Análisis de Impacto Regulatorio

Covariate livestock mortality is a key source of vulnerability among pastoralists in northern Kenya and can often drives households into extreme poverty, making it difficult for them to escape once they are destitute. Effectively managing risk should help alter these dynamics. Index based livestock insurance is designed in Chantarat et al. (2009a) as a commercially viable risk management instrument offers the promise of protecting pastoralists from the impacts of covariate herd losses and is scheduled for pilot sale in early 2010 in northern Kenya. This paper uses household-level panel data sets collected in targeted communities to provide a complete analysis of the effectiveness of IBLI in managing livestock mortality risk and improving herd and welfare dynamics of the vulnerable populations. Results and implications from this paper could provide useful information for finalizing the pilot plan.

Our analysis adds to the current literature because of our focus on asset risks – rather than income risk commonly considered – and the pastoral production system of northern Kenya characterized by the existence of bifurcation in herd accumulation, both of which combine to require a unique application of analytical tools. A dynamic model is therefore used as a basis for a suite of simulation exercises along with a modified expected utility based evaluation criterion in order to take into account the potential dynamic impact of IBLI. We use household-level variables, including household-specific risk preferences elicited from field experiment in the target areas to provide provides critical information regarding the variations and distributions of IBLI performance across households and locations needed to generate realistic simulations and explore variations in willingness to pay and aggregate demand for IBLI.

Our model and simulations show that performance of a particular IBLI contract varies greatly across households and locations with different natures of livestock asset

exposures and basis-risk factors, which determines the extent to which IBLI can provide compensations for household’s livestock losses. More strikingly, we show that IBLI’s performance is also significantly influenced by household’s herd size relative to the critical herd threshold, which potentially determines the significance of IBLI in altering herd growth dynamics under the presence of bifurcations in herd accumulation. IBLI is shown to be most valuable where it helps stem collapses into poverty of vulnerable but non-poor pastoralists following a drought shock.

In contrast to available theoretical and empirical evidence of high risk premia among the poor (Rosenzweg and Binswanger 1993, Morduch 1995, Dercon 1996, among others), IBLI performance is shown to be minimal among pastoralists with very small herds far below the critical threshold despite our elicited risk preference that also exhibits the widely evidenced inverse relationship between risk preference and wealth. In our model, IBLI is not well suited for the poorest, who already slowly collapse toward destitution over time, as the premium payment tends to further speed up such herd de-cumulation during good seasons.

This implication, however, holds true in our setting as we abstract away from other potential behavioral responses to IBLI that may lead to improved welfare outcomes. The extent to which the poor can reduce their costly risk management strategies may lead to slightly different outcomes. We also ignore the possibility that IBLI can crowd-in much needed credit for the insured pastoralists including the least well-off ones in order for them to expand their herd to achieve high-growth trajectory over time. With such possibility, the value of IBLI should be more significant.

The joint impact of ex ante herd sizes and household-specific basis-risk determinants thus results in location-averaged performances that can be ranked positively with mean beginning herd size and negatively with dispersion of unpredicted asset risk. IBLI Performance is high in the main pastoral locations of

North Horr and Kargi relative to Dirib Gombo with the lowest performance of IBLI due to smallest proportion of large-scaled pastoralists and the largest dispersion of uncovered livestock asset risk. This result holds despite the evidence that predicted mortality index, on average, over-predicts the actual location-averaged mortality losses in North Horr relative to others. Therefore, our results imply that though the out-of-sample forecasting performance of the predicted mortality index serves to determine effectiveness of IBLI, the variations and distribution of beginning herd sizes and other household-specific factors seem to play a larger role in determining overall performance of IBLI in each particular area. As such, studies that ignore household- level variations may fall short of accurately capturing the performance of similar insurance contracts.

Our result shows that 10% strike contract with the highest coverage of covariate risk out-performs others for each household and location, and is there chosen for the optimal contract used in the ensuing simulations. The district-level aggregated demand is shown to be high price elastic with evidence of potentially low demand for commercially viable contract. Willingness to pay among the most vulnerable pastoralists is very sensitive to premium loadings and lower than the commercially viable rates, on average, despite its potentially high dynamic value. We therefore illustrate that safety nets in the form of subsidizing IBLI, properly targeted based on easily observed characteristics such as herd size, can prove appropriate as a cost effective poverty reduction program. Our future empirical research to be implemented in parallel to the pilot sale of IBLI early next year will provide greater insight for the most effective way to implement IBLI as a productive safety net in northern Kenya.

Cluster/ Variable % Bad-

Location Climate

Mean S.D. Min Max Mean S.D. Mean S.D. Regime

Chalbi Mortality rate 0.1 0.2 0.0 0.7 0.0 0.1 0.1 0.2

(Pooled) Czndvi_pos -1.5 15.9 -26.3 25.9 15.8 7.4 -12.9 7.3 60%

Czndvi_pre -0.7 9.9 -19.6 21.8 8.6 7.4 -6.8 5.7

CNzndvi 6.4 4.6 0.1 18.6 2.5 1.6 8.9 4.1

CPzndvi 5.5 6.0 0.0 21.4 9.9 7.0 2.6 2.7

North Horr Mortality rate 0.1 0.2 0.0 0.6 0.0 0.0 0.2 0.2

Czndvi_pos -4.8 14.3 -26.2 17.4 9.0 5.7 -15.5 7.9 56%

Czndvi_pre -2.5 9.5 -19.6 18.3 5.0 6.7 -8.4 7.0

CNzndvi 6.9 5.0 1.6 18.6 3.3 1.3 9.7 5.1

CPzndvi 4.4 5.3 0.0 20.7 7.3 6.6 2.2 2.7

Kalacha Mortality rate 0.1 0.2 0.0 0.7 0.0 0.0 0.2 0.2

Czndvi_pos -1.5 17.9 -26.3 25.9 19.3 5.9 -14.0 7.4 63%

Czndvi_pre -0.6 10.9 -16.5 21.8 10.2 8.4 -7.1 5.9

CNzndvi 6.6 5.0 0.6 16.3 2.1 1.5 9.4 4.2

CPzndvi 5.6 6.7 0.0 21.4 11.3 7.9 2.2 2.4

Maikona Mortality rate 0.1 0.1 0.0 0.4 0.1 0.1 0.1 0.1

Czndvi_pos 1.8 15.7 -17.4 24.4 20.3 4.5 -9.3 5.8 63%

Czndvi_pre 1.0 9.5 -10.8 18.7 11.2 6.7 -5.1 4.0

CNzndvi 5.6 4.0 0.1 11.1 1.9 2.0 7.8 3.1

CPzndvi 6.3 6.1 0.0 19.9 11.4 6.8 3.3 3.0

Laisamis Mortality rate 0.1 0.1 0.0 0.6 0.0 0.0 0.1 0.2

(Pooled) Czndvi_pos -3.5 16.5 -35.3 34.9 12.9 9.0 -14.7 9.7 59%

Czndvi_pre -1.9 10.1 -20.3 23.0 6.0 7.9 -7.4 7.7

CNzndvi 6.7 5.1 0.0 19.6 2.5 2.1 9.6 4.6

CPzndvi 4.8 5.8 0.0 24.1 9.3 5.7 1.8 3.6

Overall Bad Year

Czndvi_pos<0 Good Year

Czndvi_pos>=0