As part of its information gathering, the Committee on the Effective Use of Data, Methodologies, and Technologies to Estimate Subnational Populations at Risk held a workshop on March 13-14, 2006, at the National Academies Keck Center in Washington, D.C. As part of this workshop, the committee requested voluntary technical paper contributions from each panelist related to the topic of his or her panel session (see Appendix D for workshop agenda). These papers were intended to supplement the information the panelists presented during the workshop and served as references for the committee’s use during the course of the study.
The technical papers that the committee received are presented here for reference, along with biographical sketches of the contributing authors. The statements, opinions, findings, and conclusions or recommendations made in the authored papers are those of the individual authors and do not necessarily represent positions of the committee, the National Academies, or the sponsors. Mention of trade names or commercial products does not constitute their endorsement by the U.S. government.
LIST OF TECHNICAL PAPERS
Assessing the Strengths and Weaknesses of Existing Data for Estimating Subnational Populations at Risk from Disasters Associated with Natural Hazards
Mark Pelling, King’s College, London
In Harm’s Way: Estimating Populations at Risk
Jerome E. Dobson, University of Kansas
Organizational Impediments to Estimating Populations, and Acquiring, Accessing, and Using Population Data
John A. Kelmelis, U.S Geological Survey and Department of State, Washington, D.C.
Political Geography and Emergency Relief
Wm. Glen Lauber, National Geospatial-Intelligence Agency, Washington, D.C.
Identify Ways in Which Subnational Demographic and Geographic Data and Tools Could Be Used to Help Decision Makers Provide Useful Information to Populations at Risk
Shannon Doocy, Johns Hopkins University Center for Refugee and Disaster Response, Baltimore, Maryland
Cognitive and Institutional Limits on Collecting and Processing Data on Populations at Risk: Preliminary Reflections on Southern African Responses to Displacement
Loren B. Landau, Wits University’s Forced Migration Studies Programme, Johannesburg, South Africa
Strengths and Limitations of Information and Data Analysis in Responding to Crisis in Mali
Mamadou Kani Konaté, CAREF, Bamako, Mali
ASSESSING THE STRENGTHS AND WEAKNESSES OF EXISTING DATA FOR ESTIMATING SUBNATIONAL POPULATIONS AT RISK FROM DISASTERS ASSOCIATED WITH NATURAL HAZARDS
Mark Pelling, King’s College, London
This paper offers a review of international disaster databases. It draws on Pelling (2005a,b, 2006). Four publicly accessible, international databases are described, and challenges facing the use of these data for subnational analysis are assessed.
THE INTERNATIONAL INSTITUTIONAL ARCHITECTURE FOR COLLECTING DATA ON DISASTER LOSS
Table E-1 presents a summary of the characteristics of four disaster loss databases: EM-DAT, NatCat, Sigma, and DesInventar. These are discussed in turn below.
TABLE E-1 International Disaster Databases
University/NGO (La Red)
Hazard types with which disasters are associated
Natural and technological
Natural (rapid onset)
Natural and technological
Natural and technological
Criteria for disaster entry
At least 10 deaths or 100 affected, or state of emergency, or call for international assistance
1980-present: any property damage and person injured or killed. Before 1980, only “major” events
At least 20 deaths, or insured losses of at least U.S. $15.1 million or total loss of U.S. $74.9 milliona
Any social loss
Principal data source
Humanitarian agencies, governments, international media
Insurers, international media, supported by site visits
Insurers, international media
Local/ national media, agency and government reports
Time for new entry
Around three weeks
Around one week to one month
1900-present, with goodaccuracy from 1980
Good accuracy from 1979
Data fields (not all data are available for every disaster event)b
Mortality, injured, homeless, total people affected, estimated economic loss
Insured and economic losses
Includes data on human losses for large-scale disasters
Economic loss, mortality, missing, injured, and homeless are also noted for large-scale disasters
Mortality; injured or missing victims; affected, destroyed, and affected houses; evacuated areas; roads, education centers, or livestock lost; economic losses
NOTE: NGO = nongovernmental organization.
aUsing 2004 U.S. dollars, monetary values annually adjusted.
bNo data set covers ecological loss.
THE EMERGENCY DISASTERS DATABASE
The Emergency Disasters Data Base (EM-DAT), managed by the Centre for Research on the Epidemiology of Disasters (CRED) at the Université Catholique de Louvain, Brussels, is the most complete, internationally accessible, public database on disaster loss at the national scale. It produces estimates of human and economic losses from disasters. The database includes relatively small disasters for which reliable data exist. The main aim of EM-DAT is to support research, and great effort is put into the verification of input data; nonetheless, many challenges to data quality and completeness remain. The verification process means that data often are not posted on EM-DAT until four weeks after an event. Even at this stage the data are subject to change over periods of 12 months or more when, for example, people described initially as missing are re-categorized as dead.
The contribution of EM-DAT is primarily constrained by a lack of systematic and standardized local and national disaster data collection. A particular problem from the perspective of this workshop is the lack of spatial specificity found in international reports upon which EM-DAT builds much of its database. Consequently, EM-DAT catalogues events by country, making it difficult to identify subnational patterns of disaster loss, with changes in international borders complicating historical analysis. Subnational loss patterns can be deduced as a proportion of national population or population exposed to a particular hazard type, for example, but the results are rough estimates at best.
REINSURANCE DATABASES: NATCAT AND SIGMA
NatCat is managed by Munich Reinsurance (MunichRe). Beyond holding data on insured loss, MunichRe has built a methodology for calculating total economic losses for large-scale disasters (excluding those associated with drought) from this database. Economic losses are calculated based on insured loss, with weighting for the country affected and the natural hazard trigger type. Economic losses include damage to physical assets and infrastructure but do not incorporate losses from the interruption of productive, distributive, and marketing activities and secondary effects on the national economy, all of which are difficult to calculate. The accuracy of the methodology is verified by comparison with final loss estimates from the field. The georeferencing of individual buildings means that very high resolution loss studies can be undertaken, with MunichRe having also developed an urban risk index.
Sigma, managed by Swiss Reinsurance (SwissRe), presents information on insured property losses and on economic and human loss due to natural
and technological disasters in its annual loss reports. Sigma uses the event (which may include multiple countries) as the basis for each entry. This contrasts with the use of country categorization in NatCat and EM-DAT.
Both insurance databases provide only limited information on countries and populations with low insurance density. This reduces coverage for Africa and also for Asia and Latin America. In particular, rural areas lack comprehensive data, and losses associated with drought are not covered. However, the inclusion of data from sources beyond insurance reports enables the databases to provide some information even on these areas.
DesInventar is of special interest to this workshop because it is the only internationally accessible database that holds locally georeferenced data for human and economic loss. A comparison of loss data of EM-DAT and DesInventar has shown differences. For example, DesInventar tends to record higher numbers of people affected by disaster (WG3, 2002) than does EM-DAT. Differences in data have been explained by alternative data collection approaches.
DesInventar is managed by a regional coalition of academic and non-governmental actors and covers 16 countries in Latin America and the Caribbean. DesInventar’s focus on local disasters introduces a number of specific challenges for data quality:
The media is a prominent source of information; not only is there no systematic method for media reporting on losses, but reliability is debatable.
An aim of DesInventar is to collect information on secondary impacts and losses to infrastructure, but it is found that this information is unevenly reported, even locally.
Sometimes DesInventar has to draw on national sources of data, and in these cases it is very difficult to disaggregate to determine the local distribution of losses.
Subnational DesInventar databases also exist for individual states in the U.S., Brazil, Colombia, South Africa, and India.
Lack of Standardized and Systematic Subnational Data Collection
The absence of standard guidelines for local disaster loss data collection is compounded in most countries by an ad hoc system of data collection
by local media, government, or civil society groups. Local data are collated and fed to international databases by intermediaries. Not only do intermediaries not have a standard set of definitions to order data, but they might be tempted to exaggerate or suppress data for professional, political, or economic advantage (DFID, 2004).
Mortality is the cleanest indicator of disaster loss. Even here, however, the distinction between deaths and people missing creates uncertainty, with some countries requiring that people be declared dead after they have been missing for 12 months. Wide variance in reports of mortality is common. For example, in the Bam earthquake, mortality was put at 43,200 by the UN Office for the Coordination of Humanitarian Affairs (OCHA), at 31,884 by the French Agency Press, and at 26,796 by the International Federal of Red Cross and Red Crescent Societies (IFRC).
Data are most incomplete for economic losses. Over the past three decades, economic losses were reported for less than 30 percent of all natural disasters, with the least data for developing countries (Pelling, 2005b). Scarcity of data is compounded by the lack of a standardized methodology for reporting economic losses. Generally it is only direct losses that are reported with no breakdown of losses by sector. Having few international data sources makes verification difficult. In Bam, economic loss estimates ranged from U.S. $1 billion by SwissRe to U.S. $32.7 million from the U.S. Geological Survey. A possible model for a standard methodology for reporting losses has been developed by the Economic Commission for Latin America and the Caribbean (see http://www.proventionconsortium.org/toolkit.htm); this model is becoming widely used.
When disasters are associated with hazards that impact more than one country, such as the Indian Ocean tsunami or Hurricane Mitch, potential problems with double counting in data sets may also occur. Losses might be recorded for individual countries and for the event as a whole. A harder challenge to overcome is a product of disaster cascades—this happens when an initial hazard triggers a secondary event, for example when landslides follow seismic activity or flooding. This challenge may be addressed partly by an agreement among agencies to use a common and unique Global Identifier Number (GLIDE) for each event. GLIDE has been available since 2004.
Defining Hazards and Distinguishing Events
Strict physical definitions for hazards can be proposed so that their distribution and frequency might be mapped. Associating hazards with recorded loss data to measure severity of impact is not as easy to do. In each database and for each hazard type, the quality of attributed loss data is compromised by other intervening and overlapping causes of loss.
Drought hazard and associated loss present the greatest challenge (drought is also the hazard type associated with most human loss). A common definition of drought does not exist, and it may be defined using a range of meteorological, hydrological, or agricultural variables. Thus, no common rubric is used to draw spatial and temporal limits around drought events, creating challenges for the comparative analysis of drought impacts. The slow-onset nature of drought makes it very difficult to separate the relative contribution of associated environmental and human factors, such as soil loss, armed conflict, or HIV/AIDS, to loss data. Overall this can make it difficult to judge whether a drought period is a cause, product, or context for reported losses.
The 20-year time span of reliable information from loss databases makes it difficult to use past losses as a basis for assessing current vulnerability and risk status for low-frequency, high-impact hazard types such as volcanic eruptions or tsunamis.
Projecting Past Vulnerabilities into the Future
Care should be taken when using loss data as input to vulnerability and risk indexing. Loss data speak toward past events. While strong historical correlations between events may be apparent, this should not lead to an assumption that trend lines will continue similarly into the future.
This assumption becomes more problematic at finer resolution where development pressures, such as rapid urbanization and local environmental changes linked to global climate change, have the potential to alter radically the local distributions of population, wealth, hazard, and vulnerability over a short time period relative to hazard frequency. It is also possible that losses during past disasters will lead to local learning and the building of resilience, rather than the continuation of vulnerability, so that past impacts might locally be associated with future security rather than vulnerability. Regular assessments accompanied by contextual analysis of pressures shaping hazard, vulnerability, and disaster risk management can help overcome this challenge for measuring local risk.
Measuring Human Vulnerability
Mortality is arguably the most reliable comparative indicator of local human loss at the global scale. Data on people affected, injured, or made homeless are far less reliable. Reliance on mortality gives statistical rigor but limits policy impact. This can be seen most clearly in drought events where complex interactions between drought, political violence, chronic disease, and economic poverty can make it very difficult to ascribe mortality. It would be more useful for policy makers if the impacts of a drought
on livelihood could be measured, but this can only be done on the ground and at the local level.
Where economic impacts are to be used to indicate vulnerability or risk, three challenges should be considered. First, economic loss data do not routinely include long-term economic impacts (sometimes called secondary losses) including changes in national balance of payments, international debt, or fluctuating levels of employment or price inflation in the years following disaster. Second, the destruction or erosion of household livelihoods is not accounted for in economic loss estimates. Third, a focus on gross domestic product (GDP) excludes the informal sector from analyses. Finally, economic loss as a proportion of national wealth as well as in absolute terms should be measured. Low-income countries have fewer economic assets to lose, but damage can be felt as a high proportion of national or local wealth.
Building Multihazard Measurements
The challenges discussed above are magnified when trying to build multihazard vulnerability or risk indicators. A particular challenge is how to combine hazards measured using different metrics, for example, combining the hazardous nature of drought (measured by frequency) and landslides (measured by probability).
The greatest need is for standardized collection and systematic collation of local disaster data. This should include agreed protocols for the start and end dates of disasters, for their georeferencing, for distinguishing between cascading hazards, and for measuring impacts, in particular economic impacts, which should include secondary and livelihood losses and ecological impacts. In addition to this, the quality of international disaster databases would be greatly improved by guidelines for standardization in disaster event and loss reporting among intermediaries.
Subnational baselines of social, economic, and ecological status would enhance the accuracy of disaster loss measurement and allow the verification of vulnerability and risk assessments through disaster impact. This will be enabled where the local disaster loss data are compatible with aggregation-disaggregation to the finest level of resolution routinely available for social statistics, such as a census tract. Developing baselines goes beyond the capacity of the disasters community, but is an agenda in which disasters data managers could usefully participate. The recommendations presented here in part respond to the Hyogo Framework for Action, which endorsed the
need for more work on disaster data and analysis to feed into disaster risk reduction.
DFID (Department for International Development), 2004. Disaster Risk Reduction: A Development Concern. London: DFID.
Pelling, M., 2005a. Visions of Risk: A Review of International Indicators of Disaster Risk and Its Management. Geneva: United Nations Development Programme-Bureau for Crisis Prevention and Recovery.
Pelling, M., 2005b. Disaster Data: Building a Foundation for Disaster Risk Reduction. In World Disasters Report. Geneva: International Federation of Red Cross and Red Crescent Societies, pp. 172-180.
Pelling, M., 2006. Measuring vulnerability to urban natural disaster risk. Open House International, special issue on Managing Urban Disasters 31(1):125-132.
WG3 (Working Group Three), 2002. Comparative Analysis of Disaster Databases. Geneva: International Strategy for Disaster Reduction, Final Report submitted to Working Group 3, November 30.
Mark Pelling is a Reader in Human Geography, King’s College London. Before this he lectured at the University of Liverpool and the University of Guyana. His research interests are in assessing and understanding human vulnerability, resilience, and adaptation to natural hazards including those associated with climate change. Recent publications include The Vulnerability of Cities: Natural Disaster and Social Resilience (Earthscan) and Natural Disaster and Development in a Globalizing World (Routledge). He has acted as a consultant for the United Nations Development Programme-Bureau for Crisis Prevention and Recovery, the United Nations-Human Settlements Program (UN-HABITAT), Tearfund, the UK Department for International Development, and the UK Environment Agency. He is chair of the Climate Change Research Group of the Royal Geographical Society and a member of the ProVention Consortium Advisory Committee.
IN HARM’S WAY: ESTIMATING POPULATIONS AT RISK
Jerome E. Dobson, University of Kansas
My charge is to assess the strengths and weaknesses of existing data, methods, and tools for population analysis at subnational levels. First, I ask, what is the intended purpose and what are the risks to be addressed? What are the scopes and resolutions of risks against which each database must be judged? What spatial and temporal resolutions are required to meet each data need, and what methods are required to produce appropriate databases? I conclude with recommendations on policy, institutional frameworks, methods, and delivery mechanisms.
A narrow definition of the term “populations at risk” implies that the end use will be emergency management and humanitarian relief. Usually, such risks are categorized as natural disasters, technological accidents, epidemics, and wars including regional conflicts and terrorism. A broad definition would cover all types of risk, including economic crises, environmental degradation, disease outbreaks, and social upheavals. It would be in the national and global interest for the committee to adopt a broad definition and address a wide range of risks.
SCOPE OF RISK
Risks come in all sizes, shapes, and durations. Consider, for example, extreme weather events. Tornadoes, hurricanes, and climate change place whole regions at risk on a long-term basis, but they differ greatly in the spatial and temporal extent of individual incidents. Each tornado is highly precise in duration (minutes to hours) and spatial extent (from a single building to at most a long, narrow swath of buildings, fields, and forests). Each hurricane typically impacts a coastal strip several hundred kilometers long with high winds and ocean surges, and even its most violent impacts may extend a hundred kilometers or more. Climate changes cover vast territory on the order of whole regions, continents, or even the entire planet. Population data requirements thus differ greatly depending on the type of weather event. Hence, an ideal system for tornado response would require data resolutions on the order of individual buildings, or roughly 15 meters. An ideal system for hurricanes should support evacuation planning at least as precise as neighborhood (1 kilometer) or preferably block (90 meters). An ideal system for climate change, however, might suffice with census data typically available for countries, districts, and minor civil divisions, and no disaggregation would be required.
Likewise in war, conventional wisdom during the Cold War era was that population estimates would suffice at very coarse resolutions because blast effects would cover cities, say tens of kilometers, and radiation effects would cover hundreds of kilometers. The rise of terrorism in the 1990s sparked an immediate demand for more precise population data resolutions of 1 square kilometer or finer.
Time is crucial for most types of emergency response. To what precision can the timing of each incident be predicted? Tornadoes, volcanic eruptions, earthquakes, tsunamis, and terrorist attacks often happen with little
or no warning. Hurricanes form for days, but always with uncertainty as to where and when they will make landfall. In our lifetimes, remarkable advances have been made in forecasters’ abilities to predict when and where tornadoes and hurricanes will strike, but error bands remain broad. Wars, climate changes, economic crises, and environmental catastrophes form over years, but warning signs can be interpreted in many ways with regard to the timing and certainty of their most violent manifestations. Hence, it is essential to have suitable databases prepared ahead of time and updated as often as possible.
In contrast, some incidents can be predicted with absolute certainty. If, for example, an asteroid was hurtling toward Earth, its time of impact could be predicted to the minute. That is a rare event, of course, but the same scenario has occurred several times with man-made satellites descending from orbit. Similarly predictable categories include military air strikes by allied forces and certain pre-announced or pre-detected terrorist attacks. Many other types of events are predictable after the incident has begun. A tsunami’s arrival, for instance, can be predicted with great certainty once a given surge is observed to be propagating across a given sea.
The most important determination is whether an incident will occur at night or during the day. Census counts measure where people sleep at night or, at least, on most nights. Substantial analysis and modeling are required to estimate whereabouts in daytime. Thus, census counts are preferred when the incident occurs at night, while modeled population estimates, even with broad error bands, are preferred when the incident occurs in daytime. Twenty-four hour ambient population densities are adequate, and may even be preferred, when the timing is random or unknown.
Adding to the complexity is the fact that many emergencies are combinations of large and small disasters. Paradoxically, for instance, floods often cause fires and may cause all sorts of other local disasters such as building collapses, chemical releases, power outages, and transportation accidents. Wars typically are complex emergencies consisting of disasters intended by one protagonist and unavoidable for the other. Ideally, population data would be updated in real-time as each disaster unfolds, but this would require an exceptional capability for data acquisition, analysis, computation, and dissemination that does not exist at present.
Terrain is a key factor for many types of disasters, which greatly increases the need for fine-resolution data for both population and elevation. Hurricanes, tsunamis, volcanoes, and the blast effects and airborne contaminants of terrorist attacks and industrial accidents are constrained by land surface elevation, slope, and relief. In the immediate aftermath of the 2004
tsunami in Southeast Asia, for example, humanitarian agencies desperately sought estimates of the number of victims in each narrow coastal strip, and elevation was crucial. For reference, consider that a 10-meter surge on an 8 percent slope, typical of many residential areas, would extend a mere 125 meters inland. The best available data, however, were at 30-arc-second resolution (1 square kilometer at the equator and getting finer toward the poles), suitable for flat shores where the land surface slope is 1 percent or less. Elsewhere, more spatially precise population data are needed. Even on flat shores, finer-resolution data are needed to ensure that population data can be intersected in three-dimensional space with geographic features such as streets and buildings. Not long ago, intersections of such precision were considered impractical for most professional applications; now they are expected by the public due to the advent of GoogleEarth.
METHODS AND TECHNIQUES
Various cartographic and statistical techniques are available for smoothing, interpolating, and disaggregating census counts from native census geometries to subcensus polygons or cells. A great distinction lies in whether they are performed on population variables only or, conversely, involve one or more ancillary variables. For multivariable analyses, the standard cartographic technique is dasymetric interpolation, first proposed by John K. Wright of the American Geographical Society in 1936 (Wright, 1936). This technique is applicable to many geographic distributions, but Wright’s initial application was population. He started with choropleth maps of population density, but he reasoned that the uniform densities produced by this traditional cartographic technique masked certain known or knowable factors affecting the actual distribution. He first distinguished uninhabited from inhabited areas. He then weighted the inhabited areas by land use and settlement characteristics. Wright called his technique dasymetric mapping, but Chrisman (2002) later proposed that it should be considered more properly a form of areal interpolation. By implication, Tobler et al. (1995) were advocating dasymetric interpolation when they called for a “smart interpolation or co-Kriging” and suggested that global population density estimates can be improved in accuracy and precision by incorporating ancillary data such as location and size of towns and cities, roads, railroads, natural features, and nighttime lights.
The de facto world standard for estimating populations at risk in natural disasters, technological accidents, and wars is the LandScan Global Population Database. It comes in two resolutions, 30 arc-seconds for the whole Earth and 3 arc-seconds for much of the United States. In the in-
terest of full disclosure, I note that I led the development of LandScan at Oak Ridge National Laboratory (ORNL) from its inception in 1997 until I moved to the University of Kansas in 2001. I no longer have any connection with the project. Annual updates of LandScan Global are available free on the ORNL web site to government, education, and not-for-profit users, and a fee is charged for commercial licenses. LandScan Global is copyrighted. LandScan USA has not been released by the Department of Homeland Security. A patent application was submitted for LandScan USA after my departure, and it is pending. I am one of four patent applicants.
LandScan is ideal for certain purposes, but no existing database suits all. The utility of LandScan Global could be improved by finer spatial resolution, and some steps in that direction are underway. The land-cover database, for instance, has progressed from Advanced Very High Resolution Radiometer (1 kilometer) in 1998 and 2000 to MODerate Resolution Imaging Spectroradiometer (500 meters) in 2001 to Landsat Thematic Mapper (30 meters), which covered much of the world in 2004 and most of the world in 2005. LandScan Global continues to improve through increasing use of databases that are more “local” in scale, though often less “global” in coverage. Because of updates and improvements in the resolution of land cover and other input variables, population estimates within a province may appear to “move” about when compared to earlier versions of LandScan. Hence, users are warned not to make such temporal comparisons. If a village location is corrected or an urban boundary is adjusted through imagery analysis, people appear to “migrate” when, in reality, there was no spatial redistribution at all.
For many types of risks, LandScan could be improved by finer temporal resolution, foremost by distinguishing daytime and nighttime distributions. Progress in doing so for the entire globe will demand more ancillary databases at finer resolution than presently available. LandScan is designed for first responders, whose greatest need is for in situ populations in real time. It stops, however, at the onset of each incident and has no capability to monitor changes that occur during and immediately after an unfolding incident. LandScan has many actual and potential applications in social, economic, and environmental analyses, but it has no capability to project populations forward or backward. Caution must be taken in certain types of modeling to ensure that output variables, such as land cover, are independent of LandScan’s input variables.
The Center for International Earth Science Information Network’s (CIESIN) Gridded Population of the World (GPW) was originally developed to study land use and land cover in relation to population densities. It has been used to study the distribution of population density by altitude (Cohen and Small, 1998) and in biodiversity hotspots (Cincotta et al., 2000). The GPW was also used to estimate global populations at risk from volcanic hazards (Small and Naumann, 2001) and from coastal hazards (Small et
al., 2000; Nicholls and Small, 2002), though it clearly does not meet the resolutions needed for tsunamis. Each grid cell (2.5 minutes by 2.5 minutes) is 25 times larger than a LandScan Global cell, and 2,500 times larger than a LandScan USA cell. The principal difference between the two models is that GPW employs a cartographic interpolation of population data only, while LandScan employs a dasymetric interpolation of population data plus distance to roads, slope, land cover, and (formerly) nighttime lights. Nighttime lights are no longer used because they underestimate poor people, overestimate energy-rich people, and unrealistically spread urban populations through “overglow”—the spread of city lights to surrounding areas.
GPW version 3 projects populations out to 2015. Globally, GPW includes more administrative units, many of whose digital boundary files are precise enough for georegistering with GPW cells but not with the finer LandScan cells. GPW’s uniform distributions uninfluenced by other variables make it less precise and therefore less suitable for real-time emergency management and humanitarian relief applications. These same characteristics, however, coupled with its multi-temporal projections make it more suitable for visualizing provincial-level population changes through time and for many types of geographic analyses exploring causal relationships between population distribution and other geographic factors such as land cover, terrain, and transportation.
LandScan is approaching the limits of what can be accomplished with current global databases. An alternative approach developed by Dobson, Peplies, and Dunbar would create building occupancy coefficients that can be multiplied by building area once the location and type of buildings are known for any given disaster. This approach reduces the need for additional ancillary data, relying instead on fieldwork to collect building occupancy coefficients, which vary considerably by region and type of building, plus fieldwork or image interpretation to measure buildings. The approach has been tested through fieldwork in the United States, East Africa, northwestern South America, the Middle East, the Balkans, and East Asia.
It is technically feasible to conduct a co-verification analysis of building occupancy (field-based, bottom-up) against LandScan (census-based, top-down) at the finest level of individual LandScan cells. Both databases need verification and validation, and this can best be accomplished by comparing their results for a statistically valid sample in any given region. Each database could be a check against the other, but no such analysis has been done.
RECOMMENDATIONS TO THE FEDERAL GOVERNMENT
First, do no harm. Continue to fund current efforts at present levels or greater.
Assign principal responsibility for global population geography to a single federal agency that will then be its champion.
Promote and establish global population estimates and fine-resolution spatial distributions as an essential element of national security, economic security, emergency management, and humanitarian relief. Consciously tailor the collection of databases to provide appropriate spatial, temporal, and attribute characteristics for a wide range of applications.
Establish continuing programs for financial support of researchers in demography, population geography, and settlement geography. All three fields are essential to progress. Population estimation is now a labor of love practiced long term only by those who are willing to sacrifice for their science or the good of the nation. A principal goal of this committee should be to establish a regular basis for funding to advance the science, retain established expertise, and recruit new expertise into this vital effort.
Support the development of ancillary data that are essential to the improvement of population databases. Key examples include the following:
Administrative and census geometry: Accurate boundaries are required in every effort aimed at improving population geography. Boundary databases are numerous in cyberspace, but most are spatially inaccurate and dated. The United Nations International and Administrative Boundaries Group is making an effort in this area through its Second Administrative Level Boundaries (SALB) database, but much more work needs to be done. The boundaries produced by the National Geospatial-Intelligence Agency (NGA) are usually spatially accurate, but they are limited in coverage especially for higher-order boundaries.
Land use: Land use must be inferred from land cover, with additional fieldwork and field verification in every region of the world.
Land cover: At a minimum, satellite imagery must be classed by land cover type.
Nighttime lights: At present, nighttime lights are a by-product of meteorology programs, and the resulting spatial resolution (1 square kilometer) and temporal resolution (multiple observations aggregated for periods as long as two years) are far too coarse. Addressing populations at risk justifies a satellite sensor of its own at a spatial resolution approaching that available for land cover (30 meters) and a temporal resolution approaching real time.
Support fundamental research to determine what constitutes appropriate reference data for accuracy assessment and the fitness-for-use of population databases at various spatial and temporal resolutions.
Disseminate all population databases and certain key input databases through easily accessible web sites, restricting only those that are deemed to be security threats.
Design and execute a training program for worldwide deployment of essential technologies, databases, and models to first responders, coordinators, planners, and policy makers for real-time use in disasters.
Recognize that population estimation requires scientific advancement and should rank among the most important research priorities in the national interest. Efforts need to be devoted, for example, to exploiting the enormous, ongoing explosion of data (imagery and textual information) in a timely fashion using high-performance computing.
Chrisman, N., 2002. Exploring Geographic Information Systems New York: John Wiley & Sons, Inc. 305 pp.
Cincotta, R.P., J. Wisnewski, and R. Engleman, 2000. Human population in the biodiversity hotspots. Nature 404 (27 April):990-992.
Cohen, J.E., and C. Small, 1998. Hypsographic demography: The distribution of human population by altitude. Proceedings, National Academy of Sciences 95(24):14009-14014.
Nicholls, R.J., and C. Small. 2002. Improved estimates of coastal population and exposure to hazards released. EOS, Transactions American Geophysical Union 83(28):301, 305.
Small, C. and T. Naumann, 2001. The global distribution of human population and recent volcanism. Environmental Hazards 3(3-4):93-109.
Small, C., V. Gornitz, and J.E. Cohen, 2000. Coastal hazards and the global distribution of human population. Environmental Geosciences 7:3-12.
Tobler, W.R., U. Deichmann, J. Gottsegen, and K. Maloy, 1995. The Global Demography Project. Technical Report No. 95-6. National Center for Geographic Information and Analysis. University of California, Santa Barbara, 75 pp.
Wright, J.K., 1936. A method of mapping densities of population with Cape Cod as an example. Geographical Review 26:103-110.
Jerome E. Dobson is Professor of Geography at the University of Kansas and president of the American Geographical Society. He holds a Ph.D. in geography from the University of Tennessee (1975), M.S. and B.A. in geography from the University of Georgia (1972, 1967), and A.A. from Reinhardt College (1965). Dobson previously served as a research professor in the Kansas Biological Survey at the University of Kansas, distinguished research & development staff member at Oak Ridge National Laboratory, president of the University Consortium for Geographic Information Science (UCGIS), contributing editor of GeoWorld magazine, and U.S. delegate and expert to the International Standards Organization (ISO). His principal contributions are the paradigm of automated geography and his instrumental roles in originating the National Center for Geographic Information and Analysis and in establishing the Department of Defense’s LandScan Global Population Project and the National Oceanographic and Atmospheric Administration (NOAA) Coastal Change Analysis Program (C-CAP). Employing geographic information systems and automated geo-
graphic methods, he has proposed new evidence and theory regarding the mechanisms responsible for lake acidification and new evidence and theory regarding continental drift and plate tectonics. Dr. Dobson is the author of more than 150 publications. He and ORNL colleagues produced the LandScan Global Population Database, which has become the world standard for estimating populations at risk during natural disasters, wars, and terrorist acts. LandScan recently gained widespread acclaim as the only feasible means of estimating populations impacted by the 2004 tsunami in Southeast Asia.
ORGANIZATIONAL IMPEDIMENTS TO ESTIMATING POPULATIONS AND ACQUIRING, ACCESSING, AND USING POPULATION DATA
John A. Kelmelis, U.S. Geological Survey and Department of State
Population data are vital for decisions made in national and international policy, business and marketing, economics, disaster response, military operations, health planning and response, sustainable development and many other arenas. Some activities are impossible to conduct without population estimates. The flow of population information includes acquisition of raw data, refinement and analysis of these data to produce some type of database, storage and distribution of the data, access to the data by the user community, analysis for some specific purpose, supply of the analysis to the customer, and the development of new questions or requirements for the data; these new questions or requirements track their way back through the chain so that new or refined applications can be developed and produced and, ultimately, new requirements for the data articulated. At each stage of this process, organizations of one type or another participate. The manner in which these organizations participate affects the availability, quality, and the ultimate use of the data.
The charge given me by the Committee on the Effective Use of Data, Methodologies, and Technologies to Estimate Subnational Populations at Risk was to identify the limitations of current institutional structures in using existing demographic and other data and tools for estimating subnational populations. This question is broad, and the population data flow described above is used here to help provide structure to the answer. In addition, I was asked to make specific comments on the U.S. Department of State’s (DOS) and the U.S. Geological Survey’s (USGS) activities with regard to population data. Thus, broad community and specific agency questions are to be addressed.
In order to address the broad question, I received input from more than 25 people representing the USGS, DOS, Center for International Earth Science Information Network (CIESIN), Oak Ridge National Laboratory (ORNL), Centers for Disease Control and Prevention (CDC), U.S. Bureau of the Census, a number of other federal agencies, some academics, and members of other organizations. Input was received by responses to e-mail questions, direct interviews, and telephone interviews. The people with whom I discussed this topic are experts in their respective fields and represent producers and users of population data. I also sought and received input from the specific agencies about which I was asked to comment. This paper contains detailed information on several specific agency programs using population data and concludes with comments about the flow of population data and institutional impediments to that flow.
CONTEXT IN WHICH THE DATA ARE USED
Not surprisingly, population data are used for a wide variety of applications. In the DOS, for example, data can be used as a basis to brief policy makers at all levels up to and including the Secretary of State, Congress, and the President. At the USGS, data are used to identify how the hazards of the natural environment are likely to affect the social environment. I discuss below in more detail these two agencies’ needs for population data.
DEPARTMENT OF STATE
Diplomacy is conducted in a variety of settings. Some of these are at the negotiation table, in public fora, in private conversations, or at high level meetings. Often demographic or other population data are needed to develop and support policy positions. Within the DOS, the Bureau of Oceans and Integrated Environmental and Scientific Affairs (OES) uses population data when assessing the human impact of deforestation in Africa and Latin America and applying geographic information systems (GIS) to public health issues in the western hemisphere. The Office of E-Diplomacy (E-Dip) is working with the Bureau of Intelligence and Research (INR), along with other offices, to promote the use of geographic information, including imagery, and GIS throughout the DOS. GIS applications abroad, including at embassies and our other foreign missions, need to incorporate into their databases subnational population and related socioeconomic data.
Although the DOS has a number of people trained in demographics, they work in other fields, and the DOS has hired another person with training and experience in population and demography to work specifically as a demographer. This work is in constant demand to provide high-quality demographic analysis to the Population, Refugees, and Migration Bureau
and technical liaisons with UN demographers. This is an important positive step in improving the technical skill base in the DOS. However, the great demands on time for this work hinder application of these professional skills as a demographer to many other population- and demographics-related issues elsewhere in the agency. Limited technical staff and tools further preclude the DOS’s ability to evaluate much of the existing data or to conduct more sophisticated analyses that could address some pressing issues such as making better estimates of national and subnational populations.
Application of Population Data
The DOS uses population data to meet many policy demands. The depth and intensity of use vary from bureau to bureau depending on the issues being addressed and access to the skills and technology to address them. Local to regional population counts are needed to respond adequately to natural disasters such as earthquakes and landslides or to social conflicts. The genocide problem in Darfur was officially recognized in 2005. Part of the analysis needed at the time required identifying the locations of ethnic groups within and around the region. The only map of the area available to analysts that described the ethnicity of the population was produced in the 1950s. This 50-year-old map was central to the analysis, yet was without doubt highly inaccurate in representing today’s population distribution and composition. Many other countries lack recent, accurate population data that can be tied to modern maps.
Subnational population data do not apply only to disaster response situations. For example, GIS-based subnational population data are needed to track the population of the Solomon Islands for hazard mitigation, resource allocation, and economic development. The Solomon Islands have approximately 550,000 inhabitants distributed among 6 major islands and more than 900 smaller islands. Understanding the distribution, age profile, education level, and other characteristics of the population would be useful to help the government identify pools of potential workers and convince international investors to strengthen their positions in spice growing and the vanilla trade, and to establish other industries to avert the spread of poverty and possibly the spread of crime. Similar population distribution characteristics are required when implementing programs to promote the development of civil society, target public diplomacy campaigns, and provide support for elections at local levels.
Regional population studies or integrated regional analyses should be conducted on a continuing or frequent basis not only to ensure current population data but to obtain more in-depth information about the population. Maintenance of population databases must include updated information about population composition, social dynamics, availability of and
access to societal and environmental goods and services, and the myriad of attributes and relationships affecting or affected by the region.
The crisis in Darfur is a good example in that it has resulted in a large refugee population and even larger population of internally displaced persons (IDPs). Often in times of crisis, women and children are sent away while husbands and fathers remain to protect their homes and property. Understanding and modeling this phenomenon will help identify certain populations at risk in certain types of situations. A sound understanding of the culture is important because all cultures do not respond similarly to crises or humanitarian assistance.
The DOS regularly makes use of available counts or estimates of those in need of humanitarian assistance for the purpose of identifying potential beneficiaries of aid, but at the same time there is poor knowledge of resident populations and local services, making it difficult to assess the environmental and social stress caused by large numbers of displaced persons. This limits the department’s ability to track the distribution of resources and to ensure that aid is reaching those for whom it is intended.
Monitoring reconstruction and post-conflict conditions is of importance to policy makers as well. Subnational population information is critical here because variations in redevelopment among ethnic groups may be an indicator of continued internal stress and diversion of resources, internal policies of discrimination and repression, or other conditions that can lead to nutrition and health problems. Imbalance in reconstruction can result in dissatisfaction within the population leading to the reemergence of conflict. While not yet clear how a subnational geospatial population data set could be used for the analysis of trafficking in persons or other human rights violations, some testable hypotheses could be developed with such a database to attempt to monitor these issues.
In conclusion, at the DOS, population data are required for policy making, planning, and making operational decisions. The data could be part of the planning process to respond to complex humanitarian emergencies, to understand development needs, to monitor environmental impacts, to improve medical response, and to respond to natural disasters. The department also provides assistance to refugees and monitors elections, migration, and the status of women, and evaluates nations on such issues as trafficking in persons, human rights, and the nutritional conditions of inhabitants. All of these are population-related issues, though all might not use the same database.
U.S. GEOLOGICAL SURVEY
The USGS has very program-specific uses for population data. The USGS uses population data primarily in three programs, Prompt Assess-
ment of Global Earthquakes for Response (PAGER), Volcano Disaster Assessment Program (VDAP), and the Famine Early Warning System Network (FEWS NET), and is considering using population data in other programs. Population data are also used by the USGS in a number of projects and experiments including urban dynamics modeling, local hazard and risk analysis, and response to specific Earth events such as the 2005 landslides in Pakistan. The USGS has incorporated population data into some of its databases, such as the Global GIS. Furthermore, USGS land-cover data are used by developers of population databases to help refine and disaggregate national data into subnational estimates. This paper concentrates on two programs, PAGER and FEWS NET, to describe the manner in which the USGS consistently uses population data.
Application of Population Data
PAGER. The system PAGER, a cooperative project with the U.S. Agency for International Development (USAID), is designed to transmit alarms by pager, mobile phone, and e-mail that include precise estimates of impact. The information transmitted includes earthquake location, magnitude, depth, an estimate of the number of people exposed to varying levels of shaking, a description of the region’s fragility, and a measure of confidence in the systems impact assessment. Associated maps of shaking level, population density, and susceptibility to landslides are posted on the Internet. This information is made available within minutes of the determination of the earthquake’s location and magnitude. Earthquake solutions are generally available within an hour for significant events, and improvements to the real-time earthquake detection system will soon decrease the response time to as little as 15 minutes (Earle, 2005).
Although the information processing and flow are straightforward, the scientific work to develop accurate shake maps is highly complex, involving understanding of the geologic setting, geophysical processes and properties, and overlying earth materials. Understandably, the accuracy of the shake maps is highly variable depending on the knowledge of regional geology. PAGER relies on established population databases to identify the vulnerable population; it then generates an impact statement by considering the fragility of the exposed population and infrastructure, the potential for earthquake-induced landslides, and if available, damage reports from nearby historic earthquakes. PAGER uses ORNL’s LandScan database for population estimates. Scientists on the PAGER team indicate that both LandScan and the CIESIN population databases are important and should continue to be updated and developed (Earle, personal communication, 2006).
When the March 28, 2005, earthquake struck northern Sumatra three months after the great tsunami of 2004, the data from the then-experimental
PAGER analysis were transmitted to the first responders within three hours of the event. Those data provided the basis to prepare flight plans so helicopters could depart on rescue missions at dawn after the earthquake.
FEWS NET. While PAGER is designed for response to rapid-onset disasters (e.g., earthquakes), FEWS NET is a multi-organization, international program designed to address and prevent disaster from a slow-onset hazard, famine, in 26 participating nations in Africa, Central Asia, Central America, and the Caribbean. Funded by USAID, FEWS NET is designed to strengthen the abilities of participating countries and regional organizations to manage the risk of food insecurity by providing timely analysis and early warning and vulnerability information. FEWS NET develops and uses analytical tools to combine data on climate, meteorology, hydrology, soils, land cover, vegetation, topography, agriculture, and other variables to develop forecasts of future food security conditions and provide warnings in case of pending food security issues. Although FEWS NET will continue its role in monitoring and warning about food security problems, it is increasing efforts to identify underlying causes of food insecurity and provide that information to policy makers (FEWS NET, 2006). Because the ratio of population to available food varies from place to place and time to time, and can result in localized food shortages, better subnational population estimates are critical to this effort.
FEWS NET is truly a partnership among the nations and communities involved. Advanced analytical systems have been developed by all partners and have been contributed to the program. In addition, cooperative activities at the community level have produced maps and atlases to which communities have contributed. This involvement by potential users of the warnings has the potentially positive result of leading to more informed use of FEWS NET. The analytical modules are updated as improved data become available.
FEWS NET uses the LandScan database, population data from CIESIN, and population data derived directly from participant nations. With the exception of a few governments that sometimes do not want to release existing national or subnational population data for political reasons, FEWS NET managers have not noted many other substantial institutional barriers in getting access to existing population data. Insufficient funding remains a major issue for conducting censuses in many of the nations within FEWS NET.
Weaknesses in the data can result in very inaccurate assessments of food availability, food import, and food needs. For example, whether Zimbabwe has 13 million or 10 million residents, after massive emigration to South Africa and elsewhere, has a huge impact on what the country’s government will have to import to supplement its failed harvests and what the donor community may have to plan to supplement and distribute. Such
large magnitudes of potential error may be costly in terms of direct expenditures, nutritional stress on vulnerable populations, loss of livelihoods, (Eilerts, personal communication, 2006), and for decisions on where to ship food once it is in country.
INSTITUTIONAL IMPEDIMENTS TO POPULATION DATA FLOW
As outlined earlier, the flow of population information includes acquisition of raw data, refinement and analysis of these data to produce some type of database, storage and distribution of data, access to the data by the user community, analysis for some specific purpose, supply of the analysis to the customer, and development of new questions or requirements for the data; these new questions or requirements track their way back through the chain so that new or refined applications can be developed and produced and, ultimately, new requirements for the data articulated. This section outlines barriers within institutions that restrict aspects of this data flow scheme.
Due to lack of available subnational population data, specific questions from planners and field personnel have gone unanswered. This has led to additional costs, compromised activities, and in some cases, the failure of projects. In some instances the data have existed, but the lack of corporate knowledge of their availability has prevented their use.
Acquisition of Raw Data
The physical acquisition of population data in some subnational regions can be difficult and hazardous. For instance, in March 2006 a number of people were killed in ethnic violence as Nigeria started a five-day census. Religious, tribal, and regional rivalries resurfaced in several states as nationally appointed enumerators rioted when local authorities fired them and substituted their own officials. The violence stemmed from anger that the census would not record ethnicity or religion. The census fueled mistrust over the distribution of oil wealth and constitutional reform. Methods must be developed to obtain reasonable estimates of subnational population characteristics, including ethnicity, religion, tribal affiliation, and other attributes, in spite of difficulties caused by regional, national, and local conditions.
Lack of data below the provincial level and data gaps at the intra-urban (neighborhood) level have made detailed humanitarian aid and other socioeconomic planning difficult. Increased emphasis on subnational
population estimates specific to urban areas is a growing concern, particularly as an increasing global population becomes more urbanized than at present.
Obtaining specific census data by districts or quarters for a foreign city or village is extremely difficult. National- or provincial-level census data do not have the fidelity to provide the information needed for detailed analysis or planning. Also, frequent changes in administrative boundaries pose problems in reallocating populations if population data are not available at the appropriate scale.
Legal and political considerations place limitations on the sharing of not-yet-aggregated population data.
Thresholds of chronic and acute malnutrition and infant mortality rates are a major influence in the prioritization of perceived emergencies. The basic population data that drive calculation of these rates, in many of the most critical areas, are not very robust.
The U.S. government should work more closely with nongovernmental organizations to help obtain the necessary data. This partnership should include increasing access to existing data, tapping into existing surveys to extract data, and helping to design new surveys to obtain needed information.
Highly detailed population data are occasionally made available to embassies by host governments. It is not clear that these data find their way into databases or are available for use by analysts or operations personnel outside the embassy.
Some international organizations (e.g., the World Health Organization) are required by their position to report only the data released to them by countries, which can lead to extremely inaccurate databases. In many countries, this problem has been exacerbated by a combination of inaccessible terrain and a lack of international interest outside times of crisis that has quelled any initiative for population data gathering.
In truly large and socially diverse countries, great differences in exposure to risk can exist among internal population groups. The development of subnational population databases would be extremely helpful in these situations to delineate exactly which segments of the subpopulation are truly “at risk.”
Requirements and Analysis of Data for Database Development
Land-cover data sets such as USGS Global Land Cover 2000 derived from the Advanced Very High Resolution Radiometer satellite and other sources, North American Land Cover derived from Landsat and other sources, and GeoCover LC, a commercial land-cover database derived from Landsat, are critical to refining global population databases. However,
the age of the data, their resolution, land-cover classes, and interpretation are problematic despite advances with additional analysis, high-resolution imagery, and application of geographic expertise. A global high-resolution land-cover database produced to the appropriate standards and specifications would be helpful for producing more accurate subnational population databases. Such a land-cover database should be updated periodically to reflect the rapidly changing land-surface environment. More frequent updates should be made in regions vulnerable to disasters and social upheaval.
Numerous development assistance programs require gathering of population, health, natural resources, and other data. No standard interoperable procedures exist for the data, data collection, or data sharing. These data usually do not reach official, organized databases and may be lost to the U.S. government. These assistance programs, funded by USAID and others, should be required to provide their data to a centrally operated repository to be archived and distributed, particularly since many data may already be supported by taxpayers or donors.
As storage media and technology change, data must be converted to the new storage and retrieval systems for their long-term survival. Both current and legacy data sets must be maintained. Some possible options for storage are the Library of Congress and the USGS Earth Resources Observations System (EROS) Data Center, both of which have experience in maintaining and refreshing large amounts of disparate data and have been established to store and manage information for the long term.
The International Programs Center of the U.S. Census Bureau maintains an international database containing population estimates and projections at the national level for more than 227 countries and areas. This database is updated to reflect the receipt of new data and changes in trends or methodologies. However, products for population estimates and projections are done on an ad hoc basis depending on the needs of clients. In addition, subnational population and demographic analyses, estimates, and other products are based on user requests and do not form a global database. A more effective approach for data maintenance and estimation of subnational populations would be to operate on a program basis with a consistent set of standards, specifications, methods, and revision cycle. A programmatic approach would help in establishing long- and short-term priorities as well, such as (1) performing comprehensive subnational population estimates for a group of predetermined high-priority countries on an appropriate periodic basis (in some cases annual), and (2) allowing the unit and its partners to mobilize quickly to collect and analyze subnational population data for a specific high-priority country on very short notice. Resources would be needed to accomplish this.
Refining data often requires ancillary data, including remote sensing and land cover, authenticated place names, current boundaries, and imper-
vious surfaces, and can be significantly improved at a highly detailed level with cadastral data. These types of data must be current.
Online open-source population and census databases are easily accessible and can quickly be compared to other population databases. However, these comparisons are relative, not absolute. For example, the Population Reference Bureau (PRB) provides access to a wide variety of variables useful for analysis. The PRB also produces analyses on topics as varied as population and demographic change, immigration, health, morbidity, mortality, and many others. Yet PRB is not a portal with access to most geospatial population data and does not provide an opportunity to check spatial data against some accepted “known” data set.
One major obstacle involves host country government unwillingness to release population data for various reasons. Unfortunately, donors funding these kinds of studies often agree to these restricted release terms when working with a client government. Such behavior only perpetuates this impediment. If bilateral and multilateral donors funding census work would agree to use standard data release language as a requirement for all future censuses, the release of census data funded with donor resources could be ensured. This type of stipulation should be a requirement before signing an agreement to undertake data collection work.
Neither a clearinghouse nor a portal exists for access to all population data. Within the U.S. government these data could logically be housed at the Census Bureau, Department of Housing and Urban Development International Office, or USGS EROS Data Center. In addition, a reasonably comprehensive portal would be helpful in identifying overlaps and gaps in the data.
The U.S. government has no single arbiter of population data to evaluate the accuracy and other quality indicators of various data or to identify the best data for a particular application. Although the Census Bureau performs this role for the data under its purview, many data used by the U.S. government are produced by others and are not evaluated in the same fashion as those at the Census Bureau.
The analysts’ environment is high pressure. Results are usually needed on a short turnaround. Decisions that are made are often politically charged. Therefore, there is a tendency for analysts to use data and techniques that have been accepted in the past rather than take the risk of using a new data source or analytical technique that is not fully tested or understood. Here
again, population data experts who could keep abreast of new developments in data and techniques would be a valuable addition to the DOS team, whether they reside at the DOS, the Census Bureau, or elsewhere.
Much work is “just in time” and conducted by analysts with very broad portfolios. Not all analysts can be expected to be experts in population data evaluation; these data are among the many inputs to a wide variety of analyses that must be produced on short notice. Analysts must use the data that are available. Here again, expert vetting of population data in advance of a disaster or crisis would be useful.
Confidence in data quality is a great concern and can affect analytical results. In Africa, population databases at the subnational level are based almost entirely on estimates (at best). This uncertainty affects risk estimates. For example, an analyst must constantly overestimate the presence of infectious diseases, just to provide the safest assessment to the audience. This may lead to a distortion of actual disease risk and might encourage the channeling of humanitarian assistance resources or funding away from deserving areas and toward lower-risk ones.
The DOS has limited in-house capabilities to conduct populationrelated analyses and should raise the profile of and increase access to technical skills and resources in population, demographics, and other social sciences. One option is to emulate its current effort to reincorporate the physical and biological sciences, technology, and health into its programs. In that effort a number of internship and fellowship programs have been established to ensure that scientific expertise exists within the DOS and that links are maintained with the broader scientific community. Similar programs could be developed to expand the availability of technically trained demographers, population experts, and other social science technical skills into its personnel mix.
The value of analyses of population data can be improved by developing and using understandable ways to present the uncertainty of the analysis. The source of the uncertainty must be articulated clearly and may require improving the customer’s understanding of the analytical tools and the types of data used.
Develop New Requirements and Standards
No systematic feedback mechanism exists to transmit potential user requirements for the data from the ultimate user of the data analysis (policy maker, strategist, response planner, responder) to those who analyze, acquire, and produce the data.
In many instances, no post-action analysis is conducted due to time limitations and constantly emerging issues. Post-action analysis might aid in improving decision making and the use of information to make future actions more effective.
Standard operating procedures for data collection and standards for data content, accuracy, acquisition, transfer, and interoperability should be developed. The Federal Geographic Data Committee; its international arm, the Global Spatial Data Infrastructure; and the Open GIS Consortium could provide models or a vehicle for developing standards for geospatial population databases.
Private producers of data are interested in selling the data for a profit; therefore, they divide the data into components to sell, while government producers generally provide the data at no or nominal additional cost to users because taxpayers have already borne the cost of data acquisition. Issues of trust, security, and cost arise when addressing the trade-offs between publicly and privately produced data.
SUGGESTIONS FOR CHANGES OR NEW CAPABILITIES
A repository for subnational data should be established. An entity within the government responsible for funding, collection, and storage of subnational data sets should be designated. With appropriate safeguards for privacy and security, access must be provided to original surveys from the many agencies and other organizations.
A catalog of subnational data sets held by each nation and the detail of these data should be created. For example: block-level census data should be available for any nation. Organization of ownership, content of data, and method of access should be included as part of the metadata.
Standards for data content, collection, quality, and format should be developed and used to ensure usefulness and interoperability. Requiring that funding be contingent on the use of these standards and providing the data to the appropriate accessible database or clearing house will help ensure conformity to the standards.
Collectors, policy makers, planners, and analysts should be educated on the importance and use of subnational data sets in their work.
Data collection and mapping companies and other entities such as LexisNexis, Environmental Systems Research Institute, Inc. (ESRI), Gallup, CIESIN, and ORNL should collaborate to obtain needed data.
Academic researchers should be engaged to conduct studies involving subnational data sets. Academia has the potential to access in-country information not otherwise obtained.
Access to demographic and regional studies, articles, reports, maps, and other documents produced by other countries should be improved. Translation services should be made available.
Regional population studies or integrated regional analyses should be reinstituted, modernized, supported, and funded as an important part of academia. As part of master’s and doctoral studies, fieldwork conducted in particular regions should include mapping ethnicity, religion, and other cultural variables, as well as understanding the causes and effects of change in social, political, economic, health, and environmental conditions. The relationships between subnational populations and the broader regional contexts should be explored.
To help incorporate a variety of data sources and reduce costs of field data acquisition, modeling, statistical, and other analytical techniques should be used and improved. Attention should be placed on dynamic as well as static populations and the use and development of ancillary data to help improve understanding of what is taking place at the subnational scale.
A program for systematic global subnational population data collection and estimation, administered in the Census Bureau should be developed and funded. This should complement, not replace, the customer-driven, ad hoc, project-based subnational data collection currently in place. Recognizing that sufficient funds will not be available to acquire all of the desired data, such a program would work on a set of U.S. governmentidentified priorities, including a strong field collection and survey component, advanced estimation techniques, and a surge capacity for suddenonset or unanticipated emergencies.
FEWS NET, 2006. About FEWS Net. Available online at http://www.fews.net/about/index. aspx?pageID=aboutTheApproach [accessed March 9, 2006].
Earle, P., 2005. ProtoPAGER, U.S. Geological Survey. Available online at http://earthquake. usgs.gov/eqcenter/pager/ [accessed March 8, 2006].
John A. Kelmelis is a senior science adviser at the U.S. Geological Survey and is the senior counselor for Earth science at the U.S. Department of State. He is leading studies to identify Earth science findings with foreign policy implications and advising on remote sensing, geographic information, and scientific issues for areas of national concern. He is a senior science adviser to the Humanitarian Information Unit and is also examining selected sustainable development, natural hazard, terrorism, and critical infrastructure issues. He is the former Chief Scientist for Geography at the
USGS. He led the USGS Global Change Research Program. He directed the White House Scientific Assessment and Strategy Team, managed the United States Antarctic Mapping Program, and conducted research on many geographic and other scientific topics. He is a counselor of the American Geographical Society and is on many national and international scientific boards and committees. He holds a Ph.D. in geography from the Pennsylvania State University.
POLITICAL GEOGRAPHY AND EMERGENCY RELIEF
Wm. Glen Lauber, National Geospatial-Intelligence Agency
When emergency relief is sent to a nation in crisis, responding organizations (foreign governments, nongovernmental organizations [NGOs], or the United Nations [UN]) often learn that the government accepting their aid does not have adequate data in a usable form to identify all administrative subdivisions and all populated places. Without these data, aid agencies do not know where precisely to send aid, resulting in prolonged suffering and the potential for additional illness and mortality.
Political geography should be part of the mitigation strategy for emergency preparedness. Relief agencies must know immediately what places are affected and what they are called. Villagers probably do not know the geographic coordinates of their village; they know the local village name, but this name may not be recognized by the government. If relief workers cannot find a name or the area of an administrative subdivision on a map, in a gazetteer, or in a geographic information systems (GIS) database, that place may not receive aid.
The proper place from which to develop these types of standardized geographic data is at the level of individual countries. Place names reflect millennia of culture, language, and history and rouse strong feelings in people so these names should be established and standardized by countries themselves. Well-designed databases with variant names and approved boundaries will limit the confusion for aid workers in emergency relief efforts and contribute toward generating and maintaining better estimates of subnational populations.
UN mechanisms already in place should be engaged to exhort those nations to toponymic and boundary-depicting action. This paper introduces and briefly discusses possible avenues for addressing the inadequacy of geographic place names and political boundary data in countries worldwide. Addressing these fundamental issues will aid in removing some of the structural impediments to organizations dispersing emergency aid and will facilitate international, national, and private efforts in making and retrieving population estimates that are part of emergency relief plans.
STANDARDIZATION OF GEOGRAPHICAL NAMES
The best international organization to address worldwide geographic names standards is the UN Group of Experts on Geographical Names (UNGEGN, pronounced n-g ’-g n).1 The UNGEGN mission, derived from UN resolutions recognizing the importance of geographic name standardization, is as follows:
To provide encouragement and guidance to those nations which had no national organization for the standardization and coordination of geographical names to establish such an organization and to produce national gazetteers at an early date;
To take the necessary steps to ensure the functions of a central clearinghouse for geographical names, including: the collection of gazetteers; and the tion concerning the technical procedures adopted by Member States for standardization of domestic names, and concerning the techniques and systems used by each Member State in the transliteration of the geographical names of other countries.2
According to UNGEGN approximately 25 percent of nations have a national authority for standardizing geographical names.3 Many authorities were established since 1990 and remain immature. Three authorities can trace their toponymic roots to the 19th century (Canada, Norway, and the United States). Not documented but well known is the paucity of resources these authorities have to develop and make accessible a complete record of their place names. UNGEGN documentation makes it obvious that developing countries constitute the majority of countries without a names authority. Although all countries make maps of their land, without a place name authority to document and catalog place names according to a standard, the data will be incomplete and conflicted.
The UN recognized the importance of names standardization in 1948 and held the first Conference on the Standardization of Geographical Names (UNCSGN) in 1967
to encourage national and international geographical names standardization; to promote the international dissemination of nationally standardized geographical names information; and to adopt single romanization systems for the conversion of each non-Roman writing system to the Roman alphabet.4
The UNCSGN formally established UNGEGN in 1972 (an ad hoc group had been meeting since 1960) “to carry forward the programme of cooperation between conferences.”5
To accomplish its task, UNGEGN is organized into 22 divisions of geography and linguistics,6 and 10 working groups on topics and issues of toponymy.7 Among the working groups are Publicity and Funding,8 Training in Toponymy,9 and Toponymic Data Files and Gazetteers.10
UNGEGN is organized to promulgate place name standardization to as many countries as possible but is not funded for an aggressive campaign. The various divisions are urged to seek grants and search out other sources to fund training in their part of the world.
The typical UNGEGN course in toponymy lasts for 12 days including a day of field collection. The subject areas include the following: Functions of Official Geographical Names, UNGEGN Resolutions, National Geographic Names Authorities, Language Issues, Technical Issues and Toponymic Research, Cartography and Geographic Names, Field Work Preparation, International Issues, Toponymic Data Files, and Databases and Gazetteers.
The faculty is drawn from UNGEGN members through funding from either the UNGEGN Secretariat or a member country.
One UNGEGN principle is free access to all names data. Nations are perfectly willing to grant public access to their names for many economic and cultural reasons. The problem is implementing data access solutions that are useful for both routine and emergency situations.
Much work is yet to be done in this area. At its meeting in March 2006 the UNGEGN Working Group on Toponymic Data Files and Gazetteers recognized that relief organizations have specific needs; UNGEGN has agreed to achieve progress in this area by the ninth UNCSGN in August 2007 in New York.
Ibid., see working group number four.
Ibid., see working group number six.
Ibid., see working group number two.
The UNGEGN Secretariat11 has initiated a project to develop a distributed database of place names, starting with national capitals and cities with a population of 100,000 or greater. The secretariat will not be ready to handle all places for some time. It relies on UN member nations to have their own digital database it can access. The United States and the United Kingdom will contribute their considerable experience in this field.
The Efforts of the United States
The U.S. Board on Geographic Names (BGN)12 is responsible for the official U.S. government names of foreign places as well as U.S. places and maintains the official U.S. government database of foreign place names13 as a publicly available resource.14 Part of the BGN plan to improve the database’s detail and currency is to train foreign countries in toponymy and help them maintain their capability, in exchange for access to their toponymic data files.
The BGN has been providing toponymy training to a limited number of foreign governments. This work continues in Latin America through the Pan American Institute of Geography and History (PAIGH),15 and elsewhere directly with foreign governments.
The Proposed Way Ahead
Government relief agencies and NGOs concerned with emergency relief and UNGEGN divisions and working groups should engage each other to create useful solutions to the problem of inadequate place-name data in emergency areas.
NGO and UN Development Programme (UNDP) grants should be awarded to fund training, and government foreign aid should be allocated to the startup costs of national names authorities.
Grants and aid should be awarded for a standard but flexible data management solution addressing routine and emergency management requirements.
The International Organization for Standardization should be engaged to aid in standards development.
Financing is also an issue. Many governments have already invested a great deal of money over 130 years to reach the current data level. Continued funding for these efforts is necessary to make the suggested improvements.
ACCURATE GEOSPATIAL DEPICTION OF INTERNAL POLITICAL BOUNDARIES
A leading resource for the science of political boundaries is the International Boundaries Research Unit16 (IBRU) of Durham University in the UK. No UN equivalent to UNGEGN exists for political boundaries. The UN Geospatial Information Working Group17 (UNGIWG, pronounced ŭn’-jē-wĭg) addresses boundaries but is only six years old and has other tasks as well.
Accurate and acceptable geospatial depiction at varied scales remains a challenge in work with political boundaries. Both IBRU and UNGIWG should be engaged by emergency relief organizations to help address administrative boundary inadequacies. Both organizations should be asked to lead the closing of the standards development gap.
IBRU was established in 1989 to “enhance the resources available for the peaceful resolution of problems associated with international boundaries on land and at sea, including their delimitation, demarcation and management.”18 Its chief contribution to this problem would be to host workshops providing training.
UNGIWG was created in 2000 to “address common geospatial issues—maps, boundaries, data exchange, standards—that affect the work of UN Organizations and Member States.”19 It answers to the UN Systems Chief Executives Board for Coordination20 (CEB). Its mandate stems from the UN Economic and Social Counsel Resolution 131 (VI) of 1948. Most UNGIWG work to date focused on international boundaries, and only at 1:1,100,000 scale.21 In 2001, UNGIWG began work on the world’s first and second administrative order boundaries, also at 1:1,100,000 scale.22
UNGIWG Task Group 1 addresses international and administrative boundaries. Its objectives are to “provide the UN community, and when
possible the international community, with validated information and maps regarding international and administrative boundaries for all the UN member countries.”23
As with the proposed UNGEGN names database, the issue of scale is problematic with regard to UNGIWG. However, member nations would be valuable contributors if they were able to provide reliable data at larger scales, such as 1:50,000 or 1:24,000. Organizations such as UNGIWG will be left to guide the development of standards for the accurate geospatial depiction of administrative boundaries at a variety of scales.
IBRU would have to be paid,24 and UNGIWG would require additional funding to move beyond offering just one, small-scale administrative boundaries data set.
IBRU could be contracted to conduct workshops worldwide. UNGIWG does not offer boundary analysis training but should be invited to participate as a party interested in having more countries capable of sharing reliable data.
Much work is needed to make accurate and acceptable geospatial depictions of boundary data, particularly data that traditionally have been maintained in either textual form or undocumented historical memory. Even before the advent of widely available GIS, the large-scale cartographic depiction of boundaries was contentious and open to widely varying interpretations.
Treaties were traditionally vaguely written to a small-scale map on the conference table. Small-scale treaty description frustrates accurate large-scale geospatial depiction. Many countries face this problem in today’s world (e.g., Eritrea and Ethiopia; Israel and Syria). Digitized depictions of a segment of boundary from different hard-copy sources will vary widely when displayed together on a GIS. The world must establish a set of standards for accurate and acceptable interpretations of a boundary between points in a treaty in a large-scale environment.
The Efforts of the United States
The U.S. Department of State and the National Geospatial-Intelligence Agency work to establish the recognized U.S. government depiction of international and administrative political boundaries, consistent with the treaties each government accepts. They have recognized the need and accepted responsibility for developing large-scale geospatial boundary depiction standards for U.S. government use. Their work will significantly impact political boundary science.
The Proposed Way Ahead
Government relief agencies and NGOs concerned with emergency relief should engage UNGIWG and IBRU to create useful solutions to the problem of inadequate administrative boundary data in emergency areas.
UNGIWG and IBRU should support the Department of State and the National Geospatial Intelligence Agency (NGA) by acting as the test beds and mediators for negotiating acceptable standards.
NGO and UNDP grants should be awarded to fund training, and government foreign aid should be allocated to the startup costs of national boundaries authorities.
Grants and aid should be awarded for a standard but flexible data management solution addressing routine and emergency management requirements.
Grants and aid should be awarded for the promulgation of large-scale boundary analysis standards.
Administrative boundary definition and depiction comprise a tougher problem than standardized place names. Countries seem less willing to share boundary information, and are perhaps reluctant to expose internal disagreements that might imply a weakness. Emergency relief agencies and organizations will have to convince these countries that it is in their best interest to share these administrative boundary definition and depiction data. The boundaries do not need to be declared definitive; they have always been subject to change. The boundaries do have to be accurate and acceptable for use in a large-scale environment.
Disaster preparedness is not sexy, and memories of poor preparedness do not last long in the minds of political leaders and their constituents. Attention is quickly drawn to response and recovery. The money appropriated for improved preparedness is usually spent on visible items such as fire and
rescue equipment, and not on the data required to efficiently direct those and other physical resources that NGOs and other relief agencies send.
Given the difficulty in garnering support for response requirements, it is critical that response agencies find a long-term economic solution to the problem of inadequate geographic place names and political boundary data.
UNGEGN, IBRU, and UNGIWG are a credible nexus for addressing the problems. They hold expertise, prestige, and an infrastructure directly applicable to these issues. A network should be built around them offering additional resources that would serve not only crisis needs but also the long-term good.
Wm. Glen Lauber is the former chief of the Political Geography Division of the National Geospatial-Intelligence Agency Analysis and Production Directorate (P). The division provides geospatially-based cultural and political geography analysis (“mapping of the human terrain”) for the intelligence community and the Department of Defense. He holds a bachelor of arts degree in political science from Lycoming College and a graduate certificate in systems engineering from George Washington University. Mr. Lauber received a commission as a Military Intelligence officer in the United States Army in 1981. He served on active duty in the 1st Cavalry Division and the National Photographic Interpretation Center until 1988. After leaving active duty, he worked for Hughes Aircraft Company/Raytheon in various positions on contracts with NGA and its predecessors. In 2003, Mr. Lauber left the private sector to accept a position with NGA/InnoVision, the agency’s research and development directorate. He transferred to P and served as chief from June 2005 to July 2006.
IDENTIFY WAYS IN WHICH SUBNATIONAL DEMOGRAPHIC AND GEOGRAPHIC DATA COULD BE USED TO HELP DECISION MAKERS PROVIDE USEFUL INFORMATION TO POPULATIONS AT RISK
Shannon Doocy, Johns Hopkins University Center for Refugee and Disaster Response
Over the past several decades, natural disasters have become more frequent, and the populations affected by disasters annually are following a similar and increasing trend (IFRC, 2003). The growing impact of natural disasters has triggered a need to enhance the understanding of vulnerability and to improve both the quality of data and the management
of information that guides the humanitarian response following disasters. Disasters and the subsequent humanitarian response have a large impact on population well-being, and it is important that relief and recovery efforts be adequately informed in order to target assistance appropriately and facilitate recovery.
Modeling population vulnerability and risk in natural disasters and post-disaster assessments of surviving populations enables governments and humanitarian organizations to make rapid, informed decisions under conditions of great uncertainty (McEntire, 2001; Kunreuther, 2002). Post-disaster surveys of affected populations that incorporate aspects of population movements, needs assessment, living conditions, and health data with spatial modeling are ideal tools for providing information on disaster vulnerability and impact and have a practical role in assisting governments and humanitarian organizations in disaster response by promoting informed management and decision making. Information generated from these approaches can also inform rehabilitation, mitigation, and preparedness. The advantage of using modeling before or immediately after a disaster is that it can help target and guide post-disaster activities, including surveys, to obtain higher-quality data for consequent analysis. Current advances in spatial data organization, including the availability of analysis with geographic information systems (GIS), provide an opportunity for improvements in rapid assessments as well as rapid analysis and dissemination of findings to governments and organizations involved in the ongoing disaster response (Herold et al., 2005; Mansor et. al., 2004; Zerger and Smith, 2003; Chen et al., 2003; Simonovich and Nirupama, 2002).
HUMANITARIAN INFORMATION SYSTEMS
Humanitarian Information Systems (HIS) constitute an essential element of the response to emergencies because they provide a basis for decision making. This is particularly critical in the chaotic early stages of disaster response where priorities must be established in order to quickly reduce the impact of the disaster on those affected and save the most lives. Conventional HIS components include early warning systems and response monitoring. Early warning systems identify the processes that trigger emergencies, while response monitoring supports the management and delivery of relief. Early warning systems usually focus on the size of the hazard; the impact on populations according to vulnerabilities is a weaker part of an early warning system. Tying this anticipated impact to specific vulnerabilities and providing this information to responding governments and international organizations are the functions of the information system, in which innovative informatics play a critical role.
The initial early warning response is the nucleus from which the full HIS will grow; it is constantly adjusted and reshaped as the disaster and the response unfold. The information system plays an important role in tracking the ongoing response to ensure that it effectively meets the needs of all vulnerable populations. New vulnerabilities develop as the response unfolds. An effective evidence-based system can track these as they emerge and alert responders to intervene. While HIS are critically important in disaster response, a variety of factors have been identified as limiting their utility including the ineffective and inefficient use of existing tools or methods, difficulties in accessing information, and the fact that HIS are often poorly linked with decision making, resulting in untimely or inappropriate responses (Buchanan-Smith and Davies, 1995). As a result, information systems that are put into place to help guide emergency response are often incomplete, ineffective, or inefficient (Maxwell and Watkins, 2003).
AVAILABILITY OF SUBNATIONAL DEMOGRAPHIC DATA
Demographic information is often provided on a national basis, but global environmental and other cross-disciplinary studies usually require data at the subnational level. Ideally, subnational demographic data are referenced by geographic coordinates, such as latitude and longitude, rather than by political or administrative units. In practice, major limitations exist in the availability and quality of subnational demographic data. In many cases, subnational population figures are outdated and incomplete; have been collected using imperfect methodologies; or were developed using unknown models. However, in many instances, particularly in the post-disaster context, decision makers are forced to rely on the best available information, regardless of methodological constraints. In addition to problems that may arise due to the quality of population data, subnational demographic information is typically provided by an administrative unit, which may create challenges when describing populations at risk or affected by disasters because disaster-prone regions are seldom congruent with administrative boundaries.
A recent innovation and resource for characterizing populations at risk of and affected by disasters is spatially distributed demographic data. Gridded Population of the World version 3 (GPWv3) and the Global Rural-Urban Mapping Project (GRUMP) are the latest developments in the rendering of human populations in a common georeferenced framework and are produced by the Center for International Earth Science Information Network at Columbia University (see http://Sedac.ciesin.columbia.edu/gpw/). GPWv3 depicts the distribution of human populations across the globe. GRUMP builds on GPWv3 by incorporating urban and rural information,
allowing new insights into urban population distribution of human settlements. Spatially distributed demographic data from GPWv3 and GRUMP are based on the most recent available demographic data and face similar data quality constraints as those discussed in the preceding paragraph.
Developed between 2003 and 2005, GPWv3 and GRUMP provide globally consistent and spatially explicit human population information and data for use in research, policy making, and communications. In the pre-disaster context, GPWv3 and GRUMP, when combined with environmental risk models, can serve to estimate the population at risk of disaster. This information is equally important in the post-disaster context where it can be used to verify estimates of affected and surviving populations in need of assistance (which often vary significantly by organization and time and can be highly inaccurate).
ASSESSMENT OF POPULATION STATUS AND VULNERABILITY
The availability of spatially distributed demographic data allows for the quantification of the population at risk of or affected by a disaster; however, it does not provide an indication of the status of the population or of humanitarian assistance needs in the post-disaster context. Assessment of surviving population status and conditions immediately following the disaster is an essential HIS component that should occur in the earliest phases of the disaster response. In many disaster situations, integrated needs assessments and population surveys that are representative of the entire affected population are not conducted or are implemented using inappropriate methods. The result is a dearth of information or inaccurate information that complicates decision making and planning of the humanitarian response. In addition to methodologically sound assessments of the status and needs of the disaster-affected population, the spatial distribution of selected indicators of population needs and status is an ideal means of providing information for decision making to governments and international organizations involved in the response.
Concentrating humanitarian response on the most affected populations requires an analysis of vulnerability and needs. The desire to focus assistance on the most vulnerable is based on the assumption that they have the least capacity and resilience to deal with sudden change, uncertainties, and other disaster-associated impacts. Identification of vulnerable groups is essential because characteristics of vulnerable populations vary by disaster and relate to a combination of social and environmental factors. An integrated approach to the examination of environmental and social risks in the natural disaster context is ideal.
Combined assessments that incorporate environmental risk models with post-disaster population surveys that characterize population status
and social risk factors are ideal because vulnerability to disasters is related to both human and environmental conditions. Spatial modeling of environmental risk factors can identify the levels of impact across disaster-affected areas. Human factors in vulnerability include social and demographic variables such as population age structure, sex, and socioeconomic status; the latter encompasses education, occupation, and other measures of social and economic status such as wealth or membership in certain group.
Environmental risk models can be combined with spatially distributed subnational demographic data to provide an initial characterization of disaster-affected populations. In the post-disaster context, combining environmental risk models with population status and needs data (from surveys of disaster-affected areas) can aid in identifying at-risk and vulnerable populations that require humanitarian assistance. The integration of post-disaster survey data with GIS allows the mapping of populations that report specific needs or substandard living conditions; it can facilitate the identification of population subgroups and geographic regions that are particularly at risk and can be used to guide disaster response.
DATA VISUALIZATION SOFTWARE
Data visualization software can help to present information on post-disaster population status and humanitarian assistance needs to decision makers and those involved in disaster response. An example of one data visualization and spatial statistics software package that is freely availably is GeoDa (see https://geoda.uiuc.edu/default.php), which was developed by the Spatial Analysis Lab at the University of Illinois, Urbana-Champaign. By employing visualization software, users have interactive access to a vast amount of information that is displayed in a more readily understandable form. For example, spatial analysis of demographic data can help identify regions where a particular event or need is concentrated. Information can be presented spatially, either by rates or by accounts per administrative unit, which can help to guide the disaster response. This approach targets geographic areas (for example, through visual display of unmet needs or reported service access) or age- and gender-specific population subgroups as an event density. Identification of particular population subgroups in this manner may delimit those who are at risk for certain outcomes such as a disease or malnutrition.
The integration of spatially distributed demographic data with environmental models of disaster risk and vulnerability allows for enumeration and mapping of the population initially at risk of disaster. This is particularly
important in developing countries where current and reliable subnational demographic data are not available. In addition to estimating the affected population, maps of buffer zones of disaster impacted areas, when applied to spatially distributed demographic data, provide a reference population (those exposed to immediate hazard) that not only can be considered in the design of post-disaster population surveys and impact assessments but also can delineate areas for future disaster mitigation and preparedness activities.
Post-disaster assessments are highly varied in methodology and outcomes. Key limitations to assessments in the post-disaster context relate to lack of accurate information on the size and location of the surviving population. Logistical difficulties in the rapidly changing post-disaster context can further limit the ability to make these assessments and include the destruction of infrastructure, limited availability of qualified personnel, transportation, cultural and linguistic barriers associated with working in foreign and unfamiliar environments, and other complications. Ideally, these assessments are population-based surveys that employ appropriate sampling methods to select a representative sample of individuals or households from a larger reference population. In most contexts, the sampling frame comprises displaced disaster survivors and key indicators including morbidity, mortality, living conditions, and immediate survival needs. Estimates of surviving and displaced populations are generally available from local governments and the United Nations in the post-disaster context and serve as the basis for many assessments and the planning of humanitarian assistance.
Spatially distributed demographic data, environmental risk models, and population-based assessments of disaster-affected regions are important tools that can aid decision makers in characterizing vulnerable populations and humanitarian assistance needs in the aftermath of disasters. Integrated assessments that incorporate environmental and social risks, indicators of population status, and spatial data are ideal for the identification of population subgroups and regions that are particularly vulnerable and in need of humanitarian assistance. Recent developments in GIS and spatial analysis, including spatially distributed demographic data and data visualization software, should be considered as new and innovative tools for the communication and provision of information on populations at risk of disasters as well as those in need of humanitarian assistance in the post-disaster context.
Buchanan-Smith, M., and S. Davies, 1995. Famine Early Warning and Response; The Missing Link. London: Intermediate Technology Publications.
Chen, K., R. Blong, and C. Jacobson, 2003. Towards an integrated approach to natural hazards risk assessment using GIS: With reference to bushfires. Environmental Management 31(4):546-560.
Herold, S., M. Sawada, and B. Wellar, 2005. Integrating geographic information systems, spatial databases and the internet: A framework for disaster management. In Proceedings of the 98th Annual Canadian Institute of Geomatics Conference, Canadian Institute of Geomatics, Ottawa, Canada, June 13-15.
IFRC (International Federation of Red Cross and Red Crescent Societies), 2003. World Disasters Report. Geneva: International Federation of the Red Cross and Red Crescent Societies.
Kunreuther, H., 2002. Risk analysis and risk management in an uncertain world. Risk Analysis 22(4):655-664.
Mansor, S., M.A. Shariah, L. Billa, I. Setiawan, and F. Jabar, 2004. Spatial technology for natural risk management. Disaster Prevention and Management 13(5):364-373.
Maxwell D., and B. Watkins, 2003. Humanitarian information systems in the greater horn. Disasters 27(1):72-90.
McEntire, D., 2001. Triggering agents, vulnerabilities and disaster reduction: Towards a holistic paradigm. Disaster Prevention and Management 10(3):189-196.
Simonovic, S.P., and Nirupama, 2002. Role of Remote Sensing in Disaster Management. London, Ontario: Institute for Catastrophic Loss Reduction (ICLR), Research Paper Series No. 2.
Zerger, A., Smith, D.I., 2003. Impediments to using GIS for real-time disaster decision support. Computers, Environment and Urban Systems 27:123-141.
Shannon Doocy received her Ph.D. in international health from the Johns Hopkins School of Public Health in 2004. She is currently a research associate with the Johns Hopkins University Center for Refugee and Disaster Response. Her research focuses on populations affected by disasters and conflict, including both refugees and internally displaced populations. Within the emergency context, her areas of interest include population-based assessments, mortality, nutrition and food security, livelihoods and cash interventions, and monitoring and evaluation. Recent research includes coping capacity and vulnerability in Ethiopia; an effectiveness trial of a point-of-use water treatment product in camps for displaced populations in Liberia; the demographic and nutritional assessment of conflict-affected populations in northeastern Sudan; tsunami-related research and program evaluation in Aceh, Indonesia; and technical assistance to projects in northern China.
COGNITIVE AND INSTITUTIONAL LIMITS ON COLLECTING AND PROCESSING DATA ON POPULATIONS AT RISK: PRELIMINARY REFLECTIONS ON SOUTHERN AFRICAN RESPONSES TO DISPLACEMENT
Loren B. Landau, Wits University’s Forced Migration Studies Programme, Johannesburg, South Africa
This paper explores how cognitive and institutional structures limit researchers’, policy makers’, and service agencies’ efforts to collect and process critical information on mobile and displaced populations. It be-
gins from the assertion that policy and research in all fields are shaped by assumptions and institutional structures. These boundaries, along with practical difficulties, embedded sets of ethical values, and institutional imperatives, work together to shield key aspects of vulnerable populations from consideration.
This paper explores two concerns related to displaced populations in southern Africa. Although sound research design, representative sampling, and objectivity are the hallmarks of good academic and policy-oriented research (Jacobsen and Landau, 2003), those of us attempting to meet these standards quickly discover that research soundness has little to do with our influence on policy and practice. This is generally frustrating for academic researchers and becomes even more so when the research is commissioned by governments and aid agencies. When our research is used, it rarely serves to challenge potentially faulty fundamental principles or to introduce previously unrecognized dimensions of potentially vulnerable subgroups. Instead, we often find it merely serves to refine or harden existing policy parameters. Indeed, commissioned research is often of such limited scope that it could hope to do little else. Such tendencies seem all the more pronounced in discussions of vulnerable subpopulations in times of crises. The narrow scope of data collection and the underuse and abuse of data emerge from a combination of institutional and cognitive constraints. Using examples from research and policy making regarding refugees in southern Africa, this paper outlines some of the most significant ways in which the close relationship between policy discussions and research limits our ability to learn about displaced persons.
The remaining pages proceed through four stages in highlighting the institutional and cognitive limits on production and consumption of knowledge on migration, displacement, and humanitarianism in southern Africa. The first draws on sociological treatments of formal organizations and (superficially) literature on the philosophy of science and construction of academic knowledge. The second segment discusses how stylized and incomplete information—often taking the form of unverifiable or unverified “truths”—exerts particular power in discussions surrounding immigration and asylum. The third section identifies five tenets that reflect and shape public attitudes, policy making, and research on displacement in southern Africa. Although many of these are empirically suspect, they are deeply embedded within cognitive schema and continue to shape parameters of debate. The paper ends by suggesting a pragmatic and critical approach to research on displacement and humanitarianism that recognizes (and challenges) our own cognitive limitations and the mechanisms through which knowledge is processed, consumed, and implemented as policy.
Much of the following analysis is based on personal experience gained
through participation in primary research, analysis, and policy discussions on migration, displacement, and humanitarianism in eastern and southern Africa. In particular, I draw on research and discussion with policy makers and aid workers during a year and half of fieldwork in western Tanzania (1999-2000) and through my role at the Forced Migration Studies Programme in Johannesburg and as chair of the executive committee of South Africa’s National Consortium of Refugee Affairs. In these latter roles, I have had opportunities to interact extensively with migrants, service providers, advocates, and government officials from throughout southern Africa. Many of the findings reported here are drawn from this participant observation in professional meetings and workshops with government officials in local, provincial, and national governments. They are, as such, incomplete impressions and should be read as provocations for consideration, not as firm conclusions.
ORGANIZATIONAL AND COGNITIVE LIMITS ON RESEARCH AND POLICY
In emergency situations, policy makers and operational agencies often justifiably complain that the lack of sound information on the populations they are trying to assist compromises their decisions. The scarcity of data is problematic—a point I return to later—but even when data are available, values and political priorities are often equal determinants of policy outcomes. Value systems are, after all, the keys to present problems that need to be solved and that provide many of the criteria for evaluating possible solutions. Where political systems are fragile and legitimacy is scarce, local leaders may see emergency assistance primarily as a political, not a humanitarian, resource. The greater the symbolic currency at stake, the more likely are such factors to determine an outcome, although such criteria may be hidden by post hoc justifications couched solely in technical terms. These are rational outcomes—based on considered cost-benefit analysis—but outcomes that result from priorities different from those held by international agencies and donors. Understanding outcomes requires that we account for them.
The values informing various policy fields add symbolic value to responses and frame the range of “legitimate” policy options. In domestic and international humanitarianism, elements of this normative framework are visible in legislation, treaties, and principles (e.g., SPHERE standards). The Sphere project aimed at developing minimum standards for humanitarian relief operations (for further information, see http://www.sphereproject.org). Reflecting the language of the 1951 UN Refugee Convention that established it, the office of the UN High Commissioner for Refugees (UNHCR), for example, will often ignore any aspect of humanitarian assistance
that is considered overtly political or developmental. The research the UNHCR commissions similarly focuses on displaced persons’ short-term “protection” needs, leading to a focus on emergency settings. Moreover, the UNHCR focus is almost entirely on those areas in which humanitarian action is likely: camps and border-areas. The fragmentation of agencies and organizations into those concerned with health, sanitation, security, or food creates further analytical boxes in which questions are framed and answered. Coordinating bodies may partially transcend these divisions, but they also bring with them interests and cognitive limitations. Given the close relationship between research and policy in the humanitarian field, these paradigms or problematiques often serve to identify problems worthy of scholarly attention. This has been well noted in studies of “development,” where dominant policy discourses define a set of concerns and metrics for measuring progress (see Escobar, 1995; Leys, 1996; Ferguson, 1994). Such frames are often even more important in humanitarian fields where the framework for data collection is compressed and the evident need to save lives overrides careful conceptualization and other considerations. That the language and policy that characterize southern African responses to asylum seekers so evidently echo those in Europe, Australia, and North America, illustrates the power of discourse in shaping policy frames. Regardless of their precise origins, researchers wishing to say anything about immigration or asylum cannot dodge these considerations but consider how the parameters of these paradigms shape research questions, findings, and dissemination strategies.
Contesting institutionally embedded wisdom is, of course, easier said than done. Most importantly, it requires levels of self-awareness that elude many researchers. Those explicitly working on behalf of government or aid agencies may, moreover, have few reasons to challenge the paradigms in which they work. For those without the financial security that such affiliations provide, stepping outside paradigmatic boundaries may mean losing crucial research funding and exclusion from research forums. Such exclusion may be formal and conscious, or it may occur simply because analyses asking unfamiliar questions or using unusual methods become unintelligible to policy makers. Paradoxically, the more specialized and trained the consumers are, the narrower and more formulaic is the information they require, whether that information takes the form of economic indicators, epidemiological projections, or meteorological readings. Their training and relative isolation within specialist departments can create situations where those with the authority to make decisions and positive change “often cannot even conceive of appropriate alternatives” (Powell and DiMaggio, 1991:11). Rather than draw on “radical” analyses to question the principles underlying policy (or the ways they make decisions), organizations are far more likely to “muddle through”—to make minor adjustments in
the face of failed policy in ways that leave structures and imperatives intact (see Lindbloom, 1959; Haas, 1990; Crozier, 1964; Argyris, 1982).
There are also significant limitations on organizations’ ability to process the information they do collect. Simon’s (1957) work on “bounded rationality” describes a tendency to “satiscfice”: for decision makers to accept the first plausible or acceptable solution rather to consider a range of alternatives and likely outcomes. Cohen et al.’s (1972) “garbage can” metaphor of decision making in which organizations are bins filled with concerns looking for solutions, solutions looking for issues to which they might be attached, and “decision makers looking for work.” The consequence is that policy prescriptions and problems are attached due to a mix of chance and calculation. Where decisions must be made quickly, chance will consistently trump calculation. This does not suggest chaos or total randomness within organizations or agencies that collect and process data on vulnerable subpopulations. Rather, it argues that there are always random elements and institutional constraints on how questions are framed and how they are answered. While these tendencies are present within all policy fields, they deserve particular attention in attempting to develop effective, evidence-based responses to humanitarian crises. Through its discussion of limitations on research and policy discussions on migrant populations in southern Africa, the following section illustrates these limitations and variables.
HEIGHTENED CONSTRAINTS IN MAPPING DISPLACEMENT
No policy arenas exist in which scientific inquiry consistently dominates symbolism, values, and political priorities. Although further consideration would reveal others, this paper speaks of four primary reasons why cognitive limitations, bounded rationality, political legitimacy, and institutionally embedded assumptions are particularly likely to shape research and analysis of displaced populations. Schmidt (2005) raises many of the points discussed here in a broader discussion of the relationships between researchers and the displaced, policy makers, and aid agencies.
Information is scarce and practically difficult to collect. Conventional wisdom is difficult to dislodge even when confronted with the most scientifically compelling, empirical counterevidence. Armed with patchy or irregularly collected data, the chances for effectively challenging policy presumptions are further limited. Southern Africa’s shortage of trained demographers and migration specialists is partially to blame for this (e.g., Global Commission for International Migration, 2005). However, the region’s social transformation coupled with its extended, highly porous borders makes it all but impossible to conduct accurate surveillance of human movements. As elsewhere in the world, people making these journeys often have compel-
ling reasons for remaining bureaucratically invisible (e.g., to avoid deportation, harassment, or discrimination). This applies not only to the millions of economic migrants within the region, but also to refugees who wish to avoid being warehoused in camps for years on end. Domestic migrants or those who have been internally displaced may have fewer reasons to hide, but they are also less likely to traverse established surveillance sites (e.g., borders, immigration check points) and are unlikely to encounter relatively well-resourced aid agencies that have incentives to count them. Moreover, states often actively suppress the numbers of internally displaced persons within their borders in an effort to promote their international legitimacy.
The challenges of generating good information are not limited to monitoring. Developing accurate projections on migration and displacement patterns would not only mean counting those on the move, but coding those movements in ways relevant to the policy discussions. These may include determining if individuals are involved in circular, seasonal, or permanent migration, and whether they are in transit, have been displaced or otherwise forced to move, or are simply tourists. It also means finding ways of predicting seasonal price fluctuations for labor and agricultural commodities, droughts, famines, wars, and other changes in broader economic and social conditions. Were resources available for this, the number of variables at work would limit any model’s reliability and predictive power. At most, one might hope to derive generalized trends with limited predictive power.
Accurate information about immigration and migration is not easily processed through standardized bureaucratic or political channels. Many of the same reasons that make it difficult to collect and analyze information on migration—its unpredictability, its multiple causes and effects, the desire of those moving to remain invisible—make it tricky for planners to develop empirically informed policy responses. Scott’s (1992) work on rational organizations—a category that includes most government bureaucracies and many large international agencies—is built on systems for regularly evaluating information and considering the consequences of various responses. Although few bureaucracies are as orderly as this model suggests, they nevertheless manage to filter out “noise”: inconvenient or irregular information. The more irregular the information required for making sense of the outside world, the more difficult it is to process this information, and the more bounded the organization’s rationality will become.
Given their organizational design, it is unlikely that governments and other large agencies could respond quickly and effectively even if able to overcome the practical problems of data collection and analysis. Budgetary and planning cycles, for one, make it difficult to prepare for populations that might arrive (or disappear). Perhaps more critically, the need for politicians to show short-term gains and policy responses makes it less likely that they will dedicate resources to building effective emergency response
mechanisms when crises are so uncertain or may only be critical issues for their successors. Problems in responding to Hurricane Katrina illustrate the U.S. vulnerability to a surprise storm, despite prior efforts to coordinate its emergency response apparatus. In countries facing severe research scarcities—a category including most African states—the problems are even more acute. When a target population is not considered a political constituency, as in the case of migrants and refugees, the likelihood is even less that money will be earmarked for planning.
Researchers also suffer for the unpredictability that often characterizes instances of displacement. For academics, teaching and funding cycles make it difficult to be on the scene when refugees or flood victims arrive. Even for agency or organization researchers, the challenges of enumerating mobile populations tend to lead to rough estimates and a reliance on assumptions or experience gained elsewhere. Consequently, most representative research favors protracted humanitarian emergencies or resettlement schemes where the displaced are in relatively stable settings under conditions that are easier for outside researchers to navigate.
Migration and displacement threaten deeply held values linking spatial origins with rights and identities. Governments dedicate themselves to managing people and processes within a carefully defined geographic space: villages, municipalities, districts, provinces, or even countries. Such delimitations of authority and responsibility are not only administrative, but are often linked to more fundamental definitions of community. This is most obvious at the national level, where the politico-administrative distinction of citizenship reflects membership in a quasi-fictional nation, a group whose members claim almost exclusive rights to the country’s territory and resources (see Arendt, 1958; Agnew, 1999; Malkki, 1992).
Under such circumstances, exclusionary social responses to newcomers claiming space and resources may degrade established legal protections, and people almost inevitably code and label those from outside their communities or countries as threats to prosperity, security, or sovereignty. As noted, the conflation of migration and security in the post-9/11 era means that we are limited both in the scope of our discussions and in our power to convince governments to break from the status quo unless we are advocating further restrictions on migration and asylum. That a migrant presence is so often blamed for political failings makes it less likely that politicians will support even the most well-informed policy prescriptions if they facilitate movement, provide migrants with critical social services, or promote economic and political integration.
The connection between migration and values also influences how researchers frame questions. Debates about whether immigrants have the right to continue female circumcision or genital mutilation, for example, are so difficult that many simply dodge the issue. Even choosing one term
over another can unwittingly position a person on one side of a highly contentious divide. Elsewhere, commitments to promoting the interests of the displaced among hostile policy makers may cause researchers to ignore or suppress data that portrays these groups negatively. Those who do so, or wander into any other such contentious circumstances, must be prepared for attacks from their peers and colleagues. This may be less so with government- or agency-commissioned research, but here, too, the values of the organization are likely to influence the questions asked and the kinds of data produced. Perhaps most fundamentally, many researchers simply accept the unnaturalness of refugees and implicitly endorse strategies to restrict them to camps or otherwise limit their mobility and livelihoods. Failing that, displaced populations are often treated as anomalous populations disconnected from the people and places surrounding them. Because these analytical distinctions are almost never reflected empirically, their conclusions suffer, as does what is known about displaced persons.
Academic and policy fields are interdependent. The close connections between displacement scholars, policy makers, and aid agencies create mutually reinforcing interests. Those who help shape a government or aid agency policy are far less likely to criticize it. The tendency to defend progeny—whatever form they take—cannot be dismissed. Other researchers are well aware that producing a radical or even undiplomatically worded critique may be costly in terms of the response they receive, including access to research funding or consulting projects. Poorly paid African researchers—those potentially most attuned to the sociopolitical subtleties of a displaced subpopulation—are all the less likely to rock the boat.
In cases where “hard” information is absent, difficult to collect, or carefully managed by public (or private agencies), rumor, paranoia, and political pronouncements often replace carefully established and verified causal explanations. Indeed, without empirical substantiation, socially and politically derived imperatives are the only possible bases available for policy formulation. Even more normal research, however, is bounded by cognitive limits and presuppositions. Similarly, institutions are often ill-equipped to process information that challenges fundamental operational principles or takes on unfamiliar forms. The following section outlines some of the parameters that structure research and policy discussions around displacement in southern Africa.
PARAMETERS OF THE POLICY-RESEARCH DISCOURSE ON DISPLACEMENT IN SOUTHERN AFRICA
As the previous pages discuss, decision makers and researchers tend to see only what they have predetermined is relevant or important to their policy deliberations. The following paragraphs identify five key tenets that
help frame research and policy on migration and displacement in southern Africa. In many instances these are myths around migration and displacement that are easily falsified through reasoned analysis. The point here is not whether these tenets are objectively true, but rather to illustrate that even when they are not, they frame discussions and establish a demarcation and nodal point within policy and scholarly deliberations. Although informed by my experience working in southern Africa, these tenets are likely to be familiar to those working elsewhere.
Tenet One: Immigration is Structured and Controlled by National Policy
Southern African immigration and asylum policies are founded on a misplaced faith in governments’ ability to prevent cross-border movements. The focus on identity documents, detention, and deportation is illustrative of this, as is the need for asylum seekers and refugees to report regularly to designated offices. Due in part to this belief, recent discussions about harmonizing regional instruments have tended to shy away from facilitating movements and have instead put forward new measures to control them. These include, inter alia, proposals to create asylum seeker camps and a computerized database listing all people crossing borders, refugees and asylum seekers entering the region, so that individuals may be traced and prevented from “asylum shopping” or undertaking irregular movements beyond state regulation. Apart from the ethical problems with such propositions, there are few reasons to believe that any country in the region has the capacity to track such movements when so many citizens—who have few incentives to hide from the state’s eyes—continue to live without identity documents and effectively outside of state regulation. The result is rather that migrants will find new ways of becoming invisible and the system will be further opened to corruption and irregularities (Landau, 2005). Perhaps more fundamentally, the widespread presupposition continues that national policy structures migration dynamics. Although policy is important, the power of law and the influence of policy are minimal throughout much of Africa. Rather, people tend to live largely outside state regulation. What comes to matter in these instances is the response of local leaders and peoples (Misago and Landau, 2005).
The focus on control and policy has also infiltrated the research community. The fixation on counting and identifying migrating population flows stems in part from a belief that doing so could help regulate and structure people’s movements or is necessary for developing a policy response. While such information may be useful—if the ability exists to process it—policy deliberations in many other areas continue from equally shaky empirical foundations. Unfortunately, migrant flows tend to conform to something
approximating the Heisenberg uncertainty principle where the act of counting and categorizing may change the nature of what is being studied. There is also a danger of creating legal categories for migrant populations (e.g., refugee, asylum seeker, illegal) that inappropriately draw attention away from commonalities and relationships among them (see Zetter, 1991). Most importantly, the assumption that migrant populations may best be helped through formal responses has drawn attention away from the tens of thousands of people who receive no such assistance and the factors influencing their reception by host communities.
Tenet Two: The Deluge
The fixation on control and regulation mentioned above in part stems from a sense that the region is being flooded by migrants and refugees. The ongoing conflict in the Democratic Republic of Congo, crisis in Zimbabwe, HIV/AIDS, and poverty-economic transformation throughout the region do, indeed, continue to spawn new movements of people. The fear of these movements, however, is often disproportionate to the numbers. This is most evident in South Africa where the unmanaged arrival of foreigners starkly contrasts with apartheid-era controls on human mobility. The governmentfunded Human Science Research Council’s (HSRC’s) alarmist figures of 2.5 to 5 million illegal immigrants continue to shape public discussion, despite the council’s formal retraction (Crush and Williams, 2001). That the HSRC figures remained on the Department of Home Affairs (DHA) web site throughout Mangosotho Buthelezi’s term as minister (1994-2004) is indicative of how research confirmed deeply held suspicions and was then used to justify them. That said, government officials have not restricted themselves to such “scientific” data, but have independently begun citing figures of between 7 and 8 million “illegals” in the country. These elevated numbers (and their presumed effects on everything from crime to food prices) have justified a series of disproportionate responses. Botswana has, for example, recently erected an extended electrified fence along its border with Zimbabwe, ostensibly to prevent the spread of hoof and mouth disease. South Africa continues to make plans to establish refugee camps—something prohibited except in cases of mass influx—among other efforts to restrict the movement and rights of nonnationals.
While South Africa’s economic liberalization and regional integration have added new dimensions to the region’s long-standing patterns of labor migration (see Rogerson, 1995), more objective analysis reveals figures of a different order of magnitude. The most recent census (2001), for example, indicates that there were 345,161 non-South African Africans in the country. Although this is certainly an undercount, the total number of foreigners is likely to be 500,000-850,000 (Crush and Williams, 2001); this
number is possibly slightly higher now, given the trouble in neighboring Zimbabwe. These figures are significant, but by no means overwhelming. Similarly, widely touted, figures of 2-3 million Zimbabweans in the country are simply implausible (2-3 million represents more than 20 percent of that country’s population). Even in those urban neighborhoods most affected by immigration, only one-quarter of the population (at most) is foreign born (Leggett, 2003). This represents a sizable increase compared to a decade ago, but hardly compares with cities like Toronto, Canada, where more than 40 percent of the total population is foreign born. However, perceptions—not carefully considered estimates—shape policy. Data on other countries are even harder to collect, opening additional space for myths and paranoid pronouncements.
The disparities outlined above have a threefold significance. First, through constant repetition, the elevated figures have taken on the status of Truth, something all but impossible to dislodge. In meetings where policy makers, bureaucrats, and the police are asked to justify their presumptions, they either simply reply that they know, refer to other government documents estimating migrants, or cite outdated and discredited scholarly studies. The HSRC report outlined previously features prominently in these discussions. The second reason these discussions are important is with regard to the way policy makers use research to legitimate already-known truths. In this instance, numerous studies—including some by their own statistics bureau—have been ignored simply because they do not reach findings that resonate with long-standing presumptions. When these studies are acknowledged, policy makers demonstrate an uncharacteristic awareness of their methodological limitations and dismiss them as undercounting the number of foreigners. In such instances, officials will almost always draw on anecdotal evidence of a neighbor who is employing an illegal domestic worker or of a relative who was robbed by a Zimbabwean. The last point I wish to raise through this example is how this fixation on numbers has served to structure the nature of research. Efforts continue on the part of Statistics South Africa and others to provide more precise estimates of the numbers of foreigners. Few, however, have asked whether such precise information is needed or if the South African government has the interest or capacity to respond to anything other than rough estimates. The fixation on numbers and on categorizing migrants by legal status has drawn attention away from other issues related to the composition and dynamics of migrant communities.
Tenet Three: Refugees Are Inherently Needy and Vulnerable
Not surprisingly, attitudes toward refugees and other migrants are often contradictory, with these tensions reflected in both policy pronouncements
and research. Within this discourse, refugees and immigrants are implicated in destabilization of the region’s economies while fostering crime, disease, and other social ills. Paradoxically, the power they exercise is rooted in their presumed vulnerabilities and inability to care for themselves. These sentiments were evident early in the post-apartheid era when, in his first speech to parliament following his appointment as the South African Minster of Home Affairs Mangosuthu Buthelezi reflected (and helped frame) such beliefs by proclaiming: “If we as South Africans are going to compete for scarce resources with millions of aliens who are pouring into South Africa, then we can bid goodbye to our Reconstruction and Development Program [RDP].” He went on to argue:
The employment of illegal immigrants is unpatriotic because it deprives South Africans of jobs and … the rising level of immigrants has awesome implications for the RDP as they will be absorbing unacceptable proportions of housing subsidies and adding to the difficulties we will be experiencing in health care. (Reitzes, 1994)
Such perspectives are also present at the local level. In his State of the City 2004 address, for example, Johannesburg’s executive mayor argued:
In keeping with the international trend of growing migration, our city has become a magnet for people from other provinces, the African continent and indeed the four corners of the world. While migrancy contributes to the rich tapestry of the cosmopolitan city, it also places a severe strain on employment levels, housing and public services.
In countries where 40 percent or more of the population is unemployed, an inevitable result is some resentment against any group that may potentially fill jobs or push down the price of labor. However, little evidence exists that migrants are either needy public wards or stealing jobs from South Africans. Although mine and agricultural labor (formally imported through formal guestworker schemes) have disempowered South African workers and unions, new immigration patterns appear to be increasing job opportunities for South Africans. Wits University research in inner-city Johannesburg, for example, found that non-South Africans were far more likely to have hired someone to work for them in the past year than the South Africans amongst whom they lived. While just 20 percent of South Africans report having paid someone to do work for them, 34 percent of migrants surveyed had. Even more significantly, more than two-thirds (67 percent) of those hired by migrants were South Africans. Hunter and Skinner’s (2003) work in Durban also identifies a positive economic impact from immigration, and the city government has adopted policies that allow nonnationals to apply for street-trading permits. Bakewell’s (2000)
work on self-settled refugees in Zambia confirms the potential benefits of displacement for host communities. Internationally, there is evidence that immigration provides a net benefit to national economies, although some groups are always likely to face negative consequences (see Simon, 1995; Smith and Edmonston, 1997).
Little evidence has been presented to back claims that nonnationals represent a significant drain on the state’s financial resources. Summarizing work done in South Africa and elsewhere, Meintjies (1998:20) reports:
Immigrants are, in fact, net contributors, not parasites. Immigrants are, on average, healthier, more energetic and better educated than people in the host population. Consequently, they draw comparatively less on social welfare and other social services. Many pay tax and, through their entrepreneurship, make a positive injection into local economic development.
In the South African context, refugees are more likely than other immigrants to make such contributions for the simple reason that while Zimbabweans and Mozambicans can easily enter the country, those from elsewhere must already be endowed with resources for extended travel. This creates a selection bias in favor of the wealthy, educated, and entrepreneurial. This is not to deny that the presence of additional people—whatever their origins—can burden public services. However, given the relatively small number of immigrants using these services—and their ability to contribute economically—it makes little sense to single them out as a primary cost to government or a threat to South Africans’ economic prospects. Throughout the region, cities are struggling to manage burgeoning populations, but most of those coming to the cities are citizens. Local capacity and budgets are not being overwhelmed by immigrants and refugees.
Regardless of refugees and other migrants’ actual impact on the southern African economy, the dual presumptions that refugees are inherently needy or vulnerable and steal jobs continue to shape how scholars produce and disseminate information about refugees. Much of the debate around migrants generally is about whether they have a positive or negative impact on South Africans. Such a question cannot be answered definitively and depends largely on the level of analysis and the metrics used (see Landau, 2003). Moreover, given the extraordinary dynamism within the region, identifying the impact of a single variable on the economy is almost impossible. Wherever possible, advocates will try to minimize the costs to host societies out of fear that this will harden attitudes (and policies) toward refugees. Paradoxically, many of those working in the field—whether in academia or on behalf of aid organizations—share assumptions about vulnerability or have interests in highlighting the need to enhance funding flows to cash-starved assistance programs. Research commissioned by aid agencies certainly tends to look for instances of exploitation and marginality. Since
little other research exists on communities living in remote or inhospitable areas, these findings give an impression of generalized vulnerability. Many policy-oriented researchers, even when they see self-sufficiency and adaptation, are unwilling to publish work highlighting refugees’ economic capacities out of fear that this will cause them to be classed as economic migrants or will reduce funding for assistance programs. The result is an almost bimodal portrayal of refugee populations. In some forums they are economic assets (or at least not liabilities), on the other hand, they are a vulnerable population that will not survive without external assistance.
Tenet Four: Women Are Universally Victims of Displacement
Nowhere is the discourse of vulnerability more pronounced than in discussion of refugee women. As the numbers of unaccompanied women (i.e., those without spouses or male offspring) increase, so too have deliberations around trafficking, prostitution, and collapsing families. Without denying the risks women face during displacement, assumptions are made about women’s experiences that blind both policy makers and researchers to aspects of their realities. These blinders are not surprising given the bias toward research in refugee camps and settlement schemes, sites that disproportionately attract vulnerable women. Assumptions of vulnerability, however, often cause scholars to overlook the active role women play in the migration process.
New research is slowly drawing attention to a broader range of women’s experiences and revealing activities that have previously been invisible or that reframe vulnerability in more empowering language. While scholarship on female migration often speaks of women moving only after urging from male relatives or after being abandoned by them (see Gugler and Ludwar-Ene, 1995; Adepoju, 1995), many women (refugees and otherwise) in the region’s urban centers independently chose to move and have done so without male guardianship (Kihato, 2005; Raimundo, 2005; Sander and Maimbo, 2003; Adepoju, 2006). Sexual networking provides an additional example. Although such activities present considerable dangers in an era of HIV/AIDS, many women speak nonchalantly about their ability to generate income. As Augustin (2005:378) notes, “Many thousands of women who more or less chose to sell sex as well as all women working in domestic or caring services are ‘disappeared’ when moralistic and often sensationalistic topics are the only ones discussed.” Others simply speak about how displacement has freed them of patriarchal demands levied on them at home. Even women’s stories of vulnerability and victimization can be valuable resources. Given their audience’s expectations, the articulation of predictable narratives of violent exclusion and hardship—whether founded or false—may elicit sympathy, material resources, or other economic opportunities.
Tenet Five: Migration Policy Is Only a Concern of National Governments
In describing the character of the nation-state system, Arendt (1958:279) writes that sovereignty is nowhere more absolute than in matters of “emigration, naturalization, nationality, and exclusion.” Indeed, there continues to be a widespread belief that as an issue of state sovereignty and international relations, immigration and asylum policy should be formulated and implemented solely by national government departments. However, as political decentralization and devolution continue, responsibility for responding to refugees and asylum seekers is, de facto, being transferred to the local governments already charged with overseeing and spearheading community development. Although there are examples of cooperation among government officials across the range of departments and levels (e.g., interministerial committees, links between the army and the police), responsibility for the immigration and asylum regime remains a matter almost fully within the bailiwick of the national government The tendency of international agencies to deal only with ministries enforces these boundaries.
Apart from the inability of national governments to respond at the local level, this tenet has implications for the way research is conducted and disseminated. Throughout the region, the focus of policy deliberation remains almost entirely regional or national, and researchers associated with these processes tend to focus only on aggregate statistics and blanket policy considerations. Rarely are efforts made to disaggregate refugee populations and to map spatial variations in gender, age, experience, or vulnerability. Where localized studies exist, the data they generate are typically proprietary or difficult to integrate with similar studies elsewhere. Similarly, little research attempts to integrate local considerations into national policy debates. Cape Town and Johannesburg have begun considering how migration may affect local planning, but they have yet to dedicate significant resources to understanding the dynamics of refugees and asylum seekers living in their cities. Elsewhere in the region—where resources and researchers are scarcer—efforts have been even more minimal. Perhaps more importantly, the continued international focus on camps and areas receiving international aid draws researchers away from what may well be larger numbers of self-settled refugees whether in urban or rural settings.
CONCLUSION: HOW DO WE LEARN (MORE) ABOUT MIGRANTS AND DISPLACED PEOPLE?
There is no easy way to conduct research on displaced or mobile populations. The movements of such people are almost impossible to predict, although we could do better. That the populations are so highly dynamic—normal death and birth rates combined with special health risks
and mobility—make it almost impossible to create reliable data sets with a shelf-life of more than a few weeks. In addition to these obstacles, we must negotiate hostile or unsupportive leadership and find funds to support our work. Our efforts may also be stymied by host governments, aid agencies, and refugees who all have reasons to distort the size and character of these subpopulations themselves. Making projections may be possible, but given the variables at work, their accuracy is far from guaranteed. Overcoming the technical obstacles also does little to ensure that the data generated will provide a holistic view on the populations concerned. Moreover, experiences can lead researchers to believe that the more comprehensive and timely the data produced, the less likely it is to be used to influence policy decisions.
Rather than ending with specific suggestions on improving how data are collected and used, I wish to conclude with three generalized recommendations.
Work toward mechanisms for objective tracking of forced migration while recognizing the limitations of such measures. Accepting that displacement will continue means developing institutional mechanisms within the African context that treat it as a normal rather than truly exceptional phenomenon. This not only means coordinating emergency response institutions, but creating technically equipped research teams with ready funding that can be deployed rapidly to emergency situations and to areas where displaced populations are living.
Rather than relying on politicians and or self-interested operational agencies to generate estimates of migration and its impacts, efforts are needed to train population scientists that can collect longitudinal and comparative data throughout the region. In South Africa, for example, there are fewer than 20 full-fledged demographers; the number in Mozambique can be counted on one hand. Most of these have little expertise or interest in migration, but instead focusing on fertility and mortality. There is, therefore, a need to bolster local expertise. This expertise should not, however, be in technical demography alone. In order to meet the objectives outlined above, training and research must pair those with technical demographic skills with those with skills in anthropology and organizational sociology. Until these groups work together, it is unlikely that researchers will be in a position both to identify and to challenge the parameters of existing approaches. Moreover, by complementing demographic analysis with studies of organizations and institutions, we may better target our dissemination strategies in ways that may positively influence policy and practice for the benefit of vulnerable subgroups.
To move beyond existing lines of inquiry or promote institutional adaptation (rather than learning), scholars must identify the social and
political arrangements that buttress the existing policy and research regime. Simply producing new data is not enough; there is also a need to consider what is motivating the creation of myths, their propagation, and their maintenance. Challenging these tenets means learning to present information that is at once legible to institutions—private, public, academic, or activist—but does not simply confirm that which they already know. As a corollary, there are good reasons to move beyond simply generating empirical information about migration—although this is certainly critical—and include efforts to trace how those data are produced and used. To do otherwise may provide scholars with short-term influence and professional benefits, but may risk our long-term effectiveness in assisting the displaced.
Adepoju, A., 1995. Migration in Africa. In J. Baker and Tade Akin Aina (eds.), The Migration Experience in Africa. Sweden: Nordic Africa Institute.
Adepoju, A., 2006. Internal and international migration within Africa. In P. Kok, D. Gelderblom, J.O. Oucho, and J. van Zyl (eds.), Migration in South and Southern Africa: Dynamics and Determinants. Pretoria: Human Sciences Research Council.
Agnew, J., 1999. Mapping political power beyond state boundaries: Territory, identity, and movement in world politics. Millennium 28:499-522.
Augustin L.M., 2005. Migrants in the mistress’s house: Other voices in the “trafficking” debate. Social Politics: International Studies in Gender, State, and Society. 12(1):96-117.
Arendt, H., 1958. The Origins of Totalitarianism, 2nd edition. New York: Meridian Books.
Argyris, C., 1982. How learning and reasoning processes affect organizational change. In P.S. Goodman (ed.), Change in Organizations: New Perspectives on Theory, Research, and Practices. San Francisco: Jossey-Bass.
Bakewell, O., 2000. Repatriation and self-settled refugees in Zambia: Bringing solutions to the wrong problems. Journal of Refugee Studies13(4):356-373.
Cohen, M.D., J.G. March, J.P. Olsen, 1972. A garbage can model of organizational choice. Administrative Science Quarterly 17(1):1-25.
Crozier, M., 1964. The Bureaucratic Phenomenon. Chicago: University of Chicago Press.
Crush, J., and V. Williams, 2001. Making up the numbers: Measuring “illegal immigration” to South Africa. Migration Policy Brief No. 3. Cape Town: Southern Africa Migration Project.
Crush, J., and V. Williams, 2003. Criminal tendencies: Immigrants and illegality in South Africa. Migration Policy Brief No. 10. Cape Town: Southern Africa Migration Project.
Dewey, J., 1938. Experience and Education. New York: Collier Books.
Escobar, A. 1995. Encountering Development: The Making and Unmaking of the Third World. Princeton: Princeton University Press.
Feldman, M., and J. March, 1981. Information in organizations as signal and symbol. Administrative Science Quarterly 26(2):171-186.
Ferguson, J., 1994. The Anti-Politics Machine: Development, Depoliticization, and Bureaucratic Power in Lesotho. Minneapolis: University of Minnesota Press.
Foucault, M., 1972. The Archaeology of Knowledge. London: Tavistock.
Global Commission on International Migration, 2005. Report on Regional Hearing for Africa held in Cape Town, South Africa, 28 February-1 March. Geneva: Global Commission on International Migration.
Gugler, J., and Ludwar-Ene, 1995. Gender and Migration in Africa South of the Sahara. In J. Baker and Tade Akin Aina (eds.), The Migration Experience in Africa. Sweden: Nordic Africa Institute.
Haas, E.B., 1990. When Knowledge is Power: Three Models of Change in International Organizations. Berkeley: University of California Press.
Hunter, N., and C. Skinner, 2003. Foreigners working on the streets of durban: Local government policy challenges. Urban Forum 14(4).
Jacobsen, K., and L. Landau. 2003. The dual imperative in refugee research: Dome methodological and ethical considerations in social science research on forced migration. Disasters 27(3):95-116.
Kihato, C., 2005. Invisible lives, inaudible voices? The social conditions of migrant women in Johannesburg. Paper presented at the African Migration Alliance Workshop, Pretoria, South Africa, March 10-11.
Landau, L.B., 2003. Beyond the losers: Transforming governmental practice in refugee-affected Tanzania. Journal of Refugee Studies 16(1):19-43.
Landau, L.B., 2005. Immigration and the state of exception: Security and sovereignty in refugee-affected Africa. Millennium Journal of International Studies 34(2):325-348.
Leggett, T., 2003. Rainbow Tenement: Crime and Policing in Inner Johannesburg. Monograph No. 78. Pretoria: Institute for Security Studies.
Leys, C., 1996. The Rise and Fall of Development Theory. London: James Currey. Lindbloom, C., 1959. The science of “muddling through.” Public Administration Review 19:79-88.
Malkki, L., 1992. National geographic: The rooting of peoples and the territorialization of national identity among scholars and refugees. Cultural Anthropology 7:24-44.
Meintjies, F., 1998. Immigrants are people like us. Sunday Times Business Times, September 20.
Misago, J.P., and L. Landau, 2005. Responses to displacement in Africa: The irrelevance of best practice. Conflict Trends 3:4-8.
Powell, W.W., and P.J. DiMaggio, 1991. Introduction. In W.W. Powell and P.J. DiMaggio (eds.), The New Institutionalism and Organizational Analysis. Chicago: University of Chicago Press, pp. 1-38.
Raimundo, I., 2005. Interview with Researcher, Centre for Population Studies, Eduardo Mondlane University, Maputo, Mozambique (September 5, 2005).
Reitzes M., 1994. Alien issues. Indicator SA 12(Summer):7-11.
Rogerson, C.M., 1995. Forgotten places, abandoned places: Migration research issues in South Africa. In J. Baker and T. Akin Aina (eds.), The Migration Experience in Africa. Sweden: GOTAB:109121.
Sander C., and S.M. Maimbo, 2003. Migrant labor remittances in Africa: Reducing obstacles to developmental contributions. Africa Region Working Paper Series No. 64.
Schmidt, A., 2005. ‘I know what you’re doing’, reflexivity and methods in forced migration studies. Paper presented at the 9th International Conference of the International Association for the Study of Forced Migration. São Paulo, Brazil, January 9-13.
Scott, W.R. 1992. Organizations: Rational, Natural, and Open Systems. Englewood Cliffs, N.J.: Prentice Hall.
Simon, H.A., 1957. Administrative Behaviour. New York: MacMillan.
Simon, J.L., 1995. Immigration: The Demographic and Economic Facts. Washington D.C.: The Cato Institution and National Immigration Forum, December 11.
Smith, J.P., and B. Edmonston (eds.), 1997. The New Americans: Economic, Demographic, and Fiscal Effects of Immigration. Washington, D.C.: National Academy Press.
Zetter, R., 1991. Labelling refugees: Forming and transforming a bureaucratic identity. Journal of Refugee Studies 4(1):39-6.
Loren B. Landau is director of Wits University’s Forced Migration Studies Programme in Johannesburg, South Africa. He first worked in Washington, D.C., as a policy advocate on behalf of refugees and is now chair of the South African Consortium of Refugee Affairs’ Executive Committee. Past research explores the effects of migration and aid on western Tanzania and refugee assistance strategies in western Cote d’Ivoire. He is currently coordinating a post-graduate degree program in forced migration studies and co-directing a comparative project on migration and urban transformation in Johannesburg, Maputo, Nairobi, and Lubumbashi. He has testified to the South African Parliament, and was consulted for the South African Human Rights Commission and the United Nations High Commissioner for Refugees. He is also co-editing a book series tracing the relationship between displacement, citizenship, and the African state. He holds a master’s in development studies from the London School of Economics and a doctorate in political science from the University of California, Berkeley.
STRENGTHS AND LIMITATIONS OF INFORMATION AND DATA ANALYSIS IN RESPONDING TO CRISIS IN MALI
Mamadou Kani Konaté, CAREF, Bamako, Mali
Given the natural disasters that Mali has experienced since the country’s independence in 1960, one might expect to find a real culture of preventive measures or of response to these events. Empirical evidence of the practices of Malian societies with regard to handling such crises leads to the conclusion that, in fact, responses to them are rarely foreseen.
Over the last 45 years to which we refer, two examples come to mind in relation to ecological crises and famine where research shows that the events were not anticipated. First, the cyclical nature of drought in the Sahel (the sub-Saharan African region) did not encourage early research to develop specially adapted seed varieties. The same delay was evident with regard to research on social behavior and practices to facilitate improved management that could have preserved the natural environment, despite demographic pressure that increases every year. Second, periods of food shortage have not led to the development of techniques to transform and conserve food supplies to ensure that surplus production could cover times of food shortage. To date, research in this sector is still at its early stages.
The conclusion relating to research concerning the management of ecological crisis is equally valid for the social domain. Social crises often manifest themselves in the form of conflict in the primary sector. Farmers, herders, and fishermen are often in conflict concerning management of areas
that are vital to their subsistence economies. Social equilibrium is permanently in a state of disruption and rearrangement. Traditional mechanisms to regulate conflict are disowned because of demographic growth. A more and more pronounced gap exists between national legislation (“Law in books”) and the vital and practical rules determined by communities (“Law in action”). These developments challenge the responses applied to social crises. As a result, the responses do not seem to be adapted to the situation. In this context, more pertinent and durable approaches are needed.
The political and administrative domain is also responsible for failure to predict these crises. Populations expect that the centralized state will help them during times of natural catastrophe. In these regular natural crises, the state’s duty takes the form of forecasting and planning. For exceptional events, responses are commensurate with the means available at the time of their occurrence. However, in the context of Mali, to a certain extent the inadequate nature of response to repeated ecological crises in the semidesert in the north, combined with lack of development of the state presence there, contributed to the rebellion that lasted for five years.
Whatever form they take, the crises and their effects generate a mass of information. Analysis of this information generates better knowledge relating to the crises and a deeper understanding. Data analysis can provide elements toward future forecasting. From these elements of forecasting, response strategies can be developed. While some countries or organizations can speak toward data visualization as one element of data analysis and distribution, the tools or technologies needed for visualization are largely not available in the context of Mali. Geographical information systems or global positioning systems are currently used by few research institutions. Visualization is not discussed further in this contribution.
This paper focuses on several central questions: What is the experience of Mali with regard to the crises experienced since the country gained independence? In which sectors do crises occur periodically? What types of information are used for prevention of crises? What information is used to produce responses? What methods are put in place to generate the data necessary to analyze these crises and to develop responses to them? What are the limitations of the available information or data? What approaches can we use to advance beyond current practices and methods?
STRENGTHS OF ANALYSIS FOR INFORMATION AND DATA IN RESPONDING TO CRISES IN MALI
Data collection and analysis in a poor, French-speaking country such as Mali were, until recently, limited to national-scale operations including the following, which were taken from a more exhaustive list:
National Census of Population and Habitat (1976, 1998)
Demographic and Health Surveys (1987, 1995-1996, 2001)
Budget and consumption survey (1987, 1989)
National household activity survey in 1989
Continuous agricultural surveys
Other international and national institutions based in Mali produce information on samples of different sizes. Among the international institutions are the Famine Early Warning System (FEWS), the African Sub-Saharan Economic and Statistical Observatory (AFRISTAT), and the Sahel Institute.
Most of these institutions help national structures to produce one-off information or to do routine surveillance. They also produce data themselves, often at a subregional level, together with analyses and results.
Although in certain contexts such as famine, drought, and plagues of locusts, information generated at a macro level can be useful, in more localized instances, macro data present limitations that can make their use nearly impossible. The typical example is that of census details that are limited to a certain number of variables.
Faced with the AIDS pandemic, for example, orphans have become an increasing subpopulation. Mortality analyses from census data can give the proportion of orphans who have lost one or two parents. However, these data do not allow calculation of the proportion of children orphaned due to AIDS because cause of death is not a variable considered in the census.
The same considerations apply to social crises. Variable determining factors in the consideration of social conflicts are not always available at the time these occur. Numerous ethnographic or anthropological studies of the human groups in Mali’s different regions do exist. However, the findings of these studies are little used in planning or decision making. A strong tradition of administrative or political decision making occurs and is rarely based on the results of studies; sometimes these decisions are the source of conflicts.
The most topical example is provided by the application of decentralization to Mali. According to Mr. Aboubacrine Ag Indi, technical adviser at the Commissariat for Institutional Development in Mali, this has been defined as “the transfer of power from the Central State to decentralised local government bodies. Thus it is simply a matter of enabling communities at a local level to manage their affairs through their elected representatives.”
In Mali, ethnicity is still not recorded as a variable in the census. In countries where the ethnic group variable is recorded, data from a general census of the population facilitate determination of useful indicators such as the comparative size of ethnic groups. We call this comparative because,
more than ethnicity, it is the ethnic background during socialization that determines individuals’ beliefs and behavior. Social ethnicity will therefore have a more determining influence than ethnicity inherited through birth. Mutation of family names after about two generations’ change of residence is evidence of the preeminence of social ethnicity which over time will mutate into the ethnic origins of succeeding generations.
These examples lead us to assert that the objectives of collection determine the type of information collected. Indeed, one of the advantages is that some variables can be used on later occasions for other types of analysis. In using data to proceed to secondary analysis, attention should be drawn to the degrees of relevance that can be attained. These degrees of relevance, among others, may include the following:
The time that has past since the data were collected;
The fact that the variables used allow for only indirect or approximate estimation of the phenomena being studied; and
The difference between the characteristics of the surveyed population and the characteristics of the population at the time of a crisis.
The limits inherent in analyses based on secondary use of the data are a constraint with which Mali is now confronted. Despite the considerable number of data collection operations that have been carried out, very few have been specifically conducted to document a situation that could lead to a crisis. Four examples are used to illustrate the lack of specific information to resolve crises. The first example concerns a natural crisis involving invasion by swarms of locusts in 2004. The second example discloses a social crisis relating to management of natural pasturage. A third relates to the school crisis after the social revolution of March 1991. The Touareg rebellion in the north completes the partial typology of crises that Mali has experienced during the past 10 years.
EXAMPLE OF NATURAL CRISIS MANAGEMENT
The most recent, final, nationwide report drawn up by the Ministry of Agriculture gives the following description:
Mali experienced a massive invasion of locusts during the agricultural season 2004-2005. Two-thirds of the country, particularly the area between the 14th and the 21st northern parallels, was affected between June and September 2004. The phenomenon took the form of what was called a “locust hazard” equivalent to being on a war footing. This danger affected 7 Regions, 26 Circles and 130 Communes.
It is noted that Regions, Circles, and Areas are the administrative subdivisions in Mali.
Modern Administrative Management
Existing surveillance structures at the Ministry of Agriculture technical services gave the alert before June 2004 to prevent invasion by swarms of locusts. According to Mr. Fakaba Diakité, Coordinator of the Unit for Migratory Locust Control (ULCP), and the African Project for Emergency Control of Migratory Locusts (PALUCP):
As technicians we expected this, because in 2003 we saw a build-up in the region. We treated almost 40,000 hectares in Mali…. From the month of February the United Nations Food and Agriculture Organization (UNFAO) launched an appeal to the international community announcing that as soon as rains arrived in the Sahel there would be locusts. By the month of March, we had drawn up an action plan that we presented to the Ministry of Agriculture and all the other Ministries involved, as well as to the development partners. Unfortunately there was no reaction. We then presented this plan more than ten times before the month of July. There was no response. It was from July that we began to have the first swarms, and people, or at least the authorities, began to move. First an operational headquarters was set up … from the month of August…. It was from September that we began to receive responses from partners.
The national coordinator’s explanations show the following sequence of management of the invasion by locusts in Mali in 2004:
Preventive activities are limited to treatment of a part of the surface infested by larvae, given the limited stocks of pesticides.
The alert was given in good time at both national and international levels.
Intensive advocacy was conducted at national level for about six months.
The conclusion was that no action had been taken despite all these preliminary stages.
An attempt was made to provide a response as soon as the first swarms began to move.
This model is one of centralized administrative management of the crisis. It leads to limiting the crisis for a much longer period and calls for more logistical and financial resources. In looking at the analysis of control of migratory locusts that is made by Mr. Fakaba Diakité, it is evident that “in locust control, as soon as you see them here (in the southern areas), it
means that everything we have done has failed.” In other words, prevention is the best means of managing invasion by swarms of locusts.
Before preventive control, there has to be political will. In Mali since the 2004 crisis, the government has made a commitment to provide a national budget allocation of 1 billion francs for locust control. Despite this commitment, the coordinator of the Locust Control Unit judges that “unfortunately we never have a line of credit at the level of our State budgets to enable us to conduct real, permanent surveillance, because when you look at what we spent on cricket control in 2004, it is about 7 billion CFA francs (about U.S. $13,207,547 at a rate of $1.00 to 530 West African CFA francs). In our calculations we thought that with seven billion we could conduct control activities for over twenty years if there was preventive control.”
Currently the government budget allocation going to the ULCP is sited within this structure. This institutional arrangement means that the budget is not devoted exclusively to prevention of locust invasion. Surveillance of other pests is also funded from the same allocation.
Traditional and Community Means of Management
At the level of prevention, members of communities can be very useful. Community members can be trained in techniques of unearthing eggs. In managing the infestation, instead of exposing them to chemical products used for treatment, they will be effective by using ditches to bury the locusts that fall into them because these insects always move forward in the same direction.
Early Warning System
The Early Warning System (SAP) was set up in Mali in 1982 (Tékété, 2002). Its objectives are
forecasting situations of food shortages or abundant production, and
improving availability and provision of necessary assistance.
Early warning is ensured by collaboration of several structures and institutions in Mali. These are grouped in a multidisciplinary working group including an American nongovernmental organization, FEWS. In Mali the early warning system has developed to respond to needs linked to food security.
The SAP “is based on permanent information collection concerning rainfall, crop evaluation, livestock, market prices, migration of populations,
their habits and food stocks, as well as their health status…. The SAP systematically surveys zones that are traditionally at risk, that is to say the zones having experienced and quasi-permanently experiencing severe food shortage. These zones are generally the arrondissement subdivisions located to the north of the 14th parallel” (Tékété, 2002). The arrondissement is the third-level administrative subdivision in Mali, following the Region and the Circle, which are the largest geographic entities.
According to Tékété (2002), the methodology utilised by the SAP is based on a certain number of activities:
Collection of information
Verification and rapid treatment of information
Distribution of information
These operations often vary according to whether they are situated at the regional or national level. They are produced on a 10-day basis.
FEWS has been cited as an early warning system in Mali. This structure that existed in Mali from 1985 to 2000 has been succeeded by FEWS NET. “The general aim of FEWS NET is to facilitate installation of food security and more efficient and more sustainable planning networks which will then be managed by the governments involved…. In Mali and indeed in general, FEWS NET carries out climate monitoring (Meteosat images), of vegetation (vegetation indicators), and cereal prices in various markets. This information is provided by FEWS to other working groups of which it is a member, in particular SAP and the GTPA (Multidisciplinary Working Group for Agrometeorological Assistance” (Tékété, 2002). FEWS NET publishes a monthly bulletin that is posted on an Internet site.
The basis of this early warning mechanism was access to costly logistical and technological tools. Less expensive means can now be substituted for them. In addition, the mechanism is designed and conducted by administrative structures. Little place is provided for communities in the process of data collection and analysis. It has to be recognized, however, that community members are involved in daily evaluation of their own environments. In countries such as Mali with limited means, the emphasis should be above all on human investment. Sentinel surveillance is an approach to explore. It is permanent. It can be conducted by communities themselves. The accompanying measure to ensure that it becomes effective is the training of resource persons among the communities for the collection, systematic analysis, and transmission of data in real time. This approach is just as valid for natural as for social crises and can contribute to making early warning more effective because it documents phenomena over a longer period of time through qualitative and quantitative chronological series.
MANAGEMENT OF SOCIAL CRISES
Social crises in Mali, whether or not they are documented by research studies, typically occur between two communities (or within the same community) concerning management of natural resources (pasture land, agricultural land, water, and products derived from water). Implementation of decentralization policy is currently at the origin of localized social conflicts whose management varies from one location to another. “Decentralization” is the legal framework installing modern local government bodies.
The discussion begins with the last of these aspects. A second example is cited that relates to management of natural pasturage. Specific types such as urban sociopolitical conflicts and the Touareg rebellion in the north conclude the analysis of social crises.
Conflict of Traditional Chiefdoms
The first example describes establishing local management structures in the framework of decentralization. A persistent conflict is present in the area of Bamba, situated midway between Gao and Timbuctou. Transfer of local power from the master group to the traditionally administered group following communal elections is at the origin of this situation. After numerous clashes and loss of lives, the Malian state currently maintains troops on the spot to prevent continuing clashes. Negotiations have still not led to reconciliation of the different points of view on the new approach to management of local government.
In factual terms, at stake are the defense of prerogatives and failure to accept challenge to the local traditional mode of administration. The possible causes behind the origin of this conflict could be lack of preparation through preliminary studies of local management of power and/or failure to identify potential mechanisms and degrees of acceptable change, as well as their mode of operation.
Conflict Concerning Management of Pasturage
This example describes the clash between two Fulani communities; the clash concerned disputes regarding pasturage situated in the Inner Delta of the Niger River in Mali (the Niger Bend) in 1993. The final count included 20 dead and 42 wounded. According to Maïga (2005), numerous studies of the Niger River Inner Delta have been conducted, but “on the other hand, very few have taken an interest in the stakeholders in negotiations for access to ’bourgou’ pasturage in the context of decentralisation.” Choice of the zone of his study “is justified by the recurrent conflicts that occur there, proof of an exacerbation of competition surrounding ‘bourgou’ pasturage.” The
Niger Inner Delta includes an area of 30,000 square kilometers of wetlands subdivided into three protected sites linked to the International Convention on Wetlands, which are of international importance particularly because of their biological diversity. The area is noted as the habitat of water birds.
In his approach, Maïga (2005) questions two fundamental aspects: first, the interactions between state law and customary law, and second, prospects for access to pasturage and the future of traditional chiefdoms in the context of decentralization (local government). The problem of recognition of customary law in the Interior Delta was posed during the colonial period when addressing relations between the Fulani and Touareg ethnic groups. Establishing recognition from the colonial administrator of the right to movement to and from seasonal pasturage was considered to be the reason for no longer observing the rules of those who controlled customary management of pasture land. These realities still persist and regularly bring to the fore the question of primacy of one of the two types of law or their simultaneous application.
Beyond customary regulation of pasturage conflicts, two problematic aspects are highlighted. On one hand, understanding which legal system is intelligible, applicable, or weightier for the protagonist populations is needed during conflicts. On the other hand, understanding the influence of decentralization on reorganization of the chiefdoms or on the power of the masters of the land is also important.
In pre-decentralization law in Mali, Dembélé (2005) noted that “[natural] resources are determined by the law to be part of the State’s natural public domain,” while “neighboring villages consider them to be part of their heritage and regulate their operation.” This customary tendency is still alive today. Parallel operation of the two jurisdictions is an additional source of conflict that does not make management of crises any easier.
Sociopolitical and Urban Conflicts
Two types of sociopolitical conflict have emerged in the urban setting in Mali during the past 10 years: the school crisis and the claim for democracy. The origin of the school crisis was the demand by high-school and higher education students for better educational conditions. Mali inherited responsibility from the colonial system for total coverage of school costs for all those who obtained access to secondary education. Economic difficulties at the end of the 1970s are the basis for revising this system. Installation of scholarship awards according to certain criteria and the difficulty of honoring payment of these scholarships led to disruptions that recurred and were generalized in 1991. The background clamor for democracy made use of this fertile ground, which then became the factor that triggered popular
revolution. The military regime fell on March 26, 1991. The Third Republic emerged in Mali. Democracy has been installed since this date.
Although intertwining of the school crisis and claims for democracy temporarily benefited both causes, delays in meeting students’ claims became the source of recurring disruption in schools and colleges in urban settings in Mali. The role played by students in social change became a source of demands and blackmail whose disorderly manifestations paralyzed Malian towns on a sporadic and partial basis. Public authorities did not attempt to analyze the school phenomenon in order to understand its mechanisms and consider taking appropriate measures. They chose instead the method of long and uncertain negotiations with various splinter groups of the student movement. In the end, the formula of mediation that brought about the beginning of calm and pacification of campuses and schools was established under the aegis of a committee made up of all parts of civil society. This calm is still precarious; during the twenty-third summit bringing together African and French heads of state in Bamako on December 3-4, 2005, youth delegates from the continent delivered an ultimatum (read by the Cameroonian, Marie Tamoifo NKom) that said in essence “if politicians do not take notice of youth, the wind of change, in a democratic context, will lead young people to settle the business of the politicians so that their commitments take on a meaning.”
The Touareg Rebellion
The last type of social conflict presented is one that was expressed through taking up arms. Touareg communities in the north of Mali judged that they had been ignored during the process of development since independence. The imbalance between the south, the center, and the north in investment choices and installation of infrastructure was becoming more and more marked. Faced with this situation, the Movements for Liberation of the Azaouad launched the rebellion. National and international mediation were the mechanisms that resolved this social crisis. This led to
upgrading the north of Mali with the design of dedicated development programs;
setting up a dedicated administration (the Commissariat for the North);
organization of a symbolic Flame of Peace under the aegis of the United Nations during which the arms laid down by combatants in the rebellion were burned; and
integration in March 1996 of some of combatants of the rebellion into the various branches of the military.
LIMITATIONS OF INFORMATION AND DATA ANALYSIS IN RESPONDING TO CRISIS IN MALI
Examination of macro-level data collected in Mali as well as those connected with certain types of crises and their resolution does not highlight the information to be collected in order to respond to crisis. Since this paper is not based on the findings of a systematic evaluation or a specific study, it is difficult to evaluate the contribution made by information analysis. One exception is management of the invasion of migratory locusts, which was supported by prior technical information collection. Even in this particular case, however, a lack of consideration of the population dimension was evident. This determining feature in measuring both quantitative and qualitative impacts does not appear clearly in the cases presented.
In response to the question of which types of information are used in crisis prevention, the following answer can be made: Specific and systematic information was used from the early warning system concerning agricultural crises and their impact on food security or concerning natural crises such as swarms of migratory locusts. In all the other cases considered, relevant information is largely from administrative sources.
By way of methods installed to generate the necessary data for crisis analysis and providing responses to them, the first to be noted is the early warning system. Routine information generated by the administrative system and secondary treatment of certain data collected in connection with other operations are used to provide complementary analysis. The limitations of these approaches have already been mentioned. Given the conclusions reached through the cases examined in this paper, which approaches will lead to advancing beyond practices and methods that are currently used?
CONCLUSION: APPROACHES TO ADVANCING BEYOND CURRENT METHODS AND PRACTICES
Approaches to the anticipation of natural or social crises exist, but in the current context of Mali, very little investment has been made in this area. At the level of state structures, the example of locusts has shown the limitations of rapid and appropriate reaction. Affected communities are also unprepared, either because of ignorance of what they can do or through lack of initiative, or both.
To advance beyond this situation, two levers need to be activated. The first is that of information, and communities must be provided with the means of acquiring improved knowledge regarding the natural risks that they may experience in their environment. The second is to encourage local mechanisms of social organization to take part in permanent documenta-
tion of the process through sentinel surveillance in anticipating and reacting to early warnings.
More specifically, regarding management of social crises, anticipation is also a valid general approach for both authorities and communities. It should be preceded by sociological, anthropological, and ethnographic research studies in zones with recurrent conflicts.
To go beyond these approaches, positive partnership in prevention should be emphasized. A recent example is the signature of a convention between farmers and livestock owners in the Circle of Kangaba. This convention delineates areas for keeping livestock, “pasture zones,” responsibilities of community surveillance brigades for application of the agreement and those of the forestry service agents, and a number of other provisions. All community and government partners have signed the document. Experience from its implementation will tell us if the formula is a good one and if it might be replicated elsewhere.
Dembélé, A., 2005. Decentralisation and management of natural resources in Mali–Boro Lake. In G. Hesseling, M. Djiré, and B.M. Oomen (eds.), Law in Africa–Local Experience of State Law in Mali. Karthala: Afrika-Studiecentrum, pp. 216.
Maïga, B., 2005. Access to pasturage in the Niger Interior Delta–Case of the “bourgou” pasture at Leydi Yaarlarbé. In G. Hesseling, M. Djiré, and B.M. Oomen (eds.), Law in Africa–Local Experience of State Law in Mali. Karthala: Afrika-Studiecentrum, pp. 185.
Tékété, A., 2002. Evaluation of Performance of EW’s—CLIMAG West Africa: Harmonisation of Climate Prediction for Mitigation of Global Impact in Sudano-Sahelian West Africa, May 22.
Touré, B., 2005. Youth: Burning questions, timely answers. L’ESSOR, National Daily Newspaper, No. 15592, December 5, pp. 8-9.
My thanks go to Mr. Fakaba Diakité for the time he devoted, the interview he agreed to give me, and all the documents he placed at my disposal to produce this paper.
Mamadou Kani Konaté is a rural sociologist with 20 years’ experience in the use of qualitative and quantitative research techniques. He spent 15 years working for CERPOD (Centre for the Study and Research of Population for Development). He conducted the first DHS in 1986-1987 and carried out various sociodemographic studies and evaluations in Mali. He has assisted several institutions in nine West African countries belonging to the Interstate Committee for Drought Control in the Sahel, to design and carry out research on population and development. He has been on several
WHO committees and is the chair of the committee called Strengthening Research Capability for Social Health Sciences for Tropical Disease Control in Geneva. He is also a member of UEPA (Union for the Study of Population in Africa), FASAF (Network for Schooling and Family in Africa), and IUHPE (International Union for Health Promotion and Education in Africa. In February 2003, he co-founded CAREF (http://www.caref-mali.org), a private institution whose aim is to support young researchers by providing them with an enabling environment in which to work as well as giving them the opportunity to share and acquire experience. The center is entirely financially self-reliant, its funding being derived from consultancies.