Abstract

When a damages claim arises from a regulatory measure, the expert must reconstruct a world in which the measure was not adopted. In simple cases, that world is the actual world with the measure removed. In harder cases, removing the measure would also have changed the conduct of regulators and market participants, and those responses can never be observed, even after the event. This paper argues that the distinction between these two kinds of counterfactual, which it calls the Ceteris Paribus Counterfactual (CPC) and the Response Counterfactual (RC), is what distinguishes uncertainty in the regulatory counterfactual from ordinary valuation risk. It explains why uncertainty specific to the RC cannot generally be addressed through a conventional discount-rate adjustment and proposes three measurable diagnostics: the share of damages attributable to modeled responses under a specified decomposition, the sensitivity of damages to errors in the but-for cash flows, and an uncertainty profile indexed by time since divergence between the actual and counterfactual worlds rather than from the valuation date. It shows how the growth of that uncertainty can be calibrated from the regulatory calendar, the dispersion of the parties' scenarios and the history of regulatory reviews. Together with an error model and a selected evidentiary threshold, they provide an operational answer to the question in the title: future damages can be projected for as long as the counterfactual margin exceeds the growing uncertainty about the counterfactual itself, scaled by a degree of confidence selected in light of the applicable legal standard. The paper reconciles this with the principle of full reparation: under the proposed framework, the confidence threshold determines the years for which the existence of loss is established, while loss within those years is measured at its expected value. A worked example shows how the diagnostics are computed and why an equivalent premium on the discount rate misallocates the adjustment. The framework is then applied to AES v. Argentina , where the tribunal imposed an annual uncertainty adjustment of 1.25 per cent without setting out a quantitative derivation; the paper illustrates how an adjustment could be derived and calibrated under the proposed framework. A technical appendix sets out the derivations and the full assumptions of the example.