Does Financial News Predict Stock Returns? New Evidence from Islamic and Non-Islamic Stocks

Highlights

• Financial news predicts most Islamic and conventional stock returns.

• Positive worded news has is relatively more important for stock returns.

• Shock to returns from financial news reverses only in part for some stocks.

• A mean-variance investor investing in Islamic stocks makes more profits.

• Profits are robust to a range of time-series risk factors.

Abstract

The paper extends the time-series financial news data set constructed by Garcia (2013) and uses it to examine whether financial news predicts returns of Islamic stocks differently compared to non-Islamic (conventional) stocks. We find that they do. First, while both positive and negative worded news predict most Islamic and conventional stock returns, positive words have a larger impact on both types of stock returns. Second, shock to returns from financial news reverses only in part for some stocks. Third, for a mean-variance investor, investing in Islamic stocks is relatively more profitable than investing in the corresponding conventional stocks. Fourth, we show that profits are robust to a range of time-series risk factors, namely, market risk, size-based risk, and momentum-induced risk.

Introduction

Islamic stocks are different from conventional stocks. This is clear. The relatively recent literature has been vocal on this subject. Taking Islamic banks, for instance, Beck et al. (2013) show that Islamic banks are less efficient in comparison to conventional banks but outperform conventional banks in terms of asset quality, intermediation, and capitalisation. Using simple performance measures, such as Jensen's alpha, Treynor index,' and Sharpe ratio, Ho et al. (2013) show that Islamic stock indices outperform conventional stock indices over time. Saeed and Izzeldin (2014) show how the relationship between efficiency and default risk for Islamic banks is different compared to conventional banks.

Our goal in this paper is different. We, for the first time in this literature, consider whether financial news predicts Islamic stock returns any differently compared to conventional stock returns. To accomplish this aim, we consider as many as nine Islamic stock indices and the corresponding non-Islamic (or conventional) indices. These indices are: Asia-Pacific, Developed Country, Emerging Markets, European, Global, Canadian, Japanese, the UK, and the USA indices. Our main contribution does not only rest with this test for predictability of stock returns, but includes the construction of the financial news variable also. Our first contribution is that we extend the data set constructed by Garcia (2013). The Garcia financial news daily data set was constructed for the period 1905 to 2005. We add seven years of daily time-series data on financial news to the existing database. Since our sample period in this study is 1996 to 2012, we update the database, using the steps discussed in detail in Garcia (2013), to 2012. The uniqueness of the data set is not only that we have a new and updated daily time-series data on financial news but we have data on positive (good) and negative (bad) news. This allows us to test whether positive news or negative news predict Islamic stock returns better compared to conventional stock returns.¹

Our second contribution is one that goes beyond a statistical analysis of stock return predictability. It is true that whether or not financial news predicts Islamic and conventional stock returns differently is unknown; and, as a result, nothing is known about how investors can successfully trade in Islamic stocks versus conventional stocks by tracking financial news. We, therefore, explore this angle too. Where there is evidence of predictability, return forecasts can be generated based on which investor utility and profits can be computed. This has been shown in the case of mean-variance investors by several studies in the stock return predictability literature; see, inter alia, Westerlund and Narayan, 2012, Westerlund and Narayan, 2014 and Narayan and Sharma (2015). Because we can, at best, generate forecasts based on positive news and negative news, we are able to test exactly which type of news maximises investor utility and profits for Islamic stocks versus conventional stocks.

Our empirical analysis produces several new insights, not only on the role of financial news in predicting stock returns, but also on the relevance of financial news for Islamic versus non-Islamic stocks. Our empirical findings can be summarised as follows. First, we find that of the nine indices, returns are more predictable for Islamic stocks compared to conventional stocks. All-in-all, strong evidence of in-sample predictability is found for six out of nine Islamic stocks and for eight out of nine conventional stocks when positive news is the predictor of returns. However, the magnitude of impact is higher for Islamic stocks relative to conventional stocks. This evidence of in-sample predictability is corroborated by out-of-sample forecasting tests. Similar dominance of Islamic stocks is seen when negative news and the pessimism news factor, which is the difference between the negative and positive news, is used to predict returns. Therefore, we conclude that there is robust evidence that financial news is a relatively strong predictor of Islamic stock returns.

Second, we discover that financial news has an asymmetric effect on stock return predictability. In other words, the pessimism news factor predicts stock returns for more indices than do positive news or negative news. Of the nine indices, the pessimism news factor predicts returns of eight Islamic and conventional stock indices. Also, among the positive and negative news, the magnitude of impact differs by news type.

Third, we discover that, based on the evidence of predictability, a mean-variance investor is able to gain more utility and make more profits if she holds Islamic stocks compared to holding non-Islamic stocks. On average, across all six Islamic stock indices (Asia-Pacific, Canada, Emerging Markets, European, Japan, and the UK) for which positive news predicted returns, tracking positive news allows investors to make profits (utility gain) of 11.61% (9.13%) per annum, while tracking negative news allows investors a profit (utility gain) of 16.86% (13.73%) per annum (averaged over the five predictable indices). This compares to an average profit (utility), across the eight indices for which positive news predicted returns for conventional stocks, of 8.56% (5.70%) and a profit of 12.12% and a utility of 8.72% when tracking negative news (averaged across seven indices). The main message emerging from this finding is that while financial news is important for both Islamic and conventional stocks, it is (a) relatively more important for Islamic stocks, and (b) news has an asymmetric effect on investor profits and utility regardless of the type of stock—that is, negative news offers a mean-variance investor high profits and utility gains compared to positive news.

Fourth, we conclude by testing the robustness of the evidence on profitability. We run time-series regression models where mean-variance investor profits are regressed on commonly known risk factors, such as market risk, size-based risk, and risk induced by the momentum factor. We find that not only abnormal returns are greater than raw returns for both Islamic and conventional stocks but a portfolio of Islamic stocks is relatively more profitable than a portfolio of conventional stocks. On the whole our results on the profitability of Islamic and conventional stocks are insensitive to key risk factors and are, therefore, robust.

It follows that our main contributions and findings contribute to two specific strands of the literature. First our study connects with the traditional literature on predicting stock returns. This literature is dominated by the financial ratio predictors; see for instance Narayan and Bannigidadmath (2015) for a list of studies on this. A related body of literature has begun considering other non-traditional predictors of returns. Driesprong et al. (2008) and Phan et al., 2015a, Phan et al., 2015b, for example, consider oil price as a predictor of returns²; Rapach et al. (2013) use U.S. stock returns as a predictor of OECD returns; Hsu (2009) examines how technology risk factors predict stock market returns; Narayan et al. (2014) use mutual funds to predict returns; Ni et al. (2015) considers the role of investor sentiment; and a recent study by Chava et al. (2015) shows that credit conditions predict stock returns.³ We add to this list the role of financial news in predicting returns. In this regard we join the evidence provided in Garcia (2013) but we go beyond Garcia (2013) by showing the economic significance of the importance of financial news. In our analysis apart from testing for predictability, we propose and evaluate trading strategies. Our paper, as a result, has a profitability story.

Second, our study connects with the literature on asset pricing of Islamic stocks.⁴ In this literature the consensus is that Islamic stocks are profitable. We join this literature by not only showing that Islamic stocks are profitability but providing evidence that Islamic stocks are relatively more profitable than conventional stocks. Equally importantly, we show that this evidence on profitability is robust. In addition, the evidence on profitability we obtain, based on forecasting stock returns and utilizing a mean-variance investor utility function, represents a fresh approach and is different from the approaches undertaken in Narayan et al. (2015a), Ashraf and Mohammad (2014), Bialkowski et al. (2012), and Hoepner et al. (2011).

We organise the balance of the paper as follows. The next section explains our data set, provides a snapshot of our data and offers a discussion on some preliminary findings. Section 3 presents our main results, including a discussion on the empirical framework. The final section concludes with the key findings.