People's Insurance Counsel Division v. Allstate Insurance

20 A.3d 117 (2011) 199 Md. App. 1

PEOPLE'S INSURANCE COUNSEL DIVISION
v.
ALLSTATE INSURANCE COMPANY, et al.

No. 1949, Sept. Term, 2009.

Court of Special Appeals of Maryland.

May 10, 2011.

Peter K. Killough (Susan Cohen, Michael P. McDonald, on the brief) Baltimore, MD, for appellant.

Brian D. Bolton & Jeffrey L. Shelton (Douglas F. Gansler, Atty. Gen., Michael P. Cunningham, Funk & Bolton, on the brief) Baltimore, MD, for appellee.

Panel: EYLER, JAMES R., HOTTEN, and CHARLES E. MOYLAN, JR. (Retired, Specially Assigned), JJ.

CHARLES E. MOYLAN, JR. (Retired, Specially Assigned), J.

In this challenge to a decision of the Maryland Insurance Administration ("MIA"), the appellant is the People's Insurance Counsel Division ("Division"). The Division is an entity of recent vintage, only having been created by the Maryland General Assembly during its Special Session in December of 2004. It is a subdivision within the Office of the Attorney General charged with a very special purpose. That purpose is to "evaluate each medical professional liability insurance and homeowner's insurance matter pending before the [Maryland Insurance] Commissioner to determine whether the interests of insurance consumers are affected." Maryland Code, State Government Article, § 6-306. With several differences not here pertinent, the relation of the Division to the MIA is closely analogous to the relation of the Office of People's Counsel to the Public Service Commission. People's Insurance Counsel Division v. Allstate Insurance Company, 408 Md. 336, 349-68, 969 A.2d 971 (2009).

The core problem in this case is that the Division sought to impose on the appellees, Allstate Insurance Company and Allstate Indemnity Company (collectively "Allstate"), a two-pronged set of conditions or restrictions devised by the General Assembly, beginning in 1970, essentially to combat discriminatory practices in underwriting of racial, ethnic, religious, gender and other familiar varieties. The flaw in the Division's effort is that it seeks to apply those conditions or restrictions to what was a fundamentally business decision of Allstate that did not remotely involve any of the traditional or historic discriminations.

A final threshold observation about this appeal is that the type of risk here being examined, the risk of catastrophic wind damage associated with hurricanes, is a phenomenon so completely and fundamentally dissimilar to the only types of risk heretofore dealt with in the caselaw that the case becomes one of truly first impression.

Procedural History

On December 4, 2006, Allstate advised the MIA that it intended to cease writing new property insurance policies in "certain catastrophe-prone areas" in Maryland effective January 1, 2007. Allstate believed that certain coastal areas bordering the Atlantic Ocean and the Chesapeake Bay presented an unusually high risk of loss in the event of a catastrophic hurricane. As a result, it decided that it was no longer in Allstate's best economic interest to continue to write new property insurance policies in those areas. After an "extensive six month review," this filing of its intent to stop writing new property insurance, submitted pursuant to Maryland Code (1995, 2006 Repl.Vol.), § 19-107 of the Insurance Article ("I.A."), was approved by the Maryland Insurance Commissioner ("the Commissioner") on May 31, 2007. The MIA concluded that the designation of the geographic areas within which Allstate would no longer issue new policies had "an objective basis and [was] neither arbitrary nor unreasonable."

The very next day, the Division requested a hearing before the MIA regarding Allstate's filing. The Division's request was granted, and a hearing was held on December 13 and 14, 2007. By an order issued February 2, 2008, Associate Deputy Commissioner Thomas Paul Raimondi determined that the Division had standing to request the hearing, that Allstate had the burden of persuasion, and that Allstate had sufficiently demonstrated that its filing satisfied I.A. §§ 19-107 and 27-501.^[1]

The Division filed a petition for judicial review of the Commissioner's order in the Circuit Court for Baltimore City. Allstate moved to dismiss the petition, arguing that the Division lacked statutory authority to petition for judicial review. The circuit court agreed, and dismissed the petition for lack of standing on July 15, 2008. The Court of Appeals reversed the circuit court's order, holding that the Division had standing to seek judicial review, and remanded the case for further proceedings. Allstate Insurance Company, 408 Md. 336, 969 A.2d 971.

On remand, a hearing was held before Judge Sylvester B. Cox in the Circuit Court for Baltimore City on the Division's petition for judicial review on September 24, 2009. Judge Cox, by order filed October 9, 2009, denied the Division's petition, and affirmed the Commissioner's final order. The Division filed a timely appeal to this Court on November 4, 2009.

The Contentions

On appeal to this Court, the Division raises the following questions for our determination:

1. Did Allstate's proposed decision violate § 19-107 because its designation of a certain geographic area was arbitrary and unreasonable?; and

2. Did Allstate's proposed decision violate § 27-501 because 1) it failed to provide any statistical data showing the probability of a catastrophic hurricane, and 2) it failed to provide any statistical data showing that its rating plan then in effect was not sufficient to cover losses in the event of a catastrophic hurricane?

Standard of Review

Although the appeal to us is technically from the Circuit Court for Baltimore City, we are actually reviewing the decision of the Commissioner. We look not at the circuit court, but through the circuit court. Bayly Crossing, LLC v. Consumer Protection Division, 417 Md. 128, 136-37, 9 A.3d 4 (2010).

In reviewing the decisions of the Commissioner, this Court has an austerely limited role. "Ordinarily, a final order of the Commissioner must be upheld on judicial review if it is legally correct and reasonably supported by the evidentiary record." Insurance Commissioner v. Engelman, 345 Md. 402, 411, 692 A.2d 474 (1997). In reviewing the Commissioner's decision, our role is confined to "determining if there is substantial evidence in the record as a whole to support the agency's findings and conclusions, and to determine if the administrative decision is premised upon an erroneous conclusion of law." United Parcel Service, Inc. v. People's Counsel, 336 Md. 569, 577, 650 A.2d 226 (1994).

"In applying the substantial evidence test, we have emphasized that a `court should [not] substitute its judgment for the expertise of those persons who constitute the administrative agency from which the appeal is taken.'" Our obligation is "to `review the agency's decision in the light most favorable to the agency,' since their decisions are prima facie correct and carry with them the presumption of validity."

"Even with regard to some legal issues, a degree of deference should often be accorded the position of the administrative agency. Thus, an administrative agency's interpretation and application of the statute which the agency administers should ordinarily be given considerable weight by reviewing courts." We are under no constraint, however, "to affirm an agency decision premised solely upon an erroneous conclusion of law."

Grasslands Plantation, Inc. v. Frizz-King Enterprises, LLC, 410 Md. 191, 203-04, 978 A.2d 622 (2009) (citations omitted; emphasis supplied).

On this appeal, the Division, incidentally, does not challenge any of the Commissioner's factual findings.

Section 19-107(a)

The Division's first contention is that Allstate's decision to discontinue writing new property insurance in certain catastrophe-prone areas of Maryland was in violation of Insurance Article, § 19-107. Section 19-107(a) provides:

(a) In general.—An insurer may not refuse to issue or renew a contract of motor vehicle insurance, property insurance, or casualty insurance solely because the subject of the risk or the applicants or insured's address is located in a certain geographic area of the State unless:

(1) at least 60 days before the refusal, the insurer has filed with the Commissioner a written statement designating the geographic area; and

(2) the designation has an objective basis and is not arbitrary or unreasonable.

(Emphasis supplied).

Phrased affirmatively, that section therefore provides that Allstate "may ... refuse to issue ... a contract of ... property insurance or casualty insurance solely because the subject of the risk ... is located in a certain geographic area" so long as it satisfies preconditions (1) and (2). With respect to Allstate's satisfaction of precondition (1) there is no dispute. It filed with the Commissioner a written statement designating the geographic area within which it would no longer write property insurance more than 60 days before it began to implement its intended policy. The only dispute with respect to § 19-107(a) is whether Allstate's designation of the geographic area in which it intended to discontinue writing property insurance had an "objective basis" and was not "arbitrary or unreasonable" within the contemplation of subsection (a)(2).

"The Insurance Code is basically the product of a comprehensive revision enacted by Ch. 553 of the Acts of 1963." Muhl v. Magan, 313 Md. 462, 465, 545 A.2d 1321 (1988). In the 1996 recodification of the insurance laws, what became § 19-107(a) was taken directly from what had been Article 48A, § 61A. That earlier section had been enacted by Chapter 927, Section 1, of the Acts of 1965 and in its original form provided simply:

No insurer shall decline to issue or renew contracts of motor vehicle insurance solely on account of the geographic area within this State wherein is located the subject of the risk or the applicant's or insured's address, unless such insurer not less than sixty (60) days previously shall have filed with the Commissioner a written statement designating such geographic area, which statement shall be an open filing with the Commissioner as a matter of public record.

(Emphasis supplied). That appears to have been a legislative effort to prohibit, or at least to inhibit, the practices of some insurance companies to "red line" areas in which the companies would not write motor vehicle insurance on the basis of racial demographics.

In that original manifestation, the section applied only to motor vehicle insurance and it did not prohibit discriminatory practices generally. It was only in 1970 that the Legislature undertook to curb discriminatory practices in insurance underwriting. Section 61A was amended by Chapter 746 of the Acts of 1973, at which time it took essentially its present form. The coverage of the section was expanded so as to include property and casualty insurance as well as motor vehicle insurance. The section was also amended to provide that the "designated geographic area" in which the insurer would "decline to issue or renew contracts of ... insurance" must have "an objective basis and shall not be arbitrary or unreasonable."

What clearly emerges is that within the contemplation of § 19-107, the "arbitrary or unreasonable" test is not to be applied to the fiscal wisdom or business acumen of the insurance company in discontinuing to write or renew insurance but only to its "designated geographic area." It is simply the mapping function that must have an objective basis and must not be arbitrary or unreasonable. That particular concern is not with "what," but with "where."

An Objective Basis For the Designation Of a Catastrophe-Prone Geographic Area

Allstate timely filed with the MIA its intention to stop writing new property insurance in what it had determined to be a catastrophe-prone area of Maryland. This "catastrophe-prone" region, designated by Allstate as Hurricane Bands 4, 5, and 6, consisted of Calvert, St. Mary's, Somerset, Talbot, Wicomico, and Worcester Counties, as well as portions of Anne Arundel, Charles, Dorchester, Prince George's, and Queen Anne's Counties. Allstate Insurance Company, 408 Md. at 340 n. 3, 969 A.2d 971.

According to Allstate, its decision to stop writing new policies in Bands 4, 5, and 6 was based on its significant market share in those bands, and its projections of losses that would result in the event of a catastrophic storm striking Maryland^[2] Allstate, which claims to own 12.7% of the market statewide, and 13.0% in Bands 4, 5, and 6, made the "business judgment that further growth at this time could jeopardize [the company's] anticipated long term strength." According to Allstate, controlling its exposure by not writing new property insurance in Bands 4, 5, and 6 is an action "designed to help keep our commitments to our existing customers, and is being implemented so as to minimize any disruption to the market."

Allstate's belief that Bands 4, 5 and 6 are especially vulnerable to substantial losses resulting from wind-storm damage was based on information produced by a computer-generated model provided by a catastrophe modeling service, Applied Insurance Research, Inc. ("AIR"). At the December 2007 hearing before the circuit court, counsel for Allstate explained how the AIR Hurricane Model V7.0 (the "AIR model") operated:

[W]hat AIR did, Your Honor, is they were provided Allstate's actual property information from Maryland. They actually got the hard data from December 31, 2004, that Allstate had with respect to all of our homeowners, where they were located, how much insurance they had, et cetera on the properties. That's all loaded into the AIR model. ... What it did, in order to get down to the zip codes, statistical level, generated the next year 100,000 times. That is, it's doing simulations of the next year, 100,000 times. And what they do in order to do that is they look at the last 100 years of meteorological data to try to come up with a probability of various hurricane strikes.

... The modeling specifically revealed four hurricanes making landfall in Worcester County, Maryland. Two more hurricanes making landfall in Virginia and another making landfall in Delaware and those hurricanes would cause losses in Maryland alone at $500 million dollars, under the findings of fact by the Insurance Commissioner.^[3]

(Emphasis supplied).

Counsel for Allstate next explained how Allstate used the statistics generated by the AIR model:

[W]hat happened here after the Applied Insurance Research Model completed its work, Allstate looked at the data that came out of this modeling and examined it zip code by zip code. And what Allstate came up with was what was called—they were referred to throughout much of the [Commissioner's] decision, ADRs, which are Average Damage Ratios and those were constructed, Your Honor, based on the five percent of the most severe storms.

Because the question Allstate was trying to address was catastrophic risk, all storms were not considered. Only the five—top five percent of [the] most severe storms.

(Emphasis supplied).

Using the data output representing the worst five percent of simulated storms, damage ratios were calculated for every zip code in Maryland. The damage ratios represented the degree of damage a zip code could expect in the event of a catastrophic storm, as compared to the rest of the state. "The higher the damage ratio, the higher the potential damage an area is likely to sustain in the event of a hurricane."

Those damage ratios were then used to divide the state into "Hurricane Bands." Allstate banded zip codes with like damage ratios together, and separated those zip codes with dissimilar damage ratios.^[4] It then calculated the average modeled hurricane losses for a $300,000 home by band, and the loss per $1,000 of coverage for the average home in the respective bands as compared to the whole state, referred to by Allstate as the "damage ratio relativity." Allstate submitted these statistics, as featured below, to the MIA:

----------------------------------------------------------------------
                                     Average Hurricane
                  Allstate Average   Loss Predicted for   Damage Ratio
Homeowners Band     Damage Ratio       $300,000 Home       Relativity
----------------------------------------------------------------------
Band 1                  N/A                N/A                 N/A
----------------------------------------------------------------------
Band 2                 0.202             $60.60               0.65
----------------------------------------------------------------------
Band 3                 0.275             $82.50               0.88

----------------------------------------------------------------------
Band 4                 0.442             $132.60              1.42
----------------------------------------------------------------------
Band 5                 2.323             $696.90              7.47
----------------------------------------------------------------------
Band 6                 4.155            $1,246.50            13.36
----------------------------------------------------------------------
Grand Total            0.311             $93.30               1.00
----------------------------------------------------------------------

The "damage ratio relativity" column indicates that approximately 1.4 times more damage is expected in Band 4 as compared to the whole state, 7.5 times more in Band 5, and 13.4 times more in Band 6. Allstate contended that those statistics were "compelling evidence of the potential for substantial catastrophic loss in Maryland," and believed that "actions [were] necessary to responsibly control our exposure."

The Division also takes umbrage at the fact that "Allstate's geographic restrictions are not just along the coast—they extend far inland from the Atlantic Ocean, in some cases 60 miles or more. No other insurance company imposes such restrictions beyond 1 mile from the Atlantic Ocean." The Division would have the Commissioner put on geographic blinders. To be 60 miles away from the Atlantic Ocean is beside the point if one is nonetheless on the eastern or western littoral of the lower Chesapeake Bay or along the wide estuarial mouth of the Potomac River.

The essential north-south axis of the Chesapeake Bay is in parallel alignment with the paths of hurricanes veering north along the Atlantic coast. The width of the Bay's mouth and its lower reaches make them, in time of storm, almost open sea. Whereas Puget Sound may be shielded from the open Pacific by the Olympic Mountains and San Francisco Bay can hide behind the Santa Cruz Range, the Delmarva Peninsula offers the Chesapeake Bay no such topographical jetty or windbreak. Its negligible elevation above sea level and its pencil-thin tip offer little more protection than a sand bar. A hurricane coming directly up the Bay or across Northampton or Accomack Counties, Virginia, would have nothing to stop it until it hits southern Maryland or the coastline of the Eastern Shore. If other insurers have not yet recognized this, Allstate has.

With respect to § 19-107(a), Commissioner Raimondi, with ample support in the evidence, expressly found:

[W]hen an insurer intends to refuse to issue or renew particular risks based upon the applicant's place of residency, Ins. § 19-107 requires that the insurer file a "geographic designation," or a map, in which the geographic area in which business will be refused is defined. Section 19-107 is concerned with the manner in which the map has been drawn and is designed to ensure that the insurer's mapping has an objective and factual basis.

It is undisputed that Allstate filed a written designation of the geographic [area] where new coverage would cease to be written more than 60 days before the date of filing. This meets the requirement [of] Ins. § 19-107(a)(1).

It is likewise undisputed that Allstate's designation has an "objective basis", as required by Ins. § 19-107(a)(2). See PICD's Proposed of Law, Conclusion of Law #5. Allstate's designation has an "objective basis" because it is externally verifiable by zip code and hurricane band and is not subject to an insurer's perceptions, feelings, or intentions. See State Dep't of Assessments & Taxation v. Md.-Nat'l Capital Park & Planning Comm'n, 348 Md. 2, 13-14 [702 A.2d 690] (1997). Whether a particular insured will be affected by Allstate's § 19-107 filing can be verified by determining if the property is located within or outside of the zip codes identified by Allstate as part of Bands 4-6.

The final hurdle for Allstate with respect to § 19-107 is whether its geographic designation meets the statutory requirement that it not be "arbitrary or unreasonable." The Commissioner previously has defined the word "arbitrary" to mean: "... subject to individual judgment or discretion, and made without adequate determination of principle." See Berkshire Life Ins. Co. v. Maryland Insurance Administration, 142 Md.App. 628, 671 [791 A.2d 942] (2002). Blacks Law Dictionary, Sixth Edition, defines "arbitrary" similarly as: "including something done `[w]ithout adequate determining principle,' `nonrational' and `[w]illful and unreasoning action without consideration and regard for facts and circumstances presented'...," quoted in Hurl v. Board of Educ. of Howard Co., 107 Md.App. 286, 306 [667 A.2d 970] (1995).

With those definitions in mind, I conclude that Allstate's geographic designation of Hurricane Bands 4-6 had adequate factual support and, therefore, was not arbitrary or unreasonable. Allstate's hurricane bands were developed based on objective and reasonable factors, including modeled hurricane loss data, proximity to water and geographic contiguity. Through its use of the hurricane models, Allstate developed ADRs at a zip code level.22 The higher the ADR, the higher the potential damage the area in the band is likely to sustain in the event of a catastrophic storm.

Zip codes were grouped into hurricane bands based on an analysis of variance in damage ratio data in order to create the most efficient alignment of hurricane bands. Based on input from the MIA, Allstate amended its filing to move four zip codes from Band 4 to Band 3, which was supported by Allstate's analysis of the Average Damage Ratios ("ADRs") and the within and between variances. Relying on the raw data, Allstate refused to move 17 zip codes out of Band 4 and into Band 3, because the change was not supported based on analysis of the Average Damage Ratios and the between and within variance analysis. This further demonstrates that Allstate's grouping of zip codes into hurricane bands was guided by the data, and was therefore reasonable and supported by facts.

Because Allstate's geographic designation was based upon an objective and reasonable fact based grouping of zip codes with the highest ADRs together into hurricane bands, I conclude that Allstate has complied with the requirements of Ins. § 19-107.

22 Previous modeling had always been county based, but improvements in the modeling process enabled Allstate to be more precise in grouping geographic areas with similar risk bands.

(Emphasis supplied).

We hold that there was substantial evidence to support the Commissioner's findings and his rulings. With respect to § 19-107(a), there was no error.

Section 27-501(a)

The Division's second contention is that Allstate's decision to discontinue writing new property insurance in certain catastrophe-prone areas of Maryland was in violation of Insurance Article, § 27-501(a). Section 27-501(a) provides:

(1) An insurer or insurance producer may not cancel or refuse to underwrite or renew a particular insurance risk or class of risk for a reason based wholly or partly on race, color, creed, sex, or blindness of an applicant or policyholder or for any arbitrary, capricious, or unfairly discriminatory reason.

(2) Except as provided in this section, an insurer or insurance producer may not cancel or refuse to underwrite or renew a particular insurance risk or class of risk except by the application of standards that are reasonably related to the insurer's economic and business purposes.

(Emphasis supplied).

As it was with § 19-107(a), the satisfaction of subsection (a)(1) poses no problem. The Division never suggested that Allstate's projected action involved any of the traditionally prohibited discriminatory practices. The contention focuses exclusively on subsection (a)(2). There is no dispute as to what the words of (a)(2) say. There is a critical dispute as to how broadly the words of (a)(2) were ever meant to apply. The legislative history of § 27-501(a) is the indispensable starting point for any intelligent analysis.

In the recodification that produced the present Insurance Article, what is now § 27-501(a) was taken, without any substantive change, from what had been Article 48A, § 234A. Section 234A, in turn, had been enacted by Chapter 417 of the Acts of 1970 and was aimed at prohibiting a broad range of discriminatory practices. As first enacted, § 234A(a) "made it unlawful for an insurer ... to cancel or refuse to underwrite or renew a particular insurance risk or class of risks for any arbitrary, capricious, unfair or discriminatory reason based in whole or in part upon the race, creed, color, religion, national origin or place of residency of any applicant or policy-holder." Within a year, § 234A(a) was rewritten by Chapter 789 of the Acts of 1971. "Sex" was added to the catalogue of prohibited discriminatory criteria; "any arbitrary, capricious, or unfairly discriminatory reason" became an independent catchall criterion; and "religion, national origin or place of residency" were relegated to implicit membership in the latter catch-all category. "Blindness" has since been added to the list of forbidden criteria.

The first case to interpret the new section was Insurance Commissioner v. Allstate Insurance Company, 268 Md. 428, 302 A.2d 200 (1973). Judge Barnes summed up the purpose of the new legislation, 268 Md. at 442, 302 A.2d 200:

It is clear to us that the principal thrust of this legislation was directed toward any action of an insurer in failing to underwrite or renew a particular risk or class of risk for any reason based in whole or in part upon race, color, creed or sex of an applicant or policyholder for any arbitrary, capricious or unfairly discriminatory reason like those specifically mentioned, including, but not restricted to, religion, national origin, place of residency or other similar irrelevant considerations. In short, the General Assembly intended to broaden the scope of "arbitrary, capricious, or unfairly discriminatory reason," but within the frame of reference of the specifically mentioned "reasons."

(Emphasis supplied).

The decision in that case overturned two rulings by the Insurance Commissioner that the determinations by two insurance companies not to renew automobile insurance had been arbitrary and capricious and were, therefore, in violation of § 234A. The holding of the Court of Appeals was that the "arbitrary and capricious" language was restricted to the types of discrimination spelled out in the section and did not confer any right on the Commissioner to monitor in any broader way the decision of an insurer not to renew policies. The Commissioner, in short, could not challenge established underwriting criteria on grounds other than that they violated traditional discrimination factors.

In both Nos. 205 and 212, there was no evidence either before the Commissioner or the Baltimore City Court on appeal of any arbitrary, capricious, or unfairly discriminatory reason applied by either Allstate or Aetna in declining to renew the respective policies, based upon race, color, creed or sex of the policyholders or any similar reason. The evidence established that each insurer made its decision not to renew upon its established underwriting criteria and for no other reason. The facts in each case abundantly established this and there was no evidence to the contrary.

268 Md. at 443-44, 302 A.2d 200 (emphasis supplied).

Government Employees Insurance Company ("GEICO") v. Insurance Commissioner, 273 Md. 467, 330 A.2d 653 (1975), was also a case in which the Insurance Commissioner had ruled that two insurance companies had failed to "justify" their decisions not to renew automobile insurance for two motorists. The circuit court affirmed the Insurance Commissioner but the Court of Appeals reversed, holding that the insurance companies had no such obligation under the then controlling law to "justify" their decisions. Judge Levine's opinion for the Court of Appeals emphasized that substantive burdens on the business decisions of the insurers can only come from legislative action and may not be devised by either the Insurance Commissioner or by the courts.

Nothing in § 240AA ... permits the Commissioner to substitute his underwriting judgment for that of the insurer. Similarly, it is not for the courts to decide whether a driver is a good or poor risk; nor may the courts formulate criteria for the Commissioner to follow in considering whether the action proposed by an insurer is "justified." Such measures ... must come from the Legislature. We have clearly indicated that they can originate neither judicially nor administratively.

273 Md. at 483-84, 330 A.2d 653 (emphasis supplied).

In direct response to the 1973 Allstate decision, the General Assembly had amended § 234A once again. By Chapter 752 of the Acts of 1974, the Legislature, immediately after the words proscribing traditional or historic discriminations, added the provision:

... No insurer, agent or broker may cancel or refuse to underwrite or renew a particular insurance risk or class of risk except by the application of standards which are reasonably related to the insurer's economic and business purposes.

(Emphasis supplied). The GEICO opinion was handed down six months after the 1974 amendment took effect on July 1, 1974, but had no occasion to apply it because the critical decisions being reviewed in the GEICO case had been made before the 1974 amendment took effect. GEICO, therefore, does not help us in interpreting the impact of the 1974 amendment to § 234A.

Within months, however, the interpretation of the 1974 amendment that GEICO failed to provide was provided by St. Paul Fire & Marine Insurance Co. v. Insurance Commissioner, 275 Md. 130, 339 A.2d 291 (1975). This interpretation is critical because the appeal now before us turns exclusively on the question of whether Allstate's business decision to discontinue writing new property insurance in a broad swathe of catastrophe-prone areas is or is not in violation of § 27-501(a)(2), taken directly from former Art. 48A, § 234A(a) as amended by Chapter 752 of the Acts of 1974.

We shall hold that Allstate's decision is not in violation of § 27-501(a)(2) for two separate reasons. In the first place, we hold that § 27-501(a)(2) does not apply to this broad-ranging policy decision by Allstate that did not involve any of the traditionally prohibited forms of discrimination. We also hold, alternatively, that even if § 27-501(a)(2) were, arguendo, deemed to apply, Allstate's decision would not in any event have constituted a violation of that section.

The Threshold Of Section 27-501(a)(2)'s Applicability

The decision of the Court of Appeals in St. Paul Fire & Marine Insurance Company v. Insurance Commissioner is, in our judgment, completely dispositive of the present appeal. It is, indeed, the only appellate opinion to deal with a situation comparable to the one before us. The rest of the caselaw, dealing as it does with decisions to cancel the coverage of an individual motorist because of a bad driving record, or not to renew the coverage of a doctor because of an untruthful application, bears such a strained and attenuated relationship (if any) to what is before us as to be essentially totally useless. We are dealing here, by contrast, with an across-the-board decision by Allstate to discontinue underwriting all new property damage policies in all or in significant parts of eleven of Maryland's twenty-three counties. We are not dealing with Allstate's appraisal of the driving record of an individual motorist. The two diametrically different situations are simply not comparable, and it would take a Procrustean effort to attempt to squeeze the broad-ranging analysis called for in this case into the Lilliputian framework of analysis employed in all of the cases other than St. Paul Fire & Marine.

The St. Paul Fire & Marine situation was comparable in scope to the situation before us. Just as Allstate in the present case became increasingly aware of the financial losses it might be facing in catastrophe-prone areas, St. Paul Fire & Marine, as Judge Levine explained, became increasingly aware of the risks involved in medical malpractice coverage.

With the advent of specialization and greater complexity in the practice of medicine have come increasing demands upon the insurance companies handling medical malpractice insurance. Claims adjusters and defense attorneys in this field must now possess a specialized expertise. Not surprisingly, therefore, these circumstances, together with other developments in medical malpractice litigation, have increased the underwriting risk of this form of coverage, and have reduced the national market to some 12 insurance companies.

275 Md. at 133, 339 A.2d 291 (emphasis supplied).

We see no meaningful distinction between the total termination of a type of coverage in the St. Paul Fire & Marine case and the broad curtailment of writing new coverage within the three hurricane bands in the present case. The rationale of Judge Levine's opinion would apply as surely to § 27-501(a)(2) in the present case as it applied to the 1974 amendment to former § 234A in that case. The thrust of the opinion is that the new language supplements the old language and is not a free-standing criterion.

Faced with past losses and the threat of increasing future losses, St. Paul Fire & Marine announced that "it would cease writing physicians and surgeons professional liability [coverage] in Maryland by January 1, 1975." Id. at 134, 339 A.2d 291. A protesting physician brought the matter before the Commissioner, who ruled that the company "had not met its burden of persuasion to demonstrate that the ... refusal to underwrite or renew [was] justified" under § 234A. Id. at 135, 339 A.2d 291. The Insurance Commissioner in St. Paul Fire & Marine made precisely the type of applicability ruling which the Division now urges upon us as one properly falling within the purview of the Insurance Commissioner. Judge Levine, 275 Md. at 134-35, 339 A.2d 291, summarized the Commissioner's ruling, which the Court of Appeals then vacated as one based upon an inapplicable statutory provision.

The commissioner ruled that § 234A, as amended by chapter 752 of the Laws of 1974, was applicable, and that the company had not met its burden of persuasion to "demonstrate that the... refusal to underwrite or renew [was] justified.... In so ruling, the commissioner observed:

"The statute in question specifically limits the action of an insurer (or agent or broker) in making what is termed `underwriting decisions' (refusal to write a risk or class of risk in the first instance, ...), in three distinct areas. These include:

....

"3. Underwriting decisions not based on `standards which are reasonably related to the insurer's economic and business purposes.'"

He then found that the company had violated standards "2" and "3"[.]

The Circuit Court for Baltimore City affirmed the Commissioner. Before the Court of Appeals, St. Paul Fire & Marine advanced three contentions.

1) Section 234A does not apply because the company has made a general business decision not to underwrite an entire category of insurance, rather than a specific decision involving only a "particular insurance risk or class of risk."

2) The company has met its burden of persuasion of showing its decision to have been by the application of standards which are reasonably related to its economic and business purposes.

3) That § 234A is unconstitutional.

Id. at 135-36, 339 A.2d 291 (emphasis supplied).

Significantly, the Court of Appeals never addressed the latter two contentions because it held, at the threshold, that the statutory provision in question was not even applicable to the type of broad-based business decision made by the insurer.

Since we agree that the statute is inapplicable to this case, it is unnecessary for us to decide the remaining two points.

Id. at 136, 339 A.2d 291 (emphasis supplied). The subject was thus applicability and not satisfaction.

In rejecting the argument that the 1974 amendment to § 234A(a) had created a new and broad substantive power in the Commissioner to oversee the business soundness of decisions affecting coverage generally, Judge Levine's opinion stressed that the amendment did not create an independent section in the Insurance Code but only added a provision to § 234A(a) which was historically an anti-discrimination measure. The new provision was not a free-standing or autonomous measure but simply an incremental addition to a pre-existing measure. In terms of who was covered by the new amendment, the opinion pointed out that the words "risk or class of risk" in the new language replicated precisely the words "risk or class of risk" in the original anti-discrimination measure. Judge Levine summarized the insurer's position.

In pressing its argument that § 234A is inapplicable, the company maintains that the statute is aimed at individual underwriting decisions involving "a particular risk or class of risk," not at general policy decisions, made at the highest corporate level, affecting a compete line of insurance such as medical malpractice coverage. Thus, the company contends that the words "risk or class of risk," within the contemplation of § 234A, actually mean "person or class of persons." ... As the company sees it, therefore, it is free to determine which line of insurance it will offer, but once it elects to underwrite a category or line of insurance, the statute requires that it not discriminate, without justification, between persons or classes of persons.

Id. at 138, 339 A.2d 291 (emphasis supplied).

The Commission, on the other hand, was contending for a much broader reading of the 1974 amendment, essentially the same contention now being made by the Division in the present case.

The commissioner, on the other hand,... argues that the company cannot refuse to underwrite medical malpractice insurance in this state without showing the decision to be based on "standards which are reasonably related to the insurer's economic and business purposes."

Id. at 138-39, 339 A.2d 291.

The Court of Appeals concluded that although the amendment broadened and enhanced the protection being given to certain persons and classes of person, it did not enlarge the ranks of the special group being protected. The Court accepted the company's position.

The language of the first sentence of § 234A itself suggests that "risk," as used in that section, was intended to mean person or applicant.

Clearly, then, the evil aimed at by the statute is discrimination against individuals or classes of individual. When the Legislature added a new standard to § 234A in the 1974 amendment, it did so by inserting the identical language, "particular insurance risk or class of risk," immediately following the prior enactment. This must be regarded as a manifestation of intent that this phrase, as used in the amendment, should have the same meaning that it already had in the original statute. Section 234A, therefore, although now containing broader substantive standards, continues to be aimed at discrimination against individuals or classes of individuals.

Id. at 140, 339 A.2d 291 (emphasis supplied).

What is now § 27-501(a)(2) is simply the second step in a two-step anti-discrimination process. If an action by an insurer would appear presumptively to violate § 27-501(a)(1), it is not enough for the insurer merely to disclaim a discriminatory purpose. The insurer, pursuant to § 27-501(a)(2), then bears the burden of establishing that its decision is "reasonably related to the insurer's economic and business purposes" rather than serving a discriminatory purpose. In GEICO v. Insurance Commissioner, 273 Md. at 483, 330 A.2d 653, Judge Levine explained the supplementary or auxiliary role of the new provision added by the 1974 amendment.

A necessary implication of the provision that a non-renewal not be "for any arbitrary, capricious or unfairly discriminatory reason," as those words are used in § 234A, is that a reason actually exists. We think, therefore, that no insurer refusing to renew a policy can avoid running afoul of §§ 234A and 240AA, if the stated reasons for the proposed action are not actual and true.

(Emphasis supplied).

Step Two in this process does not exist except in conjunction with Step One. If, as the Court of Appeals held, the new 1974 statutory language was inapplicable there, it is inapplicable here.

In the present case, no distinction is being made by Allstate between individuals or classes of individuals within Hurricane Bands 4, 5, and 6. The consequential inapplicability of § 27-501(a) in this case is indistinguishable from the inapplicability of Art. 48A, § 234A(a) in St. Paul Fire & Marine.

In light of the evident purposes of § 234A, the words "particular insurance risk or class of risk" mean that the statute applies to decisions aimed at individual persons or classes of persons, but not to decisions, such as the one here, which concern an entire line of insurance. In short, an insurer may determine independently of § 234A whether it will underwrite a given line of insurance, but once it elects to do so, it must insure every individual or class of individuals desiring such insurance, or else justify its refusal under § 234A.

Id. at 142, 339 A.2d 291 (emphasis supplied). Judge Levine's opinion concluded:

In sum, as its placement under the subtitle "Unfair Trade Practices" also tends to suggest, § 234A is aimed at underwriting decisions which discriminate against an individual or a class of individuals, not at business decisions to withdraw from an entire line of insurance. Since the company has not discriminated against an individual physician or a class of physicians, that is, since its decision does not single out Dr. Levine or any other individual physician or class of physicians, § 234A does not apply.

Id. at 144, 339 A.2d 291 (emphasis supplied).

With respect to this second contention by the Division, our primary holding is that § 27-501(a)(2) of the Insurance Article, at the threshold, does not apply to this broad-based business decision by Allstate that did not involve any of the traditionally prohibited forms of discrimination. If inapplicable, it, ipso facto, could not have been violated.

The Merits, Arguendo, Of Section 27-501(a)(2)

A second reason why the Division may not prevail on this contention is that even if, purely arguendo, § 27-501(a)(2) were deemed to apply to Allstate's decision to stop underwriting new property insurance in Hurricane Bands 4, 5, and 6, our holding even in that event would be that Allstate, on the merits, had fully satisfied the requirements of the section. The Division contends that Allstate failed to satisfy § 27-501(a)(2) in two separate regards: 1) that Allstate failed "to provide any statistical data validating the probability of a catastrophic hurricane occurring in Maryland," and 2) that Allstate failed "to provide statistical data showing that its rating plan in effect was not sufficient to cover losses in the event of a catastrophic hurricane[.]" We will address each of those subcontentions in turn.

A. "Hurricanes Hardly Happen"

The first subcontention is unreal. There is a repetitive drumbeat to the Division's attack on Allstate's compliance with § 27-501(a)(2). In its brief, the Division boldly asserts, "Allstate was legally required to show the probability of a catastrophic hurricane striking Maryland in order to justify its no-write decision." Quoting from the preamble to Chapter 752 of the Acts of 1974, the Division seeks to impose on Allstate an obligation to "measure the probability of a direct and substantial adverse effect upon the losses or expenses of the insurer." Its argument is that "Allstate did not present any evidence showing that Maryland is at risk of a catastrophic hurricane. Allstate presented no historical data of hurricane strikes in Maryland." In its reply brief, the Division goes on "No hurricane has made landfall in Maryland in at least 100 years." (Whatever happened to the legendary 1933 hurricane that hit Ocean City and carved out the Inlet?) The reply brief again states, "Allstate itself concedes that the administrative record is completely bereft of any evidence that Maryland is somehow prone to hurricanes."

We hear in this nostalgic echoes from Lerner and Loewe's "My Fair Lady," as it tells us that "hurricanes hardly happen." Professor Higgins to the contrary notwithstanding, however, hurricanes do happen— frequently and predictably. We can predict, as did Allstate, when they are going to happen and where they are going to happen. In terms of when they will happen, for the Eastern Seaboard of the North American continent, they are going to happen six or eight times a year and, each year, within a June 1 to November 30 window commonly called the "hurricane season." In terms of where they are going to happen (meaning where they will make landfall), they are going to happen on the east coast of North America somewhere between the Yucatan Peninsula and Nova Scotia. Even if we were to narrow that target range to one of between Brownsville, Texas and Bar Harbor, Maine, for a national insurance company such as Allstate it would all be part of a common concern that knows no state boundaries.

For tropical disturbances brewing off the west coast of Africa in the southern hemisphere and six thousand miles away from their ultimate landfalls, this is commendably good prognostication. For the Division to demand that Allstate somehow fine-tune the prognostication to one of an eight percent likelihood that a category-four hurricane will come ashore on Assateague Island in September of 2057 is absurd. The very nature of the meteorological phenomenon is not vulnerable to being thus pinned down.

It is as if the Division, stubbornly hung up over some irrelevant sentence or two in the caselaw about some accident-prone motorist who had his insurance cancelled, refuses to hear what Allstate's three expert witnesses were talking about: a problem massively bigger than Maryland but of which Maryland is nonetheless a small but undeniable part. The syllogism is simple.

THE ENTIRE EASTERN SEABOARD OF THE UNITED STATES IS AT RISK FROM HURRICANES.

MARYLAND IS PART OF THE EASTERN SEABOARD OF THE UNITED STATES.

THEREFORE, MARYLAND IS AT RISK FROM HURRICANES.

In more generic form, the syllogism is:

ALL "A" IS AT RISK.

MARYLAND IS SOME "A."

THEREFORE, MARYLAND IS AT RISK.

Although it is a continental problem of a type to be studied through a telescope, the Division wants it to be examined under a microscope. That is not a scientific way to look at a global phenomenon. We will take judicial notice that Maryland is on the Eastern Seaboard of North America. As such, it is in the gunsights of six to eight hurricanes per year. As a relatively thin slice of a much larger target, Maryland's statistical chances of not being hit in a given year are good. It cannot be predicted, however, when its turn will come. It is, in any event, prudent and fiscally wise to plan ahead for such an eventuality, which is what Allstate is doing.

Appearing before Commissioner Raimondi on December 13, 2007, Robert Grant Newbold was accepted by all parties as an expert witness in computer modeling. He was employed by AIR and worked with the computerized simulation studies that calculated the pattern of losses that might be sustained from hurricane wind damage. He explained, moreover, how the studies and the projections extend necessarily from Texas to Maine.

Q ... Can you explain how the model accounts for a potential tracking of a storm?

A Sure. We use what are called conditional probability matrices to develop the track. Effectively, for each hourly point in a storm's life, we draw from a probability distribution of potential directions that storm could go.

Q And is the entire gulf and east coast measured?

A Yes, all of the way from Texas to Maine.

Q Roughly how far apart are you measuring in terms of each area?

A The model is based on a 50 nautical mile coastal segment, but landfall points are generated each one nautical mile.

(Emphasis supplied).

In talking about one of the projections for Maryland specifically, the witness said that some of the Maryland calculations were based on four of the model storms making landfall in Worcester County, Maryland, but that Maryland would also be affected by other model storms making landfall in Sussex County, Delaware; in Virginia Beach; and in Northampton County, Virginia.

Q Then, after that, it starts talking about four of the model storms were projected to make landfall in Worcester County, Maryland, and that is talking about a Category 4 Storm making landfall in Sussex County, Delaware, and other storms making landfall in Virginia Beach, the City of Virginia, and Northampton County, Virginia. Is that the information that AIR Corporation would have provided to Allstate?

A Yes, that's Event Level Loss Information.

Q So, that's an example of Event Level Loss Information?

A Correct.

(Emphasis supplied).

The study of possible hurricane impact, the expert made clear, of necessity covered twenty-eight hurricane-prone states.

Q Now, am I correct in understanding that the modeling that is performed by AIR Worldwide Corporation on behalf of Allstate, that is not limited on a state-by-state basis, is it? In other words, the modeling, itself, is nationwide?

A It's nationwide as far as states exposed to hurricanes go. It is, approximately, 28 hurricane-exposed states.

Q There is no separate model just for Maryland, let's say?

A No, sir, there is one model.

MR. RAIMONDI: When you use Model 7.0, it covers how many states?

THE WITNESS: I believe it is 28 states, Your Honor. It is the entire eastern seaboard.

MR. RAIMONDI: All of the eastern seaboard is to be using Model 7.0?

THE WITNESS: Yes.

(Emphasis supplied).

Ryan A. Michel, the senior actuary at Allstate, also testified as an expert. He explained that Allstate considers Maryland, Virginia, and Delaware as a single identifiable catastrophe-prone area.

Q Then, further down the same paragraph in Answer No. 8, you said, "The company first looked at Damage Ratios for Virginia, Maryland, and Delaware." Now, what are we talking about when you refer to Damage Ratios? What does that mean?

A Damages Ratios are expected hurricane losses over an amount of insurance, approximately $1,000.00 of insurance. The loss due to hurricanes would have come out of catastrophe modeling.

Q Why were the Damage Ratios first looked at for Virginia, Maryland, and Delaware?

A Because when we are talking about implementation of something like a Tropical Cyclone Deductible or any kind of Catastrophe Management Act, we wanted to take into consideration kind of a region, as a whole, because we have got distribution points and agencies that fall on the border of those areas, for example, who need to write in close proximity of each other, across the boundaries, so we want to have kind of a holistic picture of how we applied those things.

Q Then, you further state there that, "The Damage Ratios were compared and the Hurricane Zones were determined based on continuity across all three states." Do you see that?

A That's correct.

Q What does continuity have to do with it?

A We want to have a consistency, for example, of what we call border agents that you might write in Virginia, as well as Maryland, or in Delaware, that they might have a consistent application of these things.

(Emphasis supplied).

Mr. Michel again explained how the hurricane-related problems cross state lines and can only be understood in that broader geographic context.

Q Has the catastrophe risk analysis changed over the past several years?

A Yes, especially with the 2004-2005 hurricane seasons, Allstate has really stepped up its desire to have a better understanding and grasp on its corporate risk to hurricane and tropical cyclones, in particular.

Q Am I correct that the assessment by Allstate is on a country-wide level?

A That's correct.

Q But the actions here relate only to Maryland, correct?

A Yes. I mean, the hurricanes and tropical cyclones, ... they don't know political boundaries, they are purely geographic events. But from an implementation standpoint, we have to look at each individual state perspective....

Q Now, with respect to Virginia, Maryland, and Delaware that you talked about before, is Allstate implementing the exact same strategy in all three states?

A It is very similar. In fact, when you look at the states that are proximate, Virginia, we have a group of hurricane bands where we are not writing new business. We have Tropical Cyclone Deductibles in place today. In Delaware, we are not writing new business in the entirety of the state. In New Jersey, exactly the same thing, we are not writing business in the entirety of the state there as well.

(Emphasis supplied).

In view of the narrowness of the "Mid-Atlantic" band that embraces Virginia, Maryland, and Delaware and the fact that hurricane-caused damage could occur in all three states regardless of which of those states enjoyed the literal landfall, we do not understand the Division's narrow focus on the history of hurricane's actually making landfall in Maryland. There is also what Robert Newbold referred to in his testimony as a "by-passing storm," which he described as a storm that "does not actually make a landfall on the U.S. coastline, yet it still generates hurricane-force winds over land." The projected damage is not restricted to the state where landfall is made.

Whereas the Division is insisting on data about historic hurricanes that have actually made landfall in Maryland, Mr. Michel explained why that is not the pertinent line of inquiry.

Q ... One question I would like to ask you about, can you explain why Allstate relies on hurricane computer modeling as opposed to historical experience with respect to hurricanes?

A Absolutely. For many of the things we do in regards to the pricing I am responsible for, we look at historical data. However, for major catastrophes, we only have a little over 100 years of data. These are major storms that occur a certain number of times in any given year. So, we really have too small of a data base to perform any kind of credible loss analysis on. As a result, using the computer models, as Mr. Newbold testified to earlier, we can actually get 100,000 years of simulated loss experience, which is good not just for individual state pricing, but all kinds of loss characteristics related to hurricanes, even down to a zip code level.

(Emphasis supplied).

Mr. Michel also explained how the nature of the risk is changing and that a hurricane today could produce losses much greater than those produced by a hurricane of precisely the same characteristics fifty years ago.

[I]n general, the catastrophic risk has increased significantly over the last 100 years because, No. 1, people are moving to the coast. I think the increase over the past 50 years has been over 50% of people moving to the coastal areas of the United States. No. 2, the property values and the increase in the values have been 2 to 10% the last 100 years. So that aggregates quite quickly in terms of our exposure. Even with stopping writing new business, our existing business will increase in exposure simply because the values of the property will increase.

(Emphasis supplied).

A number of exhibits were also submitted to the Commissioner at the two-day hearing. A letter from Allstate in response to a question from the MIA referred to the "Mid-Atlantic states of Virginia, Maryland and Delaware" as an indivisible geographic unit of measurement.

For the mid-Atlantic states of Virginia, Maryland and Delaware, damages ratios at the zip code level were used to establish the hurricane bands. First, Virginia hurricane bands were established from the data, with an adjustment to ensure that the bands were contiguous. Maryland data was then analyzed and bands established in a way that maintained the geographic continuity of the bands across state borders.

Overall, this process is highly data driven and results in risk-based objective hurricane bands.

(Emphasis supplied).

Another response to the MIA from Allstate on January 19, 2007, explained how the simulated computer studies take into account all East Coast hurricanes that have occurred since 1900.

The pattern of simulated hurricanes will match closely the pattern of historical losses because meteorological data on the actual events since 1900 were used to estimate the parameters of the AIR hurricane simulation model. The meteorological sources used to develop the model are the most complete and accurate databases available from various agencies of the National Weather Service and the National Oceanographic and Atmospheric Administration (NOAA), including the National Hurricane Center.

(Emphasis supplied).

Another response to the MIA, also in evidence, pointed to the magnitude of estimated losses in Maryland from hurricanes that made literal landfall in Virginia, in Delaware, and even in North Carolina.

A Category 4 storm making landfall in Sussex County, Delaware would result in Maryland losses of $147,241,509. A Category 2 storm making landfall in Virginia Beach City, Virginia would cause Maryland losses of $145,692,739. A Category 1 hurricane making landfall in Northampton County, Virginia would cause $24,109,824 in Maryland losses. These numbers do not reflect losses that Allstate would incur from the same events in other states, including Delaware or Virginia.

... The AIR model was also used to determine net loss (including loss adjustment expenses) for a past event, were that event to occur today. For example, in 1954, Hurricane Hazel, a category 4 storm, made landfall in North Carolina and made its way up the coast to strike Maryland. If this same event were to occur today, Allstate would incur $307,849,483 in net loss (including loss adjustment expenses) in Maryland. This number does not reflect losses that Allstate would incur from the same event in other states, including North Carolina.

(Emphasis supplied). That same study also showed that a Category 4 hurricane projected to make landfall in Worcester County, Maryland would produce expected losses of $237,189,890.

As an actuarial expert, Mr. Michel testified that the estimated wind damage in Hurricane Band 4 would be 42% higher than would be the average for the rest of the state, 650% higher in Hurricane Band 5, and 1,300% higher in Hurricane Band 6.

Based upon this wealth of evidence, Commissioner Raimondi concluded that Allstate had amply demonstrated that its decision to discontinue writing new property insurance in Hurricane Bands 4, 5, and 6 would "serve its business and economic purpose of reducing its exposure in the event of a catastrophic coastal storm."

Allstate has demonstrated objectively that the zip codes located in Hurricane Bands 4-6 represent the greatest potential for loss in the event of a catastrophic storm when compared to the rest of the State. Under these circumstances, Allstate is not required to quantify the probability of a catastrophic storm. Although Allstate has not quantified its business and economic purpose in terms of the specific amount of exposure Allstate will reduce by refusing new homeowner's business in Hurricane Bands 4-6, by refusing to accept additional insureds in the areas which pose the highest risk in the event of a catastrophe, Allstate will serve its business and economic purpose of reducing its exposure in the event of a catastrophic coastal storm. The additional risk to Allstate associated with accepting new business in the designated high risk areas is self-evident and requires no further statistical validation.

(Emphasis supplied).

Although charged with not having been sufficiently anecdotal about the "Hurricane of '33," for instance, or any other historic hurricane, Allstate's use of the AIR Hurricane Model V7.0 cranked out, zip code by zip code, predictive statistical data for 100,000 model years. We are hard-pressed to understand exactly what more the Division could want.

B. The Insubstantiality of the Crumlish Dicta

In its second subcontention, the Division does not allege that Allstate discriminated in any way against any person or class of persons. It simply challenges the showing of the soundness of Allstate's business decision. The gist of the subcontention is:

Allstate also failed to provide any statistical data showing that its February 2006 rate plan was insufficient to cover losses in the event of a hurricane. It is well-settled that an insurer must show the "direct and substantial adverse effect the supposition would have upon the insurer's losses and expenses in light of its current approved rating plan." Crumlish, 70 Md.App. [182,] 190 [, 520 A.2d 738 (1987)].

(Emphasis supplied). The Division immediately went on to charge that "the Commissioner did not adhere to this Court's decision in Crumlish." (Emphasis supplied)

Quite aside from characterizing as "well-settled" something that is far from settled and bestowing the honorific of "this Court's decision" on what was overtly nothing more than dicta, what is this Crumlish to which the Division does obeisance? To tip our hand at the outset, we note that had the Commissioner's decision on this issue gone in the other direction, Allstate would have had a very strong cross-appeal charging that the Commissioner was in error for following the Crumlish dicta.

In Crumlish, the insurer had cancelled a motorist's coverage because of two collisions by the insured resulting in property damage. The Commissioner affirmed the decision of the insurer and the circuit court affirmed the decision of the Commissioner. The Court of Special Appeals, however, vacated the order of the Commissioner and remanded the case to the Commissioner because of the failure of the original order to contain a "concise statement of the facts as found by the Commissioner and his conclusion therefrom" as required by Art. 48A, § 39. 70 Md.App. at 187, 520 A.2d 738. The Crumlish opinion then went on:

Because of our holding it is not necessary for us to reach the specific issue of whether [the insurer] provided evidence that the underwriting standard was reasonably related to its business and economic purpose; however, for the guidance of the Insurance Commissioner, on remand, we will address that issue.

Id. at 188, 520 A.2d 738 (emphasis supplied).

After opining that the standard the insurer was using was not reasonably related to its economic and business purposes, as ostensibly required by Art. 48A, § 234A, Crumlish elaborated on what must be proved to establish that an underwriting standard is adequately related to the insurer's economic and business purposes:

Conclusory statements that the standard does relate are not enough to comply. Facts must be produced which answer at least the following questions:

1. What is the statistical basis for the supposition that a person who has two or more chargeable losses within a 24 month period is more likely to have a chargeable accident within the next 12 months than a person who has had no accidents, one chargeable accident, or two or more nonchargeable accidents?

2. How valid is any such statistical evidence?

3. If there is statistical validity to the supposition, what direct and substantial adverse effect would it have upon [the insurer's] losses and expenses in light of the current approved rating plan?

Id. at 190, 520 A.2d 738 (emphasis supplied).

The Division's present charge that Allstate failed to show the "direct and substantial adverse effect the supposition [about hurricanes] would have upon [its] losses and expenses in light of its current approved rating plan" is verbatim out of that Crumlish dicta and not out of former Art. 48A, § 234A(a) or out of present § 27-501(a)(2). It was, moreover, from the Crumlish dicta that the Division got its obsession with statistical data.

What the Crumlish dicta did was to construct an elaborately detailed regulatory scheme out of thin air. Such judicial rule making, however, was in direct contravention of Judge Levine's overarching admonition in GEICO v. Insurance Commissioner, 273 Md. 467, 483-84, 330 A.2d 653 (1975), that should be the starting point for every analysis:

[I]t is not for the courts to decide whether a driver is a good or poor risk; nor may the courts formulate criteria for the Commissioner to follow .... Such measures ... must come from the Legislature. We have clearly indicated that they can originate neither judicially nor administratively.

(Emphasis supplied).

It did not take long for subsequent caselaw to question the universality of the Crumlish dicta. In Miller v. Insurance Commissioner, 70 Md.App. 355, 521 A.2d 761 (1987), the General Accident Insurance Company of America had cancelled a doctor's professional liability insurance. The Commissioner affirmed the action of General Accident and the circuit court affirmed the Commissioner. On appeal to this Court, the doctor contended that General Accident had not satisfied Art. 48A, § 234A by showing "`the application of standards which are reasonably related to the insurer's economic and business purposes.'" 70 Md.App. at 369, 521 A.2d 761 (relying on Lumbermen's Mutual Casualty Company v. Insurance Commissioner, 302 Md. 248, 487 A.2d 271 (1985)). Despite the lack of any statistical data to support the validity of General Accident's cancellation decision, this Court affirmed the soundness of the decision based on an applicant's deception or misrepresentation.

We do not read that case [Lumbermen's] as authority for the proposition that an underwriting standard providing for cancellation of a policy obtained by material misrepresentations is not reasonably related to the insurer's economic and business purposes.

70 Md.App. at 370, 521 A.2d 761. See also Erie Insurance Company v. Insurance Commissioner, 84 Md.App. 317, 579 A.2d 771 (1990).

In Mirkin v. Medical Mutual Liability Insurance Society, 82 Md.App. 540, 572 A.2d 1126 (1990), Medical Mutual had cancelled the professional insurance of a doctor whose billing records were admittedly inaccurate. The Commissioner upheld the action of Medical Mutual, but the circuit court ruled in favor of the doctor, who argued that Medical Mutual had failed to "present statistical evidence which satisfies the three part test employed in Crumlish." This Court, speaking through Judge Karwacki, overruled the circuit court and confirmed that statistical data or mathematical verification is not a universally required sine qua non.

Dr. Mirkin asserts that appellee's justification for the standard is not sufficient; that in order to comply with § 234A an insurer must present statistical evidence which satisfies the three part test employed in Crumlish v. Ins. Comm'r, 70 Md.App. 182, 190, 520 A.2d 738 (1987). In employing this three part test the Crumlish court, as the Court of Appeals noted in Muhl v. Magan, 313 Md. 462, 475-76 [545 A.2d 1321] (1988), "presupposed that the statute could be satisfied only by an objectively demonstrable, statistical basis for underwriting." There are, however, circumstances in which this is not possible. See, e.g., Miller v. Ins. Comm'r, 70 Md.App. 355, 521 A.2d 761 (1987) (An underwriting standard providing for the cancellation of a policy obtained by a material misrepresentation was held to be reasonably related to the insurer's economic and business purposes without statistical proof.). There are obviously some underwriting standards whose fairness cannot be demonstrated through statistics but are nonetheless "reasonably related to the insurer's economic and business purposes." The standards employed in this case and in Miller, supra, provide useful examples. Each of these underwriting standards reflects the insurer's reluctance to insure what it justifiably considers questionable risks. The fact that the suitability of these standards cannot be mathematically verified does not foreclose their being "reasonably related to the insurer's economic and business purpose." We do not read § 234A as requiring insurers to utilize only those underwriting standards capable of statistical validation.

82 Md.App. at 550-51, 572 A.2d 1126 (emphasis supplied). See, however, Medical Mutual Liability Insurance Society v. Magan, 72 Md.App. 330, 336-38, 529 A.2d 841 (1987), and Insurance Commissioner v. Nevas, 81 Md.App. 549, 557-58, 568 A.2d 1144 (1990), which followed the Crumlish dicta.

The Crumlish dicta was in a very real sense doomed to go wrong from the start because of its heavy reliance upon a secondary discussion in Lumbermen's Mutual Casualty Company v. Insurance Commissioner, 302 Md. 248, 487 A.2d 271 (1985), which was itself palpably unreliable. The basic decision in Lumbermen's, of which we have no criticism, turned upon the misuse by insurers of Art. 48A, § 234A when their proper approach should have been through § 242, dealing with rating plans. The actual holding in Lumbermen's was clear.

[W]hen an automobile liability insurer's complaint is over the adequacy of its filed and approved rating plan, § 242 is the specific statute, dealing with the matter, and the controversy is not encompassed by §§ 234A and 240AA.

If the two insurers in these cases desired greater surcharges for insured drivers having more than one traffic violation or accident in a three year period, they could have applied for such modification of their rating plans under § 242. Or, if the insurers decided that they did not wish to insure such risks, they could similarly have sought to modify their rating plans by deleting this classification of risks. But, in our view, proceedings under §§ 234A and 240AA were not intended to encompass challenges to the adequacy of rating plans.

Id. at 269, 487 A.2d 271 (emphasis supplied).

In earlier contrasting § 242 with § 234A, however, the opinion had recited some of the history of § 234A, including Chapter 752 of the Acts of 1974 which amended it.

By Ch. 752 of the Acts of 1974 ... the Legislature amended § 234A and went far beyond a proscription of discrimination in underwriting.

Id. at 254, 487 A.2d 271 (emphasis supplied).

The flaw in Lumbermen's, and consequently in the Crumlish dicta, is with respect to the mishandling of the preamble to Chapter 752 of the Acts of 1974. House Bill 859, which introduced the Act, originally contained the following language:

NO INSURER, AGENT OR BROKER MAY CANCEL OR REFUSE TO UNDERWRITE OR RENEW A PARTICULAR INSURANCE RISK OR CLASS OF RISK EXCEPT BY THE APPLICATION OF STANDARDS WHICH MAY BE DEMONSTRATED OBJECTIVELY TO HAVE A DIRECT AND SUBSTANTIAL EFFECT UPON LOSSES OR EXPENSES.

(Emphasis supplied).

Accordingly, the title of H.B. 859 included the phrase "objective standards" and the preamble to the bill announced the legislative purpose that insurers' underwriting decisions must

be made solely on the basis of a reasonable application to relevant facts of underwriting principles, standards and rules that can be demonstrated objectively to measure the probability of a direct and substantial adverse effect upon losses or expenses of the insurer in light of the approved rating plan or plans of the insurer then in effect[.]

(Emphasis supplied).

On the last calendar day of the 1974 legislative session, however, the State Senate amended the operative language (and the House of Delegates concurred) by deleting from the bill the words "may be demonstrated objectively to have a direct and substantial effect upon losses or expenses" and by substituting therefor the words "are reasonably related to the insurer's economic and business purposes." In the obvious crush of the rush toward adjournment, however, the Legislature, by legislative oversight, neglected to change the bill's preamble to bring it into conformity with the version of the law that ultimately passed.

A preamble, of course, is a brief preliminary statement of what a bill is intended to do. To the extent that the ultimate act does not do that or does something else, the original preamble, if not updated, loses its value as a tool of statutory construction. The interest is in what the act ultimately does, not in what it was once intended to do but didn't.

Lumbermen's nonetheless quoted the superseded and inaccurate ghost preamble, 302 Md. at 254, 487 A.2d 271, and used the excised language in its discussion of legislative intent, Id. at 267-68, 487 A.2d 271. The Crumlish dicta took its lead from Lumbermen's reliance on the inaccurate preamble and followed suit. One of the three precise requirements that it created used verbatim the very language that had been excised from the Act by the Legislature.

Facts must be produced which answer at least the following questions:

....

3. If there is statistical validity to the supposition, what direct and substantial adverse effect would it have upon [the insurer's] losses and expenses in light of the current approved rating plan?

70 Md.App. at 190, 520 A.2d 738 (emphasis supplied).

The final conclusion of the Crumlish dicta is framed in the verbatim language of the 1974 law that never was.

The preamble to Ch. 752 to which Judge Eldridge referred supra, and which the Court implicitly adopted as the legislative history of the subsection of the statute involved in the case sub judice, leads us to conclude that the statute clearly requires that [the insurer] demonstrate objectively "the probability of a direct and substantial adverse effect upon losses or expenses of the insurer in light of the approved rating plan or plans of the insurer then in effect...."

Id. at 190, 520 A.2d 738 (emphasis supplied).

The Division in this case falls into precisely the same error. It characterizes the flawed preamble to Chapter 752 of the Acts of 1974 as "the preamble to [the present] § 27-501." After quoting the phantom preamble, it contends, in the rejected language of 1974, "It simply is not possible to `measure the probability of a direct and substantial adverse effect upon the losses or expenses of the insurer' in the absence of any evidence on that subject —the `probability of the effect' occurring." (Emphasis supplied). The Division charges that "the Commissioner did not adhere to this Court's decision in Crumlish."

Let us make it absolutely clear that the definitive interpretation of both the meaning and the function of the 1974 amendment to what was then § 234A (and what is now § 27-501(a)(2)) remains Judge Levine's carefully crafted analysis for the Court of Appeals in St. Paul Fire & Marine Insurance Co. v. Insurance Commissioner (1975) and not the more casual references appearing in either Lumbermen's Mutual Casualty Co. v. Insurance Commissioner (1985) or Crumlish v. Insurance Commissioner (1987).

To forestall the risk of any further relapses into such fantasy analysis, we hereby expressly repudiate the Crumlish dicta. The Division, in reply brief, protests that "[e]ven if Crumlish's interpretation of Section 27-501 was dictum when offered, the Crumlish test has become a settled principle in insurance regulation, and the MIA has consistently applied it prior to this case." The notion of a "settled principle" may only reflect the fact that until the present case arose, there had been no appellate activity dealing with this area of law for over 20 years. If Crumlish was wrong in 1987, however, as it most assuredly was, it is still wrong 24 years later. Longevity does not make it right. In the words of Oliver Wendell Holmes in "The Path of the Law," 10 Harv. L.Rev. 457, 469 (1897):

It is revolting to have no better reason for a rule of law than that so it was laid down in the time of Henry IV.

Even if § 27-501(a), arguendo, applied in this case, we would see no error in the Commissioner's decision that Allstate had fully satisfied it.

The Unique Nature of Catastrophic Risk

In its insistence that venerable tests and measurements that have traditionally been applied to individual and ad-hoc risks, such as a particular motorist or a particular medical practitioner, should also be applied to catastrophe risk, the Division well stated that position in its reply brief.

Maryland law does not recognize "catastrophe risk" as different from any other type of risk for which insurance companies underwrite. Neither the MIA nor Allstate cites a single case, statute, rule or regulation that stands for this proposition.

That is, indeed, true, because this is the first case that has had to deal with catastrophe insurance. A sound business decision, however, can readily deal with new problems (or newly appreciated or more fully understood problems) in new ways. The fascinating development of the present case will illustrate, perhaps for the first time, the difference between short-term and long-term insurance problems and solutions and the gaping difference between ordinary insurance risk and catastrophe risk. At the hearing before the Commissioner, Ryan Michel, the senior actuary for Allstate, testified:

Because we could not charge a price that would account for the short term volatility of a risk. So, for many types of insurance, the more insurance that you bring in, you diversify your risk. So, the risk becomes less and less in any particular event. However, catastrophe risk is different. The more and more insurance that you are providing in one specific area, the greater and greater volatility that you have.

(Emphasis supplied).

Commissioner Raimondi found as a fact the diametric difference in the treatment of ordinary risk and the response to catastrophic risk.

As Allstate's actuary, Ryan Michel, explained at the hearing, rates cannot account for the short term volatility associated with a hurricane strike. For many types of insurance risk, the larger the number of the insureds, the greater the diversity of risk. Catastrophic risk is different. When the concern is catastrophic risk, the opposite occurs. The increased number of insureds in the catastrophe-prone area increases the volatility of the loss in the event of a hurricane strike.

(Emphasis supplied).

David Chernick, the actuarial consultant, also noted the diametric difference in the handling of the respective risks.

For insurance companies that write property products, however, there is also another huge risk and that is the risk of catastrophes. The risk of catastrophe is a little bit different than the concept I spoke about earlier, because adding additional catastrophe risk will not reduce the overall risk by pooling arrangement. It actually increases the overall risk of the company writing catastrophe policies.

(Emphasis supplied).

Commissioner Raimondi also accepted this analysis as a matter of fact.

Mr. Chernick offered testimony regarding how an insurance company uses underwriting guidelines to implement enterprise risk management. Although catastrophe risk is one of the risks that a property insurance company must consider, Mr. Chernick explained that catastrophe risks are different from other insurance risks. Mr. Chernick explained that adding additional catastrophe risks will not reduce the overall risk by a pooling arrangement. The additional risks in the catastrophe prone area actually increase the overall risk of the company writing the catastrophe policies.

(Emphasis supplied).

With respect to this subcontention, we hold again that, even if, arguendo, § 27-501(a)(2) applied, we would hold that it had been fully satisfied.^[5]

Conclusion

The difference in magnitudes of risk between Hurricane Katrina and Katrina Abramowitz, with two traffic infractions and three points on her driving record, is so vast as to be incomprehensible. Even to attempt to describe the one in terms of the other would be gibberish. An automobile collision or a botched operation is not a catastrophe.

JUDGMENT AFFIRMED; COSTS TO BE PAID BY APPELLANT.

NOTES

[1] Section 19-107 prohibits an insurer from refusing to underwrite certain geographical areas unless the geographic designation "has an objective basis and is not arbitrary or unreasonable." Section 27-501 prohibits an insurer from refusing to underwrite insurance risks for discriminatory reasons. Both sections are discussed more fully, infra.

[2] Allstate cited "the steadily increasing concentration and value of insured properties in harm's way along the coastline and waterways of Maryland" as a significant contributor to Allstate's risk exposure in Bands 4, 5, and 6.

[3] The record before the Commissioner indicated that AIR's Atlantic Tropical Cyclone Model V7.0, presumably the same model as AIR Hurricane Model V7.0, was determined by the Florida Commission on Hurricane Loss Projection Methodology as "sufficiently accurate and reliable for projecting hurricane loss costs for residential property in Florida" on June 3, 2005. The Commissioner, in his February 2008 order, specifically found that "Allstate's use of computer-generated hurricane model data [was] justified and the data yielded from the use of the AIR models [was] reliable."

[4] Allstate explained this methodology as minimizing the "within variance," or the difference between zip code damage ratios in a single band, and maximizing the "between variance," or the difference between the ADRs between each band, while maintaining geographic continuity. The bands were constructed in this manner in order to "pool risks that have similar characteristics together."

[5] Allstate has moved to strike all arguments and references made by the Division in relation to the February 2006 filing, as that filing was not part of the administrative record before the Commissioner. The motion is granted. We have not relied on this material in forming our judgment here.