Discount rates for long-horizon loss periods: why a single blended figure invites cross
A thirty-year front-pay calculation discounted at a single blended rate is the easiest target a defense expert has. The Treasury yield curve already contains the answer. The model should use it.
The present-value discount in an economic damages model is the most scrutinized single assumption in the report. It collapses decades of projected earnings into a trial-date number, and small changes produce large swings. For short loss periods the choice of rate is mostly a disclosure issue. For long-horizon periods, typically fifteen years or more, a defensible approach is not optional.
The single-rate problem
A single blended discount rate applied across a thirty-year loss period implicitly assumes that the time-value of money is constant across that horizon. It is not. Treasury yields are term-specific. A one-year Treasury and a thirty-year Treasury reflect different expectations about inflation, real-rate dynamics, and risk. Using a one-size-fits-all rate ignores information the market is already providing.
When a defense expert crosses the plaintiff economist on a single-blended assumption, the first question is simple: why did you discount a year-one loss at the same rate as a year-thirty loss? There is no satisfying answer.
Yield-curve matching
The defensible alternative is to discount each year’s projected loss at the Treasury rate appropriate to that year’s duration from trial. A loss realized in year one is discounted at roughly the one-year Treasury yield at trial. A loss realized in year thirty uses the thirty-year yield. This approach is standard in financial practice and is well accepted in forensic-economic practice for any loss period of meaningful length.
Real versus nominal
A parallel question is whether to discount nominal projected earnings at nominal Treasury yields or real projected earnings at real yields (typically TIPS, Treasury Inflation-Protected Securities). Both approaches work if applied consistently. What cannot work is mixing: discounting nominal projected earnings at a real rate, or vice versa. That is the source of many published errors in forensic-economic work over the past two decades.
The practical choice is usually to stay nominal throughout, because earnings projections are built from nominal historical data and the translation to real terms introduces its own estimation noise. The model that stays nominal on both sides of the calculation is easier to defend.
What the report should show
The report should state the chosen discount rate for each year as a table or reference to the published Treasury curve on a specific trial or reference date. It should state whether nominal or real treatment is used. It should show the workings of the present-value calculation as a line item per projected year, not as a single aggregate figure. And it should include a sensitivity table that shows how the damages figure moves under plus-or-minus fifty basis points at both ends of the curve. That sensitivity table protects the rest of the report: the trier of fact sees that the chosen rate is defensible, and that reasonable alternative rates produce a bounded range of outcomes.
A discount-rate section that survives cross is not a section that picks one number and defends it forever. It is a section that picks one number, defends it, and shows the math for what would happen if the number were different.