flood Across the Border: China’s Disaster Relief Operations and Potential Response to a North Korean Refugee Crisis (Washington, DC: US-Korea Institute at SAIS, 2009); Joshua Kurlantzick and Jana Mason, “North Korean Refugees: The Chinese Dimension,” in The North Korean Refugee Crisis: Human Rights and International Response (Washington, DC: The Committee for Human Rights in North Korea, 2006); Andrei Lankov, “North Korean Refugees in Northeast Asia,” Asian Survey 44, No. 66: 856-73.
  16. Although North Korea’s neighbors have begun to coordinate, the broader international humanitarian aid community has been critical of the region’s response to existing North Korean refugees and has criticized the lack of long-term planning incorporating IGOs, NGOs, and other potential partners. Marcus Noland, Stephan Haggard, and Yoonok Chang, The North Korean Refugee Crisis: Human Rights and International Response (Washington, DC: The Committee for Human Rights in North Korea, 2006).
  17. Please note that these are preliminary findings. As the research progresses, detailed simulations using varied conflict dynamics and geo-coded data will provide finer-grained estimates incorporating factors such as county-level demographics and topography, factors affecting individuals’ travel speed, and seasonal changes.