Is There Seasonality in Traded and Non-Traded Period Returns in the US Equity Market? A Multiple Structural Change Approach

This paper simultaneously examines day-of-the-week, turn-of-the-month and pre- and post-holidays calendar effects in traded and non-traded daily period returns in a group of broad-index exchange traded funds (ETFs) that track the major US stock indices (S&P 500, DJIA 30, NASDAQ 100 and Russel 2000 index). Bai and Perron (1998, 2003)’s method is employed to examine stability of the significant calendar effects over time and across ETFs. Results exhibit a high instability of the significant calendar effects over the various regimes up until 2001. From 2001 onwards and until the end of 2013, only a single regime across all ETFs is identified. In this last regime, results point to the disappearance of the previous significant effects across the US equity ETFs group. Unstable observed effects could have been motivated by market-specific conditions in such short time periods. The disappearance of these effects from 2001 onwards are consistent with the nature of this asset class: these ETFs are broadly diversified portfolios with diversification of private information, with higher liquidity and lower transactions cost, which is likely to reduce potential calendar effects.