This paper contributes to the econometric literature on structural breaks by proposing a test for parameter stability in vector autoregressive (var) models at a particular frequency ?, where ? ? [0, p]. When a dynamic model is affected by a structural break, the new tests allow for detecting which frequencies of the data are responsible for parameter instability. If the model is locally stable at the frequencies of interest, the whole sample size can then be exploited despite the presence of a break. The methodology is applied to analyse the productivity slowdown in the us, and the outcome is that local stability concerns only the higher frequencies of data on consumption, investment and output.