Weather Insurance 101, Part III

December 13, 2018
Lightning strike over a city skyline. Photo credit: Andre Furtado.Lightning strike over a city skyline. Photo credit: Andre Furtado.

Arbol focuses on disrupting insurance, specifically weather insurance. Before we discuss the specifics of weather insurance it is important to understand the concept in its simplest terms which will help set the stage for why we need to disrupt it.

In this three-part series, we will take you through the basics of insurance, different types of weather and agriculture insurance, some of the pitfalls with the current solutions as well as cost analysis.

In part one, we discussed the basic types of insurance, as well as the basics of weather and agriculture insurance.

In part two of this series, we discussed weather and agriculture insurance in the United States which lays the foundation for the global framework of these policies.

In our final post of the series, we examine the cost structure of insurance and a better solution in parametric insurance and Arbol.

Pricing Agricultural Insurance

The price of crop insurance follows the general framework of pricing an insurance contract. Specifically, the cost of insurance can be broken down into three major components:

Breaking down the components, the cost of risk refers to the actuarially fair premium, which refers to the probability distribution of outcomes for the insured, some of which require payment by the insurer. In the case of a crop insurance program, the cost of risk will be the probability of crop loss at which point the insurer is liable. The administrative costs include information costs used to minimize loss adjustment costs, asymmetric information, and other operating expenses. Loss adjustment may be significant when assessing large, geographically dispersed farms. Asymmetric information occurs when one party in an agreement has more information than the other. In agricultural insurance, the farmer has more information than an insurer on his or her risks and management practices, which can lead to adverse selection and moral hazard. Insurance is subject to a number of distortions that affect pricing and risk. Specifically, adverse selection and moral hazard are always prominent issues facing insurers.

Adverse selection refers to the tendency of only those exposed to higher risks purchasing insurance such as only sick people purchasing health insurance. Adverse selection is addressed by making insurance mandatory or using incentives to increase the size of the pool of insurance buyers. For crop insurance, especially multi-peril, which covers a range of issues, adverse selection leads to only the lower quality farms to seek insurance. This is why private sector multi-peril crop insurance has had difficulty sustaining itself. The main area with MPCI policies is the USA and other countries with high premium subsidies by the Government which keeps the insured pool large. In other countries such as Australia, which lack subsidies, MPCI programs initiated by private insurers invariably failed. Along with adverse selection moral hazard represents another problem with a wide-ranging policy such as MPCI.

Moral hazard incentivizes the insured to reduce upkeep as losses are capped. To counteract moral hazard in traditional MPCI crop insurance, insurers demand that farmers keep farms in good working condition and be able to prove they followed best practices for crop upkeep. Such an assessment is quite subjective as a multitude of decisions guide crop growth, which can often lead to insurers having inordinate power to reduce damage payments. As additional protection from moral hazard, crop insurance carries a deductible, which means farmers face the first share of damage after which insurers step in.

In addition to the asymmetric information costs, there are a number of operating costs for insurance, which include costs of damage assessment. Additionally, there are fees needed to procure clients for insurers including agent fees and broker fees. Insurers use brokers who then use local agent networks to interface with clients. Brokerage and agent fees can often total up to 30–40% of premiums.

The cost of ready access to capital adds further cost to the price of agricultural insurance. The insurer must have the capacity to pay for losses that exceed income from premiums in some years. This can be a problem with systemic risks, for example when many clients make claims at once because of a widespread drought. Insurance companies cannot deal with this type of risk very well. Insurance companies generally deal with their counterparty risks by diversifying the risk among a large pool of clients, with premium revenue from those clients not claiming in any given year offsetting losses from those clients who do claim. When the risks are systemic, insurers may not be able to diversify them away and they may suffer unacceptable losses in a single year. Insurance companies can deal with this by transferring some of their risk to reinsurers who operate on a global scale, but they, in turn, charge their own premiums, which add to the price of insurance.

Cost Advantages of Parametric Insurance and Arbol

Index or parametric insurance reduces the administrative costs meaningfully by removing asymmetric information problems as the payout is based on objective data that neither the insurer nor the farmer controls. There is also no need for a deductible since moral hazard issues are removed from the picture with parametric insurance. The data-based approach also reduces administrative costs to a certain extent though there are still brokers and agents in the parametric space. Another major benefit of parametric insurance is not having the subjective loss assessment from insurers to determine damages, which leads to fewer disputes and delays in payments. Reinsurance costs for parametric insurance should also trend down over time as the variety of policies increases and diversification for reinsurers rises as a result. For MPCI policies, farm level data is needed for reinsurance, which is not available in most places with long enough history, this is true in both developing countries and developed markets, like Australia. For parametric insurance, data availability is less of an issue as insurance contracts are generally struck on data that already has a history, which can reduce reinsurance costs.

The Arbol approach aims to take all the benefits of index insurance with far greater adoption and scale by taking out the numerous intermediaries that are currently taking fees in the insurance space. Arbol aims to take the cost reduction to the next level by directly connecting insurers with clients without intermediaries.

Parametric Insurance Globally

We discussed parametric insurance above as policies that payout based on the realization of an objective index measurement such as rainfall. Parametric insurance can cover a number of non-agricultural markets such as damages stemming from hurricanes or earthquakes. Parametric insurance has been gaining acceptance in different parts of the world and there have been a number of examples. One of the earliest forms of index insurance was crop hail insurance, which still remains a popular one. The start of index insurance were numerous pilot programs focused on a variety of developing country agriculturalists such as livestock in Kenya or Mexico and rainfall contracts in India. For example, the start of weather index insurance in the developing was likely a rainfall insurance contract underwritten in 2003 for groundnut and castor farmers in Andhra Pradesh, India. Since its early days, the availability of index insurance programs has expanded greatly showing the promise of the concept. Broadly, it is estimated there have been approximately 150 donor-supported weather index insurance pilots alone possibly spanning 50 countries. For example, one success story is The Global Index Insurance Facility (GIIF). The GIIF is a dedicated World Bank Group’s program that facilitates access to finance for smallholder farmers, micro-entrepreneurs, and microfinance institutions through the provisions of catastrophic risk transfer solutions and index-based insurance in developing countries. Funded by the European Union, the governments of Germany, Japan, and the Netherlands, GIIF has facilitated more than 3 million contracts, with $392 million in sums insured, covering approximately 15 million people, primarily in Sub-Saharan Africa, Asia, and Latin America and the Caribbean.

Notwithstanding some success stories, the potential market for index insurance is in the trillions with considerable room to grow. Unfortunately, many programs have not expanded from the pilot stage due to a mix of factors ranging from basis risk, complexity in the product, or high prices because of lack of scale. As with insurance of any type, a global network of diversified geographies is needed to get sufficient scale for insurers to reduce prices. At Arbol, we believe this can be achieved via competition on a platform facilitated by smart contracts, which keep friction costs low. We will discuss the various methods we think are needed to achieve scale below.

Parametric insurance policy detailed example

Given that Arbol is focused on expanding the adoption of parametric insurance using blockchain technology, it is instructive to understand an existing policy in detail. Both the Rainfall and Vegetation Index plans of insurance utilize a productivity factor that allows the insured to individualize their coverage based on the productivity of the acreage insured. The Insured party may elect a productivity factor between 60 percent and 150 percent, in 1 percent increments, and only one productivity factor can be selected by county, crop, and intended use.

Example: Insured A is insuring alfalfa acreage with an intended use of haying in a county with a county base value of $269.54 per acre. Insured A believes the alfalfa acreage has a greater value than the county base value and selects a productivity factor of 115, thereby increasing the dollar amount of protection per acre by 15 percent. If insured A believed the alfalfa acreage had less value than the county base value, a productivity factor less than 100 could have been selected, thereby reducing the dollar amount of protection per acre.

Insured must allocate, on their application, a percent of value to each unit. The percent of value allows the insured party with more than one unit to individualize their coverage within the requirements of the program. Using the percent of value, insureds can allocate a percentage of the total insured value to each selected index interval.

Example: Insured A has a 100 percent share in 1,000 insurable acres in the grid and elects to insure all 1,000 acres with an intended use of grazing. The county base value is $20.00 per acre. Insured A selects a 90 percent coverage level, 120 percent productivity factor, and the April — May and July — August index intervals. The dollar amount of protection per acre is $21.60 ($20.00 x .90 x 1.20), and the total policy protection is $21,600 ($21.60 x 1,000 acres). Insured A allocates 60 percent of the total value to the April — May index interval and 40 percent to the July — August index interval. Based on Insured A’s allocation, the policy protection amount for the unit comprised of the April — May index interval is $12,960 ($21.60 x 1,000 acres x 60 percent of value x 1.00 share). The policy protection amount for the unit comprised of the July — August index interval is $8,640 ($21.60 x 1,000 acres x 40 percent of value x 1.00 share). The total policy protection amount ($21,600) does not change. Regardless of how the total value is allocated between index intervals, the sum of the percentages for all index intervals, by grid ID, share, irrigated practice, organic practice (if applicable), and intended use, must equal 100 percent.

The insured must select the grid where the insured acreage is physically located, or assigned if contiguous acreage, by providing a point of reference.

Pasture, rangeland, forage acreage must be reported and insured with an intended use of either haying or grazing, as selected by the insured. Under no circumstances can the same acreage be reported or insured with an intended use of both haying and grazing in the same crop year. If the insured intends to hay and graze the acreage to be insured and the acreage meets the requirements to be insured as either, the insured must select only one intended use for the acreage.

The Rainfall Index plan of insurance is a risk management tool to insure against a decline in an index value that is based on the long-term historical average precipitation for the grid and index interval. It is best suited for producers whose production tends to follow and correlate to the historical average precipitation patterns for the grid. The Rainfall Index plan of insurance: (1) does not measure, capture, or utilize the actual crop production of any producer or any of the actual crop production within the grid; and (2) utilizes NOAA CPC gridded interpolated precipitation data.

Unlike other Federal crop insurance area plans of insurance that are based on county boundaries, the Rainfall Index plan of insurance utilizes a numbered grid system. Each grid covers an area equal to .25 degrees in latitude by .25 degrees in longitude. The grids do not follow state, county, or other geopolitical boundaries. The grids for the Rainfall Index insurance plan are created by NOAA. Each grid is assigned a specific grid ID and is individually rated based on the NOAA CPC interpolated historical precipitation data for that grid. NOAA CPC gridded precipitation data is obtained, by grid, for the following 11 specific 2- month time periods, referred to as index intervals, during a year. Not surprisingly, selecting the right months of the year is critical for effective use of index insurance and so months when rainfall matter for the farmer’s crop would be the index interval.

The USDA uses NOAA CPC precipitation grid data which we also intend to support at Arbol. The NOAA CPC precipitation data is not simply the measurement of precipitation from a specific rain gauge(s) or reporting station(s) within a grid. Each day, NOAA CPC obtains data from a minimum of four reporting stations closest to the center of the grid that report data for that day. The closest reporting station may be located outside the grid for which the data will be used. Each day, different reporting stations may be used because not all reporting stations report data every day. Accordingly, the gridded precipitation data used is an interpolated value for the entire grid and cannot be traced to a single point or reporting station(s). RMA does not receive the daily precipitation amount or which reporting station data was used. NOAA CPC data is accepted and used by other government agencies and private entities and undergoes a rigorous quality control process by the CPC to ensure accuracy.

An expected grid index is calculated for each grid ID and index interval using the long-term historical gridded precipitation data for the grid ID and index interval. The expected grid index represents the average precipitation for the grid ID during the index interval based on NOAA CPC data from 1948 to two years prior to the crop year.

Example: The expected grid indexes for 2013 crop year represents the average precipitation based on NOAA CPC data from 1948 through 2011.
A final grid index is based on NOAA CPC precipitation data and is expressed as a percentage. An index of 100 represents average precipitation, an index below 100 represents below average precipitation, and an index above 100 represents above average precipitation. Only the precipitation received during the index interval is used to determine a final grid index.

As with any index insurance contract, the USDA’s rainfall index policy only covers shortfall in rainfall and not actual crop losses.

By offering the insured parties customization and objective measurement, as well as, other benefits discussed above hopefully insured and uninsured business will find new protection opportunities and previously inaccessible capital.


Continue reading