Risk Perceptions in Northern Kenya and Southern Ethiopia
Presented by Dr. Christopher Barrett, Cornell University, SAGA Cooperative Agreement
Sponsored by USAID’s Office of Poverty Reduction/Poverty Analysis and Social Safety Nets (PASSN)
November 20, 2006
Dr. Christopher Barrett presented recent research on the evolution of risk perceptions over time, space, households and individuals in northern Kenya and southern Ethiopia. The presentation was based on a paper he co-authored with Profs. Cheryl Doss (Yale University) and John McPeak (Syracuse University) entitled, “Interpersonal, Intertemporal and Spatial Variation in Risk Perceptions: Evidence from East Africa.” Using original data from the arid and semi-arid lands of east Africa, Dr. Barrett and his co-authors were able to identify primary determinants of risk rankings and what those risks are. The authors argue that their findings have development policy and program implications, particularly for food security programs.
Background and Methodology
The paper draws from previous research under the Pastoral Risk Management (PARIMA) Project of the USAID Global Livestock Collaborative Research Support Program (GL CRSP). Explored under PARIMA were questions such as:
- With factors such as steady degradation and frequent crises how does one understand risk(s) faced by pastoralists?
- What are some methods that Kenyan and Ethiopian pastoralists use to manage risk?
- What can outsiders do to help effectively?
- How do locals of these areas studied understand risk?
From the above, other questions arose which influenced the current research. These questions include:
- How are risks prioritized?
- How can primary concerns be predicted?
- Where should primary intervention start?
The researchers picked ten communities in southern Kenya and northern Ethiopia to attempt to get at some of these questions. 300 households were then randomly selected from those communities and individuals from households were randomly selected to identify the order of priority of risks. The rankings were meant to be prospective, asking people to identify the risks they faced for the next three months. The prioritized risks were ranked 1-11 with 1 being the highest in risk.
General Findings
- Food availability and human sickness were two concerns that overall were more important than other risks. Lack of pasture was more important than water availability, although the interpretation is complicated by the fact that pasture commonly holds access to water. There were spikes in grain pricing thus food shortage was dependent on many factors. Animal loss/theft was ranked overall lowest. The standard deviation on all risks was high, signaling considerable variation over time, space and respondents.
- Men and women were generally concerned about the same things. Household head characteristics showed higher differences than gender. Wealth measured had surprisingly modest effects once controls for shocks and space were in place, so that risk perception differences between rich/poor households were modest.
- The largest variation was across time and location. For example, between June 2000 and June 2002 there were many risks ranked lower and lots of criss-crossing Only 16% of respondents ranked the same risk as a priority over time (food availability).
- People respond to information quickly. They adapt their risks perceptions primarily to shocks in their community instead of shocks faced only by their households. This shows that people are paying close attention to other community members and that community communication channels work well.
Recommendations
- Communities must be monitored regularly in order to understand current risk perceptions and interventions to be fashioned appropriately
- Prioritize community level interventions over household level interventions
- Prioritize the protection of human capital
- Across locations the differences were huge- thus engage in geographic targeting where intervention is possible
- Information dissemination can be useful since it is disseminated and absorbed quickly within the community
- Violence and security are a lesser problem than food availability (but affected by violence depending on accessibility)
Q&A
Q: What is the content behind the question of “food shortage”? Don’t the other risks from the list feed into that as an outcome? Also, consumer prices ranked 4th but holds a lower deviation than other risks, does that mean it is an on-going concern and a long-term intervention is thus worthwhile?
A: This concern is shared about risks but it’s important to be faithful to local responses. It is agreed that it seems to underscore other risks than those that threaten human capital. However there is a sense here that those other risks aren’t the first order of concern and that this is important.
Q: Does that imply that by not helping herds, more is being done to help people?
A: Yes but when resources are given to people directly they will find innovative ways to help the animals. People are used to situations like high consumer prices which are affected by the market system. They have an underdeveloped market there and marketing margins are 50-100% of terminal market prices. There is little value added to selling animals in different areas. Policies don’t also affect them because they still have other obstacles to overcome, i.e. paving roads and accessibility.
Q: Has the variable market assessment been tested? What is the availability? What is the anomaly of priorities among locations, communities? Do the accessibility differentials and gender differentials reveal pre-occupation to men? Is there a useful tool to identify actions, address those, and then have people embrace it? Why is the research focused on short-term risks and not larger concerns and long-term risks, i.e. education?
A: The research matches period and retrospectively addresses risks. This eliminates the long-term question. It is not concerned with this, since the main question is always, when intervening, what is the immediate concern? For example long-term goals such as education were not studied under this design.
On the gender question, the difference was small overall between men and women in a large number of cases. Men were only slightly more concerned with things such as livestock than women. The shade of difference is not big enough to make much of it. Cross-sectional evidence among groups suggests more of a difference, but this study offers better controls which seem to make the gender effect largely go away.
On market accessibility, different sites had different levels of accessibility. There would be more going on than just strictly market accessibility such as government services, decisions made by community and coping strategies. Bundled differences cannot just be reduced to market accessibility. One has to ask what is the full package?
Q: What are the elements of health concern across time? Were these tracked and identified? What types of illnesses were covered? Was it tracking the health of heads of households strictly or children? What about access to water/sanitation over services?
A: Risk assessment did not address this, although illness types were primarily diseases such as malaria. Findings were not separated out by age, although it would be interesting to explore this. It would be surprising if much difference of well-being among families resulted. Equally true is that family well-being depends on the health of the males; if a male can not migrate herds and loses ability to move, etc. the health of his family will be affected.
Q: When you went on interviews and reflected on the quarter, what was the verification on perception?
A: On verification-this is a good point. The trade-off is to frame risks. If you take food shortage, where is the boundary on this? There are different ways to operationalize this. On the risk rankings-human health is affected by pasture and water- eliciting ordinal data ability couldn’t track this.
The community was asked about the different market prices. Average market prices were collected. People had a good sense of distributions, which was tested by using the stone system. Price realizations are difficult to map since prices are ever changing. It’s a good guess to also say that people aren’t all paying the same price anyways. And even with the majority of people without education lots of information is still processed.
Strengthening Evaluation and Assessment of Poverty
Issues in Poverty Reduction
Publications
Conversations on Poverty
Links
Partners
Chronic Poverty Research Centre
IDS Anniversary Conference 'Reinventing Development Research'
SAGA
2-03-05 Primary Schooling Demand in Rural Madagascar
4-22-05 HIV/AIDS Prevention Knowledge & Behavior
5-18 Agricultural Technology and Food Security in Madagascar