dissenting.
A. Introduction. I respectfully disagree with Part III.B.2 of the court's opinion. It holds that payment of the "face value" of Kevin Babosky's policy exhausted his lability coverages, satisfying AS 28.20 445(e)(1) for purposes of invoking Colleen Coughlin's un-derinsured motorist (UIM) coverage.1 It consequently holds that exhaustion did not require payment of the "add-ons" which Ba-bosky's liability coverages also covered.2 This holding is contrary to the result subsection 445(e)(1) requires. I will first explain how we should resolve the issue and then explain my disagreement with the court's analysis.
B. How We Should Decide this Issue. This case poses a practical question important to insurers and their injured insureds: What triggers the underinsured motorist coverage, requiring the UIM insurer (here, GEICO) to pay? Alaska Statute 28.20 445(e)(1) answers that question: The UIM insurer must pay after "the limits of liability of all bodily injury ... lability ... policies ... have been used up."3
*993What do the legislature's unqualified words "limits of liability" mean? Read in isolation by ignoring companion sections in AS 28.20, they could have two different theoretical meanings. They could mean the amounts potentially payable under all the lability coverages of any underlying lability policy.4 This is the meaning I think they have. Or they could mean the facial (or numerical) limits specified in the declarations of any underlying policy. This is the meaning the court gives them in this case. Our job is to determine the meaning the legislature gave them.5
We need not guess. Subsection 445(e)(1) is part of the Motor Vehicle Safety Responsibility Act. Nearby passages in that act reveal the meaning the legislature intended. Alaska Statute 28.20.070(a) specifies the mandatory minimum numerical limits applicable to automobile lability policies effective in Alaska. It requires each policy to have "a limit, exclusive of interest and costs, of not less than $50,000 because of bodily injury. ..." 6 Similarly, AS 28.20.440(b)(2) specifies mandatory minimum "limits exelusive of interest and costs" of $50,000 for a vehicle owner's lability insurance policy.7 Both subsections specify the mandatory minimum facial, or numerical, limits automobile liability policies must have in Alaska.
*994The legislature's choice of terminology controls here. When the legislature meant to refer to the facial policy limits, as it did in subsections .070(a) and 440(b)(2), it did not use the unqualified words "limit," "limits," or "limits of liability." Instead, in subsection .070(a) it qualified "limit" with the phrase "exclusive of interest and costs." And in subsection .440(b)(2) it qualified "limits" with the same phrase. The legislature must have been aware of its own usage. Section A45 immediately follows section A440. There would have been no reason to use those phrases to qualify "limit" or "limits" in subsections .440(b)(@2) and .070(a) if the legislature intended its unqualified words "limit," "limits," or "limits of liability" to mean only the facial policy limits.8 Had the legislature intended the words "limits of liability" in AS 28.20.445(e)(1) to refer only to the facial limits of underlying insurance, it would have used the same qualifying phrase it used elsewhere in the Motor Vehicle Safety Responsibility Act when it meant to specify the facial limits. That it did not demonstrates that it meant the unqualified words "limits of Hiability" in subsection A45(e)(1) to refer to all amounts payable under a policy's liability coverages, and not just its facial limits.
We recently summarized our approach to statutory construction in an opinion that considered whether policyholders had exhausted the underlying liability limits of an insurance policy:
We apply a sliding scale approach to statutory interpretation: to determine the meaning of a statute we look to its legislative history, even if its language is plain on its face. But "the plainer the meaning of the language of the statute, the more convincing any contrary legislative history must be." When a statute's meaning appears clear and unambiguous, the party urging another meaning "bears a correspondingly heavy burden of demonstrating contrary legislative intent." We decline to "modify or extend a statute where the statute's language is clear and the legislative history reveals no ambiguity."[9]
Applying that approach, we must apply the language of the statute as written. 'We begin with analysis of the language of the statute.10 Statutory context, revealed in subsections A40(b)(2) and .070(a), confirms the meaning the legislature intended.
Further, reading "limits of Hability" in AS 28.20 445(e)(1) to mean the same thing as "limits exclusive of interest and costs" in AS 28.20.440(b)(2) or "limit, exclusive of interest and costs" in AS 28.20.070(a) conflicts with our rule of statutory construction under which we give effect to all words of the statute and render none superfluous.11 Reading the unadorned phrase in subsection A4b(e)(1) to refer only to facial limits makes superfluous the qualifying words in subsections .440(b)(2) and .070(a).
No legislative history suggests that the legislature meant something different. Nothing brought to our attention outweighs the plain meaning of the legislature's words. Nor, for reasons I will discuss in the next part, do the legislative purposes permit a contrary reading. Instead, two precepts of Alaska's approach to UIM confirm the reading the plain statutory language requires.
First, the requirement that policies contain underinsured motorist coverage reflects what this court has described as "Alaska's current *995excess approach."12 We have stated that this approach is
premised upon the idea that the injured person is entitled to recover under his or her own underinsured motorist coverage to the extent that the tortfeasor's lability insurance coverage is insufficient to compensate the insured person fully for his or her loss, subject only to the limits of the underinsured motorist coverage."[13]
We described the purpose of Alaska's excess UIM approach as follows: "Exeess coverage thus strives to provide additional coverage, as needed to fully compensate injured motorists, after available lability coverage has been completely exhausted." 14 Classifying UIM coverage as "excess" confirms that all underlying lability coverages, including coverage for any "add-ons," must be exhausted. In dealing with primary-excess coverage disputes, we require complete exhaustion of underlying primary liability coverages, including lability coverage add-ons, before the exeess coverage comes into play.15 There is no conceptual difference relevant here between excess coverage and UIM coverage.
The court states that UIM and excess coverage disputes are distinguishable. It asserts in support that the UIM statute's "concern with making UIM coverage widely available distinguishes it significantly from excess coverage," that "exeess coverage insures against catastrophic loss," and that "UIM coverage, unlike excess coverage, does not contemplate extreme or unusual cireum-stances."16 In my view, these assertions do not support the distinction the court draws. The noted "concern" is only one of several principles underlying Alaska's current UIM statute. Other principles-treating the UIM coverage as excess, requiring exhaustion of the underlying coverages, and aiming for full compensation-refute the proposed distinction and outweigh whatever relevance the availability principle may have here. And excess coverage is not extraordinary or limited to catastrophic claims. It is simply the coverage purchased to cover claims exceeding the primary coverage. It is routinely available to consumers, who may, for economic reasons, carry minimum primary limits and use excess coverage to extend their limits. Given that motor vehicle accidents can easily generate non-catastrophic claims exceeding the primary limits, excess insurance is not limited to "extreme or unusual cireumstances."17 It has the same ultimate purpose as UIM coverage: to satisfy the claims of an injured victim to the extent the tortfeasor's underlying liability coverage is insufficient to compensate the victim fully.18
Second, it is anomalous to speak of "complete exhaustion" in terms of "full compensation"19 unless the injured claimant is truly being compensated for the loss of use of the money damages suffered at the moment of injury, and for the costs, including attorney's fees, of obtaining that compensation. The concept of making the claimant whole to the extent of the available coverage has driven this court to reason that prejudgment interest is part of compensatory damages 20 and therefore part of the maximum limits of a policy that covers prejudgment interest.21 It has likewise driven this court to hold that a policy's coverage of the plaintiff's litigation *996costs awardable against the insured is a lo-bility coverage an insurer must tender when making a "policy limits" settlement offer.22 Our past decisions require the conclusion that "full compensation" is not achieved for purposes of triggering UIM coverage under subsection 445(e)(1) unless all applicable underlying liability coverages have been paid.
This interpretation of the statute is also compelled by our past treatment of similar terms in insurance policies. Policies often cover not only an insured's Liability for tort damage awards, but also an insured's Hability for additional awards, such as for costs and interest. We refer to such coverages as "add-ons" that must be added to a policy's numerical limits.23 Add-ons are part of a policy's liability coverage. An insurer must therefore pay them to pay "policy limits"the maximum amount the policy would cover if judgment were entered against the insured at trial.24 Because a Judgment would include these additional awards, the limits of a policy covering liability for these additional awards are not exhausted until the coverages for the insured's exposure for these additional awards are exhausted too. These coverages are part of the "limits of Hability."
Colonial's "limits of liability" consequently include everything Colonial would have had to pay had a judgment been entered against its insured after trial. It is undisputed that Colonial's policy covered Babosky not only for liability for damage awards, but also for lability for prejudgment interest, attorney's fees, and cost awards, and that with the add-ons, Colonial's liability coverage substantially exceeded the policy's $50,000 facial limits.25 No one claims that Colonial paid more than $50,000. Colonial therefore did not exhaust its "limits of liability" per subsection A4Ab(e)(1) for purposes of implicating GEI-CO's UIM coverage.26
C. The Court's Opinion. Because the opinion implicitly concludes that the legislature's words do not resolve the exhaustion question, it embarks on an avoidable analytical voyage. It then arrives in the wrong port. The opinion first translates the legislature's words into the court's words, but then chooses not to give those words the meaning we have given them in equivalent insurance disputes. It bases this choice in reliance on its perception of public policy and the legislature's purposes. The statute's clarity precludes reliance on those interpretive aids, but I think the court's policy and purpose assessment is incorrect in any event. The court ultimately reads "limits of Hability"-the controlling words in AS 28.20.445(e)(1)-to include no liability coverages except the facial policy limits.27 I disagree with each stage of the court's approach.
1. The statute's words. Because the opinion does not recognize or enforce the distinctions the legislature drew, it does not give the legislature's words the meaning intended. I discussed this proposition above.
2. The court's words. The court's opinion substitutes its words for the legislature's, *997translating .the statute's words into "policy limits."28 This substitution obscures the need to focus on the legislature's words, and forees the opinion to try to explain why "policy limits" means something different in UIM coverage cases than in other insurance exhaustion contexts. It also results in a public policy analysis that is unsupported by the words the legislature used.
But if we are going to ignore the effect of the legislature's qualifying words, our past decisions should guide us. There is no logical or functional difference between this case and our policy limits cases. Both deal with this central question: what must the tortfea-sor's insurer pay to exhaust its policy limits (or, in the statute's words, the "limits of liability")? We should answer the question the same way in both contexts.
The cases discussed above hold that the policy limits of a liability policy are not exhausted unless liability coverages for add-ons are also paid. Our usage is also instructive. When we wish to speak of the numerical limit set out in a liability policy's declarations, we have qualified our words to make our meaning clear.29
The court here uses similar qualifiers to make its meaning clear.30 That it must do so when it intends to exclude coverage for add-ons illustrates the incongruity of reading subsection .445(e)(1)'s unqualified phrase as though the legislature qualified it.
3. Public policy and legislative purposes. The opinion relies on public policy and the court's perception of the purposes underlying UIM coverage to justify its choice not to follow our policy limits cases.31 This reliance is unwarranted.
Public policy and legislative purposes are inherently weak interpretative aids. There is no need to resort to them when the statute's words are so clear. They are especially unhelpful here because they do not speak specifically to the narrow issue before us. At best, they reflect broad considerations which do not explain what the statute means in this specific context. At worst, they reflect the court's own projections about what the legislature must not have intended,32 and its own expressions of public policy not voiced by the legislature.33 Relying on public policy seems particularly chancy here, given that the result-allowing partial exhaustion to trigger the UIM coverage-is inconsistent with the legislature's express words and purposes.
Concerns about possible delays do not justify ignoring the statute's words. More significant disputes about the underlying coverage (whether exclusions apply, for example) would not justify triggering the UIM coverage. Disputes over coverage for add-ons will occasionally delay exhausting the underlying liability limits, but I doubt that they will unduly delay UIM claims. First, a lability insurer with a good-faith interest in litigating a coverage dispute about add-ons can "exhaust" the policy limits by agreeing to pay whatever the court will require when it resolves the add-ons dispute.34 This agreement has the effect of exhausting the limits, whatever they prove to be, triggering UIM coverage. A liability insurer has incentive to enter into such an agreement to avoid bad faith claims. And if the lability insurer is engaging in bad faith in a policy limits situation, the UIM insurer relies at its peril on the liability insurer's recalcitrance.35 Further, no difficulty in determining the extent *998and amount of add-ons coverage has prevented us from treating the add-ons as something an insurer must pay to maximize the liability coverage it owes its insured. We have assumed that this total figure is "quantifiable" and have imposed on the insurer a duty to tender a specific amount to discharge policy limits.36 And indeed, a UIM claim may have to await a liability and damages trial on the plaintiff's claims against the tortfeasor. Only then will it be known whether the tortfeasor was actually underinsured.
The court supposes that a statutory purpose of "making the benefit of UIM coverage broadly available" supports its reading of the statute.37 It is not apparent how disputes about the underlying insurer's obligations could reduce the availability of UIM coverage. If a claim is worth more than the underlying limits of liability, there is little chance in a given case the UIM coverage will not be implicated.38 A social policy of making UIM coverage more readily available would not justify reading the statute to say something it does not. The court also states that "it is difficult to believe" that the legislature might use terminology that would "engender litigation and uncertainty in many cases."39 Perhaps so, but this general speculation cannot justify a reading contrary to the statute's words. There is no reason to think the legislature intended to adopt a different standard for exhaustion than we have adopted.
The court also expresses concern about the uncertainty of determining policy limits. It recognizes that this "is sometimes a guess" and "is quite speculative." 40 It is "wary of starting a new litigation industry."41 But an existing litigation industry already stands prepared to dispute every aspect of personal injury claims.
In short, the court draws a bright line which the legislature did not draw. And it does so assuming that full exhaustion carries detriments with it. That is not a public policy choice open to us, given the words of the statute.
4. Marginal benefit. The result benefits Coughlin by allowing her to make a UIM claim, but the social benefits the court anticipates are hypothetical or improbable, and will likely be outweighed by the detriments.
First, I think the harm to the purpose of exhaustion outweighs any possible prejudice to UIM coverage. The legislature chose an excess UIM approach.42 The court's reading of the statute is inconsistent with making excess coverage available to fully compensate injured motorists "after available Hability coverage has been completely exhausted."43 The underlying coverage has not been exhausted. The result may mean that the claimant will never be made whole for loss of use of her prospective damages award or costs and fees incurred in recovering from the underlying insurer. And requiring UIM insurers to compensate plaintiffs prematurely may reduce recalcitrant underlying insurers' exposure 'to liability for excess judgments and bad faith claims. If so, it would deprive personal injury plaintiffs of their main tool in seeking policy limits settlements from the tortfeasors' insurers.
Second, I predict that the result will burden the wrong people. It will burden purchasers of underinsured motorists coverage (which will now be more easily implicated) and benefit tortfeasors who purchase liability coverage (whose insurers will now have less need to pay add-ons if the claimant has a UIM insurer to pursue). It may transfer from the alleged tortfeasor's insurer to the *999UIM insurer the expense of defending the other driver and challenging the plaintiff's damages claims. If so, it will make UIM coverage more expensive, again at the expense of the parties who are not at fault.
D. Conclusion. In short, I think the superior court correctly analyzed this issue. I would therefore affirm.
. Op. at 992. There is no claim that GEICO's policy terms are more favorable to Coughlin than subsection .445(e)(1).
. Op. at 992.
. AS 28.20.445(e)(1) provides in pertinent part:
(e) Uninsured and underinsured motorists coverage
(1) may not apply to bodily injury, sickness, disease, or death of an insured or damage to or *993destruction of property of an insured until the limits of liability of all bodily injury and property damage liability bonds and policies that apply have been used up by payments, judgments or settlements....
(Emphasis added.)
. I use a qualifier, such as "facial" or "numerical," with "limits" or "policy limits" to refer to the numerical lability limits specified in the policy's declarations.
The court's opinion uses these and equivalent qualifiers for the same purpose. It therefore speaks of the "$50,000 'facial limit," (Op. at 987), the "face value of the Colonial policy," (id. at 990), "the face amount of policy limits," (id. at 991), "facial coverage," (id.), "full face value of one's policy limits," (@d. at 992), and "full face value of the policy." (id.). (Although we have sometimes used "face value" when referring to the facial limits of liability policies, the concept of "value" seems better suited to describing life and property loss coverages.)
This practice is consistent with our past use of qualifying words or descriptive phrases when we refer only to the numerical limits, see, e.g., State Farm Mut. Auto. Ins. Co. v. Harrington, 918 P.2d 1022, 1025-26 (Alaska 1996) ("numerical facial limit"; "facial limits"; "facial policy limits"); Hughes v. Harrelson, 844 P.2d 1106, 1108 (Alaska 1993) ("minimum policy limits"); Tucker v. United Servs. Auto. Ass'n, 827 P.2d 440, 441 (Alaska 1992) ("base liability limit"); Wold v. Progressive Preferred Ins. Co., 52 P.3d 155, 162 (Alaska 2002) ("nominal policy limits"), or when context makes our meaning clear, e.g., Curran v. Progressive Northwestern Ins. Co., 29 P.3d 829, 830, 831, 835 n. 35 (Alaska 2001) ("policy limit of $50,000 per person"; "$50,000 policy limits"; "facial policy limits plus applicable add-ons").
These qualifiers are necessary because we have consistently used the unqualified phrase "policy limits" to include all lability coverages. See, eg., Farquhar v. Alaska Nat'l Ins. Co., 20 P.3d 577, 580-81 (Alaska 2001) (''We found that 'policy limits' meant the facial limit of the policy, plus ' Rule 82 attorney's fees.") (quoting Schultz v. Travelers Indem. Co., 754 P.2d 265, 267 (Alaska 1988)).
. Curran, 29 P.3d at 831; Baker v. State, 30 P.3d 118, 118 (Alaska App.2001).
. AS 28.20.070(a) provides:
A policy or bond is not effective under AS 28.20.060 unless it is issued by an insurance company or surety company authorized to do business in this state, except as provided in (b) of this section, and if the accident resulted in bodily injury or death, unless the policy or bond is subject to a limit, exclusive of interest and costs, of not less than $50,000 because of bodily injury to or death of one person in any one accident and, subject to the same limit for one person, to a limit of not less than $100,000 because of bodily injury to or death of two or more persons in any one accident, and if the accident has resulted in injury to, or destruction of, property to a limit of not less than $25,000 because of injury to or destruction of property of others in any one accident.
. AS 28.20.440(b)(2) provides:
The owner's policy of liability insurance must
[[Image here]]
(2) insure the person named and every other person using the vehicle with the express or implied permission of the named insured, against loss from the liability imposed by law for damages arising out of the ownership, maintenance, or use of the vehicle within the United States or Canada, subject to limits exclusive of interest and costs, with respect to each vehicle, as follows: $50,000 because of bodily injury to or death of one person in any one accident, and, subject to the same limit for one person, $100,000 because of bodily injury to or death of two or more persons in any one accident, and $25,000 because of injury to or *994destruction of property of others in any one accident....
. The legislature used similar terminology in AS 28.22, the Mandatory Motor Vehicle Insurance Act. Thus, AS 28.22.101(d) requires in part that a motor vehicle liability policy be subject to "limits exclusive of interest and costs" of "$50,000 because of bodily injury to or death of one person." AS 28.22.201(a)(1) provides that underinsured motorists coverage required by Chapter 22 "does not apply to bodily injury ... until the limits of liability bonds and policies that apply have been used up by payments or judgments or settlements."
. Curran, 29 P.3d at 831-32 (footnotes omitted).
. Bullock v. State, Dep't of Cmty. & Reg'l Affairs, 19 P.3d 1209, 1214 (Alaska 2001); Gerber v. Juneau Bartlett Mem'l Hosp., 2 P.3d 74, 76 (Alaska 2000).
. Louisiana-Pacific Corp. v. State, Dep't of Revenue, 26 P.3d 422, 427 (Alaska 2001); Kodiak Island Borough v. Exxon Corp., 991 P.2d 757, 761 (Alaska 1999); Romann v. State, 991 P.2d 186, 190 (Alaska 1999); Fancyboy v. Alaska Village Elec. Coop., 984 P.2d 1128, 1133 (Alaska 1999); Rydwell v. Anchorage Sch. Dist., 864 P.2d 526, 528 (Alaska 1993).
. Curran, 29 P.3d at 832.
. Progressive Ins. Co. v. Simmons, 953 P.2d 510, 517 n. 6 (Alaska 1998).
. Curran, 29 P.3d at 832.
. "Only when the primary insurer's limits are exhausted do obligations on the part of the excess insurers arise. Under Alaska law it is established that primary policy limits include, among other things, facial limits and attorney's fees taxed under Rule 82, to the extent of coverage." Safety Nat'l Cas. Corp. v. Pacific Employers Ins. Co., 927 P.2d 748, 751 (Alaska 1996) (citations omitted).
. Op. at 991 n. 19.
. Id.
. Cf. Curran, 29 P.3d at 832 quoted above in the text.
. Cf. Curran, 29 P.3d at 832 (describing purpose of UIM coverage as coverage needed to "fully compensate" after liability coverage has been "completely exhausted").
. Guin v. Ha, 591 P.2d 1281, 1287 (Alaska 1979).
. Tucker, 827 P.2d at 441.
. Harrington, 918 P.2d at 1025-26; Schultz v. Travelers Indem. Co., 754 P.2d 265, 267 (Alaska 1988).
. Curran, 29 P.3d at 835 n. 35.
. Wold, 52 P.3d at 162-63; see, e.g., Tucker, 827 P.2d at 440-41 & n. 3.
An insurance carrier's agreement to settle a claim for "policy limits" obligates the company to pay its maximum potential liability available under the policy....
[[Image here]]
... Whether the disputed coverage involves the base liability limit, court costs, or prejudgment interest, it is the insured's total potential liability which is of concern to the insured in a ""policy limits" settlement.
Tucker, 827 P.2d at 440-41.
. GEICO asserts on appeal that the parties do not dispute that Colonial's policy covered prejudgment interest, costs, and attorney's fees. Coughlin does not dispute this assertion, and represents that "had she held out for fees and interest," her recovery from Colonial would have been about $67,750. According to a demand Coughlin made on GEICO, the Rule 82 add-on for a $50,000 policy would be $7,500.
. I agree with the court's resolution of the medical lien issue in Part IILB.1., but my conclusion that Colonial's limits were not exhausted would moot the lien issue.
. Op. at 992 ("'We therefore hold that a policy limit is exhausted for the purposes of invoking underinsured motorist coverage when the full face value of the policy is paid...." (Emphasis added.) The court's footnote to that passage cites subsection .445(e)(1).}.
. Op. at 988-89, 990-92.
. See cases cited supra note 4.
. See supra note 4.
. Op. at 991-92.
. See, eg., Op. at 991 ("it is difficult to believe that the legislature would use"; "it would be unlikely that the legislature ... would have intended").
. See, eg., Op. at 992 ("there are strong policy reasons").
. This appears to have been the practice followed in Schultz v. Travelers Indem. Co., 754 P.2d 265, 266-67 (Alaska 1988).
. Insistence on actual payment may expose an excess insurer to a bad faith claim for anticipa-torily repudiating its contract. Cf. Grace v. Ins. Co. of N. Am., 944 P.2d 460, 467 & n. 15 (Alaska 1997) ("[The excess insurer's] duty was triggered when Bell became liable for sums in excess of INA's lower coverage limit, rather than when the underlying limits were actually paid.").
. E.g., Bohna v. Hughes, Thorsness, Gantz, Powell & Brundin, 828 P.2d 745, 768 (Alaska 1992).
. Op. at 991.
. Coughlin's demand letter to her UIM insurer alleged total losses exceeding $150,000. She asserts on appeal her damages exceeded $172,000.
. Op. at 991.
. Op. at 991, 992.
. Op. at 992.
. Curran v. Progressive Northwestern Ins. Co., 29 P.3d 829, 832 (Alaska 2001).
. Id.