“Effects of ambiguity in market reaction to changes in stock recommendations”

This study uses analyst recommendations and three ambiguity proxies, namely ambiguity in fundamentals, ambiguity in information and market ambiguity, to examine market reaction to recommendation changes in the Taiwanese stock market. The authors find that analysts’ recommendation changes have positive effects on subsequent buy-and-hold abnormal returns when market ambiguity is moderate. When ambiguity in fundamentals is low, recommendation changes have a positive influence on smaller firms. The effect of ambiguity in information on stock returns is associated with market ambiguity; market ambiguity is negatively associated with abnormal returns for firms with moderate ambiguity in fundamentals. Investors in a small firm rely more on analyst recommendations. Mei-Chen Lin (Taiwan), Chen-Yang Lin (Taiwan), Ming-Ti Chiang (Taiwan) BUSINESS PERSPECTIVES LLC “СPС “Business Perspectives” Hryhorii Skovoroda lane, 10, Sumy, 40022, Ukraine www.businessperspectives.org Effects of ambiguity in market reaction to changes in stock recommendations Received on: 10th of April, 2017 Accepted on: 6th of June, 2017


INTRODUCTION
The puzzle of mean variance premium is the difference between risk-neutral and objective expectations of market return variance. Miao et al. (2012) find that about 96% of the mean variance premium can be referred to as ambiguity aversion. 2 Illeditch (2011) shows that, due to the effects of risk and ambiguity on optimal portfolio and equilibrium asset prices, the desire of investors to hedge ambiguity will lead to portfolio inertia and excess volatility when investors receive information that is hard to link to fundamentals. Even when there is no transaction costs or other market frictions, investors may not react to price changes when news is surprising. In addition, this paper shows that small shocks to cash flow news, asset betas or market risk premia may have dramatic impacts on stock prices and excess volatility.
Investors tend to be ambiguity averse when faced with risk and uncertainty. Keren and Gerritsen (1999) argue that ambiguity aversion is a situation in which decision-makers prefer gambling with a known probability to gambling with uncertainty. The experiment of Bossaerts et al. (2010) shows that people are ambiguity averse (Heath & Tversky, 1991). Ambiguity aversion can explain the responses of investors. For example, the asymmetric response of investors to good and bad news is contributed by ambiguity aversion (Epstein & Schneider, 2008). Williams (2014) finds that investors have a greater response to bad news than to good news as ambiguity increases. Gilboa and Schmeidler (1989) argue that investors choose the worse cases when ambiguity exists. Moreover, ambiguity aversion could explain the equity premium puzzle (Mehra & Prescott, 1985;Rieger & Wang, 2012), mean variance premium 1 , portfolio inertia and excess volatility of stock prices (Illeditsch, 2011) 2 .
Driss (2013) defines two kinds of ambiguity, namely ambiguity in fundamentals (AIF) and ambiguity in information (AII). AIF is a situation in which investors know the fundamental prospect of a firm, but cannot make reasonable decisions, since there is no relevant information. AII is a situation in which investors have difficulty in updating their prior beliefs in dealing with the information, since they have doubts about the uncertain information quality. Driss (2013) finds that investors have a greater response to analyst recommendation changes, as they perceive a higher level of AIF or AII.
Investors are ambiguity averse, as they try to figure out the unfamiliar circumstances they face (Epstein & Schneider, 2008;Liu, Pan, & Wang, 2005). William (2014) finds this directly affects investor reaction to firms' earnings announcements (Drechsler, 2013;Epstein & Schneider, 2008;Hansen & Sargent, 2010;Illeditsch, 2011). Typically, investors receive earnings information and face a set probability distribution of revenues. If the market is ambiguous, they act cautiously and choose the worse information. As a result, they give more weight to the bad news and so respond more to bad news than to good.
Prior research points out that analyst recommendations convey information contexts and that investors could obtain normal or abnormal returns through investing in stocks recommended by analysts (Beneish, 1991;Bauman et al., 1995;Womack, 1996;Barber et al., 2001;Jegadeesh et al., 2004). We use ambiguity in fundamentals (AIF), ambiguity in information (AII) (Driss, 2013) and market ambiguity (the change of VIX) (Williams, 2014) to measure whether different ambiguity proxies affect investors' decisions after analyst recommendations. We find that only in the situation of low AIF, do recommendations have stronger and positive impacts on smaller firms. When market ambiguity is moderate, analyst recommendation changes have a positive effect on buy-and-hold abnormal returns. For low-level market ambiguity, larger, more analyst-recommended and younger firms have higher buy-and-hold abnormal returns. The response of stock prices to AIF is affected by market ambiguity.
The contribution of this paper is to simultaneously explore whether markets respond differently to recommendation changes for various ambiguity proxies -AIF, AII and market ambiguity. The findings will help investors to make investment decisions when considering analyst recommendations.
The remainder of this paper is organized as follows: section 1 presents the sample, variable definitions and research design, section 2 offers the empirical results; and final section summarizes the results and gives a conclusion.

RESEARCH DESIGN
where , jt r is the raw return on stock j at day , t r is the raw return on a benchmark portfolio formulated with a comparable size, book-tomarket and momentum as stock . j

Proxy for ambiguity
We use the ambiguity in fundamentals (AIF), ambiguity in information (AII) (Driss, 2013) and market ambiguity index, the change of VIX (Williams, 2014), to measure how different ambiguity proxies affect investors' decisions. The ambiguity variables are defined as follows.

Ambiguity in fundamentals
where 52 j N WL is the stock price's nearness to its 52-week low, j P is stock j's price at the end of the prior month, and, 52 j WH and 52 j WL are stock j's 52-week high and low prices, respectively, calculated from the 52-week period at the end of the prior month.

Ambiguity in information
Ambiguity in information (AII) is a situation in which a firm conveys too little or impre-cise information for investors to correctly interpret and investors have difficulty in updating their prior beliefs in response to that information. Ambiguity-averse investors are uncertain about the quality of the signal, and, thus, they give less weight to information carried by the signal and more weight to new information from analyst recommendations (Epstein & Schneider, 2008). In general, smaller firms, younger firms or firms with lower analyst coverage release less information to the public and receive little media coverage, thereby providing investors with information environments of lower quality. This study uses the reciprocals of firm size, analyst coverage and firm age as the proxies to measure AII ( A. Reciprocal of firm size (RME): smaller firms have a lower quality information environment.
Since the cost of an information release is fixed, smaller firms may offer less information. Firm size (ME) is measured as the market capitalization at the end of the previous month before recommendations are released.
B. Reciprocal of analyst coverage (RACOV): firms with lower analyst coverage have more information uncertainty. Analyst coverage is defined as the number of analysts following the firm during the year prior to the end of previous month.
C. Reciprocal of firm age (RAGE): compared to long-history firms, younger firms release less information to the public, receive little media coverage and attract less attention, thereby offering lower quality information to the public. This may lead to less influence in investors' decision-making. Firm age is measured as the number of quarters from the firm first time listed to the month prior to the announcement of recommendation change.

Market ambiguity
Following Williams (2014), we use the change of VIX ( VIX ∆ ) to measure market ambiguity. VIX is measured as the TXO (Taiwan Index Option) volatility index. A higher (lower) VIX ∆ indicates that investors expect more (less) market ambiguity.

Ambiguity aversion and recommendation changes
In order to measure whether or not the market response is affected by recommendation changes, we run the following regression model: (MAG). The variables are defined in the Appendix. Following Petersen (2009), we estimate the standard errors by allowing correlation between error terms. Furthermore, we also control for the year and industry effects to consider a fixed effects panel regression.
If investors respond to upgrades (downgrades), the subsequent BHARs are positively related to the direction of the recommendation changes, β 10 β and 11 β will be greater than zero.  Table 3 presents the regression results. Column 1 shows that the magnitude of recommendation change is significantly and positively correlated with

Level of ambiguity and analyst recommendations
This indicates that recommendation changes do affect the market reaction before controlling for ambiguity, as well as firm, analyst and recommendation characteristics. Column 2 includes the ambiguity proxy in the regression, and Column 3 further includes other control variables.
The results indicate that the coefficient of recommen-

. BHAR
However, the interaction terms of recommendation change between ambiguity in fundamentals, ambiguity in information and market ambiguity have no significant coefficients. This implies that ambiguity does not affect the market's immediate response to recommendation changes.
In order to investigate whether the above results still hold when the holding periods are extended, we use different periods of BHAR as the dependent variables. Table 4 and Table 5 show these results. The results in Table 4 include firm ambiguity and market ambiguity. The coefficient of recommendation change is positive for ( )

. BHAR
This indicates that upgrading firms have higher buy-and-hold abnormal returns than firms within the same industry and of similar size and BM ratio. The significant and positive ( ) 1 j D HAIF implies that a higher AIF firm has a lower . BHAR In addition, firms with shorter history have higher buy-and-hold abnormal returns, and this effect lasts for two days to six months. This implies that firms with strong ambiguity in information ( ) AII are affected by analyst recommendations for longer periods. Note: This table shows descriptive statistics of all variables. CHANGE is the recommendation changes, 52 N WL is the proxy for ambiguity in fundamentals, ME is the firm size, ACOV is the analyst coverage, AGE is the firm age, VIX ∆ is the change in the TXO volatility index, BM is the market to book ratio, MOM is the price momentum, IO is a common stock holding ratio of an institution, AXEP is the analyst experience, DC is the recommendation deviation from consensus, MAG is the magnitude of recommendation changes. The variables are defined in Appendix.    are the moderate ambiguity, which are dummy variables indicating ambiguity proxies ranked above their 30th and below their 70th percentiles, respectively. BM is book to market ratio, MOM is momentum of stock price return, IO is institutional ownership ratio, AXEP is analyst experience, DC is recommendation deviation from consensus, MAG is the magnitude of recommendation change. The variables are defined in Appendix. ( ) indicates the t-value. *,**,*** indicates statistical significance at 10%, 5% and 1% level, respectively.     Note: This table examine whether investors make decisions according to analyst recommendations after controlling for firm-level ambiguity and market ambiguity.
BHAR H is the buy-and hold abnormal returns for H = 1, 20, 62, 125 days. Other variables are defined in Table 3. ( ) indicates the t-value. *,**,*** indicate statistical significance at 10%, 5% and 1% level, respectively. Furthermore, in Table 5, we include the characteristics of firms, analysts and recommendations as control variables in order to examine the influence of recommendation changes on investors. We find that recommendation changes have no significant effect on investors' responses when considering the effects of ambiguity. Ambiguity in fundamentals ( ) AIF is significant and positively related to   Note: This table reports the effects of ambiguity on subsequent buy-and-hold returns over various periods. BHAR (0, H) is the buy-and-hold abnormal returns for H = 1, 20, 62, 125 days. Variables are as defined in Table 3 and Table 4. ( ) indicates the t-value. *,**,*** indicates statistical significance at 10%, 5%, and 1% level, respectively.
( ) 0.125 BHAR at a 5% confidence interval. In addition, firms with more analyst coverage have higher BHARs, an effect which persists for at least one month. Younger firms have higher two-day buy-and-hold abnormal returns. This result implies that, when analyst recommendations are changed, the effects of AII on subsequent stock price changes are mixed.
However, the interaction terms of recommendation change between AIF, AII or market ambiguity are not significantly different from zero. This implies that market response to recommendation changes is not affected by ambiguity magnitude for either short or long term. For more robust results, we divide AIF and AII into high, moderate and low firm subgroups, and further examine whether the market reaction to recommendation change is affected by different levels of ambiguity.      Note: This table presents the results when dividing the ambiguity in fundamentals into three groups. The sample is divided into low-, moderate-, and high-AIF groups according to the 30th and 70th percentiles of 52 N WL .
( ) 0, 1 BHAR is the dependent variable. Other variables are defined in Table 3 and 4. ( ) indicates the t-value. *,**,*** indicates statistical significance at 10%, 5%, and 1% level, respectively. Table 7 presents the results when market ambiguity is divided into three groups. Table 7  Moreover, the coefficient of the reciprocal of firm size and BHAR is marginally significant and negative at the 10% level, indicating that smaller firms which are recommended by analysts have lower returns during periods of low market am-biguity. The relation between the reciprocal of analyst coverage and BHAR is also significantly negative at the 10% level in times of low and moderate market ambiguity, but not significant for high market ambiguity. This implies that larger firms and firms with more analyst coverage have higher returns during periods of low and moderate ambiguity. The reciprocal of firm age is positively associated with BHAR for low ambiguity.
This implies that firms with a longer history have lower returns during periods of low market ambiguity. However, when controlling for the effect of market ambiguity, the effect of recommendation changes on stock prices is not significantly associated with ambiguity in fundamentals or ambiguity in information. Finally, the finding of price momentum remains unchanged regardless of market ambiguity.   Note: This table presents the results when dividing the market ambiguity into three groups. The sample is divided into low-, moderate-, and high-VIX ∆ groups according to the 30th and 70th percentiles of VIX ∆ . BHAR (0, 1) is the dependent variable. Other variables are defined in Tables 3 and 4. ( ) indicates the t-value. *,**,*** indicates statistical significance at 10%, 5% and 1% level, respectively.