New tool detects structural breaks in panel data for better predictions.
The xtbunitroot Stata module tests if panel data series have unit roots with structural breaks. It considers one or two breaks in series components, like intercepts and trends, for fixed-T or large-T settings. The breaks can be known or unknown, with a bootstrap method for unknown breaks. The tests are robust to nonnormal errors, cross-sectional dependencies, and heteroskedasticity. They can detect if some or all series are stationary with breaks in their specifications.