Cheating with Models

Research output: Contribution to journalArticlepeer-review


Beliefs and decisions are often based on confronting models with data. What is the largest "fake" correlation that a misspecified model can generate, even when it passes an elementary misspecification test? We study an "analyst" who fits a model, represented by a directed acyclic graph, to an objective (multivariate) Gaussian distribution. We characterize the maximal estimated pairwise correlation for generic Gaussian objective distributions, subject to the constraint that the estimated model preserves the marginal distribution of any individual variable. As the number of model variables grows, the estimated correlation can become arbitrarily close to one regardless of the objective correlation.
Original languageEnglish
Pages (from-to)417-434
Number of pages18
JournalAmerican Economic Review: Insights
Issue number4
StatePublished - 2021


Dive into the research topics of 'Cheating with Models'. Together they form a unique fingerprint.

Cite this