mSHAP: SHAP Values for Two-Part Models

Two-part models are important to and used throughout insurance and actuarial science. Since insurance is required for registering a car, obtaining a mortgage, and participating in certain businesses, it is especially important that the models that price insurance policies are fair and non-discriminatory. Black box models can make it very difficult to know which covariates are influencing the results, resulting in model risk and bias. SHAP (SHapley Additive exPlanations) values enable interpretation of various black box models, but little progress has been made in two-part models. In this paper, we propose mSHAP (or multiplicative SHAP), a method for computing SHAP values of two-part models using the SHAP values of the individual models. This method will allow for the predictions of two-part models to be explained at an individual observation level. After developing mSHAP, we perform an in-depth simulation study. Although the kernelSHAP algorithm is also capable of computing approximate SHAP values for a two-part model, a comparison with our method demonstrates that mSHAP is exponentially faster. Ultimately, we apply mSHAP to a two-part ratemaking model for personal auto property damage insurance coverage. Additionally, an R package (mshap) is available to easily implement the method in a wide variety of applications.

Derivatives or climate insurance? Risk management tools for a soybean farmer in Brazil

This paper aims to analyze whether there is a difference between the cost/benefit of climate risk reduction and cost/benefit of soybean price in Palotina, Paraná, Brazil. The data were collected from bibliography, official and private documents. Climate data were analyzed based on the releases content. The methodology to analyze de future market data was the same as the B3. Options were analyzed as Black & Scholes model. Also, this paper developed the Leismann & Bortoluzzi index to analyze the protections cost/benefit. To compare mitigation costs, there were used the t student test. Limitations were about the Black & Scholes model, which does not consider subjective variables. Cost/benefit index of price protections were compared with the climate insurance index in order to test if there was a statistical difference between them. All tests allowed to infer that the indices are statistically different. This study concluded that climate and price insurance are excellent tools for rural enterprise risk management, and there was significant evidence to infer that the protections are feasible. Or rather, that the farmer is exposed to both types of risk and that the forms of mitigation are satisfactory in both cases.

Insuring Risks Associated With the Production and Sale of Marijuana

The strong growth potential of the cannabis industry offers insurers significant new business expansion opportunities, but the industry is complex and faces many unique risks. Legalized cannabis businesses need and desire insurance to mitigate these risks. However, the availability of insurance lags behind this need for insurance. Marijuana is still illegal at the federal level, causing many insurers to be hesitate to insure a federally illegal activity. These legal uncertainties will continue until Congress settles the conflict between federal and state laws. Further, there is little data available on the cannabis industry, making it difficult to price insurance products. However, rapid expansion of the cannabis industry has produced multiple unique needs and complex risks that need to be insured. This paper examines many of the risks faced by the cannabis industry, and the issues keeping many insurance carriers from entering this growing market.

The Price Insurance Demand of Rice Producers in the Vietnamese Mekong Delta

Using the dichotomous choice contingent valuation method, this chapter helps shed light on the potential for marker-based insurance schemes in Vietnam by empirically exploring the demand for minimum price insurance among rice households. The study showed that the majority of rice farmers accepted the guaranteed price of VND 4,500 per kg, and their accepted insurance fee was about 13% of the guaranteed price and 30% of the break-even price. Farmers growing rice under a monoculture system were less likely to pay for the proposed insurance service, while those with access to any formal credits were more likely to pay for it.

Crop price insurance in China: pricing and hedging using futures market

Purpose The purpose of this paper is to demonstrate how crop price insurance premium can be calculated using an option pricing model and how insurers can transfer underwriting risks in the futures market. Design/methodology/approach Based on data from spot and futures market in China, this paper develops an improved B-S model for the calculation of crop price insurance premium and tests the possibility of hedging underwriting risks by insurance firms in the futures market. Findings The authors find that spot price of crops in China can be estimated with agricultural commodity futures prices, and can be taken as the insured price for crop price insurance. The authors also find that improved B-S model yields better estimation of crop price insurance premium than traditional B-S model when spot price does not follow geometric Brownian motion. Finally, the authors find that hedging can be one good alternative for insurance firms to manage underwriting risks. Originality/value This paper develops an improved B-S model that is data-driven in nature. Insured price of the crop price insurance, or the exercise price used in the B-S model, is estimated from a co-integration model built on spot and futures market price series. Meanwhile, distributional patterns of spot price series, one important factor determining the applicability of B-S model, is factored into the improved B-S model so that the latter is more robust and friendly to data with varied distributions. This paper also verifies the possibility of hedging of underwriting risks by insurance firms in the futures market.

Structure Specific Flood Risk Based Insurance

This paper highlights the importance of developing accurate flood hazard maps to price insurance effectively and to communicate flood risk to interested parties. Risk-based insurance premiums can encourage insurance purchase and investment in cost effective mitigation measures. We undertake a study using light imaging detection and ranging (LIDAR) technology and depth damage curves to determine risk-based rates for residential structures in three counties in the state of North Carolina. We then compare these prices with current premiums charged to homeowners by the National Flood Insurance Program (NFIP) for 11,915 single-family residences. NFIP premiums are significantly higher than risk-based premiums for over 90 percent of the homes in each of the counties in our study. Risk-based prices are higher than NFIP premiums only in instances where buildings are predicted to suffer damage from more frequent, shallow floods that are currently not considered explicitly in NFIP premium calculations. Accurate flood maps are needed to determine cost-effective loss reduction measures and to address issues of affordability and fairness for homeowners currently living in flood-prone areas.