This study examines interest rate pass-through across the conditional distribution using quantile cointegration. Using U.S. data from 1994–2024, we estimate long-run relationships between the federal funds rate and both lending rates and Treasury yields at quantiles \(\tau = 0.1\) –0.9, employing the Phillips–Hansen fully modified quantile estimator with quantile CUSUM stability tests. Quantile positions map approximately to economic conditions: lower quantiles correspond empirically to accommodative and crisis episodes, upper quantiles to tightening cycles. Two findings stand out. First, cointegration fails at the crisis extreme across virtually all rates, and at tightening conditions for lending rates and short-term Treasuries, suggesting conventional monetary policy faces its greatest challenges when economic conditions are most stressed. The medium segment—particularly the 1-year Treasury, which maintains cointegration across all quantiles—provides the most reliable transmission channel, while very short-term and long-term rates exhibit more fragmented patterns. Second, pass-through is systematically stronger during tightening than easing across all rates—a structural asymmetry confirmed in subsamples predating both the GFC and the COVID-19 episode. Temporal robustness checks suggest that full-sample tail instabilities may reflect extreme observations generated by unprecedented policy episodes at both distributional extremes, rather than permanent structural deterioration. These findings suggest that state-contingent frameworks—deploying unconventional tools at the distributional extremes where conventional transmission breaks down, and accounting for the systematically lower potency of easing relative to tightening—may enhance overall transmission effectiveness.
