On Low-Frequency Estimates of "Long-Run" Relationships in Macro- Economics

A number of recent studies have attempted to test propositions concerning "long runt" economic relationships by means of frequency-domain time series techniques that concentrate attention on low frequency co-movements of variables.The present paper emphasizes that many of these propositions involve expectational relationships that are not inherently related to specific frequencies or periodicities. Thus the association of low-frequency time series test statistics with long-run economic propositions is not generally warranted. That such an association can be misleading is demonstrated by analysis of examples taken from notable papers by Geweke, Lucas, and Summers