Salvatore Di Falco
University of Geneva
Switzerland salvatore.difalco@unige.ch

• Background
• Use of survey data for policy relevant research
• Case study => adaptation pay off and identification of the “best” strategies for food security
• Data needs
• Environment for Development (EfD): a success story
2

3
1970-2004 (IPCC 2007)

• Current agreements to limit emissions, even if implemented, will not stabilize atmospheric concentrations of greenhouse gases and climate change
• Crop productivity to decrease “for even small local temperature increases (1 – 2 ° C)” (IPCC 2007)
• In many African countries => “yields from rain fed agriculture could be reduced by up to 50% by 2020”
(IPCC 2007, p.10)
4
• Serious implications for food security

How would the future look like? (IPCC, 2007)

Can we do without adaptation?
• Probably not
• The identification of climate change adaptation strategies is therefore vital in sub Saharan Africa
1.Autonomous adaptation => pay off?
2.What is the impact on food security of farmers’ decision to adopt some strategies in response to changes in temperature and/or precipitation?
3.What are the driving forces behind farmers’ decisions to adapt to climate change?
6

7
Case study: Ethiopia Nile River Basin

8
• Agriculture accounts for about 40% of GDP, and 90% of exports, and 85% of employment (MoFED, 2007)
• Ethiopian agriculture is heavily dependent on natural rainfall, with irrigation agriculture accounting for less than
1% of the total cultivated land in the country
• Ethiopia suffers from extreme weather events:
- frequent droughts (1965, 1974, 1983, 1984, 1987, 1990,
1991, 1999, 2000, 2002, 2009);
- recent flooding (1997, and 2006)
Food insecurity

Data
• Partners: IFPRI, USA; Ethiopian Development Research Institute
• 1,000 crops farms (2,823 plots)
• 2004 and 2005 => Smallholders Agriculture
• Area: Ethiopia Nile River Basin
• Great survey on the issue!
• Use of farm specific weather data
• Thin Plate Spline method of spatial interpolation for imputation of farm and plot specific rainfall and temperature
• Perceptions and adaptation
• Extension services, tenure security, information + others
9
• In other 8 African countries

Issues for quantitative analysis
• Systematically different between adapters and non adapters
• Some farmers are better than others…
• Unobservable characteristics of farmers and their farm may affect both the adaptation strategy decision and net revenues
=> inconsistent parameter estimates
• Self selection
10

Ideally, we would like to have…
Control Group Treatment Group
Quantity produced by farmer X if DID NOT adapt
Quantity produced by farmer X if DID adapt
11
Compare the expected food productivity under the actual and counterfactual cases that farmer X adapted or not to climate change
Statistical tools: Di Falco et al. 2011 11

What Adaption to Climate Change actually is?
What are the “best” strategies that can be implemented to deal with climatic change in the field
12
What are the economic implications of different strategies?
Identify the most successful strategies by implementing a counterfactual analysis

T able 1. Climate change adaptation str ategies
13
Frequency %
Soil conser vation 1,397 72.27
Changing crop varieties 1,186 61.36
Water strategies
Building water harvesting scheme 309
Water conservation 82
Irrigating more 279
15.99
4.24
14.43
Other strategies
Early-late planting 176 9.11
Migrating to urban area 23 1.19
Finding off-farm job
Reduce number of livestock
132
Leasing the land 3
Changing from crop to livestock 71
121
6.83
0.16
3.67
6.26
Adoption of new technology 26 1.35
Note: sub-sample of farm households that adapted at the plot level (sample size: 1,933)
25

Strategies
(1) changing crop varieties only;
(2) implementing only water strategies such as water harvesting, irrigation or water conservation;
(3) implementing only soil conservation;
(4) implementing water strategies and changing crop varieties;
(5) implementing soil conservation and changing crop varieties
(6) implementing water strategies and soil conservation
(7) implementing water strategies, soil conservation, and changing crop varieties;
(8) implementing other strategies.
14

Switching Regression Model
(Di Falco et al. 2011; Di Falco and Veronesi;
2012)
• Two stage procedure
1. We estimate the probability of choosing a particular strategy (selection model where a representative farm household chooses to implement a specific strategy)
2. The information stemming from the first step is used on farm revenue, where farm net revenues are regressed against climatic variables and other control variables

16

How do we get there?
• First stage of the model tells you exactly that
• Barriers and drivers for autonomous adaptation
• Besides climate and past shocks
• Trees as tenure security is important driver
• The dissemination of information on changing crops and implement conservation strategies are very important – increase awareness
• Extension services are very important in determining the implementation of adaptation strategies (also training)

Future research needs
• More (and better) surveys – involvement of local institutions (academia, research, training)
• Going back to areas previously interviewed
• Nile Basin: How things have changed
• How perception have changed
• How the barriers and drivers have changed
• Dynamic implication of adaptation strategies
• Long term vs short term responses

A platform is necessary: EfD
• Environmental economics and advocacy in developing countries for 20 years => SIDA (Swedish Development funded)
• Weak and underfunded academic institutions
• Academic brain-drain
• Government agencies unable to attract and maintain academic capacity for policy analysis
• Weak interface between academia and government
• Two pillars: building capacity (PhD programs and courses)
• Policy relevant research => on the field
• EfD centers

Characteristics of EfD centers
• Strong research capacity and facilities
• Connection to a graduate program
• Strong policy interface
• Efficient administration
• Ethiopia – EDRI/Addis Ababa University
• Kenya –University of Nairobi/KIPPRA
• China - Peking University
• Tanzania – University of Dar es Salaam
• South Africa – University of Cape Town
• Central America – CATIE
• Sweden – EEU/UoG
• USA – Resources for the Future

Conclusions
• Adaptation in the field => food security
• Understanding what can be done to facilitate it
• Combination of surveys, methods
• Randomized controlled trials and field experiments can complement
• Building capacity is essential to face challenges of climate change
• Follow and expand experiences like EfD

choice of M mutually exclusive strategies to long term changes in mean temperature and rainfall, which can be indexed j = 1 … M . We then outline an econometric model that is used to investigate the effects of different climate change adaptation strategies on net revenues.
Stage I – Selection Model of Climate Change Adaptation Str ategies
In the first stage, let A* be the latent variable that captures the expected net revenues from implementing strategy j ( j = 1 … M ) with respect to implementing any other strategy k . We specify the latent variable as
(1) A ij
* regression model First Stage
=
V ij
+ h ij
=
Z α i j
+ h ij
ì
1 iff A i
*
1
> max( k
¹
1
A * ik
) or e i 1
<0 with A i
= í
ï
ïî
M M
M iff A * iM
> max( A * ik
M
) or e iM
<0 if strategy j provides expected net revenues greater that is farm household i will choose strategy j in response to long term changes in
23 mean temperature and rainfall if strategy j provides expected net revenues greater than any other strategy k
¹
j , that is if e ij
= max( k
¹ j
A * i k
-
A ij
* )
<
0 . Equation (1) includes a deterministic component ( V ij
=
Z α ), and an idiosyncratic unobserved stochastic i j component h i j
, the latter captures all the variables that are relevant to the farm household’s decision maker but are unknown to the researcher (e.g., farmers’ skills, motivation), it can be interpreted as the unobserved individual propensity to adapt, and it is assumed that that the covariate vector Z i
is uncorrelated with h i j
, i.e.,
E ( h ij
Z i
)
=
0 .
The deterministic component V ij
depends on factors Z i
that affect the likelihood of choosing strategy j such as farmer head’s and farm household’s characteristics (e.g., age, gender, education, marital status, if the farmer head has an
10

Second stage: Multinomial Endogenous
Switching
• if the error terms of the selection model correlated with the error terms u ij
 ij are of the net revenues functions (3a)-(3m), the expected values of u ij conditional on the sample selection are nonzero
• Estimates will be inconsistent by OLS
24