證券分析師股票推薦和盈餘預測行為面課題之研究

Abstract

本論文探討分析師行為面課題，以其所發佈之股票推薦與盈餘預測應具資訊內涵。本研究考量股票特性及行為偏誤，以探索分析師對於資訊處理及分析的能力。本篇論文之兩部分分別研究：為何在精明市場參與者主導的市場上，價值策略依舊可行？以及：證券分析師是否對復甦中的股票未即時做推薦？ 論文第一部份主要在辨明是否分析師對於明星股以及價值股有不同的分析能力。以「分析師推薦」作為觀察變數，我們發現分析師在明星股群組有較強的擇股能力，此組在三因子模型分析後，應有可能產生較高異常報酬的系統性交易策略，同時，分析師對明星股的盈餘預測也有較佳績效。此種系統性差異可能源自於分析師對明星股的評價專業能力較佳，或源自於價值股的會計一致性較低。實證顯示，控制三年的盈餘波動後，分析師對明星股之推薦及盈餘預測績效仍優於價值股。本研究結果穩健呈現：分析師對明星股有較優異的專業分析能力。 第二部分我們檢驗是否分析師在新資訊與過去期望不一致時，會延後發佈股票推薦。根據心理學的認知失調與保守主義理論，我們推測並發現：當後續證據與過去觀點不一致時，分析師可能產生認知失調。若將重複發佈負面推薦樣本做為基準，我們發現分析師對潛在有認知失調群組會更保守，及其對於前次發佈負面推薦的股票，會延遲發佈向上修正的意見。過去經歷過由正面推薦向下修正意見的資產組合，對新資訊反應不足的現象更為明顯。顯示先前推薦行為的挫折經驗所造成的保守行為，會加重其對後續新資訊反應不足的現象。本研究亦排除復甦中股票具高報酬波動性，以及分析師對股票偏好的潛在解釋。研究結果顯示，分析師對復甦中的股票需花費較長的時間來進行向上的推薦修正，此結果與反向策略之有效性相一致。

Security analysts’ recommendations and earnings forecasts convey rich information. This study investigates analysts’ ability of information processing and analyzing concerning various stock characteristics and behavioral biases. In the first parts of the thesis, we consider the puzzle: why value strategies still work when the market is dominated by sophisticated participants. In the next part, we examine: do security analysts underreact in generating recommendations for recovering stocks? The purpose of the first part is to investigate whether analysts perform differently when processing information concerning glamour compared with value stocks. We find that analyst recommendations contribute more (less) to the selection of glamour (value) stocks. Our results consistently indicate that the glamour stocks with analyst’s favorable ratings yield significantly higher abnormal return than value stocks with all kinds of ratings after control size, book-to-market, and beta. Analyst’s better expertise in valuing glamour stocks is not only in recommendations but also in earnings forecasts. The systematic bias of analyst inferior performance of recommending and forecasting in value portfolio may be owing to analyst’s differential expertise in evaluating companies or to less accounting persistency which is proxied by earnings predictability in value stocks. Our evidence shows that the results are robust after controlling three-year earnings volatility. In short, analysts have superior expertise in analyzing glamour stocks. In the second part, we examine whether the incorporation of new information that is inconsistent with prior analyst expectations leads analysts to delay the release of their new and changed recommendation. Our study is based on two phenomena documented in psychology: cognitive dissonance and conservatism. We posit and find that analysts may be subject to cognitive dissonance and underreact to new information when the evidence is inconsistent with one’s prior perception. Adopting stocks with reiterated unfavorable recommendations as the benchmarks, we show that analysts are more conservative and delay upgrades for the unfavorable category firms. Analyst underreaction appears to be more pronounced for the stocks been downgraded from the favorable category than those with reiterated unfavorable recommendations. Our finding shows that analyst underreaction to new information may be compounded by the conservatism resulted from previous disappointments. We further exclude the potential explanations that the delay may be driven by greater volatility of recovering stock returns and difference in analyst preference of coverage. The documented significantly longer duration for the recovering stocks is consistent with the effectiveness of contrarian investment strategies.