MAIN HEADING: Ontario Automobile Insurance Anti-Fraud Task Force - Review of Research Related to the Nature and Scope of Automobile Insurance Fraud in Ontario

5 October 2012

Author:
Mr. Liam M. McFarlane
Ernst & Young LLP
Insurance and Actuarial Advisory Services
Ernst & Young Tower
222 Bay Street
Toronto, Canada M5K 1J7
Phone: 416 941-7751
Fax: 416 943-3796
Email: liam.mcfarlane@ca.ey.com

1. Introduction

Ernst & Young (“EY”) was retained by the Ontario Ministry of Finance (“MoF”) to provide consulting services to the Automobile Insurance Anti-Fraud Task Force (“the Task Force”) in their mandate to determine the nature and scope of automobile insurance fraud in Ontario.  Specifically EY was asked by the MoF to:

• Complete a comprehensive review of the report on automobile insurance fraud in Ontario prepared by KPMG (“the KPMG Report”) for the Insurance Bureau of Canada (“IBC”)
• Complete a comprehensive review of three (3) other reports on automobile insurance fraud as directed by the Task Force

In the interest of time, it was agreed with the Task Force that EY would produce two reports (i) a Preliminary Report which focuses on the estimate of automobile insurance fraud in Ontario contained in the KPMG Report, and (ii) a Final Report which, in addition to the content included in the Preliminary Report, would provide additional commentary relating to the comprehensiveness of the methodology used, the objectivity of the conclusions reached and gaps in KPMG’s approach. The Final Report would also include a review of the other reports as directed by the Task Force.

This document is our Final Report as described above.

1.1 Use and Distribution

This report and the opinions and conclusions contained herein were prepared for the use of the Task Force and were based on information supplied by the Task Force.  It is not intended or necessarily suitable for any other purpose. 

No further distribution of this report may be made without the prior permission of both Ernst & Young LLP and the Task Force.  The Task Force and Ernst & Young LLP should be notified immediately following any requests for disclosure of any part of this report.  Should the report be disclosed, it must be provided in its entirety.

1.2 Restrictions

The information and findings included in this report are based on information that was provided to us to the date of this report. We reserve the right to review our comments and modify this report should additional information become available to us subsequent to the date of this report.

We understand that the Task Force may publicly communicate the findings of this report. EY will not assume any responsibility or liability for any costs, damages, losses, liabilities or expenses incurred by anyone as a result of circulation, publication, reproduction, use of or reliance on this report. Comments in our report are not intended, nor should they be interpreted to be, legal advice or opinion as we are not qualified to provide such advice or opinion.

2. Executive Summary

KPMG’s review was comprehensive in scope as they considered numerous research articles, books and journals as well as internet websites with material relevant to their review. They met with various individuals from the IBC as well as representatives from a number of the leading automobile insurers in Ontario, government insurers and organizations in Canada, the United States and internationally, who are involved in servicing the insurance industry in fraud investigation and related matters.

KPMG relied on various closed claims studies in making their estimate of insurance fraud. The conclusions that they have reached are objective since they did not perform their own analysis, but rather applied the results of other studies to recent data to estimate insurance fraud. We believe, however, that the percentage values of Opportunistic Fraud selected by KPMG should be applied to injury claims only since the studies from which the values were extracted related to injury claims only. In addition, we believe that the omission of Premeditated Fraud has the potential to materially underestimate total automobile insurance fraud in Ontario. We agree with KPMG that the level of Organized Fraud in Ontario is likely greater than that estimated by the POC’s referred to in KPMG’s report. If revisions were made to KPMG’s estimates to reflect (i) Premeditated Fraud (ii) the correct treatment of non-injury claims and (iii) a more fulsome estimate of Organized Fraud, then we believe that KPMG’s estimate of automobile insurance fraud ($769 million of $1,560 million) is not unreasonable.

The Task Force asked that we review the 2011 Closed Claim Study Concerning Auto Injury Fraud in New York State (“the New York Report”). The New York PIP insurance product and the Ontario Accident Benefits insurance product both reimburse claimants for medical expenses, loss of income and other out-of-pocket costs in the event of an accident, on a no-fault basis. In both of these jurisdictions claimants can only sue for damages if they meet a certain threshold.

Our review of the New York Report indicates that there are many similarities in the insurance environments in New York and Ontario, e.g. severity of injuries is decreasing yet cost of injury claims is increasing. Further, observations made by the authors of the New York report are similar to observations we have heard made by many participants in the Ontario insurance market e.g. the injuries are generally strains and sprains, there are multiple medical providers involved in the claims, many treatments, differences between experience in urban and rural areas etc.

The authors of the New York report estimate that fraud resulted in excess payments in the range of 26% to 34% of PIP benefits in 2010. The Ontario Accident Benefits coverage offers significantly higher benefits than the New York PIP coverage and we believe that claimants recover a greater proportion of their expenses from insurers in Ontario than claimants would in New York. The result is that we suspect that the incentive for fraud in Ontario would be at least as great as that in New York. If the figure of 26% to 34% found in this study was in fact representative of Ontario Accident Benefits fraud this would translate to approximately $1.2 billion to $1.5 billion in fraud in 2010 Ontario Accident Benefits only. When you consider potential fraud in the liability coverage as well as non-injury coverages the total potential fraud could materially exceed that estimated by KPMG.

3. KPMG Report

KPMG Forensic was retained by the Insurance Bureau of Canada (“IBC”) to undertake an examination of the extent of automobile insurance fraud in Ontario. The KPMG Report was published on June 13, 2012 and the report contained a broad review of information from Ontario and other jurisdictions relating to automobile insurance fraud as well as a quantitative estimate of automobile insurance fraud in Ontario.

This section of our report addresses our review of the KPMG Report.

3.1 Scope of Review

KPMG reviewed numerous research articles, books and journals as well as internet websites with material relevant to their review. They met with various individuals from the IBC as well as representatives from a number of the leading automobile insurers in Ontario, government insurers and organizations in Canada, the United States and internationally, who are involved in servicing the insurance industry in fraud investigation and related matters.

KPMG estimated the extent of insurance fraud by adopting findings from certain of the reports and studies included in their review; they did not perform their own analysis of insurance claims to identify those that provide indications of fraud.

3.2 Fraud Definition

KPMG indicates in their report that there are many definitions of insurance fraud, however insurance fraud is often grouped into the following three categories:

1. Organized Fraud which involves multiple participants working in concert
2. Premeditated Fraud where one individual defrauds the system
3. Opportunistic Fraud where an individual inflates an otherwise legitimate claim

We note that this definition of insurance fraud is consistent with the definition used by the Task Force.

3.3 Overview of Financial Trends

KPMG provides a useful summary of automobile insurance claims trends in Ontario over the past ten years and compares these trends to trends observed in other Canadian provinces.

The following are key observations from this information:

• The number of automobile insurance claims in Ontario declined materially (28%) over the period surveyed
• Over the same period the average Ontario automobile insurance claim increased by 135%. For accident benefits claims the increase over this period was 174%
• The average automobile insurance claim in the other provinces remained relatively flat through the same period (this was also true for accident benefits claims)
• Claims frequencies (number of claims per earned vehicle) generally declined in all provinces across the period surveyed
• The result of the foregoing is that claim costs per insured vehicle (the largest portion of the insurance premium) grew significantly faster in Ontario than the rest of the Canadian provinces
• The trends in Ontario are markedly different between the metropolitan Toronto area and the rest of Ontario
• Certain of these trends may be related, at least in part, to an increase in automobile insurance fraud

3.4 Closed Claim File Reviews

A common approach in estimating insurance fraud is to review closed claims.  These reviews are typically carried out by experienced claims personnel who review the claim file documentation for indicators of fraudulent activity. Generally a questionnaire is filled out for each claim which allows for the classification and estimation of insurance fraud.

KPMG considered a number of closed claim file reviews as follows:

1. 1992 IBC Closed Claim Study
2. 2001 Canadian Coalition Against Insurance Fraud Study
3. 2007 Closed Claim Study Concerning Auto Injury Insurance Fraud in the U.S.
4. 2011 Closed Claim Study Concerning Auto Injury Insurance Fraud in New York State
5. 2011 Closed Claim Study Concerning Auto Injury Insurance Fraud in Florida

Items number 2 and 3 were used by KPMG in estimating a range of Opportunistic Fraud in Ontario automobile insurance and item 4 was included by the Task Force as one of the three extra-jurisdictional reports for us to review. Each of these items is discussed in greater detail in what follows.

It should be noted that closed claim studies generally do not include claims closed without payment. As a result the incidence of attempted fraud may be understated as a portion of claims closed without payment may in fact represent claims where claimants attempted to defraud the insurer but were denied for that or other reasons.

3.4.1 2001 Canadian Coalition Against Insurance Fraud Study

The purpose of this study was to determine the nature and extent of accident benefits, bodily injury and personal liability fraud in Canada. This study was designed to measure the incidence of premeditated and opportunistic fraud in personal injury claims. The study defined premeditated fraud as “any action or commission resulting in illicit collection of property and casualty insurance benefits” and opportunistic fraud as “the inflation of otherwise legitimate expenses that result from a real injury”¹. This definition of Opportunistic Fraud is consistent with that as defined by the Task Force.

Over 4000 closed claims were reviewed from companies representing over 60% of the Canadian Property & Casualty insurance market. A professional approach was taken to the study where instructions were provided on how to choose representative claims and claims reviewers were provided with training by a leading industry expert. A separate section of the report dealt with Ontario.

This study relates to automobile injury insurance fraud. The focus of the study was the coverages of Bodily Injury and Accident Benefits. Fraud in automobile physical damage coverages such as Collision and Comprehensive insurance is not considered in this study.

KPMG used the Ontario specific results to develop a range of estimates of insurance fraud for opportunistic and premeditated fraud. We were able to verify the calculations made by KPMG. KPMG used the high value of the range in producing their estimate of opportunistic fraud in Ontario.

This is a comprehensive study that included a reasonably significant number of closed claims. The authors of the CCAIF study appear to be credentialed to conduct studies of this nature and hence we have no reason to question the results contained in this report. The study was a national study, however due to the size of the automobile insurance market in Ontario a separate appendix was created which deals with this province specifically.

Since this study included an Ontario specific element we believe it is reasonable to assume that the high value of Opportunistic Fraud claims calculated (14.7% of claims paid) could apply to insurance injury coverages in the current Ontario environment.

In producing their estimate of opportunistic fraud in Ontario, we note that KPMG uses the high value of the range referred to above, which is based on a study of injury claims, and applies this to the total 2010 Ontario automobile (private passenger excluding farmers) claims experience to derive their estimate.  This implies that the estimated 14.7% of fraudulent claims derived from injury claims applies to non-injury (automobile physical damage) claims as well. We question this assumption for the following reasons:

1. Unlike automobile injury coverages, the nature of automobile physical damage insurance has not changed materially over the last twenty years and hence insurers are more familiar with adjusting these claims
2. Physical damage to an automobile is typically not as subjective as an injury
3. Insurers use of technology and preferred repair shops has increased dramatically over the last ten years allowing insurers to better manage potential fraud
4. Injury claims costs consist of approximately 75% of total claims costs and the average claim is much greater (as illustrated in the table below) and hence the incentive for fraud in injury claims is likely greater.

In KPMG’s calculation of the high estimate of Opportunistic Fraud they applied the 14.7% to the estimate of ultimate claims for all coverages for Ontario Private Passenger (Excluding Farmers) Accident Year 2010 ($8,739,000,000) as estimated by GISA. This assumes that the level of opportunistic fraud in the non-injury coverages is the same as the injury coverages. We are not convinced that this is an appropriate assumption. If the high factor was applied to injury claims only the estimated amount of insurance injury fraud would be $970 million which is $314 million lower than the $1,285 million estimated by KPMG.

3.4.2 2007 Closed Claim Study Concerning Auto Injury Insurance Fraud in the U.S.

This study was completed by the Insurance Research Council (“the IRC”). The IRC is a division of the American Institute for Chartered Property and Casualty Underwriters and its purpose is to provide timely and reliable research to all parties involved in public policy issues affecting risk and insurance. The IRC is supported by leading property-casualty insurance organizations.

The study involved the collection of detailed closed claim information on over 42,000 claims from leading American insurers. Fraud is defined as “the deliberate misrepresentation of a material aspect of a claim. Fraud is distinct from buildup, which occurs when some aspect of a claim is inflated.”² KPMG considers buildup claims to be what the Task Force has considered to be Opportunistic Fraud which we think is a reasonable assumption.

The 2007 IRC Report relates to automobile injury insurance fraud. The focus of the study was the coverages of Bodily Injury, Personal Injury Protection, Medical Payment, Uninsured and Underinsured Motorist. Fraud in automobile physical damage coverages such as Collision and Comprehensive insurance is not considered in this study.

The 2007 IRC Report is a comprehensive study that included a significant number of closed claims. The IRC regularly completes studies of this nature and hence we have no reason to question the results contained in this report.

Given the nature of the insurance injury coverages provided in Ontario we believe it is reasonable to assume that the low value of buildup claims calculated (6.8% of claims paid) could apply to insurance injury coverages in the Ontario environment. 

In addition to the quantitative elements of this study there are a number of observations made by the IRC in respect of injury claims which may parallel some of the comments we have heard from insurers operating in Ontario.  Key observations, among others, are as follows:

1. The most common elements of fraud cited were fictitious injuries and injuries unrelated to the accident.
2. The claim element most likely to appear inflated was medical expenses. Among Personal Injury Protection (“PIP”) claims, inflated medical expenses were seen in 97% of apparent buildup claims.
3. Sprains and strains were more common among apparent fraud and buildup claims than among other claims.
4. Apparent fraud and buildup claims involved more types of medical providers than other claims. These claims were also much more likely to involve chiropractic treatment, Durable Medical Equipment (“DME”) such as crutches and wheelchairs, physical therapy and other alternative treatments. They exhibit higher average number of visits and higher charges form service providers. They received more Magnetic Resonance Imaging (“MRI”) treatments and used more pain clinics.
5. Among PIP claims average claimed losses with apparent fraud were 76% higher than those without the appearance of fraud or buildup.
6. Attorney involvement was much more likely with fraud and buildup claims.
7. The most common types of fraud were claims for fictitious injuries and reports of injuries not related to the accident for which the claim was filed. More than 1/3 of PIP claims involved apparent fictitious injuries.
8. Nearly ¼ of PIP claims with the appearance of fraud involved bills submitted for treatment that was not actually rendered.
9. In PIP claims there was a greater incidence of staged and caused accidents.
10. Among PIP claims with the appearance of buildup, a medical provider’s desire to obtain more compensation was cited in 69% of the claims.
11. Accidents occurring in cities with a population of over 100,000 and their suburbs were more likely to exhibit fraud or buildup than rural areas or small cities.
12. Average reported PIP losses were 76% higher for claims with apparent fraud and 69% higher for claims with the appearance of buildup.
13. Average paid losses for PIP claims with apparent fraud or buildip were more than double the average paid losses for other claims.
14. The estimate of the amount of total excess payment from fraud and buildup in 2007 ranged from 13% to 18% of all dollars paid under automobile injury coverages.

It should be noted that this study included only those claims closed with payment and hence claims closed without payment because of clear evidence of abuse were not included and hence as a result the measures of incidence of fraud and buildup are likely understated.

This was a comprehensive study involving a large number of closed claims, across a number of automobile injury coverages in a variety of states. The environmental factors cited in the study, along with the above observations parallel certain observations in Ontario, and hence we believe that it is reasonable for KPMG to use some of the findings in their quantification of automobile insurance fraud. In producing their estimate of opportunistic fraud in Ontario, we note that KPMG uses the low value of the range calculated from this study, again which is based on injury claims only, and applies this to the total 2010 Ontario automobile (private passenger excluding farmers) claims experience to derive their estimate.  This implies that the estimated 6.8% of fraudulent claims derived from injury claims applies to non-injury (automobile physical damage) claims as well. We question this assumption for the same reasons outlined in Section 3.4.1.

In KPMG’s calculation of the low estimate of Opportunistic Fraud they applied the 6.8% to the estimate of ultimate claims for all coverages (injury and  non-injury) for Ontario Private Passenger (Excluding Farmers) Accident Year 2010 ($8,739,000,000) as estimated by the General Insurance Statistical Agency (“GISA”)  . Of the $8.8 billion of Ontario automobile claims referred to by KPMG, approximately $6.6 billion relates to injury claims and $2.2 billion to non-injury claims. If the low factor was applied to injury claims only, the estimated amount of insurance injury fraud would be $449 million which is $145 million lower than the $594 million estimated by KPMG.

3.4.3 2011 Closed Claim Study Concerning Auto Injury Fraud in New York State

This study was also completed by the IRC . The report is based on a sample of 4,552 closed New York Personal Injury Protection (“PIP”) claims from 2010 (3,460 closed with payment and 1,092 closed without payment, however the results are driven by the closed with payment claims). Ten insurers, accounting for approximately 70% of the New York private passenger insurance market participated in the study.

New York’s PIP insurance provides an injured party with benefits such as reimbursement for medical expenses, lost wages and other out-of-pocket expenses related to the injury regardless of fault. The New York PIP is similar in concept to the Ontario Statutory Accident Benefits (“SABS”), however it differs in scope and limits of benefits available. As in Ontario claimants whose injuries meet a specified severity threshold are able to file a liability claim.

This study narrowly defines fraud as the material misrepresentation of the facts of loss, such as the accident, injury or treatment. Buildup, in contrast, refers to the inflation of expenses in an otherwise legitimate loss.

Key qualitative and quantitative observations from this study are as follows:

1. This report provides some background information indicating that PIP severity increased materially from 2005 to 2010 (almost 50%). Ontario accident benefit claims have also experience a marked increase in severity over the same time period.
2. There is a geographic segmenting of results into NYC Metro (“NYCM”) area versus the Rest of The State (“ROTS”). The study finds large differences between NYCM and ROTS. This is similar to assertions made by participants in the Ontario automobile insurance market regarding the Greater Toronto Area (“GTA”) and the rest of the province. Certain boroughs of NYCM are worse than others – again there is a GTA parallel based on assertions made by participants in the Ontario market relating to segments of the GTA.
3. The most common types of injuries are sprains and strains.
4. Sprains and strains are more common in NYCM than ROTS. However most measures of injury severity did not show large variations between NYCM and ROTS. Therefore differences in injuries cannot explain the large differences in claim outcomes.
5. The trend over time is to less serious injuries. Traffic fatalities and injuries fell countrywide due to improvements in safety features, traffic and road design, increased use of seat belts and fewer instances of drunk driving.
6. The large majority of claimants (82%) had no visible injury at the scene of the accident 82%. This statistic was 86% in NYCM and 78% in ROTS.
7. The report makes the assertion that “it is clear that unscrupulous medical providers have played an integral part of New York’s experience with no-fault insurance fraud and buildup”. We believe similar assertions may have been made in Ontario.
8. Most types of medical providers were more commonly used among claimants in NYCM than claimants from ROTS. For example, 40% of NYCM claimants visited acupuncturists versus 6% for ROTS.
9. Claimants generally visited multiple providers. For example, 44% of NYCM claimants visited more than 4 providers versus 14% for ROTS.
10. In NYCM claimants visited general practitioners an average of 11.6 times versus 4.4 for ROTS.
11. Chiropractic visits averaged 32 visits with little variation across the state.
12. Physical therapy also involved many visits with an average of 29.9 state wide.
13. Acupuncturists averaged 28 visits with little variation across the state.
14. Average charges from medical providers far exceeded payments made by insurers
15. The majority of providers submitted bills in excess of the relevant fee schedules.
16. Provider treatment locations and billing locations differed fairly often, ranging from 22% for chiropractor to 37% for acupuncturist.
17. The majority of claimants received at least one claim handling technique or audit feature such as independent medical exam (“IME”), peer review etc. This is more common in NYCM. Most common is IME. In most cases an IME resulted in refusal by the claimant or a reduction in the charges.
18. Other medical expenses such as diagnostic procedures and durable medical equipment (“DME” – oxygen tents, iron lungs, catheters, wheelchairs, braces, supports, canes etc) contribute to total expenses.
19. DME requires no special licensing and needs little capital and hence can be easy way to take advantage of a no-fault system. Use is more prevalent in NYCM where 37% of claimants receive DME vs ROTS at 7%.
20. Pain clinics can provide multiple treatments and diagnostics in one stop – “medical mill”. Higher in NYCM at 43% vs 12% ROTS. Pain clinic patients more likely to visit chiropractors and acupuncturists and have MRI and electromyography (“EMG”).
21. Much higher use of MRI and EMG in NYCM than ROTS. Use of Xrays is declining across the state.
22. Claimed losses for PIP claims expenses have far outpaced the rate of inflation from 2000 to 2010, especially in NYCM. Growth in PIP payments has been modest due to claimants only recovering a portion of that claimed due to other coverages, policy limits etc.
23. Medical expenses have increased as a share of total claimed losses. 92% of bills now versus 85% in 2000.
24. Less seriously injured claimants accounted for a greater share of claims in 2010 than in 2000.
25. The rate of Attorney Involvement (“AI”) has increased. Very few claimants, however, file lawsuits and the percentage that file liability claims is falling. There is a close correlation between the hiring of medical providers and the filing of lawsuits.
26. File reviewers estimated that only 15% of claims state wide could overcome the tort threshold and establish fault in 2010 vs 24% in 2000.
27. AI was associated with significantly more treatment even adjusting for similar types of injuries.
28. AI claimants were more likely to visit most types of providers. They were also more likely to received MRI’s and EMG’s, use pain clinics and have DME.
29. AI claims resulted in significantly higher claimed losses than no attorney. Payments were also much higher. This could be because more seriously injured will seek representation and required more treatment.
30. Nearly one quarter of PIP claims closed with payment were found to have the appearance of fraud and/or buildup. This was more than one third in NYCM.
31. File reviewers examined each claim for certain claim characteristics that can be associated with fraud and buildup. Most common was injuries appear excessive in relation to the accident at 23% in NYCM and 15% statewide.
32. The majority of claims were found to have no suspicion of fraud - 79%, 7% low ratings, 6% moderate ratings and 8% high ratings. Among upstate claims 92% had no suspicion of fraud.
33. The majority of claims were found to have no suspicion of buildup – 74%, 8% low ratings, 6% moderate and 12% high ratings. Among upstate claims 91% had no suspicion of buildup.
34. Nearly all buildup claims involved the inflation of medical expenses. The most common was alternative treatment provider charges, followed closely by physical therapist charges and chiropractic charges.
35. Certain claim attributes were associated with the appearance of fraud or buildup such as multiple claimants from one vehicle, where the claimant was not the named insured or a family member, involving attorneys.
36. Pain clinic treatment was reported for 86% of apparent fraud claims and 72% of apparent buildup claims. In claims with no apparent fraud or buildup only 18% visited pain clinics. DME in 66% of apparent fraud versus 11% without the appearance of fraud.
37. This study estimates that fraud and buildup added between 26% to 34% in excess payments to total PIP dollars paid in 2010.

Many of the foregoing observations are consistent with assertions that have been made by insurers operating in the Ontario automobile insurance marketplace.  We note that this study was not used by KPMG in their quantification of insurance fraud, however if the range of total fraud calculated in this report were applied to total 2010 Ontario Accident Benefits claims this would imply a level of approximately $1.2 billion to $1.5 billion of fraud relating solely to Ontario Accident Benefits claims.

3.5. Approaches to Estimating Insurance Fraud

KPMG indicates in their report that “Measuring the extent of criminal activity such as insurance fraud is not a science. There is no precise or exact way to identify all such fraudulent activities or illicit transactions and the financial impact of such activities.”³

We agree with this statement and acknowledge that estimating the extent of automobile insurance fraud is challenging. As such KPMG considered many sources in their review including the following:

1. Individual opinions obtained through surveys and interviews
2. Public opinion surveys
3. Interviews with individuals knowledgeable about insurance fraud
4. Economic Approaches
5. Closed claims studies
6. Data analytic approaches

In their report, KPMG discusses the merits and pitfalls associated with each of the foregoing approaches in some detail.

The following chart summarizes the annual estimate of automobile insurance fraud in Ontario produced by KPMG⁴.

As illustrated in the foregoing table, KPMG has not considered Premeditated Fraud in arriving at their estimate of automobile insurance fraud in Ontario.  In what follows, we will discuss the estimate of automobile insurance fraud produced by KPMG.

3.5.1 Opportunistic Fraud

KPMG considers it Opportunistic Fraud when an individual “pads” an otherwise legitimate claim. This is consistent with the definition used by the Task Force.

KPMG selects percentages of Opportunistic Fraud for both the low and high end of the range from two closed claims studies. The low end of the range is selected from the 2007 closed claim study prepared by the Insurance Research Council (“IRC”) relating to Automobile Injury Insurance Fraud in the United States (discussed in Section 3.4.2) and the high end of the range is selected from a 2001 closed claim study completed by the Canadian Coalition Against Insurance Fraud (“CCAIF”) (discussed in Section 3.4.1).

3.5.2 Organized Fraud

KPMG considers Organized Fraud to be when a group of individuals act in concert to take advantage of the insurance system. On the other hand, Premeditated Fraud involves the purposeful claiming of insurance claim benefits by an individual (rather than a group). These definitions are consistent with those of the Task Force.

As part of their review KPMG considered three “Proof of Concepts” (“POCs”) which used data analytic tools for the identification of claims that have indicators of fraudulent activity. The POCs were not undertaken for the purpose of quantifying the extent of fraud in the Ontario automobile insurance market, however extrapolations of potential fraud by two of the POCs was used by KPMG in their report. The POCs were used to provide estimates of Organized Fraud and not Premeditated Fraud, however some elements of Premeditated Fraud may be included in the estimates derived from the POCs.

The low estimate of the annual impact of automobile insurance fraud in Ontario was selected from the POC labelled by KPMG as POC-2 for confidentiality reasons and the high estimate was selected from that labelled POC-1. 

3.5.2.1 POC-1

The data used in POC-1 consisted of approximately 233,000 Ontario automobile claims (injury and non-injury) with a value of $6.8 billion from the six year period of 2005 to 2010.  This data was processed using analytics software which identified approximately 56,000 claims with $1.6 billion of payments that contained some indicators of potential fraud. These claims were then further filtered to identify some 222 “clusters” of connected groups with a total of 2,600 claims which received $54 million in payments.  According to those involved with POC-1 a review of these claims indicated that previously identified organized insurance fraud groups were flagged.

The extrapolation completed by those involved with POC-1 provided an estimated range of annual Ontario automobile insurance fraud of $200 million to $275 million.

3.5.2.2 POC-2

The data used in POC-2 consisted of approximately 1.2 million claims (injury and non-injury) with a value of $4.5 billion from the period of May 2008 to May 2011. The Ontario data set consisted of approximately 0.8 million claims with a value of $3.3 billion. The data analytic process used a scoring approach and applying this to the Ontario data produced 6,298 claims with a value of $89 million that were considered as suspicious. 

The extrapolation completed by those involved with POC-2 provided an estimated range of annual Ontario automobile insurance fraud of $175 million to $203 million.  These estimates would be consistent with an extrapolation of the Ontario figures above to total Ontario industry statistics as published by GISA.

The POCs focused on Organized Fraud where more than one entity was involved in a claim. As a result the extrapolations made by the POCs do not include Opportunistic Fraud and may have limited elements of Premeditated Fraud. There are a number of reasons enumerated by KPMG as to why they believe the POCs may produce an estimate of Organized Fraud that is understated including:

• Only a sampling of industry data was used and hence if all industry information was included it is likely that more “relationships” would have been identified increasing the number of potential Organized Fraud claims; and
• Only a limited period of data was included. Again, a longer experience period would likely increase the number of “relationships” detected.

We agree with KPMG that it is likely that their estimate of Organized Fraud is understated by the data analytics software.

3.5.3 Premeditated Fraud

KPMG has not included Premeditated Fraud in their estimate of total fraud in Ontario. We note that the 2001 CCAIF Report, which was used to select the high estimate of Opportunistic Fraud, also addressed Premeditated Fraud with respect to injury claims. The 2001 CCAIF Report classified claims as:

1. Staged Accident
2. Caused Accident
3. Fictitious Injury
4. Other Premeditated Fraud
5. Opportunistic Fraud Only
6. Fully Legitimate

KPMG used item (v) in their estimate of Opportunistic Fraud and used the POC’s for estimating Organized Fraud. If we assume that Organized Fraud equates to items (i), (ii) and (iii) above then we believe that item (iv) provides a possible estimate of Premeditated Fraud for injury claims. This would imply a range of between 2% to 4% of injury claims or between $130 million to $260 million of Premeditated Fraud related to Ontario automobile injury claims.

3.5.4 Summary Comments Relating to KPMG Quantification

We conclude the following relating to the quantification completed by KPMG.

1. We believe the percentage values of Opportunistic Fraud selected by KPMG are reasonable, however we believe that these percentages should be applied to injury claims only since the studies from which the values were extracted related to injury claims only;
2. We believe that the omission of Premeditated Fraud has the potential to materially underestimate total automobile insurance fraud in Ontario;
3. We agree with KPMG that the level of Organized Fraud in Ontario is likely greater than that estimated by the POC’s referred to in KPMG’s report;
4. If revisions were made to KPMG’s estimates to reflect (i) Premeditated Fraud (ii) the correct treatment of non-injury claims and (iii) a more fulsome estimate of Organized Fraud, then we believe that KPMG’s estimate of automobile insurance fraud is not unreasonable.

4. Other Studies

In addition to the review of the KPMG Report the Task Force has directed EY to review three (3) other reports on automobile insurance fraud. The reports that the Task Force requested we review are as follows:

1. 2007 Closed Claim Study Concerning Auto Injury Insurance Fraud in the U.S.
2. 2011 Closed Claim Study Concerning Auto Injury Fraud in New York State
3. Insurance Fraud Estimation: More Evidence from the Quebec Automobile Insurance Industry, Louis Caron and Georges Dionne, 1994-1995

Items 1 and 2 were discussed in Sections 3.4.2 and 3.4.3 respectively and hence in this section we will discuss item 3.

4.1   Insurance Fraud Estimation: More Evidence from the Quebec Automobile Insurance Industry, Louis Caron and Georges Dionne, 1994-1995

This paper uses a statistical methodology for the estimation of insurance fraud. The study used data from eighteen companies participating in the Quebec insurance market representing approximately 70% market share. This study builds on a previous study (Dionne and Belhadji) which limited its evaluation to observed fraud only. Approximately 2,500 closed claim files were reviewed in the prior study and 19 established fraud cases and 123 suspected cases (of varying degrees) were identified.

The authors in this paper ask the question “To what extent does the observed fraud underestimate the real fraud?”⁵ The authors assumption is that the fraud detection process is described by the Binomial distribution, Bin(n,p) where “n” and “p” are the unknown parameters. In this description “n” is the total number of fraud cases and “p” is the conditional probability of detecting a fraud, given that the claim is fraudulent.  They consider “p” to be the “index of efficiency” of claims adjustment staff in detecting fraud. As an example the authors contend that if the likelihood of a claims adjuster detecting fraud is 0.5 then the expected number of fraud claims should be doubled, i.e. 2n = n/0.5.

The authors use the Method of Moments approach to estimate the unknown parameters. In order to do this a number of groups of data have to be created. A stochastic process is used to sort the data into sets. There is a trade-off between the number of sets and the size of the sets; more groups gives a more stable process, however more groups implies less observations in each group and hence the weaker the conclusion that can be drawn about the number of frauds observed in each group. Various experiments are completed by varying the number of sets, the number of claims included in the sets with varying degrees of fraud suspicion.

This analysis produces an estimated range of fraud of between 2.37% and 16.53% of claim payments depending on the degree of fraud suspicion claims included in the study. The authors consider a realistic assumption to be halfway between the two extremes or approximately 9.5% of claims being fraudulent. Combining their work with that of the previous report they estimate that the monetary value of insurance fraud is between 10% to 21.8% of all claim payments.

We are not convinced that this study is overly relevant for the following reasons:

1. The study is quite dated.
2. The extrapolations are made from a small number of observed and suspected fraud claims.
3. We are not convinced that the statistical methodology is appropriate.
4. It is related to Quebec experience only.

5. Conclusions

KPMG’s review was comprehensive in scope as they considered numerous research articles, books and journals as well as internet websites with material relevant to their review. They met with various individuals from the IBC as well as representatives from a number of the leading automobile insurers in Ontario, government insurers and organizations in Canada, the United States and internationally, who are involved in servicing the insurance industry in fraud investigation and related matters.

KPMG relied on various closed claims studies in making their estimate of insurance fraud. The conclusions that they have reached are objective since they did not perform their own analysis, but rather applied the results of other studies to recent data to estimate insurance fraud. We believe, however, that the percentage values of Opportunistic Fraud selected by KPMG should be applied to injury claims only since the studies from which the values were extracted related to injury claims only. In addition, we believe that the omission of Premeditated Fraud has the potential to materially underestimate total automobile insurance fraud in Ontario. We agree with KPMG that the level of Organized Fraud in Ontario is likely greater than that estimated by the POC’s referred to in KPMG’s report. If revisions were made to KPMG’s estimates to reflect (i) Premeditated Fraud (ii) the correct treatment of non-injury claims and (iii) a more fulsome estimate of Organized Fraud, then we believe that KPMG’s estimate of automobile insurance fraud ($769 million of $1,560 million) is not unreasonable.

The Task Force asked that we review the 2011 Closed Claim Study Concerning Auto Injury Fraud in New York State (“the New York Report”). The New York PIP insurance product and the Ontario Accident Benefits insurance product both reimburse claimants for medical expenses, loss of income and other out-of-pocket costs in the event of an accident, on a no-fault basis. In both of these jurisdictions claimants can only sue for damages if they meet a certain threshold.

Our review of the New York Report indicates that there a lot of similarities in the insurance environments in New York and Ontario, e.g. severity of injuries is decreasing yet cost of injury claims is increasing. Further, observations made by the authors of the New York report are similar to observations we have heard made by many participants in the Ontario insurance market e.g.   the injuries are generally strains and sprains, there are multiple medical providers involved in the claims, many treatments, differences between experience in urban and rural areas etc.

The authors of the New York report estimate that fraud results in excess payments in the range of 26% to 34% of PIP benefits in 2010. The Ontario Accident Benefits coverage offers significantly higher benefits than the New York PIP coverage and we believe that claimants recover a greater proportion of their expenses from insurers in Ontario than claimants would in New York. The result is that we suspect that the incentive for fraud in Ontario would have to be at least as great as that in New York. If the figure of 26% to 34% found in this study was in fact representative of Ontario Accident Benefits fraud this would translate to approximately $1.2 billion to $1.5 billion in fraud in 2010 Ontario Accident Benefits only. When you consider potential fraud in the liability coverage as well as non-injury coverages the total potential fraud could materially exceed that estimated by KPMG.

¹ Premeditated and Opportunistic Fraud in Personal Injury Claims, 2001, Canadian Coalition Against Insurance Fraud

² Fraud and Buildup in Auto Injury Insurance Claims, 2008 Edition, Insurance Research Council

³ KPMG Report, Page 17

⁴ KPMG Forensic Report Dated June 13, 2012, Page 58

⁵ Insurance Fraud Estimation: More Evidence from the Quebec Automobile Insurance Industry, Caron and Dionne, 1994-1995