Financial Services Commission of Ontario. Analysis of Loss Trend Rates for Ontario

Transcription

1 Private Passenger Automobile Insurance Introduction This document provides information on the analysis of Ontario private passenger automobile loss trend rates, as prepared by the Chief Actuary, Automobile Insurance Division, Financial Services Commission of Ontario (FSCO). The document includes a description of the data, methodology and the loss trend benchmark rates adopted for use in FSCO s review of rate filings. The loss trend benchmarks found in Exhibit of the Technical Notes are provided for reference purposes. Insurers who deviate from the benchmark rates are required to submit their own analysis of loss trend rates as part of the actuarial support in rate filings. The scope of our review of the loss trend rates includes the following coverage groups: Third Party Liability Bodily Injury Direct Compensation Property Damage Medical Benefits, Rehabilitation and Attendant Care Disability Income including Income Replacement and Non-Earners Uninsured Automobile Collision Comprehensive All Perils To isolate the effect of the 0 product reforms, our loss trend analysis has considered the effect of the 0 reforms on coverage costs. Cost data prior to the 0 reforms were adjusted based on a review of actual reform impact on costs in the 0 post-reform period. In addition, cost data were further adjusted for the expected impact of the 5 and 6 reforms based on estimated cost savings in Oliver Wyman s report, Ontario Private Passenger Automobile Insurance Costing of Product Changes (September, 5). Insurers are expected to use the most recent data in the ratemaking process. The 0 reform cost adjustments are taking on less significance as actual data experience is available. July 7 Page Private Passenger Automobile Insurance

2 Industry Loss Development Data for Ontario Private Passenger Automobile Insurance and Estimated s We have relied on the most recent industry loss development data from the Automobile Statistical Plan (ASP) that was provided to FSCO in its semi-annual review of the Ontario private passenger automobile insurance experience. This data is included in the General Insurance Statistical Agency s (GISA) Ontario Loss Development Exhibit as of December 3, 6. FSCO reviews loss cost data from the ASP on a regular basis and will continue to monitor the rate of changes in loss costs for each major coverage. FSCO s loss trend benchmarks are subject to change based on updated loss cost data as it becomes available through the ASP. As noted in the GISA actuary s Incurred Loss Development Factor report for Ontario private passenger automobile insurance business as of December 3, 6, several reporting data issues from different accident year periods could have an impact on loss development changes. In particular, a major insurer s data was excluded in the past (prior to 4-) due to its impact on actual loss development changes. However, the GISA actuary has concluded that exclusion of this major insurer s data is no longer required. As a result, FSCO s loss trend benchmarks are based on the same loss development information that is published as part of the GISA Loss Development Exhibit. Certifying actuaries who are responsible for filings may want to conduct their own loss development factor analysis and include proper actuarial documentation per rate filing guidelines. Table summarizes the estimated loss costs for the Ontario private passenger automobile insurance business as of December 3, 6, as a result of FSCO s loss development review process. These estimated loss costs are used as the basis for the development of FSCO s loss trend benchmarks. Reform Adjustments We had previously analyzed the cost impact of the 0 regulatory reforms by comparing industry loss cost data available from the ASP before and after the 0 reforms. No further analysis has been conducted as the actual cost impact from the 0 post-reform is now fully reflected in the data. We applied the 5 reform cost adjustments (as estimated by Oliver Wyman), due to the effect of the reduced pre-judgment interest rate for non-pecuniary Bodily Injury claims and reduced interest rates for disputed claims in Accident Benefits. Further, we applied the 6 reform cost adjustments, due to the effect of the product changes effective June, 6, including FSCO s additional reform factor of 0.98 for cost reduction initiatives that Oliver Wyman did not quantify. Partial reform cost adjustments are applied to the second half of the 6 accident year (6- ) loss costs as claims incurred in this period were evaluated at a mix of both pre-reform and post-reform benefit levels. July 7 Page Private Passenger Automobile Insurance

3 The above reform loss cost adjustments are necessary to bring all the cost data onto a common basis before analyzing the loss trend rates. Loss Trend Benchmarks Loss trend rate assumptions are part of the actuarial support required in Ontario rate filings. The development of FSCO s loss trend benchmarks is based on actuarial standards of practice. However, filing actuaries may undertake various approaches in their independent analysis of loss trends and provide support for their own loss trend assumptions in filings. In general, development of our loss trend benchmarks is based on a review of changes in loss costs, claim frequencies and claim severities over a period of 5-8 years. We perform statistical regression analysis on estimated loss costs at semi-annual intervals, separately for claim frequencies and claim severities. Due to the interdependency of the results, consideration is given to the changes in both claim frequency and claim severity in our loss trend estimates. To isolate the effect of product reforms, reform adjustment factors are determined to adjust historical costs to a common cost level before performing statistical regressions. For Accident Benefits, we have relied significantly on 6 years of data experience. We reviewed the statistical significance of the variables such as Time and Seasonality, as well as the goodness of fit from the regression results in our selection process. The following table is a summary of the various reform factors and loss trend results for the Bodily Injury and Accident Benefits sub-coverages. Coverage 0 Reform Cost Adjustment Factor 5 Reform Cost Adjustment Factor 6 Reform Cost Adjustment Factor 6 FSCO s Reform Factor Past Trend* Future Trend* TPL - Bodily Injury % 0.0% AB- Medical Benefits % 6.4% AB- Rehab/AC %.9% AB- Disability Income % 5.3% AB - Total % 5.3% * Cut-off date for the past and future trends is October, 6. We provide additional commentary on the selected loss trend benchmark rates for the major coverages in Appendix A. Dennis Chan, FCIA, FCAS, FSA Chief Actuary Automobile Insurance Division July 7 Page 3 Private Passenger Automobile Insurance

4 Table. Estimated s for Ontario Private Passenger Automobile Insurance Accident Half-Year BI DCPD Med Reh/LTC Med/Reh Disability Death Funeral Quebec X UA Coll. Comp. AP SP OPCF 44 Past Loss Trend -0.8% 7.% 6.4%.9% 5.4% 5.3% -.5% -.% Future Loss Trend 0.0% 7.% 6.4%.9% 5.4% 5.3% 0.0% 0.0% * Cut-off date for the past and future trends is October, % -.4% 5.8%.7% 6.% 9.9% 0.9% 0.0% -.4% 5.8%.7% 6.% 9.9% 0.9% July 7 Page 4 Private Passenger Automobile Insurance

5 Appendix A Commentary on Loss Trend Estimates and Benchmarks by Sub-Coverages for Ontario Filings Third Party Liability Bodily Injury We selected a past loss trend rate of -0.8% and future loss trend of 0.0% for the Bodily Injury sub-coverage. Our loss trend estimates fall in the range of -0.9% to 0.% based on a review of 5-8 years of data. Our regression analysis indicates that both the Time and Seasonality variables are statistically significant based on 8 years of data. The R-squared value indicates a reasonable fit of the model, while high R-squared values are indicated in the range of results. accident periods of 09- to 6- in semi-annual intervals. Lo ss C o st , 6 3, 6 As compared to our previous estimates (as of June 30, 6) of the Bodily Injury loss costs, our current estimates of the actual loss costs in accident years 5 and prior are generally in line with the previous estimates. It is uncertain that the negative trend rates will continue in the future. A future loss trend rate of 0.0% is selected considering the uncertainty. July 7 Page 5 Private Passenger Automobile Insurance

6 Accident Benefits Medical Benefits We selected a loss trend rate of 6.4% for both past and future periods. Our loss trend estimates fall in the range of 4.6% to 6.5% based on a review of 5-7 years of data. Our regression analysis indicates that both the Time and Seasonality variables are statistically significant based on 6 years of data. The R-squared value indicates a good fit. accident periods of - to 6- in semi-annual intervals , 6 3, 6 The Medical loss cost in the post-reform 6- period was $93, seasonally lower than the $0 loss cost in 5-. Medical claim frequency has increased in this period but the claims severity has decreased, resulting in lower Medical loss cost as compared to the cost a year earlier. July 7 Page 6 Private Passenger Automobile Insurance

7 Accident Benefits Rehabilitation and Attendant Care We selected a loss trend rate of.9% for both past and future periods based on 6-year regression results. Our loss trend estimates fall in the range of.7% to 4.0% based on a review of 5-7 years of data. Our regression analysis indicates that both the Time and Seasonality variables are statistically significant based on 6 years of data. accident periods of - to 6- in semi-annual intervals , 6 3, 6 The Rehabilitation and Attendant Care loss cost in the post-reform 6- period was $7.3, seasonally lower than the $87. loss cost in 5-. While Rehabilitation claim frequency was relatively stable, claims severity had decreased, resulting in lower Rehabilitation and Attendent Care loss cost as compared to the cost a year earlier. July 7 Page 7 Private Passenger Automobile Insurance

8 Accident Benefits Disability Income We selected a loss trend rate of 5.3% for both past and future periods for the Disability Income sub-coverage. Our loss trend estimates fall in the range of 4.3% to 5.3% based on a review of 5-7 years of data. Our regression analysis indicates that both the Time and Seasonality variables are statistically significant based on 6 years of data. The R-squared value indicates a very good fit. accident periods of - to 6- in semi-annual intervals , 6 3, 6 The Disability Income loss cost in the post-reform 6- period was $75.0, seasonally lower than the $83.8 loss cost in 5-. While the Disability Income claim frequency was relatively stable, claims severity had decreased, resulting in lower Disability Income loss cost as compared to the cost a year earlier. July 7 Page 8 Private Passenger Automobile Insurance

9 Direct Compensation Property Damage We selected a loss trend rate of 7.% for both past and future periods for the Direct Compensation Property Damage coverage. Our selected loss trend rate was higher than our previous estimate due to the observed increase in loss cost in 6-. We note that the frequency trend is cyclical and appears to be affected by weather in certain accident periods. In the long run, such as a period of 5-7 years, the frequency trend is approximately.4%. Our estimate of claim frequency trend rate is based on a 6-year data period. We estimated that the claim severity trend rate is in the range of 4.0% to 5.6%, based on 5-7 years of data. We consider that a 4.7% severity trend rate is reasonable. The following graph displays our estimates of the actual loss costs and the fitted values over the accident periods of - to 6- in semi-annual intervals , 6 3, 6 July 7 Page 9 Private Passenger Automobile Insurance

10 Collision We selected a loss trend rate of 5.8% for both past and future periods for the Collison coverage. Our selected loss trend rate was higher than our previous estimate due to the observed increase in loss cost in 6-. We note that the claim severity trend generally fits well but the claim frequency fluctuates in recent accident periods. The actual claim frequency data shows that there was a significant reduction in the number of claims in the 0 post-reform period, followed by an increase in the recent accident period. We estimate a 6-year frequency trend rate of.5%. Our model indicates that the Time variable is statistically significant but the Seasonality variable is not for Collision frequency. accident periods of - to 6- in semi-annual intervals , 6 3, 6 July 7 Page 0 Private Passenger Automobile Insurance

11 Comprehensive We selected a loss trend rate of.7% for both past and future periods for the Comprehensive coverage. We note that the claim severity trend fits well in the range of.% to 4.3% over the 5-7 year data period, while the number of claims has generally decreased at the rate of.5% in the past 5 years. accident periods of - to 6- in semi-annual intervals , 6 3, 6 July 7 Page Private Passenger Automobile Insurance