Univariate tests of momentum on the JSE

ABSTRACT

This study intends to test whether medium-term price momentum is present on the cross-section of shares listed on the JSE. Using the methodology of Jegadeesh and Titman (1993; 2001), shares are sorted applying four estimation periods (E = 3, 6, 9 and 12 months) and held for four portfolio holding periods (H = 3, 6, 9 and 12 months) post sort. The portfolio ‘optimisation’ methodology employed in this study provides additional insight when compared with previous tests conducted in South African literature, as further robustness tests are applied in order to determine the sensitivity of momentum to variations in sorting methodology, liquidity and trading costs. The study finds that estimation and holding periods between six and nine months generate the highest levels of excess return. Consistent with the findings of Basiewicz and Auret (2009) in relation to the size and value anomalies and Oladele and Bradfield (2016), equally weighted momentum excess returns provide a premium when compared to value weighting. Bid-ask bounce and microstructure effects are present on the JSE, given that momentum profits consistently increase when applying the skipping of most recent estimation month per Asness (1997) and, lastly, momentum is consistently more sensitive to direct transaction costs when compared to liquidity.

KEYWORDS:

• momentum
• liquidity
• transaction costs
• reversal
• price filter
• bid-ask bounce
• microstructure effects
• estimation period
• holding period