BASIS Director Michael Carter (working collaboratively with Ghada Elabed and Sarah Janzen) presented at the OECD meeting in Paris, September 2015 on the topic of index insurance and its impact on small farm investment and social protection.
Before and After the Drought: Evidence on the Impact of Index Insurance on Small Farm Investment and Social Protection
1. Before and After the Drought
Evidence on the Impact of Index Insurance on Small Farm
Investment and Social Protection
Michael R Carter*
(Collaborative work with Ghada Elabed & Sarah Janzen)
*University of California, Davis, NBER & the Giannini Foundation
OECD, Paris
September 15, 2015
M.R. Carter Before and After the Drought
2. Logic of Insurance as a Development Tool
Decades of evidence that risk
Makes people poor by reducing incomes & destroying assets;
and,
Keeps people poor, by discouraging investment & distorting
patterns of asset accumulation)
The development impacts of risk reduction through insurance
should therefore be significant:
By protecting households against the worst consequences of
adverse climatic shocks, index insurance should in principal
allow households to prudentially invest more in risky, but high
returning agricultural activities.
That is, if insurance has ex post protection effects, then it
should also have ex ante investment effects
M.R. Carter Before and After the Drought
3. Today’s Talk
Evidence on the ex ante and ex post effects is just now
emerging
Review today impact results from two randomized controlled
trials of index insurance
Index-based Livestock Insurance (IBLI) protected consumption
(for poorer households) and assets (for less poor households)
after a drought in Northern Kenya
Two-trigger area yield contract in Mali induced substantial (ex
ante) increases in borrowing and cotton plantings by
small-scale farmers
Synergies between the ex post and ex ante effects of insurance
has important implications for design of effective insurance
policy
Discuss these implications in the context of specific
government programs in Kenya and Peru
M.R. Carter Before and After the Drought
4. Risk & Coping Absent Financial Markets
Households in developing rural economies are exposed to a
multitude of risks
For those facing borrowing constraints & missing insurance
markets, coping strategies reduce to two:
Reduce assets to protect or smooth consumption
Reduce consumption to protect or smooth assets
Both strategies can have prolonged economic consequences
Expanding the array of, or access to financial instruments
would seem likely to have large short & long run impacts
This paper specifically explores the impact of a novel index
insurance contract on coping after a drought
M.R. Carter Before and After the Drought
5. The IBLI Experiment in Northern Kenya
The arid pastoral regions of Nothern Kenya are an archetype
of risk & coping in the absence of financial markets
Shocks are severe
Financial instruments are sparse
Households are extremely poor
Traditional reactive emergency response appears exhausted
Index-based Livestock Insurance (IBLI) was rolled out as an
RCT in this environment in 2009
2011 drought created the opportunity to evaluate IBLI’s
effectiveness
On average, we find that after the drought insurance leads to:
A 36%-point decrease in livestock sales
A 25%-point decrease in meal reduction
M.R. Carter Before and After the Drought
6. Looking Beyond the Averages
However, average impacts can obscure as well as illuminate
Poverty trap theory suggests that coping strategies may
differentiate within communities:
Conventional consumption smoothers
(Less conventional) asset smoothers
Guided by these theoretical insights, use threshold
econometrics to split the sample along the asset continuum &
find that consistent with poverty trap theory:
Households above a critical threshold consumption smooth
without insurance; Insurance leads to a 64%-point drop in
asset sales for these less poor households
Households below that threshold “asset smooth”; Insurance to
a 43%-point reduction meal reduction as a coping strategy for
these poorer households
M.R. Carter Before and After the Drought
7. Coexistence of Asset & Consumption Smoothing
M.R. Carter Before and After the Drought
8. Index-based livestock insurance (IBLI)
Drought insurance for livestock launched in January 2010
International Livestock Research Institute, Cornell University,
Syracuse University and the BASIS Research Program at UC Davis.
“Index-based": uses satellite-based NDVI (normalized difference
vegetation index) measures of available vegetative cover to predict
livestock mortality
M.R. Carter Before and After the Drought
9. Timeline of Events
Survey 924 Households in October 2011.
All households had access to insurance.
25% of households purchased insurance.
M.R. Carter Before and After the Drought
11. Econometric Identification Strategy
IBLI had a random spatial rollout, but this analysis restricted to
treated areas & identification through an encouragement design
In each period 60% of surveyed households were randomly selected
to receive coupons offering a 10-60% discount on the first 15 TLU
insured.
Ii = Zi δ + Xi θ + vi (1)
yQ
i = β Ii + Xi φ + εi (2)
M.R. Carter Before and After the Drought
12. Threshold Model
Use Hansen’s threshold estimator (2000) to predict critical asset
threshold, L∗, at which households switch between
consumption/asset smoothing.
Estimate impact of insurance separately for “asset poor" and “asset
rich" households:
yQ
i =
βlow Ii + Xi φlow + εi,low if Li ≤ L∗
βhigh Ii + Xi φhigh + εi,high if Li > L∗
(3)
M.R. Carter Before and After the Drought
15. Impacts of Anticipated Payoffs
Steady deterioration of rangeland over 2011 made it quite
likely that a payout was coming
IBLI regularly posts index condition with a TSA-like color
scheme
See in descriptive statistics that in quarter 3, households
already deploying asset & consumption smoothing strategies
Might anticipated payouts influence this behavior (after the
drought, but before payout)?
For asset smoothers, maybe (if trust in insurance)
For consumption smoothers, no (unless could borrow against
future payout)
M.R. Carter Before and After the Drought
18. Summary of IBLI Impact Analysis
Data reveal ample evidence of costly, differentiated coping
strategies:
Households below critical threshold much more likely to asset
smooth, with predictable consequences on future human
capital (see paper with Sarah Janzen)
Households above critical threshold more conventionally
consumption smooth, with predictable consequences on
livestock prices & own future well-being
Insurance has large impacts on both strategies:
Cuts in half asset smoothing (or its severity)
Similar reduction in asset sales
M.R. Carter Before and After the Drought
19. Index Insurance in Mali (& Burkina Faso)
Farmers pursue a diversified production strategy of growing
their own food plus some cotton
Value chain credit via group loans, but consequences of default
are substantial (informal collateral)
Joint liability itself discourages investment as the more a
farmer produces, the more likely that some of his output will
be ’taxed’ away to pay for others in the group
Farmers report growing less cotton then they otherwise would,
or by reducing financial risk exposure by investing less in the
crop
Result is that risk keeps these farmers poorer than they need
be given the economic opportunities available to them
M.R. Carter Before and After the Drought
20. Two Trigger Area Yield Contract
Area yield contracts can in principle offer strong insurance
value (compared to weather-based insurance)
But over what geographic should yields be calculated?
A small area (e.g., the individual farmer’s field in the extreme)
creates a moral hazard problem
A too large area (e.g„ average yields for an entire department
or even country) lessens the quality of the insurance
So might two triggers be better than one?
Primary trigger set a small area (e.g., village)
A higher level “audit” trigger can control moral hazard
Results are encouraging:
M.R. Carter Before and After the Drought
21. Two Trigger Area Yield Contract
M.R. Carter Before and After the Drought
22. Descriptive Statistics
Original Reclassified
(1) (2) (3) (4)
Obs Control Treatment Control Treatment
Post Intervention
Area (ha) 954 2.61 2.82 2.53 2.92*
Harvest (kg) 942 2,630 2,729 2,567 2790
Loans (CFA) 967 321,571 332,645 301,740 348,661
Seed (CFA) 952 13,453 16,327 12,749 17,338**
Fertilizer (CFA) 952 135,253 145,301 133,044 148,538
Offered insurance? 25.00% 75.00% 25.00% 84.00%
Are you insured? % yes 0.00% 27.00% 5.00% 29.00%
Baseline
Cotton area (ha) 586 2.20 2.41 2.19 2.44
Cotton harvest (kg) 584 2,339 2,288 2,316 2,291
Age household head 971 55.85 55.33 55.19 55.71
Education head 948 3.50 3.98 3.89 3.77
M.R. Carter Before and After the Drought
23. Impact Results
Standard Instrumental Variable LATE Identification strategy
Look at results using original and reclassified households
See similar results if look at insurance perception as
endogenous variable as opposed to purchase insurance
Also similar results if control for baseline production
characteristics (using smaller sample on which have full
baseline information)
M.R. Carter Before and After the Drought
24. Impact Results
Reclassified Original
Loans Area Harvest Other crops Loans Area
(kCFA) (ha) (kg) (ha) (kCFA) (ha)
Insured 114.462* 1.389** 951.931 -0.511 60.387 1.106*
Constant 94.203 0.270 171.284 3.929** 95.911 0.279
Obs 949 938 925 952 949 938
R2 (adj) 0.112 0.097 0.156 0.087 0.145 0.129
First stage
Treated 0.24*** 0.254***
Penalty -.0004** -0.00038*
Constant 0.092 .0198
Obs 952 938
R2 0.351 0.449
F-stat 10.11 12.61
M.R. Carter Before and After the Drought
25. Designing Comprehensive to Promote Mobility via the Risk
Reduction Dividend
Have seen that:
Index insurance can provide real protection to consumption &
assets
Risk reduction dividend works in Mali
This “risk reduction dividend” corroborated by a few other
recent studies (Karlan et al., Emmerick et al.)
So how do we design, or not design, index insurance to reap
this dividend?
However, policy thinking is often segmented–insurance is
either thought of for social protection for the “poor,
subsistence” households or for economic advancement for
“commercial” farmers
Peru
Kenya
M.R. Carter Before and After the Drought
26. Designing Comprehensive to Promote Mobility via the Risk
Reduction Dividend
But class positions are not fixed
If do not design comprehensive program, may miss opportunity
to use insurance as tool for upward mobility, moving
households from subsistence to commercial
Need to work on more comprehensive designs that allow
poorer households to insure marginal risks that might take on
when insured
M.R. Carter Before and After the Drought
27. Conclusions
Seen that index insurance can work
I have not discussed substantial design, pricing and demand
issues for insurance
Comprehensive insurance suggests new economic logic for
cost-sharing via smart subsidies
M.R. Carter Before and After the Drought