Loading...
Please wait, while we are loading the content...
Similar Documents
Who are the sentiment traders ? Evidence from the cross-section of stock returns and demand March 14 , 2014 Luke
| Content Provider | Semantic Scholar |
|---|---|
| Author | Sias, DeVault Richard Starks, Laura |
| Copyright Year | 2014 |
| Abstract | Recent work suggests that sentiment traders shift from less volatile to speculative stocks when sentiment increases. Given that the market clearing condition requires a buyer for every seller, we exploit these cross-sectional patterns and changes in share ownership to test whether investor sentiment metrics capture individual investors’ demand shocks. In contrast to theoretical assumptions and common perceptions, we find no evidence that individual investors’ trading is responsible for sentiment induced demand shocks and mispricing. Our results suggest that either these metrics do not capture “investor sentiment” or that institutional, rather than individual, investors are the sentiment traders whose demand shocks drive prices from value. DeVault is from the Department of Finance, Eller College of Management, University of Arizona, Tucson, Arizona, 85721; ldevault@email.arizona.edu. Sias is from Department of Finance, Eller College of Management, University of Arizona, Tucson, Arizona 85721, (520) 621-3462, sias@eller.arizona.edu. Starks is from the Department of Finance, McCombs School of Business, University of Texas at Austin, Austin, Texas 78712, (512) 471-5899, lstarks@mail.utexas.edu. We thank seminar participants at the University of Arizona, the 2013 European Finance Association Meetings, Kelvin Law, Charles Lee, and Jeff Wurgler for their helpful comments. We thank Jeff Wurgler and Ken French for providing data. Copyright © 2013 by the authors. 1 Who are the sentiment traders? Evidence from the cross-section of stock returns and demand “There is simply no reason to believe that institutional investors are less subject to social influences on opinion than other investors, and there are substantial grounds for thinking that they may be even more so.” (Friedman, 1984) A burgeoning theoretical and empirical literature posits that demand shocks by uninformed “sentiment traders” impact security prices, which has important implications for both asset pricing and corporate finance. This research commonly assumes irrational individual investors are the source of sentiment-based demand shocks captured by sentiment metrics. In this paper, we examine this assumption by building upon the recent insight that investor sentiment has both crosssectional and time-series implications. Specifically, Baker and Wurgler (henceforth, BW) (2006, 2007) propose that securities with “highly subjective valuations” are more susceptible to the vagaries of sentiment. Consistent with their hypothesis, high volatility stocks display a strong positive relation between the BW metric for changes in investor sentiment and contemporaneous stock returns while low volatility stock returns move inversely with contemporaneous changes in sentiment. That is, “sentiment betas” are positive for speculative stocks and negative for safe stocks. The authors also find speculative stocks tend to underperform safe stocks following high sentiment levels, but outperform safe stocks following low sentiment levels. They conclude that the combined results are consistent with the hypothesis that sentiment traders’ demand shocks impact prices and result in 1 See, for example, the recent (May 2012) special issue of the Journal of Financial Economics devoted to investor sentiment. 2 See, for example, Shiller (2000), De Long, Shleifer, Summers, and Waldmann (1990a,1990b), Lee, Shleifer, and Thaler (1991), Nagel (2005), Barberis and Xiong (2012), and Stambaugh, Yu, and Yuan (2012a). Moreover, Baker and Wurgler (2007, p. 136) note, “The inexperienced retail or individual investor is more likely than the professional to be subject to sentiment.” A few early theoretical models, however, suggest institutional investors may engage in noise trading because clients cannot fully distinguish noise trading from informed trading (e.g., Allen and Gorton (1993), Dow and Gorton (1997), and Trueman (1988)). The introductory quote (Friedman (1984)) is the sole exception we are aware of that posits institutional investors are more susceptible to sentiment than individual investors. 3 BW (2006, 2007) propose that greater limits to arbitrage for speculative stocks (relative to safe stocks) also contributes to speculative stocks’ larger sentiment betas. We discuss this point in greater detail below. 2 pushing speculative stocks’ valuations too high relative to the valuations of safe stocks when sentiment is high (and too low when sentiment is low). The investor sentiment hypothesis is a demand shock story—it requires changes in demand (i.e., in the words of BW (2007, p. 131), “sentiment-based demand shocks”) and finite demand and supply elasticities. That is, demand shocks imply net buying or selling by sentiment traders which results in changes in their ownership levels. Moreover, because the market clearing condition requires a buyer for every seller, sentiment traders’ net demand shocks must be offset by supply from traders who are less subject to changes in sentiment. For ease of exposition, we denote these latter traders as “liquidity” traders. Of course, at least some of the liquidity traders’ supply may be motivated by fundamental trading, e.g., selling overvalued speculative stocks to sentiment traders when sentiment increases. It is these two insights from the sentiment literature—speculative stocks are more susceptible than safe stocks to the vagaries of sentiment and sentiment traders’ demand shocks must be offset by liquidity traders’ supply—that drive our primary hypothesis—changes in sentiment will be positively related to changes in sentiment traders’ demand for speculative stocks (and inversely related to their demand shocks for safe stocks). An increase in sentiment, for example, causes sentiment traders to purchase risky stocks and sell safe stocks (i.e., their buying and selling—their demand shocks—are the drivers of the mispricing in the sentiment literature). Despite the near universal assumption that, as a group, individual investors are more prone to sentiment induced frenzies while institutions are smart-money rational investors, we demonstrate that an increase in sentiment is associated with an increase in aggregate institutional demand for speculative stocks and a decrease in their aggregate demand for safe stocks. Equivalently, individual 4 In most sentiment models, market frictions (e.g., short sale restrictions, transaction costs, capital constraints, or noise trader risk) keep rational speculators from immediately correcting mispricing (see, for example, Miller (1977), DeLong, Shleifer, Summers, and Waldmann (1990a), and Shleifer and Vishney (1997)). 3 investors (as a group) buy safe stocks and sell risky stocks when sentiment increases. Thus, our key result suggests that either these metrics do not capture investor sentiment or that institutional investors (in aggregate), rather than individual investors, are the sentiment traders that drive sentiment induced mispricing. Although we primarily focus on institutional and individual investors’ demand shocks, we also investigate the relation between sentiment levels and institutional and individual investors’ ownership levels of speculative and safe stocks. Further inconsistent with the hypothesis that sentiment metrics capture irrational individual investor demand, institutional investors’ ownership levels (i.e., the fraction of shares held by institutions) of speculative stocks (relative to their ownership levels of safe stocks) are higher when sentiment levels are higher. Equivalently, high sentiment levels are associated with (relatively) lower individual investor ownership levels of speculative stocks. We conduct a number of robustness tests that continue to support the hypothesis that sentiment metrics capture innovations in institutional, rather than individual investors’ (direct), demand. First, although we focus on the BW sentiment metric because it is the dominant measure in recent research on sentiment (e.g., Antoniou, Doukas, and Subrahmanyam (2013), Rosch, Subrahmanyam, and van Dijk (2013), Moskowitz, Ooi, and Pedersen (2012), Karolyi, Lee, and van Dijk (2012), Ramadorai (2012), Hribar and McInnis (2012), McLean and Zhao (2012), Novy-Marx (2012), and Stambaugh, Yu, and Yuan (2012a, 2012b), Baker, Wurgler, and Yuan (2012), and Yu and Yuan (2011)), we find similar results using consumer confidence measures as an alternative proxy for sentiment (see, for instance, Fisher and Statman (2003), Lemmon and Portniaguina (2006), Bergman and Roychowdhury (2008), and Schmeling (2009)). 5 We focus on institutional and individual investors’ demand shocks and changes in sentiment, because both institutional investors’ ownership levels and sentiment levels are highly persistent which can lead to problems in inference (see Yule (1926), Granger and Newbold (1974), Ferson, Sarkissian, and Simin (2003), and Novy-Marx (2012)). Our tests based on changes in sentiment (and changes in institutional/individual investor ownership) largely avoid this issue. 4 Second, one of the components of the BW sentiment measure—the dividend premium—is computed from the cross-section of securities. That is, BW (2004, 2006, 2007) posit a rise in sentiment causes sentiment traders to increase their demand for speculative non-dividend paying stocks and decrease their demand for safe dividend paying stocks, resulting in a decline in the dividend premium. Further inconsistent with the hypothesis that these metrics capture sentiment trading by irrational individual investors, changes in the dividend premium are negatively related to individual investors’ demand shocks. That is, the dividend premium increases when institutions buy dividend paying stocks from individual investors and sell non-dividend paying stocks to individual investors. Although the primary focus of our study is identifying whose demand shocks sentiment metrics capture, our empirical results naturally bring up another question—why do investor sentiment metrics capture |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://gsm.ucdavis.edu/sites/main/files/file-attachments/06_who_are_the_sentiment_traders_-_evidence_from_the_cross_section_of_stock_returns_and_demand.pdf |
| Language | English |
| Access Restriction | Open |
| Subject Keyword | Aggregate data Algorithmic trading Angela McLean (biologist) Clients Copyright Cross-sectional data Darknet market Ephrin Type-B Receptor 1, human Experience Flow Hospitals, Proprietary Inference Kelvin Know-how trading Large Marx generator Ocean Observatories Initiative Paul Moskowitz SAP NetWeaver Business Warehouse Scott R. Lemmon Sentiment analysis Shock Software metric Speculative execution Time series Traders Volatility buying explanation sentiments |
| Content Type | Text |
| Resource Type | Article |
