STATE OF FLORIDA

DIVISION OF ADMINISTRATIVE HEARINGS

AIU INSURANCE COMPANY, AMERICAN ) HOME ASSURANCE COMPANY, )

BIRMINGHAM FIRE INSURANCE )

COMPANY OF PENNSYLVANIA, ) COMMERCE AND INDUSTRY INSURANCE ) COMPANY, GRANITE STATE ) INSURANCE COMPANY, ET AL., )

)

Petitioners, )

)

vs. )

)

DEPARTMENT OF FINANCIAL ) SERVICES, OFFICE OF INSURANCE ) REGULATION, )

)

Respondent. )

Case No. 03-4477

)

RECOMMENDED ORDER

A disputed-fact hearing was held February 10-13, and 17-18, 2004, in Tallahassee, Florida, before Ella Jane P. Davis, a

duly-assigned Administrative Law Judge of the Division of Administrative Hearings.

APPEARANCES

For Petitioners: Daniel C. Brown, Esquire

Kelly A. Cruz-Brown, Esquire Robert W. Pass, Esquire Carlton Fields, P.A.

Post Office Box 190

Tallahassee, Florida 32302-0190

For Respondent: Elenita Gomez, Esquire

Dennis Silverman, Esquire Michael H. Davidson, Esquire Steven Parton, Esquire Department of Financial Services

Division of Legal Services 612 Larson Building, Suite 612

200 East Gaines Street Tallahassee, Florida 32399-4206

STATEMENT OF THE ISSUES

1. Whether Respondent's proposed disapproval of Petitioners' burglary-theft rate filing (OIR File No. 03-11518) for insurance coverage under the Terrorism Risk Insurance Act of 2002, is based on an invalid agency statement of general applicability which has not been adopted as a rule pursuant to Section 120.54, Florida Statutes.

2. Whether Petitioners' burglary-theft rate filing (OIR File No. 03-11518) for insurance coverage under the Terrorism Risk Insurance Act of 2002 meets all procedural and substantive requirements so as to be approved.

PRELIMINARY STATEMENT

On September 11, 2001, international terrorists destroyed the privately-owned twin towers of the World Trade Center in New York City, New York; a wing of the governmentally-owned Pentagon in Washington, D.C.; and four commercial passenger airplanes, resulting in unprecedented loss of life and property within the United States of America.

The Federal Terrorism Risk Insurance Act of 2002 (TRIA) was enacted in November 2002. It required insurers, such as Petitioners, to provide terrorism coverage to their customers and to withdraw any exclusions that had been previously issued with regard to such coverage. TRIA reduced the exposure of terrorism loss to insurers post-September 11, 2001, by capping losses as indicated in the Act. TRIA did not require insurers to request State rate increases or make rate filings for providing such terrorism coverage, which insurers had been offering before September 11, 2001. Nonetheless, it was axiomatic that responsible insurers would make new rate filings if they deemed it necessary as the result of the mandatory TRIA coverage.

This case arises from a TRIA burglary-theft rate filing by Petitioners, which was disapproved by Respondent, Florida's Office of Insurance Regulation (OIR or Agency).

Petitioners (collectively, "AIG") submitted a "use and file" rate filing to OIR. On October 14, 2003, the Agency issued a Notice of Intent to Disapprove Petitioners' filing because it was deemed "excessive." On October 31, 2003, Petitioners timely requested a disputed-fact hearing on the merits of the filing, pursuant to Section 120.57(1), Florida Statutes.

The case was referred to the Division of Administrative Hearings on or about November 26, 2003, and was assigned the instant DOAH Case No. 03-4477. From the beginning, it has been a component of this case that AIG has asserted, pursuant to Section 120.57(1)(e), Florida Statutes, that "the real reason" the Agency rejected AIG's burglary-theft rate filing was the Agency's adherence to an alleged non-rule policy to cap all TRIA rate filings at the one percent rate proposed by a non- governmental insurance group, Insurance Services Offices (ISO).

However, on or about December 2, 2003, the same Petitioners filed a formal (non-) rule challenge to the alleged one percent cap, pursuant to Section 120.54, Florida Statutes. That case was assigned DOAH Case No. 03-4486RU.

DOAH Case No. 03-4486RU was consolidated with the instant case by an Order entered herein on December 17, 2003. Until February 4, 2004, discovery, scheduling, and continuances proceeded in the cases as consolidated under the lower case number. On that date, pursuant to Section 120.56(4)(e)2., Florida Statutes, the Agency published a Notice of Rule Development, addressing the subject area of terrorism insurance adjustments and rates, and moved to stay DOAH Case No. 03- 4468RU, the "rule case/final order case." That case was abated by an Order entered February 6, 2004.

The instant "disputed-fact case/recommended order case" proceeded to final hearing February 10-13, and 17-18, 2004. All motions, proffers, and rulings are clear on the record, either by written order or within the Transcript of Proceedings, and will not be reiterated here.

A component of this proceeding is the Agency's perception that AIG's filing, even after supplementation in response to a "standard letter" from the Agency, constituted an untimely and insufficient response to a deficiency letter.

The Agency also maintains that AIG's burglary-theft rate filing, including its response to the foregoing standardized deficiency letter, has never been supported by sufficient information to demonstrate that AIG's proposed rates are not "inadequate, excessive, or unfairly discriminatory" and that the filing is "without sufficient justification," and that for those two reasons, the Agency deemed AIG's rate filing "excessive."

Respondent's Motion in Limine at the commencement of the final hearing, to the effect that only the papers filed by AIG as constituting its rate filing could be considered in this de novo proceeding was denied for the reasons expressed within the Transcript.

At hearing, Petitioners AIG presented the oral testimony of Patricia Barrett, Richard L. Thomas, James Stephen

(Steve) Roddenberry, William Bodiford, Jack Swisher, and

David Shaffer. AIG also had depositions and/or excerpts of depositions of the following persons admitted in evidence: Sri Ramanujam (AIG Exhibit 4); Frank Douglas (AIG Exhibit 9);

Jason Simmons (AIG Exhibit 10); Shirley Kerns (AIG Exhibit 11); and William Cotter (AIG Exhibit 12). In all, AIG Exhibits 1-26 were admitted in evidence.

Respondent OIR presented the oral testimony of Shirley Kerns, Steve Roddenberry, Verne Ivarson,

William Bodiford, Jack Swisher, and John Robert Hunter. The Agency also had depositions and/or excerpts of depositions of the following persons admitted in evidence: Sri Ramanujam (Agency Exhibits 30-31); Steve Lehman (Agency Exhibit 14);

Mary Gaillard (Agency Exhibit 22); Jason Simmons (Agency Exhibit 23); William Cotter (Agency Exhibits 28-29); Adam C. Reed (Agency Exhibit 21); Brian O'Neill (Agency Exhibit 20);

Frank Douglas (Agency Exhibit 19); William J. Moller (Agency Exhibit 24); James Prendergast (Agency Exhibit 25); Carl E. Chamberlain (Agency Exhibit 27); Augustine Igwe (Agency Exhibit 26); and Richard L. Thomas (Agency Exhibits 15-16). The Table of Contents and Index of the Transcript herein are confusing, but the Transcript itself shows that Respondent OIR had admitted in evidence Agency Exhibits 1, 3-4, 6-12, 14A and 14B-16, and 18-32. Agency Exhibit 13 was marked as such but not admitted separately, and is part of Agency Exhibit 14A. The Agency's

Exhibit 14 is divided into 14A and 14B. Agency 18 was admitted at page 822 of the Transcript. An Agency proffer appears at pages 760-763 of the Transcript.

Some of the deposition exhibits include multiple volumes of testimony and/or of identified exhibits.

Official recognition was taken of various items as evidenced by either the Joint Pre-hearing Stipulation, written Orders, or oral rulings on the record.

The seven volumes of Transcript were filed on February 25, 2004.

Each party timely-filed a Proposed Recommended Order on March 22, 2004. All written objections to questions internal to depositions or excerpts of depositions which were admitted at the formal hearing are hereby overruled. The answers to those questions have been considered and weighed with the rest of the evidence.

On April 27, 2004, and repeatedly thereafter, the parties stipulated to hold this case and DOAH Case No. 03-4486RU in abeyance. The record reflects repeated Orders approving the parties' successive stipulations.

An Order on all pending motions was entered on December 3, 2004, which, among other rulings, set an aspirational date for this Recommended Order. Both Proposed Recommended Orders have been considered in preparation of this Recommended Order.

FINDINGS OF FACT

1. The Terrorism Risk Insurance Act of 2002, Pub. L. 101- 297, Nov. 26, 2002, 116 Stat. 2322, a/k/a TRIA, became law in November 2002.

2. Congress found in TRIA that:

   
   

   Widespread financial market uncertainties have arisen following the terrorist attacks of September 11, 2001, including the absence of information from which financial institutions can make statistically valid estimates of the probability and cost of future terrorist events, and therefore the size, funding, and allocation of the risk of loss caused by such acts of terrorism.

   
   

3. TRIA's purpose and objective, see Section 101(b)(2), is to:

   . . . allow for a transitional period for the private markets to stabilize, resume pricing of such insurance, and build capacity to absorb any future losses.

   
   

4. TRIA requires insurance companies to separately price and offer insurance coverage for certified acts of terrorism as defined in TRIA in most commercial lines of property and casualty (P&C) insurance. Burglary and theft insurance policies fall in this P&C category. See TRIA §§ 102, 103(a) (3), (c).

5. TRIA provides that the federal government will reimburse an insurance company for 90 percent of the amount of that company's insured losses resulting from certified acts of terrorism that exceed the insurer's TRIA-defined deductible in a

   given year. TRIA defines insurers' deductibles as seven percent of the previous year's direct earned premium in TRIA Program Year One (2003), 10 percent of the previous year's direct earned premium in TRIA Program Year Two (2004), and 15 percent of the previous year's direct earned premium in TRIA Program Year Three (2005). See TRIA §§ 102(6), (7), (9)-(11), and 103(e)(1)(A).

   Absent some further Act of Congress, TRIA's assistance to insurers ends with 2005.

6. Therefore, in the event of future certified foreign- sponsored terrorist attacks in the United States, each insurer will be required by TRIA to absorb and pay, without federal reimbursement, all losses equal to, or less than, their respective TRIA deductibles in any given TRIA year and ten percent of all losses exceeding their respective TRIA deductibles, until such time as total industry losses reach

   $100,000,000 (100 billion dollars) in any given year.

   
   

7. At § 106, TRIA addresses state laws as follows:

   
   

   Nothing in this title shall affect the jurisdiction or regulatory authority of the insurance commissioner (or any agency or office performing like functions) of any State over any insurer or other person --

   
   

   1. except as specifically provided in this title; and

      
      

   2. except that--

      
      

      * * *

      (B) during the period beginning on the date of enactment of this Act and ending on December 31, 2003, rates and forms for terrorism risk insurance covered by this title and filed with any State shall not be subject to prior approval or a waiting period under any law of a State that would otherwise be applicable, except that nothing in this title affects the ability of any State to invalidate a rate as excessive, inadequate, or unfairly discriminatory . . . [.]

      
      

8. AIG traditionally writes insurance policies for large commercial risks with high limits of coverage. Many witnesses, including several actuaries, opined that such insureds may be potential terrorist targets. However, only 8.8 percent of commercial lines insurance in the United States is written by AIG.

9. Insurance companies, like AIG, are in the business of transferring their clients' risks to themselves, but in order for the system to work, as well as for any insurance company to prosper and continue to be able to transfer such risks so as to protect its insureds, the risk assumed by the insurance company must be an intelligent risk. When an insurance company sets a rate, its goal is to accumulate enough capital from premiums and associated investment income to pay predicted losses from the eventuality insured against.

10. TRIA mandated that AIG and other insurers assume a new, specific risk. As a result of TRIA's mandated risk, AIG

    made rate filings in all 50 states and the District of Columbia, including burglary-theft filings.

11. In Florida, OIR is charged with the review and approval or disapproval of all insurance rate filings.

12. At no time has either party herein contended that September 11, 2001, did not change the perspective of governmental insurance regulators and of members of the insurance industry with regard to underwriting insurers' risk exposure to terrorism on United States' soil.

13. In one form or another, every witness in this case has opined that historical loss data, the bedrock of traditional actuarial predictions, does not exist for insurers to use in making rates for insurance coverage under TRIA. No less an authority than the American Academy of Actuaries has determined that " . . . lack of information has precluded the use of traditional ratemaking methodology." Expert actuarial witnesses and expert underwriting witnesses differ within their respective disciplines and among themselves about whether there were preceding minor terrorist events that could have foretold the two assaults on the World Trade Center (WTC) in New York City in 1993 and 2001, but the only two events that credible testimony, expert or lay, were able to pinpoint as necessary to be considered for forecasting future foreign terrorist activities within the continental United States are the 1993 bombing of the

    WTC, and the unprecedented loss of life and property that occurred on September 11, 2001, when international terrorists destroyed the WTC, a wing of the Pentagon, and four airplanes.

14. Some witnesses anecdotally opined that the presence of Governor Jeb Bush, the President's brother, in Tallahassee; Disney World and similar theme-based tourist attractions in the Orlando area; the Tampa site of The Southern Command for the United States' military efforts in Iraq and Afghanistan; and Florida's three active nuclear reactors "could" constitute potential terrorist targets in Florida. Other witnesses disagreed. No witness was authoritatively credible as to whether or not these entities raise or lower Florida's susceptibility to terrorist attack within the time frame that TRIA currently is designed to operate (the end of 2005). Some witnesses, some modeling explanations, and some rate filings expressed the belief that geographic concentration of people and/or geographic concentration of commercial buildings made a difference. Others did not. There was general agreement among the witnesses that reasonable persons could disagree on what entities and locations will be future targets. There were also disagreements among the witnesses on the severity and number of attacks that may occur. No witness could say with any reasonable certainty what type of weapon(s) future terrorists might use (i.e., conventional small arms, automatic weapons, car

    bombs, nuclear devices, bio-chemical hazards, chemical release, etc.) against these, or any other, potential targets. All of these elements affect prediction of the severity and frequency of terrorist attacks, and all of these elements suggest different levels and time periods for terrorists' planning of the next, or many future, attacks.

15. Therefore, all sources confess that ratemaking for insurance coverage under TRIA must rely on informed insurance practitioners, actuaries, lawyers, accountants, security and terrorism experts, and overall, upon informed, reasonable underwriting judgment applied to the opinions of terrorism experts, among others.

16. Each witness who was asked the question, considered it reasonable to rely, in part, on "terrorism experts" to forecast future targets and the frequency and severity of TRIA-covered losses. However, there was no consensus as to who constituted a "terrorism expert" or what might be the qualifications for that designation. The most agreed upon was that governmental or military service would be logical for qualifying a "terrorism expert" and that current or past Central Intelligence Agency (CIA) or Federal Bureau of Investigation (FBI) agents were desirable terrorism experts.

17. OIR employs no terrorism experts. Neither does AIG. No terrorism expert testified at the hearing.

18. Several modeling (simulation) schemes exist for predicting terrorism activity, but all are in their infancy in comparison to other modeling techniques for natural catastrophes, as explained in greater detail, infra.

19. ISO is an insurance rating organization, of which many commercial insurance companies are members, and which is authorized to make advisory rate filings. If the Agency approves an ISO filing, ISO's member insurance companies, and non-member insurance companies alike, need not make their own rate filings. They may simply adopt the rate OIR has approved for ISO. ISO members and non-members also are not obligated by law to adopt the ISO rate. They retain the right to make their own rate filings. ISO TRIA rate filings caution that they are not appropriate for all insurers due to the disparate effects that TRIA has on various insurers.

20. AIG is a member of ISO.

    
    

21. On or about December 26, 2002, ISO made two TRIA rate filings in Florida: one for commercial property insurance and one for general liability insurance.

22. AIG did not make a burglary-theft rate filing in Florida until September 4, 2003.

23. In early 2003, when Florida's OIR was considering how to respond to the expected flood of rating organization and private insurance company TRIA rate filings, Shirley Kerns, OIR

    Bureau Chief, memorialized decisions that OIR's senior management had agreed upon by meetings and collegial conference,

    as follows:

    
    

    Insurers will need to make a filing

    documenting whatever rate level affect [sic] is applicable . . . . Depending upon their particular situation (rate adequacy at the time of filing) and related documentation thereof, up to 1% may be justified with appropriate documentation and support regarding the insurer's expected retained losses. Reliance on a model can be a factor in this exercise.

    
    

24. Ms. Kerns passed this information on to Sri Ramanujam, a senior, and highly accredited, OIR actuary, and to others in OIR's Bureau of Property and Casualty Forms and Rates. She believed Mr. Ramanujam to be, of all OIR's actuaries, the one who would best understand the Agency's position on TRIA rate filings.

25. Regardless of State statutes or Agency rules, according to Mr. Ramanujam and other OIR witnesses, a rate filing normally proceeds through OIR review more in the nature of a negotiation than a determination by the Agency of a base level of legal sufficiency. The first review is by an insurance or actuarial analyst. It appears that actuarial analysts were assigned to all TRIA rate filings. The analysts may, and frequently do, consult their respective supervising actuaries for directions during the first review. They also routinely

    request additional or clarifying data from the filer, either by telephone, e-mail, or formal letter, and a timely response is also reviewed as part of the rate filing. When a letter is used, the time frame for the filer's response is obviously clearer. When the analyst is comfortable with making a recommendation as to whether OIR should approve or disapprove a rate filing, he or she passes the file to the supervising actuary. If the actuary has concerns, she or he may request more information from the filer directly or instruct the analyst to do so, and timely responses are again considered. When the actuary is finally comfortable with making a recommendation, the actuary does so in writing, and then, at a "red book" meeting, wherein issues raised by the actuary are discussed among several actuaries and OIR's senior staff, Ms. Kerns "signs off" on either an approval or disapproval of the rate filing. A notice of intent to approve or disapprove is abbreviated as "NOI," and is sent to the filer by an actuary or analyst after Ms. Kerns has acted. There is no legal prohibition against amendments to a rate filing up to the date of a formal NOI letter to the filer.

26. Mr. Ramanujam was assigned to review the ISO commercial property filing, which, like AIG's later burglary- theft filing, is part of the (Property & Casualty) P&C category of insurance. ISO requested a one percent rate increase.

27. At all times material, Mr. Ramanujam found it difficult to meaningfully comment upon or to advise his superiors about, ISO's rate filing, because it was based on modeling (simulation) techniques proprietary to Applied Insurance Research (AIR). Mr. Ramanujam characterized this modeling system as a "black box," because he could not examine what was inside it.

28. AIR is a wholly-owned subsidiary of ISO. AIR and its simulation modeling technique, the Delphi Method, were developed by the Rand Corporation during World War II and have gained acceptance in academia and the commercial world because of their effective use for about 30 years to predict frequency and severity of natural disasters, such as hurricanes and earthquakes. Hurricanes and earthquakes provide several hundred years' worth of past natural catastrophe data as the threshold for AIR and other modeling companies to predict the frequency, severity, and location of future hurricanes and earthquakes. However, terrorism is not a natural catastrophe. Rather than the predictability of physics and climate, which drive most natural disasters, international terrorism prediction must assess human malice, planning, and cunning, within a broad spectrum of potential perpetrators, while factoring-in diverse motives, opportunities, access to targets, and fluctuating world politics.

29. According to Mr. Ramanujam's understanding of ISO's TRIA commercial property rate filing, the inputs to the simulated events AIR model were the subjective opinions of less than 10 unnamed "experts."

30. The Agency's out-sourced actuarial expert, John Robert Hunter, thought AIR's number of experts was even less, between five and seven.

31. Mr. Ramanujam understood that as the whole construct of the AIR-Delphi terrorism model was developed, these few experts weighed their own opinions independently at set stages within the simulation construct to determine how confident each expert was of his predictions, and then, a collective weighing was done somewhere toward the end of the project. The model's final predictions, made up of many component expert opinions from a very few experts, constitute the basis on which ISO made assumptions concerning the frequency of future terrorist events, the locations of such events, the means of terrorist attacks, the severity of the events, and similar components of ISO's forecast. The ISO filing described the AIR-Delphi modeling technique, but did not reveal the number of experts, who the experts were, what their areas of expertise were (although several witnesses, including Mr. Ramanujam, presumed all or some were "terrorism experts"), their qualifications, the substance

    of their component-part opinions, or where they disagreed or agreed as the model was finally developed.

32. Mr. Ramanujam asked ISO no questions concerning how it used the AIR-Delphi model to request a one percent rate increase, in part, because AIR-Delphi was a proprietary model and, during prior non-TRIA rate filings, ISO had refused to reveal how it used the method.

33. Although he did not have any solid evidence of the contents of the AIR-Delphi model for terrorism to which he could apply his actuarial skills, Mr. Ramanujam recommended approving the ISO TRIA commercial property rate filing at one percent. He did this, in part, because the elaborate description of the AIR- Delphi modeling method provided in ISO's multi-page explanatory memorandum just "seemed reasonable" to him as a actuary, and because Ms. Kerns' memorandum (see Finding of Fact 23) and actuarial literature recognized the use of modeling techniques, generally. On his own, Mr. Ramanujam would only have asked questions of ISO if its modeling technique had resulted in a much higher increase, like six percent.

34. Mr. Ramanujam assumed that ISO's commercial property filing drew its basic data exclusively from its insurance company members, which he considered reasonable. Also, he utilized a personal "approach" to ISO's commercial property filing, figures, and information, which "approach" he resolutely

    would not characterize as either a "generally accepted actuarial technique, standard, or principle."

35. No statute, rule, or witness characterized the AIR- Delphi system as an "actuarial technique." No statute, rule, or witness established a definition of "actuarial technique" for purposes of this case.

36. Mr. Ramanujam's "approach" calculated that only .6 to

    
    

    .7 percent was justified by the ISO commercial property filing, not the one percent rate increase requested. His "approach" resulted in an estimated $3.34 premium increase on a hypothetical $100,000 policy, which he also pronounced "reasonable." Mr. Ramanujam opined that in the absence of sufficient historical data on terrorism, a rate based on any of these three percentages ( .6, .7, or one percent) "might" be "reasonable," but he thought that the one percent rate proposed by ISO would help small insurers. Even with his assumption of the validity of the AIR-Delphi method, ISO's commercial property filing had not justified to Mr. Ramanujam a one percent increase, but he approved its one percent increase request anyway, without asking any questions. He testified that in the circumstances of approving a TRIA rate, he would only have asked further questions if ISO had requested a new rate above one percent.

37. Despite ISO having an actuarial black box effect which

    he could not pierce, Mr. Ramanujam took his favorable recommendation of the ISO TRIA commercial property rate filing to a red book meeting, where the filing was approved by the Bureau Chief, Ms. Kerns. ISO was notified thereafter by formal letter.

38. When OIR approved the ISO commercial property TRIA rate filing, OIR issued an April 23, 2003, Informational Memorandum 03-007M, announcing that OIR had approved ISO "form, rule, and rate filings" relating to coverage under TRIA at one percent. This memorandum did not disclose that the ISO TRIA rate filing for general liability insurance, which had been filed the same day as ISO's commercial property filing still remained pending before OIR.

39. ISO's TRIA general liability filing, which was based on the same AIR-Delphi modeling methodology as ISO's commercial property filing, had proposed a category of rates equal to, or less than, one percent of the underlying (non-TRIA) coverage of "average risks." However, ISO's general liability filing also had proposed a second category of TRIA rates, up to 1.6 percent of the underlying coverage, for a limited set of "above average" risks.

40. OIR keeps track of how long a rate filing has been pending and generally expects rate filings to be acted upon

    promptly. Prompt review and decision is facilitated by holding red book meetings every week or every other week.

41. ISO's TRIA general liability filing did not explain in writing the basis by which ISO allocated particular types of risks to the "average" and "above average" categories of premium.

42. Jack Swisher was the actuary assigned by OIR to the ISO general liability TRIA rate filing. That filing was based on the same type of AIR-Delphi terrorism model as ISO's commercial property filing, no parts of which were explained from the filing itself.

43. ISO's TRIA general liability rate filing was not immediately disapproved as incomplete or unsupported. The Agency sent no deficiency letter seeking any additional information or clarification, even as to ISO's two proposed premium categories. Mr. Swisher just phoned an ISO representative and asked about the two-part risk categorization.

44. Mr. Swisher completed his review of ISO's general liability filing in April 2003. However, despite red book meetings approximately every one-two weeks in the interim, OIR did not approve ISO's general liability filing until October 3, 2003, after a senior management meeting. That senior management meeting and October 3, 2003, had implications for AIG's burglary-theft filing as well. (See Finding of Fact 71.)

45. In April 2003, Mr. Ramanujam drafted, and Ms. Kerns approved, what OIR witnesses referred to as "the standard letter." The standard letter was a specialized deficiency letter to be used whenever OIR made further inquiries for TRIA filings in excess of one percent.

46. Deficiency letters actually sent to TRIA filers who asked for more than a one percent increase varied as to the supplemental materials requested by OIR from the respective filers, but substantially the following language was prescribed for use in every letter:

    To the extent that the commercial policies associated with the proposed rate change is an all risks policy[sic] with noted exclusions, it is the opinion of the Office of Insurance Regulation (OIR) that all currently approved rates or loss costs do include the "terrorism" risk whether intended or not. Therefore, a rate filing that contemplates a separate premium component over and above what is already approved is considered excessive.

    
    

    Also substantially the following paragraphs were used, regardless of whether the line of TRIA insurance involved P&C:

    . . . In this context, it may not be out of place to mention that the recent submission by ISO for the so called "covered" risks has called for loss costs of

    .001 for Building and .001 for Contents based on what they consider a reasonable model but with extensive data base. . . .

    
    

    In comparison, your submission contemplates an increase in premium by % [whatever percentage more than one percent

    had been requested in the filing] for [TRIA coverage in the given line of business], which in our considered opinion and informed judgment is deemed excessive and thus cannot be supported . . .(Bracketed material added for clarity)

    
    

47. The standard letter's one percent language was used consistently, and between April 2003 and August 2003, various actuarial analysts, actuaries, or other OIR employees were sometimes even more explicit than the standard deficiency letter in stating that OIR was not going to approve any TRIA rate filing in excess of one percent for any risk category.

48. For instance, on April 17, 2003, OIR's Mr. Bodiford wrote to Lynn Staubly, Ace American Insurance Company, concerning a general liability TRIA rate filing for more than one percent, stating:

    So far, no more than 1% has been approved for a terrorism factor. I suggest that you revise to no more than 1%.

    
    

49. On April 24, 2003, Mr. Ramanujam wrote to Terri Smith, concerning Universal Underwriters Insurance Company's TRIA rate filing, stating:

    Any charge in excess of 1% for the certified acts of terrorism will be deemed excessive and I will recommend to my senior management issuance of this office action known as "Notice of Intent to Disapprove" (NOI). It is likely that this office will consider favorably if you will revise the charge downwards to 1% and amend the manual pages accordingly. You may do so this via

    an e-mail attachment [including the revised manual pages].

    
    

50. On May 21, 2003, Mr. Bodiford wrote to Jason Simmons, who is employed by the Petitioners, concerning an AIG general liability TRIA rate filing for more than one percent, stating:

    Hopefully, the rate will be 1% or less. So far, that is all we have approved.

    
    

51. On May 22, 2003, Mr. Bodiford stated to the same insurers concerning the same rate filing:

    To make a long story short, the State of Florida has not approved any rate higher than 1% for certified acts of terrorism . .

    . I advise you to make a change in your rates.

    
    

52. On June 2, 2003, Mr. Bodiford wrote to Josh Struve, Chicago Insurance Company, concerning a general liability TRIA rate filing for more than one percent, stating:

    Revise to 1% or less if you want the filing approved.

    
    

53. On June 3, 2003, OIR's Ms. Cai wrote to

    
    

    Sharon Lawrence, Hartford Insurance Company, concerning a commercial multi-peril TRIA rate filing for more than one percent, stating:

    All terrorism charges above 1% total premium have been considered excessive and required to be revised.

    
    

54. On June 16, 2003, OIR's Ms. Arnold wrote to

    
    

    Kathy Salzsiener, Westport Insurance Corporation, concerning a

    commercial multi-peril TRIA rate filing for more than one percent, stating:

    The charge for terrorism appears to be excessive. The highest that has been approved is 1%.

    
    

55. On July 22, 2003, Mr. Bodiford wrote to Emilie Fetty, Chicago Insurance Company, concerning a professional liability TRIA rate filing for more than one percent, stating:

    You must revise your terrorism rate to no more than 1%.

    
    

56. On August 5, 2003, OIR's Mr. Baltodano wrote to Emilie Fetty, Chicago Insurance Company, concerning a general liability TRIA rate filing for more than one percent, stating:

    The Office has not approved any terrorism rate over 1%.

    
    

57. Most, if not all, of the deficiency letters contained language requesting more information in order to justify a rate above one percent.

58. OIR's senior staff's testimony that reviewing staff, all of whom were experienced in communicating with insurers, wrote the foregoing letters without direct authority to do so, is not credible.

59. On April 24, 2003, Mr. Ramanujam had noted the following in the TRIA rate file record for American Association Insurance Services (AAIS), an insurance rating organization similar to ISO:

    I spoke at length with Kim Ward and convinced her that any thing [sic] above 1% will be considered excessive notwithstanding all the elaborate explanations provided by them in response to our standard letter.

    
    

60. When reviewing the AAIS filing, Mr. Ramanujam knew AAIS took an "all industry approach" in assembling data for its filing. He considered that data gathering system to be as reasonable as ISO's selection of data only from its own member subscribers. (See Finding of Fact 34.) In fact, he considered all the assumptions listed in AAIS's calculations to be reasonable. However, he decided AAIS's rate request was unreasonable because it exceeded one percent. He also considered AAIS's filing unreasonable, per his "approach" (see Finding of Fact 34), which he forecast as a $6.68 increase on a

    $100,000 policy. He disbelieved the market value assigned by AAIS in its filing, but he never checked it or inquired about it as would have been OIR's standard operating procedure in reviewing rate filings if the Agency's one percent requirement had not been in effect for TRIA rate filings. Even though he could not say that $6.68 was unreasonable on the basis of AAIS's descriptive method, and even if all parts of AAIS's filing were reasonable, Mr. Ramanujam would have pronounced AAIS's filing "unreasonable," because of the percentage rate requested. He acknowledged that even a six to ten percent increase might not

    be excessive. However, at that point, he was only approving TRIA rate filings at one percent.

61. On September 4, 2003, AIG submitted its TRIA burglary- theft rate filing, which is at issue in this proceeding, through the internet, i.e. OIR's I-file system.

62. AIG's rate filing was submitted on a "use and file" basis, pursuant to Section 627.062, Florida Statutes, which means that AIG has been charging the disputed rate from September 4, 2003, to date. If not approved in this proceeding, OIR can adjust this effect, pursuant to that statute.

63. On September 4, 2003, AIG's rate filing consisted of only a cover letter and two pricing charts, in addition to the universal data letter, which is the form filled out electronically and filed by the I-file system.

64. The copy of the universal data letter for AIG's September 4, 2003, burglary-theft filing which Mr. Ramanujam reviewed failed to correctly mark basic information. The rate change request, rate indicated, earned premium volume, and number of policies, all appeared as zeroes. Although much of the record is devoted to this issue, and a variety of theories were proposed as to how the zeroes appeared, the two most likely explanations are that an AIG employee thought "zero" was the equivalent of "not-applicable" or Mr. Ramanujam saw an altered printed copy of the I-filing. Whatever the cause, these

    deficiencies amount to non-issues, because Mr. Ramanujam testified that some of the correct answers could be discerned from the rest of AIG's filing and none of the zeroes were sufficient for him to recommend disapproval. Moreover, no request for this information was included in either

    Mr. Ivarson's standard deficiency letter to AIG or was given as the Agency's reason for ultimate disapproval. (See Findings of Fact 68-69 and 84.)

65. When AIG submitted its TRIA burglary-theft rate filing which is at issue in the instant proceeding, the first of two pages showed judgmental ranges of one to three percent and two to five percent, but the second page, a "manual exceptions" page, converted AIG's underwriters' ranges to discrete categories of one percent, two percent, three percent, and four percent, because AIG understood that Florida Administrative Code Rule 4-170.005, precluded an insurer utilizing a judgmental range of rates. Like the ISO general liability filing which was ultimately approved (see Findings of Fact 39-43, and 71), AIG's burglary-theft filing did not explain how its percentages were determined or how "low" and "high" risks were determined.

66. Despite some waffling by Mr. Ramanujam in hindsight on whether or not AIG's four categories of one through four percent constitute illegal tiers or merely inconsistent statements between two pages of AIG's burglary-theft filing, the best

    construction of the evidence as a whole is that Mr. Ramanujam did not consider AIG's use of categories fatal at the time of the filing, but became confused at his deposition. Clearly, the deficiency letter (see Findings of Fact 68-69) and the NOI (see Finding of Fact 84) did not target the discrete categories as a fatal flaw, pursuant to Florida Administrative Code Rule 4-

    170.005 (now Rule 69O-170.005).

    
    

67. The September 4, 2003, AIG rate filing also failed to provide an "explanatory memorandum," as required by Florida Administrative Code Rule 4-170.013 (now Rule 69O-170.013). AIG deliberately chose not to submit such a memorandum. This deficiency clearly hampered the Agency's review and evaluation of AIG's filing.

68. Due to the favored one percent rate, and his initial belief that AIG's burglary-theft filing was deficient in other respects, Mr. Ramanujam instructed Verne Ivarson, the actuarial analyst assigned to the file, to send AIG a September 11, 2003, deficiency letter including the one percent language quoted at Finding of Fact 46, supra. That letter also requested information on how much AIG previously had in its rate for terrorism and how AIG derived that component; what AIG's assumptions were regarding frequency and severity of terrorism acts; and if AIG were building-in a terrorism rate and an

    insured asked for a terrorism exclusion how would AIG do that, and the amounts.

69. The September 11, 2003, deficiency letter did not require AIG to file an explanatory memorandum, an explanation of its rating categories, or correction/supplementation of the information marked with zeroes.

70. AIG requested extensions of time to respond to the deficiency letter, and two of these requests were memorialized. E-mails and memoranda memorializing phone negotiations, extensions of response time, and other conversations are not always preserved by OIR. Upon the evidence as a whole, it is found OIR granted AIG at least until October 8, 2003, and possibly longer, to respond with the information requested.

71. However, on October 3, 2003, before AIG's response time to the burglary-theft deficiency letter had run, Ms. Kerns signed-off on Mr. Ramanujam's recommended disapproval of AIG's filing. This occurred after a September 30, 2003, meeting involving Kevin McCarty, Director Office Insurance Regulation; Ms. Kerns; Steve Roddenberry, Deputy Insurance Director; Jack Swisher; and Steve Parton, OIR General Counsel, in which ISO's general liability filing at 1.6 percent was a subject of discussion. (See Prehearing Stipulation.) ISO's general liability filing also was approved at 1.6 percent on October 3, 2003. (See Findings of Fact 39-44.)

72. Except for ISO's general liability line of insurance, OIR has not approved any TRIA rate filing for lines of insurance where the rate exceeded one percent of the premium for underlying (non-TRIA) coverage, although a number of insurers filed such rates.

73. Until October 3, 2003, OIR did not approve any TRIA rate filing for general liability insurance where the rate exceeded one percent of the premium for underlying (non-TRIA) coverage.

74. Since October 3, 2003, OIR has not approved any TRIA rate filing for general liability insurance where the rate exceeded the rates indicated in the ISO TRIA general liability rate filing of 1 to 1.6 percent.

75. After October 3, 2003, The Chubb Group made a general liability TRIA rate filing, communicating to OIR that Chubb understood OIR to have a "1% capping rule" regarding TRIA rates, and proposing a TRIA general liability rate of one percent. Although OIR had approved the ISO general liability rate filing at 1.6 percent for a limited set of risks on October 3, 2003, OIR did not inform Chubb that ISO's TRIA general liability rate filing had been approved with a category above one percent. OIR merely accepted Chubb's one percent filing, thereby inherently acknowledging Agency policy to approve only at a one percent cap.

76. On July 23, 2003, Robert L. Thomas, Chief Underwriting Officer for AIG's Domestic Brokerage Group, and OIR's Ms. Kerns, Steve Roddenberry, and Mr. Watford, the OIR actuary assigned to AIG's then-pending TRIA workers' compensation rate filing, had met at the Agency's Tallahassee offices. At that time, their prime focus of discussion had been workers' compensation rates. The other attendees who testified were vague in what they understood to have been discussed or accomplished at this meeting, but Mr. Roddenberry's notes tend to support Mr. Thomas' clear recollection that, in addition to discussing AIG's workers' compensation rate filing, Mr. Thomas had apprised those present of the essential elements underlying his underwriting approach to AIG's burglary-theft filing at issue herein, including AIG's seeking to fund its deductible over approximately eight years, see infra. Mr. Ramanujam was never apprised of the details of Mr. Thomas' July 23, 2003, conversation with OIR senior staff, because he was not the actuary reviewing AIG's TRIA workers' compensation rate filing. Because the July 23, 2003, meeting occurred before AIG's burglary-theft rating was filed with OIR on September 4, 2003, the July meeting does not fall in the category of an amended or supplemental rate filing or constitute any of the negotiations between an insurer and the Agency, as described in Finding of Fact 25. Because it preceded Mr. Ivarson's September 11, 2003,

    deficiency letter for AIG's burglary-theft filing, the July meeting could not be a response to the deficiency letter, either. Apparently, there was a similar meeting in March 2003, which likewise is irrelevant to the instant case for the same reasons.

77. In September-October, 2003, Jason Simmons, AIG's team leader for its burglary-theft filing, believed he had longer than until October 8, 2003, to file AIG's response to OIR's September 11, 2003, deficiency letter. He filed AIG's response on October 10, 2003.

78. Mr. Ramanujam did not disapprove AIG's filing due to Mr. Simmons' possibly tardy response. He considered it anyway, despite Ms. Kerns' October 3, 2003, disapproval "sign off."

79. AIG's October 10, 2003, response to OIR's deficiency letter on the instant burglary-theft filing consisted of only a three-page letter from Mr. Simmons and an attached copy of the American Academy of Actuaries' (AAA's) "Report to the National Association of Insurance Commissioners Terrorism Insurance Implementation Working Group on Ratemaking Issues Related to the Terrorist Insurance Act," which publication, among other things, commented on the lack of historical data for forecasting losses under TRIA. (See Finding of Fact 13.)

80. Mr. Ramanujam, who is a member of AAA, regarded the AAA Report as important, informational, and a guide for

    actuaries. He did not consider it a principle, standard, or technique of actuarial practice. He did not consider it binding on the Agency. The greater weight of the other credible evidence is to the same effect.

81. Mr. Ramanujam's and other witnesses' views that

    
    

    Mr. Simmons' letter, with the accompanying AAA Report, could not constitute an "explanatory memorandum" under the rule is rejected, because of the latitude provided by the rule.

82. Mr. Simmons' letter, however, failed to provide the underlying assumptions for AIG's rate request. It alleged a "unique risk profile," without support. It referred to Principle 4 of the "Casualty Actuarial Society 2002 Yearbook", but stated at one point that AIG's objective was to fund a "reasonable portion" of its TRIA deductible, while later stating as the company's goal that ". . . we have evaluated our TRIA deductible in relation to the maximum dollars we can potentially collect before assessing our rate needs." At hearing, AIG attempted to reconcile these two sentences, but on their face, they clearly confused OIR's reviewers, who considered them contradictory.

83. Mr. Ramanujam did not understand AIG's September 4, 2003, filing, as supplemented on October 10, 2003, to be projecting an eight-year timeline (see infra.), but he testified that even six years might be reasonable. He also considered

    AIG's basing its rates on data gathered from AIG's own market share to be reasonable. He further testified that a rate increase of six to ten percent due to TRIA might not be unreasonable. Yet, he pronounced AIG's rate filing overall unreasonable on the basis of what was omitted from the filing and upon the one percent cap.

84. Mr. Ramanujam did not ask AIG for further information in light of the prior October 3, 2003, disapproval. (See Finding of Fact 71.) AIG was notified of the disapproval by an NOI letter dated October 14, 2003, giving as the Agency's sole

    reason for denying AIG's TRIA burglary-theft rate filing that the rate increase was "excessive."

85. Upon the whole of the evidence, including but not limited to Findings of Fact 34, 60, and 83, it is found that Mr. Ramanujam applied very different analyses to ISO's commercial property filing than he did to commercial insurers or to other rating organizations. Because he considered one percent of the premium for the underlying (non-TRIA) coverage to be the only significant factor in his recommendation for approval or disapproval of a rate filing, there was one, common standard. But for the one percent Agency cap, it could be said that Mr. Ramanujam applied fundamentally disparate standards to commercial insurers and other industry rating organizations than he had to ISO.

86. However, even considering a few variations in OIR's use of the one percent cap, the inclusion in most deficiency letters of requests for more information to justify a rate increase above one percent, senior staff's incredible testimony that OIR's rate review employees lacked authority to confine themselves to a one percent analysis, and the approval of a 1.6 percent TRIA general liability rate filing for ISO on or about October 3, 2003, the evidence is overwhelming that OIR had in place a statement of general applicability (a one percent TRIA rate cap), which described the procedure or practice requirements of the Agency, and which imposed upon all rate filers a requirement not specifically required by statute or existing rule. AIG's burglary-theft filing was improperly disapproved on the basis of this non-rule policy.

87. Accordingly, AIG's burglary-theft filing is entitled to a Section 627.062(2), Florida Statutes' review in this proceeding.

88. OIR contends that AIG's TRIA burglary-theft rate filing must be denied because it did not contain past and prospective loss experience within and without Florida; any considerations of investment income reasonably expected; the cost of reinsurance; and discussion of trend factors, all of which OIR is to consider under the statute. OIR also complains

    that AIG did not provide frequency and severity assumptions in filing their response.

89. At all times material, AIG's Richard L. Thomas (see Finding of Fact 76), had significant experience in underwriting unusual risks, and risks without significant historical loss data, such as the Y2K phenomenon. He is on the Board of the Workers' Compensation Research Institute. He serves on the National Council of Compensation Insurers' Ad Hoc Terrorism Committee for Workers' Compensation, which uses actuaries and which engaged Equicat to begin developing a modeling process for TRIA workers' compensation insurance rates. Mr. Thomas was charged by AIG with the responsibility for evaluating the risk that foreign terrorism posed to the insurance written by AIG after the events of September 11, 2001.

90. Mr. Thomas concedes that AIG's TRIA burglary-theft rate filing, even as supplemented on October 10, 2003, did not include investment income, reinsurance, past and prospective loss experience within and without Florida, or trend factors (usually understood to be projections of frequency and severity of losses) as they are routinely prepared for rate filings by the actuarial department within AIG's Domestic Brokerage Group. He testified that AIG's historical loss experience was not included in the filing because it does not exist for terrorism and that his assumptions of pricing and sales were not explained

    in the filing. However, his assumptions were presented in the de novo hearing. See infra.

91. Evidence adduced at hearing shows that Mr. Thomas assumed an eight-year horizon, thereby predicting frequency, and assumed a 1.2 billion dollar loss, thereby predicting severity of loss(es), as more fully explained, infra.

92. Mr. Thomas testified that he could not find catastrophic "cover" for AIG.

93. Mr. Thomas testified that he considers TRIA itself to constitute reinsurance.

94. Mr. Thomas closely followed the testimony, committee meetings, and successive drafts of TRIA. Upon TRIA's passage, he had the responsibility to evaluate the risk that foreign terrorism posed to insurance written by Petitioners in light of the language employed in the final Act. He supervised development of AIG's TRIA rates.

95. Mr. Thomas informed himself concerning the threat of foreign-sponsored terrorism in the United States and the risk exposure of his company's several lines of business under TRIA in a number of ways. He read governmental publications on the subject. He attended insurance industry-sponsored seminars on such risks to the industry. He evaluated the insured losses that Petitioners and the domestic P&C insurance industry as a whole had experienced as a result of the September 11, 2001,

    events. He consulted people involved in the study of foreign terrorist organizations and consulted present and former governmental and intelligence personnel, but his unmemorialized conversations with terrorism experts are at least as much a "black box" as ISO's methodology.

96. Mr. Thomas reviewed, to the extent possible, the several proprietary modeling techniques created to assist insurers in evaluating the risks of catastrophic events, such as hurricanes and earthquakes. One such model was the AIR model (simulation) which incorporates the Delphi modeling technique. He reviewed the ISO Florida TRIA burglary-theft rate filing.

97. Mr. Thomas formed the opinion that modeling techniques, such as AIR-Delphi, were good at forecasting severity of future attacks, but he rejected their predictions of location and frequency.

98. Mr. Thomas directed AIG business unit and product line managers to evaluate the risk posed under TRIA to their existing books of business and to recommend rates for TRIA coverage.

99. Without consulting with any terrorism experts and with little to go on but newspapers and internal memoranda, the product line underwriters set about their task.

100. Mr. Thomas testified that he gave these managers little direction because he wanted independent thoughts,

     concepts, and input in the initial stages of this new rate pricing concept.

101. However, Ms. Gaillard, an actuary with AIG, commented that during AIG's internal data gathering stage, if an AIG division volunteered that it had no TRIA exposure, the internal response was that because TRIA said they have an exposure, that Division needed to develop some rate to support AIG's deductible. This suggests that seeking to cover AIG's deductible was Mr. Thomas' goal from the beginning.

102. Patricia Barrett is AIG's Executive Vice-President for middle market fidelity. Middle market accounts are those less than 100 million dollars. Brian O'Neill is Vice-President and Senior Fidelity Officer for National Union, an AIG component company. Mr. O'Neill oversees risk exposures for accounts over

     100 million dollars. Ms. Barrett's and Mr. O'Neill's data gathering and underwriting conclusions for burglary-theft rate setting were reviewed by National Union's Chief Underwriting Officer, William Cotter.

103. A "white paper" was consulted by Ms. Barrett and possibly was consulted by Brian O'Neill and William Cotter. The white paper estimated categories of terrorism risk by type of entity insured. Ms. Barrett did not know the origins of the white paper or how the classifications of the industries named

     in it were determined. She admitted that the assessments were "pretty subjective."

104. The white paper assumed that AIG's burglary-theft coverage "extends to any burglary and theft arising from a terrorism event and loss of money and securities due to destruction of premises from a terrorism event," and noted that burglary-theft insurance is considered "fidelity" insurance.

105. According to Ms. Barrett, the white paper shows some locations are more likely to be a target but are not necessarily more vulnerable if they are targeted. Yet, on the white paper itself, no symbol for "fidelity" or "burglary-theft" is ranked. With the exception of descriptive examples of potential activity such as "theft of propane gas trucks" and "robbery/burglary of guns, ammunition, and explosives," the white paper does not pinpoint burglary or theft, per se.

106. Taking into account the nature and risk profiles of their books of business and the provisions of TRIA, AIG underwriters Barrett, O'Neill, and Cotter analyzed burglary- theft TRIA rates. Taking into consideration the risk profile unique to AIG's insureds, the likelihood that an existing insured would purchase the terrorism coverage, the need to keep AIG's rates low enough to encourage existing insureds and potential customers to fund the terrorism coverage by purchasing it, and largely hearsay information regarding the terrorism

     threat, which all or some of the underwriters believed was more likely to occur in areas of a high concentration of people and buildings, Barrett, O'Neill, and Cotter recommended TRIA rates for burglary-theft coverage to Mr. Thomas. Mr. Cotter considered trophy landmarks located in New York City and Washington, D. C. to be more likely targets than lesser targets in more remote areas.

107. Barrett, O'Neill, and Cotter concluded that all risks in AIG's burglary-theft line of insurance presented at least some risk of loss due to a foreign-sponsored terrorism event, and therefore all risks should be charged some (minimal) amount for the terrorism coverage of burglary-theft that TRIA mandated insurers like AIG offer. They also concluded that due to the nature of their insureds' operations and the amount of coverage purchased by those insureds, probably ninety percent of the insureds/risks in AIG's burglary-theft line of business presented a low level of exposure to such losses. Although they went through several processes, Barrett, O'Neill, and Cotter also came up with the belief, as ISO had, that the vast majority of AIG insureds should pay one percent of their underlying premium for TRIA coverage.

108. Some consideration was given by Ms. Barrett to using fractions of one percent for very small risks, but this was ultimately rejected.

109. These three underwriters also concluded that some insureds, because of their nature and the coverage limits purchased, (such as large governmental organizations, middle market commercial operations with very high value, operations with easily transportable products such as jewelry, and national account financial institutions with high cash on hand, which jewelry and cast cash are subject to looters or to destruction during the actual terrorist event), presented higher risks.

110. Based on their underwriting judgment, which was primarily grounded in their own book of business and operating figures, Barrett, O'Neill, and Cotter recommended that AIG create a TRIA charge in ranges from one percent to three percent and two percent to five percent, depending upon whether the risk was judged by the underwriter to be exposed to a terrorism risk and a low or high crime-loss risk. This "range" was refined before being submitted to OIR. (See Findings of Fact 64 and 111.)

111. The manual exceptions page of the burglary-theft

     
     

     filing at issue read:

     
     

     PRICING

     
     

     WITH NO TRIA EXCLUSION, THE FOLLOWING PERCENTAGES WILL BE APPLIED TO THE AT LIMITS COVERAGE PREMIUM TO DETERMINE THE CHARGE FOR TRIA TERRORISM:

     
     

     Middle Market Accounts

     1% of gross Annual Premium

     Includes ALL Middle Market accounts (as defined) except those risks as noted below:

     
     

     2% of gross Annual Premium =

     Any Middle Market risk that has an on- premises "money" exposure of $50,000 or greater or a "securities" exposure of

     $100,000 or greater

     
     

     3% of gross Annual Premium =

     All Middle Market governmental agencies, where the limit of liability is equal to or greater than $2.5 Million, regardless of the "money" or "securities" exposure.

     Commercial/National Accounts Low = 1% of gross Annual Premium

     Includes ALL Commercial/National

     categorized with a "Low" risk assessment.

     
     

     High = 2% of gross Annual Premium

     Includes ALL Commercial/National accounts (as defined) categorized with a "High" risk assessment except those risks as noted below:

     
     

     3% of gross Annual Premium = Commercial Account:

     Financial Accounts (w/Cash exposure greater than the deductible)

     National Account:

     Case Intensive Risks (w/Cash exposure greater than the deductible)

     Jewelry Retailers/Wholesalers 4% of gross Annual Premium = National Account:

     Financial Accounts (w/Cash exposure greater than the deductible)

     
     

112. The foregoing price list is similar to the "white paper," but amounts to a shuffling of categories with enormous

     discretion left in the hands of whoever, within AIG, assigns rates on a customer-by-customer basis.

113. As a result, AIG's pricing for burglary-theft coverage was, in effect, set by rating certain risks either high or low and then assigning a specific percentage of surcharge to that risk for terrorism coverage.

114. Ms. Barrett and Mr. O'Neill did not consult with any actuaries or terrorism experts to reach their assumptions or their pricing categories. Mr. Cotter claimed to have received some advice from experts in terrorism, but he could not explain clearly what that advice was. (Another black box.) None of the three underwriters consulted any writings, treatises, research, or studies written by terrorism experts. Nor did they review any terrorism models. Moreover, none of them were aware of the eight-year timeline ultimately selected by Mr. Thomas, and their geographical and demographical considerations were inconsistent with his. (See Findings of Fact 106 and 121.)

115. At hearing, relative to AIG's conclusion that every one of its insureds had some exposure to terrorist attack and AIG's decision to assign some percentage of risk to every insured, Ms. Barrett was posed the hypothetical of a "Mom and Pop" retail establishment in Sopchoppy, a very small rural Florida town with no acknowledged target attractive to terrorists nearby. She speculated that a messenger from the

     store, with a cash deposit on his way to a financial institution could be in proximity to a terrorist-targeted truck on the same highway. Ms. Barrett's other speculation as to risk exposure was that a contaminant dropped in a targeted location could drift as a cloud, and seep into the "Mom and Pop" store 50 miles away, rendering the untargeted premises inaccessible. Under AIG's liberal construction of TRIA burglary-theft coverage (see Finding of Fact 104), Ms. Barrett's first hypothetical is barely reasonable, but the second clearly does not take into consideration the probability of other types of concurrent liability coverage.

116. Since every AIG insured may simultaneously present a burglary-theft risk, another property casualty risk, a life or disability risk, or even a multi-peril risk (of which burglary- theft is sometimes a component), it is not clear how every AIG insured entity could be reasonably predicted to present a separate burglary-theft risk, for rating purposes.

117. In light of the foregoing, it was not proven that every one of Petitioners' insureds present some burglary-theft risk.

118. In light of the foregoing, AIG did not affirmatively demonstrate the reasonableness of its assigned categories of risk.

119. Likewise, even given that AIG insureds who deal in

     jewelry, cash, or securities may present a greater risk than some other commercial enterprises, there was no adequate justification of how AIG would assign insureds to the four percentage categories ultimately listed in its filing.

120. Mr. Thomas claimed to have tested his business units' expert underwriting opinions to ensure reasonableness, but his "strategic analysis" amounts to a "one man decision."

121. On the basis of his studies, Mr. Thomas concluded that concentration of buildings and people was not determinative of where the next terrorist attack(s) might occur and that having already struck New York City and Washington, D.C., it was less likely terrorists would soon strike there again. Both of these assumptions are contrary to some of the assumptions that went into his subordinate underwriters' assignment of high/low risk categories. (See Findings of Fact 106 and 130.)

122. Mr. Thomas also concluded that although it was not possible to predict with any reasonable degree of assurance precisely when, in what geographic locale, or by what means a TRIA-qualifying foreign terrorist attack would occur, it was reasonable to conclude that sometime before September 11, 2009, such an event will occur and will result in losses equal to, or exceeding, the sequelae of losses rippling out from the several events of September 11, 2001. He selected his eight-year time horizon on the following basis: The time between the two known

     exemplar terrorist WTC events was 1993 to 2001, or eight-years. The unnamed experts he consulted thought that it would take at least another eight years for known terrorist groups to communicate, plan, and carry out another similarly severe attack. The last universally-agreed terrorism exemplar was September 11, 2001. The span of TRIA did not begin until November 2002 and is projected to end with 2005.

123. The severity level Mr. Thomas assumed was a severe event, possibly the equivalent of September 11, 2001, within eight years of that date, not every eight years. The eight years constitute his frequency assumption.

124. Mr. Thomas concluded that AIG's share of the losses from such a terrorist event would more likely than not be in proportion to AIG's market share of the United States domestic P&C insurance market. This component of his formula is a reasonable assumption.

125. Using these figures, Mr. Thomas projected that Petitioners' deductible under TRIA for the year 2003, would be approximately $1.2 billion; for the year 2004, would be approximately $2.2 billion; and for the year 2005, would be approximately $4.0 billion. His estimate for AIG's 2003 TRIA deductible was based on AIG's 2002 direct earned premium data. His estimate for 2004 and 2005 were projected from AIG's known business performance, based on assumptions concerning growth

     rates. Assumptions regarding growth rates include underwriting prognostications of which categories of insureds already on AIG's book of business will buy TRIA coverage and which potential customers will buy the coverage and when they will buy it.

126. Mr. Thomas's testimony that it is reasonable to assume that insureds with the greatest risk exposure to terrorism will purchase TRIA coverage and that only an attractive rate will draw in the insureds with less exposure is facially logical, but it is of concern that he could simultaneously testify that with AIG customers buying AIG at an increased rate, price is not an impediment to customers making the decision to buy.

127. Considering the severity indications from terrorism simulation models to the extent he had access to them, considering Petitioners' 8.8 percent share of the P&C insurance market, and considering the types of risks that AIG predominantly underwrites (see Findings of Fact 8, 104, and 116), Mr. Thomas concluded that Petitioners' losses from projected terrorist events would probably approximate AIG's deductible in any given year and that AIG should set TRIA premiums so that, over the eight-year time horizon of September 2001 to September 2009, AIG's rates would cumulatively approximate Petitioners' TRIA deductible. Any excess monies not

     utilized in a given year for a terrorist event were programmed to roll over.

128. Mr. Thomas concluded that, barring intervening terrorism events, the TRIA rates recommended by his subordinate underwriters would accumulate by the end of 2009, the end of his eight-year time horizon, to an amount that would then be available to help fund AIG losses from a terrorist attack. He intended to set AIG burglary-theft rates to result in a collection over his eight-year time horizon which would approximate AIG's deductible in the second year of TRIA's three year timeline.

129. Mr. Thomas is not an actuary and AIG employed no actuaries in the calculations for its TRIA burglary-theft rate filing.

130. All witnesses described terrorism events as high severity/low frequency risks, but all concerned differ on what is to be expected. Some of Mr. Thomas' subordinate underwriters assumed that a large September 11, 2001-type event in an area of highly concentrated buildings and people or trophy targets were more likely to be attacked, but had no concept of how soon. The AIR-Delphi model predicted smaller events in locations other than Florida in less than eight years. Ms. Galliard, one of AIG's actuaries, and OIR's outsourced expert actuary, Mr. Hunter, speaking generally, felt that the 1993 and 2001 WTC

     events were too few and too unhomogeneous to be useful, by themselves, in determining frequency. Speaking more specifically, Ms. Galliard considered any two events insufficient to discern a pattern or trend. Mr. Hunter testified similarly, but he also believed that it was more reasonable to expect events similar to September 11, 2001, but of lower severity.

131. Simplified, Mr. Thomas assumed a frequency rate of eight years and a severity rate in the second year of 1.2 billion dollars. Any of the foregoing timeframes and severity estimates might be equally "reasonable," with that selected by Mr. Thomas. However, the fact that Mr. Thomas' subordinates had a different concept of potential attack locations and event severity when they gathered the data than the concept upon which Mr. Thomas based the rest of his analysis and upon which AIG based its rate filing is contradictory and troubling.

132. Also, due to 181 third party actions still pending from the 1993 WTC bombing, AIG can only approximate its market share loss for that past event, estimated at .5 billion dollars, plus whatever the third party losses would be, possibly up to

     1.9 billion. This position clearly does not produce the predictability normally associated with past loss calculations.

133. Several learned, credentialed actuaries, including but not limited to Mr. Lehmann, Mr. Schaffer, and Mr. Hunter,

     have applied their skills to the figures provided by Mr. Thomas and have achieved a variety of results. Mr. Lehmann could not match Mr. Thomas's figures, resulting in 1.4 to 1.5 billion accumulated by 2003, instead of 1.2 billion. Mr. Hunter, who has extensive experience with low frequency, high severity federal reinsurance programs such as federal flood, riot, and crime reinsurance, came to a much higher accumulation amount, partly based on different types of investment income and tax rates, which accumulation amount he declared to be unreasonable in relation to the risk involved. Mr. Schaffer reached Mr.

     Thomas' calculations perfectly, but neither Mr. Thomas nor Mr. Shaffer accounted for investment income.

134. Mr. Hunter interpreted as unreasonable Mr. Thomas' example of investing AIG's premiums in Treasury Bills earning one percent, because such large amounts of money are normally placed in a higher yield investment, such as municipal bonds. He also disagreed with Mr. Lehmann as to the appropriate tax rate and considered Mr. Thomas's no-growth premium projection unrealistic.

135. The evidence as a whole suggests that Mr. Thomas projected one substantial terrorism event within eight years after 2001, more likely later in that time line than earlier, and attempted to set rates to accumulate the capital needed to pay the deductibles from such an event by the end of that eight

     year period, calculating that if such an event happens sooner than his 2009 horizon, AIG still would be required to pay the losses when the event occurs, even though AIG would not, by then, have accumulated all the capital needed to cover the losses. While the undersigned is not sufficiently skilled to determine which, if any, of the named actuaries is correct in calculating the accumulations, and although it appears that neither Mr. Lehmann or Mr. Hunter fully understood what Mr.

     Thomas was trying to do, the fact that several varying results were achieved by skilled and credentialed actuaries does not persuade, by a preponderance of the evidence, that Mr. Thomas' figures will produce the results he predicts.

136. It is also disturbing that the highly experienced Mr.

     
     

     Hunter could predict that using Mr. Thomas' assumptions would ultimately require that the insurance industry as a whole would have to suffer a single loss of 55.9 billion dollars at the low end, which is nearly twice that of all insurers' September 11, 2001, losses, and an unreasonable prediction of loss.

137. The American Academy of Actuaries Report attached to AIG's October 10, 2003, response to OIR's deficiency letter supports use of a combination of statistical analysis and modeling to set rates. AIG used neither of those methods as commonly understood.

138. Within AIG's October 10, 2003 response to deficiency

     letter, under the heading "Casualty Actuarial Society's 2002 yearbook" was the following verbiage:

     According to Principle 4 of the Statement of Principles Regarding Property and Casualty Ratemaking (page 321) "a rate is reasonable and not excessive, inadequate, or unfairly discriminatory if it is an actuarially sound estimate of the expected value of all future casts associated with an individual risk transfer." Part III, Considerations (page

     321) provides considerations that "are intended to provide a foundation for the development of actuarial procedures and standards of practice."

     
     

139. The Statement of Principles Regarding Property Casualty Ratemaking, published by the Casualty Actuarial Society in 1988, and apparently not rescinded, states that "a rate is an estimate of the expected value of future costs." The Statement says nothing about covering deductibles.

     CONCLUSIONS OF LAW

     
     

140. The Division of Administrative Hearings has jurisdiction over the parties and the subject matter of this cause pursuant to Sections 120.569 and 120.57(1), including but not limited to Section 120.57 (1) (e)2., Florida Statutes.

141. Herein, Petitioners AIG assert that on September 4, 2003, they made an adequate and compliant "use and file" rate filing; received an Agency deficiency letter dated September 11, 2003; sought and received three extensions for submission of their reply to the deficiency letter; submitted their timely

     response on October 10, 2003; and received an intent to disapprove letter (NOI) from the Agency dated October 14, 2003. Petitioners allege that they then discovered that OIR had a "secret" non-rule policy, providing a one percent cap on all TRIA rate filings in Florida, which had resulted in the de facto denial of their rate filing on October 3, 2003, even before AIG's time to respond to the deficiency letter had expired.

142. Respondent OIR asserts that Petitioners may have received extensions for their response to the deficiency letter up to October 8, 2003, but no further; that, therefore, Petitioners' October 10, 2003, response was untimely; that AIG's untimely response was reviewed by the Agency anyway, and at that time, the Agency found AIG's rate filing, even as supplemented, to be unsupported, pursuant to the applicable statutes and rules; and that the Agency denied the September 4, 2003, rate filing, as supplemented on October 10, 2003, on its merits by an OIR letter dated October 14, 2003. The Agency denies that any non-rule policy has existed at any time but submits that ISO's one percent commercial property rate filing, which OIR approved, meets Agency actuarial requirements under the applicable statutes and rules.

143. This case in no way affects OIR's approval of the ISO commercial property filing or any other rate filings.

144. Petitioners bear the burden to show by a

     preponderance of the evidence that the Agency engaged in a non- rule policy and disapproved of Petitioners' rate filing based on such non-rule policy. See § 120.57(1), Fla. Stat.

145. Normally, if a Petitioner demonstrates the Agency's use of a non-rule policy, the burden sifts to the Agency to establish, by a preponderance of the evidence, that the non-rule policy is valid, so that it may be applied in the case at bar. See § 120.57(1)(e)2., Fla. Stat. However, herein, the Agency denies there was ever any non-rule policy and has made no effort to prove up such a policy. Therefore, the Agency's right to rely upon a non-rule policy is not an issue herein. It remains only for Petitioners to establish the existence, use, and invalidity of the alleged one percent cap non-rule policy.

146. That said, Petitioners ultimately bear the burden to prove, by a preponderance of the evidence, that AIG's TRIA burglary-theft rate filing produces reasonable rates, i.e., rates that are not excessive (the sole reason the Agency gave in its October 14, 2003, NOI letter), that are not inadequate, and that are not unfairly discriminatory in relation to the risk involved. See § 627.062(1), (2)(b)-(e), Fla. Stat., and Florida Department of Transportation v. J.W.C. Co., Inc., 396 So. 2d 778 (Fla. 1st DCA 1981).

147. However, before that determination can be made, several threshold issues must be addressed. First, the Agency's

     position and proof that it did a complete, though cursory, review of AIG's October 10, 2003, supplementary material renders moot any suggestion that AIG filed the supplementary material "late," i.e., beyond the last extension of time granted by the Agency for the filing of a response to its deficiency letter.

     After October 8, 2003, the Agency could have rejected Petitioners' rate filing on tardiness grounds alone, but instead, Mr. Ramanujam considered the material filed on October 10, 2003. He disapproved the rate filing after consideration of the October 10, 2003, material, but without taking the issue again to a red book meeting. Under these circumstances, the Agency cannot now be heard to complain that the October 10, 2003, response was untimely filed. Other

     reasons contrary to the Agency's concept of untimeliness may be found in the statutes, rules, and customs discussed in Findings of Fact 25, 69, and 84.

148. Another threshold issue is whether the infamous zeroes are fatal to this filing. It is concluded that they are not. Most of the correct answers could be divined from the remainder of the filing. The Agency did not even mention the zeroes in its September 11, 2003, deficiency letter or

     October 14, 2003, NOI letter. Finally, Mr. Ramanujam testified that he did not consider these flaws to be fatal to the filing. (See Findings of Fact 64, 68-69, and 84.)

149. ISO's general liability filing was approved, despite use of two rate categories. Under the facts of this case, AIG's use of the mechanical exceptions page arguably avoided Florida's ban on tier-rating. Mr. Ramanujam did not disapprove AIG's filing on the basis of the four categories of risk assigned by AIG. Therefore, the use of categories was not fatal to AIG's filing then and should not be fatal now. However, how AIG's categories were derived and whether they are justified are still legitimate components of assessing AIG's entire filing, pursuant to Section 627.062, Florida Statutes.

150. The Agency's assertion that the absence of an explanatory memorandum is fatal to the filing is also rejected. Mr. Simmons' October 10, 2003, letter is within the parameters of the applicable rule, and this disputed-fact proceeding further serves the purpose of an explanatory memorandum.

151. The Agency's suggestion that AIG's supplemental October 10, 2003, material could not be considered in the instant de novo proceeding is, once again, rejected upon considerations generic to the rights afforded AIG pursuant to Chapter 120, Florida Statutes. The concept that only the four corners of the supplemented rate filing may be considered and that no explanation of the filing by testimony and exhibits should have occurred, is likewise rejected as contrary to Section 120.57(1), Florida Statutes.

152. As with any de novo hearing, each party has had an opportunity to prove-up its position on the merits of the proposed rate increase. The elements of that proof must satisfy the applicable Florida law.

153. Under Florida's Administrative Procedures Act, a "rule" is defined as:

     . . . each agency statement of general applicability that implements, interprets, or prescribes law or policy or describes the procedure or practice requirements of an agency and includes any form which imposes any requirement or solicits any information not specifically required by statute or by an existing rule. . . .

     
     

154. After careful weighing of the totality of the evidence, it is concluded that at all times material, OIR had in place a non-rule policy to the effect that any TRIA rate filing in excess of one percent was irrebuttably presumed to be unreasonable, unjustified, and excessive. The result of this non-rule policy was to first try to persuade each filer to voluntarily reduce its request to one percent or to adopt the ISO commercial property filing. Agency approval or denial of individual filings was delayed as long as possible to let persuasion work. AIG did not reduce its request, and so its filing was denied because it had proposed TRIA rates over one percent.

155. The one percent cap non-rule policy is arbitrary and

     capricious, because it transforms, for all practical purposes, an advisory rate filing for commercial property by ISO, a single rating organization, into a rate ceiling for every insurer in every line of insurance. This is particularly egregious because the ISO rate filing cautions that it is not appropriate even for all commercial property insurers, due to the disparate effects that TRIA has on various insurers. ISO's commercial property filing and one percent rate never purported to make an estimate of the cost of TRIA coverage for crime, which is the issue herein. Even so, as an ISO member, AIG has a legal option to adopt or not adopt an ISO burglary-theft filing, but OIR's non- rule policy imposes the ISO filing on AIG and all other insurers. Obviously, that filing and its underpinning, the privately-owned AIR-Delphi modeling system, have not been incorporated into Florida's statutes or administrative rules.

     This invalid policy further imposes the inaccessible AIR-Delphi proprietary modeling system upon any insurer who wishes to write burglary-theft insurance in Florida. If that were not bad enough, all witnesses concede the AIR-Delphi model is at least minimally flawed. Finally, the non-rule policy gives no consideration to rate adequacy or non-discrimination considerations, as required by Section 627.062, Florida Statutes.

156. Petitioners request that the undersigned further

     conclude that the non-rule policy requires Petitioners to provide evidence that Congress and the actuarial community have declared do not exist; to place their capital at risk by using the flawed AIR-Delphi modeling system; and to give no deference to national policy as expressed in TRIA. Such conclusions are not necessary to resolve this case and would address federal issues outside the jurisdiction of this forum. Therefore, no conclusions are made on those propositions.

157. Once the non-rule policy is discredited, Petitioners and the Agency are bound by the applicable existing Florida statutes and rules.

158. The Legislature has charged OIR with reviewing rate filings pursuant to Section 627.062(2), Florida Statutes, which provides:

     1. Upon receiving a rate filing, the office shall review the rate filing to determine if a rate is excessive, inadequate, or unfairly discriminatory. In making that determination, the office shall, in accordance with generally accepted and reasonable actuarial techniques, consider the following factors:

        1. Past and prospective loss experience within and without this state.

        2. Past and prospective expenses.

        3. The degree of competition among insurers for the risk insured.

        4. Investment income reasonably expected by the insurer, consistent with the insurer's investment practices, from investable premiums anticipated in the filing plus any other expected income from currently invested assets representing the

           amount expected on earned premium reserves and loss reserves. The commission may adopt rules utilizing reasonable techniques of actuarial science and economics to specify the manner in which insurers shall calculate investment income attributable to such classes of insurance written in this state and the manner in which such investment income shall be used in the calculation of insurance rates. Such manner shall contemplate allowances for an underwriting profit factor and full consideration of investment income which produce a reasonable rate of return; however, investment income from invested surplus shall not be considered.

        5. The reasonableness of the judgment reflected in the filing.

        6. Dividends, savings, or unabsorbed premium deposits allowed or returned to Florida policyholders, members, or subscribers.

        7. The adequacy of loss reserves.

        8. The cost of reinsurance.

        9. Trend factors, including trends in actual losses per insured unit for the insurer making the filing.

        10. Conflagration and catastrophe hazards, if applicable.

        11. A reasonable margin for underwriting profit and contingencies.

        12. The cost of medical services, if applicable.

        13. Other relevant factors which impact upon the frequency and severity of claims or upon expenses.

        
        

     2. In the case of fire insurance rates, consideration shall be given to the availability of water supplies and the experience of the fire insurance business during a period of not less than the most recent 5-year period for which such experience is available.

     3. If conflagration or catastrophe hazards are given consideration by an insurer in its rates or rating plan,

        including surcharges and discounts, the insurer shall establish a reserve for that portion of the premium allocated to such hazard and shall maintain the premium in a catastrophe reserve. Any removal of such premiums from the reserve for purposes other than paying claims associated with a catastrophe or purchasing reinsurance for catastrophes shall be subject to approval of the office. Any ceding commission received by an insurer purchasing reinsurance for catastrophes shall be placed in the catastrophe reserve.

     4. After consideration of the rate factors provided in paragraphs (b), (c), and (d), a rate may be found by the office to be excessive, inadequate, or unfairly discriminatory based upon the following standards:

     1. Rates shall be deemed excessive if they are likely to produce a profit from Florida business that is unreasonably high in relation to services rendered.

     2. Rates shall be deemed excessive if, among other things, the rate structure established by a stock insurance company provides for replenishment of surpluses from premiums, when the replenishment is attributable to investment losses.

     3. Rates shall be deemed inadequate if they are clearly insufficient, together with the investment income attributable to them, to sustain projected losses and expenses in the class of business to which they apply.

     4. A rating plan including discounts, credits, or surcharges shall be deemed unfairly discriminatory if it fails to clearly and equitably reflect considerations of the policyholder's participation in a risk management program adopted pursuant to s. 627.0625.

     5. A rate shall be deemed inadequate as to the premium charged to a risk or group of risks if discounts or credits are allowed which exceed a reasonable reflection of expense savings and reasonably expected,

        loss experience from the risk or group of risks.

     6. A rate shall be deemed unfairly discriminatory as to a risk or group of risks if the application of premium discounts, credits, or surcharges among such risks does not bear a reasonable relationship to the expected loss and expense experience among the various risks.

     
     

159. AIG's suggestion that a rate filing is only required to address all 13 considerations listed in the statute if the Agency requests them by a deficiency letter is rejected as contrary to the statute.

160. The Agency contends that Petitioners did not supply the elements discussed in subparagraphs (2)(b) 1, 4, 8, 9, and 13, above: (1) provide past and prospective loss experience;

     (4) utilize investment income in their calculations; (8) discuss reinsurance; (9) discuss trend factors; and (13), which the Agency interprets to require frequency and severity assumptions.

161. The facts are clear that even in the de novo hearing on the merits, AIG has failed to fully account for investment income.

162. Petitioners are correct that the statute does not require that the thirteen elements be developed by the filer in accord with generally accepted and reasonable actuarial techniques, but the statute also does not acknowledge that any of the thirteen elements, such as historic loss data from terrorist attacks, may not exist. It does not recognize any

     situation in which all the elements cannot be presented to the Agency in one form or another.

163. The statute does require that the Agency determine if a rate filing is excessive, inadequate, or unfairly discriminatory. The statute further requires that in making that determination, the Agency shall consider the thirteen elements "in accord with generally accepted and reasonable actuarial techniques." Unfortunately, no evidence, statute, or rule defines "actuarial techniques." The statute is also silent as to what, if any, relative weight the Agency may assign to each element.

164. The thirteenth element inherently requires the filer to provide the filer's frequency and severity assumptions upon which its filing is based, but the thirteenth element also specifically permits the Agency to consider "other relevant factors which impact upon the frequency or severity of claims or upon expenses." Presumably, that language permits the Agency to consider the absence of historical data and, in the absence of any contrary direction in the statute, the Agency may weigh one or more of the twelve remaining elements more heavily than others. This statutory interpretation is in line with evidence that each rate filing is different and that the Agency reviews each one in its totality, more as a negotiation, than as an

     assessment of minimum standards. The Agency's interpretation of its statute is entitled to great weight.

165. In the instant situation, it is reasonable to apply a weighted consideration of statutory element five, the reasonableness of the underwriting judgment reflected in the filing. That consideration follows.

166. Petitioners state that no information exists on which to make statistically valid estimates of the probability and cost of future terrorist events, and that as a result, the Agency may not require the use of statistically valid estimates of the probability and cost of loss events projected from non- existent historical information. Based on its "we cannot submit what is non-existent" theory, AIG asserts that in the absence of sufficient data to which recognized actuarial techniques can be applied, a sort of commercial feasibility test of "reasonableness" alone is appropriate. AIG's perception of underwriting reasonableness is to the effect that when TRIA ends, so does the federal backup, and AIG would then be "on the hook" for one hundred percent of its terrorism risk exposure without any federal "reinsurance," and accordingly, AIG should be entitled to collect its deductibles, rather than demonstrating the cost of its projected losses.

167. AIG's analysis may be commercially sound, but it does not comport with Florida's statute, which presumes that expected

     losses must be considered in determining rates. Unless the Legislature amends the statute, filers still must somehow demonstrate "expected losses" to show that a proposed rate is not excessive.

168. AIG's filing sets a projected deductible as the sole test, rather than any of the thirteen statutory elements, and AIG's underwriting judgment behind this approach, as demonstrated in this case, does not support AIG's proposed burglary-theft rate(s).

169. All the evidence suggests that terrorism experts should be consulted. No terrorism expert has opined in this case to predict frequency, severity, or location of expected losses. Mr. Thomas is not a terrorism expert, and his information from terrorism experts was anecdotal and acquired without any system or recognized methodology.

170. If Mr. Thomas had selected a different timeline or had anticipated multiple small terrorist attacks, his predictions on rates would have been entirely different.

171. Contrary to some of the location, frequency, and severity assumptions of his subordinates, Mr. Thomas unilaterally assumed an eight-year timeline and a single major terrorist event, upon which to make his predictions. No internal underwriting consensus of thought was reached within AIG concerning location vis a vis geographic and demographic

     concentrations of people or structures. Neither Mr. Thomas' nor his staff's underlying assumptions were justified by the evidence as a whole, but the fact that a result was assembled from conflicting underwriting assumptions renders the whole filing suspect.

172. All AIG underwriters concurred that its burglary- theft line is less loss susceptible to terrorist attack than other lines of business, but Mr. Thomas' subordinate underwriting staff, who were responsible for setting the high/low classification of risk and the percentages of terrorism surcharge disagreed with his predictions of location and frequency. This difference alone renders Mr. Thomas' filtered "one-man" assumptions unreasonable, but further, AIG did not demonstrate how burglary-theft exposure could reasonably apply to every risk and did not demonstrate a credible rationale for how various industries were assigned to the high/low risk classifications/rate percentages.

173. Finally, the inability of several actuaries to achieve substantially similar results, using Mr. Thomas' figures results in the conclusion that AIG has not proven, by a preponderance of the evidence, that its rate request is not excessive, inadequate, or unfairly discriminatory.

RECOMMENDATION

Based on the foregoing Findings of Fact and Conclusions of Law, it is

RECOMMENDED that the Department of Banking and Finance, Office of Insurance Regulation, enter a final order disapproving the subject burglary-theft rate filing.

DONE AND ENTERED this 23rd day of March, 2005, in Tallahassee, Leon County, Florida.

S

ELLA JANE P. DAVIS

Administrative Law Judge

Division of Administrative Hearings The DeSoto Building

1230 Apalachee Parkway

Tallahassee, Florida 32399-3060

(850) 488-9675 SUNCOM 278-9675

Fax Filing (850) 921-6847 www.doah.state.fl.us

Filed with the Clerk of the Division of Administrative Hearings this 23rd day of March, 2005.

COPIES FURNISHED:

Honorable Tom Gallagher Chief Financial Officer

Department of Financial Services The Capitol, Plaza Level 11 Tallahassee, Florida 32399-0300

Peter Dunbar, General Counsel Department of Financial Services The Capitol, Plaza Level 11 Tallahassee, Florida 32399-0300

Daniel C. Brown, Esquire Kelly A. Cruz-Brown, Esquire Robert W. Pass, Esquire Carlton Fields, P.A.

Post Office Box 190 Tallahassee, Florida 32302-0190

Elenita Gomez, Esquire Dennis Silverman, Esquire Michael H. Davidson, Esquire Steven Parton, Esquire

Department of Financial Services Division of Legal Services

612 Larson Building, Suite 612

200 East Gaines Street Tallahassee, Florida 32399-4206

NOTICE OF RIGHT TO SUBMIT EXCEPTIONS

All parties have the right to submit written exceptions within

15 days from the date of this Recommended Order. Any exceptions to this Recommended Order should be filed with the agency that will issue the final order in this case.

Docket for Case No: 03-004477

Orders for Case No: 03-004477