Robert F. Whitelaw

News or Noise? Internet Postings and Stock Prices

The anecdotal evidence is growing that postings in Internet financial forums affect stock prices, either because the postings contain new information or because they represent successful attempts to manipulate stock prices. From an investment perspective, knowing whether this phenomenon is pervasive is important. We examined the relationship between Internet message board activity and abnormal stock returns and trading volume in the period from mid-April 1999 to mid-February 2000. Our study focused on the RagingBull.com discussion forum, an extremely popular site whose format permits the construction of an objective measure of investor opinions. For stocks in the Internet service sector, we found that on days with abnormally high message activity, changes in investor opinion correlated with abnormal industry-adjusted returns. These event days also coincided with abnormally high trading volume, which persisted for a second day. However, we found that message board activity did not predict industry-adjusted returns or abnormal trading volume, which is consistent with market efficiency. Growing anecdotal evidence indicates that postings in Internet financial forums affect stock prices. Clearly, if forums such as message boards contain new information activity, they may help predict stock returns. Even in the absence of any value-relevant information, large numbers of investors may follow the buy and sell recommendations of message board users, thereby inducing deviations in prices from their efficient levels. Furthermore, day traders may recognize the momentum generated by investors who use message boards, thus exaggerating this effect. From an investment perspective, therefore, knowing whether postings affect prices is important. Researchers have begun to explore both the valuation of Internet stocks and the effects of Internet activity on equity valuation, but the focus has been on using accounting data or Web traffic to explain prices. In contrast, we examine directly whether the changes in opinions contained in Internet forums can predict stock returns and/or trading volume. Our study focused on the RagingBull.com discussion forum, an extremely popular site whose format permits the construction of an objective measure of investor opinion. We limited the analysis to stocks in the Internet service sector because they are natural candidates for the most discussion and thus most likely to be affected by information in Internet forums. We present both the event study and the vector autoregression analysis of the data that we carried out. The event study examined abnormal stock returns and trading volume around days with abnormal message board activity in the period from mid-April 1999 to mid-February 2000. An “event day” was defined as a day when the number of message postings exceeded the five-day moving average number of message postings by two standard deviations. The results show that days with strongly positive changes in message board opinions are preceded by a small abnormal increase in stock price. We also found that message board opinion and abnormal returns on the event day are related. We found little evidence, however, that opinion predicts future returns. Trading volume increased significantly on the event day and generally remained high for one day thereafter. The vector autoregression analysis examined whether daily returns, trading volume, the number of messages posted, and opinion can be used to predict these variables one day in the future. Consistent with the results from the event study, we found that it is not possible to predict returns using any of the variables. After controlling for the well-known effect that trading volume is positively related to the previous days trading volume, we found message board activity to have no predictive power for trading activity. Our overall conclusion is that no causal link exists from message board activity to stock returns and volume. In fact, we found that it is market information that influences message board activity rather than the other way around. These results are completely consistent with market efficiency. They may also provide some comfort to investment professionals that the value of their portfolios may not be subject to the notoriously fickle opinions of the Internet community.

Optimal risk management using options

This paper provides an analytical solution to the problem of how an institution might optimally manage the market risk of a given exposure, under the assumption that the institution wishes to minimize its Value at Risk (VaR) using options. The solution specifies the VaR-minimizing level of moneyness of the options as a function of the underlying parameters. We show that the optimal hedge consists of a position in a single option whose strike price is independent of the level of expense the institution is willing to incur for its hedging program. The optimal strike price is increasing in the assets drift, decreasing in its volatility for most reasonable parameter, decreasing in the risk-free interest rate, nonmonotonic in the horizon of the hedge, and increasing in the level of protection desired by the institution (i.e., the percentile of the distribution relevant for the VaR). Finally, we also show that the costs associated with a suboptimal choice of exercise price are economically significant.

Ex Ante Bond Returns and the Liquidity Preference Hypothesis

We provide a formal test of the liquidity preference hypothesis (LPH), that is, the monotonicity of ex ante term premiums, using nonparametric estimates that do not require a structural model for conditional expected returns. Although the point estimates of the term premiums are consistent with previous conclusions in the literature regarding violations of the LPH, the test statistics are generally insignificant, even when powerful conditioning information is used. These results illustrate the importance of correctly accounting for correlations across maturities and of formally testing the inequality restrictions implied by the LPH.

Investigation of a Class of Volatility Estimators

This article examines a class o f volatility estimation models, all ofthem based on a weighted sum ofsquared deviationsjiom the meanfor historical returns. We show how some popular methods, such as RiskMetricsTM, GARCH, and non-parametric density estimation, fa l l into this class. We also conduct a briefempirical comparison ofthese methods. Wefind density estimation and RiskMetricsTMforecasts to be the most accurateforforecasting short-term interest rate Volatility.

Stale Prices and Strategies for Trading Mutual Funds

We demonstrate that an institutional feature of numerous mutual funds—funds managing billions in assets—generates fund net asset values that reflect stale prices. Because investors can trade at these NAVs with limited transaction costs in many cases, obvious trading opportunities exist. These opportunities are especially prevalent in funds that buy Japanese or European equities. Simple, feasible strategies generate Sharpe ratios (excess return divided by standard deviation) that are many times greater than the Sharpe ratio of the underlying fund. We illustrate the potential of the strategy for three Vanguard Group mutual funds. A particular issue to keep in mind is that when the strategies are implemented, the gains from these strategies are matched by offsetting losses incurred by buy-and-hold investors in these funds. In the past few years, the financial press has produced numerous articles about large cash flows into and out of certain mutual funds over short time periods. Most of the funds have had one major identifying characteristic: They invest in international—that is, non-U.S.—assets. We attempted to explain this phenomenon and documented the performance of trading strategies that are consistent with these fund flows. Two key institutional features underlie the trading strategies that lead to the rapid in-and-out trading. First, with the proliferation of mutual funds, a U.S. investor can buy into and exchange out of no-load mutual funds at essentially zero cost. Moreover, the opportunities abound; approximately 700 no-load mutual funds invest in international equities, and a number of them are very large. For example, at least 25 international equity funds have assets under management exceeding

The Real Value of China's Stock Market

1 billion. The second institutional feature is that when U.S. investors buy/sell mutual funds during the day, they do so at the prices prevailing at the close of trading in the United States. Those prices are based on the last transaction prices of the stocks in the fund. For Japanese and other Asian equities, the last transaction could have been at the previous 1:00 a.m. (U.S. Eastern Standard Time), and for many European equities, it could have been 12:00 noon. When these markets are closed, information flow does not cease; information relevant for valuation of the securities traded in the closed markets is still being released. For example, the literature contains considerable evidence that international equity returns are correlated at all times, even when one of the markets is closed. Moreover, the magnitude of the correlations may be quite large. This phenomenon induces large correlations between observed security prices during the U.S. trading day and the next days return on these funds. In some cases, derivatives on international markets trading in the United States provide even more informative signals (than U.S. market returns) about the unobserved movements in the prices of securities in the non-U.S. equity funds. This knowledge can be used to generate considerable excess return in the buying and selling of mutual funds. With no transaction costs and perfect liquidity, an investor can purchase funds at stale prices. In the most extreme case, one can use 1:00 a.m. prices to buy a Japan fund while one has information about the “true” price some 15 hours later at 4:00 p.m. Given these facts, it is perhaps no surprise that we document extraordinarily high excess profits and Sharpe ratios for two categories of investment funds: Pacific/Japan equity funds and international/Europe equity funds. Our sample of funds was chosen for the staleness of their underlying prices, the size of the fund, and the ease of implementing the trading strategy. We studied a strategy of switching between a money market account and the underlying fund in response to signals during U.S. market hours. We also studied the effect of the various trading costs from various types of implementation procedures. Because mutual funds do place some limits on the frequency and amount of exchanges between funds, although the limits are not always enforced, we examined strategies with particularly strong signals. We found for both types of fund that, although the strategy recommended active trading only 5–10 percent of the time, the return, on average, substantially exceeded the return to a buy-and-hold strategy during an ex post very good market for equities. More interesting is the fact that for both types of fund, we could predict the next days movement more than 75 percent of the time. Sharpe ratios generally ranged between 5 and 10 on the days the investor was in the market. The range of Sharpe ratios depended on whether the strategy included hedging of equity price movements during non-U.S. trading hours. To illustrate in a detailed manner the mechanics and results of the trading strategy, we provide a case study using three mutual funds from the Vanguard family of funds. This analysis is of special interest to academic readers because these funds are available through the retirement plans of numerous educational institutions and can be easily traded on the Internet or over the phone.

A New Strategy for Dynamically Hedging Mortgage-Backed Securities

China is the world’s largest investor and greatest contributor to global economic growth by wide margins, and will remain so for many years. The efficiency of its financial system in allocating capital to investment will be important to sustain this growth. This paper shows that China’s stock market has a crucial role to play. Since the reforms of the last decade, China’s stock market has become as informative about future corporate profits as in the US. Moreover, though it is a segmented market, Chinese investors price risk and other stock characteristics remarkably like investors in other large economies. They pay up for large stocks, growth stocks, and long shots, and they discount for illiquidity and market risk. China’s stock market no longer deserves its reputation as a casino. In addition, the trend of stock price informativeness over the last two decades is highly correlated with that of corporate investment efficiency. China’s stock market appears to be aggregating diffuse information and generating useful signals for managers. On the buy side, because of its low correlation with other stock markets and high average returns, China’s stock market offers high alpha to diversified global investors who can access it. Yet this high alpha amounts to an inflated cost of equity capital, constraining the investment of China’s smaller, more profitable enterprises. Further reforms that open this market to global investors and improve stock price informativeness will be important to increase China’s investment efficiency and fuel its continued economic growth. Finally, we interpret the stock market’s recent gyrations through the lens of this research, arguing that its post-crisis lag was a rational downward adjustment to competition from the rapidly expanding shadow banking sector, and its enormous rally last year is a cheer for the roll out of deposit insurance and other Third Plenum reforms. More than ever, China’s stock market is a crucial counterpart to its extraordinary, relationship-driven, but opaque banking sector. China’s stock market may now be the world’s most important crystal ball.

Hybrid Tail Risk and Expected Stock Returns: When Does the Tail Wag the Dog?

This paper develops a new strategy for dynamically hedging mortgage-backed securities (MBSs). The approach involves estimating the joint distribution of returns on MBSs and T-note futures, conditional on current economic conditions. We show that our approach has a simple intuitive interpretation of forming a hedge ratio by differentially weighting past pairs of MBS and T-note futures returns. An out-of-sample hedging exercise is performed for 8%, 9% and 10% GNMAs over the 1990-1994 period for weekly and monthly return horizons. The dynamic approach is very successful at hedging out the interest rate risk inherent in all of the GNMAs. For example, in hedging weekly returns on 10% GNMAs, our dynamic method reduces the volatility of the GNMA return from 41 to 24 basis points, whereas a static method manages only 29 basis points of residual volatility. Moreover, only 1 basis point of the volatility of the dynamically hedged return can be attributed to risk associated with U.S. Treasuries, which is in contrast to 14 basis points of interest rate risk in the statically hedged return.

New Evidence on the Forward Premium Puzzle

We introduce a new, hybrid measure of stock return tail covariance risk, motivated by the under-diversified portfolio holdings of individual investors, and investigate its cross-sectional predictive power. Our key innovation is that this covariance is measured across the left tail states of the individual stock return distribution, not across those of the market return as in standard systematic risk measures. We document a positive and significant relation between hybrid tail covariance risk (H-TCR) and expected stock returns, with an annualized premium of 9%, in contrast to the insignificant or negative results for purely stock-specific or systematic tail risk measures.

On the Fundamental Relation between Equity Returns and Interest Rates

The forward premium anomaly (exchange rate changes are negatively related to interest rate differentials) is one of the most robust puzzles in financial economics. We recast the underlying parity relation in terms of lagged forward interest rate differentials, documenting a reversal of the anomalous sign on the coefficient in the traditional specification. We show that this novel evidence is consistent with recent empirical models of exchange rates that imply exchange rate changes depend on two key variables: the interest rate differential and the magnitude of the deviation of the current exchange rate from that implied by purchasing power parity.