Skip to content

Research

Our research develops theory, econometric tools, and empirical evidence that take unforeseeable structural change seriously — from rational expectations under change to new methods for detecting instability.

Research Themes

  • Structural change in macroeconomic relationships
    How inflation regimes shift, policy frameworks evolve, and new patterns emerge that are not repetitions of the past
  • Rational expectations under unforeseeable change
    Formalizing how rational participants form expectations when the future is genuinely uncertain, not just risky
  • Implications for inference and policy
    What unforeseeable change means for evaluating empirical claims, forecasting, and policy analysis

Working Papers

  1. Rational Expectations of Inflation Undergoing Unforeseeable Change

    WP No. 1 · September 1, 2025
    Roman Frydman and Morten Nyboe Tabor
    Keywords inflation expectations structural change unforeseeable change Knightian uncertainty Muth's hypothesis
    JEL D83 D84 E31 E37
    Abstract
    The rational expectations hypothesis (REH) is widely regarded as representing rational market participants' expectations. However, REH presumes that market participants have perfect probabilistic foresight of future outcomes. In real-world markets, nonrepetitive, and thus unforeseeable, structural change renders such foresight inherently impossible. This paper proposes a new theoretical approach, which we call the Knight-Muth hypothesis (KMH), to representing rational participants' expectations as consistent with inflation undergoing unforeseeable structural change. Because such change gives rise to Knightian uncertainty, KMH recognizes that even when rational participants have access to full information, it is impossible for them to have perfect probabilistic foresight of future inflation. In contrast to REH, this enables KMH to reconcile key features of survey-based expectations with the assumption that participants are rational and have access to full information. Behavioral models and limited-information REH can account for only some of these key features by assuming participants are irrational or lack information. KMH implies that rational participants with full information form expectations that deviate from ex post inflation in nonrepetitive ways. Consistent with empirical findings, this implies nonrepetitive structural shifts in the bias and correlation of ex post forecast errors with ex ante information. Also consistent with survey-based empirical findings, KMH recognizes that rational participants form diverse expectations, even when they base them on full information, and that psychological factors play an important role in their expectations formation, particularly during periods of unforeseeable structural change in the inflation process.
  2. Unforeseeable Change in Rational Participants' Inflation Expectations: Evidence From Forecast-Error Regressions

    WP No. 2 · September 1, 2025
    Roman Frydman and Morten Nyboe Tabor
    Keywords inflation expectations forecast errors structural shifts unforeseeable change Knightian uncertainty Muth's hypothesis
    JEL D83 D84 E31 E37
    Abstract
    We present empirical evidence that the bias in market participants’ ex post inflation forecast errors and these errors’ correlation with ex ante information shift between nonrepetitive values. This finding is inconsistent with the predictions of full- and limited-information rational expectations (REH) models. Building on Knight’s and Muth’s pathbreaking insights, our theoretical account reconciles our empirical findings with participants’ rationality by recognizing that the inflation process undergoes nonrepetitive structural change that cannot be foreseen, even in probabilistic terms. Because REH models abstract from such unforeseeable structural change, they do not represent expectations of rational participants in real-world markets.

Econometric Tools

Our research is supported by regimes, an open-source Python package for structural-change econometrics and forecasting.

Learn more →