ISO Insurance Services Office of Israel, Ltd. Amot Bituach Building, Building B, 6th floor, Derech Menachem Begin 48, Tel-Aviv 66184, Israel Telephone (972) 3 687 6536, Fax (972) 3 687 6542, E-mail info@iso-israel.co.il, Internet www.iso-israel.co.il Database for the Compulsory Vehicle Insurance in Israel Proposed Tariffs for April 2007 18 January 2007 Version Page 1 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Table of Contents ADVICE AND RECOMMENDATIONS REGARDING THE APRIL 2007 TARIFF FOR COMPULSORY AUTOMOBILE INSURANCE ............................................................................. 4 SUMMARY OF RESULTS ...................................................................................................................... 4 DATA USED FOR THE OVERALL COUNTRYWIDE TARIFF LEVEL DETERMINATION............................ 4 Losses ............................................................................................................................................. 4 Exposures ....................................................................................................................................... 5 DERIVATION OF THE COUNTRYWIDE TARIFF LEVEL CHANGE .......................................................... 6 Determination of Ultimate Losses ................................................................................................. 6 Adjustment for Missing MADA Payments ..................................................................................... 7 Discounting .................................................................................................................................... 7 Adverse Deviation Provisions........................................................................................................ 8 Loss Trend...................................................................................................................................... 9 Calculation of Pure Risk Premiums ............................................................................................ 12 COUNTRYWIDE TARIFF LEVEL INDICATIONS ................................................................................... 14 Initial Indications......................................................................................................................... 14 Additional Considerations ........................................................................................................... 15 FINAL RECOMMENDATION FOR COUNTRYWIDE TARIFF LEVEL CHANGE ....................................... 18 Derivation of Changes by Vehicle Type ...................................................................................... 18 SUPPLEMENTAL MOTORCYCLE ANALYSIS ...................................................................................... 20 FINAL TARIFFS ................................................................................................................................. 21 ANALYSIS OF THE POOL EXPERIENCE FOR COMPULSORY VEHICLE INSURANCE IN ISRAEL ........... 21 ADVICE AND RECOMMENDATIONS REGARDING THE APRIL 2007 TARIFF FOR COMPULSORY RAILROADS INSURANCE ............................................................................... 24 DATA USED FOR THE RAILROADS ANALYSIS................................................................................... 24 ISRAELI RAILWAYS ANALYSIS ......................................................................................................... 24 Determination of Ultimate Incurred Losses ................................................................................ 24 Analysis of Paid Losses ............................................................................................................... 24 Comparison of Paid Losses Developed to Nine Years to the Current Estimate of Incurred Losses ........................................................................................................................................... 25 Determination of Ultimate Incurred Losses ................................................................................ 26 Determination of Ultimate Discounted Losses ............................................................................ 27 Determination of Tariff by Segment ............................................................................................ 28 TREND ANALYSIS ............................................................................................................................. 29 Pure Premium .............................................................................................................................. 29 Calculation of Loss Per Exposure ............................................................................................... 30 Final Recommended Pure Risk Calculation ................................................................................ 30 CARMELIT HAIFA ............................................................................................................................. 31 APPENDIX A – FINAL PAID MACK’S AGE-TO-AGE FACTORS .......................................... 33 APPENDIX B -- ANALYSIS OF MISSING MADA PAYMENTS .............................................. 35 APPENDIX C - EXPERIENCE PERIOD FOR TARIFF DETERMINATION AND DETERMINATION OF TREND FACTORS ................................................................................. 36 Page 2 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance APPENDIX D - DERIVATION OF AVERAGE CLASS FACTORS .......................................... 38 APPENDIX E -- CHANGE IN THE RETIREMENT AGE .......................................................... 40 APPENDIX F - INCREASED SUBROGATION EFFORTS OF THE INII ............................... 43 APPENDIX G -- CHANGES IN LAW AFFECTING THE SETTLEMENT OF CLAIMS ...... 45 APPENDIX H -- THE "LOST YEARS" ......................................................................................... 47 EXHIBITS Page 3 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Advice and Recommendations Regarding the April 2007 Tariff for Compulsory Automobile Insurance As part of ISO Israel’s role as the operator of the Data Bank for Compulsory Automobile Insurance, ISO Israel is required to provide advice and recommendations regarding the pure risk premium for April 2007. The following contains ISO Israel’s analysis and proposal regarding the appropriate pure risk premium level for April 2007. This proposal includes input received from the Israeli insurance companies and the Ministry of Finance in 2006. Please note that this document reflects the changes for the Pool tariffs effective January 2007. As a result of those very recent changes, ISO Israel is not proposing any further revisions to the Pool tariffs at this time. Summary of Results ISO Israel performed an analysis of the indicated changes to the current ISO Israel tariff levels with consideration of discounting, since the Ministry of Finance has requested that the calculation of investment income be provided for in the pure risk premiums (i.e., the tariffs). Additionally, this analysis includes a provision to account for the possibility of adverse loss deviations. Based on ISO Israel’s initial analysis, the countrywide tariff level indication of -17.1%, as displayed on Exhibit 9, is the appropriate indicated change for the April 2007 Israel tariffs prior to consideration of recent changes in the Israeli market such as changes in the retirement age, increased subrogation efforts by the INII, changes in the law affecting the settlement of claims, the lost years court rulings, and the recent changes in hospitalization costs. While some of these changes are already in effect, they are still not adequately reflected in the historical loss experience underlying this review. Thus, as explained further in this document, ISO Israel's recommendation is to implement a -6.0% countrywide tariff change for April 2007. Details on the data and analysis methods used for this review follow below. Data Used for the Overall Countrywide Tariff Level Determination Losses ISO Israel used the following two sources of loss data for this year's tariff analysis: Loss data accumulated in the ISO Israel Database for underwriting years 2002, 2003 and 2004. The end of 2002, end of 2003 and end of 2004 reports by Professor Yehuda Kahane of Actuarial and Financial Consulting dated February 5, 2003, March 22, 2004, and February 3, 2005, respectively. The ISO Israel paid loss data contains data for the totality of the compulsory insurance market. It includes all claim payments made through December 2004 by the insurance companies, including the Compulsory Automobile Insurance Pool; loss data from Karnit is not included even though they were Page 4 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance accumulated in the ISO Israel database, as the pure risk premium used in our analysis does not include the Karnit provision. However, ISO Israel started collecting compulsory insurance information for policies having inception dates of October 2001 and later. As a result, the ISO Israel Database contained only 3 complete underwriting years of data at the time of this analysis -underwriting years 2002, 2003 and 2004. Thus, ISO Israel relied on the Kahane report to provide historical compulsory insurance loss data for earlier underwriting years. Kahane's report is based on, and contains data for, the totality of the compulsory insurance market. The data from Professor Kahane’s report includes all claim payments made through December 2004 for underwriting years 1977 to 2002, by the insurance companies and Avner, including the Compulsory Automobile Insurance Pool; loss data from Karnit is not included. Exposures ISO Israel used the following two sources of exposure data for this year's tariff analysis: Exposure data accumulated in the ISO Israel Database for the years 2002, 2003 and 2004. Ministry of Transportation vehicle counts for 1998 through 2004. ISO Israel relied on the 2002, 2003 and 2004 exposure totals derived from the ISO Israel Database for this countrywide tariff level review. Both voluntary and Pool experience were used for this analysis. The ISO Israel Database underlying this analysis includes experience reported in detail through the ISO Israel statistical plan, as well as experience reported in aggregate by some companies for specific instances where statistical plan reporting could not be completed in time for this review. Since ISO Israel exposure data has the same short-term issue as the losses -- it is currently available only for the latest 3 years (2002, 2003 and 2004) -- ISO Israel relied on the MoT data to approximate insured exposure counts and pure risk premiums for the earlier years. That is, ISO Israel determined 1998 - 2001 pure risk premiums by adjusting the 2003 ISO Israel pure risk premiums by the overall change in vehicle counts over time, obtained from the MoT data. This process is explained further in the discussion of the derivation of the current pure risk premium level below. Before proceeding with its analysis, ISO Israel reconciled the 2002, 2003 and 2004 exposure totals from the ISO Israel Databank to the MoT vehicle totals. However, since the Ministry of Transportation database includes information on all vehicles registered in Israel, whether or not compulsory insurance is ultimately purchased, while the ISO Israel Databank contains exposure information only on actual compulsory insurance policies purchased, consideration must be given to the anticipated percentage of uninsured vehicles in Israel when performing this comparison. Based on prior Karnit pricing provisions, ISO Israel expects the uninsured vehicle percentage in Israel to be around 5%. Once the uninsured vehicle population is taken into consideration, the reconciliations revealed that the underwriting year 2002, 2003 and 2004 written exposure totals derived from the ISO Israel Database (determined based on the effective and expiration dates on the detailed premium records) reconcile fairly well to the MoT data for those years on an overall basis. That is, on an overall basis, the ISO Israel exposures for years 2002, 2003 and 2004 were lower than the MoT vehicle totals, by Page 5 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance an amount that could reasonably be explained by the uninsured vehicle population in Israel, with the exception of the special vehicles. In the case of special vehicles, it is known that the MoT database does not contain information for certain vehicles types (e.g., forklifts and mechanical engineering vehicles). Exhibit 1 provides a summary of the comparison of the ISO Israel exposures to the MoT vehicle counts for the 2002, 2003 and 2004 years. Derivation of the Countrywide Tariff Level Change Determination of Ultimate Losses ISO Israel performed an independent analysis of the ultimate incurred losses for the period 19852004 using five variants of the chain ladder method. For this analysis, ISO Israel used the paid losses contained in the Kahane report, except for 2002, 2003 and 2004 where data from the ISO Israel Database was used. All data was indexed to November 2004. At this stage in the analysis, losses are not discounted and do not contain a provision for adverse deviation. Historical losses for underwriting years prior to 1985 were eliminated from this analysis, as they exhibited a very different development pattern than the more recent years, particularly in the early evaluations. Thus, ISO Israel does not believe they are appropriate for use in estimating development factors for the more recent years. However, ISO Israel does recognize that limiting the number of underwriting years in the historical loss development triangle also limits the number of evaluations used in the analysis. As a result, ISO Israel has applied a tail factor of 1.003 to the final evaluation to account for the development expected for compulsory insurance beyond 20 years. This tail factor was selected based on a review of the earlier underwriting years of data available in the Kahane report, using the Mack method recommended for the tariff evaluation (below), and is shown on Exhibit 2-6. The five variants of the chain ladder method differ only with respect to the method of calculation of the overall age-to-age loss development factors from the individual years' link ratio factors. Age-toage loss development factors were calculated as follows: Method 1: Average of all years' link ratios Method 2: Average of all years' link ratios except the largest and smallest Method 3: Average of the latest five years' link ratios Method 4: Average of the latest three years' link ratios Method 5: According to the Mack method (see Appendix A) Exhibit 2 displays, for each of the above five methods, the derivation of the ultimate loss development factors and the resulting ultimate undiscounted losses by year. The table below compares the ultimate losses generated by each of the above 5 methods for the last 7 years (losses are in thousands of NIS): Year 1998 1999 2000 2001 2002 Page 6 of 53 Ultimate Undiscounted Losses, Excluding Any Provision for Adverse Deviation Method 1 Method 2 Method 3 Method 4 Method 5 2,800,550 2,786,987 2,812,344 2,824,921 2,797,444 2,752,897 2,739,565 2,776,977 2,804,446 2,749,843 2,669,021 2,656,094 2,692,366 2,721,357 2,663,747 2,699,564 2,688,653 2,696,866 2,736,986 2,685,553 2,597,206 2,584,863 2,526,136 2,554,326 2,570,826 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Ultimate Undiscounted Losses, Excluding Any Provision for Adverse Deviation Year Method 1 Method 2 Method 3 Method 4 Method 5 2003 2,833,616 2,817,063 2,712,330 2,753,275 2,784,877 2004 2,522,686 2,506,588 2,500,308 2,626,718 2,454,632 7 Year Average 2,696,506 2,682,830 2,673,904 2,717,433 2,672,417 The Mack Method provides a seven-year average estimate that is fairly consistent with the estimates derived from the other chain ladder methods. Additionally, the Mack Method has a distinct advantage over the other methods -- it provides a theoretical estimate of the standard error and coefficient of variation for the outstanding loss reserve estimates. For these reasons, ISO Israel has decided to continue using the Mack method for its tariff analysis. Adjustment for Missing MADA Payments During our discussions with the Israeli insurance companies in 2005, it was brought to our attention that the Israeli insurance companies are in dispute with MADA, such that there has been a withholding of payments by insurers to MADA starting with underwriting year 2003. It is ISO Israel’s understanding that this dispute had not yet been resolved at the time of this analysis, which means that the paid loss data reported to the ISO Israel Database for underwriting years 2003 and 2004 as of December 2004 does not include any payments to MADA. However, historically the majority of MADA payments were made by the 12 month evaluation for any given underwriting year. Furthermore, the historical loss development factors to be applied to the 2003 and 2004 underwriting year paid losses are based on historical data that includes MADA payments. As a result, an adjustment needs to be made to the paid losses for underwriting years 2003 and 2004 to account for the missing MADA payments. Appendix B contains a detailed analysis of the impact of the missing MADA data. As Appendix B explains, ISO Israel has determined that the reported underwriting year 2003 paid losses as of December 2004 need to be increased by about 1.7%, and the reported underwriting year 2004 paid losses as of December 2004 and the reported underwriting year 2003 paid losses as of December 2003 need to be increased by about 7.4% in order to account for the missing MADA data. This MADA adjustment is included in the paid loss totals displayed on Exhibits 2-1 through 2-5. Discounting Since the Ministry of Finance has requested that the tariffs reflect investment income, ISO Israel has performed this analysis on a discounted basis. The discount rate used for the tariff calculation should be the risk-free rate that prevails in Israel. As was done for the tariff analysis that became effective in October, 2005, ISO Israel considered the yield of ten-year bonds as a proxy for the risk-free rate. Additionally, this year, ISO Israel also considered the five year bond rates, as there has been evidence that the economic conditions in Israel are changing. Historical ten year and five year bond rates in Israel are as follows: Page 7 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Date 12/01 12/02 12/03 12/04 12/05 Average Indexed Ten-Year Bond Rate 4.3% 5.6% 4.1% 4.1% 3.3% (*) 4.3% Five-Year Bond Rate 4.9% 4.8% 4.9% 3.8% 3.0% 4.3% Data Source: Bank of Israel (*): For 2005, the ten year bond rate is as of October, 2005. As shown above, both the ten and five year bond rates have averaged about 4.3% over the past five years. However, there has been some variability by year, and even more importantly, both the ten and the five year bond rates show a considerable drop in the most recent year. As a result, ISO Israel has selected 2.5% as the discount rate for this year’s tariff analysis. For this analysis, losses are discounted from the midpoint of each payment year to the end of each underwriting year. This is based on the assumption that the average date of writings, and the average loss payment date, is July 1. Even though the average date of writings in Israel is not exactly the middle of the calendar year, but rather closer to the beginning of the year, July 1 was chosen for ease of application. This assumption has no effect on the final tariff indications, as the date selected as the average date of writings is the same date used as our assumed average loss payment date -- and in both cases we are discounting to the end of the underwriting year. So, if the true average date of writings is earlier in the year, for example June 1 as opposed to July 1, then the true average loss payment date would also be earlier in the year as well (June 1). Since both premiums and losses would be adjusted accordingly, this would result in no bottom line effect on the indication. Further details on the calculation of discounted losses are provided in Exhibit 3. Exhibits 4-1 and 4-2, display the incremental ultimate discounted losses and the cumulative ultimate discounted losses for this year's tariff analysis. Adverse Deviation Provisions To account for the possibility of adverse deviation of ultimate loss amounts, ISO Israel developed indications that incorporate a provision for adverse deviation. This was done by adjusting the estimated ultimate incurred losses for each underwriting year to include one standard error of the reserves. The standard errors were calculated using the Mack method. Further details on the calculation of the standard errors can be found in the article that is referenced in Appendix A. The resulting standard errors and coefficients of deviation are displayed in Exhibit 5. The application of one standard error of the reserves to the estimated losses, to account for the uncertainty surrounding ultimate loss development, has customarily been used in other actuarial studies in Israel, such as in the reports on the Compulsory Insurance Market as prepared by Professor Yehuda Kahane. It provides an extra level of confidence in the ultimate loss estimates, to account for the possibility that actual loss payments will be higher than expected based on past experience. Page 8 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Loss Trend Since actuarial analysis involves relying on the use of historical data to price tariffs that will be in effect in the future, the historical experience must be adjusted to reflect any anticipated economic and social changes expected in the future. Such anticipated changes can be accounted for by the application of loss trend factors to the historical loss data. Loss trend factors take into account anticipated changes in both claim frequency and claim cost levels in the future. ISO Israel evaluated four items with regard to loss trend: Compulsory Automobile Frequency Trend For this analysis, ISO Israel utilized vehicle counts from the MoT data and claim counts (casualties) from the Israel Central Bureau of Statistics (CBS). ISO Israel exposure data could not be used since only three years of exposure experience are available at this time. Even though the MoT counts include uninsured vehicles, ISO Israel believes it is appropriate to use these to evaluate frequency trends, since we are measuring changes in the average claim rate over time. As long as the same data source is used for all of the underwriting years, and there is no bias in the totals provided by that data source (i.e., the uninsured motorist percentage does not change drastically over that time period), consistency should be maintained. Additionally, while the Israeli insurers have expressed concern regarding the use of CBS data since it only represents a sample of all road accidents in Israel, ISO Israel continues to believe that it can be a useful source of information for measuring relative changes in claim frequency over time, provided that the sampling method used by the CBS is fairly consistent over time. That is, when measuring trends, claim totals need not be complete for all underwriting years; rather, their measurement only needs to be consistent over time. Exhibit 6-1 displays claim frequency annual rates of change derived using various exponential least squares fits. Compulsory Automobile Claim Cost Trend Since compulsory automobile tariffs in Israel are indexed to the general CPI, changes in claim costs need to be evaluated relative to the general CPI index. Accordingly, ISO Israel evaluated claim cost trends by examining health and wage CPI indices in excess of the general CPI index. Exhibit 6-2 displays the claim cost annual rates of change that are derived using an exponential least squares fit. It should be noted that the "real wage" indices were already net of general CPI changes, and thus they did not need to be adjusted for general CPI changes like the health indices. During our discussions with the Israeli insurers in November 2006, it was brought to our attention that the most recent CPI changes for 2006 show an increasing trend of the health index beyond the general CPI index. However, an article in Leumi Economic Weekly indicates that the most recent general CPI indices are affected by three volatile components in the short term -- fuel costs which were affected by the global oil price fluctuations, the holiday, recreation and foreign travel component, and housing prices which were affected by the strengthening of the shekel vis-a-vis the US dollar. Quoting Leumi Economic Weekly, "When analyzing the rise in prices excluding the volatile components and also discounting the impact of the exchange rate, it appears as though the Page 9 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance rate of core inflation is at the center of the price stability target range of 1-3%. Therefore, the recent adjustment in prices is not likely to impact the rate of inflation over the longer term." As a result, ISO Israel is not suggesting any adjustments for this short term CPI phenomenon in the market at this time. Compulsory Automobile Pure Premium Trend For this analysis, ISO Israel utilized losses from the Kahane report for underwriting years 1998 through 2001, losses from the ISO Israel Databank for underwriting years 2002, 2003 and 2004 and vehicle counts from the MoT data. Additionally, ISO Israel used ultimate, undiscounted losses based on the Mack method, without an adverse deviation provision. ISO Israel believes that the adverse deviation provision should not be included in the pure premiums used for the trend analysis, as the adverse deviation provision accounts only for the potential of upward deviations in the ultimate losses, which increases for the most current underwriting years; thus, including the adverse deviation provision in the pure premiums used for a trend analysis would introduce a positive bias into the trend fits. All trend losses are indexed to November 2004. Because the data for each underwriting year is indexed to the same cost level, any trends derived by the comparison of one year's losses to the next is exclusive of changes in the general CPI level. That is, the trends on Exhibit 6-3 are net insurance trends. Thus, they are appropriate for application to the fully indexed losses in this tariff analysis. Exhibit 6-3 displays the pure premium annual rates of change derived using exponential least squares fit. External Forecasts The last several years in Israel have seen economic activity impacted by domestic and global security situations. The impact on tourist visits has been most profound and this seems to have had a ripple effect on the economy in general. Unemployment rates in 2002, 2003 and 2004 have exceeded 10%. Events since January of 2005 indicate that the economic picture is improving with unemployment rates for 2005 less than 10%, and future forecasts indicating a continued decline in unemployment for 2007 and beyond. Given the nature of these changes in the economy and the expected relationship of economic activity to motor vehicle use and accidents, in 2005 ISO Israel and the Insurer Panel discussed and examined various issues and scenarios for determination of the pure risk premium portion of the Tariff. (Also impacting the determination of trend factors were potential adjustments in the base data for changes in data reporting for items examined separately elsewhere in this document, such as MADA payments, INII subrogation and changes impacting the settlement of claims.) For the purpose of this analysis, items which had uniform impacts by year for the years considered in determining pure risk premiums, such as the INII subrogation change, were not included in the trend analysis. In addition, any changes in settlements that would be prospective in nature do not impact the compiled data and do not need to be considered for trend purposes. Exhibit 6-4 presents an estimate of pure premium trend for Israel, using an econometric model and data from Global Insight, a leading provider of worldwide economic and financial data and forecasts. The model uses linear regression to model ultimate pure premium as a function of the unemployment Page 10 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance rate. A similar model is used by ISO Israel in the United States in its annual publication "Alternative Trend Forecasts for Commercial Automobile Liability Frequency and Pure Premium". The rationale for the model is as follows. It is reasonable to expect that the civilian unemployment rate is inversely related to claim frequency. When there are fewer people traveling to work, traffic density is reduced, lowering the likelihood of accidents. Also, the volume of business activity is low when the unemployment rate is high, which results in a decrease in the usage of commercial vehicles, and therefore, lower pure premium. As can be seen from the model output in Exhibit 6-4, the inverse relationship between pure premium and unemployment is borne out. For each underwriting year, the pure premium trend factors shown in Column (4) are equal to the ratio of the forecast pure premium for 2008 to that year's actual pure premium. The forecast pure premium for 2008 is used because the proposed effective date of the tariff is in 2007, and thus the average date of loss for policies written under the new rates will be in 2008. However, it should be noted that the very optimistic forecasts provided by Global Insight’s econometric model were determined prior to the recent conflict with Hezbollah, which could have a significant effect on future global security situations and tourist visits, potentially flattening or even reversing some of the anticipated economic improvements. The changes in the economic conditions throughout the experience period available for analysis raise the question of what is the appropriate base data for determining the pure risk premium and what is the appropriate period for selecting the trend factor. The results of various sensitivity tests varying the source data and the different possible periods for fits are contained in Appendix C. Final Trend Selection and Application When measuring and selecting final loss trends, changes in historical claim frequency and claim cost need to be considered, in addition to external changes in the environment. The following observations of Israel compulsory claim frequency trends, claim cost trends, pure premium trends, and external trends can be made: Claim frequencies are decreasing. Claim cost is slightly positive. Pure premium trends are negative. After a period of high unemployment, the economy appears to be rebounding with a decreased unemployment rate for 2005 and continued decreases forecasted for future years. But again, the very optimistic forecasts provided by Global Insight’s econometric model were determined prior to the recent conflict with Hezbollah, which could have a significant effect on future global security situations and tourist visits, potentially flattening or even reversing some of the anticipated economic improvements. Based on the above, ISO Israel has selected an annual trend factor of 0% for this year's tariff analysis. The 0% selection reflects the fact that while negative pure premium trends are observed for the historical data studied, the Israeli economy is rebounding; thus, we do not expect the recently Page 11 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance observed decreases to continue into the future. Additionally, as explained in Appendix C, the overall trend factor of 0% is applied to the 7 years of data used in the countrywide tariff level indication. For each underwriting year, the annual trend factor is applied from the average date of loss for that underwriting year to the average date of loss under the proposed tariff (i.e., the midpoint of the time period during which losses can occur on policies written during the time that the proposed rates will be in effect). For each underwriting year, the average date of loss is one year after the start of the year (for example, 1/1/2005 for underwriting year 2004.) This is because loss data is compiled on a policy year basis. Thus, 1 year policies with effective dates between 1/1/2004 and 12/31/2004 can produce accidents in the time period 1/1/2004 to 12/31/2005. For each underwriting year, the average date of loss under the proposed tariff is one year past the assumed effective date of the new tariff, which is April 1, 2007. Using one year past the assumed effective date as the average date of loss assumes that the new tariffs will be in effect for one year, that policies have one-year terms, and that policy writings are evenly distributed throughout the year. While ISO Israel recognizes that there are some exceptions to the one-year policy term (i.e., some policies in Israel are written for less than one year), a review of the data in the ISO Israel Database has confirmed that the vast majority of compulsory automobile policies in Israel are one-year policies. So, for underwriting year 2004, the loss trend period is from January 1, 2005 to April 1, 2008, or 3.25 years. As a result, the trend periods for this review by underwriting year are as follows: 1998 -- 9.25 1999 -- 8.25 2000 -- 7.25 2001 -- 6.25 2002 -- 5.25 2003 -- 4.25 2004 -- 3.25 The trend factor applied to each underwriting year is equal to the annual trend factor raised to the trend period for that year. So, for example, the trend factor for 2004 is calculated as: (1.00) ^ 3.25 = 1.00 Calculation of Pure Risk Premiums Calculation of 2002, 2003 and 2004 Pure Risk Premiums The 2002, 2003 and 2004 pure risk premiums were calculated by multiplying the 2002, 2003 and 2004 written exposures (i.e., total vehicles insured for one year) in the ISO Israel Database by the current ISO Israel tariffs in effect in Israel. As shown on Exhibit 7, separate calculations were performed for voluntary and Pool risks, since Pool tariffs are higher than voluntary tariffs. Voluntary tariffs are those shown in Exhibit 14 of the July 21, 2005 version of the Proposed Tariffs for October Page 12 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance 2005. Pool tariffs are the Pool tariffs effective January 1, 2007, as contained in MoF General Insurance Circular 2006-1-17, dated December 11, 2006. It should be noted that provisions were also made to accurately rate short-term policies. Specifically, policies in effect for less than 7 days were rated at 5% of the annual tariff. Policies in effect for more than 7 days but less than a full year were rated as [(.05) + (.003 * the number of days in effect - 7)] * the tariff. This short-term rating was accomplished by using vehicle counts that were adjusted for short-term policy rating on Exhibit 7. So, for example, if a policy was in effect for 6 days, the vehicle count adjusted for short-term rating would equal .05 vehicles. When this adjusted vehicle count is multiplied by the annual tariff, the correct short-term pure risk premium for this risk is applied. For standard 1 year policies, there is no adjustment applied. That is, the vehicle count after adjusting for short-term policy rating is equal to the initial vehicle count and thus there is no effect on the pure risk premium calculation. Since the current ISO Israel tariffs were at a May 2005 index level, these tariffs are re-indexed in Exhibit 7 to a November 2004 level, to be consistent with the indexing used for the losses. In addition, pure risk premiums for privately owned vehicles, corporately owned vehicles and privately owned motorcycles insured in the voluntary market and privately owned motorcycles insured in the Pool were multiplied by average class factors. Average class factors are applied to the pure risk premium to reflect the current ISO Israel Risk Classification Plan, which was implemented on October 1, 2005. The average class factors reflect the following rating variables: Privately Owned Automobiles (voluntary risks only) - age and gender, years licensed, number of accidents, number of major convictions, airbags and engine size (although currently all engine size groupings have class factors of 1.00 in the ISO Israel class plan). Privately Owned Motorcycles (voluntary and pool risks) - age and gender, years licensed, number of accidents and number of major convictions. Corporately Owned Automobiles (voluntary risks only) - the only rating variable is airbags. Average class factors were derived using written exposure data from the ISO Israel Databank for underwriting year 2004. Further details on the derivation of the average class factors can be found in Appendix D. For motorcycles, the tempered class factors, as implemented with the January 1, 2007 motorcycle Pool tariff changes, were used in this analysis. However, we have also included information in Appendix D on the average class factors that would result for motorcycles if the full classification factors, as originally recommended in our 2005 proposed risk classification plan filing, are used for motorcycle risks. The total pure risk premiums for 2002, 2003 and 2004 were then derived as the sum of the voluntary and Pool pure risk premiums, as shown on Exhibit 7-4, Exhibit 7-8 and Exhibit 7-12, respectively. Page 13 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Calculation of 1998-2001 Pure Risk Premiums As discussed previously, at the time of this analysis ISO Israel exposures were available only for the 2002, 2003 and 2004 underwriting years. To determine aggregate pure risk premiums for underwriting years 1998-2001, the 2003 pure risk premiums were adjusted using MoT vehicle counts. This adjustment is necessary because the total number of vehicles, which directly affects the total pure risk premium, has changed over the years. The aggregate pure risk premiums for 19982001 were determined by adjusting the 2003 aggregate pure risk premiums for the change in the MoT vehicle counts from each underwriting year to 2003, separately by vehicle type. For example, 1998 aggregate pure risk premiums were determined by vehicle type as follows: (2003 aggregate pure risk premiums) * (1998 MoT vehicle counts/ 2003 MoT vehicle counts) This adjustment was performed by vehicle type to reflect distributional changes in vehicle counts by vehicle type. Since the tariffs vary by vehicle type, changes in vehicle counts by vehicle type will have different effects on the resulting pure risk premiums. Exhibit 8 displays the resulting aggregate pure risk premiums for each year used in this analysis. It should be noted that since the MoT data does not separately identify dealership vehicle counts, the aggregate pure risk premiums for dealerships for years prior to 2002 were determined by using the MoT motorcycle vehicle counts for the motorcycle dealerships, and a combination of the MoT automobile and motorcycle vehicle counts for the auto dealerships. Countrywide Tariff Level Indications Initial Indications Exhibit 9 displays the indicated change to the current ISO Israel tariff levels with consideration of discounting, and with consideration for adverse deviation, for all seven years of experience (underwriting years 1998 - 2004). Since premium and loss data include Pool experience, the indicated change represents an average change that is appropriate for both the voluntary market and the Pool combined. Exhibit 9-A provides additional indications for informational purposes, obtained by varying the years of experience utilized in the review. So, for example, if we use only the last five underwriting years of data to determine the indicated tariff level changes, the indication would be -20.1% rather than -17.1%. Alternatively, eliminating underwriting years 2000 - 2002 (which the Israeli insurers thought were atypical based on our meetings in 2005) from this analysis changes the indication only slightly to -14.9%. Or, eliminating underwriting year 2004 (which the Israeli insurers thought was atypical based on our November 2006 teleconference) from this analysis also changes the indication only slightly to -15.7%. All in all, Exhibit 9-A illustrates that varying the years used for this calculation leads to results in a fairly narrow range, from -14.9% to -21.2%. For stability purposes, ISO Israel is recommending that the all years indication of -17.1% be used for this year's final tariff recommendation. Page 14 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Additional Considerations During our discussions with the Israeli insurers while preparing the 2005 Tariff review, the insurers raised concerns about several items that might impact the future cost of compulsory insurance. ISO Israel recognizes that none of these items have been explicitly reflected in the tariff indication on Exhibit 9, since the underlying data periods are prior to the time of these changes, and thus require separate consideration: Change in the Retirement Age -- Several insurers expressed concern about the impact that an increase in the retirement age would have on compulsory insurance. As a result, ISO Israel has analyzed the potential impact the increase in the retirement age would have on compulsory insurance. Appendix E contains ISO Israel's analysis of this issue. As a result of this analysis, ISO Israel estimates that the change in the retirement age will result in an increase in ultimate losses of about 1.5%, which is not reflected in the countrywide tariff level indication on Exhibit 9. Increased Subrogation Efforts of the INII -- ISO Israel is aware of the fact that the Israeli National Insurance Institute (INII) has recently increased subrogation efforts for claims against insurance companies due to compulsory auto accidents. This is a relatively new initiative, and as such the increased liability due to these efforts is not fully reflected in the losses for the underwriting years used for this countrywide tariff level review. A sample of Israeli insurance companies representing about 50% of the market provided ISO Israel with information on INII payments to analyze this issue. ISO Israel's analysis of the increased INII subrogation efforts is contained in Appendix F. As Appendix F indicates, ISO Israel estimates that the recent increased subrogation efforts by the INII represent an increase in ultimate losses of about 3%, which is not reflected in the countrywide tariff level indication on Exhibit 9. Changes in Law Affecting the Settlement of Claims -- During the course of our discussions with the Israeli insurance companies in 2005, several proposed or actual changes in the settlement of claims were raised. ISO Israel has consulted with claims expert(s) and made an independent evaluation of the effects of each as well as the appropriateness of adjusting the historical data at this time. o “Chok ha-Hesderim” -- At issue is the 2005 Economic Policy Act, which includes a change to the work invalidity threshold, so that only permanent invalidity levels of 9% and above will be entitled to a one time grant. Prior to the 2005 Economic Policy Act, a level of invalidity greater than 5% qualified for this benefit. (The one time benefit prior to this Act was paid on levels of invalidity between 5% and 20%.) ISO Israel believes that the change will have a small effect on compulsory insurance tariffs. Making the assumption that levels of disability as reported by the INII in the general population are a fair approximation of the expected levels of disability in auto accidents, ISO Israel found that the group with disability levels between 0 to 20% (which really translates into 5 to 20% as currently an invalidity level of 5% or more qualifies for this benefit), comprise 4% of all pension recipients. The number of individual recipients at the end of 2003 was 970. Not all of these individuals were injured in auto accidents. 50% were auto accident injuries would mean that there are 485 individuals with a disability level between 5% and 20%. Changing the eligibility level to 9% and assuming Page 15 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance a uniform distribution would mean that there would have been approximately 130 cases that would not be compensated from INII but would now need to be compensated by the compulsory auto coverage (i.e., changing the threshold from 5% to 9% is a 4 point increase in a range from 5 to 20,or approximately 27% of the claimants if we assume a uniform distribution). These benefits are paid by the formula as follows: Invalidity percentage * ¾ of monthly income prior to accident * number of months of invalidity The resulting value would then have a maximum of 130*.07*(.75*7,000)*43 months = 2,054,325. The cost to the Paltad of this change in 2003 would have been no more than 2,054,325 NIS. This is equal to less than 0.1% of the previously estimated ultimate incurred losses. Changes in the Road Accident Law -- There currently exists a proposal to increase the limit of pain and suffering from 150,000 NIS to 280,000 NIS. This change if enacted would impact the cost of compulsory insurance. However, the specifics of this provision are in a private law proposal and the Knesset committee has considered the idea of a change without making any specific recommendation. Since the precise details of any change in this regard are not known at this time, ISO Israel would not recommend making any adjustment to the indicated tariff at this time, but would recommend that when such a proposal is final that the tariff be adjusted to reflect any increase in liability. With regard to the potential changes in the definition of a road accident, to date there are a number of proposals that have yet to be enacted and should be evaluated at the time that they are final. Page 16 of 53 National Insurance -- A change has been made to increase the employer’s share of the injury compensation from the first 9 days to the first 12 days. We believe that this impact will be small as payment for lost income for all auto accidents represents about 20% of all losses. In order to evaluate the impact of this change on compulsory insurance ISO Israel examined wage loss information by length of disability in days from US auto insurance studies conducted by the Insurance Research Council. In doing this analysis ISO Israel calculated the insurer share under a treatment consistent with the current Israel system where the employer would pay the disability for the first 9 days and under the future 12 day level. The calculation indicates that this change would produce a 20% increase in the amount of losses paid for disability compensation for work injury auto accidents. Since the entire wage compensation portion of the losses is 20% reflecting both work and non work auto accidents, this change could have no more than a 4% effect if all wage compensation were for work injury auto accidents. While Avner has indicated that historically about 10% of all road accidents are work accidents, given the lack of detailed data available for analysis on this issue, ISO Israel has used a more conservative assumption that 25% to 50% are work accidents. As a result, ISO Israel estimates that this change will increase losses no more than 1% to 2%, and thus has selected an effect of 1.5%. The details of the 20% increase are shown in Appendix G. Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance The Lost Years -- Several insurers expressed concern about the impact that recent "lost years" rulings could have on compulsory insurance in Israel. In the "lost years" rulings, the courts have awarded income benefits to claimants that were not previously entitled to income benefits. The award for "lost years" represents the lost income that would have been earned by the injured had they not died, or experienced curtailment of life as a result of a road accident. The benefit is payable to "untraditional" beneficiaries, such as the parents or the estate of the deceased. These awards represent a broadening of compulsory auto insurance in Israel, following a similar broadening of tort rulings. As a result, ISO Israel performed an analysis to estimate the impact of the lost years rulings. Appendix H provides ISO Israel's analysis of the effect of the lost years rulings on Compulsory Insurance in Israel. As Appendix H indicates, ISO Israel estimates that the recent "lost years" cases represent an increase in ultimate losses of about 5.2%, which is not reflected in the countrywide tariff level indication on Exhibit 9. Hospitalization - On September 4, 2006 the Ministry of Health published an update of its hospital services costs, effective as of September 1, 2006. Since hospitalization costs are part of compulsory automobile insurance, these changes have a direct effect on compulsory automobile tariffs. According to a simulation performed by the Ministry of Health using Government owned hospital data for all services given by the hospitals to injureds in road accidents (i.e., ER, hospitalization, differential services, institutes and clinics), the estimated impact of the hospital services changes to total hospitalization costs is an 8.25% increase on an annual basis. Additionally, based on loss data reported via the ISO Israel statistical plan for compulsory automobile insurance, total hospitalization costs represent approximately 10% of total compulsory automobile losses. As a result, ISO Israel estimates that the changes in the hospitalization services tariffs represent an increase in ultimate compulsory insurance losses of about 1%, which is not reflected in the countrywide tariff level indication on Exhibit 9. Fraud - In discussions with insurers in 2005, concern was expressed about the lack of an operational fraud detection mechanism. Early in 2006, , ISO ClaimSearch Israel initiated its fraud detection operation. Utilizing data going back to 1990, ISO ClaimSearch Israel will provide a useful tool for companies in managing their fraud exposures. As a result, no adjustment was made to the historical data to reflect possible changes in fraud activity. Additionally, the following should be kept in mind: The estimation of future cash flows based on past payments is statistically uncertain. Additional uncertainty is due to possible future social and economic developments that might affect the payment of claims under the Compulsory Automobile Insurance system. Actuarial reserve estimates are therefore reasonable estimates resulting from the application of an actuarial model. There is no certainty about future claim payments, and they may differ from the reserve estimates. For this reason, ISO Israel has provided a tariff level indication for April 2007 that includes consideration for the possibility of adverse loss deviations. Additionally, since the Ministry of Finance has requested that investment income be reflected in the final tariffs, these indications also reflect discounting. The data used in the calculation of the countrywide tariff level indication included both voluntary market and Pool experience. Because of this, the indicated change is an overall change appropriate for both the voluntary and Pool tariffs combined. Page 17 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance The analysis and recommendations of ISO Israel contained in this document are based on an analysis of historical and projected loss experience for the entire Compulsory Auto Insurance market. To the extent that insurers will use these pure risk premiums in determining tariff levels they will need to evaluate the necessary expense loadings for sales, service and administration of the policy. They will also need to evaluate the cost of reinsurance, particularly noting the type of reinsurance program they employ and what the parameters of the program’s pricing are relative to the cost levels anticipated in this analysis. Furthermore, each insurer must take into account the fact that the experience for an individual insurer will differ from the market as a whole, and that generally, one may expect the standard deviation for individual company experience to be higher than the standard deviation for the experience of the entire market combined. ISO Israel has not provided any analysis or guidance on these matters as these are beyond the scope of the contracted work. These factors are items that each company needs to focus on as they determine the appropriate loadings and deviations to file to convert the recommended pure risk base tariffs into their final filed tariff levels. Final Recommendation for Countrywide Tariff Level Change Based on the considerations noted above, ISO Israel believes that the -17.1% countrywide tariff level indication, as displayed on Exhibit 9, is the appropriate initial indicated change for the April 2007 Israel tariffs, prior to consideration of recent changes in the Israeli market such as the change in the retirement age, the increased subrogation efforts by the INII, the changes in law affecting the settlement of claims, the impact of the "Lost Years" and the recent changes in hospitalization costs. However, ISO Israel is not suggesting that this full indication be implemented at this time. Rather, ISO Israel's recommendation is to implement a -6.0% countrywide tariff change for April 2007. This selection includes consideration of the following items: 1) the estimated 1.5% effect of the change in the retirement age; 2) the estimated 3.0% effect of the recent INII claims action; 3) the estimated 1.5% effect of the changes in law affecting the settlement of claims; 4) the estimated 5.2% effect of the "lost years" cases; and 5) the estimated 1% effect of the hospitalization services cost changes. As a result, the new indication, after consideration of the additional items noted above, is as follows: (1 - 0.171 ) * (1 + 0.015) * (1+ 0.03) * (1+ 0.015) * (1+ 0.052) * (1 + 0.01) = 0.935, or -6.5%. However, ISO Israel is recommending that this change be rounded up to the nearest whole percentage, for a final recommended change of -6.0%. ISO Israel believes that the adverse deviation provision included in the review is sufficient to cover the possibility of adverse loss development beyond what is already accounted for in the tariff calculation. Derivation of Changes by Vehicle Type Once the overall countrywide tariff level indication is determined, it is necessary to distribute this change by vehicle type. To do so, experience by vehicle type is evaluated to determine whether the Page 18 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance observed experience for each vehicle type is in line with the current ISO Israel tariff differential for that vehicle type. The main source of data for the vehicle relativity analysis is the ISO Israel Database. The Kahane report cannot be used, since experience is not available in vehicle type detail. In addition, while historical Avner data was available for analysis, the data from the Avner files has limited use, because Avner's exposures appear to be understated. Since the understatements are not consistent by vehicle type, using the Avner data would bias the resulting vehicle type relativities. Moreover, vehicle type detail is missing on nearly one quarter of the Avner loss records. As a result, using Avner loss data to determine vehicle relativities could bias the results if the loss records with missing vehicle type information do not follow the same distribution as the total loss experience by vehicle type (i.e., if the loss records with missing vehicle detail under or over represent any particular vehicle type). However, the Avner loss data was utilized to analyze development patterns by vehicle type. As Exhibit 10 illustrates, differences in development patterns by vehicle type do exist. (See Exhibit 11 for the derivation of the loss development factors by vehicle type. These factors were derived using the Mack method, as explained in Appendix A.) Exhibit 10 displays the calculation of indicated changes for each vehicle type, exclusive of the overall countrywide tariff level change. For example, Exhibit 10 indicates that the current ISO Israel tariff levels for motorcycles as a group are too low, prior to consideration of the overall tariff level changes indicated by this review. It should be noted that the ISO Premium at Present Rates used in this analysis reflects the Pool tariffs that went into effect January 1, 2007; thus, the indicated changes to the motorcycles shown on Exhibit 10 are in addition to the changes that were already implemented by the Ministry of Finance in General Insurance Circular 2006-1-17, dated December 11, 2006. The paid loss data in Exhibit 10 includes all bi-section payments and recoveries as reported through the ISO Israel statistical plan. However, an additional adjustment was applied to the data by vehicle type for motorcycle bi-section. This adjustment was necessary as there is a delay in the reflection of motorcycle bi-section recoveries. As a result, the statistical data reported to ISO Israel in 2002, 2003 and 2004 does not reflect all motorcycle bi-section recovery transactions. So, while the paid losses were reported, the bi-section recoveries and payments across vehicle types are not yet fully reflected. ISO Israel performed an analysis of 2002, 2003 and 2004 underwriting year data reported through fourth quarter 2004 to estimate the full effect of the motorcycle bi-section agreement. Based on this analysis, ISO Israel estimates that the full impact of the motorcycle bi-section agreement will be a savings of 15% to the motorcycle risks. For underwriting year 2002, the data reported to ISO Israel as of fourth quarter 2004 indicated a 15% savings to motorcycle risks. Therefore, no adjustment was applied to the underwriting year 2002 losses. For underwriting year 2003, the data reported to ISO Israel as of fourth quarter 2004 reflected a 12% savings to motorcycle risks. Therefore, the 2003 underwriting year motorcycle losses were adjusted by an additional 3%. For underwriting year 2004, the data reported to ISO Israel as of fourth quarter 2004 reflected a 5% savings to motorcycle risks. Therefore, the 2004 underwriting year motorcycle losses were adjusted by an additional 10%. The savings (i.e., loss reductions) applied to the motorcycle vehicle class on Exhibit 10 were then distributed (as loss increases) proportionally to all other vehicle types, based on the distribution of losses by vehicle type. It should be noted that this adjustment has no net effect on the overall loss levels -- rather, it is a distributional adjustment across vehicle types. While Exhibit 10 indicates changes could be made for every vehicle type that is subject to compulsory insurance, it should be noted that there is quite a bit of variability in the results from year to year for some vehicle types. For example, for the buses, the three years of data reviewed are not Page 19 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance consistent. The 2002 data indicates that the current ISO Israel bus tariffs are significantly too high, whereas the 2004 year indicates that the current bus tariffs as a group are too low. Additionally, the indicated changes for dealerships are very drastic, which is most likely attributable to the lack of credibility for these vehicle types. These examples show that reliance on only one or two years of data in a fine level of detail is not advisable, as it can lead to a situation where tariffs swing back and forth from year to year. Furthermore, as already explained, due to the very recent changes to the Pool tariffs, and the fact that the majority of the motorcycle risks are insured by the Pool, ISO Israel is not recommending additional changes to the motorcycles as a whole at this time. As a result, ISO Israel is proposing changes only to the taxis, commercial vehicles and special vehicles at this time. As Exhibit 10 illustrates, ISO Israel is proposing a 15.0% increase to the taxi tariffs, a 10% decrease to the commercial vehicles tariffs, and a 15% decrease to the special vehicles tariffs. These vehicle type changes are in addition to the overall countrywide tariff level change of 6.0% that is being filed. In order to introduce these changes by vehicle type on an overall revenue neutral basis (i.e., to ensure that these changes by vehicle type do not affect the overall selected countrywide tariff level change of -6.0%), an offset factor must be applied to all vehicle types. As Exhibit 10 indicates, this offset factor is 1.02. If this offset is not applied, then overall premium levels would be reduced an additional 2% by applying the recommended changes by vehicle type. Supplemental Motorcycle Analysis At the request of the Ministry of Finance, ISO Israel performed an additional analysis of the motorcycle pure risk tariffs by engine size. This analysis is based on motorcycle tariffs in effect PRIOR to the January 1, 2007 Pool tariff changes, contained in General Insurance Circular 2006-1-17, dated December 11, 2006. Exhibit I contains the results of this motorcycle engine class analysis. That is, Exhibit I shows the actuarially indicated pure risk tariffs for various motorcycle engine size groups; the displayed changes do not take into consideration the overall countrywide tariff level change of -6.0%. The indicated changes displayed on Exhibit I are changes to the motorcycle tariffs in effect prior to the January 1, 2007 Pool changes. Since those tariffs already had some distinctions by engine size, ISO Israel has included the current and indicated motorcycle Pool tariffs (not reflecting the overall tariff level change) for privately owned "all other" motorcycles in the last two columns to illustrate the results of this analysis on the largest class of motorcycle risks (i.e., privately owned motorcycles insured in the Pool). Any relevant surcharges (e.g., for more than 1 driver, for driving lessons, etc) or discounts (i.e., for antiques) would have to be applied to the indicated tariffs displayed on Exhibit I. As you can see from Exhibit I, the motorcycles with engine sizes 251 – 500 ccs are the most under priced, followed by motorcycles with engine size 0 – 50 ccs, motorcycles with engine sizes greater than 500 ccs, motorcycles with engine sizes 51 - 125 ccs and motorcycles with engine sizes 126 - 250 ccs. In order to maintain consistency with the recently implemented motorcycle Pool tariff changes on January 1, 2007, ISO Israel is recommending the following changes for the voluntary motorcycle tariffs as well: A 25% increase for motorcycles with engine size <= 50 ccs A 10% increase for motorcycles with engine size in the range 251-500 ccs Page 20 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance A 10% increase for motorcycles with engine size > 500 ccs A 30% increase for corporately owned motorcycles (in addition to the changes by engine size) An increase in the surcharge for more than 1 driver from 20% to 25% Final Tariffs Exhibit 12 displays the derivation of the final proposed pure risk tariffs by vehicle type and subclass, at a November 2004 index level. As Exhibit 12 illustrates, final tariffs for all vehicle types except the motorcycles and trains are determined by applying the countrywide tariff level change of -6.0%, the selected vehicle type change from Exhibit 10 and the vehicle type offset factor of 1.02 to the current tariffs. The only changes being filed to the motorcycle tariffs are the changes described above for the voluntary motorcycle tariffs, to ensure consistency with the January 1, 2007 implemented changes for the Pool motorcycle tariffs. The train tariffs are derived separately in this document. Row (12) on Exhibit 12 shows that the final achieved countrywide tariff level indication resulting from the final tariffs (excluding trains) is -5.6%. The reason this is not exactly equal to the proposed countrywide tariff change of -6.0% is because of the fact that neither the voluntary motorcycle tariffs nor the Pool tariffs are being reduced at this time. In fact, there are increases being implemented for some of the voluntary motorcycle classes. Final proposed base tariffs, at an October 2006 index, are displayed on Exhibit 13. Please note that the motorcycle tariff classes displayed have been refined by splitting the 51-250cc engine size class into two classes, 51-125cc and 126-250cc. This was done to match the structure of the latest Pool tariffs effective January 1, 2007, as contained in MoF General Insurance Circular 2006-1-17, dated December 11, 2006. At this time, the tariffs for the two classes are equal but the need for different tariffs will be reviewed in the future. Analysis of the Pool Experience for Compulsory Vehicle Insurance in Israel It is important to understand that this tariff analysis is based on Pool and Voluntary data combined. Therefore, the tariff analysis, and the final ISO Israel recommended change of -6.0% automatically, and implicitly, funds the Pool deficit. That is, the overall selected change of -6.0% is the change needed to ensure that future tariffs for all voluntary and Pool risks in Israel combined will cover the losses expected to arise from all voluntary and pool risks combined. At the request of the Ministry of Finance, ISO Israel performed a separate analysis of the Pool experience for compulsory vehicle insurance. This analysis was based on data reported via the ISO Israel statistical plan through December 2004. Exhibit 14-1 displays the Pool data by vehicle type and ownership, separately for underwriting years 2002, 2003 and 2004. This exhibit provides the indicated prospective deficit of the Pool, in NIS, using computations similar to those used in the tariff analysis to determine the indicated countrywide tariff level change with discounting and adverse deviation. Specifically: Page 21 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Premiums at present rates for the pool risks are as calculated in Exhibits 7-3, 7-7 and 7-11, and discounted by multiplying these amounts by the square root of 1.025 (as done in Exhibit 9). These premiums at present rates reflect the January 1, 2007 changes to the Pool tariffs, as contained in General Insurance Circular 2006-1-17, dated December 11, 2006. Paid losses for the pool risks were adjusted similar to the loss adjustments applied in the tariff analysis. That is, the 2003 paid losses (as of December 2004) were multiplied by 1.017 and the 2004 paid losses (as of December 2004) were multiplied by 1.074 to account for the missing MADA payments, the paid losses were developed to an ultimate settlement basis using the Mack method, the losses were discounted as explained in the tariff document to account for investment income, and the losses were adjusted to include a provision for adverse deviation. Additionally, paid losses include bi-section payments and recoveries as reported through the ISO Israel statistical plan; however, motorcycle bi-section adjustments were applied to the paid losses by underwriting year to account for the delayed reflection of motorcycle bi-section recoveries. Losses were not trended in this analysis since our final selected trend factor for this tariff analysis was 0.0%, and thus has no impact on the results. An indicated prospective Pool deficit/surplus is calculated by vehicle type and ownership in Column (7) by subtracting the discounted premiums at present rates from the ultimate incurred discounted losses with adverse deviation. However, since this is based on historical data, this does not reflect the items raised in our tariff analysis that are expected to have an impact on the future cost of compulsory insurance -- namely, the recent change in the retirement age, the increased subrogation efforts of the INII, the changes in law affecting the settlement of claims, the effect of the lost years, and the recent changes in hospitalization costs. As a result, Column (8) displays the indicated prospective Pool deficit/surcharge by vehicle type and ownership after consideration of the estimated effects of these additional factors: the estimated effect of the change in the retirement age * the estimated effect of the recent INII claims action * the estimated effect of the changes in the law affecting the settlement of claims * the estimated effect of the lost years * the estimated effect of the changes in Hospitalization services fees OR (1 + 0.015) * (1 + 0.03) * (1 + 0.015) * (1 + 0.052) * (1 + 0.01) = 1.127 As Exhibit 14-1 illustrates, the overall indicated average annual prospective Pool deficit is about 132,815,290 NIS, based on an average of underwriting years 2002, 2003 and 2004 data. Exhibit 14-2 displays the corresponding experience for the Voluntary risks by vehicle type and ownership for information purposes, mostly to illustrate the effects of the motorcycle bi-section adjustments (i.e., to show that the motorcycle bi-section adjustment has an overall net zero effect on the losses). As can be seen on Exhibit 14-1, ISO Israel estimates that the indicated prospective Pool deficit will result in approximately a 4.4% Pool subsidy on the voluntary tariffs. This estimate was determined by dividing the average Pool deficit by the average current premium at present rates for Voluntary risks, or Page 22 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance 132,815,290 / 3,012,452,107 which = 4.4% Page 23 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Advice and Recommendations Regarding the April 2007 Tariff for Compulsory Railroads Insurance During 2006, ISO Israel continued to study and evaluate the methods used to develop the tariff for passenger and cargo trains in Israel. Data Used for the Railroads Analysis The loss data used for the Railroads analysis is for Accident Years 1997 through 2004. This is a change from prior Railroad Tariff determinations. The Railroads data is being reviewed on an Accident Year basis for this year's analysis due to the extension of the policy for Underwriting Year 2003 to 3/31/2004. This extension resulted in a 2003 Underwriting Year of fifteen months which is inconsistent in length with the prior Underwriting Years. The Accident Year data is being evaluated as of 12/31/2005, which is the most recent evaluation of the data available. The 2005 data as of 12 months is not used because an analysis of previous Accident Years indicated that the 12 month evaluation for this segment was quite immature and subject to large variation in the movement from 12 month to subsequent evaluations. Because of this variation the latest Accident Year used is evaluated as of 24 months to minimize the possibility of the impact on the tariff of an immature year being highly levered by the applied loss development factor and impacting the projected pure loss premium. Israeli Railways Analysis Determination of Ultimate Incurred Losses Prior ISO analyses of the Tariff for Compulsory Railroad Insurance have relied on the use of underwriting year paid loss data developed to an ultimate settlement basis. Because of circumstances mentioned above, the data in this analysis has been compiled on an Accident Year basis. In addition, data supplied to ISO shows that the current estimate of the Accident Year incurred losses for the eight years included in this analysis, evaluated as of 12/31/2005 (prior to any consideration of loss development) are greater than the Accident Year estimates of ultimate based on the paid data developed through all the available data. Because of this, ISO Israel has based our analysis of the Tariff for Railroads on the incurred losses developed to an assumed ultimate basis of nine years. Analysis of Paid Losses We have constructed a paid loss development triangle to estimate the development of paid losses to an ultimate settlement basis. In determining the average and cumulative paid loss development factors we utilized Method 5, the "Mack Method" which was selected for use in the compulsory insurance analysis in the earlier chapter. The data in Railroads Exhibit 1 are for undiscounted paid losses, evaluated as of 12/31/2005, indexed to November 2004, in NIS. While the data for the railroads is available to construct a paid loss triangle, it is available for only nine evaluations of the eight years of experience included in this analysis. Page 24 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Based on our understanding of this coverage, our a priori expectations relating to loss development patterns for railroads, given sufficient claim volume, are: 1) It is generally expected that substantial loss development should occur during the first 12 months after the initial evaluation. 2) After the first 12 months, it is expected that there should be less development in each subsequent 12 month period. 3) The underlying loss development curve is expected to be concave up until it converges to unity. In developing the April 2007 tariff for passenger and cargo trains, we have relied upon the reported railroad data to the extent that it was deemed credible based upon reasonable a priori expectations. However, when examining the average paid '6-7' year-to-year link ratio (1.228) calculated from the available data, we noted that this link ratio was higher than the average paid '3-4', '4-5' and '5-6' yearto-year link ratios in the railroad data. As mentioned above, we generally expect to see less development in the latter evaluations of a loss development triangle than is observed during the earlier evaluations. When examining the individual paid year-to-year link ratios used in the development of the average paid '6-7' year-to-year link ratio, we note that the high average paid '6-7' year-to-year link ratio is being driven by the ratio for the 1997 Accident Year (1.393). We believe this 1997 paid year-to-year link ratio is resulting in the overstating of the true underlying average paid '6-7' year-to-year link ratio due to the limited volume of railroad data available (the average paid '6-7' year-to-year link ratio is based upon only two years of experience with a limited volume of railroad claims contained within each year of experience). ISO Israel has judgmentally selected 1.080, which is in line with our a priori expectations, to be used as the average paid '6-7' year-to-year link ratio for this year's railroad tariff analysis. Similarly, when examining the average paid '7-8' yearto-year link ratio (1.006) calculated from the available data, we noted that this ratio appeared to be understated due to the limited volume of railroad data available (the average paid '7-8' year-to-year link ratio is based upon a single year of experience with a limited volume of railroad claims). While the average paid '7-8' year-to-year link ratio meets our a priori expectations discussed earlier, we believe that loss payments will continue beyond nine years. Therefore, ISO Israel has judgmentally selected 1.030 to be used as the average paid '7-8' year-to-year link ratio for this year's tariff analysis. Comparison of Paid Losses Developed to Nine Years to the Current Estimate of Incurred Losses Because we know that losses used for the automobiles Tariff determination continued to develop beyond nine years, we felt it was necessary to see if the ultimate loss determined by the paid loss development is an appropriate estimate of the ultimate losses. In addition to the paid losses by Accident Year, ISO Israel had received the undiscounted incurred losses for each Accident Year. All paid losses and case reserves are evaluated as of 12/31/2005, indexed to November 2004, in NIS. We compared the latest evaluation of the undiscounted incurred losses for each Accident Year to the estimate of undiscounted paid losses developed to nine years, from Railroads Exhibit 1, for each Accident Year. This comparison is provided below: Page 25 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Accident Year Undiscounted Paid Losses Developed To Nine Years Undeveloped Undiscounted Incurred Losses Difference 1997 1998 1999 2000 2001 2002 2003 2004 3,256,756 1,943,701 4,815,541 8,889,859 3,838,213 4,091,912 3,700,640 5,815,034 4,275,545 2,471,939 7,281,841 11,127,553 4,187,239 4,389,643 6,444,200 8,692,185 -1,018,789 - 528,238 -2,466,300 -2,237,694 - 349,026 - 297,731 -2,743,560 -2,877,151 When examining the data presented above, we noted that the undiscounted paid losses developed to nine years are lower than the current estimates of incurred losses for every year included in the railroads analysis. Determination of Ultimate Incurred Losses Since the incurred losses are evaluated as of 12/31/2005 for each Accident Year, we have also reviewed the loss development on an incurred basis. While we do not have the data in adequate detail to create a complete incurred loss development triangle, we were able to review the most recent four evaluations of data for the incurred triangle. We utilized a method that averaged all year's year-toyear link ratios equally in order to determine the average incurred year-to-year link ratios. The data in Railroads Exhibit 2 is for undiscounted incurred losses, evaluated as of 12/31/2005, indexed to November 2004, in NIS. When reviewing the available incurred loss development link ratios we noted that the average incurred year-to-year link ratios do not completely meet our a priori expectations. Specifically, we noted that the average incurred '1-2', '3-4' and possibly the '5-6' and '6-7' year-to-year link ratios may be understated based upon our a priori expectations of link ratio behavior. When examining the individual incurred year-to-year link ratios for the '1-2' evaluation, we noted that even though there is a limited volume of railroad data available (the average incurred year-to-year link ratio is based upon only three individual incurred year-to-year link ratios, in addition, as mentioned earlier there is a limited volume of railroad claims contained within each Accident Year), the year-to-year link ratios for each individual year of experience are consistent (1.158, 1.127 and 1.152). Due to the consistency in the incurred year-to-year link ratios for the '1-2' evaluation, ISO Israel decided to use the indicated average incurred year-to-year link ratio for the '1-2' evaluation for this year's tariff development. When examining the individual incurred year-to-year link ratios for the '3-4' evaluation, we noted that there was a great deal of variability among the three available ratios (they range from 1.108 to 0.881). Since one of the three available ratios was 1.010, for the 1999 Accident Year, ISO Israel decided to use the indicated average incurred year-to-year link ratio (1.000) for the '3-4' evaluation since we did not have enough information available to support a different selection. Similarly, when examining the individual incurred year-to-year link ratios for the '5-6' evaluation, we noted that there was again a good deal of variability among the three available ratios (they range from 1.084 to 0.920). Since one of the three available ratios was 1.005, for the 1999 Accident Year, ISO Israel decided to use the indicated average incurred year-to-year link ratio (1.003) for the '5-6' evaluation since we did not have enough information available to support a different selection. Finally, since there were only two available incurred year-to-year link ratios for the '6-7' evaluation, ISO Israel decided to use the Page 26 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance indicated average incurred year-to-year link ratio (0.968) for the '6-7' evaluation since we did not have enough information available to support a different selection. When reviewing the cumulative incurred year-to-year link ratios we noted that there does not appear to be significant incurred loss development after seven years. Based on the nature of the coverage provided and the type of claims involved we believed that there could be further development of the incurred losses after nine years, however, due to the low claim volume and the limited years of experience available for railroads incurred loss data, we are unable to verify any a priori assumptions as to whether the continued development would be above or below unity, based upon the known incurred year-to-year link ratios. That is, due to the limited volume of railroad experience we do not know whether the ultimate realized losses, once all claims are settled, will be greater than or less than the developed estimate of incurred losses after nine years. Since we do not know how the railroad incurred losses are expected to develop after nine years and that there does not appear to be significant incurred loss development after seven years, we have assumed that the estimates of developed incurred losses after nine years are a reasonable representation of what the ultimate realized losses will be once all claims are settled. That is, we have assumed that there will be no incurred development after nine years. The cumulative incurred loss development factors, developed in Railroads Exhibit 2, were applied to the incurred undiscounted losses in order to estimate the undiscounted incurred losses after nine years. Since the railroad data used in this year's tariff analysis is evaluated as of 12/31/2005, this is the second evaluation of the 2004 Accident Year. Therefore, we applied the '1-2' cumulative incurred loss development factor to the 2004 Accident Year, the '2-3' factor to the 2003 Accident Year, etc…. This incurred data can be found in Railroads Exhibit 2. Determination of Ultimate Discounted Losses At this point we compared the estimated undiscounted paid railroad losses after nine years, developed in Railroads Exhibit 1, to the estimated undiscounted incurred railroad losses after nine years, developed in Railroads Exhibit 2, and calculated the difference. The difference is the undiscounted estimates of future payments that will be made after nine years. This data is summarized below: Accident Year Estimated Undiscounted Incurred Losses After Nine Years Estimated Undiscounted Paid Losses After Nine Years Undiscounted Estimate of Future Payments After Nine Years 1997 1998 1999 2000 2001 2002 2003 2004 4,275,545 2,466,995 7,034,258 10,782,599 4,559,903 4,780,321 8,693,226 13,438,118 3,256,756 1,943,701 4,815,541 8,889,859 3,838,213 4,091,912 3,700,640 5,815,034 1,018,789 523,294 2,218,717 1,892,740 721,690 688,409 4,992,586 7,623,084 In order to consider the income earned by investment of the case reserves, the values have been adjusted to a discounted basis. As noted in the compulsory tariff, the selected interest rate is 2.5%. Page 27 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance The incremental undiscounted paid losses in Railroads Exhibit 3 show the losses that have either already been paid or are expected to be paid for each Accident Year, up to nine years. The incremental undiscounted paid losses in Railroads Exhibit 3 were then all modified by the selected interest rate to provide incremental discounted losses. The incremental discounted paid losses to date are shown in Railroads Exhibit 4 while the incremental discounted estimate of future payments through nine years are shown in Railroads Exhibit 5. The number of years to discount assumes that loss payments are uniformly distributed within each payment period, thus the average payment date is July 1 of that year. The discount factor was calculated by the same calculation shown on Exhibit 3 of the compulsory tariff, (1 / 1.025) ^ (N), where N is the number of years to discount. Since we are using the 12/31/2005 evaluation for this year's railroad analysis, this is the second evaluation of the 2004 Accident Year. As a result, the incremental paid losses to date in Railroads Exhibit 4 go back eight time periods instead of only seven. Similarly, the incremental estimate of future payments through nine years in Railroads Exhibit 5 go forward seven time periods instead of eight. At this point, the undiscounted estimates of future payments in time periods after nine years still need to be discounted. Since we have no historical information to rely upon to help us predict when these estimated future payments will actually be made, we have made two assumptions while discounting these future payments. The first assumption is that all claims will be settled by the end of the twentieth year after having been incurred. The second assumption is that these loss payments will be uniformly distributed throughout the future period from 12/31/2005 to twenty years after having been incurred. We have discounted the undiscounted estimates of future payments after nine years based upon these assumptions. The discounted estimates of future payments after nine years are shown in Railroads Exhibit 6. The discounted estimates of future payments after nine years have then been added to the discounted paid losses to date and the discounted future payments through nine years in order to produce discounted total losses indexed to November 2004. The discounted total losses are shown in Railroads Exhibit 7. These calculations have provided us with estimates of the appropriate discounted total losses for each Accident Year. Determination of Tariff by Segment In order to calculate the appropriate Tariff the estimate of the discounted total losses must still be split between passenger and cargo operations. To produce an estimate of the ultimate discounted losses, we developed a series of paid Loss Modification Factors (LMFs) that modify the paid losses to date for each Accident Year, in order to account for loss development and the impact of discounting. The LMFs were developed by dividing the discounted total losses by the non-discounted total paid losses to date for each Accident Year (since they will be later applied to non-discounted paid Passenger and Cargo losses). The development of the paid LMFs can be found in Railroads Exhibit 8. The data provided for the review did not include a specific breakdown into cargo and passenger operations, but did include the engine number on many records. By comparing the engine number with the roster of train engines ISO Israel has assembled, it was possible to separate the claims into four categories: Cargo, Passenger, Mixed, and Unknown. Cargo and Passenger denotes engines that Page 28 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance are used solely for cargo or passenger operations. Mixed consists of the G-12 type engines which are used for both cargo and passenger operations while the Unknown category includes data without a reported valid engine number. The Mixed and Unknown data has been split based on volume comparisons. Based largely upon the relationship between the known passenger operations losses and the known cargo operations losses for Accident Years 1997-2001, where the losses are more mature, we judgmentally assumed that 75.0% of these Mixed and Unknown losses applied to passenger operations while the remaining 25.0% applied to cargo operations. The underlying data, as well as, the resulting 75.0% / 25.0% split of the Mixed and Unknown losses into passenger operations and cargo operations, respectively, is provided in Railroads Exhibit 9. In order to produce the Ultimate Discounted Losses, the losses were then multiplied by the Loss Modification Factors (LMFs) derived earlier. Due to the low volume of data and since there does not appear to be any reason for the losses to develop differently, the same factors were used for both passenger operations and cargo operations. The Ultimate Discounted Losses can be found in Railroads Exhibit 10. Trend Analysis Pure Premium As noted in the earlier chapter of the tariff document, loss trends are a particularly important aspect of prospective ratemaking. That is, the historical experience must be adjusted to reflect any anticipated economic and social changes expected in the future (when the developed tariff will be in effect). Due to the limited volume of railroad data, we have relied upon the analyses provided in Exhibit 6 (Sheet 2) and have made a trend selection to be used in the April 2007 Israel Railroad Tariff Analysis. While the historical data compiled by the Central Bureau of Statistics shows a decreasing frequency trend for automobile claims (Exhibit 6 Sheet 1), we do not believe it is appropriate to expect that railroads will follow a similar pattern. Given the limited number of railroad claims, a consistent selection of 0.0% frequency trend will tend to stabilize the trains ratemaking process from year to year, especially given the leveraged effect that changes in trend selections have on the indications. When examining the claim cost trends (Exhibit 6 Sheet 2), we note that the "Excess Insurance Costs over General CPI" trends indicate that a positive claim cost trend may be appropriate for tariff development purposes; however, there is not a clearly defined indicated claim cost trend. The 3-point fit indicates a trend of +1.7% and the longer term 6-point fit indicates a trend of +1.5%, while the 4point and 5-point fits indicate trends of +0.8% and +1.1%, respectively. Based largely upon the 4point and 5-point fits we have selected a claim cost trend of +1.0% for use in the Railroad tariff development. Trend Selected Frequency Claim Cost (Severity) 0.0% +1.0% Pure Premium (Frequency x Severity) +1.0% Page 29 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance In order to adjust the historical Railroads loss experience so that it reflects the expected prospective loss level, we have applied a pure premium trend factor of 1.0% to the Ultimate Discounted Losses. The trended losses are equal to the Ultimate Discounted Losses, presented in Railroads Exhibit 10, multiplied by (1.010)^N, where N is equal to the number of years from the average date of loss of the experience period to the average date of loss underlying the prospective experience period. The trend effective date for the Railroad policy is 4/1/2007. The computations are provided in Railroads Exhibit 11. Calculation of Loss Per Exposure Following the methodology used in prior reviews, two values were needed to estimate losses for 2007. First the passenger–kilometers of the system were estimated for passenger operations in 2007, and then the average loss per passenger–kilometer was determined. For cargo operations, cargo–tonkilometers were used in place of passenger-kilometers. It is expected that there will be continued growth in the actual number of passengers being carried by Israeli Railways in 2007. The expected continued growth can be partially attributed to more seats and trains being offered daily. However, there have also been additional stations opened on existing railway lines, and new lines have been established. We expect the average distance traveled per passenger for 2007 to be similar to that of 2005 and 2006. In order to utilize the passenger-km estimate provided by Israeli Railways, a measure of the loss per passenger-km is needed to be calculated. In Railroads Exhibit 12, a summary of the loss data for passenger operations and the average loss per passenger-km from 1997 to 2004 is provided. Due to the high variability of the values under the loss per passenger-km, no obvious trend could be derived. A weighted average (weighted on Passenger-Kilometers) was calculated to estimate the expected Loss Per 1,000 Passenger-Kms. In 2005, it was expected that there would be a decrease in the actual cargo-ton-kms hauled by Israeli Railways in 2005 as more of the commodities such as coal, gravel and grain were being hauled by truck, as opposed to railway. The actual cargo-ton-kms hauled by Israeli Railways did decrease, though not by as much as expected (original estimate: 1,056,000 [in 000’s]; actual: 1,138,014 [in 000’s]). In 2007, it is expected that there will be an increase in the actual cargo-ton-kms hauled by Israeli Railways, when compared to 2005, due to an increase in the volume of sand, garbage and minerals being hauled. In order to utilize the cargo-ton-km estimate provided by Israeli Railways, a measure of the loss per cargo-ton-km needed to be calculated. In Railroads Exhibit 12, a summary of the loss data for cargo operations and the average loss per cargo-ton-km from 1997 to 2004 is provided. Due to the high variability of the values under the loss per cargo-ton-km, no obvious trend could be derived. A weighted average (weighted on Cargo-Ton-Kms) was calculated to estimate the expected Loss per 1,000 Cargo-Ton-Kms. Final Recommended Pure Risk Calculation The estimated loss for passenger operations is obtained by multiplying the estimated 2007 passengerkilometer value (in 000's) by the loss per 1,000 passenger-kilometers value. Likewise, the estimated Page 30 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance loss for cargo operations is the product of the estimated 2007 cargo-ton-kilometers and the loss per 1,000 cargo-ton-kilometers. Once the estimated losses were calculated, a catastrophe surcharge was assessed and the Pool subsidy was applied. The catastrophe surcharge is equal to 24.92% of the estimated losses. The catastrophe percentage was computed in 1996, based on premium requested by a re-insurer that was given the loss data of Keter (which at that time received cargo service from the Israeli Railways). It was not verified otherwise, as up to now there is no statistical data regarding catastrophes in Israel. ISO Israel will continue to investigate the possibility of updating the catastrophe surcharge. The Pool subsidy load is approximately 4.4%, as explained in Exhibit 14-1 of the compulsory tariff. It should be noted that there are two different types of exposure for catastrophic losses for both cargo and passenger operations. The first type is an upset of the train itself. In this case, a cargo train might be hauling hazardous materials, while a passenger train could have a full load of passengers who could be injured. We believe that the exposure associated with this type of occurrence should be roughly equal. That is, while the losses caused by hazardous materials would likely have a higher claim payout, we also believe that it is less likely than a passenger train upset, simply because only a limited subset of the cargo trains are involved with hazardous materials and the overall volume of cargo traffic is smaller. The second type is if the train collides with another vehicle carrying hazardous materials. We believe that the exposure associated with this type of occurrence should also be roughly equal for cargo and passenger trains. That is, in this instance, either a cargo or a passenger train could impact the vehicle, and the resulting catastrophic losses due to a chemical spill from the vehicle would have to be absorbed, at least partially, by the train. We note that the passenger train would have an additional exposure associated with the passengers on board, while the cargo train may have an additional exposure to reflect that the cargo train may also be hauling hazardous materials that could spill. As mentioned above, while the losses caused by hazardous materials hauled by the cargo train would likely have a higher claim payout, we also believe that it is less likely than a passenger train occurrence, simply because only a limited subset of the cargo trains are involved with hazardous materials and the overall volume of cargo traffic is smaller. For these reasons, the catastrophe surcharge has been applied equally to both cargo and passenger operations. The computation of the final tariff for the pure risk is provided in Railroads Exhibit 13. Once the catastrophe surcharge and the Pool subsidy load were applied, the estimated losses for cargo operations needed to be split between ICI, Sherut and the Israel Railway system. According to estimates provided by Israeli Railways, using data by ton-kms, ICI is expected to carry 59% and Sherut 22% in 2007. The losses were then split according to this assessment. Carmelit Haifa In addition to the tariffs for Israeli Railways, Sherut and ICI, a tariff was also developed for the Carmelit. An implicit exposure was developed for the Carmelit by dividing the recommended ISO Israel tariff for the Carmelit, from April 1, 2001, by the corresponding Tariff per Exposure for passenger operations, developed at that time. The calculation and the resulting implicit exposure are provided in Railroads Exhibit 14. Page 31 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Reviewing the data received from the Carmelit, we noted that the total paid losses from 1993 through 2004 were 34,600 NIS. In addition, we noted that there were no claims incurred for 2003 and only a single claim was incurred for 2004, which has been settled for 3,000 NIS. These paid losses were generated from a total of 10 claims. This experience results in an average of 2,883 NIS per year and an average of 3,460 NIS per claim. Please note that these losses have not been indexed and are therefore at the level at which they were paid. In addition, these losses have not been discounted. We also noted that there are inherent differences in the risk assumed by the Carmelit as compared to Israeli Railways. Specifically, the Carmelit is a very limited cable railway system that is entirely enclosed underground (it does not cross roads or intersections). As a result, the Carmelit cannot be involved in any occurrence with another vehicle (except possibly one of their own rail cars - due to safety systems that are in place, there should be minimal risk of this occurring). To better reflect the consistently good overall experience, as well as, to better reflect the differences in risk assumed by the Carmelit and Israeli Railways, the tariff per exposure used to calculate the April 2007 tariff for the Carmelit was judgmentally reduced by 40%. The resulting tariff per exposure and the final recommended Carmelit tariff are provided in Railroads Exhibit 14. Page 32 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Appendix A – Final Paid Mack’s Age-to-Age Factors The final paid age-to-age factors were calculated using the chain ladder method as described in “Measuring the Variability of Chain Ladder Reserve Estimates” by Thomas Mack. The paper can be found on the Casualty Actuarial Society’s website at http://www.casact.org/pubs/forum/94spforum/94spftoc.htm. The age-to-age factors, fk, were obtained by taking a weighted average of the observed individual development factors. The age-to-age factors are calculated by I k fk C j 1 j , k 1 , I k C j 1 j ,k where I is the total number of accident/underwriting years, and Cj,k is the Year j accumulated total losses at development year k. Using data from Exhibit 2-5, the following is an example of how the f1 age to age factor was calculated. Year 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Total Age-to-Age Factor Age 0 1 34,791 43,252 61,985 61,627 62,097 75,143 93,626 108,732 100,822 94,365 105,002 122,066 115,882 101,776 109,025 103,852 100,421 77,983 97,628 1,670,074 207,057 240,094 311,480 345,135 370,787 423,215 506,758 555,200 545,855 575,329 541,885 560,201 596,071 586,721 579,625 574,626 551,339 508,072 563,302 9,142,750 9,142,750 1,670,074 = 5.474 The cumulative loss development factors by underwriting year are calculated by multiplying all of the age-to-age factors for the years not yet paid out for each specific underwriting year. As shown on Exhibit 2-5, the cumulative paid loss development factor for the 2004 losses, 27.069 , was calculated by taking the product of all of the age-to-age factors ( 5.474 x 1.814 x 1.395….x 1.003). Multiplying the 2004 paid losses by the cumulative paid loss development factors produces the ultimate Page 33 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance undiscounted loss estimate without adverse deviation for the year 2004, as displayed in the last column on Exhibit 2-5. That is, the 2004 undiscounted ultimate losses are calculated as follows: 90,702 x (5.474 x 1.814 x 1.395….x 1.003) = 2,454,632. A major benefit of the Mack Method is that it provides a means to calculate standard errors and coefficients of variation for the outstanding loss reserve estimates. This is important, as ultimate losses can never be forecasted exactly, as they are estimates. The equations and steps used to derive the standard errors for Israel can be found in the paper referred to on the previous page. Exhibit 5 shows the final results, including estimated total reserves, estimated ultimate losses, standard errors, and coefficients of variation for each of the underwriting years from 1985 to 2004. Page 34 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Appendix B -- Analysis of Missing MADA Payments During our discussions with the Israeli insurance companies in 2005, it was brought to our attention that they are currently in dispute with MADA, such that there has been a withholding of payments by insurers to MADA starting with underwriting year 2003. It is ISO Israel’s understanding that this dispute had not yet been resolved as of the end of 2004, which means that the paid loss data reported to the ISO Israel Database for underwriting years 2003 and 2004 as of December 2004 do not include any payments to MADA. However, historically the majority of MADA payments were made by the 12 month evaluation for all prior underwriting years. Furthermore, the historical loss development factors to be applied to the reported 2003 and 2004 underwriting years are based on historical data that includes MADA payments. As a result, an adjustment needs to be made to the reported underwriting year 2003 and 2004 losses to account for the missing MADA payments. In 2005 ISO Israel was able to obtain historical MADA payments from one large Israeli insurer active in the compulsory auto market. ISO Israel divided this data by that companies' market share to obtain an estimate of MADA payments on a full market basis. Based on prior years’ MADA payout patterns, ISO Israel estimated the amounts of the missing MADA payments at a 12 month and 24 month evaluation. ISO Israel then calculated the effects of the missing 12 month and 24 month MADA payments on the reported losses for underwriting years 2004 and 2003, respectively. These calculations can be found in Exhibits B1 and B2. As the Exhibits show, ISO Israel estimates the amount of the missing MADA payments as of 24 months to be 9,166,754 NIS for an effect of about 1.7% on the 2003 reported losses (as of 24 months). Similarly, the amount of the missing MADA payments as of 12 months is about 6,271,989 NIS for an effect of about 7.4% on the 2004 reported losses (as of 12 months). ISO Israel used these estimated amounts to adjust the underwriting year 2003 and 2004 losses in this tariff review. Page 35 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Appendix C - Experience Period for Tariff Determination and Determination of Trend Factors The last several years in Israel have seen economic activity impacted by domestic and global security situations. The impact on tourist visits has been most profound and this seems to have had a ripple effect on the economy in general. Unemployment rates in 2002, 2003 and 2004 have exceeded 10%. Events since January of 2005 indicate that the economic picture is improving with unemployment rates for 2005 less than 10%, and future forecasts indicating a continued decline in unemployment for 2007 and beyond. Given the nature of these changes in the economy and the expected relationship of economic activity to motor vehicle use and accidents, in 2005 ISO Israel and the Insurer Panel discussed and examined various issues and scenarios for determination of the pure risk premium portion of the Tariff. (Also impacting the determination of trend factors were potential adjustments in the base data for changes in data reporting for items examined separately elsewhere in this document, such as MADA payments, INII subrogation and changes impacting the settlement of claims.) For purpose of this analysis items which had uniform impacts by year for the years considered in determining pure risk premiums, such as the INII subrogation change, were not included in the trend analysis. In addition, any changes in settlement that would be prospective in nature do not impact the compiled data and do not need to be considered for trend purposes. The changes in the economic conditions throughout the experience period available for analysis raise the question of what is the appropriate base data for determining the pure risk premium and what is the appropriate period for selecting the trend factor. The results of the various sensitivity tests are as follows: Analysis Type Forecasted Unemployment Forecasted Unemployment Adjust 2003 & 2004 for MADA payments Adjust 2003 & 2004 for MADA payments Trend Selection 3.5% Adjust 2003 & 2004 for MADA payments -3.0% Adjust 2003 & 2004 for MADA payments Adjust 2003 & 2004 for MADA payments -3.0% 4.9% -4.0% -2.4% 0% Data 7 year base; 7 year fit for trend 5 year base; 5 year fit for trend 7 year base; 7 year fit for trend 7 year base; 5 year fit for trend (exc. hi 1998 & low 2004) 7 year base (first 2 years untrended); 5 year fit for trend 5 year base; 5 year fit for trend 7 year base; 7 year trend selection Indicated Change + 3.1% +2.9% -35.6% -28.7% -25.7% -31.8% -17.1% Conclusion ISO Israel has examined the results under the various years of data to be used and the source for the evaluation of the appropriate trend factor. Much discussion has occurred and a good deal of consideration has been given to the question of including the earlier (1998, 1999) mostly pre- Page 36 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance recession years in the determination of the trend factors. While it is notable that the economic activity in those two years is at a different level than the later years, any analysis that selects the trend factors based on data from the period 2000 to 2003 because it is believed to be more “homogeneous” should also recognize that to then include the earlier “non-representative” years unadjusted for any difference in the base data for purpose of calculating the risk premium would be biased. After careful reflection ISO Israel concludes that after adjustment of the 2003 and 2004 years for MADA payments and consideration of the changes in economic activity level, the appropriate overall trend factor is 0% applied to the 7 year data base. Page 37 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Appendix D - Derivation of Average Class Factors To calculate the average rating factors underlying the current risk premiums, ISO Israel used underwriting year 2004 data from the ISO Israel Database. In the case of motorcycle risks, Pool data was used to determine average rating factors for both the Pool and the voluntary market, as the voluntary market motorcycle data was too sparse to produce meaningful distributions. Additionally, the average class factor used for the personally owned motorcycle risks in this analysis is based on the tempered classification factors for motorcycle risks implemented in the January 1, 2007 Pool tariff changes. However, for your information, ISO Israel has also included an "alternate" average class factor for personally owned motorcycle risks which is based on the motorcycle classification factors originally recommended in our 2005 class plan filing. It should be noted that in some cases, the total exposure counts may vary from one rating variable to another for a given vehicle type, due to data being bypassed for problematic statistical reporting. Rated Driver Age and Gender Distribution The age and gender distribution, and resulting average rating factors are displayed on Exhibit D-1 for privately owned private vehicles and D-8 for privately owned motorcycles. Years Licensed in Israel Distribution The years licensed distributions, and resulting average rating factors are displayed on Exhibit D-2 for privately owned private vehicles and D-9 for privately owned motorcycles. Number of Accidents Distribution The number of accidents distribution, and resulting average rating factors are displayed on Exhibit D-3 for privately owned private vehicles and D-10 for privately owned motorcycles. Number of Major Convictions Distribution The number of major convictions distribution, and resulting average rating factors are displayed on Exhibit D-4 for privately owned private vehicles and D-11 for privately owned motorcycles. Air Bags Indicator Distribution The air bag indicator distribution and resulting average rating factors are displayed on Exhibit D-5 for privately owned private vehicles, and D-7 for corporately owned private vehicles. Engine Size Page 38 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance The engine size distribution is displayed on Exhibit D-6 for privately owned private vehicles. Since the engine size class factor is strictly 1.00 for all engine size classes, the average rating factor is 1.00 by default. Once the average rating factors for each variable were determined, the overall average rating factors by vehicle type were calculated as follows: Privately Owned Private Vehicles Average Rating Factor = (Age and Gender average rating factor + Years Licensed average rating factor + Number of Accidents average rating factor + Number of Major Convictions average rating factor) x Air Bags Indicator average rating factor x Engine Size Rating average rating factor = (1.020 + 0.028 + 0.001 + 0.001) * 0.988 * 1.000 = 1.037 Corporately Owned Private Vehicles Average Rating Factor = Air Bags Indicator average rating factor = 0.983 Privately Owned Motorcycles Average Rating Factor = (Age and Gender average rating factor + Years Licensed average rating factor + Number of Accidents average rating factor + Number of Major Convictions average rating factor) = 1.006 +0.019 + 0.000 + 0.001 = 1.026 Alternate Privately Owned Motorcycles Average Rating Factor For your information, ISO Israel has calculated the average rating factor for the motorcycle risks using the 2004 data from the ISO Israel DataBank and the motorcycle classification factors originally recommended in our 2005 class plan filing. The alternate average rating factors by rating variable are contained on Exhibits D-12, D-13, D-14 and D-15. The overall alternate average rating factor for privately owned motorcycles equals: (Age and Gender average rating factor + Years Licensed average rating factor + Number of Accidents average rating factor + Number of Major Convictions average rating factor) = 1.016 + 0.024 + 0.001 + 0.002 = 1.043 Page 39 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Appendix E -- Change in the Retirement Age In 2005 several insurers expressed concern about the impact that an increase in the retirement age would have on compulsory insurance. As a result, ISO Israel has analyzed the potential impact the proposed increase in the retirement age would have on compulsory insurance. For this analysis, ISO Israel started with the number of seriously injured and deaths by age and gender from the 2003 CBS Casualties in Road Accidents reports. ISO Israel believes that these are the claimants that would most likely be eligible for long-term income benefits from compulsory insurance. ISO Israel then calculated estimates of the total income benefits payable for these claimants under 2 scenarios -- first with retirement ages of 65 (males) and 60 (females), and then with retirement ages of 68 (males) and 63 (females). Comparing the total income losses in the second scenario to those in the first scenario provides an estimate of the percentage increase in income losses resulting from the retirement age change. Even though the legal retirement age was raised to 67 for males and 62 for females, one has to take in account that the courts often consider higher retirement age for professional and independent business people. Exhibits E-1 and E-2 provide an estimate of the increase in income losses due to the proposed change in retirement age using the following underlying assumptions and parameters: All death claims for claimants other than minors (where minors are defined as individuals age 18 or less) result in income payments from the date of accident up to the lower of the retirement age or the total life expectancy. 10% of the seriously injured claimants will be eligible for income payments for the rest of their working life (i.e., from the date of accident up to the lower of the retirement age or the total life expectancy.) The remaining seriously injured claimants will not receive income benefits for the remainder of their working life, and therefore are not affected by the proposed change in retirement age. For minor claimants that are severely injured, the starting age of work would have been 18 for income benefit purposes. Income benefits are based on the national average wage -- or 7,000 NIS per month. Income benefits are payable on a present value basis using a discount rate of 2.5% and the following formula: The present value of a stream of future payments (PV) = PMT [(1 - (1 / (1 + i) ^ n)) / i], where PMT = amount of each payment (on an annual basis) i = interest rate per period (on an annual basis) n= number of periods (on an annual basis) So, for example, for a male claimant that is killed at age 22, with an assumed retirement age of 68, the present value of the income benefit equals: (7,000*12) * [(1-(1 / (1.025) ^ 46 )) / .025] = 2,280,950 NIS Page 40 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance For further details on the calculations underlying Exhibit E, please refer to the Exhibit E "Notes" sheet. ISO Israel believes that excluding income payments that are not remainder of life payments (i.e., are not affected by the change in retirement age) from this analysis gives us the "worst case scenario" -as the partial income payments would be included in both scenarios 1 and 2 and would serve as a ballast in the calculation, reducing the estimated increase. As Exhibit E-2 illustrates, the estimated increase in income losses resulting from the change in retirement age (i.e., from 65 to 68 for males and from 60 to 63 for females) using the above assumptions and parameters is equal to 7.5% on a present value basis. However, this increase is the anticipated increase in income losses; the bottom line effect on total compulsory losses is then calculated as: The increase in income losses resulting from the retirement age change * the percentage of compulsory losses that are currently income losses Assuming income losses are currently 20% of all compulsory losses (based on data from PIP coverage in the state of Michigan -- the US coverage most similar to Israel compulsory auto), this results in an increase of .075 * .20, or a 1.5% increase in total compulsory losses. ISO Israel then performed a sensitivity analysis to test the impact of varying the following underlying parameters on the results of this analysis: Assuming 1% of seriously injured claimants would be eligible for income benefits for the rest of their working life. Assuming 50% of seriously injured claimants would be eligible for income benefits for the rest of their working life. Assuming all seriously injured claimants would be eligible for income benefits for the rest of their working life. Using a starting working age of 20 Using an average wage of 21,000 NIS (3 times the average wage) rather than 7,000 NIS for the benefit calculation Using a discount rate of 0% (i.e., not taking present value into consideration) Using a discount rate of 5% Assuming income losses currently represent 10% of total compulsory losses The results for these different scenarios are contained on Exhibit E-3 . As Exhibit E-3 illustrates, varying the percent of serious injuries resulting in income benefits for life, the starting age of work and the monthly wage benefit have small effects on the resulting increase in Page 41 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance income losses resulting from the retirement age change. The parameter that has the largest effect on the results is the discount rate used in the analysis -- with the increase in income losses decreasing as the discount rate used increases. ISO Israel believes that 2.5% is a reasonable discount rate to use in this analysis, for the same reasons noted in our proposed April 2007 countrywide tariff level indication. Using a 2.5% discount rate, this analysis indicates that the change in retirement age will have approximately a 7.5% increase in income losses under compulsory insurance. This in turn translates to about a 1% effect on total compulsory losses if we assume income losses are currently 10% of all compulsory losses and about a 1.5% effect on total compulsory losses if we assume income losses are currently 20% of all compulsory losses. In Michigan -- the state in the US that has a coverage most comparable to Israel's compulsory auto -- income losses represent about 20% of all losses. Thus, ISO Israel believes that 20% is a more reasonable estimate of the percentage of total losses that are income than 10%. As a result, ISO Israel estimates that the change in the retirement age will have an increase in ultimate losses of about 1.5%, which is not reflected in the countrywide tariff level indication on Exhibit 9. Page 42 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Appendix F - Increased Subrogation Efforts of the INII ISO Israel is aware of the fact that the Israeli National Insurance Institute (INII) has recently increased subrogation efforts for claims against insurance companies due to compulsory auto accidents. This is a relatively new initiative, and thus the increased liability due to these efforts is not reflected in the losses for the underwriting years used for this countrywide tariff level review. A sample of Israeli insurance companies representing about 50% of the market provided ISO Israel with information on INII payments in 2005 to analyze this issue. Exhibit F calculates ISO Israel's estimate of the impact of the increased subrogation efforts of the INII, using the claim payment data provided by these insurers. Column (1) represents INII payments made by the insurers, as reported to ISO Israel; Column (2) represents all payments made by these same insurers as reported to Avner, on a fully developed basis; and Column (3) calculates the corresponding percentages. Table A shows INII payments as of December, 2001; Table B shows INII payments as of December, 2002, and Table C shows INII payments as of December, 2004. All three tables relate these INII payments to total payments reported to Avner as of December 2002, fully developed to ultimate levels and indexed to December, 2004. Therefore, the percentages in Tables A through C can be directly compared to assess the changing significance of INII liabilities over time. It is ISO Israel's understanding that subrogation efforts can go back at most 7 underwriting years once subrogation activity is initiated. Furthermore, based on a review of historical INII payments in Israel, it appears that subrogation efforts begin no later than 3 years after the December evaluation of an underwriting year, or in other words, INII activity goes back 10 underwriting years at most at any given date. Thus, on Table A, where INII payments are as of December, 2001, underwriting years 1992 and prior should be fully developed with regard to INII payment liabilities. For Table C, as of December 2004, underwriting years 1995 and prior should be fully developed with regard to INII payment liabilities. Additionally, a review of INII payments from 2001 to 2004 indicates that the INII increased their subrogation efforts significantly in 2002 and 2003. Thus, INII payments as of December 2001 can be considered "pre-increased INII subrogation efforts". In Table C, underwriting years 1993 through 1995 are of particular interest, as these underwriting years were still "open" when increased INII subrogation efforts began in 2002, but should be fully developed with regard to INII payment liabilities by December 2004. By comparing the INII percentages for underwriting years 1993-1995 in Table C with those of underwriting years 1992 and prior in Table A, we obtain an estimate of the increased INII percentage due to subrogation efforts. The percentage increased from 2.4% to 4.5% However, it can also be seen in Table C that underwriting year 1997 has an INII percentage that is above 4.5%. This is not surprising, since this later underwriting year has had the benefit of 3 years of increased subrogation efforts by the INII. Consequently, ISO Israel estimates that the ultimate INII percentage for future underwriting years will be between 5.0% and 5.5%. Due to the 10 year restriction on subrogation efforts, it follows that the INII subrogation efforts in any calendar year will concentrate most heavily on the earliest underwriting years in that time period. Thus, it follows that the subrogation efforts as of December 2004 (the evaluation underlying our tariff analysis) would not be fully reflected in the most recent seven underwriting years used in our countrywide tariff level indication. However, ISO Israel believes that since these underwriting years Page 43 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance are developed to an ultimate basis, the historical loss development factors that are applied have built in the historical INII liability provisions, such that the percentage of INII losses built in for the underwriting years used in Exhibit 9 equals the percentage that existed prior to the increased subrogation efforts of the INII -- or 2.4%. Therefore, ISO Israel estimates that the increased subrogation efforts by the INII represent an increase in ultimate losses of about 3%, which is not reflected in the countrywide tariff level indication on Exhibit 9. Page 44 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Appendix G -- Changes in Law Affecting the Settlement of Claims During our discussions with the Israeli insurers while preparing the 2005 Tariff review, several proposed or actual changes in the settlement of claims were raised. ISO Israel has consulted with claims experts and made an independent evaluation of the effects of each as well as the appropriateness of adjusting the historical data at this time. “Chok ha-Hesderim” At issue is the 2005 Economic Policy Act, which includes a change to the work invalidity threshold, so that only permanent invalidity levels of 9% and above will be entitled to a one time grant. Prior to the 2005 Economic Policy Act, a level of invalidity greater than 5% qualified for this benefit. (The one time benefit prior to this Act was paid on levels of invalidity between 5% and 20%.) ISO Israel believes that the change will have a small effect. Making the assumption that levels of disability as reported by the INII in the general population are a fair approximation of the expected levels of disability in auto accidents, ISO Israel found that the group with disability levels between 0 to 20% (which really translates into 5 to 20% as currently an invalidity level of 5% or more qualifies for this benefit), disability comprise 4% of all pension recipients. The number of individual recipients at the end of 2003 was 970. Not all of these individuals were injured in auto accidents. Assuming that 50% were auto accident injuries would mean that there are 485 individuals with a disability level between 5% and 20%. Changing the eligibility level to 9% and assuming a uniform distribution would mean that there would have been approximately 130 cases that would not be compensated from INII but would now need to be compensated by the compulsory auto coverage. (i.e., changing the threshold from 5% to 9% is a 4 point increase in a range from 5 to 20, or approximately 27% of the claimants if we assume a uniform distribution). These benefits are paid by the formula as follows: Invalidity percentage* ¾ of monthly income prior to accident* number of months of invalidity (at most 43) The resulting value would then have a maximum of: 130* .07*(.75*7,000)*43 months = 2,054,325. The cost to the Paltad of this change in 2003 would have been no more than 2,054,325 NIS. This is equal to less than 0.1% of the previously estimated ultimate incurred losses. Changes in the Road Accident Law There currently exists a proposal to increase the limit of pain and suffering from 150,000 NIS to 280,000 NIS. This change, if enacted, would impact the cost of compulsory insurance. However, the specifics of this provision are in a private law proposal and the Knesset committee has considered the idea of a change without making any specific recommendation. Since the precise details of any change in this regard are not known at this time, ISO Israel would not recommend making any adjustment to the indicated tariff at this time, but would recommend that when such a proposal is final that the tariff be adjusted to reflect any increase in liability. Page 45 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance With regard to the potential changes in the definition of a road accident, to date there are a number of proposals that have yet to be enacted and should be evaluated at the time that they are final. National Insurance A change has been made to increase the employer’s share of the injury compensation from the current first 9 days to the first 12 days. We believe that this impact will be small as payment for lost income for all auto accidents represents about 20% of all losses. In order to evaluate the impact of this change, ISO Israel examined wage loss information by length of disability in days from US auto insurance studies conducted by the Insurance Research Council. In doing this analysis ISO Israel calculated the insurer share under a treatment consistent with the current Israel system where the employer would pay the disability for the first 9 days and under the future 12 day level. The calculation indicates that this change would produce a 20% increase in the amount of losses paid for disability compensation for work injury auto accidents. Since the entire wage compensation portion of the losses is 20% reflecting both work and non work auto accidents, this change could have no more than a 4% effect if all wage compensation were for work injury auto accidents. While Avner has indicated that historically about 10% of all road accidents are work accidents, given the lack of detailed data available for analysis on this issue, ISO Israel has used a more conservative assumption that 25% to 50% are work accidents. As a result, ISO Israel estimates that this change will increase losses no more than 1% to 2%, and thus has selected an effect of 1.5%. The details of the 20% increase are shown on Exhibit G. Page 46 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Appendix H -- The "Lost Years" In the "lost years" ruling, the Israeli courts have awarded income benefits to claimants that were not previously entitled to income benefits. The award for "lost years" represents the lost income that would have been earned by the claimant had they not died or experienced curtailment of life as a result of a road accident. This benefit is payable to "untraditional" beneficiaries in some cases, such as the parents or the estate of the deceased. These awards represent a broadening of compulsory auto insurance in Israel, following a similar broadening of tort rulings. The lost years ruling is applicable to all new cases, and also retroactively, to cases that have been submitted in the past but have not yet been decided at this time. While in theory the lost years could generate additional payments in curtailment of life cases, after consultation with ISO Israel’s legal counsel (Advocate Alon Belaga), ISO Israel believes that the effect of the lost years on curtailment of life cases will be insignificant. That is, in many curtailment of life cases it is very difficult to estimate the life expectancy of the injured, and thus periodical payments during the life of the injured - rather than a one-time payment - are often made. In the Varnevski case, the Supreme Court decided to uphold the periodic payments decided on by the regional court for this very reason. Furthermore, the Supreme Court clarified that the lost years doctrine does not apply while periodical payments are made. In these cases a lump sum compensation is applicable only for the lost “remainder of life”, from the actual death to the expected non-injured life span. Consequently, even if a claimant's lifespan is expected to be shortened directly as the result of injuries sustained from a road accident, in order for the lost years to come into play, such a claimant would have to die before the income benefits they currently receive would stop, which is at retirement age (age 67). (According to the Wassergrug Supreme Court Case, for curtailment of life cases there are – if any - lost years only between the actual year of death and the minimum age of retirement, 67). For these reasons, our legal expert has advised, and ISO Israel agrees, that we should expect very little, if any, effect of the lost years on compensation for curtailment of life cases. Thus, ISO Israel believes the major impact of the lost years ruling on compulsory auto insurance will be from death claims resulting from road accidents. For death claims, there are 3 different scenarios for determining the lost years benefits: Scenario 1 -- Claimant is a deceased minor Scenario 2 -- Claimant is a deceased single adult, with no dependents Scenario 3 -- Claimant is a deceased adult with dependents Page 47 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Scenario 1 -- Claimant is a deceased minor Exhibit H-1 provides ISO Israel's estimate of the additional Compulsory Insurance losses expected to result from the lost years ruling for scenario 1 -- minors who have been killed in a road accident. Following are some notes on the analysis contained in Exhibit H1: This analysis starts with the total number of road accident death claims expected per year, which is estimated by using the number of deaths by age and gender from the 2004 CBS Casualties in Road Accidents reports. For this scenario, only the deaths for minors are considered. For this analysis a minor is defined as someone up to age 21, as individuals are in the army until that time. Per the April 2006 Supreme Court Decision on the Hagai Fintz case, "no compensation should be given for the lost army service years". That is, for minors, the lost years provide for loss of income benefits from the starting age of work to retirement. Per the April 2006 Supreme Court Decision on the Hagai Fintz case, the starting age of work is 21, and retirement age is 67. The lost years benefits payable for minors are 30% of the average salary in the market. Again, this is per the Supreme Court Decision on the Hagai Fintz case which indicated that the court must study the status of the injured person at the time of death; if at the time of death he has no dependents, he should be compensated as an injured with no dependents, using a global rate of 30% of the average salary in the market. The court also decided that the working assumption for minors, whose working career and earning patterns have not yet been established, is that the income loss compensation should be based on the average salary in the market. The national average salary in Israel is 7,472 NIS per month, based on the 2006 average through March 2006, from the Bank of Israel; when converted to a November 2004 index it is 7,277 NIS per month. . Income benefits are payable on a present value basis using an interest rate of 2.5%. For minors, there is "double discounting" -- meaning that first the present value of the stream of annual payments from age 21 to age 67 is determined, and then that result is discounted to get the present value of the benefit at the age of the claimant’s death. Again, this is per the Supreme Court Decision on the Hagai Fintz case. The present value of the stream of annual payments from age 21 to 67 is determined using the following formula: The present value of a stream of future payments (PV) = PMT * [(1 - (1 / (1 + i) ^ n)) / i], where PMT =amount of each payment (average lost wage on an annual basis) i = interest rate per period (.025 on an annual basis) n = number of periods (in year; in this case = 46 years from age 21 to age 67) So, for minors, this benefit equals: (7,277 * .30 * 12) * [(1 - (1 / (1.025) ^ 46 ) ) / .025] = 711,363 NIS Page 48 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance Next, this total amount is discounted once again, to bring it to the present value at the time of the claimant's death, using the following formula: (PV) =BENEFIT FROM ABOVE * [ 1 / (1 + i) ^ n)], where i = interest rate per period (.025 on an annual basis) n = number of periods (in years; in this case from age at death to age 21) So, for example, for a claimant that was age 2 at the time of death, this benefit equals: 711,363 * (1 / (1.025) ^ 19) = 444,977 NIS The total present value of the benefits resulting from the lost years by age and gender is determined by multiplying the lost years benefit by age and gender times the number of estimated claimants by age and gender. The overall estimated lost years benefits for scenario 1 are then determined by summing up the total benefits resulting from the lost years for all ages for males and females combined. Scenario 2 -- Claimant is a deceased single adult, with no dependents Exhibit H-2 provides ISO Israel's estimate of the additional Compulsory Insurance losses resulting from the lost years ruling for scenario 2 -- single adults, without dependents, that have been killed in a road accident. Following are some notes on the analysis contained in Exhibit H-2: This analysis starts with the total number of road accident death claims expected per year, which is estimated by using the number of deaths by age and gender from the 2004 CBS Casualties in Road Accidents reports. For this scenario, only the deaths for single adults without dependents are considered. An adult is defined as someone aged 21 or higher. In order to determine the number of adults that are single and do not have dependents, ISO Israel multiplied the number of adult deaths by the percentage of individuals never married for that age group (statistics on the percentage never married are from the 2003 CBS Population by Marital Status report as 2004 data for this statistic were not readily available). For adults, the lost years ruling provides income benefits from the age of accident to the lower of the retirement age or the total life expectancy. Retirement age is 67, as per the Supreme Court Decision on the Fintz case. The lost years benefits payable for single adults without dependents is 30% of the salary at the time of death. Again, this is per the Supreme Court Decision on the Hagai Fintz case which indicated that the court must study the status of the injured person at the time death; if at the time of death he has no dependents, he should be compensated as an injured with no dependents, using a global rate of 30% of the average salary in the market. Additionally, the court noted that since the actual salary for each claimant is not known, the income benefits for these claimants is estimated using the national average salary in Israel -- which is 7,472 NIS per month, based on the 2006 average through March 2006, from the Bank of Israel, Page 49 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance converted to a November 2004 index; when converted to a November 2004 index it is 7,277 NIS per month. Income benefits are payable on a present value basis using an interest rate of 2.5%. For adults, the present value of the stream of annual payments from age at accident to age 67 is determined using the following formula: The present value of a stream of future payments (PV) = PMT * [(1 - (1 / (1 + i) ^ n)) / i], where PMT =amount of each payment (average lost wage on an annual basis) i =interest rate per period (.025 on an annual basis) n =number of periods (in years, from age at accident to age 67) So, for example, for a claimant that died at age 32, the benefit equals: (7,277 * .30 * 12) * [(1 - (1 / (1.025) ^ 35) ) / .025] = 606,338 NIS The total present value of the benefits resulting from the lost years by age and gender is determined by multiplying the lost years benefit by age and gender times the number of estimated claimants by age and gender. The overall estimated lost years benefits for scenario 2 are then determined by summing up the total benefits resulting from the lost years for all ages for males and females combined. Scenario 3 -- Claimant is a deceased adult with dependents Exhibit H-3 provides ISO Israel's estimate of the additional Compulsory Insurance losses resulting from the lost years ruling for scenario 3 -- adults with at least 1 dependent, that have been killed in a road accident. Following are some notes on the analysis contained in Exhibit H-3: This analysis starts with the total number of road accident death claims expected per year, which is estimated by using the number of deaths by age and gender from the 2004 CBS Casualties in Road Accidents reports. For this scenario, only the deaths for adults with at least 1 dependent are considered. An adult is defined as someone aged 21 or higher. In order to determine the number of adults that have at least 1 dependent, ISO Israel multiplied the number of adult deaths by 1 minus the percentage of individuals never married for that age (i.e., 1 - the percent assumed to have no dependents in scenario 2). For adults, the lost years ruling provides income benefits from the age of accident to the lower of the retirement age or the total life expectancy. Retirement age is 67, as per the Supreme Court Decision on the Fintz case. The lost years benefits payable for adults with dependents is based on the hands approach and the salary at the time of death. However, the hands approach was changed such that a savings hand was introduced effective with the Supreme Court Decision on the Hagai Fintz case, which determined that if there are dependents the compensation will be based on the Page 50 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance hands methods using the known family status with the addition of one savings hand. That is, prior to the lost years, income benefits were paid using the hands approach as follows: Average income * (# dependents hands + 1 household hand) after tax (# dependents hands + 1 injured hand + 1 household hand) However, after the lost years (specifically, with the Supreme Court Decision on the Fintz case), income benefits are payable as follows: Average income * (# dependents hands + 1 household hand + 1 savings hand) after tax (# dependents hands + 1 injured hand + 1 household hand + 1 savings hand) Since the addition of the savings hand increases the income benefits payable, the effect of the lost years on this scenario is estimated based on the additional losses expected to be generated by the change in the hands calculation. In both cases, income benefits are payable on a present value basis using an interest rate of 2.5%. For adults, the present value of the stream of annual payments from age at accident to age 67 is determined using the following formula: The present value of a stream of future payments (PV) = PMT * [(1 - (1 / (1 + i) ^ n)) / i], where PMT =amount of each payment (average lost wage on an annual basis) i =interest rate per period (.025 on an annual basis) n =number of periods (in years, from age at accident to age 67) Since the actual salary for each of the individual claimants is not known, the income benefits for these claimants is estimated using the national average salary in Israel -- which is 7,472 NIS per month, based on the 2006 average through March 2006, from the Bank of Israel; when converted to a November 2004 index it is 7,277 NIS per month. A 10% income tax then applies. Since the actual number of dependents for each of the individual claimants is not known, it is estimated using the average household size in Israel. According to the 2004 CBS Report on Households, the average number of persons per household in Israel is 3.4; thus, an injured person would have 2.4 dependents on average. So, for example, for a claimant with dependents that died at age 32, prior to the lost years the income benefit equaled: (7,277 * .90 * 12) * [(2.4 + 1) / (2.4 + 1 + 1)] * [(1 - (1 / (1.025) ^ 35) ) / .025] = 1,405,652 NIS After the lost years, that same claimant is eligible for a benefit equal to: (7,277 * .90 * 12) * [(2.4 + 1 +1) / (2.4 + 1 + 1 +1)] * [(1 - (1 / (1.025) ^ 35) ) / .025] = 1,482,031 NIS Page 51 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance The difference in these benefits is the additional loss resulting from the lost years ruling. For example, for the 32 year old claimant: 1,482,031 - 1,405,652 = 76,379 NIS The total present value of the benefits resulting from the lost years by age and gender is determined by multiplying the additional losses generated by the lost years ruling by age and gender times the number of estimated claimants by age and gender. The overall estimated increase due to the lost years for scenario 3 is then determined by summing up the total additional benefits resulting from the lost years for all ages for males and females combined. For further details on the calculations underlying Exhibits H-1, H-2 and H-3, please refer to the Exhibit "H Notes". As Exhibit H-4 illustrates, the estimated total additional losses resulting from the lost years ruling are equal to the sum of the losses anticipated for scenarios 1, 2 and 3. Thus, the estimated effect of the lost years ruling on compulsory auto insurance is calculated as: the total losses resulting from the lost years ruling for 2004 claimants total ultimate, discounted compulsory losses for an average underwriting year, with an adverse deviation provision For this analysis, ISO Israel used an average of the losses for the last 7 underwriting years to determine the overall compulsory losses for an average underwriting year, so the calculation is: 131,536,103 = 0.052 2,544,919,341 That is, we expect the lost years to have a +5.2 % effect on compulsory auto insurance losses. Sensitivity Analysis In order to test the sensitivity of using one year of CBS data, ISO Israel performed this analysis three separate times using different years of CBS data -- 2002, 2003 and 2004. The results of these three analyses were very similar, yielding results ranging from +4.8% to +5.8%, and averaging +5.3%. Thus, the use of a single year of 2004 CBS data appears to provide an accurate and reasonable estimate of the impact of the lost years. Page 52 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483 18 January 2007 Version Proposed 2007 Base Tariffs Israel Compulsory Vehicle Insurance EXHIBITS Page 53 of 53 Copyright, ISO Insurance Services Office of Israel, Ltd., 2007 106759483