Barriers in U.S. Benchmark Bond Yields

In this paper, we examine United States Treasury benchmark bond yields for the existence of barriers. We find evidence of yield barriers at integer multiples of 0.25% (e.g., 6.00%, 6.25%, 6.50%, 6.75%) in the widely followed 30-year and 10-year benchmark bonds. The robustness of these results are confirmed with Monte Carlo simulations and by the absence of barriers in several control series. However, using market prediction tests we find that barrier effects do not lead to significant predictable influences in the conditional mean or volatility of these yield series. Using Monte Carlo simulations of yield series with built-in barriers, we demonstrate that the absence of predictability in the conditional mean or volatility are not necessary conditions for the existence of barriers. We also employ the built-in barrier simulations to demonstrate the high sensitivity of barrier test results to the choice of alternative hypotheses.