Sophia Trevino and her parents sued the United States for medical malpractice, pursuant to the Federal Tort Claims Act (FTCA), 28 U.S.C. Secs. 1346(b), 2674, arising from the negligent treatment given to Sophia's mother during Sophia's birth. The government concedes liability but appeals the award of over $6.3 million in damages. We modify in part, reverse in part, and remand.

On November 3, 1981, Rachael Trevino gave birth to Sophia at Madigan Army Medical Center in Tacoma, Washington (Madigan). During labor, Rachael suffered from a condition known as abruptio placentae, meaning that her placenta was partially detached from the uterine wall. The medical care that she received during labor did not take this condition into account. As a result, Sophia was born severely disabled. She has permanent brain damage, a form of cerebral palsy that involves all four extremities, and a seizure disorder. Evaluations by the University of Kansas Children's Rehabilitation Unit (the Kansas team) and by the government's expert witnesses suggest that Sophia will be mildly mentally retarded, but that she will be able to attend school and attain a fourth-grade level of reading and writing. Her emotional development should be normal. It is her set of physical disabilities, rather than her mental or emotional disabilities, that affect her the most. Experts assessed her gross motor skills as being at the level of a twelve-month-old child and her fine motor skills as being at the level of a twenty-four-month-old child. Although she probably will be able to walk unassisted at home, she will need crutches or a wheelchair outside the home. The Kansas team predicted that, with proper training and encouragement, Sophia will be able to function in a fairly independent manner. They contend that she probably will be able to work in a sheltered workshop setting and that she might even be able to work in the competitive market. The Trevinos contend that she will never be able to work in the competitive market. Her life expectancy is normal.

Sophia and her parents filed an FTCA action against the United States, alleging that Sophia's injuries had been caused by the negligent treatment provided by Madigan. After a five-day bench trial, the district court entered a judgment for the plaintiffs. It awarded Sophia $2,000,000 in nonpecuniary damages and $3,932,504 in pecuniary damages, and it awarded her parents as a unit $200,000 for loss of love and companionship and $200,000 for injury to the parent-child relationship. The United States moved for a new trial or, in the alternative, for an amended judgment. The court denied the motion, and the United States appeals.

We review damage awards in FTCA cases for clear error. Shaw v. United States, 741 F.2d 1202, 1205 (9th Cir.1984). The award is clearly erroneous if, after a review of the record, we are " 'left with the definite and firm conviction that a mistake has been committed.' " Id. (quoting United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S. Ct. 525, 542, 92 L. Ed. 746 (1948)). To determine whether a given award is excessive, we look to the relevant state's case law on excessive awards. Id. The state of Washington considers awards excessive "only if the amount shocks the court's sense of justice or sound judgment" and if it "appears that the trial judge was swayed by passion or prejudice." Id. at 1209. To make that determination, we compare the challenged award to awards in similar cases in the same jurisdiction. Id. The choice of a discount rate, used to adjust to present value an award based on an income stream spread over time and, at the same time, to adjust for the effects of inflation, should be reviewed for an abuse of discretion. Cf. Jones & Laughlin Steel Corp. v. Pfeifer, 462 U.S. 523, 546-47, 103 S. Ct. 2541, 2555, 76 L. Ed. 2d 768 (1983) (suggesting caution be used in determining a discount rate, given the uncertainties of future economic events).

The district court awarded Sophia $2.0 million for her pain and suffering and her mental anguish. See Record Excerpt (R.E.) at 14. The government contends that this award is excessive in light of Shaw v. United States, 741 F.2d 1202 (9th Cir.1984) and comparable Washington cases.

Washington law permits damages for mental anguish and for pain and suffering. See Shaw, 741 F.2d at 1208 (citing Parris v. Johnson, 3 Wash.App. 853, 860 n. 2, 479 P.2d 91, 95 n. 2 (1970)). Our review of an award for nonpecuniary injuries is a "delicate and difficult" process, see Felder v. United States, 543 F.2d 657, 674 (9th Cir.1976), precisely because we must review an award that compensates for damage to intangibles. Since we cannot examine this type of award with any type of mathematical precision, we compare awards in similar cases in order "to maintain some degree of uniformity." Shaw, 741 F.2d at 1209.

In Shaw, Scotty Shaw was injured by the negligent obstetrical care that his mother received. His injuries included "spastic quadraparesis, blindness, a seizure disorder, and profound mental and physical retardation." Id. at 1204. The district court awarded $5.0 million in nonpecuniary damages. We reduced the award to $1.0 million after noting, first, that the highest reported verdict for medical malpractice at that time was $1.1 million, id. at 1209, and second, that Scotty had not been left wholly incapable of feeling and perceiving life, see id.

It may be true that medical malpractice awards in Washington have increased much beyond the $1.1 million noted in Shaw. Cf. Bingaman v. Grays Harbor Community Hosp., 103 Wash.2d 831, 699 P.2d 1230 (1985) (en banc) (reinstating the district court's award of $412,000 for the plaintiff's pain and suffering during the thirty-five hours preceding her death). Nonetheless, an award of $2.0 million is excessive for a child who will be able to attain a fourth-grade reading and writing level, who will be able to work during her adult life, and who has been described as "a delightful little child who obviously expects to be responded to positively by others," R.E. at 46 (evaluation by the Kansas team), and who is "imaginative in her play," id. at 40. Such an award indicates a mistake has been made and it shocks our sense of sound judgment. Accordingly, we reduce the nonpecuniary award to Sophia to $1.0 million.

C. The Nonpecuniary Award to Sophia's Parents

As already pointed out, the district court awarded Sophia's parents as a unit $200,000 for the loss of her love and companionship and $200,000 for the injury to the parent-child relationship, a total amount of $400,000. See R.E. at 13. The government argues that this award, too, is excessive. We agree. The award shocks our sense of sound judgment also.

An award for loss of love and companionship is analyzed separately from an award for injury to the parent-child relationship. Under Washington law,

[r]ecovery for loss of companionship compensates the parents for the value of the child's mutual society and protection. Damages for destruction of the parent-child relationship are intended to alleviate parental grief, mental anguish, and suffering.

Shaw, 741 F.2d at 1209 (citation omitted). In Shaw, the district court awarded $1.0 million on each of these two grounds to the parents, and we reduced the award to a total of $50,000. See id. at 1209-10. Once more we acknowledge that damage awards may have risen dramatically in the last few years. Nonetheless, Sophia's parents will be able to enjoy her company to a far greater degree than would Scotty Shaw's parents. Cf. id. at 1210 (recognizing that raising a severely handicapped child would be difficult but would not constitute a total loss of the child's company). We reduce the nonpecuniary damages awarded to Sophia's parents as a unit to $50,000 on each of the two theories of damages, a total of $100,000.

The district court awarded $1,757,667 in lifetime attendant care for Sophia. See R.E. at 13. The government contends that this award is clearly erroneous on two grounds: first, because the court improperly permitted an unqualified witness to testify as an expert, and second, because the award itself was clearly erroneous in light of the record as a whole.

The government's first argument must fail. "[I]n civil actions and proceedings, with respect to an element of a claim or defense as to which State law supplies the rule of decision, the competency of a witness shall be determined in accordance with State law." Fed.R.Evid. 601. Washington grants broad discretion to the trial judge in ruling on the competence of expert witnesses. See Balmer v. Dilley, 81 Wash.2d 367, 372, 502 P.2d 456, 459 (1972) (en banc). Admittedly, the Trevinos' expert witness, Paula Behrhorst, was far from being well-qualified to testify. She has never treated a child with cerebral palsy and has not treated many mentally retarded children. See 2 Reporter's Transcript (R.T.) at 109. She has never evaluated developmentally disabled children. Id. at 116. But these deficiencies in the witness's qualifications go to the weight to be given her testimony and not to her ability to testify. See In re Young, 24 Wash.App. 392, 397, 600 P.2d 1312, 1315 (1979).

The government's second argument, however, is persuasive. Viewing the record as a whole, the district court's award of attendant care was clearly erroneous. Behrhorst's training was spotty at best; her examination of Sophia was cursory; her conclusions were not well-founded. See 2 R.T. at 107-25. For example, after noting that Sophia could ride a tricycle, Behrhorst dismissed that accomplishment, saying that "[t]he fact that she can ride the bicycle, or tricycle, doesn't have any bearing on whether she can feed herself, and dress herself, and toliet [sic] herself." Id. at 115. Behrhorst also contends that Sophia, when working, will need ten hours of bilingual attendant care a day. Id. at 118-20. The government's expert witness, Marianne Taylor, is the Director of Training and Nursing for the University of Kansas Children's Rehabilitation Unit. 4 R.T. at 43. She has evaluated between 700 and 1000 handicapped children, with approximately 200 of those children having "mixed" handicaps. Id. at 43-44. Taylor, after observing and evaluating Sophia, contended that constant attendant care would be harmful; such care would foster dependence. Id. at 49, 51-52. She compared Sophia's condition with the condition of a child who would normally receive ten hours of attendant care a day. The latter type of child, Taylor pointed out, would "not even [be] able to hold their [sic] head up, or roll over independently, is not able to assist at all in feeding [himself], and may in fact need to be fed by a tube." Id. at 52-53. The district court's grant of attendant care is wholly unsupported by the record. It shocks our sense of sound judgment. We eliminate this award entirely.

The district court applied a -2% discount rate to the pecuniary component of the damage award. It explained its reason for this choice as follows:

The -2% discount rate was determined by the application of a set formula which assumes as [sic] constant relationship between inflation and discount rates. The differential between the growth rates and the tax-free rate of return has been stable and has averaged more than two percentage points during the 30-year period 1954-1984. To obtain the present value of economic loss using tax-free short-term interest as the appropriate discount rate, a net discount rate of -2% was applied (wage and cost increases exceed tax-free interest rates by 2%) and calculated [with] the amounts testified to by the economic expert, Dr. Lowell Bassett.

R.E. at 14. We hold that the district court abused its discretion in applying a negative discount rate to the estimated pecuniary losses. Such a rate relies on assumptions not generally accepted by the courts or economists.

The Supreme Court addressed the discount rate issue in its opinion in Jones & Laughlin Steel Corp. v. Pfeifer, 462 U.S. 523, 103 S. Ct. 2541, 76 L. Ed. 2d 768 (1983). Its analysis of the methods of computing damage awards, including the choice of the proper discount rate, need not be repeated except to the extent necessary to reveal our thinking. We quickly point out, however, that the Supreme Court did not approve the use of a negative discount rate even though its survey of the economic and legal literature touched on material that might have suggested it. See 462 U.S. at 548 n. 30, 103 S. Ct. at 2556 n. 30.

We begin by recognizing that awards based on income streams spread over time are usually discounted to present value to account for the fact that a plaintiff, by receiving the money in a lump sum, "up front," will invest the sum and earn additional income from the investment. See Pfeifer, 462 U.S. at 536-37, 103 S.Ct. at 2550; Fitzgerald, Economic Loss in Wrongful Death: Principles of Evaluation, 44 Ins.Couns.J. 427, 431 (1977). The discount rate should be based on " 'the best and safest investments.' " Pfeifer, 462 U.S. at 537, 103 S.Ct. at 2550 (quoting Chesapeake & O. Ry. v. Kelly, 241 U.S. 485, 491, 36 S. Ct. 630, 632, 60 L. Ed. 1117 (1916).1 Were it not for inflation, that simple calculation would end the matter. With inflation, however, dollars received in the future will buy less than would those same dollars if received today. A net positive discount rate implies that the gains from safe investments exceed the losses induced by inflation--in other words, that the rate chosen to reflect the interest rate on the safest investments over a fixed period of time will be a larger number than the rate chosen to reflect the rate of inflation over the same period of time. Professor Jerome Sherman's analysis of the relationship between government bond yields (the investment factor) and inflation illustrates the point:

Yields on treasury bonds have stayed slightly ahead of the rate of inflation during the past 32 years [1947-78] (4.4 percent vs. 3.6 percent). This means that the real return on treasury bonds was 0.8 percent.

Sherman, Projection of Economic Loss: Inflation v. Present Value, 14 Creighton L.Rev. 723, 732 (1981).

Obviously it is possible for the true rate of inflation to outstrip the return on the safest investments for some period of time. This would justify for that period of time a negative discount rate. The district court's use of a negative discount rate is based on this possibility.

The district court's choice is flawed, however. First, it relied on an unrepresentative timespan. That period was from 1954 to 1984. R.E. at 14. This span includes the aberrational years 1974-82: years in which oil prices tripled from 1974 to 1979, inflation reached double-digit proportions, and "[r]ecessions in 1974, 1975, 1980, and 1982 mark[ed] the 1974-82 period as the most recession-plagued nine years since the Great Depression." Formuzis & Pickersgill, Present Value of Economic Loss, Trial, Feb. 1985, at 22, 23. As one article summarized the state of the economy for the past thirty-five years or so,

In addition to the upward trend in inflation and interest rates evident in most of the post-1947 period, there has been an increase in the variability about the trend. Between 1947 and 1967, the average yearly inflation rate was 2 percent, and the range between the high and low values was 4.6 percent. For interest rates on long-term government securities, the corresponding figures were 3.37 average and a range of 2.6 percent. In the following sixteen-year period, however, the average figures for inflation and interest rates were 7.27 percent and 7.93 percent while the ranges were 10.5 percent and 7.62 percent respectively. In fact, the interest rate on ten-year government bonds fluctuated nearly as much in 1981 alone as it did for the entire period 1947 to 1967. This volatility indicates that forecasters have had a more difficult assignment in recent years and that forecast errors, which have been large in recent years, can be expected to remain that way.

Mead, Calculating Present Value, Trial, July 1984, at 16, 18. The truth of this observation is borne out when one examines the period 1926-1981, in which the average annual inflation rate was around 3 percent. R. Ibbotson & R. Sinquefield, Stocks, Bonds, Bills, and Inflation 15 (1982 ed.).

We cannot deny history, nor can history provide an always reliable basis for predicting the future. However, we can base our estimates on long time periods that will diminish the effect of shorter aberrational periods. Fluctuations that are great for a short time span are less dramatic, and skew results less, when they are seen as part of a longer period. We have no confidence in the ability of experts, the district court, or this court, to predict inflation or interest rates over the period of Sophia's life other than by extrapolating from the past.

The district court must also select an accurate measure of historical inflation as the basis for its prediction of future inflation. Here, the district court used the historical increase in wages and medical costs. Supplemental Record Excerpts at 48. But this gives the illusion of greater inflation, because some portion of wage increases is not due to inflation:

Because the growth rate of wages includes a component attributable to pay increases due to increased education, age and maturity, and increases in productivity, as well as a component attributable to inflation, the difference between the interest rate on a secure investment and the rate of growth of wages understates the real interest rate by whatever proportion of the growth rate of wages is not attributable to inflation.

Sauers v. Alaska Barge & Transp., Inc., 600 F.2d 238, 246 n. 15 (9th Cir.1979). The district court might have supposed that the historical rise in wage rates was a proper measure of inflation for the purpose of calculating a discount rate to be applied to the wage portion only of Sophia's award, because Sophia deserved to be compensated for future increases in her wages due to her projected increased productivity, as well as for increases in her wages due to projected inflation. But Pfeifer denies that supposition. A plaintiff may be compensated for projected future increases in his wages due to increases in his productivity if he proves that he would have been likely to receive wage increases by reason of his increased productivity. 462 U.S. at 534-36, 103 S.Ct. at 549-50. Because Sophia made no such showing, the district court erred when it used the historical rise in wages as a measure of inflation for the purpose of calculating the discount rate for the lost wage portion of Sophia's award. Furthermore, the rate of increase in wages may not be the same as the rate of increase in medical costs over the same period. For this reason, the measure of inflation for the purpose of calculating the discount rate to be applied to the medical expense portion of Sophia's award may be different than that employed in fixing the discount rate applicable to the lost wage portion of her award.

Pfeifer 's discussion of discount rates technically is only an interpretation of section 5(b) of the Longshoremen's and Harbor Workers' Compensation Act (LHWCA). See 462 U.S. at 547, 103 S.Ct. at 2555. But the general discussion of alternative methods of calculating discount rates is extensive. The Court mentions several options:

1. Calculate the lost income stream by excluding the effects of inflation and the "real" interest rate by fixing the difference between the market rate of interest and the anticipated rate of inflation, id. at 541-42, 548, 103 S. Ct. at 2552-53, 2556;

2. Calculate the size of the lost income stream by including the effects of inflation and discounting by the market interest rate, id. at 543, 547-48, 103 S. Ct. at 2555-56; and,

3. Calculate the value of pecuniary damages by employing a zero discount rate (the total offset approach), id. at 544-46, 549-50, 103 S. Ct. 2554-55, 2556-57.

The last option (the total offset approach) was considered unacceptable as a uniform method to calculate damages, although it could be stipulated to by the parties, or applied by the trial court in an appropriate case. See id. at 550-51, 103 S.Ct. at 2557. After examining a plethora of economic studies, the Court concluded as to the "real" interest rate option that: "Although we find the economic evidence distinctly inconclusive regarding an essential premise of those approaches, we do not believe a trial court ... should be reversed if it adopts a rate between 1 and 3% and explains its choice." Id. at 548-49, 103 S.Ct. at 2556 (footnote omitted). Obviously a decision interpreting section 5(b) of the LHWCA is not binding in an FTCA case; however, the Court's guidance on the issue of economic predictions and discount rates cannot be disregarded. We accept this guidance and choose to regard it as controlling in this case. In reliance on Pfeifer, and in the absence of Washington state law to the contrary, we decline to embrace a negative discount rate on the basis of this record.

We do not hold, however, that any discount rate above 3% or below 1% is impermissible. Supported by credible expert testimony, a court could certainly adopt a different rate. Cf. Culver v. Slater Boat Co. (Culver II), 722 F.2d 114, 121-22 (5th Cir.1983) (en banc) (requiring courts to use a below-market rate calculation and permitting courts to use a -1.5% discount rate if supported by expert opinion). We do hold, however, that it is impermissible to select a time period over which to compare inflation and interest rates that provides a decidedly aberrational result. We also point out that it is impermissible either (1) to exclude the effects of inflation in determining the size of the lost income stream and employ a discount rate equal to the market rate of interest, or (2) to include the effects of inflation in determining the size of the lost income stream and employ a discount rate measured by the difference between the market rate of interest and the rate of inflation. The former denies the injured party any adjustment for inflation while making available such adjustment to the party deemed responsible for the injury. The latter, on the other hand, provides to the injured party an adjustment for inflation in determining the size of the lost income stream and denies to the party deemed responsible for the injury any benefit of that adjustment in determining the proper discount rate. See Pfeifer, 462 U.S. at 538-52, 103 S.Ct. at 2251-58, passim. Put more succinctly, the former accords the party deemed responsible for the injury a "double benefit," while in the latter the "double benefit" passes to the injured party. Neither party is entitled to a "double benefit."

Finally, we point out that the choice of an interest rate yield on Sophia's lump sum award, or for other purposes in fixing the final amount of the award for the loss of earning capacity, is not dictated by the manner in which taxes are treated. The yield of tax-free securities enjoys no special status. That yield merely indicates what a yield would be to an investor who is strongly averse to both risk and taxes. No sensible investment counselor would advise Sophia to invest the entire lump sum in tax-free securities. Nor is it probable that she will do so. For these reasons the use of the yield of tax-free securities to arrive at the "real rate" of interest is suspect. The tax-free rate of interest is substantially below that yielded by securities not enjoying that privilege. Its use, when coupled with a period of unrepresentative high inflation, as was done in this case, yields an unnaturally low discount rate. This should be avoided on remand.

The necessity to treat inflation consistently in fixing the size of the future income stream and the proper discount rate suggests a somewhat similar problem in connection with taxes. Sophia's economist made certain adjustments in his calculations with respect to taxes that would have been imposed on Sophia's earnings had she not been injured and taxes that would be paid with respect to the earnings the lump sum would earn. The purpose of these adjustments was to make certain that Sophia did receive after taxes in the future what she would have received after taxes had she not been injured.

These adjustments must be consistent. That is, if Sophia's future earnings are reduced by taxes, as is permitted by Norfolk & Western Railway Co. v. Liepelt, 444 U.S. 490, 100 S. Ct. 755, 62 L. Ed. 2d 689 (1980), then the interest yield of the lump sum award should be reduced by taxes. Id. at 495, 100 S.Ct. at 758. On the other hand, if Sophia's future earnings are not reduced by taxes, then the interest yield also should not be reduced for taxes. When both the earnings and the interest yield are adjusted for taxes, the earnings adjustment reduces the amount of the lump sum while the interest yield adjustment increases it. When neither is adjusted for taxes, the absence of an adjustment to earnings increases the lump sum, while the absence of an adjustment to interest yield diminishes the after tax value of the lump sum.

The award of $2.0 million to Sophia for her nonpecuniary damages is reduced to $1.0 million. The award of $200,000 to Sophia's parents as a unit on each of two grounds--for loss of love and companionship and for injury to the parent-child relationship--is reduced to $50,000 per ground, a total of $100,000. The award of $1,757,667 for Sophia's attendant care is eliminated. We remand for the recalculation of Sophia's pecuniary loss in accordance with the Part II.E. of this opinion.

MODIFIED IN PART, REVERSED IN PART, AND REMANDED.