Searching for a Break in Gnp

Abstract

It has been suggested that existing estimates of the long-run impact of a surprise move in income may have a substantial upward bias due to the presence of a trend break in post war U.S. GNP data. This paper shows that the statistical evidence does not warrant abandoning the no trend null hypothesis. A key part of the argument is that conventionally computed significance levels overstate the likelihood of the trend break alternative hypothesis. This is because they do not take into account that, in practice, the break date is chosen based on pre-test examination of the data.