Presentation is loading. Please wait.

Presentation is loading. Please wait.

The case of automobile insurance in Canada

Similar presentations


Presentation on theme: "The case of automobile insurance in Canada"— Presentation transcript:

1 The case of automobile insurance in Canada
An Alternative Collection Procedure and Its Impact on Accuracy, Reliability and Efficiency of the CPI Program: The case of automobile insurance in Canada Alan Chaffe By Walid Ezzaouali Mathieu Lequain

2 Outline Introduction and Background
Overview of the previous data collection method: Rate books Overview of the new data collection method: Compu-Quote Data confrontation of the two data sources Advantages and disadvantages of the new data collection method Conclusion

3 Introduction and Background
“The main changes that are likely to occur at statistical agencies over the next two decades will be centred around data collection techniques.” Erwin Diewert (1999) Taking advantage of the large amount of electronic data is becoming an increasingly important task Statistical agencies use a variety of data collection techniques and new methods are continually becoming available as technology advances The scope for improving the efficiency of the data collection process can increase with the arrival of technological advancements - As we know, there exists a wide range of data collection techniques ranging from telephone calls to scanner data, surveys, internet price collection and databases. 3. The underlying objective of any approach is to collect the greatest amount of reliable data, using the most efficient means and given the environmental constraints faced by statistical agency. Often, it is the level of resources that are available to the statistical agency that puts a constraint on the sample.

4 Introduction and Background (continued)
Services are typically difficult to price They are often heterogenous Tend to be delivered in quite a disaggregated way Quality adjustments are particularly challenging Pricing services will only become more complicated as markets evolve The importance of the service sector has increased Expenditures on auto insurance in the basket increased by over 22% and the share of auto insurance increased by nearly 16% between 2001 and 2005 (W2005=2.96, W2001=2.54) Services such as insurance are typically more difficult to price compared to goods. Some reasons why this is the case is because: services are often heterogenous, they tend to be delivered in quite a disaggregated way and quality adjustments are particularly challenging as it is usually a complicated process to quantify the characteristics of services. 2.Pricing services will only become more complicated as markets evolve. Vehicles themselves are constantly evolving and the options and coverage of insurance policies are continually be modified. 3.The importance of the service sector has been on the rise and thus have become an increasingly important part of the basket. In 1981 the service sector accounted for approximately 65 per cent of real GDP and in 2006 the service sector contributed nearly 70 per cent to total GDP. 4. In particular, total expenditures on auto insurance increased by over 22 per cent between 2001 and 2005 and the share of auto insurance in the basket increased by nearly 16 per cent over the same period. Out of all insurance categories (homeowners’ insurance, tenants insurance and insurance of recreational vehicles) in the basket of goods and services, auto insurance is by far the largest.

5 Introduction and Background (continued)
The Consumer Price Index (CPI) measures price change by comparing, through time, the cost of a fixed basket of commodities The basket contains commodities of unchanging or equivalent quantity and quality, and therefore, the CPI reflects only pure price movements For the case of auto insurance this means that the profiles used to price auto insurance do NOT change

6 Introduction and Background (continued)
Statistics Canada (STC) uses the following characteristics to construct different profiles to price auto insurance: Location of residence by province and postal code Driver Profile (age, gender, marital status, etc.) Vehicle make, model and year Fixed policies (collision, liability etc.)

7 Overview of the Previous Data Collection Method: Rate Books
Rate books (RBs) are manuals produced by auto insurance companies and contain detailed information on auto insurance prices STC relies on the goodwill of insurance companies to provide updated RBs Until June 2007 STC relied solely on RBs to price auto insurance for all provinces Using RBs to price auto insurance is a mundane and labour intensive task - 2. Statistics Canada relies on the good will of insurance companies. Insurance companies are often reluctant to provide Statistics Canada with up-to-date rate books on a regular basis 3. Up until June 2007, Statistics Canada relied solely on rate books to collect insurance price data, however, beginning in June rate books were used to collect insurance prices for only New Brunswick, Newfoundland and Labrador, Prince Edward Island and Alberta. For Ontario and Nova Scotia a new database was used as will be discussed next. 4. Using rate books to price insurance is a mundane and labour intensive process. It requires and employee of Prices Division to first locate the insurance quote for the rate book, a difficult task on its own, and then manually transfer this price into an excel spreadsheet. Once all insurance quotes for each policy and for every region of the country have been entered into the spreadsheet the index is computed using excel.

8 Overview of the New Data Collection Method: Compu-Quote
Compu-Quote (CQ), an electronic database, provides Canada-wide price data for automobile and property insurance CQ was employed in June 2007 to price auto insurance for Ontario and Nova Scotia The data in CQ is updated on a bi-weekly basis CQ provides insurance rates for all vehicle makes and models and for every type of policy available in the market 1. It provides rates for over 100 insurance carriers. 2. For all other provinces the rate books are still used. 3. The Compu-Quote database is updated on a bi-weekly basis providing access to the latest market information. 4. -

9 Overview of the New Data Collection Method: Compu-Quote (continued)
CQ allows the user to generate price comparison reports and download large batches of pre-specified policies Reports provide a breakdown of the price for the various components of each policy Capacity to download large datasets in digital format CQ acts as an intermediary between the licensee and insurance companies and is used by more than 30,000 insurance professionals, 2,500 consultants and 75 insurance companies every day Data is downloaded in digital format, which is readily available for price index computation. -

10 Data Confrontation of the two Data Sources
In a previous STC study, using the current profiles, methodology and insurance companies, price movements for auto insurance using both sources of data were compared The underlying trends of the price movements and the year-over-year price changes were similar for all provinces In a previous study conducted by Statistics Canada auto insurance prices were collected assuming the same fixed policies as those insurance prices collected using the rate books. The index from the two sources of data was compared from April 2005 to August 2007. It is important to perform data confrontation studies as we move away from existing data sources and towards new ones to ensure that both sources deliver the same underlying message.

11 Data Confrontation of the two Data Sources (continued)
Compound Monthly Growth Rate of Auto Insurance Using Rate Books (RBs) and Compu-Quote (CQ) Here in this graph, RBs, is the compounded monthly growth rate using rate books (the published series) and CQ is the compounded monthly growth rate using Compu-Quote as a data source. The chart provides a comparison of the two data source for only those provinces which offer private auto insurance. As Compu-quote does not offer price data of auto insurance for provinces with public auto insurance (BC, MAN, SK, QUE). 2. The provinces of Quebec, Manitoba, Saskatchewan and British Columbia are excluded from the chart because of the public provision of insurance in those provinces. 3. From the chart it is evident that both data sources follow the same underlying trends. Compu-Quote is consistently higher than the index produced using the rate books for all provinces studied expect New Brunswick.

12 Advantages of Using Compu-Quote
Using CQ will reduce reliance on insurance companies to provide RBs CQ will offer STC a higher degree of flexibility possibility to develop new and improved methodologies Conduct experiments on insurance prices with ease The use of CQ data will result in a more credible and representative sample One in which covers a more diversified set of insurance companies and policy types INTRODUCTION: There are several advantages of implementing Compu-Quote as a data source. All of the advantages are centred around the common themes of increased quality assurance, credibility, accuracy, and flexibility. 1.Since many insurance carriers are reluctant to provide updated rate books on a timely basis, the problem of obtaining reliable data will be significantly reduced. In addition, it also reduces the burdensome on insurance companies. 2. With quick and easy access to unlimited amounts of auto insurance prices for any type of policy, Statistics Canada has the ability to evaluate their current methodology for pricing auto insurance and the possibility to develop new and improved methods. In addition, the database will allow Statistics Canada to conduct experiments on the data with ease, such as a review of the characteristics which act as the driving forces in determining the price of auto insurance. 3. Due to this labour intensive process and the limited up-to-date information the use of rate books only allows for a limited sample size and is likely to have poor representative properties, leading to the possibility of inaccurate results.

13 Advantages of Using Compu-Quote (continued)
CQ will require less human resources and manual intervention Using RBs it took three weeks to compute the auto insurance price index for Ontario; whereas using CQ it takes only two days Using a database instead of manually entering data into a spreadsheet will significantly lower the probability of human error 4. – 5. The final advantage follows from the fourth one. Using Compu-Quote will reduce the probability of human error, as manual intervention will be significantly reduced. CONCLUSION: Statistics Canada is very much concerned with quality assurance and is always seeking ways to address issues in constructing a reliable and consistent consumer price index. Implementing Compu-Quote is one means of ensuring a high degree of accuracy in the prices of auto insurance.

14 Disadvantages of Using Compu-Quote
The coverage of CQ is not comprehensive CQ does not provide auto insurance prices for QUE, MAN, SK and BC due to the public provision of auto insurance in these provinces The direct cost associated with using CQ is reasonable but is much higher than the use of RBs, which are free INTRODUCTION: In practice there is a trade-off between cost and accuracy. Typically, the high level of accuracy that is desirable requires very large sample sizes and more resources, which are often not affordable. Although all minor, there are a few disadvantages of implementing Compu-Quote. This problem, however, exists whether Statistics Canada uses rate books or Compu-Quote and is thus very minor. The costs of purchasing Compu-Quote is approximately equal to the cost of a half person year and is thus a disadvantage considering that the rate books are free.

15 Disadvantages of Using Compu-Quote (continued)
Using CQ may increase the degree of verification Necessary to implement outlier detection programs More verification required with an increased sample The risk of using a single provider Production is dependent on being able to obtain this database on a continuing basis 3. Using Compu-Quote will increase the degree of verification. Some insurance companies, for instance, will refuse to sell to specific regions of the country and will thus quote an extremely high auto insurance price in order to avoid such sales. However, Statistics Canada can overcome this problem by creating outlier detection programs. In addition, increasing the sample size will also require more verification. Overall, this is not expected to require much more effort than other quality assurance procedures which were in place when computing the auto insurance index using rate books. 4. This risk, however, is minimal due to the widespread use of Compu-Quote in both the United States and Canada and its almost 25 years of existence.

16 Conclusions It is pertinent that STC move in the direction to adopt new technologies and procedures around the data collection process Markets for goods and services are becoming increasingly complex and the data collection process will only follow suit The concerns of using CQ are completely outweighed by the advantages Decreased costs over time Increased quality assurance and credibility of the index Increased flexibility Increase sample size effortlessly Improve methodology and enhance quality analysis STC is constantly seeking ways to address issues associated with constructing a reliable and consistent CPI, not only from a methodological standpoint but also from a data sources perspective. A CPI is only as accurate as are the underlying concepts, methods and data sources. Once a revised methodology has been set and additional verifications put in place, using Compu-Quote will significantly reduce person hours. CONCLUSION: In light of the advantages that Compu-Quote will bring to Statistics Canada we believe that it is a very worthwhile investment and is one that will improve the overall efficiency of the price collection process for auto insurance while maintaining of enhancing the accuracy of the auto insurance index.

17 Conclusions (continued)
STC is very much concerned with quality assurance CQ is currently in use for Ontario and Nova Scotia but the plan is to implement the use of CQ for all possible provinces STC will still need to obtain RBs from insurance companies for the provinces not available in CQ Need to build a good relationship with them Thanks to CQ, a review of the current methodology is currently under way STC is constantly seeking ways to address issues associated with constructing a reliable and consistent CPI, not only from a methodological standpoint but also from a data sources perspective.


Download ppt "The case of automobile insurance in Canada"

Similar presentations


Ads by Google