POSNER, Circuit Judge.
Section 5(b) of the Illinois Workers’ Compensation Act, Ill.Rev.Stat.1981, ch. 48, 11138.5(b), provides that if an employee is injured in circumstances requiring his employer to compensate him under the Act, and if he later recovers damages from a tortfeasor for the same injury, he must pay from his damage award “to the employer the amount of compensation paid or to be paid by [the employer] to such employee”; but “where the services of an attorney at law of the employee ... have resulted in or substantially contributed to the procurement by suit, settlement or otherwise of the proceeds out of which the employer is reimbursed, then ... the employer shall pay such attorney 25% of the gross amount of such reimbursement.” In this diversity suit, we are required to decide when and how payment of this attorney’s fee shall be made when compensation is payable over a future period that is uncertain — a question on which the statute is silent and the Illinois Supreme Court has not yet spoken.
We shall simplify the facts by omitting supernumerary parties and rounding off all money figures to the nearest dollar. In 1979 Mr. Dunbar was killed in circumstances making his employer liable under the Illinois Workers’ Compensation Act. The commission that administers the act ordered the employer to pay Dunbar’s widow $1,750 in burial expenses plus $303 a week for 1,043 weeks (plus $174 for one additional week — which we shall ignore), unless she remarried or died, either of which events would terminate the payments. She hired a law firm to prosecute a common law tort claim arising out of her husband’s accident, and the firm negotiated a settlement for $325,000. The contingent-fee contract that she had signed with the firm provided for a fee of 33 percent — but only of the excess of the settlement over the present value of [621] her compensation award. When the settlement went through, the employer had paid 116 weeks of compensation to Mrs. Dunbar plus the burial expenses. The parties agree that the Illinois Workers’ Compensation Act (1) required Mrs. Dunbar to reimburse the employer for the payments the employer had already made ($36,881 = $1,750 + 116 x $303), minus 25 percent representing the statutory attorney’s fees to which her lawyers are entitled, and (2) entitled the employer to make no further payments to her until his credit, based on the settlement, was used up, at which point he would resume payments if Mrs. Dunbar was still entitled to them under the compensation award. The parties differ only on how the statutory 25 percent fee to which Mrs. Dunbar’s lawyers are entitled for creating the employer’s credit should be paid. She and the lawyers argue that it should be paid to the lawyers at the rate of $76 a week, representing 25 percent of the weekly compensation payment of $303, until either the credit is exhausted or Mrs. Dunbar dies or is remarried — whichever occurs first. The district court accepted this argument. The employer argues for deducting from the credit the total amount to which the lawyers would be entitled if the credit were used up (that is, if Mrs. Dunbar does not die or remarry before the credit is exhausted) — $79,445 ($1,750 X .25 + 1,043 X $303 X .25) — minus what the lawyers have already received in connection with the reimbursement to the employer of the compensation and burial payments made to Mrs. Dunbar before the tort settlement was made — $9,220 (one fourth of $36,881).
It might seem odd that Mrs. Dunbar’s lawyers would be arguing for deferred rather than immediate payment of the attorney’s fee. But the employer is not proposing immediate payment; he just wants the attorney’s fee deducted from his credit; he couldn’t care less whether the lawyers will actually get anything. The lawyers say, and the employer does not deny, that the effect of their contingent-fee agreement with Mrs. Dunbar is to waive their entitlement to any fee paid out of the settlement proceeds beyond one third of the amount by which the $325,000 settlement exceeds the present value of the compensation award. As the contingent-fee contract does not indicate what interest rate will be used to compute the present value of the compensation award, it is not easy to figure out what the lawyers’ fee will be, but it will bear no necessary relation to the $79,445 that the employer argues should be deducted from his credit as attorney’s fees. Under the district court’s approach, the lawyers will get, in addition to their entitlement under the contingent-fee contract, $76 a week from the employer; for that amount is denominated as the fee to which the Act entitles them, and is over and above the amount to which the compensation award entitles their client.
Even if the lawyers did stand to get immediately the full $79,445 that the employer wants to deduct from his credit, the essential issues in this case would be unchanged. Only then it would be the widow rather than her lawyers complaining that the employer was refusing to pay the statutory fee. If the 25 percent fee comes off the top of the settlement proceeds, the settlement is worth less to the widow; it is as if she paid the fee, rather than the employer. True, she would make up part of the difference because she would resume receiving periodic compensation payments from the employer sooner. But since a long-deferred benefit is worth less than a current benefit, the employer will have succeeded in shifting a portion — and, we shall see, a large portion — of the 25 percent statutory fee from his shoulders to the widow’s (or to the lawyers’, if it is true, as they assert without contradiction, that they will not receive the attorney’s fee that the employer proposes to deduct from his credit).
To understand why the employer takes the position he does, we must compare the costs to him under the district court’s approach and under his approach. Under the district court’s approach the employer will have to lay out $76 in cash (actually $79, because the district court added on $3 in court costs — but we shall ignore this addi[622] tion as it raises no separate legal issue) week after week until his remaining credit of $288,119 ($325,000-$36,881, the amount he has already been reimbursed) is exhausted. If Mrs. Dunbar does not die or remarry before her 1,043 weeks of compensation entitlements are up, of which some 927 remain, then since the credit will cover the employer’s entire obligation under the compensation award (927 x $303 = $280,-881), he will end up having paid the lawyers the full $79,445 to which the statute entitles them. If she dies or remarries before the 927th week, he will pay less, but it will still be some positive and possibly quite large amount. Under the employer’s approach, with $70,225 subtracted from his remaining credit of $288,119 (he has, we noted, already paid $9,220 in attorney’s fees, for the amounts he was reimbursed), his net credit is only $217,894, which will last only 719 weeks. At the end of that time, with his credit exhausted, he will have to resume the $303 weekly payment to Mrs. Dunbar, and continue till the 1,043rd week should she live that long and not remarry. But should Mrs. Dunbar die or remarry before the 835th week (a number derived by adding the 116 weeks that elapsed before the tort settlement to the 719 weeks of the remaining credit), the employer will end up paying not a cent to the lawyers beyond the $9,220 that he paid when he was reimbursed for the compensation payments that he made to Mrs. Dunbar before the tort settlement. Suppose, for example, that she dies in the 636th week, which is to say 10 years after the settlement. Under the district court’s approach, the employer during these 10 years will have paid the lawyers $39,520 in legal fees, compared to zero under the employer’s approach.
Even if Mrs. Dunbar lives to the 1,043rd week without remarrying, the two approaches yield dramatic differences, once the time value of money is taken into account. Under the district court’s approach, the employer pays the $70,225 in attorney’s fees still due, evenly over the next 927 weeks. Under the employer’s approach, he pays the same amount (to her, by resuming payments sooner), and ends at the same time, but he does not begin to pay for 719 more weeks — almost 14 years. In comparing these values, we shall use a discount rate of 12 percent (and with the simplifying assumption that the employer makes all payments at the end of the year, rather than weekly). In many legal contexts, for example in computing the present value of lost future earnings in a personal-injury case, the proper discount rate is a real (i.e., inflation-free) interest rate, usually estimated to be about 2 percent. See, e.g., O’Shea v. Riverway Towing Co., 677 F.2d 1194, 1199 (7th Cir.1982). But that is on the assumption that if inflation continues, wages will rise to compensate for it, so that it is not necessary (or for that matter proper) to build inflation into the discount rate. Mrs. Dunbar’s compensation award is fixed for all time to come, and the lawyers’ fee is a fixed percentage of that award and therefore also not subject to an inflation adjustment. So a market rather than real interest rate is proper. Which interest rate is suitable would be in other contexts a challenging question, but our purposes here are illustrative and it will not matter whether 12 percent (the current interest rate on long-term government bonds) is just right.
If we discount the different streams of payments contemplated by the district court and the employer respectively to present value, using a discount rate of 12 percent, the present-value cost equivalent of the district court’s approach to the employer is about $28,000; under the employer’s approach it is less than $10,000 — little more than a third as much. Of course, the fact that the employer proposes an approach under which he pays much less (in present-value terms) than the district court thought proper does not answer the question of statutory interpretation. We have to decide which approach better serves the purpose of the statute. The purpose is evident; it is to give lawyers who confer a benefit on a third party, as these lawyers quite plainly did, a claim against the third party for a 25 percent fee. See Reno v. [623] Maryland Casualty Co., 27 Ill.2d 245, 247-48, 188 N.E.2d 657, 658 (1963). Had it not been for these lawyers’ efforts on Mrs. Dunbar’s behalf, the employer would have had to pay her $303 a week for a very long time. As a result of their efforts, the employer has gotten a credit which (attorney’s fees to one side) will eliminate his entire obligation to Mrs. Dunbar, because the tort settlement, which Mrs. Dunbar must deduct from her compensation payments, exceeds the sum of those payments. The statute says that the lawyers are entitled to 25 percent of “the reimbursement” which the employer receives or is to receive. That figure is $303 a week until Mrs. Dunbar marries, or dies, or until the 1,043 weeks is up — whichever occurs first. Every week that the employer, just by virtue of the credit he has gotten as a result of the tort settlement that Mrs. Dunbar’s lawyers procured, does not have to pay her $303, he obtains a benefit of $303, and the statute says that the fee for conferring that benefit is 25 percent — $76 for each and every such month. The district court’s approach is logical. It gives the lawyers the 25 percent they are entitled to.
The employer’s approach is not logical. To begin with, it gives the lawyers (or Mrs. Dunbar, if, contrary to our earlier suggestion, the employer’s approach contemplates that his attorney’s fee will be paid immediately, out of the proceeds of the tort settlement) less than 25 percent of the benefit conferred should Mrs. Dunbar die or remarry before the credit computed by the employer runs out. True, the way the employer computes his credit it will run out sooner than under the district court’s approach, because $70,225 has been deducted from it. But it will still last many years. Should Mrs. Dunbar die or remarry during that period (her age is not in the record), the employer will have incurred a benefit from the tort settlement equal to the number of weeks until she died or remarried, times $303, discounted to present value; but he will have paid nothing to the lawyers for that benefit. That cannot be what the statute contemplates.
Suppose Mrs. Dunbar lives the entire period without remarrying, so that the employer has to resume payments to her when his credit runs out. (His credit will run out under his approach, as we have explained, because of the deduction of the attorney’s fee from it.) In this case, because the benefit to the employer from the credit and the cost to him from having to resume payments earlier than he would have had to do but for the attorney’s fee come at different times, it is once again necessary to calculate the discounted present value of both figures, and compare them. The benefit to the employer on this basis is the discounted present value of $303 a week received for 719 weeks (i.e., until the credit, as calculated by the employer, runs out). At a 12 percent discount rate, this value is about $103,000. The cost to the employer on this basis is the discounted present value of the $303 a week that he will pay beginning after 719 weeks and continuing till the end of the compensation period. This present-value figure is, as we said earlier, less than $10,000, which is much less than 25 percent of the benefit to the employer, computed on a present-value basis — indeed less than 10 percent.
In short, the employer pays no statutory fee if Mrs. Dunbar marries or dies before the end of the thirteenth year after the tort settlement, and much less than 25 percent of the real value of his credit if she does not. And the percentage we calculated above actually exaggerates the generosity of the attorney-fee award under the employer’s approach. The percentage is artificially increased because, by subtracting the fee from the credit, the employer has made the credit smaller and thus changed the ratio.
We can estimate the actual shortfall from the 25 percent statutory fee by bringing into the present-value calculation some estimate of the risk of Mrs. Dunbar’s dying or remarrying during the 14 years at which the lawyers are at risk under the employer’s approach. Presumably the probability of her dying increases with age, and the probability of her remarrying decreases, so let us assume that the sum of the probabili[624] ties is a constant 1 percent for each of the 14 years. This estimate is arbitrary but the number doesn’t matter, only the principle. At this figure, the probability of her still being alive and unmarried after 14 years can be calculated to be about 87 percent (.9914; see Freedman, Pisani & Purves, Statistics 215 (1980)). Assume for the sake of simplicity that if she is still alive and unmarried after 14 years she will remain in that state till the end of the compensation period. Then the present value of the attorney’s fee under the employer’s approach is little more than $8,000. Yet expressed as a fraction of the real benefit to the employer, which is avoiding the whole $280,881 that he would have to pay Mrs. Dunbar in the last 927 weeks of the compensation period but for her lawyers’ efforts to obtain a tort settlement, a figure having a present value computed at a 12 percent discount rate of about $113,-000, the fee that the employer will have to pay is less than 7 percent of the benefit he obtained — a far cry from the statutory 25 percent.
When at argument we asked the lawyer for the employer to justify his suggested approach, he could offer us no reason based on the purpose of the statute; he admitted that the district court’s approach was reasonable; but he told us that the Illinois Appellate Court (Illinois’ intermediate appellate court) had adopted his suggested approach, and rejected that adopted by the district court, in four cases. Ordinarily we would give great weight to the Illinois Appellate Court’s interpretation of an Illinois statute, as evidence of Illinois law, which is binding on us in this diversity suit. But bearing in mind that the Illinois Supreme Court is the final authority on the meaning of Illinois statutes, and that an intermediate appellate court decision is not binding evidence of state law in circumstances when it is not a good predictor of what the state’s highest court would do in a similar case, we have decided that we ought not consider ourselves bound by the decisions that the employer has cited to us. See, e.g., West v. American Tel. & Tel. Co., 311 U.S. 223, 237, 61 S.Ct. 179, 183, 85 L.Ed. 139 (1940); White v. United States, 680 F.2d 1156, 1161 (7th Cir.1982); 19 Wright, Miller & Cooper, Federal Practice and Procedure § 4507, at p. 95 (1982).
In the first of these cases, Vandygriff v. Commonwealth Edison Co., 68 Ill.App.3d 396, 25 Ill.Dec. 47, 386 N.E.2d 318 (1979), the trial court indeed used a method quite similar to that proposed by the employer here. But on appeal the propriety of that method was not questioned; the only issue was whether the employer could be required in the circumstances to pay a 40 percent attorney’s fee rather than a 25 percent fee. In the second case, Denius v. Robertson, 98 Ill.App.3d 83, 53 Ill.Dec. 701, 424 N.E.2d 336 (1981), the same method that had been used in Vandygriff was commended as “straightforward and uncomplicated,” id. at 87, 53 Ill.Dec. at 90, 424 N.E.2d at 339, but again had not been challenged on appeal; the appeal involved unrelated issues. The third case, Jones v. Melroe Division, 102 Ill.App.3d 1103, 58 Ill.Dec. 934, 430 N.E.2d 1385 (1981), while taking issue with Vandygriff s computation in one particular, employed an approach similar to that urged by the employer here. An alternative approach proposed by the plaintiff, whereby “the costs of recovery paid by the third party plaintiff [the employee, in his suit against the tortfeasor] should be recovered by the plaintiff [the employee] from the employer as the benefit to the employer accrues,” was rejected as “but a variation on the theme we have adopted,” a variation “not mandated in Illinois” and one that “adds unnecessary complexity to an already complex computation.” Id. at 1111, 58 Ill.Dec. at 939, 430 N.E.2d at 1390. But from the court’s discussion it is impossible to tell on what ground the plaintiff had urged the rejected method, how closely it resembled the method used by the district court in this ease, wherein it would have added complexity to the computations, and in short why it was rejected. Here the district court’s method is no more complicated than the employer’s suggested method and the employer concedes that the district court’s method com[625] ports with the policy of the statute. The last case, Lewis v. Riverside Hospital, 116 Ill.App.3d 845, 72 Ill.Dec. 358, 452 N.E.2d 611 (1983), is like Vandygriff and Denius: it describes with approval an approach similar to that urged by the employer here, but the approach had not been challenged on appeal. The appeal dealt with such unrelated questions as whether the attorney’s fee should be 33V3 percent or the statutory 25 percent.
From this review of the cases it will be seen that only Jones provides support for a prediction that the Illinois Supreme Court might reject the approach taken by the district court in this case. But although Jones did reject an approach that bears some family resemblance to that taken by the district court in this case, the rejected approach was not described and the reasons for rejecting it may have no bearing on the present ease. Maybe the plaintiff had proposed a very complicated and confusing method which the court rejected for that reason; we cannot tell. In these circumstances, even Jones is a poor predictor of what the Illinois Supreme Court would do in a case like the present one.
The statute’s purpose is clear: it is to make the employer pay 25 percent of the benefits which he derives from the employee’s tort recovery against a third party, as compensation for conferring those benefits. The district court’s approach does that; the employer’s does not. We think the Illinois Supreme Court would agree if the issue were presented to it. The judgment of the district court is therefore
AFFIRMED.