Low-Frequency Components and The Weekend Effect Revisited: Evidence from Spectral Analysis

Abstract

We revisit the well-known weekend anomaly (Gibbons and Hess, 1981; Harris, 1986; Smirlock and Straks, 1986; Connolly, 1989; Giovanis, 2010) using an established macroeconometric technique known as spectral analysis (Granger, 1964; Sargent, 1987). Our findings show that using regression analysis with dichotomous variables, spectral analysis helps establishing the robustness of the estimated parameters based on a sample of the S&P500 for the 1972-1973 period. As further evidence of cycles in financial times series, we relate our application of spectral analysis to the recent literature on low-frequency components in asset returns (Barberis et al., 2001; Grüne and Semmler, 2008; Semmler et al., 2009). We suggest investment practitioners to consider using spectral analysis for establishing the ‘stylized facts’ of the financial time series under scrutiny and for regression models validation purposes.