Sell‐Side Analyst Research and Stock Comovement

We document that a stock's price around a recommendation or forecast covaries with prices of other stocks the issuing analyst covers. The effect of shared analyst coverage on stock price comovement extends beyond analyst activity days. A stock's daily returns covary with the returns of other stocks with which it shares analyst coverage. These links between stock price comovement and shared analyst coverage are consistent with the coverage‐specific information we find in earnings forecasts; analysts who cover both stocks in a pair expect future earnings of the stocks to be more highly correlated than do analysts who cover only one stock from the pair. Collectively, our evidence indicates that analyst research produces coverage‐specific spillovers that raise price comovement among stocks that share analyst coverage. The strength of these spillovers is comparable to spillovers from broad industry and market information in analyst research.     

Learning and predictability via technical analysis: Evidence from bitcoin and stocks with hard‐to‐value fundamentals

What predicts returns on assets with “hard‐to‐value” fundamentals such as Bitcoin and stocks in new industries? We are the first to propose an equilibrium model that shows how technical analysis can arise endogenously via rational learning, providing a theoretical foundation for using technical analysis in practice. We document that ratios of prices to their moving averages forecast daily Bitcoin returns in and out of sample. Trading strategies based on these ratios generate an economically significant alpha and Sharpe ratio gains relative to a buy‐and‐hold position. Similar results hold for small‐cap, young‐firm, and low analyst‐coverage stocks as well as NASDAQ stocks during the dotcom era.     

The Piotroski F‐score: evidence from Australia

A market‐neutral strategy that is long [short] stocks with a high [low] Piotroski F‐score generates an index‐weighted 0.8 percent pm on S&P/ASX 200 stocks and 1.4 percent pm on smaller stocks. Equal‐weighted returns are higher and in all cases returns are statistically significant. However, the Carhart model alphas are not statistically significant except in the case of equal‐weighted small cap portfolios. For such portfolios, however, most of the alpha comes from the short side and most institutional investors would find them uninvestable due to capacity constraints. A range of tests indicate that analyst neglect does not explain the F‐score premium.     

TRADING COSTS OF NON‐U.S. STOCKS ON THE NEW YORK STOCK EXCHANGE: THE EFFECT OF INSTITUTIONAL OWNERSHIP,ANALYST FOLLOWING,AND MARKET REGULATION

We investigate the differences in market microstructure between U.S. and non‐U.S. stocks cross‐listed on the New York Stock Exchange using a sample of 316 pairs of matched stocks. We find that non‐U.S. stocks have wider spreads and larger adverse‐selection costs than U.S. stocks even after controlling for macro‐level institutional differences. Regression analysis shows that spreads and adverse‐selection costs are negatively correlated with institutional ownership and analyst followings. Thus, the higher spreads and adverse‐selection costs for non‐U.S. stocks can be partly explained by the lower institutional ownership and analyst following of non‐U.S. stocks. In addition, we find that although the spreads and adverse‐selection costs for non‐U.S. stocks are significantly higher before the implementation of Regulation Fair Disclosure (FD), the differences become even greater after Regulation FD, suggesting that Regulation FD has improved the information environment for U.S. stocks.     

Analyst Disagreement,Mispricing, and Liquidity*

This paper documents a close link between mispricing and liquidity by investigating stocks with high analyst disagreement. Previous research finds that these stocks tend to be overpriced, but that prices correct downwards as uncertainty about earnings is resolved. Our analysis suggests that one reason mispricing has persisted through the years is that analyst disagreement coincides with high trading costs. We also show that in the cross‐section, the less liquid stocks tend to be more severely overpriced. Additionally, increases in aggregate market liquidity accelerate the convergence of prices to fundamentals. As a result, returns of the initially overpriced stocks are negatively correlated with the time series of innovations in aggregate market liquidity.     

A New Measure of Equity and Cash Flow Duration: The Duration‐Based Explanation of the Value Premium Revisited

This article reexamines the duration‐based explanation of the value premium using novel estimates of the firms' equity and cash flow durations based on analyst forecasts. We show that the value premium can be explained by cross‐sectional differences in the shares' equity durations, but not by their cash flow durations. Different from the duration‐based explanation of the value premium that explains the value premium with cross‐sectional differences in the firm's cash flow timing, we find that short‐horizon stocks have lower (expected) returns than long‐horizon stocks. This result is consistent with an upward‐sloping equity yield curve.     

An adaptive hierarchical fuzzy logic system for modelling of financial systems

In this paper an intelligent hierarchical fuzzy logic system using genetic algorithms for the prediction and modelling of interest rates in Australia is developed. The proposed system uses a hierarchical fuzzy logic system in which a genetic algorithm is used as a training method for learning the fuzzy rules knowledge bases that are used for prediction of interest rates in Australia. A hierarchical fuzzy logic system is developed to model and predict three‐month (quarterly) interest rate ?uctuations. The system is further trained to model and predict interest rates for six‐month and one‐year periods. The proposed system is developed with ?rst two, three, then four and ?nally ?ve hierarchical knowledge bases to model and predict interest rates. A novel architecture called a feed forward fuzzy logic system using fuzzy logic and genetic algorithms is also developed to predict interest rates. A back‐propagation hierarchical neural network system is also developed to predict interest rates for three‐month, six‐month and one‐year periods. The results obtained from these two systems are then compared with the hierarchical fuzzy logic system results and conclusions are drown on the accuracy of all systems for prediction of interest rates in Australia. Copyright © 2004 John Wiley & Sons, Ltd.     

Cross‐firm return predictability and accounting quality

We test the hypothesis that if poor accounting quality (AQ) is associated with poor investor understanding of firms’ revenue and cost structures, then poor AQ stocks likely respond more slowly than good AQ stocks to new non‐idiosyncratic information that affects both sets of firms. Consistent with this, results indicate that stock returns of good AQ firms significantly positively predict one‐month‐ahead stock returns to industry‐ and size‐matched poor AQ firms. In testing a delayed‐information‐processing mechanism behind the cross‐firm return predictability, we find that: (i) analyst earnings forecast revisions (FR) mimic the return patterns, as FR of good AQ firms significantly positively predict one‐month‐ahead FR of matched poor AQ firms; (ii) cross‐firm return predictability is concentrated in months with substantial news arrival, including months with Federal Open Market Committee (FOMC) rate announcements, but not in no‐news months; (iii) cross‐firm return predictability is stronger when the good AQ predictor firms have a richer information environment than poor AQ firms as proxied by analyst following, institutional ownership, and the presence of a Big 4 auditor. Collectively, the results uncover a new relation between accounting quality and stock return dynamics.     

What Drives Sell‐Side Analyst Compensation at High‐Status Investment Banks?

We use proprietary data from a major investment bank to investigate factors associated with analysts’ annual compensation. We find compensation to be positively related to “All‐Star” recognition, investment‐banking contributions, the size of analysts’ portfolios, and whether an analyst is identified as a top stock picker by the Wall Street Journal. We find no evidence that compensation is related to earnings forecast accuracy. But consistent with prior studies, we find analyst turnover to be related to forecast accuracy, suggesting that analyst forecasting incentives are primarily termination based. Additional analyses indicate that “All‐Star” recognition proxies for buy‐side client votes on analyst research quality used to allocate commissions across banks and analysts. Taken as a whole, our evidence is consistent with analyst compensation being designed to reward actions that increase brokerage and investment‐banking revenues. To assess the generality of our findings, we test the same relations using compensation data from a second high‐status bank and obtain similar results.     

Comparing the Post–Earnings Announcement Drift for Surprises Calculated from Analyst and Time Series Forecasts

Post–earnings announcement drift is the tendency for a stock's cumulative abnormal returns to drift in the direction of an earnings surprise for several weeks following an earnings announcement. We show that the drift is significantly larger when defining the earnings surprise using analysts' forecasts and actual earnings from I/B/E/S than when using a time series model based on Compustat earnings data. Neither Compustat's policy of restating earnings nor the inclusion of “special items” in reported earnings contribute significantly to the disparity in drift magnitudes. Rather, our results suggest that this disparity is attributable to differences between analyst forecasts and those of time‐series models—or at least to factors correlated with these differences. Further, we document that analyst forecasts lead to return patterns around future earnings announcements that differ from those observed when using time‐series models, suggesting that the two types of surprises may capture somewhat different forms of mispricing.     

Market Segmentation and Cross‐predictability of Returns

We present evidence supporting the hypothesis that due to investor specialization and market segmentation, value‐relevant information diffuses gradually in financial markets. Using the stock market as our setting, we find that (i) stocks that are in economically related supplier and customer industries cross‐predict each other's returns, (ii) the magnitude of return cross‐predictability declines with the number of informed investors in the market as proxied by the level of analyst coverage and institutional ownership, and (iii) changes in the stock holdings of institutional investors mirror the model trading behavior of informed investors.     

The Extreme Future Stock Returns Following I/B/E/S Earnings Surprises

We investigate the stock returns subsequent to quarterly earnings surprises, where the benchmark for an earnings surprise is the consensus analyst forecast. By defining the surprise relative to an analyst forecast rather than a time‐series model of expected earnings, we document returns subsequent to earnings announcements that are much larger, persist for much longer, and are more heavily concentrated in the long portion of the hedge portfolio than shown in previous studies. We show that our results hold after controlling for risk and previously documented anomalies, and are positive for every quarter between 1988 and 2000. Finally, we explore the financial results and information environment of firms with extreme earnings surprises and find that they tend to be “neglected” stocks with relatively high book‐to‐market ratios, low analyst coverage, and high analyst forecast dispersion. In the three subsequent years, firms with extreme positive earnings surprises tend to have persistent earnings surprises in the same direction, strong growth in cash flows and earnings, and large increases in analyst coverage, relative to firms with extreme negative earnings surprises. We also show that the returns to the earnings surprise strategy are highest in the quartile of firms where transaction costs are highest and institutional investor interest is lowest, consistent with the idea that market inefficiencies are more prevalent when frictions make it difficult for large, sophisticated investors to exploit the inefficiencies.     

Determinants of Investor Demand for Cross‐Listed Firms

By focusing on the decisions of investors to invest in cross‐listed stocks, this paper presents new evidence on why we observe striking differences in the percentage of trade in foreign markets for cross‐listed stocks. With a large sample of Toronto Stock Exchange (TSX) stocks cross‐listed in the U.S. and Canada, we document the effect of investor recognition and risk characteristics on the distribution of trading volume. Firms that are more visible to American investors are traded more heavily in the U.S. At the same time, firms that offer diverse risk characteristics are attractive to Americans. While investors understand the benefits of international diversification, as they are attracted to stocks that are different (e.g., the stock of small firms with few assets in the U.S.), they also seek stocks that provide them with high returns.     

Risk,return and mean‐variance efficiency of Islamic and non‐Islamic stocks: evidence from a unique Malaysian data set

We find that Islamic stocks are more mean–variance efficient than non‐Islamic stocks and the market because they reduce risk of the same level of returns. We combine a unique Malaysian data set of individual Islamic stocks (as opposed to aggregate stock indices) since 1997 with a new method where we apply Islamic business activity and financial ratio screens to the universe of Malaysian stocks. Both data sets show that Islamic stocks have an annualised standard deviation that is on average 3.43–3.78 percent points lower compared to non‐Islamic stocks. This lower variance of Islamic stocks is exclusively driven by financial ratio screens.     

Active investment manager portfolios and preferences for stock characteristics

The present study investigates the stock characteristic preferences of institutional Australian equity managers. In aggregate we find that active managers exhibit preferences for stocks exhibiting high‐price variance, large market capitalization, low transaction costs, value‐oriented characteristics, greater levels of analyst coverage and lower variability in analyst earnings forecasts. We observe stronger preferences for higher volatility, value stocks and wider analyst coverage among smaller stocks. We also find that smaller investment managers prefer securities with higher market capitalization and analyst coverage (including low variation in the forecasts of these analysts). We also document that industry effects play an important role in portfolio construction.     

The Components of the Bid‐Ask Spread: the Case of the Athens Stock Exchange

We analyse the components of the bid‐ask spread in the Athens Stock Exchange (ASE), which was recently characterised as a developed market. For large and medium capitalisation stocks, we estimate the adverse selection and the order handling component of the spreads as well as the probability of a trade continuation on the same side of either the bid or the ask price, using the Madhavan et al. (1997) model. We extend it by incorporating the traded volume and we find that the adverse selection component exhibits U‐shape patterns, while the cost component pattern depends on the stock price. For high priced stocks, the usual U‐shape applies, while for low‐priced ones, it is an increasing function of time, mainly due to the order handling spread component. Furthermore, the expected price change and the liquidity adjustment to Value‐at‐Risk that is needed are higher in the low capitalisation stocks, while the most liquid stocks are the high priced ones. Moreover, by estimating the Madhavan et al. (1997) model for two distinct periods we explain why there are differences in the components of the bid‐ask spread.     

NEURO‐GENETIC PREDICTIONS OF CURRENCY CRISES

We create a neuro‐genetic (NG) model for predicting currency crises by using a genetic algorithm for specifying (1) the combination of inputs, (2) the network configuration and (3) the training parameters for a back‐propagation artificial neural network (ANN). The performance of the NG model is evaluated by comparing it with standalone probit and ANN models in terms of utility for a policy decision‐maker. We show that the NG model provides better in‐sample and out‐of‐sample performance, as well as provides an automatic and more objective calibration of a predictive ANN model. We show that using a genetic algorithm for finding an optimal model specification for an ANN is not only less laborious for the analyst, but also more accurate in terms of classifying in‐sample and predicting out‐of‐sample crises. For a sufficiently parsimonious, but still nonlinear, model for generalized processing of out‐of‐sample data, the creation and evaluation of models is performed objectively using only in‐sample information as well as an early stopping procedure. Copyright © 2011 John Wiley & Sons, Ltd.     

Economic value of analyst recommendations in Australia: an application of the Black–Litterman asset allocation model

This study empirically examines the investment value of security analyst recommendations on constituent stocks of the S&P/ASX 50 index. We find that stocks with favourable (unfavourable) recommendations on average outperformed (underperformed) the benchmark index. An investment strategy using the Black–Litterman asset allocation model that incorporates consensus analyst recommendations, in conjunction with daily rebalancing, outperforms the market in terms of return and risk‐adjusted performance measures. The investment strategy involves high levels of trading, and no significant abnormal returns are achieved after transaction costs. Less frequent rebalancing, under most situations, causes a decrease in both performance and turnover. Filtering of dated recommendations causes an increase in turnover, while having mixed effects on investment returns.     

Words versus Deeds: Evidence from Post‐Call Manager Trades

We examine the impact of conference call tones on the direction and magnitude of subsequent manager trades. Our univariate results show that corporate insiders buy company shares following negative‐tone conference calls and sell shares following positive‐tone conference calls. This inverse call tone–trading pattern holds for both managers’ introductory sessions and subsequent question‐and‐answer (Q&A) sessions. Our multivariate results confirm the univariate call tone–trading patterns and show that contrarian manager trades are mostly driven by managerial selling activity. In contrast to the consistent and strong evidence of managers trading in the opposite direction of their call tones, we find no evidence of managers trading in the same direction of their call tones. We also examine the impact of analyst Q&A challenges on post‐call manager trades. Our findings suggest that managers learn from analyst feedback and adjust their post‐call trades accordingly.     

Pattern recognition and financial time‐series

This paper investigates financial time‐series from the perspective of a practitioner in artificial intelligence methods and pattern recognition. It presents results from statistical experiments which suggest that financial markets operate with a measure of inefficiency and predictability. However, identifying the nature of any regularities and patterns presents a difficult challenge to the artificial intelligence community, in that established techniques make assumptions about the underlying process that mostly prove to be invalid for this class of data. Copyright © 2007 John Wiley & Sons, Ltd.