Realistic Assumptions in Economic Models

Expert testimony backed by economic models is essential for showing discrimination in employment cases, proving causation in antitrust suits, and measuring damages in an array of complex business litigation. Courts lack consistency and coherence in admissibility decisions regarding expert testimony based on economic models, in particular when the opponent targets the model's assumptions as unrealistic. This article surveys the body of scholarship by economics methodologists on realism and assumptions, in particular the “reorientation” of the issue starting in the 1990s. The methodologists recognize different ways in which assumptions are realistic or unrealistic: by excluding factors, models are partial representations of the real world and therefore incomplete; assumptions may be confirmed or disconfirmed by evidence; and idealizations and omissions might be plausible or implausible to the target audience. The have also identified different functions for assumptions, such as excluding some factor as negligible, limiting the domains in which the model can be applied, and idealizing or omitting for purposes of tractability. Concern about whether assumptions are realistic is heightened in the context of applied models, such as those used in litigation. This survey leads to guidelines backed by theory for evaluating economics expert testimony for admissibility. All models are necessarily incomplete representations of the real world because they isolate and simplify, but so long as they offer insight about some relevant question, they are not unrealistic. If a modeler offers no explanation for a challenged assumption's purpose or if evidence disconfirms a key assumption, then the judge as gatekeeper is justified in declining to admit the testimony. If the modeler makes idealizations or omissions for the purpose of empirical tractability, then the realisticness of those assumptions is a question of plausibility for the factfinder. Sometimes common sense cannot help with evaluating a complex theoretical model or model built upon numerous questionable assumptions, so the judge should admit the model but take a heavier hand in evaluating it for sufficiency