This paper considers a seasonal adjustment procedure that is capable of preparing data to the use in applied general equilibrium models. It is shown that standard seasonal adjustment procedures do not satisfy the property of invariance to deflating, that hinders their use in applied general equilibrium models. A system of axioms that describes the desired properties of a seasonal adjustment procedure is suggested. The impossibility of simultaneous fulfillment of additivity and invariance to deflation properties is shown. Therefore, one needs to choose the desired property depending on the type of the task that is solved.

The proposed procedure models the seasonality as a set of seasonal multiplicative dummy variables, so it can not only remove the seasonality, but also return it to the data in order to obtain forecasts. The procedure also has a built-in outlier detector, which enables it to handle noise and outliers in data of different types. It is compared to the popular X12 seasonal adjustment procedure using Monte-Carlo method. It is shown that the preciseness of the proposed procedure is comparable to X12 in terms of resistance to outliers and preservation of statistical properties of the series in the specific set of problems connected to the estimation of general equilibrium models. Several examples of its application to real data are shown. The obtained results allow us to make a conclusion about applicability of the suggested procedure to the removal of seasonality from the data that is used in the estimation of macroeconomic models.