The performance of stock sangomas: evidence from the JSE

Margolis, Shaun
Journal Title
Journal ISSN
Volume Title
The study explores the performance of following analyst recommendations for equities listed on the Johannesburg Stock Exchange (JSE) during July 2007 to June 2013. Specifically, broker consensus recommendation levels are used to divide a sample of covered stocks into three portfolios. Using a calendar time based approach, the study provides a thorough analysis of raw and risk adjusted portfolio returns in order to establish whether value adding investment strategies can be formed around broker consensus recommendations. The overarching questions that the study asks of broker consensus recommendations are: can they provide market participants with significant portfolio returns over and above traditional return drivers and transaction costs? Do investors have to respond timeously once they are issued? Do they reliably deliver consistent value through varying economic cycles? Contrary to the results of Prayag and van Rensburg (2006) and Barber, Lehavy and Trueman (2001), stocks rated with average consensus levels outperformed stocks in the best rating category over the full sample period. However, this result appears to be driven by top rated stocks underperforming in the financial crisis. The study thus highlights that the value of consensus recommendations fluctuates over time. Further evidence of this is provided by examining the return spread between the best and worst rated stocks, which varied drastically, from a mean of 26 basis points during July 2007 – December 2009 to a mean of 138 basis points during the subsequent sub-period of the study ended June 2013. Whilst prior South African literature only adjusts returns for market risk, this study employs a wider range of asset pricing models, using an Arbitrage Pricing Theory (APT) methodology, so that returns can also be benchmarked in terms of other traditional determinants such as value, growth, size and momentum. After accounting for the above drivers of stock returns and transaction costs, it was determined that the value that analysts provide investors is likely to be embedded in equities which have ii poor (sell) consensus ratings. This is likely the result of the increased reputational cost that analysts face when issuing sell recommendations, as highlighted in Womack (1996). The study’s results suggest that in order for short-sellers to extract value from poorly rated stocks; they need to base their trades on the most recent information as the poor performance of the worst rated stocks were reversed in the long run. Furthermore, the study also casts doubt on the availability of abnormal returns after applying market capitalisation limits to account for short selling restrictions.
Thesis (M.Com. (Finance)--University of the Witwatersrand, Faculty of Commerce, Law and Management, School of Economic and Business Sciences, 2014.