Earnings forecasts are an important source of information for security valuation. The accuracy of analyst forecasts obviously has important practical consequences for portfolio performance, and the authors contribute to the understanding of the efficiency of information dissemination and pricing of securities markets.

Generally issued on behalf of brokers, analysts’ forecasts are widely used by fund managers to make portfolio allocation decisions. Most empirical studies use analysts’ consensus forecasts as a proxy for the market’s expectation of future earnings and are predicated on the assumption that such forecasts are unbiased and efficient. Analysts are supposed to provide investors with accurate and truthful research reports, but many studies have demonstrated that analyst forecasts are upwardly biased and inaccurately reflect available information.

The authors attempt to determine how important corporate affiliations, investment banking, and compensation conflicts are when explaining analysts’ research optimism—relative to the portion of error attributable to the incomplete rationality of analysts. Mutual funds and commission business seem to be more important businesses than underwriting IPOs, so analysts write their reports with these audiences in mind.

Analysts tend to have optimistic forecasts to please their clients, which are largely long-only mutual funds that could underperform when downgrades are issued. The authors’ central premise is that analysts affected by sentiment unwittingly amplify these shocks and move prices away from fundamental values. Sell-side analysts represent a subset of sophisticated market participants that should be the most informed about a stock’s fundamentals, so they provide a lower bound to the effect that sentiment might have on non-professional, less informed investors.

Most academic studies are based on developed markets, but this study is one of the first to provide direct evidence of the role of investor sentiment in explaining forecast errors in an emerging economy. China is the world’s largest emerging and transitional economy. Furthermore, it is associated with a poor information environment and complicated conflicts of interest.

There are two types of explanations for the forecast bias—one based on irrationality and the other on rationality. The goal of this research is to unravel and control for the effect of rational and irrational factors to determine which impose a greater influence on analysts’ forecast bias when both factors are present. Investor sentiment is considered an unintentional irrational factor, whereas a conflict of interest is a rational bias.

The analysts’ earnings forecast bias measures the gap between predicted and actual earnings after adjustments for nonrecurring gains and losses and dilution by share issuance. To measure the irrational factors, the authors construct their investor sentiment index with six basic components: mainboard turnover of A shares in Shanghai and Shenzhen, new account, live account, finance increase, bank inflow, and IPO first-day return. To identify rational factors, the analysts’ characteristics and fee affiliations are categorized into four classes—commission, conflicts with affiliated funds, analyst reputation, and connections to underwriters—in an effort to identify and measure the strength of the bias. The model enables the authors to test for interaction and to quantify the magnitude of each bias.

Using a multivariate regression that controls for these conflicts of interest, the authors show that sentiment significantly affects the forecast bias.

The authors’ findings suggest that analysts could be important information intermediaries in the price momentum effect that poses one of the greatest challenges to the semi-strong-form efficient market hypothesis. The optimism observed in US equity research during the tech boom of the late 1990s raised a number of important research questions about the sources of analyst forecast error.

Regulators’ focus on investment bank analysts suggests that they believe banking conflicts were the primary source of research optimism. Regulatory responses like Regulation FD and the Global Research Analyst Settlement that diminished analysts’ motives to inflate their forecasts had a meaningful impact on analyst behavior: The mean forecast bias declined significantly, and the median forecast bias essentially disappeared.