Hartman v. Duffey, 8 F. Supp. 2d 1 (D.D.C. 1998)

*2 Roger E. Warin, Mindy A. Kaiden, Steptoe & Johnson, L.L.P., Washington, DC, Bruce A. Frederickson, Jeffrey E. Fallon, Linda M. Correia, Susan L. Brackshaw, Jonathan C. Puth, Webster & Frederickson, Washington, DC, David L. Rose, Washington, DC, Douglas B. Huron, Stephen Z. Chertkof, Heller, Huron, Chertkof, Lerner & Salzman, Washington, DC, Laurel P. Malson, Terrence F. Flynn, Crowell & Moring, Washington, DC, for Plaintiffs.

Daniel F. Van Horn, Assistant U.S. Attorney, Washington, DC, for Defendants.

ROBERTSON, District Judge.

This employment discrimination suit was brought in 1977 by a class of women who alleged that they were wrongly denied employment at the United States Information Agency ("USIA"). In 1984, the late Judge Charles R. Richey found that defendant had "discriminated against women as a class with regard to hiring" in six occupational categories at USIA.^[1]Hartman v. Wick, 600 F. Supp. 361, 375 (D.D.C.1984), aff'd, 88 F.3d 1232 (D.C.Cir.1996). In January 1987, Judge Richey conducted a trial to determine appropriate remedies and thereafter issued an opinion and order requiring that individual Teamster hearings be held before a Special Master. Hartman v. Wick, 678 F. Supp. 312, 344-45 (D.D.C.1988). Judge Richey's order provided that back pay awards would be calculated on the basis of a "proxy-salary model."The proxy salaries are to reflect the "average salaries actually paid to persons currently employed by the agency who were hired into positions similar to the one that the class member would have filled absent discrimination." Id. at 337. The Special Master was given the responsibility for developing the proxy-salary model and for determining what fringe benefits (vacation leave, sick leave and medical coverage) and overtime and shift differentials would be included in awards of back pay. Id.

On January 31, 1992, with the consent of the parties, Special Master Stephen Saltzburg selected Drs. Cheryl and Martin Asher as the experts who would create the proxy-salary damages model. For some five years thereafter, the parties were intimately involved in the development of the damages model. In a May 22, 1995 Order the Special Master ordered the parties to "simultaneously provide [the experts] with all objections, suggestions and additional data relating to the proposed damages model no later than 5:00 p.m., Friday July 21, 1995." Objections made after that date would be considered waived.

After receiving and considering timely filed objections, the Ashers delivered their proposed damages model to the Special Master on June 15, 1996. Both sides had objections, defendant's being much more extensive than those of plaintiffs. The Special Master heard most of the objections on November 13, 1996, and rejected all the challenges considered at the hearing.

*3 The experts made further changes in response to the parties' objections and delivered the final damages model to the Special Master on March 1, 1997. After reviewing the model, the experts' explanations of their work, and the history underlying the model's formulation, the Special Master concluded in his report to this Court that this is "as fair a model as could possibly be fashioned." SM Rep. at 1. The Special Master "strongly recommend[ed] that [the damages model] be approved by the Court and used to compute back pay as to all claims." Id.

After considering the parties' objections to the damages model I have decided to affirm the recommendations of the Special Master in every respect except two: the rate of pre-judgment interest, and the application of an across-the-board factor to reduce health and life insurance benefits.

The parties agree that each successful claimant is entitled to receive interest on her back pay award from November 21, 1991, the effective date of the Civil Rights Act of 1991, until judgment is entered in her favor. They disagree, however, about what rate of interest to apply. Defendant argues for the three-month Treasury bill rate averaged over each year. The experts, who originally proposed the one-year Treasury bill rate, changed their opinions after Forman v. Korean Air Lines Co., 84 F.3d 446 (D.C.Cir.), cert. denied, ___ U.S. ____, 117 S. Ct. 582, 136 L. Ed. 2d 513 (1996), and now argue for the prime rate.

In Forman, which was a tort action between private parties, the court observed that "the prime rate is not merely as appropriate as the Treasury Bill rate, but more appropriate." Id. at 450. The court quoted In the Matter of Oil Spill by the Amoco Cadiz Off the Coast of France, 954 F.2d 1279, 1332 (7th Cir. 1992), which favored the "market rate" of interest for prejudgment awards because that is the rate "the victim must pay ... and the rate the wrongdoer has available to it." 84 F.3d at 450-51. Cf. Gorenstein Enterprises, Inc. v. Quality Care-USA, Inc., 874 F.2d 431, 436 (7th Cir.1989) (the risk of default bears directly on the appropriate rate of interest); Cement Division, National Gypsum Co. v. City of Milwaukee, 31 F.3d 581, 587 (7th Cir.1994) (allowing the district court to consider the City's status as a municipality when setting the interest rate), aff'd, 515 U.S. 189, 115 S. Ct. 2091, 132 L. Ed. 2d 148 (1995); In Matter of Oil Spill by Amoco Cadiz Off Coast of France, 954 F.2d at 1332 (noting that the market rate reflects in part the risk of non-payment).

Plaintiffs acknowledge that a government defendant does not borrow money at prime, but they argue, relying on 42 U.S.C.2000e-16(d), that they should have the benefit of the same interest rate that would apply in a case between private parties. Section 2000e-16(d) declares that in Title VII cases involving the government "the same interest to compensate for delay in payment shall be available as in cases involving nonpublic parties." 42 U.S.C. § 2000e-16(d). In the context of this case, that provision requires, at most, that the interest rate payable by the government be determined using the same Forman analysis that would be applied if the defendant were a nonpublic party. Money is "available to" the government at the rate of interest established by the sale of Treasury bills. Granted that victims cannot borrow money at the government's rate, there is no unjust enrichment to the government in selecting the one-year Treasury Bill rate as the rate of pre-judgment interest, and I believe it an appropriate use of my discretion to do so. That rate was the pre-Forman recommendation of the experts and has been, over time, about one-half percent greater than the three month T-bill rate urged by the government.

The damages model calculates the value of health and life insurance benefits for each claimant by applying a multiplier, known as the H & L Factor, to the claimant's imputed base salary for a given year. The H & L Factor reflects the percentage of the USIA's total personnel budget spent on those benefits in any given year. Defendant challenges the use of the multiplier, arguing that it incorporates the faulty assumption that health and life insurance benefits increase *4 in proportion to salary.^[2]

Defendant points out that the federal government does not pay for life insurance coverage but instead offers its employees an option to buy life insurance underwritten by the government. Agency spending on life insurance is for underwriting costs, not premiums, and is not correlated with individual salaries. An individual claimant may be entitled to an award representing her increased life insurance costs if she proves that she did buy life insurance in the private market and had to pay higher premiums than would have been available to her in government service, but any such award would have to be based upon individualized proof.

As for health insurance, defendant's evidence is that 84 percent of government health care plan enrollees receive 95 percent or more of the maximum government contribution allowable. Def.Obj. at 13. The government benefit increases in proportion to salary, but only to a point, and not beyond a maximum allowable contribution. Moreover, those individual plaintiffs who have been employed will be presumed to have received health insurance during their employment. The presumption may be rebutted in the individual case by a showing that the claimant did buy health insurance, or that the employer did not provide it, or both. Damages for lost health insurance for periods when a claimant was not employed, or when she was employed but did not receive health benefits, will be no more than the government's maximum allowable contribution.

Defendant suggests that each individual's recovery should be limited by the probability that the individual would have left the agency prior to the date of judgment in her case. To account for this probability, defendant suggests that the back pay award for each year should be reduced according to the agency-wide separation rate of 6 to 11 percent annually. Defendant argues that, without such an "average tenure analysis," the model "would be inconsistent with the Court's proxy salary approach and [would] impermissibly exaggerate back pay awards." Def.Mot. at 20.

The Special Master once rejected the average tenure analysis request in a February 9, 1993 order and has rejected defendant's renewed request. See SM Rep. at 35-43.

The grounds for the Special Master's decision are clearly correct:

1) Judge Richey's remedial order unambiguously stated that each prevailing plaintiff was entitled to back pay measured from the date of judgment: "In keeping with the remedial, make-whole purpose of back pay, the cut-off date of each back pay award will be the day on which the award is approved for the particular claimant." Hartman v. Wick, 678 F. Supp. at 336.

2) The damages model already cuts off back pay for any particular claimant at the date of her death or disability. As the Special Master noted, an average tenure model would have required the Ashers to make predictions as to the likelihood that any individual would die or become disabled. He concluded that the method adopted by the model "may actually favor Defendant rather than Plaintiffs," because, it "contains a tenure approach that imposes strict liability on the claimants...." SM Rep. at 40-41.

3) The cases erect a strong presumption against average tenure analysis. Thorne v. City of El Segundo, 802 F.2d 1131, 1136 (9th Cir.1986) ("absent compelling circumstances, ... the court *5 should compute the backpay award from the date of the discriminatory act until the date of final judgment"); Walker v. Ford Motor Co., 684 F.2d 1355, 1361 (11th Cir.1982) ("the injury continues until reinstatement.... [I]n the absence of evidence to the contrary, no reason exists to assume that employment for an indefinite term would not have lasted indefinitely"); EEOC v. Monarch Machine Tool Co., 737 F.2d 1444, 1453 (6th Cir.1980) ("under Title VII, the award of backpay will be computed from the date of first refusal until final judgment," unless "compelling considerations" warrant an earlier cutoff). Contra, EEOC v. Mike Smith Pontiac GMC, Inc., 896 F.2d 524, 530 (11th Cir. 1990) (upholding departure from the presumption in favor of back pay until date of judgment because evidence showed that no person had ever held the relevant position for that amount of time). The present case does not involve "compelling circumstances" for an earlier cut-off. Indeed, there is evidence in the record case that some individuals have held their positions at USIA for the entire period at issue.

A fourth reason for rejecting average tenure analysis is that the damages model already takes account of individual tenure. Each claimant's back pay will be limited to the time she was actually in the workforce (back pay is cut off by death, disability, retirement, or voluntary departure from the workforce), and limited for any year to the difference between an individual claimant's earnings and the earnings of her proxy at USIA. Factoring in the agency-wide separation rate would double count these events against plaintiffs.

The damages model developed by the Ashers and affirmed by the Special Master treats employee-paid retirement contributions as it does all elements of back pay without regard for the tax system. The rationale for disregarding taxes is that the parties would have had to spend excessive amounts of money to reconstruct the individual tax histories of each claimant even to begin considering the tax consequences of her award. Even if that were done, there would still be the problem of a lump sum recovery to contend with. The Ashers rationally determined that the costs of attempting tax calculations outweigh the benefits and developed the model without regard for taxes.

For the most part, the decision to ignore taxes favors the government. For example, the dollar value of fringe benefits such as health coverage would be included in an award of back pay and would be taxable, although such benefits would not have been considered income in the normal course. One aspect of the model does work against defendant: it assumes that employees would have contributed the maximum of seven percent of their salaries into their retirement accounts, and thus deducts seven percent from salaries (while giving claimants the benefit in their retirement accounts). The model deducts pre-tax dollars, however, when in actuality this money was not tax-deductible. Plaintiffs would have needed to earn more than the amount of the contribution in order to contribute seven percent of their salaries in after-tax dollars. Defendant's expert, Dr. Jonathan Walker, suggests that the damages model should reflect the reality that retirement contributions would have been made with after-tax dollars. The suggestion is not adopted. Defendant has not put forward a persuasive reason for taking account of the tax consequences of this one part of the model when it ignores the tax system in all other respects.

Judge Richey's remedial order directed the experts to develop proxy salary cohorts for "positions similar to the one that the class member would have filled." In developing their model, the Ashers decided to count, not only the initial entrance of an employee into a relevant occupation, but also any subsequent competitive promotions during that employee's career. Thus, the model in some instances counts the same employee more than once within the same cohort (if she filled a new competitive position in the same cohort) and includes the same employee in more than one cohort (if she filled a *6 position in a different occupational series code).

Defendant challenges these decisions and submits that a claimant should be placed only in one cohort for purposes of the model. Defendant contends that the Ashers, method skews the average salaries upward because a person already employed by USIA will likely have more seniority, and a higher salary grade, when she fills any new position that opens up. Analysis by defendant's expert, American Institute for Research, shows that second and later entrants into cohorts (meaning internal hires who take second and later competitive positions in a cohort) have salaries 29.3 percent higher than first entrants.

The Ashers contend that their approach best reflects the salary that these plaintiffs would have received absent discrimination because it is reasonable to assume that they would have been the ones who competed for and received the promotions to the positions reflected by inclusion of the multiple entrants. The Special Master accepted the Ashers' argument, reasoning that, while the model may permit double-counting and thereby inflate the average salaries in some of the cohorts, the double-counting is offset by the fact that the cohorts were "infected by discrimination." SM Rep. at 49. Thus, the Special Master argues, "the fact that some people working for USIA may have had experience or qualifications that produced slightly higher back pay is offset by the hiring of individuals based on gender who were not the most qualified for the position." Id. In addition, the Special Master noted that the Ashers "excluded step-ladder promotions," and focused only on positions that were competitively filled, thereby limiting cohort membership to those hires at issue in this case. I agree with the Special Master that "against this backdrop, the Ashers made a sound choice." Id.

The damages model does not deal with mitigation, leaving it to case-by-case determination. Nonetheless, the parties and the Special Master agree that it is appropriate at this time to decide an important issue about the calculation of mitigation: Will a claimant's earnings in any given year be used only to offset back pay in that year? Or should her earnings over the period of recovery be aggregated and subtracted from the proxy's total earnings over the period of recovery? The Special Master ruled in favor of the year-by-year approach. That ruling will be affirmed.

The Special Master's conclusion is solidly grounded in the case law and in the public policy underlying Title VII. Title VII's back pay provision was expressly modeled after that of the National Labor Relations Act, see Albemarle Paper Co. v. Moody, 422 U.S. 405, 419, 95 S. Ct. 2362, 45 L. Ed. 2d 280 (1975). The Supreme Court upheld the periodic approach to mitigation, NLRB v. Seven-Up Bottling Co. of Miami, Inc., 344 U.S. 344, 73 S. Ct. 287, 97 L. Ed. 377 (1953), reasoning that aggregation would give employers the incentive to delay reinstatement for as long as possible, "since every day the employee put in on the better paying job [would] reduce[ ] back pay liability." Id. at 347, 73 S. Ct. 287. Because timely reinstatement is an important remedy under Title VII, periodic mitigation is the preferred method for determining back pay liability in discrimination cases. See Darnell v. City of Jasper, Alabama, 730 F.2d 653, 656-57 (11th Cir.1984) (district court erred in using aggregate rather than quarter-by-quarter approach); Berger v. Iron Workers, 1994 WL 151292, *8 (D.D.C. Apr.14, 1994) (adopting year-by-year approach); Green v. United States Steel Corp., 640 F. Supp. 1521, 1553 (E.D.Pa.1986) (same), aff'd in part, rev'd in part, 843 F.2d 1511 (3rd Cir.1988), vacated on other grounds, 490 U.S. 1103, 109 S. Ct. 3151, 104 L. Ed. 2d 1015 (1989); see also Leftwich v. Harris-Stowe State College, 702 F.2d 686, 693 (8th Cir. 1983) (ADEA case); Dyer v. Hinky Dinky, Inc., 710 F.2d 1348, 1350-52 (8th Cir.1983) (Vietnam Era Veterans Act case).

Judge Richey's remedial order required that the damages model incorporate the value of fringe benefits, noting that "[a]mong the fringe benefits whose monetary value could be included in the back pay award are vacation leave, sick leave, and medical coverage." Hartman v. Wick, 678 F. Supp. at 336-37 (emphasis added).

*7 The Ashers concluded that this language did not require the inclusion of vacation leave and sick leave. SM Rep. at 53. They determined that the inclusion of vacation and sick leave could be equitably accomplished only by reconstructing the leave history of each claimant. Because such analysis would be costly and would yield only a limited benefit for plaintiffs,^[3] they decided to exclude annual and sick leave from the fringe benefits calculation. The Special Master affirmed that decision on the ground that it was not made "with an intent to favor one side or to disfavor the other ... [but to] make the entire model more understandable and more easily used by the parties." SM Rep. at 53. Plaintiffs object.

Early in the development of the model, plaintiffs advocated increases in back pay awards to account for leave. They have now abandoned that approach and contend that the model should provide for individualized leave accounts. The Special Master found that plaintiffs waived this argument by not raising it until after the cut-off date in 1995. I agree that the argument was waived. In any case, however, the Ashers reasonably determined that they were not required by the remedial order to include the value of annual and sick leave and that the cost of developing individualized leave histories would outweigh the benefits of such an analysis.

[1] Originally, there were six occupational series codes relevant to this case: Electronic Technician (856), Production Specialist (1071), Foreign Language Broadcaster (1048), Writer/editor (1082), Foreign Information Specialist/Foreign Affairs Specialist (1085), and Radio Broadcast Technician (3940). Because of the changes in USIA job classification, action in the 1001 occupational series code for those having International Radio Broadcaster job titles were added.

[2] The Special Master's ruling that defendant has waived its objection to the H & L Factor is overruled. It is true that neither side objected to use of the multiplier before the July 21, 1995 cutoff. Defendant points out, however, that the original multiplier included retirement benefits as well as health and life insurance benefits. The Ashers decided to disaggregate retirement benefits from insurance benefits in 1996. This change "simultaneously eliminated from the multiplier the bulk of the fringe benefits and the only fringe benefits that are actually proportional to salary. The net result was a pronounced increase in the multiplier's tendency to upwardly bias the damage model." The change materially affected the accuracy of the multiplier, and I consider the objection timely.

[3] For example, the Ashers pointed out, those plaintiffs who were not employed during the mitigation period received the benefits of leave. Moreover, most of those who were employed probably had leave benefits similar to what they would have enjoyed at USIA. The only damage would be to those plaintiffs who were employed, but whose employers provided less leave than USIA.