We granted certiorari to determine whether the Maryland Insurance Commissioner properly approved a filing by Allstate Insurance Company and Allstate Indemnity Company (collectively Allstate) giving notice of its intent to cease writing new property insurance policies in certain geographic areas of the State. In a memorandum and order filed on his behalf by an Associate Deputy Commissioner, the Commissioner concluded that the filing was subject to administrative review under Maryland Code, §§ 19-107(a) and 27-501(a) of the Insurance Article (INS) and that it satisfied the pertinent criteria under both statutes.

In an action for judicial review filed by petitioner, People’s Insurance Counsel Division (PICD), a unit within the Office of the Attorney General, the Circuit Court for Baltimore City affirmed both aspects of the Commissioner’s ruling. On PICD’s appeal, the Court of Special Appeals affirmed the Circuit Court judgment, but its conclusion as to § 27-501 rested on alternative grounds — first, that the statute did not apply to the filing and second, that, even if it did, the statute was not violated. People’s Insurance v. Allstate, 199 Md.App. 1, 20 A.3d 117 (2011). We shall hold that § 27-501 does apply *447to the Allstate filing and, with that caveat and for the other reasons expressed in this Opinion, we shall affirm the judgment of the Court of Special Appeals.

BACKGROUND

Section 19 — 107(a), which is part of the Title of the Insurance Article dealing with property and casualty insurance, provides, in relevant part:

“An insurer may not refuse to issue or renew a contract of ... homeowners insurance ... solely because the subject of the risk or the applicant’s or insured’s address is located in a certain geographical area of the State unless:

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

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

Title 27 of the Insurance Article deals generally with unfair trade practices in all lines of insurance. Subtitle 5, of which § 27-501 is a part, prohibits a range of discriminatory practices. 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 risk or class of risk except by the application of standards that are reasonably related to the insurer’s economic and business purposes.”

On February 13, 2006, Allstate filed with the Maryland Insurance Administration (MIA) a new underwriting rule stating that, from and after April 10, 2006, due to “catastrophe management actions,” it would no longer write new homeowners insurance policies on properties lying within one mile *448of the Atlantic Ocean. That was significantly expanded, and effectively made moot, on December 4, 2006, when Allstate filed a new underwriting rule stating that, effective January 1, 2007, it would cease writing new homeowners and renters policies in all ZIP Codes located within what it defined as Hurricane Bands 4, 5, and 6. The exclusion, as so identified, would include all of St. Mary’s, Somerset, Talbot, Wicomico, and Worcester Counties and significant parts of Anne Arundel, Calvert, Charles, Dorchester, Prince George’s, and Queen Anne’s Counties.

Allstate agreed to delay the effective date of the new rule until MIA had an opportunity to review it. Thereafter, MIA posed a number of questions to Allstate regarding the filing, which Allstate answered. On May 31, 2007, MIA advised Allstate that it had reviewed the filing under § 19-107 and, based on the information submitted, concluded that the geographic designation had an objective basis and was neither arbitrary nor unreasonable. No mention was made in the letter about INS § 27-501(a). The next day, PICD, which had received a copy of the filings, requested a hearing, which the Commissioner granted, and a stay, which the Commissioner denied.1 Allstate was permitted to intervene. With the denial of the requested stay, the new underwriting rule was put into effect on June 4, 2007; Allstate stopped accepting new applications on that date.

The hearing was conducted by the Associate Deputy Commissioner (who, for the sake of convenience, we shall refer to as the Commissioner) on December 13 and 14, 2007.2 Three *449principal issues were addressed. The first, upon Allstate’s challenge, was whether the PICD had standing to request a hearing. The Commissioner concluded that the Division had standing, and that issue is not before us.3 The second was whether, in a filing under § 19-107, the general prohibition against discrimination in underwriting in § 27-501(a) also was relevant and, if so, who had the burden of proof with respect to both sections. In seeming contrast to the initial letter of acquiescence in the filing, the Commissioner concluded that § 27-501 was applicable and that Allstate had the burden of proof to show that its filing complied with both sections. The third, which proceeded from the resolution of the second, was whether Allstate had met its burden, largely through the presentation of computer modeling evidence, and the Commissioner found that it did.

Most of the evidence presented at the hearing came from the testimony, some of it pre-filed, of three expert witnesses called by Allstate — Robert Newbold, an expert in computer modeling employed by Applied Insurance Research, Inc. (AIR), Ryan Michel, an Allstate actuary, and David Chernick, an expert in actuarial science employed by Milliman, Inc.— along with documents supporting their testimony.

At the hearing, Allstate sought to establish, and, in the view of the Commissioner did establish, several sequential propositions. The first was that managing catastrophic risk presents a unique challenge to insurance companies, for at least two reasons, one dealing with the distribution of the risk and the other dealing with the ability to measure the risk. As to the first, Mr. Michel explained that, for many types of insurance, risk is diversified. The more policies that are written, the less the overall risk in any particular event. Catastrophic risk is different, he said — the more insurance written in a specific area that may be subject to a catastrophic event, the greater *450the volatility. Mr. Chernick confirmed that point — that adding additional catastrophe risk does not reduce overall risk because of pooling but actually increases the overall risk.

With respect to the ability to measure the risk, Mr. Newbold, in his pre-filed testimony, explained that natural catastrophes can cause devastating damage to homes and businesses. Because such occurrences are rare, however, historical loss data is scarce, making it difficult to estimate losses. From 1900 to 2004, he said, there were only 163 hurricanes that made landfall in the United States. Moreover, given the changing landscape of insured properties, even the historical data that exists has limited usefulness. Rules of thumb based on previous loss experience constitute an inexact approach that may either understate the risk, leaving insureds vulnerable to loss, or overstate the exposure, resulting in an inflated cost.

Mr. Newbold observed that, because of the lack of reliable information, it has become standard practice for insurance companies to use catastrophe models to anticipate the likelihood and severity of potential future catastrophes before they occur. Those kinds of models can be used to address such questions as where future catastrophes are likely to occur, how big they are likely to be, how often they are likely to occur, what level of loss a company can expect to incur in each year over a long term, and the probability of incurring a large loss in the current year. Many insurance companies, he said, find catastrophe models “to be essential tools for decision-making regarding risk selection, ratemaking, underwriting, risk transfer, loss mitigation and portfolio management.”

Mr. Newbold listed a number of advantages to computer modeling:

(1) It was able to capture the effects on catastrophic loss distribution of changes over time in population patterns, building codes, amounts insured, and construction costs;

(2) It provides a complete picture of the probable distribution of losses rather than just estimates of probable maximum losses;

*451(3) Because simulation models can be tested more easily than other approaches, it leads to greater stability in estimating expected annual losses;

(4) It provides a means to determine the impact of new scientific information; and

(5) It provides a framework for performing sensitivity analyses and “what-if ’ studies.

Mr. Michel confirmed some of that testimony. He noted, in regard to pricing, that he does look at historical data but that, for major catastrophes, there was only 100 years of data, which was too small a base to perform a credible loss analysis. By using computer models, they can get 100,000 years of simulated loss experience, which is good not just for Statewide pricing but also for loss characteristics related to hurricanes down to the ZIP Code level. About 10 years earlier, he said, Allstate began using the computer models created by AIR because it provided that level — 100,000 years — of simulated loss information.

Mr. Newbold described how the model actually works. We need not recite all of the technical details. The model simulates hurricanes from the genesis point to the point where they no longer generate 40 miles per hour of wind speed. For each simulated year, it determines the number of landfalls that will occur that year, which is checked against the historical data, and, using a probability distribution, determines the location of landfall. Hurricane severity, he said, depends principally on five variables — minimum central pressure, the radius of maximum winds, the forward speed, the angle at which the storm enters the coast, and the track of the storm after landfall — and values are determined for each of those factors. In essence, that aspect of the model estimates the force to which structures in the subject areas will be exposed.

A second phase of the model estimates the damage likely to be suffered. AIR experts developed mathematical functions that describe the interaction between buildings and their contents and the local intensity to which they are exposed. These “damageability relationships,” incorporating well-docu*452mented engineering data, estimate a complete distribution around the mean level of damage for each local intensity and structural type. Losses are calculated by applying the appropriate damage function to the replacement value of the insured property. These calculations are continually monitored based on engineering literature and post-disaster field studies. According to Mr. Newbold, AIR has analyzed over $10 billion of actual claims data from recent hurricanes. In a nutshell, Mr. Newbold said that “AIR’s probabilistic catastrophe models simulate thousands of scenario years of catastrophic events, and calculate loss estimates for each one.” The statistics generated by that analysis, he said, include expected loss estimates and loss exceedance probabilities.

The model is updated annually. The one used by Allstate in this proceeding was Model 7.0, which is used in 28 States, including the entire Eastern Seaboard.

The decision by Allstate to cease writing new property insurance policies in Hurricane Bands 4, 5, and 6 was based on the AIR model. Allstate took the top 5% of events from the AIR model and calculated “damage ratios” for each ZIP Code in the State (as well as in contiguous States). The damage ratio is the amount of annual loss per $1,000 of insurance. Thus, if the expected annual loss is $100 and $1,000 of insurance is in force, the damage ratio would be 0.1. Each ZIP Code was assigned to one of six Hurricane Bands based on its damage ratio. The bands were designated so that (1) they comprised contiguous ZIP Codes, (2) the variances between the ZIP Code damage ratios within the band were minimal, and (3) there was maximum variance between the ZIP Codes in the band and those outside it. Allstate then calculated the average modeled loss for a $300,000 home (excluding deductibles) by band and the loss per $1,000 of coverage for the average home in each band as compared to the ratio for the entire State, which it characterized as the “damage ratio relativity.”

This data was presented in the following form:

*453 Homeowner Band Average Damage Ratio Average Hurricane Lossl$300,000 Home Damage Ratio Relativitg

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

What the data from the model thus shows is that the average damage expected in Band 4 for a $300,000 home will be 1.4 times that in the State at large. In Band 5, the damage for such a home will be nearly 7.5 times that in the State at large, and in Band 6, it will be more than 13 times that for the State as a whole.

Mr. Michel and Mr. Chernick both opined that the designation of Hurricane Bands 4, 5, and 6 had an objective basis within the meaning of INS § 19 — 107(a)(2) because they are defined by ZIP Code and thus are easily determined, and that the designations were not arbitrary because they were clearly related to the risk of hurricane losses. In its initial review of the filing, MIA asked Allstate to respond to a number of questions, including what other business options — underwriting, marketing, reinsurance — Allstate had considered in lieu of ceasing to write new policies in the designated areas. That was addressed at the hearing, principally by Mr. Chernick.

Mr. Chernick stated that Allstate could not manage its Maryland catastrophe exposure simply by purchasing reinsurance. In light of the amount of reinsurance that would be needed, given Allstate’s significant market share, Allstate would need a consortium of 15 to 20 reinsurers, but there is not the capital available to purchase that amount of insurance. After Katrina, the cost of reinsurance soared. In addition, he pointed out that reinsurance is negotiated and purchased on an annual basis, and the market was unstable. “There is no guarantee,” he said, “that reinsurance is going to be available, especially for a company Allstate’s size, especially at the layers we are talking about here ...” Finally, he observed that the cost of reinsurance would have to be reflected in Allstate’s premiums, which would require rate filings and extended regulatory proceedings.

*454Mr. Chernick also explained why, in his view, additional catastrophic risk could not be managed through increased premiums. He gave as an example a home that suffers a total loss of $300,000. The premium for the insurance is $1,000, of which $250 represents coverage for the catastrophic risk. At that rate, it would take Allstate 1200 years to recover the loss. Even if Allstate were to increase the premium 10-fold and charge $2,500 instead of $250 for the catastrophic risk, it would take 120 years collecting that premium to recover the loss.

The only witness produced by PICD was Anthony Grippa, an actuarial consultant in Florida. Mr. Grippa admitted that he had no experience whatever in catastrophic models. The Commissioner accepted him as an expert in actuarial science but would not allow him to express any opinion with respect to the model that formed the basis of Allstate’s amended filing. Mr. Grippa agreed that the designation of the various hurricane bands had an objective basis but contended that the designation was arbitrary or unreasonable because of “the absence of any quantifiable goal in implementing the filings and the absence of any quantitative analysis with respect to the tools.” When asked to elaborate, he said that the goal is catastrophic risk reduction and there was “no measurement that I have seen that compares catastrophic risks to some benchmark, some surplus, some premium fund.” He added that the decision not to write new policies in the designated areas was unreasonable “because there is no showing that it is reasonable.”

When asked about the tools to reduce risk, Mr. Grippa identified four — no new business, non-renewal of existing policies, reinsurance, and higher deductibles — but, when asked which of them, in his opinion, Allstate should have used, he declined to choose, claiming only that Allstate should have shown some reason for the tool it selected. He later agreed that damage ratios were a legitimate means of drawing distinctions between areas of risk and that it was an appropriate actuarial measure to consider variances within the bands and between the bands, as Allstate had done.

*455In his Memorandum of Law and Final Order, the Commissioner summarized the evidence and, as noted, concluded that (1) PICD had standing, (2) INS § 27-501(a) applied and Allstate had the burden of satisfying the criteria under that statute as well as under INS 19-107(a), and (3) it had met that burden. As noted, we are not concerned here with the first of those rulings

Rejecting Allstate’s contention that INS § 27-501(a) does not apply because § 19-107(a) makes no reference to it, the Commissioner concluded that Allstate’s determination to cease writing new homeowner’s policies in the designated areas, aimed at reducing risk in an effort to maintain overall financial strength, constituted an underwriting decision, and, because Allstate will apply underwriting standards in order to implement that decision, those standards are subject to regulatory scrutiny. Under § 27-501(a), when an insurer refuses to underwrite an insurance risk, it must do so based on standards that are reasonably related to the insurer’s business and economic purposes and are not discriminatory.

Noting that § 19-107(a) contains no direct prohibition against discrimination, the Commissioner was concerned that, if § 27-501(a) did not apply, an insurer would be able to make geographic designations that were both objective and not arbitrary or unreasonable but which gave effect to racial or other discrimination prohibited by § 27-501. In support of his conclusion, the Commissioner examined the legislative history of the two statutes and determined, in the end, that “absent some clear expression of legislative intent to carve out underwriting standards implemented through § 19-107 filings from the general applicability of § 27-501, the two statutes must be read together.

In considering whether Allstate had met its burden of showing compliance with both statutes — the third basic issue — the Commissioner addressed a number of points made in PICD’s post-hearing submission, the principal one being that Allstate had failed to produce statistical evidence showing why cessation of new business would accomplish its economic and *456business decision. Although noting that in some cases insurers have been required to provide statistics to demonstrate the relationship between the underwriting standard applied and the insurer’s business and economic purposes, as directed in Crumlish v. Insurance Comm’r, 70 Md.App. 182, 520 A.2d 738 (1987), the Commissioner concluded that that kind of evidence is not required in all cases. The standard under § 27-501(a), he held, is whether the relationship is objectively demonstrable, and that there was no exact formula by which that requirement can be satisfied.

The Commissioner’s ultimate findings were that (1) the process used by Allstate to arrive at its business decision produced an underwriting decision that is reasonable and supported by reliable data, (2) it therefore complied with § 27-501(a), and (3) the geographic designations had an objective basis and were not arbitrary or unreasonable, and (4) they therefore complied with § 19-701(a).

DISCUSSION

Issues; Standards of Review

PICD, the petitioner before us, agrees with the Commissioner that INS § 27-501(a) applies to Allstate’s filing but argues that, by “relieving” Allstate of its burden to provide statistical validation of its “asserted justification for otherwise impermissible geographic discrimination,” the Commissioner erred as a matter of law. Allstate has a contrary view. It believes that the Commissioner erred in holding that § 27-501(a) applied but that he otherwise reached the right result. The Commissioner, of course, finds no error of any kind in his decision.

In Insurance Com’r for the State v. Engelman, 345 Md. 402, 411, 692 A.2d 474, 479 (1997), we confirmed that “a final order of the [Insurance] Commissioner must be upheld on judicial review if it is legally correct and reasonably supported by the evidentiary record.” We added that, out of deference to agency expertise, “we are required to affirm an agency’s findings of fact, and its application of law to those facts, if *457reasonably supported by the administrative record, viewed as a whole,” but that we “are under no constraint to affirm an agency decision premised solely upon an erroneous conclusion of law.” Id.

Application of INS § 27-501(a)

The question of whether § 27-501(a) is applicable to Allstate’s amended filing is a threshold one in terms of PICD’s challenge to the Commissioner’s ultimate decision. PICD’s principal complaint is that the Commissioner erred in not requiring Allstate to produce statistical evidence showing a probability that the designated geographic areas actually would be subjected to hurricanes of the force feared by the company. That kind of evidence, it argues, is necessary to demonstrate that the underwriting standard is reasonably related to Allstate’s economic and business purposes, a requirement found in § 27-501(a)(2).

Whether that statute applies and, if so, how it should be interpreted, are questions of law, and, although an agency’s interpretation and application of a statute that it administers ordinarily is given considerable weight by reviewing courts, the court must make the ultimate legal determination. Thomas v. State Retirement, 420 Md. 45, 54-55, 21 A.3d 1042, 1047 (2011); Grasslands v. Frizz-King, 410 Md. 191, 204, 978 A.2d 622, 629 (2009); AT&T v. Comptroller of Treasury, 405 Md. 83, 92-93, 950 A.2d 86, 92-93 (2008).

After considering the textual language and the legislative history of §§ 19-107(a) and 27-501(a), both of which are part of the Insurance Code administered by the Commissioner, the Commissioner concluded that both statutes were applicable— that the Legislature did not intend to exclude from the anti-discrimination and nexus requirements of § 27-501(a) a filing made under § 19-107. The Court of Special Appeals, examining the same textual language and legislative history, came to a directly contrary conclusion, believing that this Court’s decision in St. Paul Fire & Mar. Inc. Co. v. Ins. Comm’r, 275 Md. 130, 339 A.2d 291 (1975) (St.Paul) was “completely dispos*458itive.” People’s Insurance v. Allstate, supra, 199 Md.App. at 20, 20 A.3d at 127-28.

There are three strands to Allstate’s argument. First, it complains that PICD never mentioned § 27-501 in its request for a hearing, which we take to be an oblique argument of waiver. Second, it argues that, because § 19-107 does not mention § 27-501, the Legislature must have intended that the latter not apply to a filing under the former. Allstate’s main contention, however, is that § 27-501(a) applies only to the cancellation, non-renewal, or refusal to underwrite “a particular insurance risk or class of risks,” which, borrowing language from St. Paul, it interprets to mean an underwriting decision “aimed at individual persons or classes of persons” and not what it regards as “a prospective future application of an underwriting standard to an unknown insurance risk.”

We need not dwell long on Allstate’s first argument, which was effectively dealt with by the Commissioner. Although PICD did not mention § 27-501 in its request for a hearing, it made clear in its submissions its view that the statute applied, and Allstate responded. The issue was clearly presented, considered at length, and determined by the Commissioner. There was no discernible prejudice to Allstate.

As noted, both the Commissioner and the Court of Special Appeals examined the legislative and judicial history of §§ 19-107(a) and 27-501(a) in reaching their respective conclusions. That is appropriate when a clear result cannot be determined solely from the text of the statutes. As we confirmed in Gardner v. State, 420 Md. 1, 9, 20 A.3d 801, 806 (2011), quoting from State v. Johnson, 415 Md. 413, 422, 2 A.3d 368, 373 (2010):

“Where the words of a statute are ambiguous and subject to more than one reasonable interpretation, or where the words are clear and unambiguous when viewed in isolation, but become ambiguous when read as part of a larger statutory scheme, a court must resolve the ambiguity by searching for legislative intent in other indicia....”

*459Title 27 of the Insurance Article, as we observed earlier, deals with unfair trade practices and other prohibited practices in all lines of insurance. Section 27-501 is part of the subtitle defining and prohibiting discrimination. Those laws emanated from a 1945 Act of Congress (the McCarran-Ferguson Act) that allowed the States to regulate the business of insurance and declared that silence on the part of Congress shall not be construed to impose any barrier to such regulation. See P.L. 15, 79th Congress. The General Assembly responded two years later with the enactment of a new subtitle to what was then Art. 48A of the Code, the subtitle containing 15 sections on unfair and deceptive practices. In § 258, the Legislature declared that the purpose of the new Act was “to regulate trade practices in the business of insurance” in accordance with the Congressional intent by defining practices which constitute unfair methods of competition or unfair or deceptive acts or practices. See 1947 Md. Laws, ch. 757.

Section 27-501 had its direct origin in a 1970 enactment intended, in part, “to establish standards of fairness in insurance underwriting.” See 1970 Md. Laws, ch. 417. The Act added a new § 234A to Art. 48A that prohibited insurers from cancelling or refusing 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 an applicant or policyholder.” The next year, the Legislature rewrote the statute to prohibit insurers from cancelling or refusing to renew or underwrite those risks not only for those enumerated reasons but “for any arbitrary, capricious, or unfairly discriminatory reason.” See 1971 Md. Laws, ch. 789.

In Insurance Commissioner v. Allstate Ins., 268 Md. 428, 302 A.2d 200 (1973), this Court was presented with orders of the Commissioner requiring two insurers to renew individual automobile policies that they had declined to renew based on their insured’s driving record. The Commissioner had concluded that the non-renewals on that ground were arbitrary, *460capricious, and unfairly discriminatory. In judicial review actions, the Circuit Court for Baltimore City reversed those decisions, and the Commissioner appealed. In affirming the Circuit Court’s rulings, we read the new language very narrowly, as adding only arbitrary, capricious, or unfairly discriminatory reasons “within the frame of reference of the specifically mentioned ‘reasons.’ ” Notwithstanding its seemingly plain language, we said, the law did not prohibit any arbitrary, capricious, or unfairly discriminatory reason but only those like the specific ones mentioned.

The legislative response to that decision was swift. In its next (1974) session, in addition to rewriting the statute to repudiate this Court’s decision, the General Assembly included in the bill, in the form of “Whereas” provisions, a very clear expression of its intended public policy. After reciting what the Court had done, the Legislature declared, in relevant part:

“That the general welfare of the People of Maryland in the vital area of insurance requires that all underwriting decisions of insurers ... with regard to both eligibility and acceptability of applicants for insurance and insureds, be made solely on the basis of 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 ...”

and that the general welfare of the People

“is insufficiently assured and inadequately promoted merely upon a showing by insurers ... that their underwriting decisions are made without reasons based in whole or in part upon such irrelevant considerations as race, color, religion or creed, sex, national origin or place of residency and the like.”

1974 Md. Laws, ch. 752 (Emphasis added).

To implement that policy, the Legislature added to § 234A the provision that “no insurer ... may cancel or refuse to renew a particular insurance risk or class of risk except by the *461application of standards which are reasonably related to the insurer’s economic and business purposes.” As part of the general code revision process, that provision, with minor style changes, was recodified in 1997 as § 27-501(a)(2) of the new Insurance Article.

The main thrust of Allstate’s argument, which found favor in the Court of Special Appeals, is that, even if the forms of prohibited discrimination are broad, they apply only to the refusal to renew or underwrite “a particular insurance risk or class of risk,” which, in St. Paul, this Court held had reference to individual applicants or policyholders or classes of applicants or policyholders. There is, unquestionably, language in the St. Paul opinion that facially supports that argument. We think, however, that St. Paul was dealing with a different set of facts and a different problem and that some of the language the Court used to support its decision in that case does not mandate the result Allstate seeks here.

By the early 1970s, St. Paul Fire & Marine Insurance Company had become the predominant supplier of medical malpractice insurance in Maryland, insuring over 85% of the physicians practicing in the State. In 1973, it sought a 59.7% rate increase, but had to settle for an increase of 45.9%, effective January 1, 1974. Within a few months, it sought another increase of 48%, which was denied in May 1974. One month later, it announced that, effective January 1, 1975, it would cease writing any medical malpractice policies in the State. No existing policy would be renewed and no new ones would be written.

On a complaint by one of the insured physicians, the Commissioner, acting under what was then § 234A of Art. 48A and is now § 27-501, ordered St. Paul to renew the complaining physician’s policy and to continue to accept medical malpractice business generally at current rates, i.e., he forbid St. Paul from withdrawing from the medical malpractice insurance market. The Commissioner construed St. Paul’s action with respect to the one doctor as an unjustified refusal to renew “a risk” and its more general decision as a refusal to underwrite *462or renew a “class of risks.” The St. Paul case before us was a judicial review action testing the validity of that order.

St. Paul’s argument was that § 234A was aimed at individual underwriting decisions, not general policy decisions “affecting a complete line of insurance such as medical malpractice coverage.” St. Paul, supra, 275 Md. at 138, 339 A.2d at 296. The Commissioner acknowledged that § 234A would not preclude St. Paul from ceasing to write all insurance in Maryland — to withdraw completely from the Maryland market— but that is not what St. Paul intended to do. He viewed medical malpractice coverage not as a “line of insurance” but as a “class of risk.” to which § 234A was applicable. Id. at 139, 339 A.2d at 296. This Court agreed with St. Paul that the statute was aimed at “discrimination against individuals or classes of individuals” and that medical malpractice insurance constituted a “line of insurance” and not a “class of risks.” Id., at 140, 339 A.2d at 297.

The Commissioner’s view, we noted, would lead to absurd results. It would mean that “no insurer could ever refuse to offer any category of insurance to an individual applicant, even though it had previously never underwritten such coverage” and that every insurance company writing any business in Maryland would be required to write medical malpractice policies. Id., at 141, 143, 339 A.2d at 297-98. The Court’s ultimate holding was that St. Paul had made a business decision to withdraw from an entire line of insurance and had not discriminated against either an individual physician or a class of physicians, and, for that reason, § 234A did not apply. Id., at 144, 339 A.2d at 299.

That is not what we have before us now. Allstate was not proposing to cease writing homeowners’ insurance in Maryland. It intended to remain very active in that market, including properties that it currently insured in the designated geographic areas. What it determined to do was to refuse to underwrite new business in those areas, which quintessentially constitutes discrimination against a class of applicants — the class being homeowners who live in the designated areas but *463were not current Allstate policyholders. Given that critical difference, St. Paul is not at all dispositive.

Unlike § 27-501(a), which applies to all lines of insurance, § 19-107 deals only with motor vehicle, property, and casualty insurance, and, even in that constricted area, it is much more focused than § 27-501(a). It prohibits an insurer from refusing to issue or renew a contract of such insurance solely because the subject of the risk or the insured’s address is located in a certain geographic area of the State unless (1) the geographic designation is filed with the Commissioner 60 days in advance of its implementation, and (2) the designation has an objective basis and is not arbitrary or unreasonable. The parties agree that the “objective basis” requirement relates simply to whether the designated area is reasonably ascertainable and that, as the hurricane bands here are identified by ZIP Codes, that requirement is satisfied.

What is less clear is the scope of the requirement that the designation not be “arbitrary or unreasonable.” If that language is interpreted broadly, it could well include, in the context of examining geographic designations, not just what is in § 27-501(a)(l) but also what is in subsection (a)(2), on the theory that any discriminatory practice that is not reasonably related to the insurer’s economic and business purpose is necessarily arbitrary or unreasonable. That issue has not been raised here, however, and we shall not, on this record, assume such a broad reading. Barring such a broad interpretation, § 27-501(a) takes on a greater importance, as it adds a condition applicable to all insurers in all lines of insurance that would not otherwise be implicit in § 19-107(a).

We can find nothing in either the texts of the two statutes or in their respective legislative histories to suggest an intent on the part of the General Assembly to give an insurer desiring to discriminate against insureds or applicants on a geographic basis a free pass to violate the prohibitions in § 27-501(a). Accordingly, we hold that the Commissioner did not err in finding § 27-501(a) applicable.

*464 Sufficiency of the Evidence

Although not precisely articulated as an insufficiency of evidence, the thrust of PICD’s complaint is essentially that. The Division argues first that, in order to prove that its decision not to write new homeowners policies in the designated areas was reasonably related to its economic and business purposes, as required by § 27-501(a)(2), Allstate was required to produce valid statistical data demonstrating the probability of a hurricane sufficiently strong to cause catastrophic damage actually making landfall in Maryland, and that it failed to do so.4 It complains as well that Allstate failed to demonstrate that its rating plan in effect at the time of its amended filing was insufficient to cover catastrophic losses should they occur. With respect to both arguments, PICD relies heavily on Crumlish v. Insurance Com’r, supra, 70 Md.App. 182, 520 A.2d 738.

Crumlish involved a decision by an insurer to cancel an automobile liability policy because the insured had two accidents within two years. The notice stated that the insured was 2.93 times as likely to have another accident within the next year as a driver with no prior accidents. At a hearing *465before the Commissioner, the insurer merely said that one of the company’s underwriting standards was that the company would not insure any individual who had two or more surchargeable losses involving a payout of more than $300 over a two-year period. No evidence was produced to show how the 2.93 probability was determined or how, in light of the ability to surcharge for that kind of driving record, that standard was reasonably related to the insurer’s economic and business purposes.

The Court of Special Appeals did not need to reach that issue, because it concluded that the Commissioner’s order was deficient for other reasons, but, in dicta, the Court made clear that conclusory statements that the standard at issue has the required relationship are insufficient. Rather, the insurer must produce facts showing (1) the statistical basis for the asserted relationship, (2) the validity of the statistical evidence, and (3) the direct and adverse effect the asserted increased risk would have on losses and expenses in light of the company’s current approved rating plan.

PICD contends that this Crumlish dicta governs the analysis required under § 27-501(a)(2). The Commissioner rejected that argument, and so do we. It made sense in the context of what was before the Crumlish Court — an ex cathedra assertion of a quantified probability of future accidents based on two prior accidents, without any evidence establishing the validity of the assertion. It is not, and never has been, a universal requirement applicable to every underwriting standard. The Court of Special Appeals itself has recognized that (see Mirkin v. Medical Mutual, 82 Md.App. 540, 551, 572 A.2d 1126, 1132 (1990) and Miller v. Ins. Comm’r, 70 Md.App. 355, 521 A.2d 761 (1987)), as, indeed, has the Legislature.

In 1998, the General Assembly addressed the issue directly by adding to § 27-501 a new subsection (j) which, in the case of homeowners insurance, lists a number of standards that the Legislature declared were reasonably related to an insurer’s economic and business purpose and did not need statistical validation. Among them, with an exception not relevant here, *466was the “catchall” — “any other standard approved by the Commissioner that is based on factors that adversely affect the losses or expenses of the insurer under its approved rating plan and for which statistical validation is unavailable or is unduly burdensome to produce.” See § 27 — 501(j)(l)(vi).

That is what the Commissioner did in this case. We already have summarized the testimony given by Messrs. Newbold, Michel, and Chernick that, given the paucity of historical data, insurers throughout the country have found sophisticated computer modeling to be a more accurate predictor of when, where, and how often catastrophic hurricanes will strike particular areas of the country, including Maryland. They described in significant detail how the particular model used by Allstate was devised and tested, what factors it took into account, and how the meteorological conclusions translated into reliable damage estimates. This was not a Crumlish situation. PICD had the opportunity to cross-examine those witnesses, and it did so. The Commissioner found that evidence relevant and compelling, as he had a right to do, and, based on that evidence, concluded that the geographic designations were reasonably related to Allstate’s economic and business purposes and were not arbitrary or unreasonable. We find no error.

As a final note, Allstate moved to strike certain parts of the record extract and PICD’s appendix as presenting material outside of the record in the case. PICD is admonished for doing so. As we have not relied on any of that material, there is no need to rule on the motion.

JUDGMENT OF COURT OF SPECIAL APPEALS AFFIRMED; PETITIONER TO PAY THE COSTS.

HARRELL, J., Concurs and Dissents.

concurring and dissenting.

History shows again and again how Nature points out the folly of Man.

Blue Oyster Cult, Godzilla (1977)

*467Imagine a parallel universe with an Alternate Maryland, where an Alternate Allstate is approved to write homeowners’ and lessees’ property insurance (and does so to the tune of approximately a 13% market share). One Saturday at home, the head of Alternate Allstate’s underwriting department settles-in to watch an all-day fest of Godzilla movies. As he watches Godzilla destroy large portions of Tokyo and other coastal areas of Japan again and again (and concerned that climate change could induce the giant amphibious lizard to migrate to other climes, perhaps along a warming mid-Atlantic coastline of the Alternate United States), he imagines Alternate Allstate’s financial exposure should Godzilla come ashore in Alternate Maryland. First thing the next morning, he inquires confidentially of the head of Alternate Allstate’s actuarial department how likely it might be that, assuming Godzilla exists, it might meander ashore along Alternate Maryland’s Atlantic coastline or lower Alternate Chesapeake Bay. The actuary’s response is: “How the hell should I know, but if it does, the property damage will dwarf Tokyo’s fate.” Now really worried and being an eminently cautious person, the underwriter decides that, at a minimum, Alternate Allstate should cease writing new policies of property insurance (but not abandon its existing policy holders) to a distance in Alternate Maryland along its Atlantic coastline and the southern Alternate Chesapeake Bay corresponding to how far inland Godzilla had ravaged Japan. Upon learning of Alternate Allstate’s decision, the Insurance Commissioner of Alternate Maryland, no fan of science fiction, thought to himself quite logically: “Well, how likely is it that such a force of nature will savage Alternate Maryland?” Alternate Allstate, anticipating such a question, included the following in its new business plan: “It is too hard for us to calculate whether, when, and how often over time a giant amphibious lizard will attack Alternate Maryland, but if it does the damage will break our company. We prefer not to take that chance.” The Insurance Commissioner, quixotically, is satisfied by this response, excuses Alternate Allstate from having to justify the probability of the projected catastrophic event occurring, and *468approves Alternate Allstate’s plan. Shortly thereafter, a vast majority of the judges on the Alternate Maryland Court of Appeals affirms that decision.

In the real Maryland, the real Allstate decided not to write new homeowner’s insurance policies in nearly one-third of Maryland because of unsubstantiated fear of a hypothetical force of nature, a Category 2 or greater hurricane making landfall in Maryland. Although my opening analogy is silly, it is so intentionally to illustrate my view that the Commissioner’s and this Court’s approval of Allstate’s discriminatory decision is wrongheaded. Recorded history on the subject shows (again and again) that a catastrophic hurricane of the order of magnitude described in Allstate’s plan justification has not made landfall in Maryland yet. The Commissioner and the Court, nonetheless, approve Allstate’s decision to cease writing new homeowner’s insurance policies in the artificial hurricane bands (neatly coinciding with zip-code areas) because Allstate’s computer models predict such a hypothetical hurricane would make landfall in Maryland every 25,000 years. The basis for this decision is folly. The decision contravenes precedent, which requires insurers to justify the withdrawal of a line of insurance in less than the entire state with a statistical basis grounded in probability, not hypotheticals. The conclusions reached in the majority opinion will transform Maryland Code, Insurance Article § 27-501(a)(2) into a figurative Maginot Line, which insurers may sweep around and avoid. I dissent.

I.

In order to reduce its exposure to risk of losses due to a catastrophic hurricane making landfall in Maryland, Allstate decided to cease issuing new homeowner’s policies in all of St. Mary’s, Somerset, Talbot, Wicomico, and Worchester Counties and significant swaths of Anne Arundel, Calvert, Charles, Dorchester, Prince George’s, and Queen Anne’s Counties. Allstate grounded its decision in hurricane prediction models, which computed a relative risk of a category 2 or higher hurricane making landfall in Maryland four times in 100,000 *469years. Allstate’s battery of evidence did not include, however, an assessment of the probability of such a catastrophic hurricane landing in Maryland. This omission is significant because, as Allstate’s own evidence points out, none of the 163 hurricanes recorded in meteorological history that made landfall in the United States did so in Maryland. I agree with the majority that § 27-501 applies to Allstate’s decision; however, I disagree with the majority that Allstate provided sufficient evidence to justify the Commissioner’s decision in Allstate’s favor.

In 1974, the General Assembly amended then Maryland Code, article 48A, § 234A (the predecessor to § 27-501), expanding the prohibition of discriminatory insurance underwriting beyond prejudice (e.g., race, color, creed, sex, ability) to include underwriting decisions based on any reason not reasonably related to the insurer’s economic or business purposes. Lumbermen’s Mut. Cas. Co. v. Ins. Comm’r, 302 Md. 248, 255, 487 A.2d 271, 274 (1985). The language of the amendment is substantially similar to the present language in § 27-501(a)(2).5 The General Assembly was concerned at that time about a developing pattern where automobile insurers justified cancelling insurance policies because the insured had as few as one accident, an underwriting policy that “might not be justified by the actual risk experience of those insurers.” St. Paul Fire & Marine Ins. Co. v. Ins. Comm’r, 275 Md. 130, 143-44, 339 A.2d 291, 298-98 (1975). The preamble to Chapter 752 of the Acts of 1974 expresses this concern and states that an insurer’s 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 *470probability 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....”

Lumbermen’s Mut. Cas. Co., 302 Md. at 254, 487 A.2d at 274 (quoting 1974 Md. Laws 752). The preamble is an integral part of the aim of § 27-501. We relied on it as authority for requiring insurers to justify objectively the termination of an insurance policy. Lumbermen’s Mut. Cas. Co., 302 Md. at 267-68, 487 A.2d at 280-81; see also St. Paul Fire & Marine Ins. Co., 275 Md. at 137, 339 A.2d at 295. Thus, the 1974 amendment represents the legislative intent to protect the public from arbitrary underwriting practices. Gov’t Emps. Ins. Co. v. Ins. Comm’r, 273 Md. 467, 478, 330 A.2d 653, 659 (1975).

Crumlish v. Insurance Commissioner, 70 Md.App. 182, 520 A.2d 738 (1987), expanded upon Lumbermen’s analysis of § 27-501(a)(2) and the statute’s requirement that “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.” In order to establish that an underwriting standard is reasonably related to the insurer’s economic and business purposes, the insurer must produce facts that answer sufficiently the following questions: (1) What is the statistical basis for the supposition? (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 insurer’s losses and expenses in light of its current approved rating plan? Crumlish, 70 Md.App. at 190, 520 A.2d at 742; see also Lumbermen’s, 302 Md. at 266, 487 A.2d at 280 (stating that an insurer seeking to not renew an insurance policy must show the available rating plan “would not have adequately compensated the insurance companies for the asserted increased risk”). Although the Crumlish questions were concededly set-out in dicta (which the majority here notes), several later cases incorporated the Crumlish questions into their analyses (which the majority here ignores), giving this principle precedential weight and bolstering an important regulatory tool to prevent unsubstantiated, discrimi*471natory insurance underwriting practices in Maryland. See Fromberg v. Ins. Comm’r, 87 Md.App. 236, 245, 589 A.2d 544, 548-49 (1991); Ins. Comm’r v. Nevas, 81 Md.App. 549, 557-58, 568 A.2d 1144, 1148-49 (1990); see also Stavely v. State Farm Mut. Auto. Ins. Co., 376 Md. 108, 110, 829 A.2d 265, 267 (2003); Andrew Janquitto, Maryland Motor Vehicle Insurance § 14.5 (Matthew Bender 3d. ed. 2011).

Because § 27-501 applies to the present case, Allstate was required to produce a sufficient statistical basis justifying its cessation of writing new homeowner’s insurance policies in its so-called hurricane bands four, five, and six. Allstate must show also that its current rating plan, were it to continue to write polices in those areas, would leave it compensated inadequately. Further, Allstate’s supposition 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’ Allstate’s current insurance rating plan. Preamble to 1974 Md. Laws 752 (emphasis added). In sum, even as the Commissioner conceded, “insurers have been required to provide statistics to demonstrate the relationship between the underwriting standard applied and the insurer’s business and economic purposes.”

Allstate failed to advance a statistical basis, derived from historical meteorological evidence, climate science, and/or actual claims history, that demonstrated the probability of a catastrophic hurricane making landfall in Maryland. Even if it would be difficult or lacking in some other actuarial way to make that effort, merely asserting that difficulty or impression, without making a bona fide effort to marshal the available evidence, should not be good enough to justify moving on to Allstate’s alternative modeling approach. Allstate justified its decision based on hypothetical hurricanes, i.e., computer models based on the worst 5% of hurricanes that made landfall in North Carolina, Virginia, and Delaware, but not Maryland.6 *472These models estimated that the relative risk of such a hurricane landing in Maryland is once every 25,000 years.7

Allstate justified its decision using the relative risk of hypothetical hurricanes and thus failed to justify its “no-write” decision for new policies and to show the current rating plan would fail to compensate Allstate sufficiently should it continue to write policies in the hurricane bands. Allstate fails also to illuminate why there is an increased risk of a catastrophic hurricane. If the science of climate change is at the bottom of its worries, Allstate does not say so. Finally, Allstate’s decision enlarges considerably the portion of Maryland where Allstate will not write new homeowner’s policies. Previously, Allstate excluded only those properties within one mile of the Atlantic Ocean. Allstate’s new policy excludes properties as far inland as 60 miles from the Atlantic Ocean (approximately one-third of Maryland), including cities such as Cambridge, Prince Frederick, and Leonardtown. Allstate should not be allowed to implement its new plan until it shoulders its statistical burden of showing probability.

The majority opinion excuses the Commissioner’s failure to find fatal Allstate’s lack of probabilistic evidence by distinguishing the Crumlish line of cases. It relies on Mirkin v. Med. Mut., 82 Md.App. 540, 572 A.2d 1126 (1990) and Miller v. Ins. Comm’r, 70 Md.App. 355, 521 A.2d 761 (1987), as authority for the supposition that the Crumlish analysis is not a standing requirement applicable to every underwriting stan*473dard.8 Reliance on Mirkin and Miller for this proposition is misplaced, however.

In Mirkin, the insurer cancelled Dr. Mirkin’s professional liability insurance because he altered a patient’s billing records, a fact discovered during a malpractice lawsuit against Dr. Mirkin. 82 Md.App. at 546-47, 572 A.2d at 1129-30. In Miller, the insurer cancelled Miller’s professional liability insurance because he made material misrepresentations on his application for the insurance. 70 Md.App. at 361, 521 A.2d at 764. Drs. Mirkin and Miller posited that cancelling insurance because the insureds made misrepresentations is not an underwriting standard reasonably related to the insurers’ economic and business purposes. The panels of the intermediate appellate court rejected their arguments because cancelling an insurance policy because the insureds mislead the insurers is a underwriting standard incapable of objective justification, but nonetheless related reasonably to the insurers’ economic and business purposes. Mirkin, 82 Md.App. at 551, 572 A.2d at 1132 (citing to Miller, 70 Md.App. at 368-70, 521 A.2d at 768-69). The Legislature codified the Mirkin and Miller exception as Maryland Code, Insurance Article § 27-501(j)(1)(i).9

Mirkin and Miller are inapposite to the present case. Unlike Drs. Mirkin and Miller, homeowners in Southern Maryland and the Eastern Shore obviously did not act inten*474tionally to jeopardize their prospect for obtaining insurance coverage from Allstate. Second, declining to write new homeowner’s insurance policies because of the risk of catastrophic hurricane damage is an underwriting policy that can be justified by “an objectively demonstrable, statistical basis.” Mirkin, 82 Md.App. at 551, 572 A.2d at 1132. The probability of a category two or greater hurricane making landfall in Maryland should be able to be determined by using historical data (which shows no hurricanes making landfall there), predicting meteorological change based on quantitative or qualitative factors from climate change or other evidence, as well as using computer models (which predict one hypothetical catastrophic hurricane every 25,000 years).

The majority opinion finds refuge (unsuccessfully, in my view) in Maryland Code, Insurance Article § 27 — 501(j)(l)(vi), which it labels a “catchall” exception to § 27-501(a)(2). Section 27 — 501(j)(l)(vi) provides:

In the case of homeowner’s insurance, standards reasonably related to an insurer’s economic and business purpose under subsection (a)(2) of this section, include, but are not limited to, the following and do not require statistical validation: ... any other standard approved by the Commissioner that is based on factors that adversely affect the losses or expenses of the insurer under its approved rating plan and for which statistical validation is unavailable or is unduly burdensome to produce[.]

Md.Code Ann., Ins. § 27-501(j)(1)(vi). The use of § 27-501(j)(1)(vi) here is problematic. The Commissioner did not invoke expressly § 27-501(j)(1)(vi) when he excused Allstate’s lack of statistical evidence to support its decision not to write new insurance policies in hurricane bands four, five, and six. Rather, the Commissioner invoked (erroneously) Mirkin and Miller to distinguish Crumlish. Second, statistical validation as to the probability of a catastrophic hurricane making landfall in Maryland is available, and is not unduly burdensome to produce, as pointed out supra. Therefore, § 27-501(j)(1)(vi) does not relieve Allstate of its burden to demon*475strate objectively the probability of a catastrophic hurricane making landfall in Maryland.10

For these reasons, I would reverse the judgment of the Court of Special Appeals and remand this case to that court with directions to reverse the judgment of the Circuit Court for Baltimore City and to remand the case further to the Insurance Commissioner with further directions to vacate his decision and conduct further proceedings consistent with this concurring and dissenting opinion. Allstate should be given the opportunity, if it wishes to seize it, to produce proper evidence to justify its new plan.