No. 03-1129

General Remarks and official action taken:

On this day came on for consideration by the Commissioner of Insurance (Commissioner) the matter of the 2003 manual rate filings by the Texas Windstorm Insurance Association (the Association) with the Texas Department of Insurance (TDI), as required by the TEX. INS. CODE ANN. Art. 21.49 §8(h), for commercial and residential risks. Notice of an open meeting concerning the 2003 manual rate filings for all types and classes of risks submitted by the Association was posted by the Office of the Secretary of State on its internet website on September 10, 2003 . On September 22, 2003 , an open meeting was convened before the Commissioner in Room 100 of the Hobby Building to consider the Association's rate filings and to allow all interested persons to present written and oral comments. The following persons presented comments at the open meeting: Mr. Jim Shawn and Mr. James Murphy on behalf of the Association; Mr. Rod Bordelon and Mr. Ken Lovoy on behalf of the Office of Public Insurance Counsel (OPIC); and Mr. Jay Thompson on behalf of The Insurance Council of Texas (ICT).

After considering the Association´s rate filings, staff´s analysis, and all written and oral comments presented, the Commissioner adopts the following findings of fact and conclusions of law.

Findings of Fact

1. The Association's rate filings (reference number P-0803-18) were filed with TDI on August 12, 2003. In its rate filings, the Association proposed an increase of ten percent for its commercial property rates and an increase of ten percent for its residential property rates.

2. On September 17, 2003 , actuaries representing the Association, OPIC, and TDI shared additional information concerning the Association's rate filings in a telephone conference call.

3. On September 22, 2003 , OPIC filed with TDI calculations documenting its review of the Association's rate filings. OPIC´s calculations indicated a uniform decrease of 14.8 percent for the commercial property rates and a uniform decrease of 19.6 for the residential property rates.

4. OPIC recommended uniform decreases of ten percent for both commercial and residential rates.

5. Neither the Association nor OPIC recommended changes in the classification relativities for the commercial and residential rates.

6. The Association and OPIC presented comments and recommendations concerning the Association's rate filings at the open meeting on September 22, 2003 . ICT presented a statement in support of the rate filing of the Association.

7.The Association's proposed rates were based on three components of its rates: non-hurricane losses and associated loss adjustment expenses (LAE); hurricane losses and associated LAE; and expenses other than LAE.

COMMERCIAL PROPERTY

Non-hurricane Losses and Associated LAE

8. Both the Association and OPIC used ten years (1993 through 2002) of the Association´s non-hurricane commercial premium and paid loss experience data, valued as of December 31, 2002, to project the component of the rates intended to cover non-hurricane losses and associated LAE.

9. TEX. INS. CODE ANN. Art. 21.49 §8(h)(13) requires the use of the ten years of experience described in Finding No. 8 to project the provision for non-hurricane losses in the Association rates.

10. The Association and OPIC selected their paid loss development factors from arithmetic averages of the loss development indications of between nine years of data (the 12 to 24 months´ indication) and five years of data (the 60 to 72 months´ indication).

11. The Association and OPIC adjusted the premiums earned during each experience year to current rate levels by the application of current rate level factors.

12. The Association calculated loss ratio trend factors by dividing the product of historical and prospective loss trends by the product of historical and prospective premium trends. However, the Association´s originally filed loss ratio trend factors were not calculated correctly. On September 18, 2003, the Association filed corrected loss ratio trend factors, which resulted in a slight decrease (0.2 percentage points) in their originally filed non-hurricane loss ratio.

13. OPIC adopted the Association´s originally filed loss ratio trend factors for its calculations.

14. The Association's corrected loss ratio trend factors are reasonable.

15. The Association calculated a factor of 1.260 to include associated non-hurricane LAE based on the experience for non-hurricane years in the 27-year period from 1976 through 2002.

16. OPIC adopted the Association´s LAE factor for its calculations.

17. It is reasonable to apply a factor of 1.260, as recommended by the Association and adopted by OPIC, to non-hurricane losses in order to include associated LAE.

18. The Association and OPIC differed in their calculations of the 10-year average loss ratios in that the Association used a premium-weighted average while OPIC used an arithmetic average.

19. The resulting Association´s average loss and LAE ratio was 0.154 while OPIC´s loss and LAE ratio was 0.145, based on the Association´s corrected loss ratio trend factors.

20. Either averaging approach described in Finding No. 18 would be reasonable given the limited number of years used to project non-hurricane losses and LAE.

21. Based on Finding Nos. 8-20, a non-hurricane loss and associated LAE ratio of 0.154 is reasonable and should be used to calculate the Association´s commercial property rates.

Hurricane Losses and Associated LAE

22. The Association utilized six different estimates to project its rate needs to cover hurricane losses and associated LAE: a 32.3-year historical analysis; a 32.3-year plus one hurricane historical analysis; a 32.3-year adjusted to the actual 103-year frequency level historical analysis; a computer model prepared by Applied Insurance Research (AIR); a computer model prepared by EQECAT; and a computer model prepared by Risk Management Solutions (RMS). The first three estimates were based on 90 percent of the industry (excluding the Association) extended coverage experience and 100 percent of the Association´s experience. The Association averaged the last three estimates to arrive at a single indication.

23. OPIC utilized four different estimates to project its rate needs to cover hurricane losses and associated LAE: a 32.3-year historical analysis and an arithmetic average of the hurricane loss and LAE ratios; a 32.3-year historical analysis and a premium-weighted average of the hurricane loss and LAE ratios; a 36.3-year historical analysis and an arithmetic average of the hurricane loss and LAE ratios; a 36.3-year historical analysis and a premium-weighted average of the hurricane loss and LAE ratios. The four estimates were all based on 90 percent of the industry (excluding the Association) extended coverage experience and 100 percent of the Association´s experience. OPIC recommended using the 36.3-year historical analysis and arithmetic average of the hurricane loss and LAE ratios.

24. TEX. INS. CODE ANN. Art. 21.49 §8(h)(11) requires the use of not less than the most recent 30 years of 90 percent of the industry (excluding the Association) extended coverage experience and 100 percent of the Association´s experience to project the provision for hurricane losses in the Association rates.

25. In developing its various 32.3-year historical analyses, the Association used 10 years (1970-1979) of industry-wide experience for the five first and second tier county territories combined and 22 years (1980-2001) of industry-wide experience for each of the three first tier county territories and for the combined two second tier county territories, weighted to reflect the Association´s distribution of business by territory.

26. OPIC used 10 years (1970-1979) and 14 years (1966-1979) of industry-wide experience for the five first and second tier county territories combined and 22 years (1980-2001) of industry-wide experience for each of the three first tier county territories and for the combined two second tier county territories, weighted to reflect the Association´s distribution of business by territory.

27. It is necessary to use combined experience of the first and second tier county territories for the years prior to 1980 because data at the territorial level for that time period is not readily available.

28. Industry-wide experience for each of the three first tier county territories and for the combined two second tier county territories is available separately for 1980 and subsequent years.

29. Using an average of the territorial loss ratios for each year weighted by the Association´s current distribution of business, as proposed by the Association and adopted by OPIC, produces a more accurate measurement of the Association's catastrophe exposure than using all five coastal rating territories combined, which reflect the industry-wide distribution of business.

30. An average of the territorial loss ratios weighted by the Association´s current distribution of business is reasonable for those years for which the necessary data is available.

31. The Association calculated its hurricane loss ratio for the most recent 32.3 year period by taking an arithmetic average of the loss ratios for years in which there were hurricanes, subtracting the estimated average non-hurricane loss ratio, and spreading the result over 32.3 years by multiplying by the quotient of the number of years in which there were hurricanes and 32.3. The estimated average non-hurricane loss ratio is an average of the average loss ratio of all non-hurricane years and the average loss ratio of all non-hurricane years excluding year 1991.

32. OPIC calculated the hurricane loss ratio for the most recent 36.3 year period by taking an arithmetic average of the loss ratios for years in which there were hurricanes, subtracting the estimated average non-hurricane loss ratio, and spreading the result over 36.3 years by multiplying by the quotient of the number of years in which there were hurricanes and 36.3. The estimated average non-hurricane loss ratio is an average of the loss ratios of all non-hurricane years.

33. The simple arithmetic weighting methodology described in Finding Nos. 31 and 32 is reasonable.

34. Based on Finding Nos. 22-33, the indicated hurricane loss ratio based on the 32.3-year historical analysis proposed by the Association is 0.393 and the indicated hurricane loss ratio based on the 36.3-year historical analysis proposed by OPIC is 0.364.

35. The hurricane loss ratio of 0.393 described in Finding No. 34 is reasonable.

36. The Association proposed to adjust the hurricane loss ratio described in Finding No. 35 due to the lower than normal hurricane frequency in the 32.3-year period in two ways: adding one more average hurricane to the 32.3 year period; adjusting the 32.3 year loss ratio for the difference in the actual 103-year hurricane frequency (0.398) to the hurricane frequency (0.248) of the 32.3-year period.

37. It is reasonable to add one more average hurricane to the 32.3 year period described on Finding No. 36.

38. Based on Finding Nos. 35-37, the indicated hurricane loss ratio is 0.442.

39. A factor of 1.092 applied to projected hurricane losses to provide for associated LAE, as proposed by the Association and adopted by OPIC, is reasonable.

40. Based on Finding Nos. 38 and 39, the indicated hurricane loss and associated LAE ratio is 0.483 for commercial property.

41. The hurricane loss and associated LAE ratio described in Finding No. 40 as the Association´s provision for hurricane losses and associated LAE is reasonable and should be used to calculate the Association´s commercial rates.

42. The Association proposed the following expense provisions: 15.8 percent for commissions and brokerage; 6.3 percent for general expense; 1.9 percent for premium taxes, licenses, and fees; 22.0 percent for the net cost of reinsurance; and 10.0 percent for contribution to the catastrophe reserve trust fund (trust fund).

43. OPIC adopted expense provisions of 15.8 percent for commissions and brokerage, 6.3 percent for general expense, and 1.9 percent for premium taxes, licenses, and fees.

44. It is reasonable to use expense provisions of 15.8 percent for commissions and brokerage, 6.3 percent for general expense, and 1.9 percent for premium taxes, licenses, and fees.

45. The Association and OPIC considered general expenses as fixed expenses.

46. General expenses include expenses for activities such as the cost of policy issuance and are not directly related to losses.

47. It is reasonable to consider general expenses as fixed expenses .

48. In 2003 the Association purchased reinsurance protection to cover, in part, catastrophe claims payments in excess of 200 million dollars, with the first 100 million dollars covered by assessment of its members and the second 100 million dollars covered by the trust fund.

49. The Association included in its rate filing a fixed expense provision of 22.0 percent to reflect the estimated reinsurer underwriting costs, brokerage fees incurred in conjunction with the reinsurance contract, and the reinsurer profit; and a 10.0 percent variable expense provision for its contribution to the trust fund.

50. OPIC recommended a variable provision of 15.0 percent to cover the net cost of reinsurance and the contribution to the trust fund, prior to any consideration of investment income.

51. OPIC recommended reducing the provision for the net cost of reinsurance and the contribution to the trust fund by 4.0 percentage points to account for the Association´s investment income.

52. It is well accepted that investment income be taken into account in setting insurance rates for coverage written by for-profit insurers.

53. The Association is not a for-profit insurer.

54. The Association´s investment income is available to increase the amount in the trust fund.

55. It is desirable that the Association build up the trust fund so that the funds will be available to pay claims from future major catastrophes.

56. It is not reasonable to reduce the Association´s rates to account for its investment income.

57. The net cost of reinsurance of 22% was paid when the treaty was placed and is not directly related to losses.

58. It is reasonable to consider the net cost of reinsurance as a fixed expense .

59. It is reasonable to use 22.0 percent for the net cost of reinsurance and 5.0 percent for contribution to the trust fund.

Commercial Property Rates

60. Based on Finding Nos. 21, 41, 44, 47, 58, and 59, the indicated rate change for commercial risks and classes of commercial risks is +19.0 percent.

61. The Association´s existing rates for commercial risks and classes of commercial risks should be increased by the statutory maximum rate level change of ten percent.

RESIDENTIAL PROPERTY

62. Both the Association and OPIC used ten years (1993 through 2002) of 90 percent of the industry (excluding the Association) extended coverage experience and 100 percent of the Association´s experience for each of the three first tier county territories and for the combined two second tier county territories, valued as of December 31, 2002, to project the indicated loss and LAE ratios for each of the territory groups. The component of the rates intended to cover non-hurricane losses and LAE is the weighted average of the territory loss and associated LAE ratios using the Association´s 2002 written premium distribution by territory.

63. TEX. INS. CODE ANN. Art. 21.49 §8(h)(12) requires the use of the ten years of 90 percent of the industry (excluding the Association) extended coverage experience and 100 percent of the Association´s experience described in Finding No. 62 to project the provision for non -hurricane losses in the Association rates.

64. The Association selected its paid loss development factors from arithmetic averages of the loss development indications of between nine years of data (the 15-27 months´ indication) and two years of data (the 99 to 111 months´ indication).

65. OPIC selected its paid loss development factors from arithmetic averages of the loss development indications, excluding the highest and the lowest loss development, of between ten years of data (the 15-27 months´ indication) and three years of data (the 99 to 111 months´ indication).

66. Either approach in Finding Nos. 64 and 65 would be reasonable for calculating the ultimate loss development factors.

67. The selected ultimate loss development factors by the Association and OPIC were the same except the 15 months to ultimate factor, which is 1.098 selected by the Association and 1.096 selected by OPIC.

68. It is reasonable to use the loss development factors proposed by the Association.

69. The Association and OPIC adjusted the premiums earned during each experience year to current rate levels by the application of current rate level factors.

70. The Association calculated loss ratio trend factors by dividing the product of historical and prospective loss trends by the product of historical and prospective premium trends .

71. OPIC adopted the Association´s loss ratio trend factors for its calculations.

72. The Association's loss ratio trend factors are reasonable.

73. The Association calculated a factor of 1.260 to include non-hurricane associated LAE based on the experience for non-hurricane years in the 27-year period from 1976 through 2002.

74. OPIC adopted the Association´s LAE factors for its calculations

75. It is reasonable to apply a factor of 1.260, as recommended by the Association and adopted by OPIC, to non-hurricane losses in order to include associated LAE.

76. The Association and OPIC differed in their calculations of the 10-year average by-territory loss ratios in that the Association used a premium-weighted average while OPIC used an arithmetic average.

77. The resulting Association´s average loss and LAE ratios for all territories combined was 0.207 while OPIC´s loss and LAE ratio was 0.192.

78. Either averaging approach described in Finding No. 76 would be reasonable given the limited number of years used to project non-hurricane losses and LAE.

79. Based on Finding Nos. 62 - 78, a non-hurricane loss and associated LAE ratio of 0.207 is reasonable and should be used to calculate the Association´s residential property rates.

Hurricane Losses and Associated LAE

80. The Association utilized six different estimates to project its rate needs to cover hurricane losses and associated LAE: a 39-year historical analysis; a 39-year plus one hurricane historical analysis; a 39-year adjusted to the actual 103-year frequency level historical analysis; a computer model prepared by Applied Insurance Research (AIR); a computer model prepared by EQECAT; and the Risk Management Solutions (RMS) computer model. The first three estimates were based on 90 percent of the industry (excluding the Association) extended coverage experience and 100 percent of the Association´s experience. The Association averaged the last three estimates to arrive at a single indication.

81. OPIC utilized two different estimates to project its rate needs to cover hurricane losses and associated LAE: a 39-year historical analysis and an arithmetic average for hurricane loss and LAE ratios; and a 39-year historical analysis and premium-weighted average for hurricane loss and LAE ratios. The two estimates were all based on 90 percent of the industry (excluding the Association) extended coverage experience and 100 percent of the Association´s experience . OPIC recommended using the 39-year historical analysis and arithmetic average for hurricane loss and LAE ratios

82. TEX. INS. CODE ANN. Art. 21.49 §8(h)(11) requires the use of not less than the most recent 30 years of 90 percent of the industry (excluding the Association) extended coverage experience and 100 percent of the Association´s experience to project the provision for hurricane losses in the Association rates.

83. In developing the various 39-year historical analyses described in Finding Nos. 80 and 81, the Association and OPIC used 16 years (1964-1979) of industry-wide experience for the five first and second tier county territories combined and 23 years (1980-2002) of industry-wide experience for each of the three first tier county territories and for the combined two second tier county territories, weighted to reflect the Association´s distribution of business by territory.

84. It is necessary to use combined experience of the first and second tier county territories for the years prior to 1980 because data at the territorial level for that time period is not readily available.

85. Industry-wide experience for each of the three first tier county territories and for the combined two second tier county territories is available separately for 1980 and subsequent years.

86. Using an average of the territorial loss ratios for each year weighted by the Association´s current distribution of business, as proposed by the Association and adopted by OPIC, produces a more accurate measurement of the Association's catastrophe exposure than using all five coastal rating territories combined, which reflect the industry-wide distribution of business.

87. An average of the territorial loss ratios weighted by the Association´s current distribution of business is reasonable for those years for which the necessary data is available.

88. The Association calculated its hurricane loss ratio for the most recent 39 year period by taking an arithmetic average of the loss ratios for years in which there were hurricanes, subtracting the estimated average non-hurricane loss ratio, and spreading the result over 39 years by multiplying by the quotient of the number of years in which there were hurricanes and 39. The estimated average non-hurricane loss ratio is an average of the average loss ratio of all non-hurricane years and the average loss ratio of all non-hurricane years excluding year 1991.

89. OPIC calculated the hurricane loss ratio for the most recent 39 year period by taking an arithmetic average of the loss ratios for years in which there were hurricanes, subtracting the estimated average non-hurricane loss ratio, and spreading the result over 39 years by multiplying by the quotient of the number of years in which there were hurricanes and 39. The estimated average non-hurricane loss ratio is an average of the loss ratios of all non-hurricane years.

90. The simple arithmetic weighting methodology described in Finding Nos. 88 and 89 is reasonable.

91. Based on Finding Nos. 80-90, the Association´s indicated hurricane loss ratio is 0.295 while OPIC´s indicated hurricane loss ratio is 0.291.

92. The hurricane loss ratio of 0.295 described in Finding No. 91 is reasonable.

93. The Association proposed to adjust the hurricane loss ratio described in Finding No. 92 due to the lower than normal hurricane frequency in the 39-year period in two ways: adding one more average hurricane to the 39 year period; adjusting the 39 year loss ratio for the difference in the actual 103-year hurricane frequency (0.398) to the hurricane frequency (0.231) of the 39-year period.

94. It is reasonable to add one more average hurricane to the 39 year period described on Finding No. 93.

95. Based on Finding Nos. 92-94, the indicated hurricane loss ratio is 0.328.

96. A factor of 1.092 applied to projected hurricane losses to provide for associated LAE, as proposed by the Association and adopted by OPIC, is reasonable.

97. Based on Finding Nos. 95 and 96, the indicated hurricane loss and associated LAE ratio based on the 39-year historical analysis is 0.358 for residential property.

98. The hurricane loss and associated LAE ratio described in Finding No. 97 as the Association´s provision for hurricane losses and associated LAE is reasonable and should be used to calculate the Association´s residential rates.

99. The Association proposed the following expense provisions: 15.8 percent for commissions and brokerage; 6.3 percent for general expense; 1.9 percent for premium taxes, licenses, and fees; 22.0 percent for the net cost of reinsurance; and 10.0 percent for contribution to the catastrophe reserve trust fund (trust fund).

100. OPIC adopted expense provisions of 15.8 percent for commissions and brokerage, 6.3 percent for general expense, and 1.9 percent for premium taxes, licenses, and fees.

101. It is reasonable to use expense provisions of 15.8 percent for commissions and brokerage, 6.3 percent for general expense, and 1.9 percent for premium taxes, licenses, and fees.

102. The Association and OPIC considered general expenses as fixed expenses.

103. General expenses include expenses for activities such as the cost of policy issuance and are not directly related to losses.

104. It is reasonable to consider general expenses as fixed expenses .

105. In 2003 the Association purchased reinsurance protection to cover, in part, catastrophe claims payments in excess of 200 million dollars, with the first 100 million dollars covered by assessment of its members and the second 100 million dollars covered by the trust fund.

106. The Association included in its rate filing a fixed expense provision of 22.0 percent to reflect the estimated reinsurer underwriting costs, brokerage fees incurred in conjunction with the reinsurance contract, and the reinsurer profit; and a 10.0 percent variable expense provision for its contribution to the trust fund.

107. OPIC recommended a variable provision of 15.0 percent to cover the net cost of reinsurance and the contribution to the trust fund, prior to any consideration of investment income.

108. OPIC recommended reducing the provision for the net cost of reinsurance and the contribution to the trust fund by 4.0 percentage points to account for the Association´s investment income.

109. It is well accepted that investment income be taken into account in setting insurance rates for coverage written by for-profit insurers.

110. The Association is not a for-profit insurer.

111. The Association´s investment income is available to increase the amount in the trust fund.

112. It is desirable that the Association build up the trust fund so that the funds will be available to pay claims from future major catastrophes.

113. It is not reasonable to reduce the Association´s rates to account for its investment income.

114. The net cost of reinsurance of 22% was paid when the treaty was placed and is not directly related to losses.

115. It is reasonable to consider the net cost of reinsurance as a fixed expense .

116. It is reasonable to use 22.0 percent for the net cost of reinsurance and 5.0 percent for contribution to the trust fund.

Residential Rates

117. Based on Finding Nos. 79, 98, 101, 104, 115, and 116, the indicated rate change for residential risks and classes of residential risks is +9.6 percent.

118. The Association´s existing rates for residential risks and classes of residential risks should be increased by 9.6 percent.

Conclusions of Law

1. The Commissioner of Insurance has jurisdiction over this matter pursuant to TEX. INS. CODE ANN. Art. 21.49.

2. TEX. INS. CODE ANN. Art. 21.49 §8(h)(6) provides that the Commissioner shall approve, disapprove, or modify the Association´s rate filings in writing on or before November 15 of the year in which the filings are made, or the filings are deemed approved.

3. TEX. INS. CODE ANN. Art. 21.49 §8(h)(9) limits an average rate change to ten percent (higher or lower) of the rate for commercial and residential windstorm and hail insurance in effect on the date the filing was made.

4. TEX. INS. CODE ANN. Art. 21.49 §8(h)(9) provides that a rate may not reflect a rate change for an individual rating class that is 15 percent higher or lower than the rate for that individual class in effect on the date the filing was made.

5. TEX. INS. CODE ANN. Art. 21.49 §8(h)(11) requires that the catastrophe element used to develop rates applicable to risks written by the Association is uniform throughout the seacoast territory. The catastrophe element must be developed using: (a) 90 percent of both the monoline extended coverage loss experience and related premium income for all insurers, other than the Association, for covered property located in the seacoast territory, using not less than the most recent 30 years of experience available; and (b) 100 percent of both the loss experience and related premium income for the Association for covered property using not less than the most recent 30 years of experience available.

6. TEX. INS. CODE ANN. Art. 21.49 §8(h)(12) requires that the noncatastrophe element used to develop the Association´s residential property rates must be developed using: (a) 90 percent of both the monoline extended coverage loss experience and related premium income for all insurers, other than the Association, for covered property located in the catastrophe area of the seacoast territory using the most recent 10 years of experience available; and (b) 100 percent of both the loss experience and related premium income for the Association for covered property using the most recent 10 years of experience available.

7. TEX. INS. CODE ANN. Art. 21.49 §8(h)(13) requires that the noncatastrophe element used to develop the Association´s commercial rates must be developed using 100 percent of both the loss experience and related premium income for the Association for covered property using the most recent 10 years of experience available.

IT IS THEREFORE THE ORDER of the Commissioner of Insurance that the above findings of fact and conclusions of law be adopted.

IT IS FURTHER ORDERED that the manual rate filing request for 2004 commercial rates by the Association be approved and the manual rate filing request for 2004 residential rates be modified consistent with the findings and conclusions set forth in this order.

IT IS FURTHER ORDERED that the manual rates to be charged by the Association for insurance for commercial risks and classes of commercial risks written by the Association in accordance with TEX. INS. CODE ANN. Art. 21.49 be increased by ten percent from its 2003 rate level.

IT IS FURTHER ORDERED that the manual rates to be charged by the Association for insurance for residential risks and classes of residential risks written by the Association in accordance with TEX. INS. CODE ANN. Art. 21.49 be increased by 9.6 percent from its 2003 rate level.

IT IS FURTHER ORDERED that the Association's rates become effective January 1, 2004.