What are the pieces of a lost earnings calculation?
A lost earnings model is a subtraction between two projected paths, with six recurring terms. The earnings base is what the person earned, established from documents, and the growth trajectory carries it forward. The duration term is how long the earnings would have continued, which in this field means worklife expectancy rather than an assumed retirement age. The compensation that travels with work, employer-paid benefits and, in injury matters, household services, is added, and the residual path is subtracted. The stream is then converted into an award through discounting, prejudgment interest and any tax adjustment, covered at Adjustments and present value.
The structure is not where cases are won. Opposing experts rarely disagree about the shape of the model; they disagree about the base, the trajectory, the worklife figure, the rate and the residual. The useful deposition question is therefore not “what method did you use” but “which table, which release, and which year,” followed by “what in this person’s record justifies departing from it.” An input that cannot answer both has traded the individual for an average, or cited nothing at all. The department page is Lost earnings and capacity.
Where does the earnings base come from, and what makes a self-employed claim different?
The base comes from documents rather than testimony: tax returns, W-2 forms, payroll registers, pay stubs, employment agreements and benefit statements, plus the personnel file in an employment matter. Those records establish what the person earned, over what period, in what job, and with what raise history, itself evidence of trajectory. Where the record is short, the trajectory has to be supported statistically, and published earnings data enters there.
A self-employed claim is a different analysis, and skipping the difference is the standard defect. Business profit is not personal earnings: it blends the owner’s labor with returns on invested capital, the labor of other people, and business risk. Separating out what the owner’s labor commanded, in effect what it would cost to replace that person, is the work, and it draws on the business’s tax returns and financial statements as well as occupational wage data for the role. Where the dispute is instead about what the business itself is worth, that is a valuation question under a standard of value imposed by law and belongs to the Business Valuation Institute at businessvaluationsinstitute.com. Where records are incomplete or contested, reconstruction belongs to the Forensic Accounting and Fraud Institute at forensicaccountinginstitute.com.
Front pay is the term whose endpoint is contested rather than statistical. Back pay runs over a period that already happened and can be tested against records; front pay projects forward from judgment, and its duration turns on how long the effects of the termination will plausibly persist and when comparable employment is likely. Whether it is available at all, and who decides it, varies by claim and is counsel’s question; the economic contribution is to give the endpoint a reasoned basis rather than an arbitrary horizon.
Which published data sources do these experts actually use?
Nearly all are federal statistical releases with a publisher, a reference period and a release date, and naming all three is what makes an input checkable. For earnings by education, the U.S. Census Bureau publishes the Current Population Survey Annual Social and Economic Supplement personal income tables, of which PINC-03, “Educational Attainment-People 18 Years Old and Over, by Total Money Earnings, Work Experience, Age, Race, Hispanic Origin, and Sex,” is the one cited in earnings-capacity work; the accompanying report, Income in the United States: 2025, was released September 15, 2026 from the 2026 CPS ASEC. For wages by occupation and area, the Bureau of Labor Statistics publishes Occupational Employment and Wage Statistics, whose most recent estimates cover May 2025 and were released May 15, 2026. For the wage-growth environment, the Bureau of Labor Statistics publishes the Employment Cost Index, most recently for the period ending June 2026 (USDL-26-1270, July 31, 2026), and its Employment Projections program released 2025 to 2035 projections on August 27, 2026.
For duration, two different tables do two different jobs. Worklife expectancy, the expected remaining years in the labor force, comes from the Markov model literature: the current reference is Gary R. Skoog, James E. Ciecka and Kurt V. Krueger, “The Markov Model of Labor Force Activity 2012-17,” Journal of Forensic Economics (2019), DOI 10.5085/JFE-457, updating their 2011 study at Journal of Forensic Economics volume 22, number 2, pages 165 to 229, with estimates by sex, age and education from 2012 to 2017 labor force data. Life expectancy, which governs a life care plan rather than an earnings stream, comes from the National Center for Health Statistics: the most recent national table as of September 2026 is United States Life Tables, 2024, National Vital Statistics Reports volume 75, number 5, released August 25, 2026. It replaced United States Life Tables, 2023, volume 74, number 6, which is the vintage still cited in a good deal of current work. The two are not interchangeable, and worklife is the shorter because it nets out exits from the labor force. The Markov structure is described at How the models work.
For the components that travel with work, the sources are narrower. Employer-paid benefits are reported in the Bureau of Labor Statistics release Employer Costs for Employee Compensation, June 2026 (USDL-26-1494, September 9, 2026), which states that for private industry workers benefit costs “averaged $14.07 and accounted for the remaining 30.0 percent” of employer costs. That aggregate describes a sector, not a plaintiff. The defensible practice is to value the person’s actual coverage and employer retirement contributions from their own benefit records, using the release as context rather than as a markup. Household services are valued from The Dollar Value of a Day, published by Expectancy Data, most recently in its 2023 edition released April 2025, which prices daily activity by demographic grouping from the American Time Use Survey and Bureau of Labor Statistics wage data. Personal consumption, the deduction that appears in wrongful death matters rather than injury ones, rests on household expenditure data from the Bureau of Labor Statistics Consumer Expenditure Survey, most recently Consumer Expenditures 2024 (USDL-25-1586, December 19, 2025).
Two cautions attach. A release is superseded on a schedule, and an expert citing a vintage already stale at the report date has handed over a free point: the Employer Costs release above names December 16, 2026 as its next publication date. And these tables describe populations, not people. This Institute cites them so counsel can check sourcing; it selects no values from them and estimates no earnings base, growth rate or worklife expectancy for anyone.
How do mitigation and offsets enter the calculation?
As the subtracted path, and in employment matters it is statutory. Title VII provides at 42 U.S.C. § 2000e-5(g)(1) that “[i]nterim earnings or amounts earnable with reasonable diligence by the person or persons discriminated against shall operate to reduce the back pay otherwise allowable,” and caps accrual in the same subsection: back pay liability “shall not accrue from a date more than two years prior to the filing of a charge with the Commission.” In Ford Motor Co. v. EEOC, 458 U.S. 219 (1982), the Supreme Court of the United States held that an unemployed or underemployed claimant “is subject to the statutory duty to minimize damages,” and that “absent special circumstances, the rejection of an employer’s unconditional job offer ends the accrual of potential backpay liability.” Whether that governs a claim is counsel’s question; for the model it means the search record is evidence.
In an injury matter the residual path comes from vocational evidence about what work the person can still perform and what it pays, rather than from the job search. The discipline is the same in both settings: the subtracted path deserves the evidentiary standard of the projected one, because the claim is the difference between them. Partial recovery is the normal case, and the model nets each period’s actual earnings against that period’s but-for earnings. The costs of mitigating, the search, the retraining, the relocation, belong in the loss.
Offsets are the separate question of what the conduct incidentally produced or saved, and they split into two screens. The economic screen asks whether the benefit flowed from the same conduct that caused the loss. The legal screen governs what the economics cannot resolve, most prominently collateral sources, where whether a payment from the plaintiff’s own insurance reduces the award depends on rules that vary by jurisdiction and claim. Carrying collateral items separately lets the model survive a ruling either way. The deduction layer is covered at Mitigation, offsets and tax.
When do taxes actually change the number?
In two situations, and which applies depends on the claim. The first is the treatment of the lost stream itself. In Norfolk & Western Railway Co. v. Liepelt, 444 U.S. 490 (1980), the Supreme Court of the United States held that in a wrongful death action under the Federal Employers’ Liability Act the trial court erred in excluding evidence of the effect of income taxes on the decedent’s estimated future earnings, reasoning that “[i]t is his after-tax income, rather than his gross income before taxes, that provides the only realistic measure of his ability to support his family.” The same opinion held it error to refuse the requested instruction that “your award will not be subject to any income taxes, and you should not consider such taxes in fixing the amount of your award.”
The second is the treatment of the award, and the claim families point in opposite directions. Under 26 U.S.C. § 104(a)(2), gross income does not include “the amount of any damages (other than punitive damages) received (whether by suit or agreement and whether as lump sums or as periodic payments) on account of personal physical injuries or physical sickness,” and the statute adds that “emotional distress shall not be treated as a physical injury or physical sickness.” An employment recovery generally sits outside that exclusion: in Commissioner v. Schleier, 515 U.S. 323 (1995), the Supreme Court of the United States applied a two-part test requiring that the cause of action be “based upon tort or tort type rights” and that the damages be received “on account of personal injuries or sickness,” and held that the taxpayer there “has failed to satisfy either requirement.” Whether a particular recovery is taxable, and whether a gross-up is available, are questions for counsel and the client’s tax advisers. Nothing here is tax advice.
What the economist owes in either regime is internal consistency. If earnings are projected after tax, the discount rate should be an after-tax rate, because the plaintiff investing the award pays tax on the interest it earns. Liepelt anticipated the point, saying of the tax on the income to be earned by the award that “[l]ogically, it would certainly seem correct that this amount, like future wages, should be estimated on an after-tax basis,” and Jones & Laughlin Steel Corp. v. Pfeifer, 462 U.S. 523 (1983), said the discount rate “should also represent the after-tax rate of return to the injured worker.” The mirror error is mixing real and nominal frames. Each defect is invisible in the bottom line and obvious in the workpapers, which is why rebuttal experts start there, and why the rate is treated at Discount rates and present value and at Worklife and discounting.