Improving Explainability of Major Risk Factors in Artificial Neural Networks for Auto Insurance Rate Regulation

In insurance rate-making, the use of statistical machine learning techniques such as artificial neural networks (ANN) is an emerging approach, and many insurance companies have been using them for pricing. However, due to the complexity of model specification and its implementation, model explainability may be essential to meet insurance pricing transparency for rate regulation purposes. This requirement may imply the need for estimating or evaluating the variable importance when complicated models are used. Furthermore, from both rate-making and rate-regulation perspectives, it is critical to investigate the impact of major risk factors on the response variables, such as claim frequency or claim severity. In this work, we consider the modelling problems of how claim counts, claim amounts and average loss per claim are related to major risk factors. ANN models are applied to meet this goal, and variable importance is measured to improve the model’s explainability due to the models’ complex nature. The results obtained from different variable importance measurements are compared, and dominant risk factors are identified. The contribution of this work is in making advanced mathematical models possible for applications in auto insurance rate regulation. This study focuses on analyzing major risks only, but the proposed method can be applied to more general insurance pricing problems when additional risk factors are being considered. In addition, the proposed methodology is useful for other business applications where statistical machine learning techniques are used.

Estimating Major Risk Factor Relativities in Rate Filings Using Generalized Linear Models.

Predictive modeling is a key technique in auto insurance rate-making and the decision-making involved in the review of rate filings. Unlike an approach based on hypothesis testing, the results from predictive modeling not only serve as statistical evidence for decision-making, they also discover relationships between a response variable and predictors. In this work, we study the use of predictive modeling in auto insurance rate filings. This is a typical area of actuarial practice involving decision-making using industry loss data. The aim of this study was to offer some general guidelines for using predictive modeling in regulating insurance rates. Our study demonstrates that predictive modeling techniques based on generalized linear models (GLMs) are suitable in auto insurance rate filings review. The GLM relativities of major risk factors can serve as the benchmark of the same risk factors considered in auto insurance pricing.

Estimating Major Risk Factor Relativities in Rate Filings Using Generalized Linear Models.

Predictive modeling is a key technique in auto insurance rate-making and the decision-making involved in the review of rate filings. Unlike an approach based on hypothesis testing, the results from predictive modeling not only serve as statistical evidence for decision-making, they also discover relationships between a response variable and predictors. In this work, we study the use of predictive modeling in auto insurance rate filings. This is a typical area of actuarial practice involving decision-making using industry loss data. The aim of this study was to offer some general guidelines for using predictive modeling in regulating insurance rates. Our study demonstrates that predictive modeling techniques based on generalized linear models (GLMs) are suitable in auto insurance rate filings review. The GLM relativities of major risk factors can serve as the benchmark of the same risk factors considered in auto insurance pricing.

Real-Time Flood Mapping on Client-Side Web Systems Using HAND Model

The height above nearest drainage (HAND) model is frequently used to calculate properties of the soil and predict flood inundation extents. HAND is extremely useful due to its lack of reliance on prior data, as only the digital elevation model (DEM) is needed. It is close to optimal, running in linear or linearithmic time in the number of cells depending on the values of the heights. It can predict watersheds and flood extent to a high degree of accuracy. We applied a client-side HAND model on the web to determine extent of flood inundation in several flood prone areas in Iowa, including the city of Cedar Rapids and Ames. We demonstrated that the HAND model was able to achieve inundation maps comparable to advanced hydrodynamic models (i.e., Federal Emergency Management Agency approved flood insurance rate maps) in Iowa, and would be helpful in the absence of detailed hydrological data. The HAND model is applicable in situations where a combination of accuracy and short runtime are needed, for example, in interactive flood mapping and supporting mitigation decisions, where users can add features to the landscape and see the predicted inundation.

The Effect of Changes in Cost Sharing on the Consumption of Prescription and over-the-Counter Medicines in Catalonia

Many universal health care systems have increased the share of the price of medicines paid by the patient to reduce the cost pressure faced after the Great Recession. This paper assesses the impact of cost-sharing changes on the propensity to consume prescription and over-the-counter medicines in Catalonia, a Spanish autonomous community, affected by three new cost-sharing policies implemented in 2012. We applied a quasi-experimental difference-in-difference method using data from 2010 to 2014. These reforms were heterogeneous across different groups of individuals, so we define three intervention groups: (i) middle-income working population—co-insurance rate changed from 40% to 50%; (ii) low/middle-income pensioners—from free full coverage to 10% co-insurance rate; (iii) unemployed individuals without benefits—from 40% co-insurance rate to free full coverage. Our control group was the low-income working population whose co-insurance rate remained unchanged. We estimated the effects on the overall population as well as on the group with long-term care needs. We evaluated the effect of these changes on the propensity to consume prescription or over-the-counter medicines, and explored the heterogeneity effects across seven therapeutic groups of prescription medicines. Our findings showed that, on average, these changes did not significantly change the propensity to consume prescription or over-the-counter medicines. Nonetheless, we observed that the propensity to consume prescription medicines for mental disorders significantly increased among unemployed without benefits, while the consumption of prescribed mental disorders medicines for low/middle-income pensioners with long-term care needs decreased after becoming no longer free. We conclude that the propensity to consume medicines was not affected by the new cost-sharing policies, except for mental disorders. However, our results do not preclude potential changes in the quantity of medicines individuals consume.

Hemodialysis Clinics in Flood Zones: A Case Study of Hurricane Harvey

Introduction: Hurricane Harvey (2017) forced the closure of hemodialysis centers across Harris County, Texas (USA) disrupting the provision of dialysis services. This study aims to estimate the percentage of hemodialysis clinics flooded after Harvey, to identify the proportion of such clinics located in high-risk flood zones, and to assess the sensitivity of the Federal Emergency Management Agency (FEMA) Flood Insurance Rate Maps (FIRMs) for estimation of flood risk. Methods: Data on 124 hemodialysis clinics in Harris County were extracted from Medicare.gov and geocoded using ArcGIS Online. The FIRMs were overlaid to identify the flood zone designation of each hemodialysis clinic. Results: Twenty-one percent (26 of 124) of hemodialysis clinics in Harris County flooded after Harvey. Of the flooded clinics, 57.7% were in a high-risk flood zone, 30.8% were within 1km of a high-risk flood zone, and 11.5% were not in or near a high-risk flood zone. The FIRMs had a sensitivity of 58%, misidentifying 42% (11 of 26) of the clinics flooded. Conclusion: Hurricanes are associated with severe disruptions of medical services, including hemodialysis. With one-quarter of Harris County in the 100-year floodplain, projected increases in the frequency and severity of disasters, and inadequate updates of flood zone designation maps, the implementation of new regulations that address the development of hemodialysis facilities in high-risk flood areas should be considered.