Jump Testing and the Speed of Market Adjustment

Asymptotic properties of jump tests rely on the property that any jump occurs within a single time interval no matter what the observation frequency is. Market microstructure effects in relation to news-induced revaluation of the underlying variable is likely to make this an unrealistic assumption for high-frequency transaction data. To capture these microstructure effects, this paper suggests a model in which market prices adjust gradually to jumps in the underlying effcient price. A case study illustrates the empirical relevance of the model, and the performance of different jump tests is investigated here and in a simulation study. Evidence indicates that tests based on the largest of scaled price increments perform better than tests comparing measures of variability. Resolving the matter by testing at lower frequencies turns out to be less straightforward.jumps, hypothesis tests, market microstructure noise, high-frequency data.