Chapter 1 Introduction
1.0 Introduction
Equity market return predictability is one of the dominant themes in the finance literature given the fact that investors in the equity market are interested in predicting future equity returns and would like to take every opportunity to increase their wealth (Kandel and Stambaugh, 1996). Its concept is stunningly simple; any systematic variable that affects the economy’s pricing operator (i.e., interest rates) or that influences dividends must be related to equity market returns (Chen, Roll, and Ross, 1986). However, testing for the presence of return predictability is not as easy as it would initially seem.
In the academic literature on equity market return predictability, the prevalent view until the 1970s was that equity prices are closely described by a random walk, suggesting that no economically exploitable predictable patterns exist and any attempt to obtain performance superior to that of the overall “market portfolio” by picking and choosing among securities would fail (e.g., Fama, 1965; Van Horne and Parker, 1967; Jensen and Benington, 1970). More recent empirical work, however, reports evidence that equity returns are to some extent predictable, either from their own past or from other publicly available information, such as inflation, money growth and interest rates (e.g., Flannery and Protopapadakis, 2002; Rapach, Wohar, and Rangvid, 2005; Humpe and Macmillan, 2009).
Doubt is cast on these statistical investigations, defined as tests based on statistical criteria including (out-of-sample) mean squared forecast error and R2, of many of the predictable patterns that have been suggested as they do not necessarily capture the economic importance with which investors regard return predictability. This leads to studies beginning to explicitly address the economic value of this predictability (e.g., Pesaran and Timmermann, 1995; 2000; Marquering and Verbeek, 2004), which measures whether there are tangible economic gains from using forecasts in an asset allocation context.
Furthermore, the statistical evidence of equity return predictability is also questioned as some of the above-mentioned results are not robust across countries and time periods (e.g., Bossaerts and Hillion, 1999; Ang and Bekaert, 2001). Flannery and Protopapadakis (2002) argue that this might be due to the assumption of a stable prediction model whereas in fact the parameter estimates vary through time. There are a number of factors that might lead to model instability (Paye and Timmermann, 2006), which include major changes in market sentiment or regime switches in monetary policies (for example, from money supply targeting to inflation targeting). Institutional changes or large macroeconomic shocks that give rise to changes in economic growth or affect risk premia may also cause a break in financial return prediction models. Empirical studies, such as Bossaerts and Hillion (1999), Goyal and Welch (2003), and Ang and Bekaert (2007), all find that return predictability can be very different across partitioned samples. Therefore, this highlights the importance of accounting for possible model instability because it introduces new sources of risk and because it fundamentally affects the extent to which returns are predictable.
While not enough attention has been paid to addressing the issues of the economic value of return predictability and forecasting under model instability until quite recently, the consideration of these two areas is very important as it might provide new evidence on the investment opportunities of interest to investors in equity markets. This book examines the predictability of equity market performance (both equity market downturns and equity market returns) while accounting for model instability, based on three different equity markets in China, i.e. the Shanghai A-share index (SHA), the Shanghai Composite index (SHC) and the Shenzhen Composite index (SZC). Using Chinese equity market data provides an exceptional opportunity for such an examination owing to the uncertainty about government policies and volatility in market sentiment (Su, 2003). In addition to statistical investigations, this book also examines whether the predictability of equity market performance in China could have been historically exploited by investors to earn profits in excess of a buy-and-hold strategy in the market index, where trading is based on information that is genuinely available at each time an investment decision is made, taking into account, for example, transaction costs. Furthermore, it also considers the potential issues of model instability, such as the estimation of break dates, the determination of the presence and the number of breaks, and the statistical analysis of the resulting parameter estimates. Such an examination provides an insight into the behavior of different domestic equity markets and trading strategies in an economy with a unique investment environment.
The next section moves on to describing some specific/technical details about the two areas of return predictability mentioned above, which further motives the empirical work undertaken later in this book.