Title: Testing for deterministic seasonality in mixed-frequency VARs
Authors: Alain Hecq - Maastricht University (Netherlands) [presenting]
Tomas del Barrio Castro - University of the Balearic Islands (Spain)
Abstract: The mixed-frequency VAR is a multivariate modelling that stacks time series at different frequencies in order to jointly model the behavior of a low- and the high-frequency variables (respectively LF and HF hereafter). This approach complements the MIDAS regression in which a single non-linear equation from the LF to HF series is estimated. The MF-VAR modelling, although to some extent inherited from the periodic autoregressive models (i.e. PAR($p$)), is often estimated on seasonally adjusted data or, at least, the consequences of such seasonality components is not really accounted for. A strategy is provided to estimate a full range of interesting hypotheses about deterministic seasonal features within raw data. We also examine the presence of common seasonal features. We apply our testing framework on the relationship between quarterly employment and monthly tourist arrivals in the Balearic Islands.