Financial Analysts and Speculative Bubble in Emerging Stock Market[*]
Chi-Wen Jevons Lee
Tulane Universityand Zhejiang University
Chang Liu
TsinghuaUniversity
August 31, 2008
Financial Analyst and Speculative Bubble in Emerging Stock Market
Abstract
This paper examines the speculative bubble regularlyset off by financial analysts in China.We document significant positive two-day excess returns right before analyst’s buy recommendation and then significant reversals afterward. The securities with greater initial price rise suffer greater subsequent reversals. After controlling the structure of analyst market, we find that the initial price rise and subsequent reversal are positively correlated with the liquidity of the security.Due to the peculiar trading restriction, we find a way to document naive trader’s behavior. Our results suggest that analyst’s recommendations incite intensive noise trading by naive investors as well asconvey fundamental information.Thinness of market and associated illiquidity create bubble and profitable trading opportunity for informed traders. This research captures the rare episode of speculative bubbles associated with market learning toward information efficiency.
Keywords: financial analyst, stock recommendation, price pressure, market inefficiency
I. Introduction
This paper studies the market activities measured by stock returns and trading volumes around the publication of analysts' stock recommendations in China.[1]Based on alarge sample of recommendations in the weekly column "Stocks Recommended by The Most Analysts This Week" in Shanghai Securities News on every Monday during 2000 to 2005, we find apparent pattern of surging and receding of stock prices around the recommendation. Positive abnormal return begins the day before the analyst recommendation. Cumulative abnormal return reaches the peak onthe announcement date, declines sharply the next day, and never rebounds thereafter.Almost half of the two-day initial rise is reversed in the subsequent 5 days after announcement. We find evidence that dealers primarily trade with naïve traders, not between themselves. The trading activities are negatively associated with liquidity. This paper finds repetitive 5373 episodes of mini-bubbles set off by analyst recommendation. The magnitude of bubbles is positively associated with the number of analysts recommending the stocks. We provide large sample evidence to the bubbles predicted byAllen, Morris and Postlewaite (1993) and Abreu and Brunnermeier (2003)
Detailed analyses reveal that the post-announcementcumulative abnormal returns are negatively correlated with the surge before the announcement, i.e., securities with greater price surge suffer larger reversals. Moreover, we find a positive correlation between abnormal returns and abnormal trading volumes on the announcement date. More active trading is associated with greater swing of returns surges and reversals. Using market value of stocks, dollar volume, standard deviation of daily raw returns and ratio of absolute stock return to dollar volume as liquidity measures, we find that, with informational effect on returns being controlled for, the on-announcement returns decrease, while post-announcement returns increase, with the liquidity of the recommended security. Less liquid securities suffer greater fluctuation, larger initial rise and larger subsequent fall. These phenomena suggest speculative bubbles ignited by analyst forecasts due to market illiquidity. Sudden and large shifts in demand can lead to a temporary order-flow imbalance and price is adjusted to encourage investors to accommodate the demand shifts. Our empirical evidence reveals price pressure around the announcement of stock recommendation.We find speculative bubbles regularly set off by the “coarse andstale” information of analyst recommendations.
Price pressure drives price away from fundamental value. The existence of pricepressure suggests that not all the abnormal returnsare information driven. Price adjustments with little information contentarecharacterized as bubble. By bubble, we mean that price systematically deviates from its fundamental value; market overreacts beyond what fundamentals would imply.[2]Price pressure can arise fromnoise trading and naïveherding.[3]Shleifer and Summers(1990) suggested that noise traders regard the analysts as market gurus whose recommendationsstimulate herding behavior. The noise traders herd after the public signal and the actions of one another. Their bursting sentimentshifts the demand curve.Allen, Morris, and Postlewaite (1993) suggested that short-sales constraints and symmetric information can create finite bubbles. These two attributes are much more prominent in an emerging market such as China than those developed market documented in the literature.
On the announcement date for a buy recommendation, noise traders rush to buy the stocks, pushing up prices. Due to the "t+1" trading rule in China, they cannot sell the stocks on the same day as they buy them, but can do so on or after the next day. Therefore, they sell these stocks in haste after the announcement date to realize the profits. The consequence of their speculation activities is the temporary imbalance between demand and supply for the recommended securities. In order to resolve the order imbalance, prices have to be adjusted away from the equilibrium value so that some investors would be willing to provide liquidity service. Thus the mean-reversion of the stock returns comes about. Liquidity measures a security's vulnerability to short-run demand shocks. Less liquid security prices are more subject to temporary demand shifts, resulting in larger price fluctuations.
In addition, we also find that the magnitude of the on-announcement run-upis declining as years go by, greatest in year 2000 (the first sample year) and smallest in year 2005 (the last sample year). The year-median of the post-announcement reversals, in general, also declines with time, largest in year 2000 and second smallest in year 2005, suggesting that investors are learning from past experience and thus initial enthusiasm fades over time.This research captures the rare episode of bubbles associated with market learning toward information efficiency.
As Grossman and Stiglitz (1980) argued, the existence of price pressure is antithetical to the efficient market hypothesis, but crucial for market equilibrium under rational expectation. There are two critical elements for assets market equilibrium and an additional one for bubbles. The first is the existence of noise traders. In China's security market, most individual investors lack the knowledge of investment and the ways of obtaining information. They trade according to their individual sentiment or by herding. The second element is the informed traders. In this study, we can capture the information as the analysts' recommendations on the newspapers. In the youthful China stock market, listed corporations' disclosure and transparency are so opaque that individual investors have to rely on public analysts' recommendations.One would find it difficult to provide significant evidence to Grossman-Stiglitz Theorem in the U.S. and Britain. In this paper, we clearly identify the evidence of price pressure in China. Beyond the first two elements for Grossman-Stiglitz Theorem, a crucial component for bubble is the limitation of arbitrage. In China, short-arbitrageurs aiming at arbitrage profits have to own the stock before the announcement date. Long-arbitrage needs large amount of cash, which is quite difficult to acquire in China's rigidly regulated financialsystem. Hence, informed traders are only able to arbitrage to the extent of cash in pocket.This paper provides a unified empirical study for Grossman and Stiglitz (1980) and Abreu and Brunnermeier (2003).
A rational market cannot be totally blind. Price pressure in Grossman-Stiglitz Theorem required fundamental innovations. We find that the market reactions beginbefore the information release, implyinginformation leakage, even perhaps manipulation.On-announcement returns increase withthe number of recommending analysts, and decrease with the frequency of being recommended and firm size, consistent with the information hypothesisthat analyst's recommendation releases relevant information for fundamental revaluation. The number of recommending analysts measures the precision of information. More analysts' recommendationsindicate higher informationprecision. Frequency of being recommended measures the amount of information already released to the market before recommendation.Similarly, firm size is a measure of information environment and the extent of market attention.The information of frequently recommended securities and large firms is gathered in greater detail, processed with greater care, and disseminated more widely. Hence the shock caused by a single piece of information at its announcement would be small.Price pressure, arising from investors' incapability of accurate valuation, is closely associated withthe nature of information.
In China, institutional traders can trade when the market is closed whereas individual traders can only trade when the market is open. Institutional traders are better informed and more sophisticated. We have micro data to show trading behavior within trading day and between trading date. These data allow us to show the different trading behavior of informed traders and noise traders. We find that the regular speculative bubbles in our case are driven by the informed traders extracting rent of information from the noise traders. In this paper, we provide empirical linkage between speculative bubbles and noisy rational expectation.
This paper proceeds as follows. The next section is a review of the related literature, followed by hypotheses development. Section III provides evidences of speculative bubbles. Section IV conducts series of empirical rests to figure out the driving force behind speculative bubbles. Conclusions are provided at the end.
II. Literature Review and Hypotheses Development
Financial analysts,serving in brokerage houses, independent research institutes, fund corporations and banks, play important roles in allocating resources in capital markets. They are primaryinformation intermediaries in the market:collecting private information, forecasting firms’prospects, and conducting retrospective analysis that interprets past events (Beaver (1998), p.10). Their labors enhance the informational efficiency of the capital markets(Frankel et al. (2006)).
Lee (1986, 1987) found preliminary evidence thatthe value of analyst’s recommendation is positively associated withinformation costs, consistent with Grossman-Stiglitz Theorem. However, considerable evidence suggests that not all price changes are fully justified by information. Shleifer and Summers (1990) proposes the noise trader approach to finance, as an alternative to the efficient market paradigm. This approach rests on two critical assumptions: investor irrationality and limited arbitrage. First, some investors are not fully rational, subjecting to the influence of sentiments. Second, arbitrage is limited because of the risk arbitrageurs have to assume. They developed several implications consistent with the evidence, such as closed-end funds changes, price pressure, and historical episodes. One of the implications suggests that some demand changes may be a response to changes in sentiment not justified by information or pseudo-signals perceived to convey information, and thus are irrational. If many investors base their trading strategies on pseudo-signals, e.g., advice of financial gurus, aggregate demand shifts will result.
Barber and Loeffler (1993) analyze returns and volume around the announcement of analysts' recommendations appearing in the monthly "Dartboard" column of the Wall Street Journal. The stocks selected by analysts experience a 4.06 percent (t=10.77) abnormal return over the two-day period consisting of the journal's publication day and the subsequent day, and a -2.08 percent (t=-1.56) return from day 2 through day 25. Firms experiencing positive abnormal volume on the announcement have significant post-announcement price reversal, while those with no positive abnormal volume on the announcement have no reversal. Less liquid firms have a larger price reaction on days 0 and 1. Their evidence suggests that the price response on the announcement of recommendations is at least partially driven by buying pressure in the recommended securities. Greene and Smart (1999) argue that "Dartboard" column generates temporary price pressure by increasing noise (i.e., uninformed) trading. This conclusion comes from two pieces of evidence: most of the abnormal returns following the column's publication disappear within a few weeks, and abnormal trading volume and temporary abnormal returns are greater for stocks recommended by analysts with successful records, which are not at all predictor of their future success and long term performance. Securities with the greatest initial price run-up and the greatest increase in trading volume experience the largest price reversals. Liang (1999) also documents the reversion pattern in the "Dartboard" column stock prices. The two-day announcement abnormal return of 3.52 percent starts to reverse from the third day. Higher trading volume and reduced bid-ask spread are indicators of more noise trading by naive investors, which further supports the price pressure hypothesis. Stickel (1995) finds that the magnitude of change in recommendation, analyst reputation，and broker size have temporary effects on stock prices.
Nevertheless, identifying price pressure caused by analysts' recommendations may be inconclusive because these events usually convey new information to the market and it is quite difficult to distinguish the price pressure from the informational effect. Harris and Gurel (1986) avoid this objection by examining price reactions to changes in the composition of Standard and Poor's list of 500 stocks, an event that does not plausibly reflect any new fundamental information. Changes in the composition of S&P 500 should not reveal new information about performance potential because the changes are based only on publicly available information and on well-known criteria, not on forecast returns. But they do shift demand as many large index funds' portfolio holdings just represent the S&P index. Harris and Gurel (1986) find that immediately after an inclusion into the index is announced, price increases by more than 3 percent, which is nearly fully reversed after 2 weeks. Besides, the magnitude of these increases has risen over time, paralleling the growth of index funds. Ritter (1988) studies the January effect, another situation not confounded by the informational problems. January effect, one of the anomalies in stock returns, refers to the phenomenon that low-capitalization stocks have unusually higher returns than large ones in every January. Ritter finds that individuals have a below-normal buy/sell ratio in late December and an above-normal ratio in early January because small stocks usually held by individuals are sold during December to realize losses for tax purposes and reinvested in later in early January.
In this article, two joint hypotheses are empirically tested: the information hypothesis and the price pressure hypothesis. The information hypothesis asserts that analysts' recommendations reveal relevant information, with which the market revaluates the security, and thus the abnormal performance on the announcement represents a fundamental revaluation of the security. In contrast, the price pressure hypothesis maintains that the recommendations create temporary buying or selling pressure by naive traders in the recommended securities, and this pressure leads to the observed abnormal performance.
The information hypothesis suggests that analysts provide useful assessments to market players. The price pressure hypothesis suggests that investors who accommodate demand shifts must be compensated for the transaction costs and portfolio risks that they bear when they agree to immediately buy or sell securities which they otherwise would not trade (Harris and Gurel (1986)). Liquidity and transaction costs are the root of price pressure. The noise traders compensate the informed traders through price pressure. This phenomenon is commonly called as speculative bubbles. The joint evidence for information hypothesis and price pressure hypothesis will support Grossman-Stiglitz Theorem. The evidence of price pressure without the support of information hypothesis will favorthe sentiment argument made in Shleifer and Summers (1990)
Financial analysts can generate both useful information service and speculative bubbles, especially in emerging markets. As pointed out in Aharony, Lee, and Wong (2000) and Lee (2001), because of high transaction costs,opaque information system, and evolving market structure, China provides better setting to examine speculative bubbles than all other markets of same economic significance.
III. Trace of Speculative Bubbles
The weekly column "Stocks Recommended by Most Analysts This Week" has been published on every Mondayin Shanghai Securities Newssince January 10, 2000. This column summarizes the securities recommended by the largest number of analysts in major security newspapers in China on former Friday, such as Shanghai Securities News itself,China Securities Journal, Securities Times,and Information Morning Post.Each security in a column receives somewhat favorable comments from at least 2 analysts. In a typical column, each security's name is, in sequence, followed by the number of recommending analysts, analysts' names and short quotations of their opinions.Most of the opinions are technical analyses with little fundamental contents. There is scarcely an analyst claiming to have access to inside information.[4] These reports are “coarse andstale.” Market reactions to them are viewed as speculative bubbles by most observers.