On the Interpretation of Seasonally Adjusted Data.

This paper suveys several recently proposed regression-based methods that can be used to evaluate the usefulness of seasonally adjusted data, with specific focus on Census X-12 adjustment. These methods examine whether seasonality has indeed been removed and whether key properties (like a trend and a nonlinear business cycle) are left untouched by seasonal adjusment. If not so, one may doubt whether it is wise to use seasonally adjusted data for interpretation, that is, for evaluating the current state of the economy. The methods are applied to six quarterly macroeconomic variables for The Netherlands.