Assessing the value of climate forecast information for pastoralists: Evidence from southern Ethiopi

1
Assessing the value of climate forecast
information for pastoralistsEvidence from
southern Ethiopia and northern KenyaWinnie K.
Luseno, John G. McPeak, Christopher B.
Barrett,Getachew Gebru and Peter D.
LittleApril 2002
2
1. Introduction

• Roughly 2/3 of Africa is drylands, an area home
  to 50 mn people, the poorest on the continent.
• Tremendous opportunities for climate forecasting
  since most poverty remains rural and most
  Africans work in crop or livestock agriculture,
  where optimal management regimes depend
  fundamentally on rainfall
• - This is especially true among pastoralists
  whose semi-nomadic grazing systems are based on
  climate-induced spatial and temporal variability
  in forage and water availability and in disease
  patterns.

3

• Study area
• -Avg rainfall 200-750mm/year.
• Serious flooding in 97/98 and droughts in 92/93,
  96/97, 99/2000.
• Few nonpastoral options available

4
2. Data

• Collected by USAID GL CRSP PARIMA project
• As a subcontract for the "Regional Climate
  Prediction and Applications for the Greater Horn
  of Africa" project undertaken by the
  International Research Institute for Climate
  Prediction at Columbia Universitys
  Lamont-Doherty Earth Observatory and the
  University of Nairobis Department of Range
  Management
• 2 stage survey among 323 households in 11 sites.
  
• -Stage 1 fielded in March 2001 after DMC-Nairobi
  released long rains 2001 forecast.
• -Stage 2 fielded in June-July, after rains ended.
• Survey was the year following a major drought,
  but also followed furmat rains in Jan-Feb.

5
3. The Basics of Information Theory

• The value of the exogenous information depends on
  the
• Correlation between forecast information and
  state of nature (i.e., that the forecast is
  indeed informative)
• Information changing agents subjective beliefs
  as to the likely state of nature (hence the
  importance of confidence in forecast information
  received)
• Agents capacity and willingness to adjust
  decisions in response to changing beliefs.
• Our analysis aims to elucidate these latter two
  points among east African pastoralists.

6
4. The Value of Climate Forecast Information for
Pastoralists

• Our behavioral and informational assumptions
• -Pastoralist households form prior beliefs about
  the upcoming seasons climate based on past
  experiences and indigenous climate forecasts.
• -These beliefs are then subject to revision
  following reception of new forecast information
  from external sources.
• -Pastoralists act on their posterior beliefs as
  to seasonal climate patterns.
• -Actions based only on indigenous climate
  information are uninformed relative to those
  based on an information set supplemented with
  external forecast information.

7
4. The Value of Climate Forecast Information for
Pastoralists(i) Pastoralists comprehension of
probabilistic forecasts

• - We elicited pastoralists trinomial forecasts
  of seasonal rainfall to generate distributions
  directly comparable to DMC forecasts.
• Pastoralists form and communicate probabilistic
  forecasts themselves.
• Only 10.2 percent offered a degenerate forecast.
• Almost half didnt put majority weight on any
  one outcome.

Table 1 Mean probabilistic expectations of
rainfall volume for the 2001 long rains season

8
4. The Value of Climate Forecast Information for
Pastoralists(ii) Pastoralists use of indigenous
climate forecasting methods

• Pastoralists employ an extraordinary variety of
  indigenous climate forecasting methods offering
  different bases for predicting different
  variables (e.g., start date, duration, volume)
  and at much finer spatial resolution.
• Indigenous forecast skill seems acceptable.
• 80 of sample heard some indigenous forecast
• Much confidence in indigenous forecasts and
  widespread belief in their accuracy ex post.
• 94 expressed at least some confidence in
  indigenous forecast of rains start date
• 84 had at least some confidence in indigenous
  forecast of rainfall volume

9
4. The Value of Climate Forecast Information for
Pastoralists(iii) Awareness of and access to
external climate forecasts

• A sharp minority of pastoralists receive
  external, model-based forecasts only 1/5 of
  sample heard an external forecast.
• Radio is the most common source. TV, newspapers,
  other print and extension services (NGO or govt)
  reaching no more than 2-3 of respondents each.
• Only 5 of the Ethiopians and only 23 of the
  Kenyans owned a radio, suggesting the possibility
  of a material constraint to external forecast
  receipt.
• But content and timeliness are key too. Mean
  lead time requested is 4-5 weeks

10
4. The Value of Climate Forecast Information for
Pastoralists(iv) Confidence in external climate
forecasts

• Even controlling for access to forecast,
  pastoralists express less confidence in external
  forecasts than in traditional ones. Educational
  attainment and location (market access) key
  correlates of confidence in external forecasts.

(v) Do external forecasts differ from indigenous
forecasts?
External forecasts that merely reinforce prior
beliefs necessarily do not change behaviors and
are thus of limited value. Significant
differences in 9/10 sites between the DMC
forecast and the prior beliefs of pastoralists.
(vi) Do pastoralists update beliefs based on
external forecasts?

• Even pastoralists who receive and express
  confidence in external climate forecasts do not
  seem to update their beliefs in response.
• - Spatial resolution may be an issue here.

11
4. The Value of Climate Forecast Information for
Pastoralists(vii) Do pastoralists adjust
behavior in response to climate info?

• Instrumental (management) vs. direct
  (consumption) value of information
• Many pastoralists lack the material means to
  adjust behavior in response to changing climate
  expectations loss of spatial refugia, minimum
  herd size necessary for migration, water mgmt
• Pastoralists generally did not change behavior
  due to forecasts

Table 15 Percent who received a climate forecast
who changed any behaviors
12
4. The Value of Climate Forecast Information for
Pastoralists

• Behavioral change in response to changed climate
  expectations was predominantly in crop
  cultivation (agro-pastoralists) and in prayers
  and ceremonies. Little change in herd or
  financial management in our sample, either in
  LR01 or past expectations.
• Flexibility of production systems is key.
  Pastoralism is highly flexible in response to
  climate variation, so little need for climate
  forecast products as compared to crop
  cultivation. Might this change as forecasts
  spatial resolution and accuracy improve?

13
5. Concluding Remarks

• Pastoralists understand and communicate
  probabilistic seasonal climate forecasts.
• As market access, education and wealth increase
  and ICT becomes more widespread, it is reasonable
  to expect growth in access and use of external
  forecast information.
• Majority find external forecasts useful and
  accurate and express at least some confidence in
  them.
• Building confidence in or communication of
  forecasts low priority.
• Key needs are improved spatial resolution,
  developing onset forecasts, and general economic
  advancement