Title | Short interest and aggregate stock returns handout |
---|---|
Course | Investment Management (2) |
Institution | 逢甲大學 |
Pages | 12 |
File Size | 204.8 KB |
File Type | |
Total Downloads | 61 |
Total Views | 145 |
a handout for oral presentation....
•Short interest and aggregate stock returns • David E. Rapach, Matthew C. Ringgenberg, Guofu Zhou • (2016) • Outline • 1. Research issue • 2. Main results • 3. Data • 4. Methodology • 5. Interpretation of results • 6. Conclusion • 1. Reasearch Issue • Short interest, aggregated across securities, is arguably the strongest predictor of the equity risk premium identified to date • 1.1. Research context
• The equity market risk premium impacts many fundamental areas of finance, from portfolio theory to capital • A voluminous literature attempts to predict changes in future aggregate excess stock returns • 1.2. Research contribution • Short interest • The first: • Using a modern time-series approach, and show that aggregate short interest is arguably the strongest known predictor of the equity risk premium • Explicitly examine the relation between aggregate short interest and aggregate stock returns • Examine a long time series of short interest data (January of 1973 - December 2014)
• Extend the literature: evidence that short sellers (informed traders) are also skilled at processing information about macroeconomic conditions (not only firmspecific fundamental conditions)
• 2. Main results • Short interest • When aggregated across firms and appropriately detrended, is a statistically and economically significant predictor of future market excess returns over 1973:01–2014:12 sample period • Outperforms a host of popular return predictors from the literature in both insample and out-of-sample tests • Generates substantial utility gains and Sharpe ratios that exceed those provided by popular predictors
• Ability of short interest to predict future market returns stems
predominantly from a cash flow channel • Short sellers are informed traders • who are able to anticipate changes in future aggregate cash flows and associated changes in future market returns
• 3. Data 3.1. Short interest • Short interest data: Compustat • the equity risk premium • popular predictor variables from the existing literature
• Short interest • • • •
Publicity and frequency Normalize (CRSP) Clean up (exclude and drop) Result: over two million observations at the firm-month level for the 42-year
period from January 1973 through December 2014. • Cover a variety of asset classes: common equities, American Depositary Receipts (ADRs), Exchange Traded Funds (ETFs), and Real Estate Investment Trusts (REITs) • Each month: calculate aggregate short interest as the equal-weighted mean of all asset-level short interest data (EWSI)
• 3. Data • Compare: • predictive ability of aggregate short interest v.s. • that of 14 monthly predictor variables
• Consistent with the existing literature on predicting aggregate returns • predicting the excess return on a valueweighted market portfolio
• market excess return = log(rS&P 500 index) log(rone-month Treasury bill) • 3. Data 3.2. Sample properties
• The increase in short interest • development of the equity lending market • growth of the hedge fund industry
• uˆt : (has a mean of zero) • standardize the series to have a standard deviation of one.
• The standardized series: SII • a measure of market pessimism based on short interest data
• ll
• 3. Data 3.3. Relation to other predictors
• The SII measure appears to contain substantially different information from many of the stock return predictors used in the existing literature • 4.1. Predictive regression analysis 4.1.1. In-sample tests •predictive regression model (standard framework for analyzing aggregate stock return predictability)
•rt is the S&P 500 log excess return for month t •one-sided alternative hypothesis (as theory often suggests the sign of β under predictability) •The last row: • : the first three principal components •to test the predictive power of SII after controlling for the entire group of popular predictor variables taken together
• 4.1. Predictive regression analysis 4.1.2. Alternative detrending • “stochastic detrending” based on a five-year window, where SII for month t
is the difference between log(EWSI) for month t minus the average of log(EWSI) from month t − 59 to month t.
• 4.1. Predictive regression analysis 4.1.3. Out-of-sample tests • Out-of-sample R2 • predictive regression forecast • natural benchmark: the constant expected excess return model, i.e. β = 0 (returns are not predictable, as in the canonical random walk with drift model for the log of stock prices) • the out-of-sample R2 statistic (R2OS)
• Forecast encompassing tests •
: predictive regression forecast based on one of the popular predictors (SII)
• 0 ≤ λ ≤ 1: • λ = 0: predictive regression forecast based on the popular predictor already encompassed the information in the forecast based on SII • λ > 0: SII provides information that is useful for forecasting excess returns beyond the information already contained in the popular predictors • λ = 1: the forecast based on SII already encompassed the information in the predictive regression forecast based on the popular predictors
• 4.2. Asset allocation • measure the economic value of SII’s predictive ability • mean-variance investor allocates between equities and risk-free bills using a predictive regression forecast of excess stock returns • Rebalance every end of month t • Realizes an average utility or certainty equivalent return (CER): • R-p and σ2p are the mean and variance, respectively, of the portfolio return over the forecast evaluation period • The CER is the risk-free rate of return that an investor would be willing to accept in lieu of holding the risky portfolio
• Compare portfolio performance independently of relative risk aversion
• Annualized Sharpe ratios for the entire 1990:01– 2014:12 forecast evaluation period • The 14 predictors from the literature rarely outperform the prevailing mean in terms of the Sharpe ratio
• SII produces Sharpe ratios • approximately 1.5 to two times larger than those of the prevailing mean • always greater than those for the popular predictors (as well as the buy-and-hold strategy)
• 5. Interpretation of results • Short sellers as informed traders • possess an information advantage regarding future aggregate cash flows • are also skilled at processing aggregate information • with respect to both the idiosyncratic and systematic determinants of equity valuations
• Time-varying equilibrium aggregate risk premium
• Fluctuations in SII relate to time variation in the equilibrium aggregate risk premium (thought SII is largely orthogonal to popular predictor)
• Yet the predictive power of SII hasn’t declined • Short selling has become easier and more common over time • Under the time-varying equilibrium aggregate risk premium, it is not necessarily expect SII’s predictive ability to decline over time • Momentum anomaly persists, despite publication of the result (investors often learn about anomalies after studies are published) • Long-short portfolios based on strategies that are more costly to arbitrage experience smaller reductions in returns after publication
• 6. Conclusion • The information contained in the SIIbased forecast dominates the information found in forecasts based on popular predictors • SII also generates substantial utility gains for a mean-variance investor with a relative risk aversion coefficient of three, and the gains are especially large during the recent Global Financial Crisis • The information content of short selling is more important economically than previously believed • Short sellers are also skilled at processing information about macroeconomic conditions...