在学术界，关于分析师预测行为的影响因素及其预测报告的预测质量一直是学者广泛讨论的课题，众多国内外文献均对此进行了深入研究。本文在已有研究的基础上拓展和深化了这一议题，采用行为金融学中的迎合理论作为理论基础，着重探讨了在中国股票市场的背景下，分析师进行盈利预测时存在的认知偏差及其与预测误差和未来股价之间的关系。具体来说，分析师预测受股票历史表现的影响不仅是考虑了公司的基本面因素，还可能是由于股价变化导致分析师情绪变化，进而影响其对未来盈利的预测。本文将分析师进行盈利预测时受过去股价变动影响的行为称为分析师的外推偏差，首先，检验分析师是否存在外推偏差，即分析师预测是否受近期的历史股价变化影响，然后，探讨这种外推偏差是否放大了分析师的预测偏差，最后，分析分析师的这种外推预期对股票未来回报的影响。

本文主要选取2010年1月1日-2022年12月31日A股市场中关注度较高的1139只股票作为研究对象，考察截面上分析师个体如何形成对股票未来收益的预期。为进一步研究清楚该问题，本文以五个交易日为一期，引入多期股票价格的变化率，以反映过去一段时间内股票价格的波动情况。我们将这些变化率作为股价历史变化指标，构建多元回归模型，通过盈余预测与股价变化率进行回归，探究了分析师盈利报告发布前股价涨跌对其盈利预测的影响，同时，还考察了熊市和牛市两种不同市场态势下的情况。在此基础上，本文进一步定义了外推偏差变量，以量化分析师的外推程度。我们将分析师的预测偏差与外推偏差进行回归，实证检验分析师外推对预测报告质量的影响。然后，将分析师的外推偏差与盈利公告发布日附近3天的累计异常收益率进行回归，观察股价波动情况，检验分析师外推预期对股票未来回报的影响。

本文通过实证研究发现，分析师在进行盈利预测时存在外推偏差：对于过去股价上涨的公司，分析师会以外推的方式高估其盈利能力，对于过去股价下跌的公司，分析师会以外推的方式低估其盈利能力，并且这种现象在牛市期间更加明显。由于分析师会不自觉地对传统的高收益公司带有积极情绪，对传统的低收益公司带有消极情绪，在进行盈利预测时，近期股价上涨公司的正面影响会被过分放大，近期股价下跌公司的负面影响会被过分放大，即分析师预测报告的误差会被放大，质量会有所下降。由于分析师对近期股价上涨公司的增长预测明显高于实际，对近期股价下跌公司的增长预测明显低于实际，随着盈余公告发布，公司的股票价格会发生反转，市场在意识到外推偏差后会进行修正调整，即分析师的这种外推预期会对公司未来股价产生负面影响。尤其值得注意的是，当市场中出现更多持有类似观点的分析师时，这种负面影响会被进一步放大。本文关于外推偏差的研究不仅丰富了对中国股票市场中分析师预测偏差的理解，还为投资者、证券分析师和监管部门提供了实践意义。通过揭示外推偏差的存在和影响，帮助投资者更理性地看待分析师预测报告，帮助证券分析师更公正地评估市场信息，帮助监管部门更规范地提升市场效率。

In academia, the influencing factors of analysts' forecasting behavior and the quality of their forecast reports have long been a widely discussed topic, with numerous domestic and international studies delving into this area. Building upon existing research, this paper expands and deepens the discourse by utilizing the representativeness heuristic from behavioral finance as its theoretical foundation. It focuses on investigating the cognitive biases present when analysts make profit forecasts within the context of the Chinese stock market and their correlation with forecast errors and future stock prices. Specifically, analysts' forecasts are not only influenced by the fundamental factors of the company but also by their emotional response to changes in stock prices, which subsequently affects their bias in forecasting future profits. This paper terms the behavior of analysts being influenced by past stock price changes when making profit forecasts as "analysts' extrapolation bias." Firstly, we examine whether analysts exhibit extrapolation bias, i.e., whether their forecasts are affected by recent historical stock price changes. Subsequently, we explore whether this extrapolation bias amplifies analysts' forecast errors, and finally, we analyze the impact of analysts' extrapolation expectations on future stock returns.

This study mainly focuses on 1,139 extensively tracked stocks in the A-share market from January 1, 2010, to December 31, 2022, as research subjects, investigating how individual analysts formulate expectations of future stock returns across various cross-sections. To delve further into this issue, the analysis is segmented into five trading days per period, introducing multi-period stock price change rates to reflect the volatility of stock prices over specific intervals. These rates serve as historical stock price change data, enabling the construction of a multivariate regression model to correlate earnings forecasts with stock price change rates. This model probes the impact of stock price fluctuations on analysts' profit forecasts preceding profit report releases. Additionally, scenarios in both bear and bull markets are explored. Building on this groundwork, an extrapolation bias variable is subsequently defined to quantify analysts' degree of extrapolation. Regression analyses are conducted to assess the influence of analysts' extrapolation on forecast report quality by correlating analysts' forecast errors with extrapolation biases. Furthermore, analysts' extrapolation biases are regressed against cumulative abnormal returns around profit announcement dates to scrutinize stock price fluctuations and evaluate the impact of analysts' extrapolation expectations on future stock returns.

Through empirical research, this study finds that analysts exhibit an extrapolation bias when making profit forecasts: they tend to overestimate the profit capabilities of companies whose stock prices have risen in the past and underestimate those of companies whose stock prices have fallen, with this phenomenon being more pronounced during bull markets. Due to analysts' unconscious positive sentiment towards traditionally high-profit companies and negative sentiment towards low-profit ones, the positive impact of recently rising stock prices on earnings forecasts of such companies is excessively magnified, while the negative impact of recent stock price declines is similarly exaggerated, leading to amplified errors in analysts' forecast reports and a decrease in forecast quality. As analysts' growth forecasts for recently rising stock companies are significantly higher than actual figures and forecasts for recently falling stock companies are significantly lower, stock prices of companies experience reversals upon earnings announcements. The market gradually adjusts after recognizing the extrapolation bias, resulting in negative effects on the future stock prices of companies. It is particularly noteworthy that when more analysts holding similar views emerge in the market, this negative impact is further magnified. This study enriches understanding of extrapolation bias in the Chinese stock market, providing practical significance for investors, securities analysts, and regulatory authorities. By revealing the existence and impact of extrapolation bias, this research helps investors view analysts' forecast reports more rationally, assists securities analysts in evaluating market information more fairly, and aids regulatory authorities in promoting market efficiency more effectively.