Over the years, a large number of literature reviews have been documented to examine the ability of a number of cross sectional anomalies to predict future stock returns. Fama and French (1993), noted that some of the risks relating to these phenomena include risk, transaction costs behavioural biases and Liquidity as documented by Shleifer & Robert in 2012. In addition, Cajueiro and Tabak (2004), argued that transaction costs have been declining greatly over the last three decades. This is an important factor behind the shifting investment patterns. It is also important to mention that trading costs are the main constituents in performance. A paper by Schwerts in 2003, questions the viability of anomalies and how long a variable’s predictive ability can persist. With regard to this, a new literature is developing that documents a reduction in stock return predictability due to some of these well-established variables. For instance, according to Chordia and Subrahmanyam 2013, they provide evidence that majority of the market anomalies have attenuated and the average returns from a portfolio strategy based on prominent anomalies have decreased by almost half after decimalization. They also find that the predictive ability of the most prominent cross-sectional predictors such as firm size, standardized unexpected earnings, idiosyncratic volatility, changes in shares outstanding, accruals, past twelve month return and book-to-market ratio are considerably weaker especially for liquid firms. This is attributable to decline in predictive ability to increased arbitrage activity (i.e. increased asset management by
 Amihud and Mendelson, 1986  Lakonishok et al, 1994
hedge funds and increases in short term interest rates), and reduced transaction costs (decrease in tick sizes and increases in the volumes of trading). As per Charles Jones, a professor of Finance and Economics at Columbia Business School who tracked transaction costs historically, puts it that spreads have fluctuated widely over the 20th century. This is mainly attributed to episodes of market turmoil such as the Great depression of 1930 and the energy crisis of 1970s. He adds that spreads in the recent financial times have been experiencing an downward trend and decreased gradually from an average of 1.5% in the early 1930s to as low as 0.2% in the 21st century, especially on the Dow Jones Industrial Average stocks.
A Plethora of finance articles have also documented that the decrease in excess returns to various cross-sectional trading strategies have been attributed to investor learning, increased arbitrage activities and decrease in transaction costs. Lopez de Prado et al (2011), found that as a result of lower trading costs, more money and trading volume to equity increased. In addition liquidity increased from the mid-20th century levels. As the shifts in transaction costs took along, James Porterba, an economist at Massachusetts Institute of Technology, found that households have also increased their equity holdings during 1990s. This implies that retail investors have been injecting sizeable inflows of money into equity mutual funds. This led McLean and Pontiff to study the return predictability of approximately 97 different anomalies and find out that on average the post-publication return predictability decrease by about 35% with a larger decrease of
 (Easley, Lopez de Prado, & O'Hara, 2011)
more than 35% attributable to the post-earnings announcement drift (PEAD) and accrual anomaly. In addition, they discovered that the abnormal returns to accrual anomaly strategy have not been reliably positive since 2001. On the other hand, the abnormal returns to the post-earnings announcement drift (PEAD) strategy had been eliminated between the periods of 1991-1997 for large cap companies, instead, the PEAD persists where arbitrage costs are highest and also among companies with little or no analyst information or with low stock prices. Lastly, according to Scott Richardson et al (2010), they evidence that after the elimination of transaction costs; returns to the PEAD and accrual anomaly have decreased over the years (Richardson, Irem, & Peter, 2010). The papers are consistent with the efficient market hypothesis (EMH), as it learns about new information and then reduces or eliminates profitable trading opportunities.
Consistent with Richardson et al (2010), Cornell University professors David Easley, Maureen O’Hara and Soren Hvidkjaer developed the probability of informed trading measure (PIN) which follows the notion that as the number of retail investors increases the relative size of informed trading decreases which in turn produces tighter spreads as a result of the market makers feeling less exposed to fundamental risk. Ideally, this implies that specialists setting the range between bid and ask spread tend to lose only against informed traders as a result of their skill set in detecting permanent changes in fundamental stock values. With regard to this, there is a positive relationship between liquidity and uniformed trading. The probability of informed trading measure (PIN) has
 (Johnson & William, 2001)
been documented by Canisius College Professor Yuxing Yan and Hong Kong University’s Shaojan Zhang, where they measured PIN using a higher frequency data between 1993 and 2004. Their results found support for the predictive capability of market microstructure and the post earnings announcement drift. They also found that in these microstructure patterns and periods of elevated enthusiasm for stocks, households inject high inflows into the stock markets. This causes informed traders to exploit emerging arbitrage opportunities and therefore the ratio of informed trading increases. These results to market making specialists widen the gap between bids-ask spread and as the spreads keep increasing liquidity decreases, and demand and supply move away from the equilibrium and subsequently a market correction occurs.
Elimination of cross-sectional patterns in stock returns is not an attractive opportunity to an arbitrageur. This is due to the problem that arises when trying to implement a trading strategy that to exploit one of these documented cross sectional patterns. The trading strategies involves buying and selling of the stocks while relying on the relative values rather than the fundamental values. This implies that without the estimate of the firm’s fundamental value then the arbitrageur has no idea to what extent is the security mispriced, if any. According to Stein (2009), given that an arbitrageur solely depends the relative value of a stock they cannot know how many of his peers are simultaneously entering the same trade. Stein and Lundholm (2008) modelled a setting in which investors use only relative values without considering the fundamental value.
 (Brennan, Huh, & Subrahmanyam, 2013)
Their research show that when arbitrage activity is very high in the market, arbitrageurs are able to push security prices beyond the efficient level, thereby causing the opposite of the expected return pattern rather than eliminating the pattern. This implies that it is possible to profit by taking the opposite positions. Nevertheless, Andrew Lo argues that the number and lot size of trades placed can cause a profitable strategy to become unprofitable.
Over the years, the different research that has been documented by investment professionals and financial researchers indicate that cross-sectional market anomalies still exist. However, it is very difficult for investors and arbitrageurs to profit from these strategies. This is mainly due to the introduction of electronic trading, very low transaction fees and increased liquidity levels in the financial markets. Given Africa’s stock exchanges are not as developed as the western world, finance professionals are still researching on the impact of these cross-sectional anomalies on Africa’s exchange and their profitability.
Ken is a Quantitative Trader with experience in investments, quantitative finance, financial modelling and algorithmic trading in Global Investable Markets (GIM). He enjoys using Bayesian Statistics, Time Series and Machine Learning in developing Robust consistent Alphas in Equities Market, FX, ETPs and Derivatives instruments. He enjoys deep dives in understanding High Frequency Trading infrastructures and improving how the African financial markets work. He holds a Bachelor's in Actuarial Science from Strathmore Institute of Mathematical Sciences : An Executive Program in Algorithmic Trading (EPAT) certificate in Algo Trading from QuantInsti : A current MSc student in Financial Engineering at World Quant University.