Analyst’s Target Price Revision and Dealer’s Trading Behavior Analysis: Evidence from Taiwanese Stock Market

Abstract

This work utilizes the Taiwanese data primarily focused on retailing investor behavior to examine whether Taiwanese brokerage analysts issue target price revisions, whether implicit information connotation exists and whether their own brokerages use the market reaction brought about by target price revisions to conduct conflict of interest operations. The event study is used to verify whether the above results exist. The empirical results show that analysts may publish information that includes investment recommendations, earnings forecasts, or price target forecasts. Whether investors with immediate and post-event media coverage revise their relevant investment strategies and avoid serious losses caused by this news is established. The research results show that the target price revision has implicit information content no matter the target price being revised. In addition, a conflict of interest between dealers’ trading behavior and analysts’ target price revisions exists. The major contribution of this work is to fill the research gaps concerning which retail investors are easily influenced by social media and herding behavior, as well as target price forecasting. The most efficient use of resources relates to the satisfaction of everyone’s interests on a fair basis, and thus greater contribution. The governance mechanism and check and balance function can help maximize the value of the company, not only by enhancing the competitiveness of the enterprise, but also by increasing the value of shareholders’ rights and interests and better fulfilling corporate social responsibility.

1. Introduction

Motivations arising from self-interest are more likely to influence people’s judgments and decisions than professional responsibilities. Self-interest is often automatic and unconscious and is therefore more difficult to reduce through external control or regulation ([1]). In July 2020, over 5 million related documents and materials were available for the English term “conflicts of interest” in Google Scholar (nearly 400 million search results). Early literature on conflicts of interest focused on analyzing analysts versus underwriting or brokerage departments ([2,3,4,5,6,7]). Research on the conflict of interest between analysts and the proprietary sector began at a later stage and is still not as rich as that in the underwriting and brokerage sectors.

Conflicts of interest arising from services provided by brokerages and exchanges are common and the academic literature is abundant. When a company has an underwriting business, analysts tend to issue more bullish recommendations, and the stocks being recommended will underperform in the future ([7,8]). Research by [9] shows that analysts, by issuing buy recommendations, increase brokerage firms buying and selling transactions, increasing their company’s commission income. Shen and Chih (2009) [10] aimed at the conflict of interest between analysts and the company’s proprietary department and found that before the analyst issues a buy recommendation, the operations department will buy the recommended stock first and sell it after the recommendation is released for short-term compensation. Mehran and Stulz (2007) [11] defined a “conflict of interest” as party exploiting some behavior that would adversely affect the opponent to gain a potential benefit. The self-operated department may use analysts to release forecast information to buy and sell related stocks based on their own interest motives, resulting in conflicts of interest. When one is motivated to maximize the company’s interests, or when the proprietary department acts as a market maker, there may be a need to maintain or create liquidity profit from it. Fecht et al. (2018) [12] find that when stock liquidity is relatively low, investment banks may sell stocks to customers, which further confirms that dealers may have conflict of interest behaviors under self-interest motives.

The current literature on the conflict of interest between analysts and the proprietary sector can be divided into analysts’ buy and sell signals before and after issuing investment recommendations ([10,13]). Analysts may publish information that includes investment recommendations, earnings forecasts or price target forecasts. In the research of Brav and Lehavy (2003) [14], after excluding information such as investment advice and profit forecast, the target price still has information content for investors’ reference. Asquith et al. (2005) [15] further point out, since target prices are usually calculated by multiplying forecast earnings and financial ratios, target prices are more informative than earnings forecasts and investment recommendations. Earnings forecasts, investment recommendations and target prices published by analysts have informational connotations ([15,16,17,18,19,20,21,22,23,24,25,26,27,28]). Investors usually rely on the information released by analysts as a reference for decision makers, which is the so-called information content ([29,30,31,32,33]). Retail investors refer to a large amount of information when analyzing individual stocks, but the information that influences investors’ decisions is the key factors related to future expectations ([34]). Therefore, analysts’ earnings forecasts, investment recommendations and target prices are the key factors for investors’ investment making decisions. (The research reports released by sell-side analysts are mainly provided to external investors and are more likely to attract market attention. Therefore, this research will use it as the main discussion object, and the term “analyst” hereafter refers to the “sell-side analyst”. According to the “Analyst Conflict of Interest Report” issued by the Technical Committee of the International Organization of Securities Commissions Organization (IOSCO) in September 2003, the types of analysts can be mainly divided into sell-side analysts, buy-side analysts and independent analysts. The research reports issued by sell-side analysts are mainly provided to clients of their brokerage companies; buy-side analysts’ research reports are primarily provided to their employers to express opinions on their buying, selling or holding of securities; independent analysts’ employers are generally research firms, and their published research reports are usually the company's primary research report the source of income.)

To sum up, most of the literature focused on earnings forecasting and investment advice, and there is still a lack of research on target price forecasting. Based on this, this study will use the information of target price revisions to observe whether there is a conflict of interest in the trading behavior of the self-operated department of the securities company to which the analyst belongs before and after the release. To sum up, this paper will resolve the following points of interest: (1) Whether the analyst’s target price revision has reference value; (2) before and after the analyst’s target price forecast, whether there are corresponding excess trading operations in the securities company’s proprietary department, resulting in conflicts with investors’ conflict of interest.

The structure of the paper is as follows. Section 2 provides literature review and research hypothesis, while data sources and variable definitions in Section 3. The empirical results are shown in Section 4. Section 5 concludes our findings.

2. Literature Review and Research Hypothesis

2.1. Information Content of Analyst Target Price

Whether the research reports provided by analysts have rich content has been discussed by many scholars in the literature. Most revolved around the contents provided in the research reports, which were nothing more than investment recommendations, earnings forecasts and target price forecasts ([17,18,19,20,21,22,23,24,25,26,27,28]) until [16] began to study the information content of target prices.

Since target prices are often derived from financial ratios ([14,15,35]), analysts use target prices to support their investment recommendations, wherein positive investment recommendations are often accompanied by higher target prices ([16]). The more the target price is higher than the current stock price, the more overvalued the target price and the more positive investment advice will be associated with it.

Past studies have explained the informational connotation of target prices by observing short-term or long-term market reactions brought about by the release of target prices. Brav and Lehavy (2003) [14] conducted research using the target price data provided by the First Call database from 1997 to 1999 to explore the short-term market reaction after the target price was released and the relationship between long-term target price and direction of the market price. The results show that a more positive investment recommendation is more likely accompanied by a target price. Research results show that the positive target price revision has abnormal returns, and the target price has incremental information content. In addition, Asquith et al. (2005) [15] collect 1126 reports from 56 analysts between 1997 and 1999 from the Zacks and Investext database (formerly known as the First Call database), which was used for analysts’ research reports. Research results show that the target price forecast has a far greater impact on the stock price than the profit forecast and investment advice. Feldman et al. (2012) [36] further use data from 1999 to 2010 to analyze immediate and delayed market reactions to analysts’ revisions to earnings forecasts, target prices and investment recommendations. The results show that the short-term market returns of target price and investment recommendation revisions are higher than earnings forecast revisions, indicating that the former two revisions do provide more information than the latter.

It can be seen that Target Price Revision has an information connotation ([14]), which can influence investors’ decision-making and make them follow the information released by analysts. When the analyst adjusts the target price upward (downward) compared with the previous period, it indicates that the target price contains beneficial new information, and the market will also react and correct accordingly, resulting in a positive(/negative) abnormal return. Therefore, the following hypothesis is formed:

Hypothesis 1.

Analysts issue target price revisions, resulting in abnormal returns.

2.2. Conflicts of Interest between Analysts and Their Own Securities Companies

Lin and Kuo (2007) [13] use the investment recommendations issued by the top five brokerage analysts on the recommendation column of the Taiwan news media—“Economic Daily”—every Sunday from June 2000 to the end of January 2003, excluding IPOs (initial public offerings) individual stocks. The research investigates the impact of investment advice on the company’s stock price and examines whether a conflict of interest between Research Departments and Dealer Departments exists. Two ratios are used: (1) forecast error ratio: the number of stocks with a negative cumulative abnormal return 1 week after the release of the brokerage investment advice, except the number of stocks that the brokerage recommends to buy in the current week; and (2) change in selling volume: the number of stocks, which is subtracted from the number of oversold stocks in the two weeks before the release, is divided by the number of oversold stocks in the two weeks before the release. The empirical results show that when the analyst’s research department issues a buying recommendation, the proprietary department has quietly sold stocks and the research department has quietly sold stocks. There is a significant conflict of interest with the self-operated sector.

Shen and Chih (2009) [10] discussed the conflict of interest between the brokerage sector and the proprietary trading sector using individual stocks issued by securities firms in Taiwan’s Economic Daily Database from 2000 to 2003. It uses the observation of the number of excess shares traded by the self-operated department in the eight weeks before and after the released date as a measure. Two conflict of interest indices are used: (1) a securities company’s conflict of interest index: the frequency of securities companies’ release, which are related to the occurrence of conflict of interest; and (2) individual stock’ conflict of interest: individual stock characteristics such as liquidity, scale, growth, systematic risk, frequency of publication, number of companies in the same industry, number of insider shareholdings.

It can be seen that, out of the self-interest motivation, such as trading performance or maintaining liquidity, analysts release the target price to carry out corresponding operations before and after the transaction release. When the target price is high, the operated department of securities company buy the stock before the release but reduce buying or even sell the stocks in reverse after the information releases, thus resulting in the phenomenon of conflict of interest. Therefore, the following hypothesis is formed:

Hypothesis 2.

Before and after the analyst releases the target price, there is a conflict of interest between the self-operated department of the securities company and the investor.

3. Data Sources and Variable Definitions

3.1. Data Sources

The data required for this research are all taken from the Taiwan Economic Journal (TEJ) database, including target prices announced by analysts, the company’s actual stock price, individual stock returns, market returns and over-trading in proprietary industries. Due to the inclusion of the information database released by analysts—the “Investment Advice of Securities Companies”—the information has been provided since March 2007, and this research is set from March 2007 to December 2019. The following is a description of each data source and sample selection method.

3.1.1. Information Released by Analysts

The target price forecast information released by analysts is taken from the “Investment Advice for Brokers” database of Taiwan Economic News, which includes research reports and newspaper media information of various securities companies, annual EPS forecasts, investment recommendations and target price forecasts for listed companies. To unify the sample types, the criteria are set as follows: (1) since there are two types of target prices (single value or an interval range) issued by analysts, the median is used as the target price if range-type target price data is taken; (2) samples whose target price exceeds three standard deviations of the current stock price; (3) there is not enough information on the trading of the proprietary sector, the daily rate of return of individual stocks and the daily rate of return of the Taiwan weighted index; (4) if a brokerage continuously releases target prices for the same stock within the same week, only the first sample will be taken. Therefore, the number of target price revision data is 85,497; after the adjustment, the number of samples is 4198, which covers a total of 397 listed OTC companies and 18 securities firms.

3.1.2. Individual Stocks and Stock Market Returns

In order to explore the information content of analysts’ target price revisions, both the daily return of individual stocks and the daily return of the weighted stock price index are used to calculate the abnormal return (AR) and cumulative abnormal return (CAR) are used. Daily rate of return of individual stock and daily weighted rate of return is taken from the “Stock Price Database” of Taiwan Economic News. The number of samples is 172,118.

3.1.3. Trading Operations of Proprietary Departments

The trading operations of the self-operated sector are measured by its trading volume, and the data is taken from the database of “Weekly In-Out and Inventory of Proprietary Traders” of the Taiwan Economic News. The research refers to the event window (1 week to 8 weeks before and after the release date of the target price) (Shen and Chih, 2009) [10]. The numbers of sample are 67,168.

3.2. Variable Definition

3.2.1. Analyst’s Target Price Revision Measure

Brav and Lehavy (2003) [14] confirmed that the target price revision has an information connotation, so this study refers to its measurement method of target price revision, which is defined as follows:

$${TPRev}_t=\frac{{TP}_t}{{TP}_{t-1}}-1$$

(1)

where ${TPRev}_t$ is the target price revision value in period t; and ${TP}_t$ is the target price published by analysts in period t. The ${TPRev}_t$ value is less(/greater) than 0, indicating that the analyst adjusted target price issued lower(/higher) relative to the previous period; while ${TPRev}_t$ value is equal to 0, the analyst’s target price is not adjusted.

3.2.2. Information Content of Target Price Revision

To measure whether target price revisions are informative, this study examines whether abnormal returns from t days before the target price announcement to t days after the target price announcement in the market model used, the method of which is the same as pertains to the previously defined. t.

3.2.3. Conflict of Interest Measurement in Self-Operated Sector

Before and after analysts release target price revisions, the self-operated departments belonging to the same company may reverse buying and selling operations, and conflicts of interest occur. The conflict of interest index ([10]) is defined as the number of shares bought (${BS}_{i,\tau }^j$) and the number of shares sold (${SS}_{i,\tau }^j$). The data is taken from the “Weekly In-Out and Inventory of Dealers” database of Taiwan Economic News Data, a measure of the super-operational behavior of buying and selling before and after the release. The equation is as follows:

$${NTS}_{i,\tau }=\frac{{\sum }_{j=1}^{M_i}{BS}_{i,\tau }^j-{\sum }_{j=1}^{M_i}{SS}_{i,\tau }^j}{M_i}$$

(2)

where $M_i$ is the total number of target prices issued by the i-th securities firm; and ${NTS}_{i,\tau }$ is the net trading shares, i.e., the number of net trading excess shares of the i-th securities in the τ week before or after the release of the target price. When $NTS$ is greater(/smaller) than 0, the number of shares traded by securities firms in the τ week before or after the release of the target price is net overbought(/oversold).

4. Empirical Results

Descriptive Statistics

According to the results in

Table 1. Sample of Descriptive Statistics.

	# Sample	Mean	Std. Dev.	Max.	Medium	Min.
The number of companies with published target prices	397	-	-	-	-	-
The number of analysts and self-operated departments	18	-	-	-	-	-
The number of times the brokerage has released the target price	4198	210	-	1462	111	2
The number of times the company has issued a target price	4198	11	-	117	4	1
Target price revision	4198	1.46%	20.31%	100.00%	0.00%	−83.78%
Company stock price	4198	152	186	1435	75	10

, within the sample of this study, a total of 18 analysts issued target prices (396 listed and OTC companies). According to the results in

Table 2. Upward revision, downward revision and flat revision of target price in each year.

Year	Sample	Mean	Upward Revision		Downward Revision		Flat Situation
			Number	Ratio	Number	Ratio	Number	Ratio
2007	331	0.062 ***	163	49%	75	23%	93	28%
2008	388	−0.106 ***	87	22%	222	57%	79	20%
2009	302	0.113 ***	165	55%	64	21%	73	24%
2010	532	0.021 ***	229	43%	194	36%	109	20%
2011	365	−0.039 ***	135	37%	191	52%	39	11%
2012	141	0.005	59	42%	71	50%	11	8%
2013	48	0.050	19	40%	16	33%	13	27%
2014	162	0.042 ***	60	37%	53	33%	49	30%
2015	392	0.011	165	42%	166	42%	61	16%
2016	403	0.013	156	39%	168	42%	79	20%
2017	430	0.072 ***	202	47%	87	20%	141	33%
2018	456	−0.029 ***	142	31%	187	41%	127	28%
2019	248	0.054 ***	116	47%	68	27%	64	26%
Total	(n = 4198)	0.015 ***	1698	40%	1562	37%	938	22%

Note: (1) *** represent the significant levels of and 1%, respectively. (2) This study uses NTS (net trading shares) to measure the proprietary of sectors’ buying and selling for individual stocks. When NTS is greater than 0, it means that the net trading shares of the current week are positive; otherwise, the net trading shares of the current week are negative. The sample size is 4198.

, the target prices published by analysts in each year have the highest proportion of upwardly revised target prices, and the average target price revisions in each year are mostly positive, which shows that analysts are usually more willing to issue positive news. The average annual target price revision shows that the target price revision is negative during periods such as the financial crisis in 2008, the European debt crisis in 2011 and the China-US trade war in 2018, it can be seen that analysts are not optimistic about the performance of individual stocks. Therefore, the overall average target price for the current year is revised downward. (The results are omitted for sake of brevity but are available upon request.)

This paper mainly explores whether a conflict of interest in the buying and selling behavior before and after the release exists, wherein the revision of target price released by security companies is related to the buying and selling behavior of the self-operated department. The results show the highest proportion of samples with upward revision to target price (

Table 3. Changes between the weekly average net buying and selling of excess shares of the proprietary sector before and after the announcement of the target price.

	Period	Average NTS with the Upward Revision to Target Price	Average NTS with the Downward Revision to Target Price
Before Release	8 Weeks	−0.159	−0.460 **
	7 Weeks	0.038	−0.484
	6 Weeks	0.102	−0.442 *
	5 Weeks	0.615	−0.639 *
	4 Weeks	0.399	−0.509 *
	3 Weeks	−0.953	−0.136
	2 Weeks	−0.494	0.575 *
	1 Week	0.536 *	−0.298
After Release	1 Week	−1.354 *	−0.085
	2 Weeks	0.060	0.276
	3 Weeks	−0.253	0.035
	4 Weeks	−0.167	−0.215
	5 Weeks	−0.513	−0.036
	6 Weeks	0.041	0.191
	7 Weeks	0.285	0.403
	8 Weeks	−0.349 **	−0.345

Note: (1) *, ** represent the significant levels of 10% and 5%, respectively. (2) NTS (net trading shares) is defined to measure the proprietary of sectors’ buying and selling for individual stocks. When NTS is greater than 0, it means that the net trading shares of the current week are positive; otherwise, the net trading shares of the current week are negative.

), which is consistent with previous research results ([14]).

In this study, the individual stocks with target prices published by analysts are classified based on the classification defined by Taiwan Stock Exchange, and the differences between the electronics industry and the non-electronic industry are compared. First, it was found that the samples were mainly distributed in the electronics industry, with 2881 samples, accounting for 69% of the firms. (According to the industry classification announced by the Taiwan Stock Exchange, all the firms which re belong to the semiconductor industry, optoelectronics industry, communication network industry, electronic components, computer and peripheral equipment industry, electronic channel industry, information service industry and other electronic industries, are defined as the electronic industry.) If the target price announcement in each industry is further broken down, the semiconductor industry has the highest proportion in the electronics industry, with 1139 records, accounting for 27% of the total sample. In

Table 4. Changes in the cumulative net trading excess shares of the proprietary sector before and after the announcement of the target price.

Period	The Cumulative Average NTS of the Upward Revision of the Target Price		The Cumulative Average NTS of the Downward Revision of the Target Price
	Before Release	After Release	Before Release	After Release
1 Week	0.536 *	−1.354 *	−0.298	−0.085
2 Weeks	0.042	−1.293	0.277	0.191
3 Weeks	−0.911	−1.546 **	0.142	0.226
4 Weeks	−0.512	−1.714 ***	−0.367	0.011
5 Weeks	0.104	−2.227 ***	−1.006	−0.025
6 Weeks	0.205	−2.186 ***	−1.448 **	0.166
7 Weeks	0.244	−1.901 **	−1.933 **	0.568
8 Weeks	0.085	−2.250 **	−2.392 ***	0.224

Note: (1) *, ** and *** represent the significant levels of 10%, 5% and 1%, respectively. (2) The average accumulation from 1 week to τ weeks before (after) the target price is calculated. When the cumulative average NTS is greater (smaller) than 0, indicating that the cumulative net trading shares from 1 week to τ weeks before (after) the adjustment of target price is positive (negative).

, Analysts are more inclined to publish target price information for electronics industry-related stocks and even the semiconductor industry. (The results are omitted for sake of brevity but are available upon request.)

5. Conclusions

This study explores Taiwan’s listed companies from 2007 to 2019 as a sample to observe whether the target prices released by analysts have informational connotations. Moreover, whether shares are pre-purchased by investors before the release and then sold after launch, thus engendering a conflict of interest situation, is examined. The event is used to verify the above results.

Based on this work, the information content of the target price revision and both the upward revision and the downward revision of the target price are verified. Secondly, analysts’ target price revisions will bring corresponding market reactions. The target price revisions imply information connotations, and the self-operated departments of the brokerage firms and their analysts may have conflicts of interest before and after the target price revisions are released. Taiwanese retail investors are easily influenced by social media and herding behavior exists.

This study can provide investors with immediate and post-event media coverage to revise their relevant investment strategies and avoid serious losses caused by this news. If the conflicts of interest between investors are reduced, the firms can devote themselves to corporate social responsibility and increase the benefits of stakeholders. The corporate governance mechanism is a common benefit value among the company’s shareholders, managers, board of directors and other stakeholders. Through efficient governance mechanisms and checks and balances, it can help maximize the company’s value, not only by enhancing the competitiveness of the company, but also by improving the equity of shareholders’ value and better fulfilling corporate social responsibility. In the future, we aim to use the algorithms PSO, GA and BAS to explore the meta-heuristic algorithms. Moreover, cross-country data and long study periods are encouraged to overcome the sample biases. Due to the COVID-19 pandemic, the data for the years 2020–2022 are excluded. This was to avoid the results being affected by other factors.