Staff Report, Federal Reserve Bank of New York 232
The econometrics literature contains a variety of two-sided tests for unknown breakpoints in time-series models with one or more parameters. This paper derives an analogous one-sided test that takes into account the direction of the change for a single parameter. In particular, we propose a sup t statistic, which is distributed as a normalized Brownian bridge. The method is illustrated by testing whether the reaction of monetary policy to inflation has increased since 1959.