Analytical Review: Manitoba Public Insurance Corporation Motorcycle Risk Study Prepared for: Motorcycle and Moped Industry Council Prepared by: UNIVERSITY OF MANITOBA TRANSPORT INSTITUTE Department of Supply Chain Management, I.H. Asper School of Business June, 2005
Analytical Review: Manitoba Public Insurance Corporation Motorcycle Risk Study Prepared for: Motorcycle and Moped Industry Council Allister Hickson, Ph.D., CFA The author expresses his appreciation for comments on drafts of this report made by Dr. B. Prentice and Dr. K. Darko-Mensah. - 2 -
Table of Contents Page 1.0 Introduction.. 4 2.0 Overall Review. 5 2.1 Data Time Span Inconsistencies... 5 2.2. Incomplete Statistics 5 2.3 Credibility of Results 7 2.4 Inconsistencies with Other Studies 8 3.0 Review of Specific Conclusions... 8 3.1 The Cost of the Average Motorcycle Claim is Six Times the Cost of the Average Private Passenger Vehicle Claim. 9 3.2 Motorcycles are twice as likely to be in Single Vehicle Crashes... 9 3.3 The Percent of Large Displacement Motorcycles is increasing 10 3.4 Sport Bikes are the Riskiest Type of Motorcycle... 10 3.5 Since 1999, 53.4 percent of all Motorcycle Accidents Involved Injuries while Injuries Resulted from 12.3 percent of Passenger Vehicle Accidents... 11 3.6 Claims Frequency is growing faster for Motorcycles than Private Passenger Vehicles... 12 3.7 The Motorcycle Pool in Manitoba is growing faster than the Private Passenger Pool 12 3.8 The Number of Motorcycle Claims is increasing significantly.... 13 3.9 Motorcycles Have Higher Accident Rates during Night Time Driving Compared to Private Passenger Vehicles.. 14 4.0 Conclusions 15 Endnotes 18 References. 20 Appendix A... 22-3 -
1.0 Introduction As section AI.16 of its rate application for 2005 rates, the Manitoba Public Insurance Corporation (MPI) released a study titled Motorcycle Risk Study (the study) 1. The study, a response to order number 4 of the Public Utilities Board s Order 173/03, was designed to provide the Board more information on the elements and characteristics of risks associated with motorcycles 2. In addition to its use to satisfy the requirements of the Public Utilities Board, the Corporation has also used the study to support a number of claims about the insurance cost profile of the motorcycle riding community. The Motorcycle and Moped Industry Council approached the University of Manitoba Transport Institute (UMTI) in 2005 to test the veracity of the results of the study. The review conducted by the UMTI involved the following activities: Review of the study. Review of information requests filed by MPI as part of the 2005 Public Utilities Board rate review process. Review of transcripts of cross-examination of MPI s witnesses from the October 2004 hearings related to the Corporation s 2005 Rate Application (the hearings). Preparation of a synopsis outlining statistical and other issues, if evident, in the study. This report provides the UMTI s review of the study. The report consists of three additional sections. The following section provides an overall commentary on methodology used in the MPI study. The third section examines specific comments made by the Corporation in the report and in a related press release. The final section provides conclusions. - 4 -
2.0 Overall Review At an overall level, the analysis prepared by MPI suffers from several weaknesses that undermine the conclusions that are drawn. These are as follows: Data time span inconsistencies. Incomplete statistics. Credibility of results. Inconsistency with other studies. 2.1 Data Time Span Inconsistencies The study is categorical in nature comparing a number of categories (motorcycles to private passenger vehicles, sport bikes to all motorcycles) using data over a multiyear time span. A consistent set of time frames was not used in the analysis. This is evident right from the first several pages of the study. The first set of data reported is for 4 years, which is then followed by two tables of five year average data, then a table of eight years data, followed by three years data. While this may seem to be a trivial issue, good analytical practice is to use data sets with consistent time spans over as long a time frame as feasible. The length of the data set is often dictated by availability of the data as well as the need to avoid changes in the data arising from structural changes. For example in the latter case, the introduction of no fault insurance in Manitoba in 1994 could be considered a large enough structural event to eliminate use of data prior to its introduction. MPI has data for many of the categories, for example private passenger vehicles compared to motorcycles for a relatively long time period. Failure to use the entire data set is critical as results over the longer time frame may differ substantially from those over the selected time period. Inconsistent time frames also may result in incorrect comparisons, merely reflective short term transitory effects not underlying differences and thus lead to wrong inferences being made. 2.2. Incomplete Statistics The study relies heavily on the comparison of means (or the average) between the various groups. The mean is only one standard statistical measure of central location, and is used typically with a measure of the dispersion of the outcomes, namely the variance or standard deviation 3. In the study the variance is not used, thereby information on the range of outcomes is not included. The problem is illustrated in the following example. Example 1: Suppose for the last 10 years private passenger vehicles and motorcycles had the following number of claims per 1,000 insured vehicles (or frequency). - 5 -
Year: 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Private 59 40 35 54 56 64 66 67 71 68 Passenger Motorcycle 54 48 55 51 50 56 62 68 70 66 The mean in both cases is 58 claims. Private passenger vehicles have a wider range of claims per year, from 35 to 71 compared to motorcycles which range from 48 to 70. If one were to ask the question about which group was more variability over time, the answer would be private passenger vehicles, since the range is wider. Such comparisons are absent from the study. Typically the two groups do not have identical means. In this case the standard approach to compare two groups is to calculate the mean, variance and coefficient of variation 4. If this statistical measure is used, in the example, claims per 1,000 insured units is more variable for private passenger vehicles than for motorcycles. The MPI report also lacks direct inferences and related tests of hypotheses. Example 2 outlines how a hypothesis test would apply on this case. Example 2: Suppose for the last 10 years the data for private passenger vehicles and motorcycles is the same as example 1. The hypothesis that could be tested is that the number of claims for private passenger vehicles and motorcycles is the same. In the jargon of statistical testing: o o The null hypothesis is that the mean of private passenger claims equals the mean of motorcycle claims. The alternate hypothesis is the mean of private passenger claims does not equal the mean of motorcycle claims. Depending on the outcome an additional test could be performed such as testing to see if the mean of the motorcycle class being greater than the mean of the private passenger class. The advantage of using statistical tests is they incorporate all the information included in the data and allow for a statistically valid determination of inferences and hypotheses. Standard test for comparisons of different groups are widely known to researchers and included the z 5 and t tests 6. The study also does not use statistical approaches to assessing trends in the data. An example is claims frequency per 1,000 insured units from page 7 of the study. A standard approach to trend assessment is regression analysis which is readily available in most spreadsheet software packages. Using eight years of data, a statistical trend is evident as estimated by the following simple regression equation. - 6 -
Y = 47.1 + 2.75 Trend Where Y = the number of claims per 1,000 units Trend = a time trend variable that takes the value of 1 for 1996, 2 for 1997, etc. Standard statistical techniques, namely the R 2 statistic and the t-statistic, confirm the trend 7. The regression and conclusion that claims are trending upwards is limited however by the use of relatively a short time frame. If the data was extended for two years, such as adding 1994 and 1995 as shown in example 1 a different result emerges. In this case with the additional two years of data, statistically a trend is not confirmed. Further details of the regression analysis are provided in Appendix A. 2.3 Credibility of Results Credibility as described in actuarial literature is the weight given to data relative to the weight given to other data 8. Credibility simply boils down to a method of weighting to account for small amounts of data or short time spans and randomness of results. In conformance with actuarial practice MPI uses credibility in its rate setting methodology. Of the possible choices for credibility weighting MPI uses the following formula: (N/N+K) Where: N is the number of insured units (vehicles) in the class K = 6,000 For a small group such as motorcycles this means the data is given a weight of.6 (or is 60% credible) calculated as follows: N = 9,000 insured units (or vehicles) K = 6,000 (9,000/ (6,000+9,000)) =.6 The implication is that data for the entire motorcycle pool is only 60% useful from an analytical viewpoint. The motorcycle risk study fails to credibility weight any of the data for motorcycles, even though credibility weighting is used elsewhere in the MPI methodology. Credibility weighted results may be different than those presented in the study. A further difficulty is where the study splits sport bikes from the rest of the motorcycle pool. In the case of sport bikes, the credibility of the data is: N = 1,350 insured units (or vehicles) K = 6,000 1,350/ (6,000+1,350) =.18-7 -
That is the data is 18% credible. Section C of the study which is more than one third of the analysis, is devoted to sport bikes, even though the data has such low credibility. 2.4 Inconsistencies with Other Studies The study included a variety of documents related to motorcycle risk and safety. These included, studies by the National Highway Traffic Administration 9, Oxford University Traffic Studies Unit 10, and the Parliament of Victoria, Australia 11. Noticeably absent from the supporting documentation is the Hurt 12 study, which is one of the most widely known studies related to motorcycle risk. Also absent were important studies by Bednar, et al. 13 and the Royal Society for the Prevention of Accidents 14 and other experts in this field. The literature contains studies with results that are inconsistent with the MPI results. Some examples of these include: Weather is not a factor in 98% of motorcycle accidents. 15 Failure to account for safety factors which are not controllable by the motorcyclist, such as the state of the roadway. 16 Findings that most motorcycle accidents are caused by other vehicles. 17 The study indicates that about 30% of accidents occurred during rainy periods 18. The results from the study are interpreted incorrectly since it is not specific to the time of the accident merely that it was raining sometime during the day of the accident. MPI confirmed this in materials filed with its 2005 Rate Application 19. The MPI report provides no information on accident causation factors, including the state of the roadway. Finally, MPI indicates that most motorcycle accidents are caused by the motorcyclist. 20 The MPI report would have been buttressed by a thorough review of the literature. 3.0 Review of Specific Conclusions MPI published, in a press release, a number of specific comments related to motorcycles. These were: The cost of the average motorcycle claim ($14,000) is nearly six times higher than the average cost of average passenger vehicle claim ($2,500). Motorcycles are more than twice as likely to be in single vehicle crashes than private passenger vehicles (a three-year average of 50.2 percent compared with 20 percent). The percentage of motorbikes with 1250 cubic centimeter or larger engines has nearly doubled since 1996, from 15 to 27.3 percent in 2003. Sport bikes are the riskiest category of motorcycle, with a claims frequency four times higher than other bike styles. Sport bikes make up 16% of the motorcycle - 8 -
class yet account for 47% of the claims. Sport bikes are also the most likely style of motorcycles to be in crashes under bad weather conditions such as rain. Since 1999, 53.4 percent of all motorcycle accidents involved injuries, while injuries resulted from 12.3 percent of passenger vehicle accidents. 21 These are derived from the findings of the study. The report as well provides commentary on matters such as: Claims rates are growing faster for motorcycles than private passenger vehicles. 22 The motorcycle pool in Manitoba is growing faster than the private passenger (cars and light trucks) pool. 23 The number of motorcycle claims has been increasing significantly. 24 Higher accident rates during night time driving. 25 Accident rates are higher for sport bikes during inclement weather. 26 A review of each of these findings follows. 3.1 The Cost of the Average Motorcycle Claim is Six Times the Cost of the Average Private Passenger Vehicle Claim This is an interesting finding; however, from an insurance cost perspective the finding is incomplete. It does not measure the insurance cost of motorcycles versus private passenger vehicles. The frequency of claims also is needed, with the two combined reflecting the overall cost of providing insurance to motorcycles. For example, if motorcycle claim rates were 10% the rate of private passenger vehicles then the risk from motorcycle riding would be $1,400 ($14,000 x.10) versus $2,500. The result also does not consider the responsibility for the claim. If the motorcyclist was responsible, then 100% of the cost should be assigned to motorcycles. If 100% of the cost is the responsibility of the private passenger vehicle all the cost should be assigned to that group. 3.2 Motorcycles are twice as likely to be in Single Vehicle Crashes The frequency of single vehicle accidents is an important factor in the overall cost of an insurance pool. This is because the costs of such losses are normally assigned fully to the risk of the pool. The study suggests a very high level of single vehicles accidents. This data is contrary to that typical of other jurisdictions. An example is the Hurt study which indicates that 75% of motorcycle accidents are multi-vehicle, meaning that 25% are single vehicle 27. A similar result was found by the Road Safety Committee of the Parliament of Victoria 28. The study would have benefited from a plausible explanation of the why these differences occurred. The difference in results compared to other jurisdictions may be due to the peculiarities - 9 -
of the claim handling in Manitoba. As noted in the MPI s filings for 2005 rates, the Corporation does not provide policies to its adjusters to define single or multi-vehicle accidents 29. This results in inconsistent application of definitions. Another potential explanation is that in Manitoba accidents caused by bad roads and hit and runs are classified as single vehicle claims 30. The MPI result is quite dependent on the short time span of data used. Data on page 9 of the study shows the percent of single vehicle claims to all motorcycle claims is as follows: 1996 1997 1998 1999 2000 2001 2002 2003 % Single Vehicle 28.3 20.8 22.9 36.4 46.3 48.9 50.6 51.1 The average percent of single vehicle accidents varies based on the time frame selected. If all eight years are selected, it is 38.2%. If the first 4 years are selected it is 27.1%. Only when the last years are selected do we see results that are in the 50% range. It appears there was a distinct change in the percent of single vehicle accidents in 1999 when the Corporation rolled out a new claim system. It is interesting that prior to that date MPI s data was similar to that found elsewhere. The report does not provide any explanation of the change from 1999 to 2003. 3.3 The Percent of Large Displacement Motorcycles is increasing The MPI result in this case is similar to that found in other jurisdictions. For example, the Royal Society for the Prevention of Accidents found that new motorcycles licensed with engine displacements of 500 c.c.'s or more nearly tripled between 1988 and 1998. 31 Many studies have indicated that there is no relationship between risk and engines displacement. As Langley, Mullin, Jackson and Norton 32 noted in their study of motorcycle accidents in New Zealand there was no consistent pattern of increasing risk as cubic capacity increased 33. MPI was also silent on the relationship in the report and made no assertion that there was a relationship. 3.4 Sport Bikes are the Riskiest Type of Motorcycle The fundamental problem with MPI s conclusion with regard to sport bikes is the very limited data on which the conclusion was drawn. The foundation of the MPI analysis is 3 years. The results are based on 1,231 34 vehicles per year and only 150 claims per year. 35 As noted in Section 2.3, above, based on MPI s credibility formula the data is less than 20% credible. The Corporation does not highlight other outcomes for sport bikes. For example 36 : The percent of injury claims for sport bike is less than for other types of motorcycles. The average severity of injury claims for sport bike is lower than for other motorcycles. - 10 -
Finally, data related to sport bikes prepared by the Corporation is questionable, simply because motorcycles that have been categorized as sport bikes by MPI are in fact not sport bikes. The Manitoba Coalition of Motorcycle Groups polled the Corporation on this matter during hearings related to MPI s 2005 Rate Application. The errors were serious enough to cause the Public Utilities Board of Manitoba to rule that: Manitoba Public Insurance refine its category of sport bikes to ensure only sport bikes are included within that category for the next general rate application and consult with CMMG prior to finalizing its categorization 37. MPI also makes the claim that sport bikes are also the most likely style of motorcycles to be in crashes under bad weather conditions such as rain 38. The problem with MPI s assertion in this regard is that their data does not show that the motorcycle was being ridden during the inclement weather, in this case rain. All the data shows is that there was rain sometime during the day when the crash occurred. For example it may have rained in the morning and the crash occurred at night. 3.5 Since 1999, 53.4 percent of all Motorcycle Accidents Involved Injuries, while Injuries Resulted from 12.3 percent of Passenger Vehicle Accidents Most studies have obtained outcomes similar to the study with motorcycle injury rates being higher than those of other types of vehicles. The problem with the MPI results is they imply that the motorcyclist was responsible for the crash and consequent injury. The degree of at fault is an important consideration. As Peek-Asa and Kraus 39 note: Approaching turn collisions are especially common between a left-turning vehicle and an approaching motorcycle and account for up to half of all motorcycle-vehicle intersection collisions (Hancock et al. 1986, Rahimi 1989 and Hurt et al. 1984). MPI believes accidents that are not the responsibility of the motorcyclist should be included in the costs attributable to motorcycling. Unfortunately MPI did not provide information for motorcycles and private passenger vehicles on the share of all accidents that was the responsibility of each group. A conservative estimate might be that the motorcyclist is responsible half the time and the private passenger vehicle class responsible the other half. This would imply that one half of the injury claims experienced by motorcyclists should not be counted against motorcyclists but be attributed to private passenger vehicles, making the total percentage for motorcycles only 26.7% 40. There would be an increase in private passenger vehicle claims, to include the motorcycle claims, perhaps raising the proportion somewhat. - 11 -
3.6 Claims Frequency is growing faster for Motorcycles than Private Passenger Vehicles. The study makes this statement based on four years of data from 2000 to 2003 41. During that time period motorcycle claim frequency increased on average 5.3%, increasing 8.3% between 2000 and 2001, 3.2% between 2001 and 2002, before decreasing 5.8% between 2002 and 2003. Private passenger vehicle claims frequency over the same year periods, increased 1.8%, and then 5.4% before decreasing 5.4% for an overall average of 2.5%. The short time span of the data selected is a major factor in the outcome. At the hearings into its 2005 rate application, MPI filed data showing the 6 year average increase in motorcycle claims. Over the six year time period the growth in claims was only marginally higher than for private passenger vehicles, 4.6% compared to 4.3% 42. MPI fails to highlight an important aspect of the study data. Motorcycle claim frequency is lower than private passenger claim frequency. During the hearings MPI testified to this as shown in the transcript pages 1134 to 1135: 20 MR. RAYMOND OAKES: With respect to the issue 21 of motorcycle accidents per one thousand (1000), is it not 22 true, Ms. McLaren, that the frequency for motorcycles is 23 still less than private passenger automobile? 24 25 (BRIEF PAUSE) 1 2 MS. MARILYN MCLAREN: Yes, there's a -- in 3 the -- in the '03 year, for example, there was a difference 4 of about two (2) claims per thousand with the private 5 passenger having sixty-seven point eight five (67.85) claims 6 per thousand (1,000) and motorcycles having sixty-five point 7 six (65.6) claims per thousand (1,000). 8 MR. RAYMOND OAKES: If I refer you to 9 Undertaking 15, I believe you'd have to agree with me that 10 the frequency of motorcycle claims relative to private 11 passenger automobile has been less for the past seven (7) 12 years for motorcycles than for private passenger? 13 MS. MARILYN MCLAREN: Taking one (1) year at 14 a time and comparing the claims per thousand (1,000) units 15 for motorcycles compared to passenger vehicles shown in 16 Undertaking 15, yes, each year the motorcycle is somewhat 17 less than the private passenger vehicles. 1134 1135 3.7 The Motorcycle Pool in Manitoba is growing faster than the Private Passenger Pool On page 6 of the study MPI makes comments that the motorcycle pool is growing faster than private passenger vehicles. This comment is based on the following data for insured units (exposures). - 12 -
1999 2000 2001 2002 2003 Motorcycles 5,906 6,131 7,259 7,739 8,033 Private Passenger Vehicles 608,060 612,932 618,819 628,386 640,345 The number of motorcycle exposures increased by 2,127 units from 1999 to 2003. During the same time period the number of private passenger vehicle exposures increased by 32,285 from 608,060 to 640,345. In terms of the number of risks insured there has been greater growth for private passenger vehicles than motorcycles. The MPI conclusion is drawn from the percent growth rate, not direct count of underlying risk exposures. 3.8 The Number of Motorcycle Claims is increasing significantly MPI based this conclusion, apparently, on a table on page 4 of the report which shows motorcycle claims rising from 329 in 1999 to 527 in 2003 as shown below. 1999 2000 2001 2002 2003 Motorcycle Claims 329 382 490 539 527 Motorcycle claims increased by 198 over the 5 year time period. For the 2000 to 2003 time period the increase was 145 claims. The Corporation did not provide data directly on the number of private passenger vehicle claims for comparative purposes in the report. However, it did provide data on exposures and claims per 1,000 private passenger vehicles for the May through September time period for 2000 through 2003. Based on this data the number of claims for private passenger vehicles is as follows: 2000 2001 2002 2003 Exposures (000) 612.9 618.9 628.4 640.4 Claims/ 1000 exposures 66.19 67.05 70.65 67.85 Claims 40,567 41,497 44,396 43,451 Claim numbers for private passenger vehicles increased by 2,884, 20 times greater than the number of claims increase for motorcycles. Over the 2000 to 2003 time period motorcycles averaged 484 claims per year while private passenger vehicles averaged 42,478 claims. The number of motorcycle claims is about 1.1% of the number of private passenger vehicle claims. The use of the May to September time period excludes the winter period where accident rates rise substantially for cars and trucks. If the winter months are included 43, the number of claims increased by 6,905 from 2000 to 2003 as shown below. - 13 -
2000 2001 2002 2003 Exposures (000) 612.9 618.9 628.4 640.4 Claims/ 1000 exposures 158.87 160.91 169.57 162.83 Claims 97,371 99,587 106,558 104,276 On an annual basis motorcycle claims are about.5% of private passenger vehicle claims. 3.9 Motorcycles Have Higher Accident Rates during Night Time Driving Compared to Private Passenger Vehicles This conclusion is drawn on Page 5 of the study where night time is defined as the period from 10 P.M. to 6 A.M. In Manitoba motorcycles are driven primarily during the May to September time period as is indicated in the report. This is the peak of summer characterized by long days with greater outdoor late night activity. Compared to the winter, it would be expected that accident frequency during the night-time, as defined above, would be higher in the summer. Appendix A.2 of the MPI report provides some information on the standardization of private passenger data to match motorcycles used for exposures, claim frequencies, claims severity and incurred per exposure unit. It is silent on matters related to standardization to the motorcycle riding season in terms of the time of day of claims. This silence suggests that standardization was not done. The private passenger vehicle data is likely annual, while motorcycles are reflective of the summer and consequentially the comparison is on different bases. The results would be expected to yield an outcome with motorcycles having higher accident rates than private passenger vehicles. A better comparison would be the percent of private passenger vehicle claims between 10 P.M. and 6 A.M. for the May to September time period. - 14 -
4.0 Conclusions Overall the study prepared by MPI provides some interesting data related to motorcycle risk. It uses a rudimentary approach to analyzing experience and trends in motorcycle risk. The study has analytical weaknesses that affect the interpretation that can be put on the results. From an overall perspective the study is lacking in the following areas: Inconsistent data time spans are used for the analysis. This leads to inaccurate conclusions. Good research practice is to use consistent time spans. Conclusions drawn were based on limited use of statistical methods and procedures. Typically research studies include statistical test criterion to verify the results are statistically valid. Conclusions are drawn based on very limited amounts of data. Using the MPI credibility formula, in the case of sport bikes, the credibility of the data would be less than 20%. The results are inconsistent with other studies of motorcycle risk. For example other studies have found little relationship between weather and motorcycle risk, and much lower levels of single vehicle accident occurrence. MPI did not provide information on these differences or an explanation of why the differences occur. With respect to specific comments in the study and related press release the following conclusions are evident: The Cost of the Average Motorcycle Claim is Six Times the Cost of the Average Private Passenger Vehicle Claim This comment implies motorcycles are 6 times riskier than private passenger vehicles. The statement fails to account for the frequency of claims or the responsibility assigned; it is merely the cost of a claim when it occurs. If motorcycle claim frequency and responsibility for accidents is considered this relationship could potentially be very much different. Motorcycles are Twice as Likely to be in Single Vehicle Crashes The rate of single vehicle accidents for motorcycles portrayed in the MPI study is double that found in other studies, including the widely known and distributed Hurt report. There is no explanation for this apparent inconsistency to results typically found elsewhere. - 15 -
The Percent of Large Displacement Motorcycles is increasing This result is similar to other studies. It does not mean that motorcycle risk is increasing. As Langley, Mullin, Jackson and Norton 44 noted in their study of motorcycle accidents in New Zealand there was no consistent pattern of increasing risk as cubic capacity increased 45. Sport Bikes are the Riskiest Type of Motorcycle The study draws this conclusion based on a very small data set, of less than 20% credibility. The results also are based on assignment of motorcycles to the sport bike category which are not sport bikes. The error in this regard caused the Public Utilities Board of Manitoba to order MPI to review which vehicles are assigned to sport bikes. Since 1999, 53.4 percent of all Motorcycle Accidents Involved Injuries, while Injuries Resulted from 12.3 percent of Passenger Vehicle Accidents This result is of limited use in assessing motorcycle risk since it does not account for who was responsible for the accident. The degree of fault is an important consideration and could result in significantly different results. Claims Frequency is Growing Faster for Motorcycles than Private Passenger Vehicles. MPI drew this conclusion based on four years data. Data for a longer time period results in claim frequency for motorcycles growing about the same as private passenger vehicles. As well MPI does not note that motorcycle claim frequency in most years is less than for private passenger vehicles. The Motorcycle Pool in Manitoba is Growing Faster than the Private Passenger Pool The actual increase in the number of motorcycles was about 2,000 during the time period of the study. Private passenger vehicles increased by about 32,000. The Number of Motorcycle Claims is Increasing Significantly Between 2000 and 2003 there was an increase of 145 in the number of motorcycle claims. Comparatively the increase in the number of private passenger vehicle claims during the same time period was 20 times greater, at 2,884. Motorcycles Have Higher Accident Rates during Night Time Driving Compared to Private Passenger Vehicles The comparison is based on accidents occurring after 10 P.M. It appears that the - 16 -
information for private passenger vehicles was based on annual data. Comparatively the motorcycle data is for the summer. It would be expected that accident rates at night are higher in the summer than winter (and naturally higher than average annual rates) simply due to the longer daylight hours and later night activities. With private passenger vehicle rates not determined on the same (summer) basis, the results are biased against motorcycles. - 17 -
Endnotes 1 Manitoba Public Insurance (2004). Motorcycle Risk Study. Winnipeg. 2 Public Utilities Board of Manitoba (2003). Order 173/03. Page 57. Winnipeg. 3 The standard deviation is the square root of the variance. 4 The coefficient of variation is the standard deviation divided by the mean times 100. By dividing by the mean the dispersion of outcomes is standardized. For private passenger vehicles, the coefficient of variation is 20.9 percent and for motorcycles it is 13 percent. 5 Z test: a statistical test of hypotheses in making inferences about the data, normally used where the sample size is large and the data is normally distributed. 6 T test: Similar to the z-test, the t-test is a statistic used for making inferences about data when the sample size is small. 7 The summary statistics on the estimated regression equation are provided in Appendix A. 8 Dean, Gary C. (1997), An Introduction to Credibility. Casualty Actuary Forum. Casualty Actuarial Society, Arlington, Virginia. 9 Miller, T. R. and Lawrence, B.A. (2003), Motorcycle Vehicle Insurance in the United States: A 1998-99 Snapshot with Emphasis on Motorcycle Coverage. Washington, D.C. National Highway Traffic Safety Administration. 10 Huang, B. and Preston, J. (2004) A Literature Review on Motorcycle Collisions: Final Report. Oxford University. Traffic Studies Unit. 11 Road Safety Committee, Parliament of Victoria (1993) Relationship Between Accident Risk and Engine Capacity. Melbourne. 12 Hurt, H.H. Jr., Ouellet, J.V. & Thom D.R. (1981). Motorcycle Accident Cause Factors and Identification of Countermeasures. Washington, DC: National Highway Traffic Safety Administration. 13 Bednar, F., Billheimer, J.W., McRea, K., Sabol, S.A., Syner,J. and Thom, D.R. (2000) Motorcycle Safety. Transportation Research Board. Transportation in the New Millennium Series. Washington, D.C. 14 Royal Society for the Prevention of Accidents (2001). Motorcycle Position Paper. Birmingham United Kingdom. 15 Hurt, op.cit., Page 12. 16 Bednar, op.cit., Page 4. 17 Royal Society for the Prevention of Accidents Motorcycle, op.cit., P.22. 18 Derived from Page 13 of the Motorcycle Risk Study. 19 Public Utilities Board of Manitoba (2004), MPI Response to information request GMMG/MPI 2-15. Winnipeg. 20 As shown on Page 4, more than one half of motorcycle accidents according to MPI are single vehicle, meaning they were the fault of the motorcyclist. If multiple vehicle accident where motorcyclists were at fault are added then motorcyclists would likely be responsible well over 60% of the time. This is the exact opposite of the findings noted. 21 Manitoba Public Insurance. News Release, October 8, 2004. http://www.mpi.mb.ca/english/newsroom/articles/2004/nr_oct8_04.html 22 Manitoba Public Insurance, op.cit. Page 2. 23 Ibid. Page 6. 24 Ibid. Page 7. 25 Ibid. Page 5 26 Ibid. Page 13 27 Hurt, op.cit., Page 12. 28 Road Safety Committee, Parliament of Victoria (1992). Section 3.1. Melbourne. 29 Refer to Utilities Board of Manitoba (2004), MPI Response to information requests CMMG 2-5 and GMMG/MPI 2-6. Winnipeg. 30 Public Utilities Board of Manitoba. Transcripts of Hearing: 2005 Manitoba Public Insurance Basic Autopac Rates. Pages 797 and 798. Winnipeg. - 18 -
31 Royal Society for the Prevention of Accidents, op.cit. Page 2 32 Langley, J. Mullin, B., Jackson, R. and Norton, Robyn. Motorcycle Engine Size and the Risk of Moderate to Fatal Injury from a Motorcycle Crash. Accident Analysis and Prevention. Volume 32 Pages 659 to 663. Elsevier Publishing. 33 Ibid., Page 659 34 The number of units is somewhat vague. In MPI Response to information request CMMG/MPI 2-9 MPI says there are 1,231 units on average for 2001 through 2003. In the Appendix of the Motorcycle Risk the implied number is 1,350. 35 Public Utilities Board of Manitoba. Transcripts of Hearing: 2005 Manitoba Public Insurance Basic Autopac Rates. Page 753. Winnipeg. 36 Manitoba Public Insurance, op.cit. Page 11. 37 Public Utilities Board of Manitoba (2004). Order 148/04. Page 92. Winnipeg.. 38 Manitoba Public Insurance. News Release, October 8, 2004. http://www.mpi.mb.ca/english/newsroom/articles/2004/nr_oct8_04.html. 39 Peek-Asa, C. and Kraus, J. (1996) Injuries Sustained by Motorcycle Riders in the Approaching Turn Crash Configuration. Accident Analysis and Prevention, Vol.28 No.5, Page 561. 40 It is unlikely that there would be a transfer of injury claims from private passenger vehicles to motorcyclists, simply since the weight differential between private passenger vehicles and motorcycles favours the private passenger vehicle. 41 Manitoba Public Insurance, op.cit. Page 3. 42 Refer to Public Utilities Board of Manitoba (2004), MPI Response to undertaking 15(a). Winnipeg. 43 The annual frequency data is provided in Public Utilities Board of Manitoba (2004), MPI Response to information request CMMG/MPI 2-18. 44 Langley, J. op.cit., Pages 659-663. 45 Ibid. Page 659. - 19 -
References Bednar, F., Billheimer, J.W., McRea, K., Sabol, S.A., Syner, J. and Thom, D.R. (2000) Motorcycle Safety. Transportation Research Board. Transportation in the New Millennium Series. Washington, D.C. Begg, D., Langley, J., and Redder, A., Motorcycle Crashes in New Zealand Resulting in Death and Hospitalisation I: Introduction to Methods and Overview. Accident Prevention and Analysis. Volume 26, Issue 2. Pages 157-164. Langley, J., Begg, D., and Redder, A., Motorcycle Crashes in Resulting in Death and Hospitalisation II: Traffic Crashes. Accident Prevention and Analysis. Volume 26, Issue 2. Pages 165-171. Cercarelli, l., Arnold, P., Rosman, D., Sleet, D., and Thornett, M., Travel Exposure and Choice Comparisons for Examining Conspicuity by Analysis of Crash Data. Accident Prevention and Analysis. Volume 24, Issue 4. Pages 363-368. Dean, Gary C. (1997), An Introduction to Credibility. Casualty Actuary Forum. Casualty Actuarial Society, Arlington, Virginia. Huang, B. and Preston, J. (2004) A Literature Review on Motorcycle Collisions: Final Report. Oxford University. Traffic Studies Unit. Hurt, H.H. Jr., Ouellet, J.V. & Thom D.R. (1981). Motorcycle Accident Cause Factors and Identification of Countermeasures. Washington, DC: National Highway Traffic Safety Administration. Langley, J., Mullin, B., Jackson, R. and Norton, Robyn. Motorcycle Engine Size and the Risk of Moderate to Fatal Injury from a Motorcycle Crash. Accident Analysis and Prevention. Volume 32, Issue 5. Pages 659 to 663. Elsevier Publishing. Langridge, D. (2004) Introduction to Research Method sand Data Analysis in Psychology. Pearson Education Limited, London. Lin, M., Chang, S., Lu, P., and Key, P., A Longitudinal Study of Risk Factors for Motorcycle Crashes among Junior College Students in Taiwan. Accident Analysis and Prevention. Volume 35. Issue 2. Pages 243-252. Manitoba Public Insurance Corporation (2004). 2005 Basic Rate Application. Winnipeg. Manitoba Public Insurance. News Release, October 8, 2004. Manitoba Public Insurance Corporation (2004). Motorcycle Risk Study. Winnipeg. Mason, R., Lind, D. and Marchal, W. (1999). Statistical Techniques in Business and Economics. Tenth Edition. Irwin McGraw Hill. Boston. Miller, T. R. and Lawrence, B.A. (2003), Motorcycle Vehicle Insurance in the United States: A 1998-99 Snapshot with Emphasis on Motorcycle Coverage. Washington, D.C. National Highway Traffic Safety Administration. Peek-Asa, C. and Kraus, J. (1996) Injuries Sustained by Motorcycle Riders in the Approaching Turn Crash Configuration. Accident Analysis and Prevention, Vol.28 No.5, Page 561. - 20 -