BYTE SIZED

BASIS RISK & WBCIS

ABOUT Presenting the ‘byte-sized’ series from the GIZ-Rural Insurance Services Programme. Easily digestible technical insights on topics across crop insurance and agricultural risk management. The ‘byte-sized’ series endeavours to present its findings in a neat and conclusive format, backed by extensive desk and field research. Information is accentuated by interesting graphics and posters with storylines. I hope you find these an interesting read. We look forward to your feedback and suggestions. Please feel to write to us at [email protected]

EXECUTIVE SUMMARY Crop insurance is an invaluable financial instrument for mitigating risks inherent to agriculture, providing producers the means to transfer and share risks – whereby the losses suffered by few are met from funds accumulated through contributions made by many exposed to similar risks. This document aims to present a succinct overview of the Indian Weather index-based Crop Insurance Scheme (WBCIS), its potential and challenges, examining in particular the inherent phenomenon of “basis risk” prevailing within these schemes. The report features a field analysis of basis risk by GIZ & AICI in Rajasthan on the basis of which essential policy recommendations are provided.

1

BASIS RISK & WBCIS Deciphering weather index-based crop insurance schemes (WBCIS) 1. What is WBCIS? Weather index-based crop insurance scheme (WBCIS) is intended to protect insured farmers against financial implications of crop loss resulting from adverse weather conditions such as rainfall, temperature, frost and relative humidity. While traditional crop insurance directly indemnifies individual producers against shortfalls in crop yields, WBCIS uses indices calculated on the basis of weather parameters as proxies for crop yields. Compensation for crop loss takes place across designated ‘Reference Unit Areas’ (RUA – usually a tehsil or gram panchayat), linked to reference weather stations. Through the empirical analysis of historical weather data, thresholds (triggers) are estimated beyond which crops are deemed to be adversely affected. Pay-out structures are then developed to compensate producers to the extent of losses deemed to have been suffered on the basis of the recorded weather conditions.

13%

20%

13%

WBCIS 24%

4% 13%

6%

7%

Oilseeds Groundnut Other cereal Wheat Paddy Fruit/plantation Commercial crop Pulses

Source: Report of the committee to review the implimentation of crop insurance schemes in india, MoA, GoI, May 2014

CROP SEGMENTS Rice, wheat

1 Food Grains

Jawar, bajra, ragi

2 Other cereals / Millets

ABOUT The WBCIS scheme is applicable to a variety of crops including seasonal, perennial, low-value and high-value crops, covering the entire gamut of food crops, oilseeds, annual commercial crops and horticultural crops. Risk cover is available for the sowing to harvesting period for standing (food) crops for non-preventable risks, partial coverage for preventive sowing/transplanting risk, and for post-harvest losses for particular crops due to extreme weather events such as cyclones.

3 Pulses

Gram, tur, arhar, urd, mung, masur, kulthi, matar (peas)

Cotton, sugarcane

4 Commercial crops

Tea, coffee

5 Plantation crops (non-fruit)

6

Rapeseed, mustard, linseed Oilseeds

7 Fruits

Mangoes, grapes, apple, apricot, orange, banana fresh, avacados, guava, litchi, papaya, sapota, water melons

Source: Basic risks & WBCIS - A Byte Sized-Report from GIZ, September 2015

2

BASIS RISK & WBCIS Relevant weather indices are calculated from a time-series of parameters derived from data collected at reference weather stations. For example, the weather index for rainfall is a function of deficit rainfall, consecutive dry days, number of rainy days, excess rainfall and consecutive wet days. For a given coverage period (e.g. sowing, harvesting etc.), deviations from the threshold level of an index for a particular RUA ‘triggers’ a pay-out for the insured producers based on a pre-defined structure.

TRIGGERS USED IN WEATHER BASED CROP INSURANCE SCHEMES (WBCIS)

DISEASE PROXY Combination of weather parameters like rainfall, temperature and humidity

?

WIND SPEED High wind speed

?

RAINFALL Deficit rainfall, consecutive dry days, no. of rainy days, excess rainfall, consecutive wet days

RELATIVE HUMIDITY High humidity

% TEMPERATURE Max temp (high), min temp (low), frost, mean temperature, hourly chilling units Source: Report of the Working Group on Outreach of Institutional Finance, Cooperatives and Risk Management, 12th FYP, Planning Commission, November 2011

3

BASIS RISK & WBCIS 2. Outreach of WBCIS in India? Weather index insurance was originally piloted in India in 2003 by ICICI Lombard General Insurance Company Limited with subsequent pilots by the Agricultural Insurance Company of India (AICI) and IFFCO-Tokio across select states in 2004. Using the feedback from its pilot, AICI designed the first WBCIS product, and implemented it in the Kharif 2007 season as an alternative to existing insurance products available to producers. Private sector insurers were allowed to participate in the implementation of WBCIS from the Rabi 2007 season. Initially, the scheme did not gain adequate traction due to a lack of customer appeal for a useful yet perceivably expensive product, but interest was reinvigorated after the introduction of affordable pricing policies, leading to its current status as the most widely used crop insurance instrument in Indian agriculture. In the period 2007-08 to 2012-13, nearly 47 million farmers and 63 million hectares of land were covered under WBCIS cumulatively.

WBCIS - Outreach Parameters Year

Farmers Insured (`000)

Area Insured (`000 ha)

Sum Insured (Rs. Cr)

Claims (Rs. Cr)

Farmers Benefitted (%)

Claim Ratio as %

Loss Cost (claims as % of sum insured)

07-08

679

1068

1792

105

33

70.8

5.86

08-09

376

482

887

49

61

59.9

5.52

09-10

2363

3422

4974

345

64

77.1

6.94

10 -11

9305

13148

14331

635

46

49.2

4.43

11 -12

11675

15733

20725

1176

54

63.8

5.67

12 -13

13614

18117

23604

1931

79

86.8

8.18

13 (K)

8927

11230

14638

1043

62

70.4

7.12

Total

46939

63200

80951

5284

62

70.2

6.5

Source: Report of the Committee to Review the Implementation of Crop Insurance Schemes in India, MoA, GoI, May 2014

“The Bajra crop has amounted to nothing due to lack of rain, the crop developed no kernels. With even a little bit more rain we might have grown something but there’s nothing now.”

“You can see the damage for yourself, it’s all gone.”

“We pay the insurance premium every year, but either the insurance product should be based on an evalution of each farm, or closer weather stations need to be installed.”

Radha Kishanji

Bhagchand Motilal

Keshra Ram

4

BASIS RISK & WBCIS 3. Why WBCIS? Weather index - based crop insurance has gained popularity over traditional agri- insurance products as it bypasses the need for discrete yield data for crop-loss estimation, a resource and time-intensive process overwhelmingly dependent on sample crop cutting experiments (CCE). Weather data is routinely collected in most parts of the world. Correlating weather indices to yield outcomes provides an immediate and effective insurance product in areas where reliable or regular yield data is not available. Apart from this, compared to area-yield insurance products such as the National Agricultural Insurance Scheme (NAIS), the process of claims pay-outs is relatively transparent and expedient. A comparison of yield-based and weather-based insurance is presented in the table below: NAIS (YIELD-BASED) 1

Captures more aspects of crop production risk, particularly idiosyncratic factors unique to a particular area (e.g. soil type, fertility)

WBCIS (WEATHER INDICES-BASED)

Only covers parametric weather-related risks like rainfall, temperature and humidity

Other parameters captured: pest infestation, drought, etc. 2

Relatively less historical data required for design (10 years)

Designing robust weather indices and correlating weather indices with yield losses requires additional data (25 years)

3

Objectivity and transparency relatively low for both insured and insurer

Objectivity and transparency relatively high

Payments of claims often delayed due to the time taken to arrive at yield estimates 4

5

Losses in quality not considered

Claims are calculated based on deviations beyond the predefined ‘trigger’ thresholds in weather indices Faster claims settlements

Pay-outs are determined by deviations in expected yields (purely yield-based)

Quality losses are inherently reflected to some extent through weather indices (adverse weather most significant determinant of changes in size and look of produce)

Loss assessment: inconveniently time intensive (time-consuming and expensive crop cutting experiments)

Losses based on indices can be calculated immediately given that weather data is constantly monitored and collected

Source: Weather Based Insurance Scheme FAQ, Agricultural Insurance Company of India Limited (AIC)

5

BASIS RISK & WBCIS 4. Present scenario and areas of dissatisfaction WBCIS is now a part of the National Crop Insurance Program (NCIP) introduced by the Government in the Rabi 2013-14 season, along with the Modified National Agricultural Insurance Scheme (MNAIS). The premiums for WBCIS, as well as for MNAIS are higher when compared to the National Agricultural Insurance Scheme (NAIS), as these schemes are seen to provide improved insurance coverage along with additional benefits (see previous table). Another advantage of WBCIS in comparison to NAIS, which remains the largest scheme in terms of number of insurance contracts, is that WBCIS premiums are calculated on an actuarial basis. Further, the liability for claims pay-out is enjoined on the insurance company and not on the insured. Any breach of pre-defined weather parameters triggers a pay-out and is easily audited, benefitting less educated smallholder farmers. To widen coverage, there is a provision for an upfront subsidy of up to 50% on premiums divided between the central and state governmnet for WBCIS as part of NCIP. In the field, location and proximity of reference weather stations to insured beneficiaries in a Reference Unit Area (RUA) is a crucial element in a weather index insurance contract. Density of weather stations is beneficial to the insurer and insured both – with more accurate and localized data for more targeted compensation, and a reduction in the spatial risk of the insurance product. There is growing unison amongst institutional stakeholders that coverage under rainfall insurance should not be offered to a farm located at a radial distance of more than 10 km, or ideally 5km, from the nearest weather station. According to an evaluation study conducted by the Agricultural Finance Corporation Ltd. in 4 states in 2011, all key stakeholders including farmers, banks, insurance intermediaries, and technical agencies, unanimously support the minimization of basis risk (see explanation below) through an improved network of reference weather stations. This is further supported by the fact that 77% of respondents from the beneficiary sample of 1200 farmers enrolled in WBCIS were not satisfied with the location of their reference weather stations. Apart from proximity to reference weather stations, grievance redressal mechanism (50%), convenience in enrolment (50%), quantum of sum insured (43%) and period of risk coverage (38%) were the other areas that respondents were dissatisfied with. Only 17% and 19% were dissatisfied with transparency and reliability of weather data respectively, while about 31% were unsatisfied with weather-index as a substitute for yield-index insurance.

6

BASIS RISK & WBCIS 5. Basis risk & weather index-based crop insurance Basis risk signifies the possibility that beneficiaries will either be undercompensated or overcompensated for the losses suffered. This means that the insurance scheme will not pay out adequately even if the producer has suffered losses, or conversely that a producer may be compensated even if she hasn’t suffered any losses. 3 types of basis risk inherent in weather insurance are to be discussed: design-effect or product basis risk, spatial basis risk and temporal basis risk (see figure below). As explained above, the hydro-meteorological models used in weather index-based insurance use a combination of weather parameters to bring out the effect of weather on crop yields. However, crop yields are moderated by a myriad of complex interactions among weather parameters. A number of other factors, including the timing of operations, soil properties, access to irrigation facilities, level of agri-inputs, technology, and crop management practices, play a role, too. While ideally one would incorporate as many significant variables into modelling the relationship between weather parameters and yield, severe data shortages and the difficulty in implementing predictive weather simulations continue to prove a major stumbling block for improving the design and performance of weather index-based crop insurance. Therefore, the source of basis risk arises out of the imperfect modelling of the relationship between weather parameters and crop yields which leads to deviations in predicted yield outcomes that in turn distorts insurance pay-outs. The figure here illustrates the deconstruction of basis risk. The relationship between the weather parameters and crop yields are determined through highly controlled crop experiments, which often translate into incorrect predictions and are the source of design error risk. The same holds true for temporal basis risk, which arises out of the inability of models to adequately account for changes in the timing of seasons. The portion of variance that is accurately explained by the hydro-meteorological model is the risk that farmers are actually insured against.

7

BASIS RISK & WBCIS 6. GIZ & AICI Case Study: A study of Basis Risk (WBCIS) in Rajasthan GIZ in association with AICI conducted a study across 148 farms in 5 districts in Rajasthan (Baran, Churu, Jaipur, Jodhpur, Udaipur) during the 2014 Kharif season, including crops like maize, soya, and beans. The study comprised of audio-video recordings during sowing and harvesting, for one reference weather station in each district. Each area served by the relevant weather station was classified into 4 zones based on proximity; with zone 1 within a radius of 1 km and zone 4 within a radius of 5-10 km. The analysis of basis risk entailed a comparison of the yield or loss estimations calculated by the expert team, the surveyor team, and the farmers. The analysis was further broken down according to a “crop-type proximity matrix” (see table below). Comparisons were done according to crop yields of the same crop in the same zone, crop yields for different crops in the same zone and yields for same crops in different zones, so as to understand the effects of crop type and proximity on spatial and design-effect basis risk. Analysis based on Crop, Zone, and Pay-out Same Crop Same Zone

Same Crop Different Zone

Event: excess rainfall. One farmer suffered severe losses The other managed to minimize his losses to less than 20% through the use of crop management practices

Events: water shortage; flooding. Farms with access to irrigation facilities could partially neutralize the adverse impact of low rainfall. For farmers without irrigation facilities, crop losses varied widely, with one farmer suffering marginal losses and the other losing his entire crop In the case of extreme weather events, crop losses tend to be homogenized, as was the case for soya bean cultivators Conclusions:

Conclusions: This suggests that basic crop management practices can greatly offset crop losses for mild weather events.

Irrigation is a high-impact crop management tool, especially in the case of mild weather events. Secondly, extreme weather events tend to have a homogenized effect on crop outcomes.

Different Crop Same Zone

Crop Losses vs. Pay-out

Event: flooding. Soya bean farmers that were used for the same crop- different zone analysis were also surveyed here, along with farmers cultivating cluster bean and pearl millet Relevant weather parameters and idiosyncratic factors determining the crop yields for different crops are starkly different For example, although farmers suffered high losses in soya bean production due to flooding, paddy cultivation was not affected

Event: shortage of rainfall. In Churu, an acute shortage of rainfall led to losses in all crop types; however, no compensation was due as per the assessment for either pearl millet or cluster beans Heavy rains in Jaipur caused severe losses in cluster beans, but no losses in pearl millet; yet the pay-out for both crops was identical This is indicative of a failure to robustly correlate crop losses to weather parameters in the WBCIS product design.

Conclusions:

Conclusion:

Any weather index-based insurance product must be highly customized and tailored for specific crops.

The first case is representative of spatial basis risk as the homogenized losses across crop type point towards a severe but localised weather event. The second case represents imperfect modelling for specific crop types, a deficiency that manifests itself as a part of design effect basis risk.

8

BASIS RISK & WBCIS

Source: Concept Note - Profiling Farms and Farmers for Crop Insurance, GIZ - RISP (2014)

7. Recommendations to diminish basis risk Spatial basis risk: Depending of the density of reference weather stations, each RUA should be allowed to reference multiple weather stations to help reduce spatial basis risk. The government plans to set up over 5000 Automatic Weather Stations (AWS) in the country, although to maintain a 5km radius for each RUA would require approximately 40,000 weather stations. There is an immediate need for public-private partnership models with mechanisms such as viability gap funding

9

BASIS RISK & WBCIS for AWS construction. Due to the lack of adequate weather station capacity, new weather stations should be installed in such a manner as to align them with agro-climatic zones. A national-level organization for approving and monitoring the quality of data acquired from AWSs should be established. This is to ensure that private agencies follow World Meteorological Organization or other approved guidelines in acquisition and reporting of data

Design effect basis risk: The effect of idiosyncratic factors must be isolated – this requires risk profiling of farmers based on parameters like crop management, irrigation etc. Risk profiling becomes the critical first step to mitigate design-effect basis risk. Although it is impractical to identify all the factors that influence crop yields, there is significant scope for risk profiling based on factors that are known to have an influence on yields. The diagram below identifies a number of significant factors that should be part of the proposed risk profiling exercise. Ideally, risk profiling should be done on the basis of individual farms, as they constitute the smallest unit of coverage under crop insurance.

Temporal basis risk: Temporal basis risk is best analysed as a combination of factors contributing to spatial basis risk and design-effect basis risk; it arises as an outcome of inadequate data sources as well as a lack of dynamism in the WBCIS product design .

8. Conclusion Weather index-based insurance is best suited to cover high-severity, low-frequency risks. This is because torrential rain or other extreme weather events are not localised and the effects of a catastrophe are similar in most areas regardless of idiosyncratic factors. On the other hand, the effects of mild weather events, such as a short delay in seasonal rains, can have a varying influence on yields depending on idiosyncratic factors and crop management practices used. This is corroborated by evidence from the GIZ-AICI study which found that the effects of marginally lower rainfall can be almost completely offset by irrigation facilities. On their own, neither weather index-based insurance nor yield-based insurance are adequate tools for effective risk mitigation. Instead, crop insurance should ideally be a combination of yield and index-based insurance schemes. For example, GIZ launched a crop insurance pilot scheme in Karnataka where weather indices work as a signalling instrument which is then followed up by a physical on site visit to assess the actual extent of the incurred loss The specialized nature of product development, the complex terminology used, and the concoction of agro-meteorology, statistics and econometrics within the underlying parameters have the undesirable effect of turning weather insurance into an incomprehensible financial instrument. The Government of India must initiate extension programs coordinated with the Technical Support Unit within the Ministry of Agriculture to raise awareness about WBCIS, as well as guide the insurance companies in product and service standardization.

10

BASIS RISK & WBCIS 9. Glossary Weather Index-Based Crop Insurance Scheme

1

WBCIS

2

RUA

Reference Unit Area

3

CCE

Crop Cutting Experiment

4

NCIP

National Crop Insurance Program

5

MNAIS

6

NAIS

National Agricultural Insurance Scheme

7

AWS

Automated Weather Stations

Modified National Agricultural Insurance Scheme

10. References Concept Note: Profiling Farms and Farmers for Crop Insurance, GIZ-RISP 2014. Weather Index Insurance: Is it the right model for providing crop insurance? ASCI Journal of Management 2011. Report of the Committee to Review the Implementation of Crop Insurance Schemes in India, MoA, GoI, May 2014. Report of the Working Group on Outreach of Institutional Finance, Cooperatives and Risk Management, 12th FYP, Planning Commission, November 2011.

11. Disclaimer This report has been prepared under the framework of the bilateral project "Rural Insurance Services Programme" supported by the Ministry for Economic Cooperation and Development (BMZ), Federal Republic of Germany, and supported by the Department of Financial Services, Ministry of Finance, Government of India. We thank all stakeholders who have provided their valuable inputs. The findings, interpretations, and conclusions expressed in this paper are those of the authors and do not necessarily reflect the views of Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH. GIZ India does not guarantee the accuracy of the data included in this work. The boundaries, colours, denominations, and other information shown on any map in this work do not imply any judgment on the part of GIZ concerning the legal status of any territory or the endorsement or acceptance of such boundaries. No part of this report may be reproduced without the prior consent of GIZ India. Rural Insurance Services Programme Published by Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH Contact Dr Detlev Holloh A2-18/Safdarjung Enclave, New Delhi 110029/India Email: [email protected] Homepage: www.giz.de

Responsible Aniruddha Shanbhag Authors Dhritiman Murti Tanja Matheis Editor Tanja Matheis

Design/Layout IJR Experiential Design Pvt. Ltd

11