Alternative seasonal adjustment methods for aggregate Irish macroeconomic data

Three distinct strands can be identified in the literature on seasonality. Economists have long been interested in removing high-frequency "noise" from individual economic time series, or "deseasonalizing the data" in common parlance. The second strand, on which an extensive technical literature has been developed over recent decades, treats seasonality as just one element to be encompassed in multivariate dynamic time series modeling, while a final strand seeks to model the economics of seasonality as the outcome of maximizing behavior by producing and consuming agents. Direct and indirect approaches have been compared to the seasonal adjustment of Irish macroeconomic series published by the Central Statistics Office. The presumption in the literature is that the indirect method is likely to be preferred in most real-world situations, but it is an empirical question whether it makes any great practical difference. The results indicate that it makes a very big difference indeed, with one-period growth rates frequently changing sign.