U.S. AGENCY FOR INTERNATIONAL DEVELOPMENT
USAID Program and Operations Assessment Report No. 27
Center for Development Information and Evaluation
December 2000
PNñACGñ605
COMPLEX HUMANITARIAN EMERGENCIES
And USAIDíS HUMANITARIAN RESPONSE
This report and others in the evaluation publication series
of the Center for Development Information and Evaluation
(CDIE) can be ordered from:
USAID Development Experience Clearinghouse (DEC)
1611 N. Kent Street, Suite 200
Arlington, VA 22209ñ2111
Telephone: (703) 351ñ4006
Fax: (703) 351ñ4029
E-mail: docorder@dec.cdie.org
To access CDIE documents from the Internet, key in
www.usaid.gov. Click on Publications/Partner
Resources, then on USAID Evaluation Publications.
The CDIE Evaluation Publications Catalog and notices of
recent publications are also available from the DEC.
U.S. AGENCY FOR INTERNATIONAL DEVELOPMENT
The views and interpretations expressed in this report are those of the author
and not necessarily those of the U.S. Agency for International Development.
USAID Program and Operations
Assessment Report No. 27
Complex Humanitarian Emergencies
And USAID’s
Humanitarian Response
By
Donald G. McClelland
Center for Development Information and Evaluation
with
Elizabeth Adelski
Richard Hill
John Mason
Robert Muscat
International Resources Group, Inc.
Center for Development Information and Evaluation
U.S. Agency for International Development
Washington
December 2000

Contents
Preface ........................................................ v
Summary ................................................ vii
1. Introduction ........................................ 1
Number of Emergencies
And People Affected ....................... 1
Resource Implications ........................ 2
Legislative Authority .......................... 4
Complex Humanitarian
Emergencies ....................................... 5
General Evaluation
Approach ............................................ 6
2. Country Context ................................. 9
Causes of Complex
Emergencies ..................................... 9
USAID’s Humanitarian
Response ......................................... 12
3. Results: Humanitarian Effects ........ 15
Haiti .................................................... 15
Mozambique ...................................... 17
Rwanda .............................................. 19
Bosnia–Herzegovina ......................... 20
Targeting ............................................ 21
Efficiency ............................................ 22
Conclusion ......................................... 23
4. Results: Political Effects and
Effects on Hostilities ....................... 25
Haiti .................................................... 25
Mozambique ...................................... 26
Rwanda .............................................. 27
Bosnia–Herzegovina ......................... 28
Policy Implications ........................... 29
Conclusion ......................................... 32
5. Results: Economic Effects ............... 33
Haiti .................................................... 34
Mozambique ...................................... 35
Rwanda ............................................... 36
When Is the Emergency Over? ........ 38
Conclusion ......................................... 40
6. Conclusions, Lessons Learned,
And Recommendations .................. 43
Conclusions and
Lessons Learned ............................... 43
Recommendations ............................ 46
Annex A Humanitarian Emergencies
And Donor Assistance
Annex B Evaluation Objectives
And Constraints and Implications
For Donor Coordination
Annex C Are Complex Emergencies
Predictable?
Annex D Implications for the Kosovo Crisis
Bibliography

OMPLEX HUMANITARIAN emergen￾cies are by their nature multifaceted
and involve many actors. The relief inter￾ventions are often undertaken in a context
beyond the control of the implementing
agencies. Moreover, the interventions are
generally conditioned by overall foreign
policy considerations, which means that
political objectives help define the re￾sponse.
Operational coordination is compli￾cated because multiple players are in￾volved (various U.S. government agencies,
other bilateral and multilateral develop￾ment agencies, nongovernmental and pri￾vate voluntary organizations, and the host
country) and these actors often have di￾vergent approaches to strategic planning,
decision-making, and delivery mecha￾nisms. That relief experts have different
views of the purpose of emergency assis￾tance—whether it is for relief only, or for
rehabilitation and economic development
Preface
as well—exacerbates an already complex
situation.
Given these circumstances, it is not
surprising that humanitarian assistance
has been subjected to less rigorous and
extensive monitoring and evaluation than
development assistance. In addition to the
complexities just noted, this reflects the
fact that until recently there was no stan￾dard methodology for evaluating humani￾tarian assistance. Some have likened this
situation to “methodological anarchy.”
Nevertheless, it is possible to assess
the impact of humanitarian assistance on
vulnerable populations to some degree
and to shed light on the relationship be￾tween emergency assistance and the po￾litical and development processes at work.
This assessment and its lessons learned
should contribute to formulating more ef￾fective policies and interventions in re￾sponse to complex emergencies.
C

Summary
N 1998, some 32 million people needed
humanitarian assistance because they
were caught up in complex emergencies
(armed conflicts or civil wars as distinct
from natural disasters). That is triple the
number of a typical year from the early
1980s. Most of these people are refugees
or internally displaced persons. About 40
percent reside in Africa. The value of hu￾manitarian assistance worldwide has typi￾cally been less than 2.0 percent of official
development assistance (ODA). But in 1994
it jumped to 6.8 percent ($4.3 billion) ow￾ing to the crisis in the Great Lakes region
of Africa. It fell to an estimated 5.7 per￾cent of ODA in 1998 but is likely to peak
again in 1999 owing to the Kosovo and
East Timor crises. The fact remains that hu￾manitarian assistance has more than
doubled since 1990 despite diminishing
foreign assistance.
In 1990, U.S. ODA totaled nearly $13.6
billion in real terms (1998 dollars); by 1997,
it had fallen by half to $7.0 billion, the low￾est level since World War II. By contrast,
U.S. humanitarian assistance has in￾creased. In 1990, it was $263 million (1.9
percent of ODA). In 1994 it peaked at $1.2
billion (11.4 percent of ODA). By 1997 it had
gradually decreased to $344 million (4.9
percent of ODA), but in 1998 it more than
doubled to $898 million (10.2 percent of
ODA). Although U.S. ODA as a percentage
of total ODA has been falling steadily since
the 1970s, the United States continues to
be a generous provider of humanitarian
assistance. USAID’s Food for Peace Office,
Office of Foreign Disaster Assistance, and
Office of Transition Initiatives are prima￾rily responsible for administering U.S.
emergency assistance.
This evaluation seeks to assess the ef￾fectiveness of U.S. humanitarian assistance
in nations afflicted by complex emergen￾cies. It addresses three principal questions:
Did U.S. emergency assistance save lives
and alleviate suffering? Did it affect social
tensions and political hostilities? Did it
contribute to long-term economic devel￾opment? The findings are based on field￾work carried out in three countries (Haiti,
Mozambique, and Rwanda) as well as on
evaluation results in other countries.
Evaluating relief programs in the con￾text of armed conflict or civil war intro￾I
viii Complex Humanitarian Emergencies and USAID’s Humanitarian Response
duces politically sensitive issues concern￾ing sovereignty, international law, the
appropriate balance of aid between oppos￾ing sides, and donors’ foreign policy
interests. This assessment treats the politi￾cal effects of humanitarian assistance
independently of U.S. foreign policy con￾siderations. Nevertheless, these and other
issues peculiar to complex emergencies
made the evaluation methodologically
more difficult.
Complex emergencies are typically
political in nature, characterized by vio￾lent conflict (often war) and a breakdown
of institutions. But their underlying causes
vary. Predatory governance was the prin￾cipal cause of Haiti’s complex emergency.
By contrast, ethnic and ideological factors
were pivotal in Mozambique and Rwanda.
In all three countries, poverty was a con￾tributory factor. In Haiti, per capita income
was $250 in 1994; in Mozambique, $80 in
1986, the lowest in the world. Poverty was
just as severe in Rwanda. Moreover, the
distribution of income and wealth was
highly skewed in all three countries.
Haiti’s 200-year history has been charac￾terized by oppressive governments that fa￾vored the rich at the expense of the poor.
Mozambique was characterized by a
highly dualistic economy. In Rwanda,
where the proportion of people living in
poverty increased from 40 percent to 70
percent during 1990–93, a winner-take-all
mentality has benefited a tiny elite at the
expense of the poor majority.
Civilians in all three countries suf￾fered widespread and systematic human
rights abuses. Tens of thousands of refu￾gees fled Haiti (often as boat people). Mil￾lions fled Mozambique and Rwanda to
escape indiscriminate terror. Hundreds of
thousands were the victims of wholesale
massacre or, in the case of Rwanda, geno￾cide.
Donors, including USAID, responded
with increased emergency assistance, both
food and nonfood (water, seed, farming
tools, medical supplies). Nongovernmen￾tal organizations were the main im￾plementers of the humanitarian response.
In Haiti, the international community was
feeding 1.3 million people—one in seven
Haitians—each day, with the United States
providing 68 percent of the food. In
Mozambique in 1989, an estimated one
third of the population of 16 million de￾pended on food aid for 60 to 70 percent of
their food needs; again, the United States
provided about 60 percent of total food aid
during 1987–95. In Rwanda 1.3 million
beneficiaries received emergency food aid
in 1996–97.
What were the results? The assess￾ment concluded that emergency assistance
programs funded by USAID and imple￾mented by U.S. nongovernmental agencies
(NGOs) clearly helped save lives and alle￾viate suffering—which, after all, is their
overarching objective. Except for Haiti,
though, data collection and monitoring
were not done (or were done poorly), so it
is difficult to quantify results. In fact, most
evaluations of humanitarian assistance tell
a “mission accomplished” story but are
Summary ix
unable to substantiate that story with hard
data.
Distributing relief supplies was a
problem to some extent in all three coun￾tries. Food aid, in particular, was highly
valued and became a source of violent
competition—not only for its value as food
but also as a source of political power for
those controlling access. There were re￾ports of corruption, theft, and political or
personal favoritism in food aid distribu￾tion. And target populations did not al￾ways receive timely and sufficient food.
NGOs addressed these problems with vary￾ing degrees of success. In Haiti they were
highly successful in limiting diversion to
5 to 10 percent. In Mozambique, leakage
was typically 30 percent when the govern￾ment was in charge of distribution, and at
one point reached 50 percent. But after the
NGOs took over, losses fell to under 5 per￾cent. In Rwanda the military and former
political leaders controlled much of the
relief distribution. They were able to di￾vert substantial quantities of food (more
than is usually the case in complex emer￾gencies) from the intended beneficiaries
for their own purposes.
While no aid is apolitical, humanitar￾ian assistance, in particular, can result in
substantial and unpredictable political ef￾fects, since it is provided in the context of
conflict. Though designed to relieve suf￾fering and promote peace, it sometimes,
inadvertently, fuels, sustains, or worsens
complex emergencies by making more re￾sources available to warring parties. This
is because aid does not just keep people
alive in a political vacuum but affects the
local power structure and environment in
which it is given.
In Haiti, massive quantities of emer￾gency food aid reduced the probability of
food riots during a period of political and
economic stress and may have had a
dampening effect on political tensions; but
it also may have contributed to a political
status quo that enabled the de facto mili￾tary regime to stay in power longer. In
Mozambique, external military assistance
provided by the Soviet Union and South
Africa fueled the civil war; food aid, by
comparison, had relatively little effect on
the country’s political dynamics, although
food diverted to soldiers may have con￾tributed to the war effort. In Rwanda,
genocidal killers were mixed with legiti￾mate refugees in camps; targeting became
problematic, and substantial quantities of
food aid were diverted by Hutu extrem￾ists and militia resident in the camps. That
had the unintended effect of prolonging
the conflict.
The notion that relief assistance can
be made more developmental in the con￾text of ongoing armed conflicts is problem￾atic. Unlike with natural disasters, during
complex emergencies there is no institu￾tional framework to provide physical se￾curity and political stability—both of
which are necessary preconditions for eco￾nomic development. On the contrary, com￾plex emergencies are often characterized
by a total breakdown of state institutions
and social and economic structures.
x Complex Humanitarian Emergencies and USAIDís Humanitarian Response
Nevertheless, emergency assistance
programs can help shape the pattern and
direction of subsequent economic devel￾opment. In Haiti, Mozambique, and
Rwanda USAID and the NGOs not only pro￾vided immediate relief (food, medicine)
but also agricultural inputs (seed, tools)
and household goods to encourage refu￾gees and internally displaced persons to
return to their villages, resume food pro￾duction, decrease their dependence on
food aid, and maintain their livelihoods.
They also implemented food-for-work
programs in all three countries. These and
other programs created short-term jobs
and helped rehabilitate productive infra￾structure (roads, irrigation) needed for
economic development. The development￾oriented objectives were clear: to restart
subsistence agriculture and to restart the
rural economy.
The assessment offers 4 management￾oriented recommendations (summarized
below) and 18 recommendations specific
to the Kosovo crisis as of May 1999
(annex D).
■ Monitoring and evaluation. Establish a
central monitoring and data collection unit
to serve all donors during the early weeks of
a complex emergency. This is needed,
among other things, to help managers
identify appropriate kinds of emer￾gency relief, target its distribution,
evaluate its effectiveness, and enhance
donor coordination.
■ Adverse political consequences. Be alert
to potential undesirable political or social ef￾fects that relief aid may cause. Control of the
distribution of food aid, in particular, can
reinforce the power of local authorities
or political factions; it can also facilitate
their self-aggrandizing, often exploitive,
behavior toward the intended noncom￾batant beneficiaries.
■ Reducing dependency. Give refugees in￾centives to return home, and impose disin￾centives on those remaining outside their
country of origin. After populations have
been repatriated and are settled, the ag￾ricultural base begins to be reestab￾lished, dependency on free food drops,
and long-term food security is en￾hanced.
■ Capacity building. Train technocrats to
manage the postconflict economic transition,
and train others in skills for which there is
employment demand. Economic recovery
requires a cadre of high-level techno￾crats with management and conceptual
skills; it also requires the unemployed
(especially demobilized soldiers) to be
trained in marketable skills.
Finally, however one assesses the ef￾fectiveness of humanitarian assistance in
response to complex emergencies, one
thing cannot be emphasized too strongly:
it is far better to prevent complex emer￾gencies from occurring in the first place
than it is to respond to victimsí needs af￾terwards.
INCE THE END of the Cold War, the na￾ture of international assistance needs
has changed dramatically. Ethnic and na￾tional tensions have led to increased civil
strife and an explosion in the number of
complex humanitarian emergencies. As a
result, the number of civilian casualties has
increased, as has the level of emergency
assistance allocated in response to their hu￾manitarian needs. In 1998, USAID’s Center
for Development Information and Evalu￾ation (CDIE) initiated an assessment of the
effectiveness of the Agency’s humanitar￾ian assistance interventions. The assess￾ment examined USAID programs in three
countries afflicted with complex emergen￾cies: Haiti, Mozambique, and Rwanda.
This report synthesizes the findings of the
three separate country studies.
Number of Emergencies
And People Affected
According to the U.S. Mission to the
United Nations (1997, 5), the number of
humanitarian emergencies in which at
least 300,000 civilians depended on inter￾national humanitarian assistance to avoid
serious malnutrition or death peaked dur￾ing 1993–95. With improved situations in
several countries (including Armenia,
Cambodia, and Mozambique), the num￾ber of emergencies dropped to 20 in 1996
and remained at that level in 1997 (p. 5).
In 1998, Russia (Chechnya) was dropped
from the list (reducing the number to 19),
but Colombia and Uganda were added
(boosting it to 21) (U.S. Mission to the UN
1998, 7, 9). Annex A (table A1) lists ongo￾ing humanitarian emergencies in 1996,
1997, and 1998 using data from the U.S.
Committee for Refugees.
Worldwide, roughly 33 million
people needed emergency assistance in
January 1996. That increased to 34 million
in January 1997, then decreased to an esti￾mated 32 million in April 1998 (table A1).
These levels are triple those typical of the
early 1980s. They include both internally
displaced persons who have remained
within their own borders and refugees
who have fled across international bor￾ders. During the 1990s most have been in￾ternally displaced persons rather than
refugees. In 1996, 52 percent of those re￾quiring emergency assistance owing
S
1 Introduction
2 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
mainly to armed conflict or government
repression resided in sub-Saharan Africa;
in 1997 that percentage decreased to 48
percent, and in 1998 it decreased still fur￾ther to 39 percent (table A1).
Resource Implications
The Development Assistance Com￾mittee of the Organization for Economic
Cooperation and Development monitors
levels of official development assistance
and humanitarian assistance. Table 1 re￾ports these data for all donors in 1998 dol￾lars during 1988–98. In real dollar terms,
official development assistance declined
gradually from a high of $68.5 billion in
1991 to a low of $53.4 billion in 1997, a de￾crease of 22 percent. In 1998 it increased
for the first time since 1994; the increase
was 8.1 percent. (The small increase in 1994
was probably due to the Rwanda crisis.)
Humanitarian assistance peaked at
$4.3 billion in 1994 owing to the crisis in
the Great Lakes region of Africa. After that
it fell to $3.3 billion as of 1998. However,
it was expected to peak again in 1999 be￾cause of the emergencies in Kosovo and
East Timor and the Turkish and Taiwan￾ese earthquakes. Humanitarian assistance
had typically been less than 2.0 percent of
official development assistance—until
Source: OECD DAC/o database
Note: Annex A includes two figures that graphically depict dollar levels of foreign assistance
and humanitarian assistance (figure A1), and humanitarian assistance as a percent of overall
assistance (A2) during the 30-year period 1969–98.
Year
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
Official Development
Assistance
61,570
60,172
63,791
68,503
67,792
62,659
63,176
56,968
56,530
53,424
57,774
Humanitarian Assistance
955
969
1,270
3,503
2,880
3,863
4,303
3,401
3,206
2,921
3,288
HA as a Percent Of ODA
1.6
1.6
2.0
5.1
4.2
6.2
6.8
6.0
5.7
5.5
5.7
Table 1. Official Development Assistance and Humanitarian
Assistance, in Millions of 1998 US$, 1988–98
Introduction 3
1991, when it jumped to 5.1 percent (ow￾ing to violence in the Balkans). It peaked
at 6.8 percent of official development as￾sistance in 1994 with the Rwanda crisis,
but afterward fell to 5.7 percent of ODA in
1998. The fact remains that within a dimin￾ishing overall foreign aid budget, humani￾tarian assistance has nearly tripled since
1990.
What about the United States? In 1990,
U.S. official development assistance to￾taled nearly $13.6 billion in real terms
(1998 dollars); by 1997, it had fallen by half
to $7.0 billion in real terms (see table 2).
This was the lowest level of U.S. assistance
since World War II (Miller 1997, 1). U.S.
humanitarian assistance, by contrast, has
increased. In 1990 it was $263 million (1.9
percent of ODA); in 1994, it peaked at $1.2
billion (11.4 percent of ODA). Since then it
has gradually decreased to $344 million
in 1997 (4.9 percent of ODA). In 1998 both
U.S. ODA and U.S. humanitarian assistance
increased, and in 1999 humanitarian as￾sistance was expected to increase again
given current humanitarian needs.
Although U.S. official development
assistance as a percentage oftotal ODA has
been falling steadily since the 1970s, the
United States continues to be a generous
provider of humanitarian assistance. In
1998, for example, the United States allo￾Source: OECD DAC/o database
Note: Annex A graphically depicts trends in U.S. ODA and U.S. humanitarian assistance in 1998
dollars during 1971–98 (figures A3 and A4).
Year
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
Official Development
Assistance
13,141
9,547
13,580
12,951
13,146
11,099
10,662
7,743
9,669
6,959
8,786
Humanitarian Assistance
220
261
263
685
585
733
1,216
829
603
344
898
HA as a Percent of ODA
1.7
2.7
1.9
5.3
4.4
6.6
11.4
10.7
6.2
4.9
10.2
Table 2. U.S. Development Assistance and Humanitarian
Assistance, in Millions of 1998 US$, 1988–98
4 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
cated over 10 percent of its ODA to humani￾tarian assistance needs. However, to meet
humanitarian needs, U.S. policymakers
have had to divert resources away from
sustainable development programs
(Messer 1998, 15, citing USAID’s FY97 Con￾gressional Presentation). Any proliferation
of complex emergencies is likely to tighten
the squeeze on sustainable development
programs even further. Ironically, devel￾opment assistance programs designed to
spur economic growth and reduce poverty
may help mitigate the need for more ex￾pensive responses to complex emergen￾cies, since they reduce the likelihood of
their occurring in the first place.
Legislative Authority
The United States donates food aid to
victims of floods, earthquakes, droughts,
and civil strife under Title II of the Agri￾cultural Trade Development and Assis￾tance Act of 1954 (PL 480). Other types of
humanitarian assistance are provided un￾der the Foreign Assistance Act of 1961 (as
amended). Normally, the duration of both
food and nonfood emergency assistance
is limited, and only countries that lack re￾sources to purchase commodities commer￾cially are eligible recipients (GAO 1986, 10).
USAID’s Office of Food for Peace ad￾ministers the PL 480 Title II food aid pro￾gram. Under this program the United
States provides emergency food aid to co￾operating sponsors who in turn distribute
it to disaster victims. Food commodities
are distributed in areas of greatest need
with priority given to people suffering
from malnutrition. Cooperating sponsors
can be (1) governments, (2) multilateral or￾ganizations such as the World Food Pro￾gram, or (3) nonprofit U.S. private volun￾tary organizations (PVOs) such as Catho￾lic Relief Services, Cooperative for Assis￾tance and Relief Everywhere (CARE),
Adventist Development and Relief
Agency, and World Vision. Cooperating
sponsors are responsible for establishing
distribution networks to reach disaster
victims and for properly storing and ac￾counting for commodities. USAID is re￾sponsible for the overall administration
and management of the program.
An emergency response normally re￾quires not only food but also nonfood as￾sistance, including medicine, sanitation,
potable water, agricultural inputs, and
shelter. USAID’s Office of Foreign Disaster
Assistance and Office of Transition Initia￾tives are primarily responsible for admin￾istering nonfood humanitarian assistance.
OFDA coordinates the allocation of funds
appropriated under the International Di￾saster Assistance account. The largest per￾centage of funds goes to relief and reha￾bilitation project grants managed by pri￾vate voluntary, nongovernmental, and in￾ternational organizations.
OFDA has an internal policy not to ob￾ligate funds for longer than 12 months at
a time. That allows it to respond to unex￾pected crises worldwide. Food for Peace
has a similar policy for emergency food
aid. The Office of Foreign Disaster Assis￾tance is widely respected for its respon-
Introduction 5
siveness to disasters. Although it is
permitted to provide responses aimed at
relief, rehabilitation, and reconstruction,
OFDA’S lifesaving emphasis is paramount.
OFDA is concerned about potential con￾gressional criticism if it permits missions
or embassies to use emergency assistance
for longer term developmentally related
interventions (Miller 1997, 27).
Complex Humanitarian
Emergencies
The term “complex emergency” was
first coined in UN circles, probably in
Mozambique, as a diplomatic euphemism
for a “chronic political” rather than “natu￾ral” emergency. The Joint Evaluation of
Emergency Assistance to Rwanda (1996,
study 2, 5) points out that complex emer￾gencies
tend to have multiple causes, but are
essentially political in nature and entail
violent conflict. They typically include a
breakdown of legitimate institutions and
governance, widespread suffering, and
massive population displacements, and
they often involve and require a range
of responses from the international com￾munity, including intense diplomacy and
conflict resolution efforts, UN policing
actions, and the provision of multilat￾eral and bilateral humanitarian assis￾tance by official and private agencies.
A complex emergency tends to be very
dynamic, characterized by rapid
changes that are difficult to predict.
Brandt (1995, 1) suggests that complex
humanitarian emergencies are often wars.
In the post–Cold War period, most have
been conflicts that have taken place within,
not between, countries. These internal
struggles among warring factions are usu￾ally defined by ethnicity, religion, or lan￾guage. According to Apthorpe (1997, 91),
complex emergencies have deep roots and
dense branches and cannot be understood
from a Western-ethnocentric perspective.
Similarly, Kleist (1994, 45) suggests that a
disaster may be defined as complex when
its origins are multiple and its effects com￾pound one another. As USAID’s 1998 Per￾formance Report (131) notes, complex emer￾gencies are manifested by “armed conflict,
death, displaced populations, hunger, and
injury.”
Humanitarian aid (donations of food
and other commodities and services) is in￾tended to save lives in situations where
virtually everyone is at exceptionally high
risk. These situations require getting as￾sistance to where it is needed—urgently.
The humanitarian response is primarily an
act of rescue. Under the umbrella of hu￾manitarian assistance, there is a distinction
between relief and rehabilitation (Kleist,
47). Relief helps people survive; rehabili￾tation helps people get back on their feet
so they can reestablish their livelihoods.
Development aid, by contrast, is nor￾mally long term and sustainable in nature.
In economic terms, it can be characterized
as investment rather than consumption.
There is no clear, operational definition of
when short-term relief ends and long-term
development begins. For all practical pur￾poses, though, relief activities end with the
termination of emergency resources. And
this occurs when donors decide to cease
providing these resources. This suggests
that the relief-to-development continuum
may exist conceptually, but not operation￾ally.
General Evaluation
Approach
The overall objective of the CDIE
assessment was to examine the effective￾ness of U.S. humanitarian interventions,
especially emergency food aid, in nations
afflicted by a complex emergency. It
addressed three principal questions:
1. Did U.S. emergency assistance
save lives and alleviate suffering during
the complex emergency (humanitarian
effects)?
2. Did U.S. emergency assistance
affect social and political hostilities or
tensions associated with the complex
emergency (political effects)?
3. Did U.S. emergency assistance
contribute to long-term development
(economic effects)?
The assessment examined the results
of humanitarian assistance. It did not delve
into the various agencies involved in
implementation. In short, it did not evalu￾ate the implementing agencies, but rather
the results of their activities. The findings
are based on fieldwork carried out in three
countries: Haiti (McClelland 1999),
Mozambique (Lieberson 1999), and
Rwanda (Renison 2000). They also draw
on syntheses of related evaluation results
including in particular Apthorpe (1997)
covering six evaluations of humanitarian
assistance in Africa; Borton and Macrea
(1997) covering 28 evaluations worldwide;
and the UNHCR/WFP (1998) evaluation of
the experience in Bosnia and Herzegovina.
CDIE evaluation teams carried out key
informant interviews with beneficiaries
and a broad range of experts who had
managed or implemented emergency as￾sistance programs. Information was col￾lected in both urban and rural areas; site
visits within each country also produced
valuable insights. Illustrative questions
asked during the interviews included the
following:
■ What was the political, economic,
and social context in which humanitarian
assistance was provided?
■ What were the perceived results of
the assistance in terms of saving lives,
affecting hostilities, and contributing to
development?
■ Were the results achieved those that
were intended (i.e., what was the relation￾ship between results and objectives)?
■ Were there unintended effects,
positive or negative?
■ Were the interventions sustainable
(in the case of rehabilitation assistance as
distinct from relief aid)?
■ What were the key strong and weak
points of the assistance; (i.e., the major
successes and failures)?
The very nature of complex emergen￾cies imposed certain methodological limi￾tations on the study. For example, there
was little evidence of long-term develop￾ment-oriented effects, since achieving
short-term effects was the principal objec￾tive of the humanitarian assistance. Many
of the people involved (beneficiaries, man￾agers, implementers) had moved on and
could not be reached, and some of the
institutional mechanisms had been dis￾mantled. Because action came first, paper￾work was frequently not given priority, so
relevant data were lacking or conflicting.
Finally, the contribution of the United
States to operations that were cofinanced
could not be separated from the contribu￾tions of other donors.

Country Context
HE TERM “complex humanitarian
emergency” is relatively new in the
American lexicon. Complex emergencies
are generally characterized by a break￾down of institutions and governance. They
always involve conflict, often war. What
causes them, and what humanitarian tools
has USAID used to alleviate widespread
suffering? This section examines these
questions in the context of the three coun￾try case studies: Haiti, Mozambique, and
Rwanda.
Causes of Complex
Emergencies
Haiti. Most studies have singled out
predatory governance as the principal
cause of Haiti’s complex emergency; eth￾nic and ideological factors appear less im￾portant. Haiti has almost no history of
democratic governance or strong public in￾stitutions. Instead, during its nearly 200-
year history, oppressive governments have
favored the rich at the expense of the poor.
The country’s military has controlled a
subservient police, and both institutions
have engaged in widespread and system￾atic human rights abuses with nearly com￾plete impunity.
The situation boiled over during
1991–94, shortly after a military coup re￾moved democratically elected President
Jean–Bertrand Aristide from office. Al￾though never an all-out civil war, this pe￾riod bore all the hallmarks of a complex
emergency with political, social, and eco￾nomic collapse. Human rights violations
swelled to unprecedented levels, prompt￾ing a series of UN-backed sanctions in￾cluding the U.S.–led international em￾bargo. By September 1994, an estimated
300,000 of Haiti’s 7 million people were
displaced internally; another 60,000 to
70,000 were refugees, some as the highly
publicized boat people (World Bank
1998b); thousands had fled across the bor￾der to the Dominican Republic; and 4,000
had been killed (Dupuy 1997). Gross do￾mestic product fell by 35 percent during
this period and inflation increased to 50
percent by 1994 (Buttari 1997). An esti￾T
2
10 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
mated 143,000 jobs were lost in the private
sector (Maguire 1996). The inflation-ad￾justed value of the minimum wage was
less than it had been 10 years earlier, and
per capita GNP was $250.
Mozambique. In 1975, after a 10-year
war for independence, Mozambique in￾herited a highly dualistic colonial
economy that lacked schools, health facili￾ties, and other public services. With the
end of colonial rule, most of the Portu￾guese and many skilled Mozambicans
fled, leaving the country without the tech￾nical skills needed to operate factories or
the transport system, to manage commerce
or government, or to provide professional
services. The Front for the Liberation of
Mozambique (Frelimo), the insurgent
group that had fought for independence,
took control after the Portuguese left. The
Frelimo government established a one￾party state and a centrally planned
economy modeled after those in Eastern
Europe.
Mozambique’s neighbors—white￾ruled Rhodesia and South Africa, which
supported apartheid—were alarmed by a
black-ruled, antiapartheid socialist coun￾try on their borders ready to export revo￾lution. In 1976, white Rhodesian military
officers opposed to the Marxist-leaning
Mozambique government formed the
Mozambican National Resistance
(Renamo). Renamo guerrillas sought to
disrupt Mozambique’s economy in an ef￾fort to keep the new government from sup￾porting guerrillas who were trying to over￾throw the white Rhodesian government.
They cut railway and power lines, de￾stroyed roads and bridges, and sabotaged
oil-storage depots. They raided towns and
villages and sometimes engaged in the
wholesale massacre of civilians. Mozam￾bique’s socialist allies countered by pro￾viding the Frelimo government with
weapons and financial support.
The result was civil war—nominally
based on ideology but actually supported
by foreign countries in the context of Cold
War politics and fueled by a drive for
power by local military and political lead￾ers. It was mainly a low-intensity, hit-and￾run guerrilla war fought largely with small
arms and land mines to destroy economic
and social infrastructure. Both armies ter￾rorized the rural population by seizing
food and killing people.
Over 2 million people fled to neigh￾boring countries and 4 to 6 million moved
to areas of relative safety within
Mozambique. As many as 8 million people
in a country of 16 million were affected—
a reflection of the large-scale human suf￾fering and economic dislocations that took
place. In 1986 the economy hit bottom: per
capita GNP was $80, the lowest in the
world; real GNP growth was a negative 2.3
percent; inflation was 41 percent. In 1992,
after 16 years of fighting, General Peace
Accords were signed. The country’s first
democratic, multiparty elections were held
two years later. Why did the war end?
Among the more important reasons was
that foreign military support ended. As￾sistance to Frelimo dropped sharply with
the collapse of the Soviet Union, and
Country Context 11
Renamo lost its external support when
apartheid ended in South Africa. In addi￾tion, foreign (mainly Italian) intermediar￾ies helped to bring Frelimo and Renamo
to the negotiating table and to facilitate
their reaching a settlement.
Rwanda. Rwanda’s wholesale geno￾cide of 1994 was a desperate attempt by
the government and Hutu extremists to
prevent the Tutsi-dominated Rwandan
Patriotic Front from seizing power. Based
in Uganda, the RPF had already tried to
topple President Juvénal Habyarimana
and his Hutu-dominated government in
1990. That unsuccessful effort set the stage
for a second attempt in 1993, which ended
with a cease-fire and later a peace agree￾ment, the Arusha Accords. However, it
soon became clear that key elements of the
accords, including access to land and po￾litical power sharing for the Tutsi living
in both Rwanda and Uganda, would not
be honored. This reignited the military
campaign, and by July 1994 the Rwandan
Patriotic Front had defeated the army of
the government of Rwanda.
In a 100-day period during April–July
1994, more than 800,000 people were mas￾sacred in a genocide historically un￾matched in its intensity. The killing, or eth￾nic cleansing, eliminated close to three
fourths of the Tutsi population of Rwanda.
The international community (including
the United States) ignored, then acknowl￾edged, the genocide, but did little to pre￾vent it. The U.S. secretary of state apolo￾gized for this failure to act in December
1997, as did the president in March 1998.
Rwanda remained in a state of ten￾sion and instability throughout 1995 and
1996, as génocidaires came from neighbor￾ing countries, particularly from Idjwi Is￾land in Lake Kivu in the Democratic Re￾public of the Congo (formerly Zaire). By
1999 the Rwandan Patriotic Front had se￾cured its borders and established security
throughout most of the territory within
them. (Some killing continues, primarily
in the northwest.) The bulk of the popula￾tion in exile or refugee camps has returned
to Rwanda. However, there remains a
small group of Hutu-power extremists, de￾termined to overthrow the current govern￾ment and finish their work of genocide.
Many of these génocidaires have found
sanctuary in the Democratic Republic of
the Congo, the Central African Republic,
and Angola.
Scholars have documented the exist￾ence of a culture of political impunity in
Rwanda—including ethnically based mass
killing that is sanctioned or planned by
government (Uvin 1998, Prunier 1995, Af￾rica Rights 1998). Violence has been part
of a winner-take-all mentality that has
dominated Rwanda’s governments during
the colonial and postcolonial periods. Both
Hutu and Tutsi have used violence to ob￾tain, and then maintain, absolute control
over political and economic decisions.
This mentality has benefited a tiny
elite, exclusively and handsomely, at the
expense of the poor majority. During 1990–
93, the proportion of Rwanda’s population
living in poverty increased from 40 per￾cent to 70 percent. During 1994–98, eco-
12 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
nomic activity declined sharply. As a re￾sult, 75 percent of rural households (or 90
percent of all households) currently live
below the poverty line, compared with 53
percent five years earlier. Structural adjust￾ment programs had not pulled Rwanda
out of its economic crisis.
But this is not surprising, since most
elements of these programs were not
implemented. That prompted the World
Bank and the International Monetary Fund
to halt disbursements. According to the
World Bank, “rising poverty undoubtedly
played some role in exacerbating social
tensions leading up to the genocide”
(World Bank 1998, i). But as Uvin states,
“structural adjustment did not cause these
[economic] problems; rather, it was irrel￾evant to their resolution” (Uvin 1998, 59).
USAID’s Humanitarian
Response
Haiti. The need for humanitarian as￾sistance was sharpened by the economic
embargo imposed by the international
community in response to the 1991 coup
and the military’s subsequent political re￾pression. USAID responded with an ex￾panded program that included food aid,
potable water, and health and sanitation
assistance.
USAID increased Haiti’s PL 480, Title II
program by 60 percent, from $15.4 mil￾lion in 1993 to $24.6 million in 1994. In
1995 it was increased by another 37
percent to $33.6 million (USAID/Haiti
1992–96). At its peak, the international
community was feeding 1.3 million
people—one of seven Haitians—each
day at 3,100 distribution points through￾out the country. It was also providing
most of the country’s health services
(USAID 1995).
Private voluntary organizations
implemented the humanitarian response
in Haiti. The Adventist Development and
Relief Agency operated 1,100 feeding cen￾ters in poor urban neighborhoods in Port￾au-Prince and in northern and central
Haiti. CARE worked in the northwest and
Artibonite regions through 1,200 school
feeding centers as well as hospitals, clin￾ics, and other distribution centers. Catho￾lic Relief Services operated 800 feeding
centers in the Port-au-Prince area and in
the south and southwest. International
Lifeline implemented food aid programs
for two years, and the UN World Food
Program, working through local and Eu￾ropean nongovernmental organizations,
provided food to vulnerable children,
pregnant and lactating mothers, and the
destitute.
USAID’s Office of Foreign Disaster As￾sistance awarded seven grants to private
voluntary organizations totaling $5.4 mil￾lion. Catholic Relief Services, the princi￾pal grantee, distributed essential drugs,
medical supplies, and agricultural inputs
(tools, seed, and fertilizer) and contributed
to UNICEF’s oral rehydration therapy and
measles immunization programs. The Of￾fice of Foreign Disaster Assistance also
supported efforts to purify drinking wa￾ter throughout Haiti, to purchase equip-
Country Context 13
ment for the Port-au-Prince municipal wa￾ter system, and to buy fuel needed to trans￾port emergency assistance to beneficiaries.
The Pan American Development Founda￾tion played a pivotal role in the humani￾tarian response by implementing a $38
million jobs creation project.
Finally, the Office of Transition Initia￾tives provided $17.3 million in 1994–95 to
support Haiti’s transition to democratic
governance. It funded the demobilization
of the armed forces (not covered in the case
study) and over 1,900 microprojects de￾signed to bridge the gap between relief
and development.
Mozambique. In the mid-1980s, USAID
assistance to Mozambique was under $50
million a year, with emergency aid a small
proportion of total aid. During 1988–91,
as the civil war and humanitarian suffer￾ing escalated, USAID assistance doubled to
an average of $100 million a year. In 1989,
an estimated one third of the population
depended on food aid for 60 to 70 percent
of their food needs. In 1992, in the final
throes of the war, total U.S. aid doubled
again to $200 million annually. During
1993–95, after the 1992 peace accords, U.S.
assistance averaged $125 million a year, of
which emergency assistance was a large
part.
USAID’s relief-to-development pro￾gram in Mozambique included several key
components: resettlement packages (food,
seed, farming tools, household goods); re￾building rural transport infrastructure;
support for elections and civic education;
demobilization of the two armies; and
mine clearance. The number of internally
displaced emergency food aid beneficia￾ries was reduced from 1.5 million in 1993
to only 600,000 in 1995. During 1996–97,
USAID assistance dropped to about $50 mil￾lion a year, and the mission resumed its
emphasis on development.
Rwanda. USAID provided food and
other types of emergency assistance to
Rwanda. According to USAID/Kigali, the
value of the assistance was almost $118
million in 1997 and over $56 million in
1998. Most of it was food commodities.
Beans, cornmeal, and vegetable oil were
provided to genocide survivors, including
widows, orphans, unaccompanied minors,
and refugee-returnees. Most of the assis￾tance was channeled through the World
Food Program, Catholic Relief Services,
the International Committee of the Red
Cross, and World Vision Relief and Devel￾opment. The most vulnerable (children
under 5 and pregnant and lactating moth￾ers) were reached through wet feeding
programs in nutrition centers, inpatient
feeding, and feeding programs in centers
for unaccompanied children and orphans.
The Office of Foreign Disaster Assis￾tance simultaneously provided potable
water, sanitation, and health services (as
well as emergency food aid) along the
route of returning refugees. Part of this
assistance targeted orphans and unaccom￾panied minors. For example, a grant to the
International Rescue Committee helped
establish transit camps for such children.
14 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
OFDA also assisted over 50,000 vulner￾able farm families by providing seed,
tools, and food rations for three months
through the World Food Program. The in￾tent was to jump-start agricultural produc￾tion. The Rwanda Emergency Seeds and
Tools project, also OFDA funded, helped
90,000 families for one month following
repatriation. Another agricultural project,
implemented by Food for the Hungry, In￾ternational, distributed seed packages to
25,000 vulnerable farmers. The project en￾couraged farmers to move onto rehabili￾tated marshlands, trained farmers in new
practices, and rehabilitated rural infra￾structure. The Seeds of Hope project sup￾ported agricultural experts who identified
appropriate seed stock, which then was
multiplied.
Finally, OFDA provided $26 million to
fund rapid-impact activities. These in￾cluded a shelter program to help meet the
needs of some of the 1.3 million returning
refugees and to preempt a potentially un￾stable security situation in the northwest.
At the same time, the Office of Transition
Initiatives funded the Women in Transi￾tion program (reaching over 162,000
women) and various activities to educate
local leadership and support local demo￾cratic processes.
Results:
Humanitarian Effects
AVING LIVES and alleviating suffering
are key objectives of humanitarian as￾sistance. When judged in terms of these
criteria, most evaluations of humanitarian
assistance tell a “mission accomplished”
story. For example, Apthorpe’s review of
six evaluations of humanitarian assistance
in Somalia, the Horn of Africa, Rwanda,
Liberia–Sierra Leone, and Sudan con￾cludes as follows:
. . . despite the horrendous and horren￾dously difficult circumstances, what we
read in these consultancies on the whole
is that the humanitarian aid does actu￾ally get through. Against all the odds,
the job of getting it there is actually
done, if not always at the times sched￾uled or as suitably composed as
planned. (Apthorpe 1997, 101–2.)
However, this message of overall suc￾cess is highly qualified. Apthorpe writes:
All accounts appear to find that, shall
we say, making our own brave leap into
the quantitative blue yonder, normally
more than probably 90 percent of as￾sistance has not failed to get through
and be duly delivered with, say, usually
very much more than probably 60 per￾cent of this being duly distributed, if not
to the intended beneficiaries, then at
least to their representatives. (Apthorpe
1997, 97.)
What about the complex humanitar￾ian emergencies in Haiti, Mozambique,
and Rwanda?
Haiti
Emergency indicators. One indicator of
the magnitude of Haiti’s complex emer￾gency is the national food supply. Haiti
historically has had a structural food defi￾cit that makes the country dependent on
imports. The deficit increased during the
crisis years, 1992–94. Domestic production
was reported at 90 to 94 percent of nor￾mal, while commercial imports decreased
by one third. Food aid increased by an
average of 29 percent during 1993–95, but
this was insufficient to compensate for
decreased domestic production and com￾mercial imports. As a result, Haiti’s food
deficit increased to an estimated 20 per￾cent of national food requirements, nearly
S
3
16 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
three times the deficit in a normal year
(World Bank 1998a, WFP 1998).
Another indicator is food prices. Food
prices in Haiti rose sharply as commodi￾ties became scarce. In Port-au-Prince, rice
prices increased by 126 percent during
1991–94; bean prices, by 167 percent; and
corn prices, by 184 percent (USAID/Haiti
1992–94). Although average food prices
decreased during 1994–96, they were still
more than twice their 1991 levels.
Malnutrition rates of children under
5 is a third key indicator of the severity of
a complex emergency. Most studies agree
that Haiti’s historically high malnutrition
rates increased in 1991–94 owing to the
combination of economic stress and a to￾tal breakdown of the public health sector
(IDB 1994, Ianotti 1997, World Bank 1998a).
Data from OFDA’s monitoring reports also
show the trend of increased malnutrition
but indicate the changes may not have
been significant. Nationally, nutritional
status (based on weight-for-age) of 50 per￾cent of Haitian children was normal in
1992. This figure increased to 52 percent
in 1993 but then declined to 49 percent in
1994 and dropped still further to 47 per￾cent in 1995. It then rose to 51 percent in
1996 (USAID/Haiti 1992–96).
Of course, these national rates mask
regional differences. For example, severe
malnutrition at the national level increased
from 3.3 percent (1992) to 3.9 percent (1994)
to 4.1 percent (1996). In the northwest,
though, it increased from 11.7 percent
(1992) to 14.4 percent (1994), then de￾creased to 10.8 percent (1996) (USAID/Haiti
1992–96).
Humanitarian response. International
donors increased food aid deliveries to
Haiti by a third, primarily to address mal￾nutrition. The United States contributed an
average of 68 percent of total food aid
(World Bank 1998a). The United States also
initiated programs to provide short-term
employment, agricultural inputs, fuel, and
medicine. Three major U.S. nongovern￾mental organizations (CARE, Catholic Re￾lief Services, and Adventist Development
and Relief Agency, as previously noted)
implemented most of the emergency as￾sistance efforts funded by the United
States. According to the NGOs’ figures, ben￾eficiary levels nearly doubled during the
emergency. In 1995, food aid was reach￾ing 1.3 million direct beneficiaries, or 16
percent of Haiti’s population (World Bank
1998a).
Efforts to target Haiti’s vulnerable
populations generally worked well, but
the problems of looting and armed theft
were always present. The Adventist De￾velopment and Relief Agency, for example,
distributed dry rations for only six months
in the low-income neighborhood of Cité
Jasmine in Port-au-Prince in 1992 because
of violence. Catholic Relief Services also
had difficulties in urban areas. Overall,
though, the estimated amount of leakage
was 5 to 10 percent, regarded as normal.
The Jobs Creation project was imple￾mented during 1993–96 primarily to off￾set the embargo’s economic pressures. It
Results: Humanitarian Effects 17
created almost half a million person￾months of short-term employment during
its 34 months of operation, of which 20
percent was for women (Brown 1996).
Another USAID-funded activity supported
agricultural production and reduced
decapitalization of farm households. This
project loaned funds to farmers for seed
and fertilizer and sold them tools at half
price. Approximately 13,000 farming
households and 47 farmers’ associations
participated (Naval 1995). USAID’s Office
of Foreign Disaster Assistance funded the
purchase of fuel needed to deliver emer￾gency medical supplies, potable water,
and food to more than 400 health centers
throughout the country.
Most studies agree that the embargo
seriously exacerbated Haiti’s historically
high malnutrition rates. But they also con￾clude that emergency food relief alleviated
that effect. And unlike the six evaluations
reported by Apthorpe, there is quantita￾tive evidence to support this conclusion.
The data from Haiti are approximate be￾cause the sample populations and meth￾ods of data collection were not standard￾ized during the emergency. Nevertheless,
they provide a basis for making an in￾formed judgment about the humanitarian
effects of the emergency assistance. This
was not the case in Mozambique.
Mozambique
The United States was Mozambique’s
major donor during its complex humani￾tarian emergency, contributing a total of
$636 million during 1987–95. Of this, $529
million was food aid, which accounted for
60 percent of total food aid provided dur￾ing this period.
Indicators. The Mozambique evalua￾tion team found no valid quantitative data
to assess objectively the impact of U.S.
emergency assistance. Although estimated
rates of malnutrition, mortality, and mor￾bidity at the national level showed some
improvement, first in the late 1980s and
then again in 1994 after the emergency, it
was impossible to attribute these improve￾ments to emergency assistance. NGOs re￾ported having little empirical basis for tar￾geting food aid because of the lack of sys￾tematic information about food insecurity
and nutritional status in rural areas. Bul￾letins issued by Médecins Sans Frontières
(Doctors Without Borders) contained
health and nutrition information, but they
were produced only after 1992. Moreover,
their data were based on small samples
and different methodologies and there￾fore, according to Médecins Sans
Frontières, “must be interpreted with cau￾tion.”
Notwithstanding the lack of accurate
quantitative data, there was consensus
among donors, relief workers, Mo￾zambican government officials, and
Mozambicans who received food aid that
the assistance aided people’s survival dur￾ing the emergency. All agreed that many
more people would have suffered and
died without food aid, although it was
impossible to estimate the number of lives
saved. Consistent with Apthorpe’s six-
18 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
evaluation review, “Mission accom￾plished.”
Effectiveness. However, the impact of
emergency assistance varied greatly de￾pending on (1) where the beneficiaries
sought refuge (within Mozambique or in
nearby countries such as Malawi), (2) who
delivered the assistance (the relief unit of
the Mozambican government or NGOs),
and (3) when people were uprooted and
received assistance (before 1987, during
1987–92, or during 1992–95).
Mozambicans who fled to nearby
countries had a hazardous journey but
generally received adequate food and
medical care once they reached the refu￾gee camps. They were also relatively se￾cure from Renamo or Frelimo harassment.
By contrast, those who were internally dis￾placed within Mozambique received less
adequate relief food that was supplied ir￾regularly. And they were often threatened
and harmed by Renamo or Frelimo sol￾diers.
The internally displaced who re￾ceived aid directly from NGOs or the World
Food Program reportedly received more
adequate and regular supplies than those
aided by the Mozambican government.
According to former internally displaced
beneficiaries, the government did not pro￾vide enough emergency food aid, and usu￾ally there was a two- to four-month time
lapse between distributions. They also re￾ported that the government’s distributions
were unfair, often influenced by the recipi￾ents’ political affiliation and social status.
Former internally displaced persons re￾ported that everybody was hungry dur￾ing the war years, so those who received
less food preyed on those who received
more, and when quantities were insuffi￾cient, people in some areas knifed open
sacks and fought for a share.
Access to assistance by the internally
displaced also varied depending on
whether they were in Frelimo- or Renamo￾controlled areas. U.S. policy was to pro￾vide emergency assistance only to govern￾ment- (Frelimo-) controlled areas—except
for limited quantities provided to Renamo
territory through the International Com￾mittee of the Red Cross. Regardless of its
source, food aid attracted both Frelimo and
Renamo soldiers. People from several vil￾lages said they lost their food aid to
Frelimo by day and to Renamo by night.
Finally, the effectiveness of emergency
assistance varied over time. Before 1987,
the war was disruptive, but few people
had to flee their homes. Relief efforts were
relatively small and localized. The war
intensified during 1987–92. People fled to
the relative safety of neighboring countries
or the Beira corridor. (The corridor is an
east–west swath across the country’s waist;
it was guarded by Zimbabwean troops
and thus served as a safe haven for
Mozambican civilians.) This was an espe￾cially difficult period for internally dis￾placed persons. After the 1992 peace ac￾cords were signed, people began return￾ing home. Resettlement packages (food,
Results: Humanitarian Effects 19
tools, and services) were provided by
NGOs through 1995. This final stage of the
emergency was the smoothest.
Rwanda
In late 1996 and early 1997, some 1.3
million refugees were repatriated to
Rwanda from the border camps, either
voluntarily or by force. Massive starvation
and human suffering would have occurred
without substantial infusions of predomi￾nately U.S. emergency food aid. As in
Mozambique, though, this is difficult to
quantify because data were not systemati￾cally collected.
Monitoring. Following repatriation of
refugees, the World Food Program carried
out a six-month general-distribution food
program. This was intended to last only
until the harvest in June 1997 but was re￾instated in November 1997 in five prefec￾tures. In those areas local food prices had
increased by a factor of three, signaling a
significant food shortage. Starvation and
death were on a sharp incline, and emer￾gency food aid continued to be provided
until the next harvest in June 1998.
Food insecurity was especially acute
in Ruhengeri Prefecture owing to politi￾cal instability as well as poor harvests and
high prices. As many as 573,000 people of
an estimated population of 869,000 were
displaced, living in camps and awaiting
resettlement. The Ministry of Health did
a nutritional survey in January 1999 based
on a sample of 900 children under 5 living
in camps in Ruhengeri. It found what Save
the Children Foundation/UK described as
“alarmingly high rates of malnutrition, in
particular severe malnutrition among chil￾dren 6–59 months.” Specifically, it found
8.0 percent of children with acute malnu￾trition, 4.7 percent with edema, 40.6 per￾cent underweight, and 59.9 percent with
chronic malnutrition. These 1999 figures
were substantially higher than those re￾flected in a 1996 National Nutrition Sur￾vey. Thus, the situation in Ruhengeri, at
least, was not getting any better—despite
emergency food assistance.
Targeting. Targeting assistance to the
intended beneficiaries in pre-1996 Rwanda
was mixed. In the Bukavu area in south￾ern Zaire, government soldiers formed
separate camps from the very beginning.
By contrast, in the Goma area the army,
militia, and civilian refugees were all
mixed together (forming “refugee–warrior
camps”), and the military and former gov￾ernment leaders controlled relief distribu￾tion. In Tanzania, the military was not as
visible among the refugees, but the mili￾tia and former officials were. There was
little security in these camps, and food and
other relief supplies were diverted from
the intended beneficiaries.
It was painfully obvious that the per￾petrators of human rights abuses and
genocide were fed and assisted in the
camps. Médecins Sans Frontières believed
the only alternative was to leave the camps
and suspend most services. By contrast,
most other NGOs and the UN High Com￾missioner for Refugees decided to stay.
20 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
They recognized the humanitarian im￾perative to protect and assist the vast
population of refugees, even if that meant
assisting people guilty of crimes against
humanity (Joint Evaluation 1996, study 2,
58–9).
Poor monitoring (which did not be￾gin in earnest until 1998) also contributed
to ineffective targeting. U.S. government
officials reported considerable double￾counting of refugees by former military
and government leadership, particularly
in the cross-border camps. As a result,
many experts believe that more food aid
was supplied both inside and outside
Rwanda during this period than was nec￾essary—and that more food aid was mis￾appropriated in Rwanda than is usual in
emergency situations.
Interahamwe (Hutu militia respon￾sible for the genocide) and former soldiers
of the Rwandan army diverted food from
women and children for their own pur￾poses. There was also evidence that camp
rosters were sometimes not updated to
remove the names of those deceased,
which resulted in the accusation that the
UN High Commissioner for Refugees
“feeds dead people.” Although this is an￾ecdotal, there is little doubt that there were
abuses in the feeding program.
Bosnia–Herzegovina
The humanitarian assistance opera￾tion in the former Yugoslavia was one of
the largest initiatives ever undertaken by
the international community. Although
Bosnia–Herzegovina was not included as
one of the country case studies for this as￾sessment, the UN High Commissioner for
Refugees and the World Food Program
had completed a joint evaluation of the
assistance program in 1998 covering the
entire period from 1992 to June 1997
(UNHCR/WFP 1998, 1). The evaluation ex￾amined the effectiveness of targeting in re￾sponse to beneficiary needs, the impact of
the emergency operation on the war itself,
and the relevance of food aid in a period
of reconstruction—the same issues covered
in the Haiti, Mozambique, and Rwanda
evaluations.
According to the joint evaluation, an
average of 2.6 million people were reached
annually during 1992–96: 1.2 million in￾ternally displaced persons and 1.4 million
“war affected” (people who had no means
of support, although they were neither
refugees nor displaced). By September
1997, 1.14 million tons of food had been
provided at a cost of $710 million. An esti￾mated 80 percent of the population of
Bosnia–Herzegovina had been beneficia￾ries of food aid supplied by the World
Food Program and the UN High Commis￾sioner for Refugees at one time or another
(UNHCR/WFP 1998, 2).
The evaluation concluded that there
was no widespread hunger or malnutri￾tion in Bosnia–Herzegovina. However,
security-related problems hindered access
Results: Humanitarian Effects 21
and distribution to isolated communities
and cities under siege. Air transport was
used when access by land was denied, as
in Gorazde, Sarajevo, Srebenica, and Zepa.
The Sarajevo airlift was the longest run￾ning humanitarian air bridge in history,
lasting from 3 July 1992 until 9 January
1996. These operations were generally suc￾cessful. In fact, it was only in Bihac dur￾ing late 1994 and throughout 1995 that air￾drops did not succeed in averting hunger
(UNHCR/WFP 1998, 6).
Targeting
Efforts to target humanitarian aid to
intended beneficiaries often run into ma￾jor problems. Several ways to improve tar￾geting and the overall effectiveness of hu￾manitarian assistance programs involve
alternative distribution channels, com￾modity selection, and planning.
Commercial channels may offer an al￾ternative to other types of distribution. For
example, the World Food Program con￾tracted with Somali merchants in
Mombasa, Kenya, to transport commodi￾ties to targeted sites in Somalia. This in￾volved selling the commodities to the mer￾chants and then buying them back with a
10 percent profit margin. WFP paid a 10
percent markup to avoid having 60 per￾cent of the food looted. This was because
bags printed GIFT OF THE USA were more
likely to be looted than commodities
stored and transported by businessmen.
A variation of this mechanism was to
use the proceeds from the sale of the com￾modities to establish wage-generation pro￾grams rather than buy back the commodi￾ties. This gave people cash, creating a mar￾ket that the merchants then supplied. But
commercial channels are not always the
solution. In Mozambique, most private
transporters refused to transport food aid
because of bandits and land mines and
also because many that did had had their
trucks stolen.
Military involvement in complex
emergencies—both as protector and pro￾vider of commodities—has been a mixed
blessing. The tremendous costs of military
operations are generally disproportionate
to the value of the commodities protected.
Military humanitarianism can also get
wrapped up with geopolitics and foreign
policy objectives. In Liberia, for example,
the Nigerian-dominated regional military
force sent to Liberia for peacekeeping pur￾poses found itself in conflict with the larg￾est rebel force in the country (Prendergast
and Scott 1996).
Carefully selecting commodities for
emergency assistance can reduce looting
and improve targeting. For example, rice
and other high-value commodities are
typically much more attractive to looters
than sorghum, maize, or blended foods.
But this varies by region: substantial quan￾tities of maize were looted in Somalia.
Similarly, looters are rarely interested in
cooked food distributed in numerous
22 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
kitchens that are widely dispersed. In￾stead, they are generally interested in com￾modities for which there is market de￾mand and which they can turn into cash.
Diversifying entry points for emer￾gency supplies can help guard against
empowering a particular authority. In
Liberia, for example, all commodities
came into Monrovia rather than across the
border upcountry. By contrast, substantial
food assistance was delivered to affected
sites in Ethiopia (Eritrea and Tigray) via
the Sudanese border.
There is an important distinction be￾tween distributing emergency commodi￾ties to affected areas and to affected
populations within areas. Food aid was dis￾tributed to Rwandan refugees in camps
through prefectures, communes, and fi￾nally cells; but it was not targeted to indi￾vidual families. As a result, the emergency
aid perpetuated the authority of the mili￾tary and political leadership that had
planned the genocide.
Humanitarian aid is more easily
diverted when population figures are in￾flated. For example, to achieve purely po￾litical and economic objectives, warring
factions in Liberia and Rwanda overesti￾mated the need for food. But underestimat￾ing need may lead to violent competition
for food. In assessing need, it is important
to understand people’s coping strategies
and their desire to preserve their liveli￾hoods. This may be as important as en￾suring short-term hunger alleviation
(Borton and Macrea 1997, 27).
In planning emergency assistance
programs, experts need to be sensible and
beware of over-complexity. It is also
important to address capacity-building
questions early on. In anticipation of re￾habilitating a collapsed health service, for
example, capacity building might take the
form of training medical personnel. Ac￾cording to Anderson (1996), capacity
building should be a central part of any
emergency response.
Finally, Prendergast and Scott (1996)
point out it is important to plan up front
for monitoring and evaluation: “A com￾mitment to adequate, independent, and
continuous monitoring and evaluation of
programs may reduce aid’s contribution
to conflict.” In Rwanda, as a result of moni￾toring, diversion of food was reduced from
120 tons per month to 5 tons per month
between July 1993 and January 1994. “It’s
monotonous, boring, but critical in cutting
down mismanagement” (Prendergast and
Scott 1996).
Efficiency
In addition to targeting, another mea￾sure of the effectiveness of humanitarian
assistance is efficiency. Borton and Macrea
(1997) synthesized the results of a broadly
representative sample of 28 evaluations,
mostly of complex emergencies, under￾taken since 1991 by bilateral donors, UN
agencies, and the European Community
Humanitarian Office (1–2, 12). They exam￾ined humanitarian assistance in terms of
cost-effectiveness, which, unlike cost–ben-
Results: Humanitarian Effects 23
efit analysis, does not involve the valua￾tion of lives in economic terms. What they
found is instructive: very few of these
studies even considered the issue of cost￾effectiveness—partly for methodological
reasons and partly because of reservations
about whether it should be a criterion for
providing humanitarian aid. According to
Kleist (1994, 301–02), the reality is that
monetary costs are less important as an
evaluative criterion than the number of
lives saved and the security of the person￾nel delivering the humanitarian assistance.
Transportation (rail, road, air) is a
major cost of emergency operations that
varies enormously. For example, commer￾cial air transport within the Great Lakes
region was approximately 4 to 5 times
more expensive than road transport, and
10 to 20 times more expensive than rail
transport (Joint Evaluation 1996, study 3).
Therefore, on efficiency grounds it is bet￾ter to transport food to conflict areas by
road or rail rather than by plane. On the
other hand, using ground rather than air
transportation could involve negotiating
with rebel groups, thereby granting them
a degree of legitimacy they otherwise
would not enjoy (Hallam 1998, 21).
Cost also varied according to distri￾bution channel and type of commodity.
Military channels were estimated to be
four to eight times more expensive than
civilian channels (Borton and Macrea 1997,
2, 23). Moreover, replacing high-value rice
with alternative less expensive cereals in
coastal West Africa (bulgur wheat in Si￾erra Leone and Liberia, and maize meal
in Guinea and Côte d’Ivoire) improved
cost-effectiveness (Apthorpe 1996). As
noted above, using lower value commodi￾ties also reduced the likelihood of diver￾sion as well as the disincentive effect on
local production (Borton and Macrea
1997, 24).
Conclusion
The international community is gen￾erally unable to assess with any degree of
certainty the number of lives saved
through humanitarian assistance provided
in a complex emergency. The case of Haiti,
where data were available, is the excep￾tion; the case of Mozambique, which
lacked data, is the norm. Only 1 of the 28
evaluations reviewed by Borton and
Macrea (1997) attempted to estimate the
number of lives actually saved by inter￾national assistance interventions. This was
an analysis of the 1990–94 response to the
crisis in Somalia (Hansch 1994). It found
that 330,000 Somalis were at imminent risk
of death in 1992 and 1993. An estimated
110,000 of these were sustained (that is,
their deaths were averted) by health, food,
and other interventions. At least 70 per￾cent (154,000) of the famine-related deaths
that did occur in 1992 probably could have
been prevented had primary health strat￾egies been implemented earlier and more
widely (Hansch 1994, cited in Borton and
Macrea 1997, 2, 25).
However, another analysis of the ef￾fectiveness of humanitarian assistance in
Somalia reports results with far less quan-
24 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
titative precision. It concludes that “sig￾nificant numbers of lives were saved, se￾vere malnutrition and vulnerability to in￾fectious disease declined, the suffering of
the displaced was eased, and refugee
movement was slowed—though these effects
are all difficult to quantify” (Kleist 1994, 305;
emphasis added). According to Kleist, the
inability to measure results with any de￾gree of certainty reflects the fact that sav￾ing lives is of the utmost priority. Requir￾ing answers to detailed questions about
humanitarian assistance provided during
a complex emergency could cause delay
and cost human lives. Therefore, little up￾front planning or data collection is done.
In Somalia, for example, many proposals
from NGOs lacked such basic information
as who the target group was and where it
was located (Kleist 1994, 294–95). Notwith￾standing inadequate information, donors
typically allocate resources to relief orga￾nizations to meet urgent humanitarian
needs.
This raises a question about the de￾gree of planning and data collection that
is both feasible and desirable when an
immediate humanitarian response is
needed to save lives. USAID’s Bureau for
Humanitarian Response has determined
that one of its strategic objectives is to en￾sure that “critical food needs of targeted
groups are met.” Two indicators are speci￾fied to determine if the objective has been
achieved: (1) the percentage of target
populations reached by food aid programs
and (2) the impact of the assistance on the
nutritional status of beneficiaries (USAID
1998b, 4). However, data must be collected
for these two indicators, and data collec￾tion in the throes of a complex humanitar￾ian emergency takes time when time is of
the essence.
Results:
Political Effects and
Effects on Hostilities
UMANITARIANS TRADITIONALLY
have tried to remain impartial and
thus apolitical. There are at least two
good reasons for this. The first is pragmatic:
providing relief is often facilitated when
indigenous political actors perceive hu￾manitarian agents to be without political,
religious, cultural, or other agendas. Im￾partiality helps humanitarians gain access
to victims. The second reason to remain
impartial is principled: each society has the
exclusive prerogative and responsibility to
shape its own destiny. Outsiders should
not interfere except in nondisruptive ways
to save lives. By helping everyone and re￾fusing to take sides, humanitarians place
themselves above the fray (Pasic and Weiss
1997, 198–99).
Nevertheless, the increasingly obvi￾ous reality is that humanitarian relief and
its consequences are inevitably political,
often in ways that are not self-evident. That
is because aid does not just keep people
alive in a political vacuum but also affects
the local power structure and changes the
environment in which it is given. Aid is
rarely neutral.
Each of the six evaluations reviewed
by Apthorpe asked whether humanitarian
aid had had the perverse effect of prolong￾ing the war or contributing to the war
economy; that is, whether such aid in ef￾fect feeds conflict as well as its victims.
Each concluded that although food and
other humanitarian assistance are not
meant to feed conflict, they often do
(Apthorpe 1997, 95). In short, humanitar￾ian assistance can have adverse political
effects. Was this the case in Haiti,
Mozambique, or Rwanda?
Haiti
The large quantities of aid-financed
food injected into resource-starved Haiti
were conspicuous and highly valued—ei￾ther for direct consumption or as a politi￾cal tool for those who controlled their dis￾tribution. Control over access to food aid
became a new source of tension and
power, and violent elements—local gangs
or groups connected to political factions—
hijacked food supplies. Fighting some￾times erupted among beneficiaries when
food was dropped off in urban neighbor￾H
4
26 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
hoods for distribution by volunteers. Some
municipal mayors used their access to food
aid to favor supporters of one political fac￾tion or another or to promote their per￾sonal aggrandizement.
To their credit, the NGOs limited leak￾age and diversion (estimated at less than
10 percent) through regular monitoring,
convoy protection, and timely adjustment
of their stocking and distribution methods.
They also stopped distributions in some
neighborhoods in Port-au-Prince, such as
Cité Jasmin. Violence, political exploita￾tion, and local tensions consequently were
reduced to manageable if not minor pro￾portions.
Did the emergency assistance help the
de facto regime in Haiti withstand diplo￾matic pressures and the effects of the eco￾nomic embargo—before the international
community finally resorted to military
force? Many Haitians believe this is the
case. In their view, exempting humanitar￾ian aid from the embargo worked at cross￾purposes with the policy of economic
isolation. By reducing food distress the
emergency assistance dampened public
pressure that might otherwise have risen
to uncontrollable levels against the regime.
Thus, food aid may have permitted a de￾lay in the intervention by external forces,
intervention that finally proved unavoid￾able in order to eject the Haitian military
and return President Aristide to power.
This view implies that humanitarian
aid, by extending the duration and extent
of the emergency and the consequences for
its victims, ironically could have caused
more humanitarian distress than it allevi￾ated. The point, of course, is speculative,
and one cannot conclude with any degree
of certainty that humanitarian assistance
prolonged the conflict in Haiti.
Mozambique
The Reagan and Bush administrations
debated whether to support the Frelimo
government or the Renamo rebels when
Mozambique’s civil war broke out in the
mid-1980s. Many American conservatives
viewed the war as an ideological battle
over communism and believed that the
United States should therefore support
Renamo. This view was buttressed by the
fact that the Frelimo government had in￾stalled a socialist system and was receiv￾ing support from its socialist allies. Oth￾ers, however, believed the United States
should assist the Frelimo government in
recognition of its support of the antiapart￾heid movement in South Africa. In the fi￾nal analysis, the United States provided
limited humanitarian assistance to
Mozambique’s socialist government.
By the late 1980s, the Frelimo govern￾ment had abandoned most of its socialist
ideology and initiated a program of mar￾ket-based economic reforms supported by
USAID and the World Bank. In response,
the United States greatly expanded its hu￾manitarian assistance, much of it in the
form of food aid. This U.S. assistance was
provided on a government-to-government
basis, which meant it went only to inter￾nally displaced persons in Frelimo-con-
Results: Political Effects and Effects on Hostilities 27
trolled territories. The only U.S. assistance
provided to Renamo populations was sup￾plied indirectly through the International
Committee of the Red Cross.
What was the impact of emergency
food aid on the length of the civil war in
Mozambique? Expatriates and Mo￾zambicans alike consistently reported that
foreign political and military support for
Frelimo and Renamo—rather than hu￾manitarian assistance—was the primary
resource that fueled Mozambique’s 16-
year civil war. U.S. humanitarian assis￾tance had relatively little influence on the
course of the war when compared with the
military assistance provided by the Soviet
Union (to Frelimo) and by South Africa (to
Renamo).
Nevertheless, both Frelimo and
Renamo soldiers tried to steal food aid by
intimidating PVO workers and hijacking
trucks. Food aid distributions in rural vil￾lages were a magnet for looting by both
militaries. Thus, food aid helped support
the military forces to some extent, but the
effect was relatively small. As in Haiti,
emergency food aid in Mozambique was
sometimes politicized. Politicians at both
the national and local levels reportedly
used their influence over food aid distri￾butions to favor particular factions and
reinforce their political power.
Rwanda
The massive influx of Rwandans into
refugee camps in former Zaire was not
only a movement of people but also a
transplantation of a well-organized politi￾cal, social, and security structure. The
Hutu extremists (ex-Rwandan army regu￾lars, former government officials, and al￾lied militia) planned to use the refugee
camps as a staging area for their eventual
return to political power through
Rwanda’s northwest. They assumed, cor￾rectly, that the international humanitarian
relief agencies (and the national govern￾ment and regional authorities of Zaire)
would not separate them from bonafide
refugees. This meant they could consoli￾date both military and political control
over most of the camp population. In De￾cember 1994 a new government of
Rwanda in exile was declared, and incur￾sions from the camps into Rwanda began.
Mixing bonafide refugees with those
who were probably guilty of genocide and
other high crimes was seen by the UN
High Commissioner for Refugees, the In￾ternational Committee of the Red Cross,
and the international humanitarian com￾munity as a conundrum, difficult to re￾solve in the context of maintaining neu￾trality and without military intervention.
As a result, UNHCR did nothing. It was im￾portant to maintain stability in an inher￾ently unstable situation, and separating
the refugees from the Hutu extremists was
considered risky. Moreover, UNHCR did
not believe the new government of
Rwanda would welcome the refugees back
home, certainly not those involved in the
genocide. The Hutu refugees, themselves,
understandably feared retribution (“re￾verse genocide”) if they returned, whether
or not they were guilty of crimes and
28 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
atrocities committed in Rwanda. These
fears and concerns helped create a dead￾lock that lasted over two years.
The inability of the international com￾munity to resolve the deadlock gave the
Hutu extremists in the refugee camps a
false sense of enhanced legitimacy. They
used this opportunity to regroup, rearm,
and revitalize themselves with food ra￾tions intended for refugees—all in the rela￾tive safety of the camps. As indicated ear￾lier, once the camps became militarized,
targeting became problematic. Beneficiary
figures were significantly inflated. Food
aid was diverted to and consumed by the
ex–Rwandan army regulars and inter￾ahamwe militia resident in the camps.
That had the unintended effect of prolong￾ing the conflict. According to the U.S. am￾bassador to the United Nations, Bill
Richardson, “the failure of the interna￾tional community to respond adequately
to both genocide and the subsequent mix￾ing of genocidal killers with the legitimate
refugee population in the former eastern
Zaire only served to prolong the crisis”
(October 1996). It was not until 1996 that
USAID became sufficiently concerned and
ceased providing food aid to the World
Food Program for use in the camps. The
idea of providing humanitarian aid to the
planners and implementers of genocide
was seen as inconsistent with the stated
objectives of humanitarian aid.
The crisis is not yet over. Rwandan
ex-political and ex-military leadership is
using former Zaire as a staging ground to
destabilize and overthrow the present gov￾ernment of Rwanda. Their objective is to
complete the unfinished work of genocide,
using a campaign of propaganda and ter￾ror to destroy the political and economic
structures of the northwest (and beyond)
and to gain support of the local Hutu
population. In 1997, 30,000 to 40,000 sol￾diers began arriving in the northwest,
while several thousand remained in Zaire
to maintain the camps as a base of opera￾tions. In 1998, several commune offices in
the northwest were looted and burned,
and the officials were murdered or terror￾ized. By March 1999, many services had
ground to a halt. Water sources had been
destroyed and health problems multiplied.
The government of Rwanda requested
massive food assistance from the World
Food Program. In response, USAID is pro￾viding both development assistance and
humanitarian assistance to the northwest.
Bosnia–Herzegovina
The conflict in former Yugoslavia also
raises concerns about emergency aid en￾abling protagonists to prolong the conflict.
As early as 1994, some evidence, though
inconclusive, suggested that aid supported
troops, thereby releasing the authorities
from responsibilities they might have had
to civilians (Prendergast and Scott 1996,
11, citing Minear 1994a). This earlier evi￾dence was examined by the recent UN
High Commissioner for Refugees/World
Food Program joint evaluation of the
Bosnia experience (UNHCR/WFP 1998).
Results: Political Effects and Effects on Hostilities 29
This joint evaluation offers two pos￾sible arguments supporting the thesis that
humanitarian aid to Bosnia helped pro￾long the conflict (UNHCR/WFP 1998, 3).
Both arguments continue to be a source of
controversy. The first suggests that by giv￾ing generous support in the form of food
aid, donors were able to defend them￾selves against the charge of inaction. But
doing so in effect postponed the military
intervention that ultimately was needed
to end the conflict. This reasoning paral￾lels that of the Haiti evaluation, and as
noted there, is purely speculative.
The second argument suggests that
humanitarian assistance prolonged the
war in Bosnia because it was diverted to
the combatants and thus supported their
military efforts. Even if humanitarian sup￾plies were not diverted to combatants, the
aid still would have allowed resources oth￾erwise needed to sustain the noncomba￾tant population to be used instead to sup￾port the war effort. (Of course, combatant
organizations in some complex emergen￾cies have shown little concern over the con￾dition of “their” noncombatants.)
It is possible that without food aid the
consequent civilian suffering might have
hastened the cessation of hostilities in
Bosnia. However, while the conflict might
have been shorter, the suffering would
likely have been greater. Moreover,
UNHCR/WFP suggests that the outcome
would likely have been unsatisfactory: a
world without Bosnia, with the country
instead divided among its neighbors. Ac￾cording to UNHCR/WFP (1998, 7), there is
no convincing evidence that the increased
suffering that would likely have occurred
in the absence of humanitarian support
would have been justified by an increased
prospect for a swifter and satisfactory out￾come of the conflict.
The provision of aid did have some
unavoidable negative political effects.
First, since the authorities on the ground
controlled distribution of the assistance,
bargaining with them and agreeing to use
the channels they controlled inevitably
reinforced their authority. Anderson (1996,
3–4, 16–17) suggests that this is not unusual
and that aid agencies often must negoti￾ate with army leaders to gain access to
civilian populations or to hire armed
guards to protect the goods they bring.
Second, in some cases military authorities
levied food taxes to allow convoys to pass
and seized food when they were not
paid, sometimes at gunpoint (UNHCR/WFP
1998, 7).
Policy Implications
This assessment of the political effects
of humanitarian assistance raises key ques￾tions: Should aid be given if some of it is
being diverted to armed participants in the
conflict? Should aid be distributed through
local structures if these are considered
predatory or biased? USAID clearly would
not operate a development assistance pro￾gram if security deteriorated to the degree
30 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
it did in Somalia, for example. Yet the
United States can nearly always be
counted on to provide emergency assis￾tance, even under such volatile conditions.
Prendergast and Scott (1996) are sen￾sitive to the possibility that humanitarian
aid designed to relieve suffering and pro￾mote peace often, inadvertently, fuels, sus￾tains, or exacerbates such conflicts by mak￾ing more resources available to warring
parties. They like others recognize that
humanitarian aid may be given without a
political agenda, but it rarely escapes hav￾ing political consequences. Moreover, it
can be deliberately manipulated to serve
as an instrument of war by providing a
means for sustaining the conflict. This can
occur by (1) manipulating access to the aid,
(2) manipulating population movements,
and (3) diverting or looting the aid. Ex￾amples:
1. Manipulating access. Warring parties
often manipulate humanitarian aid to en￾hance their power over civilian popula￾tions or weaken their opponents by deny￾ing them food. Throughout the 1980s the
Ethiopian and Sudanese governments lim￾ited the amount of aid going to rebel-held
areas. Warring factions also have manipu￾lated access to aid in Bosnia, Liberia,
Mozambique, Rwanda, and Somalia.
Relief agencies generally abide by the
principles of neutrality and impartiality.
They don’t take sides, and instead give
both sides equal access to relief aid. But
providing aid in rebel areas necessarily
helps legitimize the rebel organizations,
because it enables them to feed the popu￾lations they seek to control. The same is
true of the government and the agencies
they mandate to distribute relief. As a re￾sult, organizations (often NGOs) that at￾tempt to provide humanitarian aid neu￾trally may support forces that carry out
violence against civilian populations, as in
Mozambique. But the alternative may be
no better: to subject civilian populations
to the double punishment of violence and
hunger.
2. Manipulating population movements.
Warring factions have used civilians as
shields or “vehicles” to obtain food and
other types of humanitarian aid. They po￾sition civilians near airstrips to enhance the
ability of their troops or militia to remain
in areas they otherwise would abandon for
lack of supplies or difficulty in defending.
For example, aid supplied to the refugee
camps in Zaire helped maintain the former
Rwandan government’s control over a
population that otherwise might have dis￾persed or returned home.
3. Diverting or looting aid. Warring fac￾tions tax, steal, or divert humanitarian as￾sistance—especially food and drugs, given
their easy monetization—for their own
consumption, for barter, or for sale. This
is a principal means to buy arms. Liberian
rebels looted relief supplies and stole re￾sources, especially vehicles and fuel to use
for hit-and-run campaigns. In Somalia,
food distribution operations were looted
so often that as little as 12 percent of inter￾national food aid destined for refugees in
1986 reached the intended recipients (GAO
Results: Political Effects and Effects on Hostilities 31
1986). Much of the remainder was taken
by the Somali army and associated mili￾tia. Agencies had to negotiate for their own
security because there was no national se￾curity force, and this led to diversion of
goods, especially food (Kleist 1994, 298).
Similar occurrences took place during the
famine in northern Ethiopia in the 1980s
and the wars in Mozambique and south￾ern Sudan (de Waal 1993).
Given this negative and widespread
experience, under what conditions, if any,
should humanitarian aid be halted?
Prendergast and Scott suggest three fac￾tors to consider in deciding whether to
cease providing humanitarian assistance:
lack of progress in peace negotiations; sup￾port of undesirable political factions and
human rights abusers; and danger to aid
personnel.
During the civil war in Ethiopia in the
late 1980s, Lutheran World Relief urged
agencies seriously to consider withhold￾ing aid if peace efforts failed (Prendergast
and Scott 1996, 44). Should we pay end￾less millions for humanitarian aid, they
asked, when this may only exacerbate and
prolong the conflict? Perhaps we should
disavow the principle that food should not
be used as a political weapon and instead
use it to force peace negotiations. How￾ever, others have argued that political de￾cisions should be separate from the basic
human right to humanitarian aid, and that
relief should not be used as a political
weapon. The rationale underlying the lat￾ter point of view is summarized in
Prendergast and Scott (1996, 44) in these
terms: “The people who don’t care and are
not affected are the rulers.”
Food and medical aid are particularly
valuable to combatants. But aid agencies
rarely withhold such assistance despite
human rights abuses and looting, and
most will not completely withdraw from
an area unless the emergency is completely
over or the security situation is untenable.
In Liberia during 1990–93, aid agencies
were subjected to harassment and robbery
by warring factions. Diversion rates were
close to 50 percent, according to some re￾ports, and debate raged within the aid
community about how much aid was too
much. All agencies eventually withdrew
operations until security improved. How￾ever, it remains unclear whether the with￾drawal contributed to the 1995 peace
agreement. And even if it did, was it at
the cost of more human suffering? There
is simply no way to measure the effect of
stopping the aid.
De Waal points out that war has be￾come synonymous with famine in much
of Africa. In fact, war is often designed to
create famine. War, according to de Waal,
has received far less attention than it de￾serves—compared with other contributory
causes of famine such as drought, environ￾mental degradation, and inappropriate de￾velopment strategies. This brings into
question the tendency for donor govern￾ments to fund humanitarian assistance but
not to address the underlying causes of
war (Hallam 1998, 5). Humanitarian agen￾cies are sometimes needed less than po￾litical or military actors. In Rwanda in
32 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
1994, for example, a well-armed UN
peacekeeping force may have been able to
prevent or mitigate the genocide. The
multitude of NGOs responding to humani￾tarian needs once the genocide was over
was much less effective. The issue of con￾flict prevention, as distinct from cure, is
briefly introduced in annex C.
Though war may be a principal cause
of famine (as de Waal suggests), Sen (1993)
provides a compelling argument that de￾mocracy and a free press are great forces
in preventing famine. Sen points out that
a government cannot ignore famine con￾ditions if (1) it has to face reelection, (2) it
cannot censor the terrible facts of starva￾tion, disease, and death that accompany
famines, and (3) it has to face criticism
from opposition parties and newspapers.
“It is not surprising that even though fam￾ines have happened in colonial economies
and in modern authoritarian states, never
has a famine occurred in a democratic
country with a relatively free press” (Sen
1993, 88).
Conclusion
 While no aid is apolitical, humani￾tarian assistance provided during complex
emergencies can result in substantial and
unpredictable political effects, since it is
provided in the context of conflict (Hallam
1998, 12–13). The political effects of
humanitarian assistance in Haiti, Mo￾zambique, and Rwanda were mixed.
According to most accounts, the assistance
clearly prolonged the conflict in Rwanda.
In Mozambique, external military assis￾tance, rather than humanitarian assistance,
fueled the civil war for more than a de￾cade. In Haiti, evidence of the political ef￾fects of the assistance is inconclusive. In
Bosnia, airlifting food to Sarajevo probably
prolonged the war, but this does not mean
that the increased suffering that would
have occurred by withholding food aid
would have been justified by the possibil￾ity of a shorter war.
Perhaps what is most important to
keep in mind is the underlying principle
that, at the very minimum, aid that is in￾tended to help victims in war settings
should not cause additional harm (Ander￾son 1996, 6). The challenge, therefore, is to
specify, on a case-by-case basis, clear ob￾jectives and to monitor closely the extent
to which the humanitarian assistance is
achieving those objectives. Allowing flex￾ible implementation and encouraging ef￾fective communication and coordination
among other involved authorities (diplo￾matic, military) is equally important.
Results:
Economic Effects
HE RELATIONSHIP between short-term
relief and long-term development has
been viewed as a continuum in which re￾lief operations, in response to a humani￾tarian crisis, are followed by rehabilitation
and then development activities (USAID
1998a, 18–19). More recent literature ques￾tions the utility of this concept of a relief￾to-development continuum. According to
the Department of State’s Bureau for Popu￾lation, Refugees, and Migration, relief and
development assistance have significantly
different aims, and implementers of each
type of assistance should address the ap￾propriate aims. Messer (1998, 15) notes that
“although relief officials try to make relief
function as development assistance, the
‘relief-to-development continuum’ they
talk about appears to be more wishful
thinking than fact. The bulk of emergency
food assistance is devoted to meeting ba￾sic human welfare needs.” Miller (1997, 15)
believes too much emphasis and attention
have been given to the concept of relief￾to-development, given the relatively brief
period when relief assistance overlaps
with longer term development assistance.
None of the six evaluations reviewed
by Apthorpe (1997, 92) concludes that re￾lief is poorly done if it is not specifically
forwardly linked to development. Two of
the six explicitly reject the linear linkage
as too simplistic and do not see it as the
best guide to what is needed in relief aid.
A third notes that the continuum concept
derives from natural disasters and there￾fore has only limited application to com￾plex emergencies, which are often politi￾cal in nature. Apthorpe concludes that
much of the literature on the relief–devel￾opment continuum has little operational
value.
Nevertheless, emergency programs
can have an important effect on shaping
the pattern and direction of subsequent
economic development. Societies recover￾ing from disastrous conflict are in the pro￾cess of remaking themselves, and the eco￾nomic opportunities that emerge from this
process can be influenced by how emer￾gency assistance is designed and deliv￾ered. What was the experience in Haiti,
Mozambique, and Rwanda?
T
5
34 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
Haiti
In Haiti, USAID emergency assistance
supported economic development in two
main ways: employment generation and
agricultural production. In addition,
USAID’s Office of Transition Initiatives pro￾vided small grants to various organiza￾tions to fund numerous microprojects.
Employment generation. The 1991–94
crisis worsened rural poverty in a coun￾try where poverty levels already were
among the highest in the world. When the
assembly plants in Port-au-Prince closed
because of the U.S.–led economic em￾bargo, 400,000 urban poor returned to the
countryside. That increased pressure on
rural households’ scarce resources. USAID’s
Jobs Creation project was implemented
during 1993–96 primarily to generate em￾ployment needed to maintain household
incomes and thereby offset the embargo’s
economic pressures. Its secondary objec￾tive was to rehabilitate productive infra￾structure.
The project created half a million per￾son-months of short-term employment
during 34 months. More than 120 indi￾vidual projects were carried out, resulting
in the repair of 1,000 miles of roads, 2,000
miles of irrigation canals, and 4,500 miles
of soil conservation barriers (PADF). The
project achieved its main objective—em￾ployment creation. However, maintenance
and long-term sustainability of the infra￾structure was not a project objective, and
over time the infrastructure has deterio￾rated. Although more durable infrastruc￾ture could have been built, that would
have required purchasing materials (rather
than hiring labor), and the primary objec￾tive of generating employment may have
been compromised.
Agricultural production. Another emer￾gency activity, funded by USAID’s Office
of Foreign Disaster Assistance and admin￾istered by Catholic Relief Services, was de￾signed to support agricultural production
and reduce decapitalization of Haitian
farm households. The project loaned agri￾cultural inputs (seed and fertilizer) to
farmers who were otherwise being forced
to sell their productive assets to buy food;
it also sold them tools at half price. Some
13,000 farming households (less than 10
percent of all farmers) and 47 farmer as￾sociations participated (Naval 1995). By
supplying inputs needed for food produc￾tion, the program provided emergency
assistance in a way that helped maintain
beneficiaries’ incomes and livelihoods and
at the same time reduced their dependence
on short-term relief.
Microprojects. The Office of Transition
Initiatives typically funds programs de￾signed to bridge the gap between short￾term relief (often managed by the Office
of Foreign Disaster Assistance) and long￾term development (typically managed by
resident USAID missions). These programs
often inject cash into an economy to gen￾erate employment or provide commodi￾ties to meet peoples’ most pressing needs
quickly. They are meant to create precon￾ditions for development and at the same
time facilitate the phaseout of emergency
Results: Economic Effects 35
assistance. This is important, because suc￾cessful rehabilitation is more difficult the
longer relief is provided (Kleist 1994,
300, 307).
The Office of Transition Initiatives
funded an $11 million program in Haiti
that supported more than 1,900 micro￾projects over a 27-month period during
1994–96. (This means that, on average,
more than two microprojects were initi￾ated each day during this period.) They
ranged from rehabilitation and construc￾tion of community schools, roads, markets,
canals, and bridges to the organization and
implementation of literacy, public health,
sanitation, reforestation, and civic educa￾tion activities. The projects were explicitly
designed to deliver assistance rapidly,
have high visibility, provide tangible ben￾efits, and support the legitimacy of local
grass-roots organizations. According to a
midterm evaluation, “there is absolutely
no doubt that this program has had un￾precedented success in mobilizing highly
valued resources to tens of thousands of
needy beneficiaries all over Haiti” (Chan￾dler 1996). The evaluation also notes, how￾ever, that the program emphasized instal￾lation more than maintenance.
In sum, employment generation ac￾tivities provided short-term benefits but
not permanent, off-farm sources of in￾come. Rehabilitated infrastructure contrib￾uted to increased economic activity in the
short term, but links to long-term eco￾nomic development were tenuous at best.
This is understandable. Relief agencies
were working in an environment of social
and economic chaos in which the goal of
long-term sustainable development was
eclipsed by the immediate short-term need
for relief. Haitians themselves were con￾cerned primarily with physical security
and survival rather than development.
Therefore, emergency assistance programs
generally were designed with only inci￾dental links to economic development, and
they had minimal developmental impact.
Although it is desirable to incorporate
long-term development objectives when
designing short-term emergency re￾sponses, the Haiti experience highlights
the difficulty of doing both well.
Mozambique
Apart from the Beira and Tete corri￾dors, which were secured by the Zimba￾bwean Army, there was no functioning
infrastructure (roads, bridges, or rail lines)
or rural markets in Mozambique in 1990.
They had been destroyed by years of civil
war. During the war USAID provided
emergency assistance to save lives and
alleviate suffering. When the war ended,
assistance programs turned toward reha￾bilitation and reconstruction. They had
two objectives: to restart the rural economy
and to restart subsistence agriculture.
Roads were demined, rebuilt, and re￾opened. That helped restore the private
transport sector and facilitated free
movement of goods (especially food) from
surplus to deficit areas, which spurred
market development. At the same time,
assistance was provided to displaced
farmers to help them resume agricultural
36 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
production and reduce their dependence
on food aid.
Seeds and tools. Farmers needed seeds
and tools to recapitalize their farms. The
total cost was small in absolute terms, less
than $50 per household. But it was large
relative to per capita income. PVOs located
seeds and tools (generally not available in
local markets), purchased them, and made
them available to farmers. Food for the
Hungry International also provided agri￾cultural inputs, conducted field trials to
identify higher yielding varieties with
shorter growing seasons, and introduced
improved farming practices. Both pro￾grams helped restart subsistence produc￾tion.
Food for work. USAID gradually
stopped providing relief food and began
supporting food-for-work projects. Doing
so helped break the dependency mental￾ity. The program supported labor-inten￾sive rural road construction and rehabili￾tation, construction of schools and health
clinics, and rehabilitation of small-scale ir￾rigation works. The quality of construction
was generally satisfactory, but mainte￾nance was questionable. As economic re￾covery continued, the food-for-work
projects evolved into cash-for-work
projects. That helped create sustainable
market mechanisms for supplying food
and other consumer goods as the cash
economy developed.
Economic liberalization. In 1990 Mo￾zambique’s rural markets were function￾ing poorly, and trade was limited. Civil
war and socialist economic policies had
taken their toll. USAID and the World Bank
supported efforts to reduce state control
of markets and prices and to promote
privatization of state-owned enterprises.
These measures helped establish the foun￾dation for rapid growth of small markets
and increased activity of private traders.
They also encouraged refugees and inter￾nally displaced persons to resettle. At the
same time, USAID funded a commodity
import program that supplied imports
needed to support economic liberalization.
These initiatives together with the pro￾gram to demine and rehabilitate roads
helped open up trade in rural areas.
Rwanda
Approximately one million Hutu
refugees returned to Rwanda within a one￾month period in late 1996. The humani￾tarian community immediately was faced
with the difficult task of helping this mass
of humanity put their lives back together.
The task of linking relief to development
assumed new dimensions. USAID ad￾dressed not only development needs in ag￾riculture, health, education, and commerce
but also the preconditions for develop￾ment: political stability, physical security,
justice, and legitimacy of the new govern￾ment.
Seeds and tools. The Office of Foreign
Disaster Assistance supported the distri￾bution of seeds and tools in almost every
region of Rwanda in 1995 and 1996. The
Results: Economic Effects 37
largely untargeted distribution was report￾edly successful. In 1997 the size of the pro￾gram was reduced, and by 1999 it was ter￾minated, except in the two northwest pre￾fectures where continued high levels of
insecurity had led to widespread looting
of farm supplies. Although livestock tra￾ditionally had been used to maintain soil
fertility and provide an important source
of nutrition to farm families, relief organi￾zations (including OFDA) were slow to in￾clude livestock in emergency packages.
The conventional wisdom was that ani￾mals represented a level of assistance far
beyond “emergency” requirements. How￾ever, most families are now acquiring ani￾mals, either through loan programs or
with their own savings.
Seeds of Hope. Agricultural experts
from the international agricultural re￾search centers recognized that productive
cropping in the various microclimates of
Rwanda required adapted seed. They
identified appropriate seed and rootstock
from their own seed banks as well as in
Rwanda where adapted seed still existed.
Seed stocks were multiplied and made
available in 1995. Because the experts had
alerted nongovernmental organizations
about the importance of planting adapted
seed, more local seed was used than oth￾erwise would have been the case. Seed
multiplication is now being expanded
under Seeds of Hope II. At the same time,
food-for-work programs were imple￾mented to reclaim wet lowland farming
areas (marais) and to improve terracing and
land productivity.
Capacity building. Most complex hu￾manitarian emergencies have occurred in
countries where the government was very
weak (as in Somalia). This meant nongov￾ernmental organizations could operate in
an environment relatively free from gov￾ernment intervention. Rwanda was an ex￾ception. The government preferred min￾istry-administered programs rather than
NGO-administered programs, and it found
it difficult to incorporate NGOs into its pro￾grams. Communication between the two
deteriorated, and in late 1995, 16 of about
60 humanitarian NGOs were expelled or
asked to suspend operations.
USAID was an early and strong sup￾porter of the government’s effort to claim
control of relief and development pro￾grams. Believing that a stable, fair, com￾petent government was key to Rwanda’s
successful agricultural and economic de￾velopment, USAID helped strengthen the
government’s capacity. Addressing justice
in response to the genocide had high pri￾ority. USAID funded the Rwandan-initiated
International Genocide Conference in
1995; trained court clerks in the Ministry
of Justice; supported a media campaign on
the genocide trial process; and developed
a central database for genocide prosecu￾tors. USAID also supported decentraliza￾tion of the Ministry of Health and helped
the ministry establish an emergency re￾sponse unit. Finally, USAID provided ba￾sic equipment to 10 ministries including
justice, health, interior, and the president’s
office. Most NGOs now work more closely
with government officials and seek oppor-
38 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
tunities to help build government capac￾ity.
Democratic initiatives. The Office of
Transition Initiatives has funded two ac￾tivities designed to support decentraliza￾tion and educate local leadership. One is
the Women in Transition program, which
has reached over 160,000 women. It en￾courages commercial interaction among
different ethnic groups. As part of a broad￾based program to support women in
postconflict situations, it also assists
women’s groups in the northwest, where
there has been an increase in the number
of women farmers. The other democratic
initiative is administered by Africare in the
Ministry of Interior. It supports election
education by building local decision-mak￾ing processes and grass-roots organiza￾tions. The program operates in 15 com￾munes in four prefectures.
All the activities described above
were designed to meet immediate needs
while preparing for follow-on programs.
In contrast to long-term development-ori￾ented programs whose success depends on
their being sustainable, this was not a cri￾terion for success in Rwanda. As the evalu￾ation states, “relief and transition pro￾grams . . . do not have to be sustainable.”
When Is
The Emergency Over?
At some point the emergency ends
and development resumes. But the demar￾cation between the two is not always clear.
In Haiti, emergency food assistance (dry
rations) was still being distributed in the
northwest as recently as July 1998—years
after the end of the crisis. (However, it was
scheduled to terminate in September
1998.)
In 1998, four years after the emer￾gency in Mozambique ended, most of the
village groups interviewed by the evalua￾tion team in the Beira corridor were still
asking NGOs for free seed, tools, food, and
even tractors. Although the NGOs had in￾formed recipients that free food would end
by a specific date, many did not expect that
to happen. They remained in the refugee
camps until free food was actually termi￾nated.
Donor pressure increased after the
Dayton Peace Accords in 1995 to reduce
emergency assistance to Bosnia. The logic
was that peace would bring stability and
economic recovery, making large reduc￾tions in food aid possible. There were also
the usual arguments about avoiding food
aid dependency and disincentives to ag￾ricultural production. In response to do￾nor demands, the joint UN High Commis￾sioner for Refugees/World Food Program
mission recommended reducing the num￾ber of direct beneficiaries from 1.6 million
to 600,000 (from over 50 percent of the
population to about 20 percent as esti￾mated by WFP in 1996) (UNHCR/WFP 1998,
8). That would help ensure that food aid
was not seen as an alternative to a social
welfare system. However, the joint evalu￾ation also pointed out the importance of
not scaling down too rapidly because an
Results: Economic Effects 39
effective social welfare system was not yet
in place.
In 1996–97, as many as 118,000 farm
families in Liberia received food rations
as well as seed and tools under a program
funded by the Office of Foreign Disaster
Assistance, the European Union, and the
Food and Agriculture Organization to
support the transition from war to recov￾ery. This ensured that rice seed was
planted rather than consumed and also
that farmers had the energy to work. As a
result, rice production increased from ap￾proximately 30 percent of prewar levels in
1996 to 60 percent of prewar levels in 1997
(USAID/Liberia 1998, 13). At the beginning
of 1997, 350,000 beneficiaries were receiv￾ing emergency food aid in IDP camps; by
the end of the year, 150,000 had been per￾manently resettled in rural areas. The do￾nors developed a plan to end general food
aid distribution in IDP camps after Febru￾ary 1998 and to rechannel these resources
to targeted activities in rural areas. These
included rural resettlement of refugees
and internally displaced persons, agricul￾tural recovery, school feeding, and food for
work.
 In Somalia, as elsewhere, humanitar￾ian assistance and related relief operations
generated substantial local employment
and purchasing power. Over 50,000 Soma￾lis found cash or food-for-work employ￾ment. But this all ended when the relief
ended, and the relief ended with termina￾tion of the donor-funded contracts with
NGOs that provided the relief (Kleist 1994,
295). This suggests that the amount of food
aid provided in a complex emergency is
driven not only by an assessment of needs
but also by the availability of donor re￾sources. The implication is that complex
emergencies may end too quickly (with
humanitarian needs still unmet) or not
quickly enough—depending on what the
various political interests stand to gain or
lose.
Hill (1997) reports that many conflicts
are unreconciled and have not ended. This
is true, for example, in Afghanistan, Iraq–
Kurdistan, Lebanon, and Somalia. When
a conflict does subside, it is for a reason.
Sometimes it is because one side wins. In
the post–Cold War era, though, this has
rarely been the case. More often it is be￾cause the war is no longer profitable. Of
course, economic issues are not the only
motives perpetuating complex emergen￾cies. Religious or ethnic hegemony have
often driven conflict (in Bosnia and Sri
Lanka, for example). But whatever its
causes, war is expensive and must be fi￾nanced. And for those who invest in the
war, it must be seen as profitable, at least
eventually; otherwise, they would cease
their support.
Economies in post–Cold War con￾flicts, as with economies in all wars, re￾volve around scarcity. With the disruption
of outside trade, loss of incomes, and sev￾ered or restricted corridors for delivery of
goods, food and other essentials become
very expensive. Profits are enormous for
those who have access to scarce resources
and can deliver them to areas where they
are needed. The combatants and their al-
40 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
lies are in the best position to manipulate
and profit from this trade.
Relief organizations are not. On the
contrary, relief organizations attempting to
operate in conflict situations provide a rich
source for exploitation. There are several
reasons. They have to operate with large
amounts of cash, which can be stolen. War￾ring factions can exact exorbitant fees from
relief organizations in return for provid￾ing them protection and ensuring their
access to insecure areas (as in Somalia).
And they can charge rents for warehouses
that are higher than market value. Some
of the most obvious spoils of war derive
from the blatant looting of infrastructure.
In Bosnia, for example, an entire
Volkswagen production plant near
Sarajevo was dismantled and sold. In So￾malia, almost every phone line, electric
cable, and water line was taken or ripped
up, put on a ship, and sold in some port
on the Indian Ocean (Hill 1997, 5). Thus,
peace can threaten a very profitable situa￾tion for the combatants.
Can donors help end wars by dem￾onstrating that more can be gained
through peace? According to Hill, the ba￾sic elements of reconstruction include (1)
identifying and creating markets for
manufactured and agricultural goods; (2)
reestablishing the rule of law, especially
in economic issues; (3) increasing oppor￾tunities for entrepreneurs; (4) supporting
self-sustaining lending institutions; and (5)
reestablishing acceptable levels of water,
power, heat, sanitation, and other basic
services. Education, often overlooked, is
also important. External assistance poten￾tially has an important role to play in all
of these areas.
Conclusion
The notion that relief assistance can
be made more “developmental” or that it
can be linked to development activities is
highly problematic in the context of ongo￾ing armed conflicts owing to the frequent
lack of local social and economic structures
that might legitimately be strengthened
(Borton and Macrea 1997, 8). Unlike the
case of natural disasters, with complex
emergencies there is no institutional
framework to provide security and justice,
both of which are necessary preconditions
for successful development activities. On
the contrary, complex emergencies are of￾ten characterized by (1) a total breakdown
of state institutions (Somalia, Liberia, Af￾ghanistan); (2) large areas of territory held
for prolonged periods of time by rebel
movements (Eritrea, Tigray, south Sudan);
or (3) a situation in which the occupying
regime had not received international rec￾ognition or was subject to international
sanctions (Sudan, Rwanda, Cambodia)
(Borton and Macrea 1997, 30).
In the absence of physical security and
political stability, support of long-term
economic development can be risky. In
Sudan, for example, water and health in￾frastructure that had been rehabilitated
was later destroyed by military action. In
Sri Lanka, the assets rebuilt by a World
Bank–led emergency rehabilitation and
Results: Economic Effects 41
reconstruction program in 1987–90 at a cost
of $125 million were destroyed when hos￾tilities resumed (Borton and Macrea 1997,
32). These examples reflect the unpredict￾ability of warfare. They also caution
against prematurely supporting long-term
investments in economic development
before political stability exists.
This does not suggest that relief agen￾cies blindly ignore development-oriented
opportunities that may arise. For example,
many NGOs providing emergency assis￾tance also routinely provide funds for ba￾sic agricultural inputs, food-for-work pro￾grams, and housing construction. But ex￾perience suggests that “large rehabilitation
financing may be more appropriately pro￾vided after resolution of the political frame￾work, rather than during the process of
political transition itself” (Borton and
Macrea 1997, 30–31).

Conclusions,
Lessons Learned, and
Recommendations
OME OBSERVERS CONSIDER each com￾plex humanitarian emergency unique.
The implication is that past experience is
not applicable to future crises. By contrast,
evaluation analysts tend to look for com￾mon themes—even when assessing the ef￾fectiveness of interventions in response to
complex emergencies. Their underlying
premise is that past experience is, in fact,
relevant for future crises. On the basis of
the three country case studies (Haiti,
Mozambique, and Rwanda) and evalua￾tions of other complex emergencies, at
least six common themes and four recom￾mendations emerge. Implications specific
to the Kosovo crisis are summarized in
annex D.
Conclusions and
Lessons Learned
1. Saving lives. Emergency assistance
programs funded by USAID and implemented
by American nongovernmental organizations
appear to deliver sufficient assistance to ensure
the survival of a country’s vulnerable poor,
though inadequate monitoring makes it diffi￾cult to quantify results.
One-half million to 1.3 million Hai￾tians (as many as one in seven) received
food aid during 1991–96. In Mozambique
an estimated one third of the population
of 16 million depended on food aid for 60
to 70 percent of their food needs in 1989.
In late 1996 and early 1997, 1.3 million
refugees were repatriated to Rwanda from
neighboring countries and received food
aid. Without massive infusions of pre￾dominately U.S. emergency assistance,
more Haitians would have fled Haiti seek￾ing refuge in the United States. Massive
starvation and human suffering would
have occurred in Mozambique and
Rwanda. Emergency assistance clearly
helped save lives and alleviate suffering.
However, except in Haiti, data collection
and monitoring were not done (or were
done poorly), so it is difficult to quantify
results.
2. Relief distribution. Effective distri￾bution of emergency assistance requires orga￾S
6
44 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
nization and control to limit theft, minimize
abuse, guard against political manipulation,
and protect beneficiaries.
Distributing relief supplies was a
problem to some extent in all three coun￾tries. The large quantity of food aid, in
particular, became a source of violent com￾petition—not only for its value as food for
consumption but also as a source of po￾litical power for those controlling access.
In Haiti, fighting among beneficiaries
sometimes erupted when food was distrib￾uted. Distribution points used to stockpile
food supplies were looted and supplies
were hijacked. Local authorities some￾times used food to favor certain political
factions or for their personal aggrandize￾ment. In Mozambique as well there were
reports of corruption, theft, and political
or personal favoritism in food aid distri￾bution. Target populations did not always
receive timely and sufficient food aid. In
Rwanda the military and former govern￾ment leaders controlled much of the relief
distribution. Thus they were able to divert
food from the intended beneficiaries for
their own purposes.
NGOs were mainly in charge of relief
distribution in Haiti. They addressed these
problems by stocking and distributing
food aid in neutral settings (schools, fac￾tory yards), using ration cards to track the
receipt of food aid, and having NGO per￾sonnel and occasionally police present to
monitor distribution. These measures lim￾ited diversion to less than 10 percent and
helped reduce violence. In Mozambique
the government emergency relief agency
lacked the technical expertise to plan, or￾ganize, and manage the distribution of
massive supplies of relief aid. Leakage was
typically 30 percent, and at one point 50
percent was lost, stolen, or diverted. In
response, donors, NGOs, and the private
sector took over much of the distribution,
and losses dropped to under 5 percent. In
camps in Tanzania and Zaire, more food
aid was supplied than was necessary, and
more than usual was misappropriated.
Some NGOs suspended their operations
because they knew they were assisting
people guilty of crimes against humanity.
3. Political and social unrest. Emer￾gency assistance can help maintain social calm
and mitigate political instability. Conversely,
it can exacerbate political tensions. Rarely is it
politically neutral.
The international community pro￾vided massive quantities of emergency
assistance to Haiti, Mozambique, and
Rwanda. The political effects of the assis￾tance varied. In Haiti, food aid reduced
the probability of food riots during a pe￾riod of political and economic stress and
may have had a dampening effect on po￾litical tensions; but it also may have re￾sulted in a political status quo that enabled
the de facto military regime to stay in
power longer. In Mozambique, external
military assistance provided by the Soviet
Union and by South Africa fueled the war.
Food aid, by comparison, had relatively
little effect on the country’s political dy￾namics, although food diverted to soldiers
may have contributed to the war effort. In
Rwanda, where genocidal killers were
Conclusions, Lessons Learned, and Recommendations 45
mixed with legitimate refugees in camps,
humanitarian assistance served to prolong
the emergency.
4. Demobilization. Demobilization,
disarmament, and reintegration of armed forces
is vital in ending a complex emergency and
beginning a period of recovery.
Demobilization of Haiti’s armed
forces removed one source of violence in
the country. However, many of the demo￾bilized soldiers retained their arms, and
because most were unemployed owing to
the weak economy, they are believed to
have caused at least part of the post-1994
rise in theft and street violence. In
Mozambique, demobilization of Renamo
and Frelimo armed forces and their rein￾tegration into civilian life was essential for
the transition from relief to recovery. As
in Haiti, though, many weapons were not
turned in and that contributed to a rise in
crime. In Rwanda, soldiers and militia
loyal to the former government remain
armed. They are still trying to destabilize
the present government of Rwanda by us￾ing a campaign of propaganda and terror
to destroy the political and social struc￾tures of the country, beginning in the
northwest.
5. Relief to development. Emergency
assistance that enables people to protect their
livelihoods (as well as meet immediate needs)
helps reduce dependency and contributes to
long-term economic development.
In Haiti many urban factory workers
lost their jobs, and some farmers were
obliged to sell their agricultural and house￾hold assets to survive. Numerous farmers
in Mozambique and Rwanda also lost their
productive resource base when they fled
their villages. These people became depen￾dent on emergency relief. USAID and the
NGOs responded—not only with food as￾sistance but also with agricultural inputs
(seeds and tools) and household goods.
That assistance encouraged refugees and
internally displaced persons to return to
their villages. It enabled them to resume
food production and decreased their de￾pendence on food aid. NGOs in all three
countries also implemented food-for-work
programs that created short-term jobs and
helped rehabilitate productive infrastruc￾ture (roads, irrigation) needed for eco￾nomic development. Often, though, the
infrastructure was not maintained.
6. Donor coordination. A clearly des￾ignated, agreed-upon central authority can
make the delivery of humanitarian assistance
more effective.
In Haiti the United Nations officially
designated the Pan American Health Or￾ganization as the coordination point for
overall health planning and services dur￾ing the U.S.–led embargo. That enabled
numerous NGOs to deliver medical sup￾plies and food to vulnerable populations
more effectively. In Mozambique, by con￾trast, donor efforts at times overlapped or
worked at cross-purposes. One donor was
giving free seed while another was sell￾ing it; one donor was shifting to develop￾ment assistance while another was still
providing grant relief. That confused ben-
46 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
eficiaries and undermined efforts to re￾duce dependency. Similarly, lack of donor
coordination was a serious problem in
Rwanda.
Recommendations
Four key recommendations emerge
from these six conclusions and lessons
learned. They are mainly management
oriented.
1. Monitoring and evaluation. Estab￾lish a central monitoring and data-collection
unit to serve all donors during the early weeks
of a complex emergency.
Baseline data for socioeconomic indi￾cators (e.g., malnutrition rates, food prices,
population displacement) can help man￾agers identify appropriate kinds of emer￾gency relief, target its distribution, and
subsequently measure and evaluate its ef￾fectiveness. Close monitoring enhances
donor coordination and is essential for
assessing aid needs, avoiding work at
cross-purposes, identifying recipient
groups no longer needing emergency aid,
shifting from relief to reconstruction and
development, and designing and adjust￾ing economic policies.
2. Adverse political consequences. Be
alert to potential undesirable political or social
effects that relief aid may cause.
Emergency food distribution, in par￾ticular, can have unintended and undesir￾able political consequences. Control over
final distribution often has reinforced the
power of local authorities or political fac￾tions. It has strengthened their relative
position during or after the conflict and
facilitated their self-aggrandizing, often
exploitive, behavior toward the intended
noncombatant beneficiaries. Decisions to
continue, withdraw, or modify aid distri￾bution should be made as a matter of de￾liberate policy on a regular basis by each
individual donor.
3. Reducing dependency. Give refu￾gees incentives to return home and impose dis￾incentives on those remaining outside their
country of origin.
Generally, the longer encampment or
temporary foreign residence lasts, the less
willing refugees are to return home. A
combination of “push” factors (such as ter￾minating free food distribution) and
“pull” factors (such as including seeds and
tools in resettlement packages) is likely to
accelerate the repatriation process. But for
reasons of political and bureaucratic self￾interest, local governments may not re￾move from the rolls those no longer need￾ing relief. Therefore, donors must moni￾tor each situation closely, recognizing that
both relief and development assistance
may be needed if some areas remain in
emergency status while others stabilize
more quickly. After populations have been
repatriated and are settled, the agricultural
base begins to be reestablished, depen￾dency on free food distribution drops, and
long-run food security is enhanced.
Conclusions, Lessons Learned, and Recommendations 47
4. Capacity building. Train technocrats
to manage the postconflict economic transition,
and train others in skills for which there is
employment demand.
Complex emergencies seriously
weaken the capacity of governments to
provide basic public services. Economic
recovery requires a cadre of high-level
technocrats with both management and
conceptual skills, especially in macroeco￾nomic and sectoral policy formulation.
Such skills are likely to be in short supply,
especially if preconflict professionals and
the intelligentsia were targeted for delib￾erate elimination or have permanently left
the country. Recovery also needs to be
dovetailed with postconflict economic re￾alities. Job training is fruitless if unemploy￾ment in the depressed economy remains
high. Training is especially critical for de￾mobilized soldiers who, because they of￾ten remain unemployed, tend to turn to
destabilizing criminal activity.

Annex A
Humanitarian
Emergencies and
Donor Assistance
HE TABLE BELOW indicates the num￾ber of people affected by humanitar￾ian emergencies in recent years. The four
figures in this annex show changing
levels of official development assistance
and humanitarian assistance provided by
the donor community and the United
States over the past 30 years.
Source: U.S. Mission to the United Nations, April 1997 and September 1998. Figures are based on
data provided by the U.S. Committee for Refugees.
Note: The table includes only major emergencies, those in which at least 300,000 people required
international humanitarian assistance to avoid severe malnutrition or death.
January 1997
3.5
2.5
0.78
3.1
1.0
0.48
0.7
2.5
0.3
>0.5
2.1
2.0
>5.0
0.35
0.63
1.5
1.0
0.85
4.4
0.63
33.82
April 1998
4.1
2.5
0.77
1.5
0.75
1.0
0.45
0.30
0.85
0.3
0.5
1.7
1.1
>7.4
0.3
>1.0
1.0
0.7
4.4
0.9
0.4
31.92
Table A1. Ongoing Humanitarian Emergencies and Number of
People Affected (Millions), 1996, 1997, 1998
Afghanistan
Angola
Azerbaijan
Bosnia and Herzegovina
Burundi
Colombia
Croatia
Eritrea
Ethiopia
Georgia
Haiti
Iraq
Liberia
North Korea
Russia (Chechnya)
Rwanda
Sierra Leone
Somalia
Sri Lanka
Sudan
Tajikistan
Uganda
Total
January 1996
4.0
2.5
0.95
3.7
0.8
0.5
1.0
3.5
1.0
1.1
2.65
1.5
0.3
1.0
1.8
1.0
0.85
4.0
1.0
33.15
T
A2 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
Figure A1
Official Development Assistance and Emergency/Distress Relief, All Donors, 1969–98
(billions $US 1998)
Source: OECD/DAC
Figure A2
Emergency and Distress Relief as a Percent of Official Development Assistance,
All Donors, 1969–98
Source: OECD/DAC
0
10
20
30
40
50
60
70
80
1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997
ODA
0
1
2
3
4
5
E/DR
ODA
E/DR
0
1
2
3
4
5
6
7
8
1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997
percent
Annex A: Humanitarian Emergencies and Donor Assistance A3
Figure A3
U.S. Official Development Assistance and Emergency/Distress Relief, 1971–98
(billions $US 1998)
Source: OECD/DAC
Figure A4
Trends in U.S. Official Development Assistance and Emergency/Distress Relief,
1971–98
Source: OECD/DAC
0
2
4
6
8
10
12
14
16
1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997
ODA
0.0
0.5
1.0
1.5
2.0
E/DR
U.S. ODA
U.S. E/DR
0
10
20
30
40
50
60
1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997
percent
U.S. E/DR % of all E/DR
U.S. ODA % of all ODA
U.S. E/DR % of U.S. ODA

Evaluation Objectives
And Constraints and
Implications for
Donor Coordination
Evaluation Objectives
HE OVERALL OBJECTIVE of the CDIE
evaluation was to assess the effective￾ness of U.S. emergency assistance in
response to complex humanitarian emer￾gencies.*
 It covered all types of relief as￾sistance including (1) food, (2) water and
sanitation supplies, (3) medical services
and health care, and (4) clothing, shelter,
and resettlement assistance. The evalua￾tion in each country examined the politi￾cal and historical events that caused the
emergency as well as the effectiveness of
the U.S. humanitarian response. Conclu￾sions and lessons learned are cast in terms
of management recommendations to help
guide USAID’s future policy, program, and
budget decisions in humanitarian assis￾tance.
Evaluation Methodology
The evaluation looked at three main
topics framed in the form of these ques￾tions:
1. Did U.S. emergency assistance save
lives and alleviate suffering during the
complex emergency?
2. Did U.S. emergency assistance af￾fect social and political hostilities (or ten￾sions) associated with the complex emer￾gency?
3. Did U.S. emergency assistance con￾tribute to economic development?
The first question derives directly
from legislation that gives USAID primary
responsibility within the U.S. government
for responding to overseas disasters. Ques￾tions 2 and 3 are not directly related to the
overarching mandate to “save lives and
alleviate suffering.” The second question
concerns the potential political impact of
emergency assistance and whether such
assistance has had the unintended effect
of prolonging complex emergencies. The
third question addresses the potential eco￾nomic impact of the assistance and the so￾called relief-to-development continuum.
Background work for the evaluation
was carried out in Washington, and field￾work was conducted in Haiti, Mo￾Annex B
T
*The “Emergency Assistance” concept paper
(11 December 1997) provides the underlying
rationale for the assessment.
Complex Humanitarian Emergencies and USAID’s Humanitarian Response
zambique, and Rwanda. During the first
phase, desk studies for each of the three
countries were completed, a workshop
was convened to review the desk studies,
and two topical guides were developed to
help structure key informant interviews
and field observations. One topical guide
was for implementers and experts and the
other was for beneficiaries.
The second phase involved fieldwork
in each of the three countries. Evaluation
teams sought answers to the three ques￾tions above through careful analysis of
secondary sources not available in the
United States (reports, evaluations, agri￾cultural production surveys, health statis￾tics), interviews with key informants and
focus groups, and site visits. The results
of the three case studies were synthesized
during the third phase.
During the second phase, evaluators
talked with a broad range of national and
expatriate experts (field technicians, pro￾gram directors, administrators) who had
managed or implemented emergency as￾sistance programs. These people typically
included donor agency staff, NGO partners,
host government officials (both national
and provincial), UN agencies, and some
academics and journalists. Beneficiary￾level information was collected in both
urban and rural areas. Site visits produced
valuable insights and helped corroborate
information from other sources and to
ground-truth the teams’ interpretations.
The process of interviewing and listening
produced answers and also generated sec￾ondary questions.
This mainly qualitative methodologi￾cal approach does not produce statistically
valid proof of impact. Rather, it allows an
interpretation of the links between a USAID
intervention and various effects that plau￾sibly can be associated with that interven￾tion. Given the characteristics of humani￾tarian assistance, a more scientific ap￾proach is rarely feasible (Hallam 1998, 28).
One can judge the validity of the interpre￾tations on the basis of several criteria: the
logic and consistency of the arguments
substantiating them, the strength and qual￾ity of the evidence, triangulation, and the
reputations of those involved. This is
sometimes called the common-sense
school of evaluation: impact is deduced
from a combination of information from
key informants and from the evaluators’
own sense of how the world works.
Evaluation Constraints
Donor agencies that provide relief in
the context of conflict or civil war must
consider factors they normally would not
need to. These factors include sovereignty,
international law, the appropriate balance
of aid between opposing sides, and per￾haps national foreign policy interests. The
political and legal questions associated
with humanitarian assistance in conflict
areas make the evaluation of relief pro￾grams far more sensitive than that of de￾velopment programs (Borton 1994, 11–12).
Moreover, difficulties typically asso￾ciated with evaluating development assis￾tance programs seem to be magnified
B2
Annex B: Evaluation Objectives and Constraints and Implications for Donor Coordination
when evaluating relief programs. These
include (1) lack of adequate baseline data;
(2) difficulty in identifying control groups;
(3) the dynamic context of relief programs,
which makes it difficult to isolate the ef￾fect of the relief intervention; (4) the lim￾ited utility of cost–benefit analysis; (5) the
large number of agencies involved; and (6)
the high political and media profile of re￾lief programs (Borton 1994, 12).
There is yet another difficulty with
evaluating relief programs: the prevalent
attitude among many relief agencies that
assessments of their programs are unnec￾essary. To paraphrase their view: our mo￾tives were well intentioned, we did our
best under difficult circumstances, why
should we now subject ourselves to a criti￾cal examination? (Borton 1994, 1). Relief
agencies also may believe that “all disas￾ters are different, so what is the point in
trying to learn the lessons of our response
to this particular disaster?” (Borton 1994).
According to Larry Minear (1998),
four characteristics of humanitarian orga￾nizations make them resistant to change:
1. As just noted, humanitarian orga￾nizations tend to approach every crisis as
unique. “As long as every crisis is per￾ceived as wholly without precedent or
parallel, there will be little scope for insti￾tutional learning (10).” Correcting this ten￾dency will require greater institutional
memory, greater attention to comparative
analysis, and more support for in-house
evaluation.
2. Humanitarian organizations are
action oriented. As a result, evaluations
generally are not read and lessons are not
learned (11).* Although the action orien￾tation of NGOs is a positive attribute, it
should be informed by past experience as
well as current political, military, and so￾cial realities. Unfortunately, NGOs some￾times ignore this experience rather than
use it as a basis for more strategic inter￾vention.
3. Humanitarian organizations are
often defensive about criticism, even con￾structive criticism. As such, they are un￾likely to use it to improve their effective￾ness (12). Nevertheless, constructive criti￾cism can be valuable, which highlights the
need for independent research and evalu￾ation.
4. Humanitarian organizations often
lack accountability. They are not held re￾sponsible for their actions: everybody—
and thus nobody—is responsible. This lack
of accountability will be difficult to rem-
*The multidonor Joint Evaluation of Emergency As￾sistance to Rwanda demonstrates the importance of
evaluations but also demonstrates their limited
ability, in and of themselves, to produce institu￾tional change (6). This comprehensive evaluation
reviewed aid programs that cost $1.4 billion from
April through December 1994. It was done by 37
institutions (governmental, intergovernmental,
and nongovernmental), enlisted 52 consultants,
produced a five-volume report, and cost $2 mil￾lion. The broadest of its 64 recommendations were
ignored; those stressing “coordination by com￾mand” were rebuffed; and only the least radical
options were acted on to some degree.
B3
Complex Humanitarian Emergencies and USAID’s Humanitarian Response
edy because of the accepted approach of
coordination by consensus (or default)
rather than coordination by command (12–
14).
Implications for
Donor Coordination
This raises the issue of donor coordi￾nation. Coordination is a concept ap￾proved by all but defined by few
(Prendergast and Scott 1996). A major
emergency assistance operation can in￾volve numerous bilateral and multilateral
donors, hundreds of NGOs, a range of UN
agencies, military contingents, and na￾tional governments. An effective division
of labor among these and other actors is
needed to maximize the comparative ad￾vantage and impact of each.
Donor coordination is often perceived
as a role for the United Nations. That body
expanded its peacekeeping operations at
the end of the Cold War when complex
emergencies proliferated. But peacekeep￾ing operations required endorsement by
the Security Council. This meant the Per￾manent Five could control UN peacekeep￾ing and enforcement operations given
their veto power and control over finances.
In the case of Rwanda, the major powers
on the Security Council (except France)
made clear they were not interested in a
small African country that was marginal
to their economic or political concerns and
peripheral to international strategic rival￾ries. This experience suggests that the
United Nations is not always best
equipped to coordinate an international
response to a complex emergency.
The only country with a demon￾strated ability to energize the United Na￾tions and the Security Council in a crisis is
the United States. But in the case of
Rwanda, even the United States, haunted
by the memories of Somalia, was deter￾mined not to get involved in another Afri￾can conflict. Not crossing the “Mogadishu
line” became the guiding principle. Wash￾ington was also preoccupied with crises
elsewhere, especially in Bosnia and Haiti,
and the potential financial burden of
Rwanda was a major concern. The United
States at the time was assessed 31 percent
of the costs of all UN peacekeeping op￾erations. (Joint Evaluation 1996, study 2,
11). In short, even the United States is not
always well positioned to coordinate an
appropriate humanitarian response to all
complex emergencies.
Lack of donor coordination in provid￾ing humanitarian assistance is well docu￾mented. This is true within a single orga￾nization as well as among several organi￾zations. For example, within the UN sys￾tem, UNICEF has responsibility for women
and children; but UNHCR has responsibil￾ity for refugees, including women and
children. Similarly, WFP has responsibil￾ity for food assistance; but UNICEF and
WHO may be involved in actual program
operations. Lack of coordination can re￾sult in inefficiency. For example, in most
situations each agency arranges local
B4
Annex B: Evaluation Objectives and Constraints and Implications for Donor Coordination
transport for its own supplies, but this pro￾cess only bids up prices. Another issue
having cost-related implications is that hu￾manitarian organizations are heavily
weighted toward mop-up operations; they
are reactive, not proactive. Too little atten￾tion is given to investments in conflict pre￾vention (Minear 1994b, 4, 6); see annex C.
The creation of the UN Department
of Humanitarian Affairs (and subse￾quently, the Office for the Coordination of
Humanitarian Affairs) is generally seen as
a positive step. Nevertheless, the UN is
often criticized. For example, each of the
six evaluations reviewed by Apthorpe calls
for some type of reform in the United Na￾tions. James Ingram (former head of WFP)
is more draconian: he believes there is no
reason that a coordinated international re￾sponse to future complex emergencies
should necessarily be built around the
UN—and suggests there are a variety of
reasons why it should not. He recom￾mends the International Committee of the
Red Cross or a new organization outside
the UN system (Apthorpe 1997, 96, 98).
In disaster after disaster, local insti￾tutions and people provide the first line
of response. By contrast, the world’s hu￾manitarian system relies heavily on out￾side resources that marginalize local re￾sources and expertise. The idea seems to
be fixed that in complex emergencies there
is simply no alternative to using donor
resources delivered by NGOs. This may be
true. If so, donor resources should be pro￾vided to support existing livelihoods or
coping strategies as much as possible.
B5

Are Complex
Emergencies
Predictable?
N REFERENCE to the Rwanda crisis,
James Kunder notes that “an ounce of
prevention is worth 25,000 tons of food
aid” (Minear 1994, 9). The point is well
taken. It is far better to prevent complex
emergencies from occurring in the first
place than it is to respond to victims’ needs
afterwards. What factors seem to be re￾sponsible for igniting complex emergen￾cies? Are complex emergencies predict￾able? Regions that are particularly prone
to seasonal natural disasters (for example,
cyclones in India, hurricanes in Florida,
drought in the Sahel) have developed
early-warning systems designed to predict
the next natural disaster and to mitigate
its effects. Such is not the case with com￾plex emergencies. However, studies have
been undertaken to help explain the causes
of civil wars and complex emergencies,
and that may be the first step in predict￾ing their occurrence. This annex summa￾rizes the results of several of these stud￾ies.*
The ethnic model is the prevailing ap￾proach to explaining complex humanitar￾ian emergencies. It postulates that
ethnicity (based on differences of lan￾guage, race, tribe, religion, national origin,
or some other cultural sense of identity) is
the primary factor underlying a complex
humanitarian emergency. By contrast, the
economic model views complex humani￾tarian emergencies and ethnic conflict in
the context of economic development and
structural change. In this model, economic
factors are pivotal in shaping conflicts,
though these conflicts may be triggered by
political or ethnic causes. That is, economic
factors create the conditions for ethnic or
political explosions that in turn lead to
complex emergencies (Nafziger 1996,
3–4).
Paul Collier and Anke Hoeffler (1998)
are among those who explain the phenom￾enon in economic terms using economic
analysis. They identify four variables that
they hypothesize are associated with civil
wars: per capita income, size of the natu￾ral resource base, population size, and
ethnolinguistic fractionalization. Using a
sample of 98 countries (of which 27 had
civil wars during 1960–92), they analyze
the relative importance of these four vari￾ables. Their results show that all four vari-
*A substantial body of literature has developed on
this subject. This annex only scratches the surface.
I
Annex C
Complex Humanitarian Emergencies and USAID’s Humanitarian Response
ables are significant determinants of both
the probability of civil wars occurring and
their duration. Specifically:
1. Per capita income. The study shows
that civil war is overwhelmingly a phe￾nomenon of low-income countries. Other
things being equal, the probability of civil
war is substantially greater in countries
with a very low per capita income than in
countries with a relatively high per capita
income. Moreover, the predicted duration
of civil war is much shorter in countries
with a higher per capita income. This is
because a high-income population has
more to lose during a conflict, and the costs
of rebellion increase with its duration.
Conversely, the opportunity cost of being
a rebel and prolonging a conflict is low for
a low-income population.
2. Natural resource base. The effect of
natural resource endowments is not as
straightforward. Initially, increased natu￾ral resources increase the risk and dura￾tion of civil war. This is because the tax￾able base of the economy constitutes an at￾traction for rebels wishing to capture the
state. But at a high level, natural resources
start to reduce the risk of civil war. This is
due to the government’s greater financial
capacity to defend itself through military
expenditures.
3. Population. Countries with larger
populations have a higher risk of civil war
and wars that are likely to last longer. This
reflects the greater likelihood that coun￾tries with larger populations will have a
larger number of diverse groups, which
in turn suggests there would be a greater
likelihood of various ethnic groups want￾ing to secede owing to cultural and lin￾guistic disparities. (Note, however, that the
effect of population size is ambiguous be￾cause potentially it could be inconsistent
with point 4, next.)
4. Ethnolinguistic fractionalization. This
characteristic is an index that ranges from
0 (complete homogeneity) to 100 (maxi￾mum fractionalization). The index would
be 100 when each individual in a country
was in a different ethnolinguistic group.
Conversely, the index would be 0 in a so￾ciety with a single ethnolinguistic group.
Contrary to conventional wisdom (and to
point 3, above), the authors find that more
fractionalized societies are no more prone
to civil war than highly homogeneous
ones. Rather, the danger of civil war arises
when a society is polarized into two
groups. Polarized societies have a much
higher probability of civil war than either
homogeneous or highly fractionalized
ones. Thus, a country with two similar-size
ethnolinguistic groups could reduce the
risk of civil war either by partition or by
union with other countries.
Not surprisingly, Collier and Hoeffler
conclude that the “ideal society” (one en￾dowed with the most favorable of each of
these characteristics) has less risk of civil
war than the “catastrophic society” at the
other end of the spectrum (one with the
least favorable of each characteristic). They
also conclude that poverty is the main
cause of civil war. Table C1 lists the sample
of 98 countries in their analysis. Countries
C2
Annex C: Are Complex Emergencies Predictable?
Algeria
Burundi
Chad
Dominican Republic
El Salvador
Ethiopia
Guatemala
India
Indonesia
Iraq
Liberia
Mauritania
Morocco
Mozambique
Myanmar
Nicaragua
Nigeria
Pakistan
Peru
Philippines
Somalia
Sri Lanka
Sudan
Turkey
Uganda
Zaire
Zimbabwe
Argentina
Australia
Austria
Barbados
Benin
Bolivia
Brazil
Burkina Faso
Cameroon
Canada
Cent. African Rep.
Chile
Congo
Costa Rica
Denmark
Ecuador
Egypt
Finland
France
Gabon
Gambia
Germany
Ghana
Greece
Guyana
Haiti
Honduras
Hong Kong
Iceland
Ireland
Israel
Italy
Ivory Coast
Jamaica
Japan
Kenya
Korea
Madagascar
Malawi
Malaysia
Mali
Malta
Mauritius
Mexico
Nepal
Netherlands
New Zealand
Niger
Norway
Panama
Papua New Guinea
Paraguay
Saudi Arabia
Senegal
Sierra Leone
Singapore
South Africa
Spain
Sweden
Switzerland
Syria
Tanzania
Thailand
Togo
Trinidad and Tobago
Tunisia
United Kingdom
United States
Uruguay
Venezuela
Zambia
With a Civil War Without a Civil War
Table C1. Countries With and Without a Civil War, 1960–92
Source: Collier and Hoeffler 1998, 573.
C3
in column 1 experienced civil war during
1960–92; the rest did not.
Nafziger also identifies economic fac￾tors as primarily responsible for complex
emergencies (Nafziger 1996, v). He speci￾fies four, some of which parallel those
identified by Collier and Hoeffler:
1. Prolonged stagnation. The majority
of countries with humanitarian emergen￾cies have experienced several years (or
even decades) of negative or stagnant eco￾nomic growth. Below a given threshold, a
protracted decline in incomes is likely to
trigger increasingly fierce competition for
scarce resources, jobs, and opportunities
(5).
 2. Unequal growth. The situation is
likely to deteriorate more rapidly if income
and asset distribution worsen. Skewed
C4 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
economic growth increases the relative
deprivation of substantial sections of the
population, even if it does not cause abso￾lute deprivation (6).
3. Population pressure on resources.
Rapid population growth coupled with en￾vironmental degradation and resource
depletion can contribute to diminishing
returns to agricultural land. Declining ag￾ricultural returns, often exacerbated by
maldistribution of land and water, are a
source of conflict (8).
4. Distributional shifts owing to adjust￾ment programs. Large and abrupt shifts in
the distribution of income and wealth dur￾ing stabilization and liberalization pro￾grams can affect the distribution of power
within a country (9).
According to Nafziger, the way in
which elites react to these four factors in￾fluences the probability of political con￾flict and humanitarian disasters occurring
(Nafziger 1996, 10).
De Soysa and Gleditsch of the Inter￾national Peace Research Institute, Oslo,
analyzed 103 armed conflicts that occurred
during 1989–97, since the end of the Cold
War. They found that political instability
that led to violent conflict has sprung
mainly from economic concerns, rather
than political or ideological differences.
They also found that most contemporary
armed conflicts have occurred in impov￾erished countries where agriculture was
the mainstay of the economy. Armed con￾flicts were often fought over issues related
to agriculture, such as land ownership,
environmental change, water scarcity, and
food shortages. However, divisions over
these issues often fall along ethnic lines,
obscuring the fundamental causes.
The researchers conclude that condi￾tions in poor countries that undermine the
rural economy can generate political griev￾ances that result in endemic armed con￾flict. Agriculture is the dominant economic
sector in most poor countries. Poor coun￾tries that invest in their agricultural sec￾tors provide livelihoods to people and
thereby lower the incidence of conflict.
People do not need to turn to violent move￾ments as a means of survival. These find￾ings, like those of Collier and Hoeffler and
Nafziger, stress the importance of eco￾nomic factors in explaining the outbreak
of complex emergencies—and in particu￾lar, the importance of agriculture.
Finally, Mary Anderson also identi￾fies economic considerations among the
key underlying factors that contribute to
civil war. She suggests that some engage
in civil war because they have little to lose;
others, because they have something to
gain. Those with few economic alterna￾tives (the poor) are often the rebels who
have little to lose. They are often supported
by those who have something to gain
(arms merchants and other profiteers).
Others who stand to gain include those
whose employment depends on continu￾ation of the war and the funding it gener￾ates.
Annex C: Are Complex Emergencies Predictable? C5
Although many leaders claim to be
engaged in a struggle against past injus￾tices, the evidence available suggests that
justice and fairness are neither the funda￾mental motives nor the likely outcomes of
their wars. Rather, these leaders seek
power. However, once a civil war starts,
the war itself creates a spiral of atrocities
and reprisals and hatred that become the
legitimate root cause for its continuation
(Anderson 1996, 10–12). As reported in
one anthropological study, vengeance is
the major reason for civil war in 75 per￾cent of the cases.
In sum, complex emergencies clearly
are not predictable. Moreover, the incon￾sistencies among the findings of the stud￾ies just noted illustrate that there is no gen￾erally accepted theory of conflict. It seems
equally clear, though, that economic fac￾tors are of major importance in generat￾ing these crises, and they may well have
an important role in preventing them.

Implications for
The Kosovo Crisis
N MAY 1999 the Kosovo crisis embraced
much of the Balkans in one way or an￾other. Various proposals were being
mooted for international and regional ac￾tion to secure a stable peace when the con￾flict ended. The length of the conflict, the
extent of casualties and population dis￾placement, the amount of physical de￾struction, the postconflict sovereignty con￾figuration, and the lingering political and
social effects on the states most affected
were all unpredictable at that time.
This annex identifies lessons from this
and other evaluations of complex humani￾tarian emergencies that might be useful in
coping with the Kosovo crisis. Applicabil￾ity of any individual lesson depends, of
course, on how the crisis develops, how it
is resolved, and what postconflict politi￾cal architecture emerges. The lessons and
observations have been culled from docu￾ments prepared by the following: CDIE, the
Department for International Develop￾ment of the UK, the Directorate General
of International Cooperation of Nether￾lands, the Danish International Develop￾ment Agency, the Organization for Eco￾nomic Cooperation and Development/
Development Assistance Committee, the
UN High Commissioner for Refugees/
World Food Program, the International
Bank for Reconstruction and Develop￾ment/Operations Evaluation Department,
the International Crisis Group, and this
author.
Crisis Management
1. Coordination. The Kosovo crisis de￾mands an effective coordination mechanism.
Virtually all evaluations dealing with
complex humanitarian emergencies cite
inadequate donor coordination as causing
serious problems—inefficiencies, waste,
lost leverage. Complex emergencies typi￾cally face unusually complex coordination
problems: large numbers of multi- and
bilateral donors and nongovernmental
organizations working in a context of con￾Robert J. Muscat prepared the original version of
this annex (25 May 1999). In anticipation of the
end of the Kosovo crisis, Muscat synthesized les￾sons from this and other evaluations of complex
humanitarian emergencies for possible application
in Kosovo.
I
Annex D
Complex Humanitarian Emergencies and USAID’s Humanitarian Response
tested or collapsed governance, sometimes
with external military organizations in￾volved (Bosnia, Cambodia, Kosovo, Soma￾lia). Coordination has ranged from volun￾tary, committeelike structures to more hi￾erarchical, or command, systems involv￾ing lead agencies or even a UN or other
sanctioned authority with powers of di￾rection (Bosnia, Cambodia).
Postconflict Kosovo is likely to have
an interim international administration
under which a centrally guided coordina￾tion system would be appropriate. Of all
recent complex humanitarian emergen￾cies, the Bosnian and Cambodian experi￾ences have the most relevant lessons in this
regard, especially concerning the security–
political–economic interfaces and the
exercise of authorized powers. (Extensive
independent and internal agency evalua￾tions of the international administration
experience of these two cases are avail￾able.)
2. Monitoring. A central monitoring
and data collection unit should be set up to
serve all donors.
Complex emergency information￾sharing and monitoring systems have also
been evaluated negatively. A central unit
to collect and analyze socioeconomic data
and program information, one that serves
all donors (again, as in Bosnia or Cambo￾dia), is needed to ensure donor coordina￾tion. Close monitoring is essential for as￾sessing needs, avoiding work at cross-pur￾poses, identifying recipient groups no
longer needing emergency aid,
transitioning from relief to reconstruction,
and designing and adjusting economic
policies. The Rwanda experience demon￾strated that good information on the con￾cerns and expectations of encamped refu￾gees, and on the power and leadership
structures in the camps, is critical for main￾taining orderly relations and confidence
among refugees, assistance authorities,
and host governments.
3. Food distribution. Local food distri￾bution organizations commonly need close
monitoring to avoid factional diversion or
politicization.
Judging by the international
community’s generally successful hu￾manitarian response record, malnutrition
or disease is unlikely to become a signifi￾cant complication of the current crisis, with
two important caveats. First, given inter￾nal displacement, exposure to ethnic
cleansing, and the bombing of supply in￾frastructure, those remaining inaccessible
inside Kosovo could suffer severe priva￾tion, depending on how long the conflict
lasts. Second, Kosovo, Albania, and
Macedonia each have deeply divided po￾litical factions and parties. In similar cases
(Ethiopia, Sudan), factions that controlled
food aid distribution have withheld food
from entitled beneficiaries loyal to rival
factions. Thus far, the political dynamics
in Albania and Macedonia have reportedly
facilitated rather than hindered aid distri￾bution to refugees. Given the fragile poli￾tics in both countries and the stresses gen￾erated by the crisis, the integrity of emer￾gency aid distribution should be closely
D2
Annex D: Implications for the Kosovo Crisis
monitored. If postconflict arrangements
retain Kosovo’s ethnic Serb minority, the
problem could arise there, unless distri￾bution is well controlled by external agen￾cies.
4. Refugee repatriation. Refugees may
require incentives to return home.
Generally, the longer encampment or
temporary foreign residence lasts, the less
willing refugees are to return home
(Mozambique). A combination of incen￾tives for returning to Kosovo and disin￾centives for remaining outside may be re￾quired if NATO’s expressed optimism
about early return does not materialize.
5. World Bank coordination of recon￾struction. Several mechanisms can be used to
strengthen World Bank performance.
The international community has re￾cently relied on the Bank to take the lead
in reconstruction, monitoring, and other
nonpolitical and nonmilitary functions. On
the basis of the Bank’s own evaluation, its
effectiveness in complex humanitarian
emergencies has varied, depending to a
considerable extent on senior manage￾ment’s level of interest and commitment
as reflected by (a) authorities and size of
staff of the in-country resident represen￾tative’s office, (b) speed of headquarters’
processing and willingness to cut bureau￾cratic corners, and (c) size of administra￾tive budget.
The Bank’s performance in some cases
has been enhanced by (a) receipt of (grant)
trust funds from donors interested in in￾ducing and supporting a larger Bank ef￾fort than might otherwise be the case, and
(b) strong advocacy, by one or more board
members, for maximum Bank perfor￾mance.
Relief to Development:
Transition and Links
1. Funding discontinuities. USAID
should avoid interruptions in funding and
operations during the transition phase.
Relief funding affords considerable
flexibility compared with the procedural
complexity and slower pace of long-term
development operations. Thus, USAID and
NGOs have experienced funding gaps and
program interruptions between the phase￾out of relief assistance and the phasein of
development assistance (Mozambique).
The two funding systems may need to
operate simultaneously in countries where
some areas remain in emergency status
while other areas have stabilized. The ex￾istence of land mines in Kosovo could ne￾cessitate operating in both modes at once.
Projects undertaken as rapid transition
responses should not ignore longer run
reconstruction and development objec￾tives. This may require close coordination
among USAID’s units responsible for dif￾ferent funding sources and operations.
2. Resettlement planning. Realistic
planning for resettlement in Kosovo should al￾ready be under way.
D3
Complex Humanitarian Emergencies and USAID’s Humanitarian Response
The needs of returning refugees dur￾ing the initial stages of resettlement may
seem obvious. In practice, though, resettle￾ment planning has often been deficient,
based on unrealistic assumptions, inad￾equate information, and poor analysis of
the conditions prevailing in repatriation
destinations (Cambodia, Haiti, Mo￾zambique). If repatriation begins as early
as autumn 1999 (as NATO spokesmen have
asserted), then planning, preparation, and
financial provision should already be un￾der way, including mobilization for com￾munity and home demining.
Since repatriation will occur at the
onset of winter, the repatriation package
will need to include housing reconstruc￾tion (fraught with problems of finance and
implementation), food aid until the next
harvest, agricultural assets and inputs for
next season’s cultivation, and household
items including fuel, livestock restocking,
and perhaps cash (if press accounts are
correct that departing Kosovars have been
stripped of their money and valuables). If
early return proves infeasible, contingency
planning for winterizing the camps should
also be undertaken immediately.
3. Agenda for negotiating settle￾ment. Economic dimensions should constitute
a key element of future negotiations to settle
the conflict.
Conflict-settlement negotiations and
arrangements should address not only
political and security dimensions but also
economic implications, with due attention
given to the policy framework and man￾agement requirements. Otherwise, reha￾bilitation may be impeded. Development
was delayed and opportunities missed in
Cambodia because the interim UN admin￾istration interpreted its mandate—against
the advice of its economics unit—as limit￾ing its authority in economic development;
the intent was to leave such decisions to
the successor national authorities yet to be
established. The economic recovery of the
resettled Kosovar population should not
be delayed by making the interim inter￾national administration’s economic scope
too restrictive.
4. Economywide distortions. Moni￾tor the effects of external interventions to de￾tect inflationary and other economywide dis￾tortions.
Local expenditures made by large
numbers of international military, civilian,
and NGO personnel can have major posi￾tive and negative effects on fragile crisis
and postconflict economies. Such expen￾ditures can help stimulate the recovery of
local production and service sectors, but
they can also have inflationary and other
adverse effects. For example, they can af￾fect local housing markets, wage levels of
scarce local professional and technical per￾sonnel, and wages and perks (including
supplements) of civil servants seconded as
aid-project staff. Well-positioned officials
and other elite may capture much of this
expenditure, producing new, large income
disparities and consequent resentment.
The obverse may also be a problem: eco￾nomic downturn and employment loss
when stabilization allows major reduc￾D4
Annex D: Implications for the Kosovo Crisis
tions in external military and other per￾sonnel. Monitoring such effects should be
initiated right from the start.
5. NGO experience. Benefit from the ex￾perience of nongovernmental organizations.
NGOs play a major role in complex hu￾manitarian emergencies, and lessons from
their experience should be identified, es￾pecially in Bosnia and Croatia. Among
those lessons:
a. NGOs emerging in response to the
unprecedented availability of funds for
civil society often have only shallow ex￾perience; they should be encouraged with
small grants for an initial testing period.
b. Technical assistance is often essen￾tial; it can bring even small NGOs to the
point of effective management.
c. Most Bosnian and Croatian NGOs
were founded and managed by women—
teachers, mental health workers—re￾sponding to the emergency need to assist
displaced families and then to promote
resettlement, normalization, and commu￾nity reconciliation.
d. Youth and women often appear
more ready than adult males to reconcile
across ethnic lines. If this is also true in
Kosovo (assuming postconflict Kosovo is
not monoethnic), the bias toward reach￾ing youth and women and neglecting
adult males needs correction for reconcili￾ation to work.
e. NGOs dedicated to specific benefi￾ciaries and discrete activities (traumatized
women, youth voluntarism, sports, fam￾ily therapy, interethnic reconciliation, cul￾tural revival) should not be pressed into
new activities in which they lack exper￾tise (e.g., microenterprise) merely because
donor priorities have shifted. Instead, do￾nors should encourage the creation of
groups of NGOs (“strategic” rather than ad
hoc) with complementary capabilities and
objectives suitable for the array of prob￾lems that need attention in specific com￾munities.
f. Financial support for local NGOs
from the ethnic Albanian Diaspora can fa￾cilitate the gradual phaseout of aid depen￾dence, but political capture of such remit￾tances can create disillusionment and
choke off these funds, as demonstrated in
Croatia.
g. Local staffs of the International Res￾cue Committee and other strong NGOs in
Bosnia and Croatia could be valuable re￾sources for aid-funded projects utilizing
(or assisting) Kosovar (or other) local
NGOs. For example, the experience of agen￾cies helping reconstruct housing in Bosnia
(such as Mercy Corps International) also
may be valuable in Kosovo. Finally,
Kosovar NGOs that may be intact in refu￾gee status, outside Kosovo, should be sup￾ported to ensure their survival pending
return.
6. Safety nets. Anticipate a need for aid￾ing widow-headed households.
D5
Complex Humanitarian Emergencies and USAID’s Humanitarian Response
When the fate of the large number of
missing Kosovar males becomes known,
Kosovo may join the ranks of other coun￾tries that have emerged from complex
emergencies with tragically distorted de￾mography. A major loss of adult males
could create a severely disadvantaged
group of widow-headed households, es￾pecially for Kosovo’s large rural popula￾tion. The need for effective safety nets and
the need to restore such households to eco￾nomic viability should be anticipated.
Other disadvantaged groups needing tai￾lored support may include the elderly,
widower heads of households, single
mothers, and the disabled.
7. Employment-oriented training.
Train people in skills for which there is em￾ployment demand and link training with
startup capital.
Recovery activities need to match
postconflict economic realities. Job train￾ing for youth, women, and demobilized
soldiers (who have often turned to desta￾bilizing criminal activity when they re￾mained unemployed) is virtually fruitless
if unemployment in the depressed
economy remains high. Although oppor￾tunities for self-employment in microen￾terprise may also be limited, they are likely
to be enhanced if the training is linked to
the provision of seed capital.
8. Commodity procurement. Maxi￾mize regional procurement to help stimulate
economic recovery.
Purchasing aid-funded goods and
services internally and from the affected
neighboring countries (with an eye to the
possible price effects, as noted above) can
contribute to regional employment, eco￾nomic recovery, and the restoration of
cross-border economic relations. This may
require flexibility in the application of pro￾curement regulations in the affected coun￾tries.
9. High-level technocrat training.
Training technocrats to manage the
postconflict economic transition should begin
immediately.
Reconstruction and development in
Kosovo (under non-Yugoslav administra￾tion) is likely to be associated with a new
set of economic institutions and rules of
the game that mirror Western European
norms. As with many complex humani￾tarian emergencies, overall macroeco￾nomic and sectoral policy formulation
during recovery-cum-transition is likely to
require broad conceptual and manage￾ment skills. Those skills are likely to be in
short supply, especially if the preconflict
professional and intellectual cadre have
been targeted for deliberate elimination.
Although Kosovo’s civil society had
developed considerable institutional expe￾rience (mainly in social and cultural sec￾tors) during the autonomy period of 1974–
89, the fate of many civil society leaders is
unknown. If possible these cadre should
be found, perhaps among the refugees,
and readied for return to Kosovo to help
D6
Annex D: Implications for the Kosovo Crisis
revive the social infrastructure and prepare
for economic recovery. A balance will be
needed between foreign implementing
agencies and contractors on the one hand
and Kosovar organizations on the other.
Otherwise, the recovery may be jeopar￾dized either by overburdening the
Kosovars or by sinking them with over￾bearing disregard.
Political Effects
1. Potential political consequences.
Monitor emergency aid distribution for pos￾sible unintended and undesirable political con￾sequences.
Emergency aid (especially food com￾modities) are a source of power. A flawed
distribution process can have unintended
and undesirable political consequences. In
several cases (especially Rwanda), control
over final food distribution reinforced the
power of local authorities or factions,
strengthened their relative position, and
facilitated their self-aggrandizing, often
abusive, behavior toward the intended
noncombatant beneficiaries. Conse￾quences of this sort would not be surpris￾ing under prevailing conditions in Kosovo
and neighboring countries. Reconstruction
aid in the Balkans will far exceed emer￾gency food aid in amount and potential
political consequences. It also will need
careful monitoring.
2. Peace conditionality. Use aid as le￾verage to enforce adherence to the peace accords
and responsible governance.
In an evaluation of its own
postconflict experience, the World Bank
asserted that donors could have and
should have exerted more forcefully the
leverage they had during the initial
postconflict period. This is typically a pe￾riod of almost total dependence on exter￾nal financing and security. The study re￾fers to governance functions within the
normal scope of the Bank’s mandate (e.g.,
fiscal practices). The point applies as well,
though, to the notion of peace condition￾ality—that is, making provision (or with￾holding) of aid flows dependant on local
authorities’ adhering to the political com￾mitments in the peace accords (or compa￾rable instruments). Lack of donor consis￾tency in this regard can undermine such
potential (Cambodia), which reinforces the
need to design an effective coordination
structure. The Bosnian experience with
peace conditionality is perhaps the most
pertinent for postconflict Kosovo.
3. Regional destabilization. Non￾project aid for budget support can help front￾line states maintain domestic stability.
There is concern that the Kosovo cri￾sis may destabilize Albania and Mace￾donia. The presence of Kosovar refugees
and the burdens they impose jeopardize
the ability of the two governments to con￾tinue to finance and sustain precrisis lev￾els of service delivery to their citizens. Aid
in the form of nonproject budget support
could help them sustain domestic budget
outlays. It could also help diminish the po￾tential for ethnic polarization in
D7
D8 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
Macedonia (e.g., by enabling the govern￾ment to maintain pensions and civil ser￾vice salaries) and the deepening of the so￾called left–right polarization in Albanian
politics. Fast-disbursing aid flows may
also be critical for sustaining the positions
of these governments regarding the con￾flict itself, a political point beyond the
scope of this note. Understandings and
commitments regarding allocation of the
local currency counterpart should be ex￾plicit and should be monitored to ensure
compliance. Disbursements in tranches
should be considered to encourage com￾pliance.
4. Resentment of refugees. To avoid
resentment, keep refugee support standards
modest in relation to host-population stan￾dards.
Political problems have arisen in host
countries when encampment extends for
some considerable period. This may well
be the case with the Kosovar refugees
owing to mines, housing destruction, and
other problems. (One recent estimate as￾sumed 3–5 years, security considerations
aside.) Evaluations have stressed that refu￾gees should not be supported at a stan￾dard that evokes resentment among the
host population. Resentment can also stem
from refugees competing for scarce local
employment. After resettlement, resent￾ment could arise if country allocations of
reconstruction aid are perceived as unfair
or unjust. Aid to refugees should therefore
conform with regional living standards.
Bibliography
Africa Rights. 1998. Rwanda: The Insurgency
in the Northwest. London: Africa Rights.
Anderson, Mary B. 1996. Do No Harm: Sup￾porting Local Capacities for Peace Through
Aid. Cambridge: Collaborative for De￾velopment Action.
Apthorpe, Raymond. 1997. “Some Relief
from Development: Humanitarian
Emergency Aid in the Horn of Africa
(Including Sudan), Rwanda, and
Liberia.” The European Journal of Devel￾opment Research, vol. 9, no. 2 (Decem￾ber): 83–106.
Borton, John. 1994. “Assessing the Impact
and Effectiveness of Relief Assistance.”
Paper prepared for the 1995 World
Disasters Report.
Borton, John; and Joanna Macrea. 1997.
“Evaluation Synthesis of Emergency
Aid.” London: Department for Inter￾national Development.
Brandt, Donald P. 1995. “War Work: Hu￾manitarian Assistance in Complex
Human Emergencies.” Journal of Hu￾manitarian Assistance (23 October).
Brown, D.G.; P.W. Bartholomew; J–C.
Lucas; and M. Jolivert. 1996. “Final
Evaluation: USAID/GOH-funded JOBS
Creation Program.” Prepared for Pan
American Development Foundation
and USAID/Haiti.
Buttari, Juan J. 1997. “Haiti: Intervention,
Reform, and Politics.” Draft report.
Chandler, Dennis; and others. 1996.
“Evaluation of the Haiti Communal
Governance Program.” Management
Systems International. Washington:
USAID.
Collier, Paul; and Anke Hoeffler. 1998. “On
Economic Causes of Civil War.” Oxford
Economic Papers 50. Oxford University
Press.
de Soysa, Indra; and Nils Petter Gleditsch.
“To Cultivate Peace: Agriculture in a
World of Conflict.” Oslo: International
Peace Research Institute.
Complex Humanitarian Emergencies and USAID’s Humanitarian Response
de Waal, Alex. 1993. “War and Famine in
Africa.” IDS Bulletin, vol. 24, no. 4.
Duffield, Mark. 1998. “Aid Policy and
Post-Modern Conflict: A Critical Re￾view.” School of Public Policy. Bir￾mingham: University of Birmingham.
Dupuy, Alex. 1997. Haiti in the New World
Order: The Limits of the Democratic Revo￾lution. Boulder, Colo.: Westview Press.
Government Accounting Office. 1985.
“The United States’ Response to the
Ethiopian Food Crisis.” Washington.
———. 1986. “Famine in Africa: Improv￾ing Emergency Food Relief Programs.”
Washington.
Hallam, Alistair. 1998. “Evaluating Hu￾manitarian Assistance Programmes.”
Prepared for the DAC Expert Group on
Aid Evaluation and reviewed by the
DAC Task Force on Conflict, Peace, and
Development Cooperation, February
2–3.
Hansch, Steve; and others 1994. “Lives
Lost, Lives Saved: Excess Mortality
and the Impact of Health Interventions
in the Somalia Emergency.” Washing￾ton: Refugee Policy Group.
Hill, Richard. 1997. “Post–Cold War Re￾construction Programs: Observations
From the Field on the Sword and the
Ploughshare.” Paper presented at a
Conference on Urban Triumph or Ur￾ban Disaster?: Dilemmas of Contem￾porary Post-War Reconstruction. Cam￾bridge, Mass.
Ianotti, L. 1997. “Nutrition in Haiti: Con￾straints and Solutions.” Washington:
Inter-American Development Bank.
Inter-American Development Bank. 1994.
“Emergency Economic Recovery Pro￾gram: Haiti.” Report of the Joint As￾sessment Mission. Washington.
Joint Evaluation of Emergency Assistance
to Rwanda. 1996. “Early Warning and
Conflict Management.” Study 2. The
International Response to Conflict and
Genocide: Lessons from the Rwanda Ex￾perience.
———. 1996. “Humanitarian Aid and Ef￾fects.” Study 3. The International Re￾sponse to Conflict and Genocide: Lessons
from the Rwanda Experience.
Kitay, Michael. N.d. “Land Tenure Issues
in Postconflict Countries.” Washing￾ton: USAID.
Kleist, Ted. See Netherlands Development
Cooperation. 1994.
Lautze, Sue; Bruce D. Jones; and Mark
Duffield. 1998. Strategic Humanitarian
Coordination in the Great Lakes Region
1996–97. Office for the Coordination of
Humanitarian Affairs. New York:
United Nations.
Biblio2
Bibliography
Lieberson, Joseph; and others. 1999. “Pro￾viding Emergency Aid to Mozam￾bique.” Center for Development Infor￾mation and Evaluation. Washington:
USAID.
Maguire, Robert; and others. 1996. Haiti
Held Hostage: International Responses to
the Quest for Nationhood, 1986 to 1996.
Occasional Paper no. 23. Providence:
Watson Institute.
McClelland, Donald G.; and others. 1999.
“Providing Emergency Aid to Haiti.”
Center for Development Information
and Evaluation. Washington: USAID.
McHugh, Heather S.; and Joshua M.
Kaufman. N.d. “A Summary of Prin￾cipal Lessons Learned from USAID Ex￾perience in Humanitarian Assistance.”
Washington: USAID.
Messer, Ellen; Marc J. Cohen; and Jashinta
D’Costa. 1998. “Food from Peace:
Breaking the Links Between Conflict
and Hunger.” Washington: Interna￾tional Food Policy Research Institute.
Miller, John M. 1997. “USAID/Washington
and Field Collaboration in the Use of
Central Resources.” Washington.
Minear, Larry. 1994a. “Humanitarian Ac￾tion in the Former Yugoslavia: The
UN’s Role 1991–93.” Providence R.I.:
Watson Institute.
———. 1994b. “The International Relief
System: A Critical Review.”
———. 1998. “Learning to Learn.” Discus￾sion Paper prepared for a seminar on
Lessons Learned in Humanitarian Co￾ordination. Office for the Coordination
of Humanitarian Affairs and the Min￾istry of Foreign Affairs of Sweden.
Nafziger, E. Wayne. 1996. “The Econom￾ics of Complex Humanitarian Emer￾gencies: Preliminary Approaches and
Findings.” UNU/WIDER.
Naval, G. 1995. “Evaluation of the Crisis
Program of the Catholic Relief Services
in Haiti.” Final report.
Netherlands Development Cooperation.
1994. Humanitarian Aid to Somalia.
Evaluation Report directed by Ted
Kleist. Operations Review Unit.
Organization for Economic Cooperation
and Development. 1999. “Guidance for
Evaluating Humanitarian Assistance
in Complex Emergencies.” Develop￾ment Assistance Committee. Paris.
Organization for Economic Cooperation
and Development. DAC/o Database
(http://www.oecd.org/dac/).
Pan American Development Foundation.
N.d. “Jobs Creation Project: Final Re￾port.” Washington.
Biblio3
Biblio4 Complex Humanitarian Emergencies and USAID’s Humanitarian Response
Pasic, Amir; and Thomas G. Weiss. 1997.
“Humanitarian Recognition in the
Former Yugoslavia: The Limits of Non￾State Politics.” Security Studies 7, no. 1
(autumn 1997): 194–227.
Prendergast, John; and Colin Scott. 1996.
“Aid with Integrity: Avoiding the Po￾tential of Humanitarian Aid to Sustain
Conflict: A Strategy for USAID/BHR/
OFDA in Complex Emergencies.”
Occasional Paper no. 2. Washington.
Prunier, Gérard. 1995. The Rwanda Crisis:
History of a Genocide. New York: Colum￾bia University Press.
Renison, Bill; and others. 2000. “Humani￾tarian Assistance in Rwanda: Hard
Choices, No Exits.” Draft Report.
Washington: USAID.
Sen, Amartya K. 1993. “The Political
Economy of Hunger.” Overcoming Glo￾bal Hunger. Ismail Serageldin and
Pierre Landell–Mills, eds. Proceedings
of a Conference on Actions to Reduce
Hunger Worldwide. Washington.
UNHCR/WFP. 1998. “Joint Evaluation of
Emergency Food Assistance to Return￾ees, Refugees, Displaced Persons, and
Other War-affected Populations in
Bosnia and Herzegovina.” Rome.
USAID. Congressional Presentation, FY 1990–
FY 2000. Various years. Washington.
———. 1995. “Survey Report on Losses of
Public Law 480 Commodities in Haiti
During the Political Transition Period
September 15, 1994, to January 15,
1995.” Audit Report No. 1–521–95–008.
———. 1998a. Agency Performance Report,
1997. Center for Development Infor￾mation and Evaluation. Washington.
———. 1998b. “Results Review and Re￾source Request: Fiscal Year 2000.” Bu￾reau for Humanitarian Response. Pre￾liminary draft.
———. 1999. Agency Performance Report,
1998. Center for Development Infor￾mation and Evaluation. Washington.
USAID/Haiti. 1992–96. “USAID Monitoring
Report” 1–5. Port-au-Prince.
USAID/Liberia. 1998. “Fiscal Year 2000 Re￾sults Review and Resource Request.”
Bureau for Humanitarian Response.
U.S. Mission to the United Nations. 1997.
“Global Humanitarian Emergencies,
1997.”
———. 1998. “Global Humanitarian
Emergencies, 1998–99.”
Uvin, Peter. 1998. Aiding Violence: The De￾velopment Enterprise in Rwanda. West
Hartford: Kumarian Press.
Bibliography
World Bank. 1998a. “Haiti: The Challenges
of Poverty Reduction,” vols. I and II.
Technical papers. Washington.
———. 1998b. “Haiti: The World Bank’s
Experience With Postconflict Recon￾struction,” vol. v. Desk Reviews of
Cambodia, Eritrea, Haiti, Lebanon,
Rwanda, and Sri Lanka. Report no.
17769. Washington.
Biblio5
———. 1998c. “Rwanda: Poverty Update.”
Washington.
World Food Program. 1998. “The Food Aid
Monitor.” 1997 Food Aid Flows. Spe￾cial issue. Rome: International Food
Aid Information System.
U.S. AGENCY FOR INTERNATIONAL DEVELOPMENT