Analysing the Analysts

Investors depend on security analysts to make forecasts and recommendations on stocks. SMU Assistant Professor Roger Loh looks at how accurate analysts’ predictions are and how exactly investors should interpret their reports.

Photo Credit: Daniel Tham

By Rebecca Tan

SMU Office of Research (21 Mar 2014) – Equity prices are driven by security analysts who forecast projected earnings and recommend whether to buy or sell a company’s stock. Investors, whether retail investors investing their own money or institutional investors like hedge funds, rely on these valuation reports to make informed decisions.

What are some of the factors that influence the accuracy of analysts’ reports, and how should investors use them to make decisions? These are the two key questions Assistant Professor Roger Loh of the Singapore Management University (SMU) Lee Kong Chian School of Business addresses in his research on empirical asset pricing and security analysts.

To answer these questions, Professor Loh utilised the large amount of data available in analysts’ reports and linked them to companies’ stock performances. By focusing on the behaviour of security analysts, he is able to go beyond widely available aggregate data such as stock price and volume.

In a recent study that was featured in Bloomberg and The Economist, Professor Loh found that although the accuracy of analysts’ predictions was less accurate during economic crises, they were paradoxically taken more seriously by investors and had a larger impact on stock prices.

“It may seem puzzling that advice based on poorer information is more influential, but if you think of it in another way, investors know even less than analysts in times of uncertainty. So even though the signal provided by analysts is noisy, investors will react to the information because it is still more than what they know. The analogy is that if it is foggy you will pay more attention to the map, even if the map isn’t very good,” he explains.

Crises can increase the inaccuracy of forecasting by as much as 10 to 20% , says Professor Loh, and the error can be even higher for firms that are strongly affected by the crisis, such as financial firms.

Educating investors to interpret analyst reports

Investors should also be aware that there are inherent biases in analysts’ recommendations, Professor Loh cautions. He notes that recommendation categories are coarse; there are only five types of recommendations that analysts make: strong sell, sell, hold, buy and strong buy. However, how their recommendations are distributed across these five categories is uneven.

“If you count all the stocks that have buy and strong buy recommendations, they would make up 70% of all stocks. Hold recommendations comprise about 25%, and those that have sell or strong sell recommendations less than 5%. Such a distribution would indicate that the recommendation is not very useful. If every recommendation is ‘buy’, it cannot be that I should buy every stock,” he says.

This implies that investors should discount the optimism inherent in the forecasts, Professor Loh says. “If an analyst says buy, I shouldn’t necessarily buy it; sell is very rare, so if I see hold I should actually sell. However, retail investors will respond more literally to the recommendation, which is incorrect.”

Another issue that investors should be aware of is biases in their own decision making process. For example, people tend to assume that more volatile stocks will earn higher returns than more stable stocks. However, the reverse is true; stocks with higher volatility earn lower returns on average, Professor Loh explains.

“What is the reason for that? A major factor is lottery preference. Consider the penny stocks that caused a problem in Singapore last year: stocks can move from one cent to two dollars and then back to two cents. If there are 100 stocks that are two cents, then maybe one of them would move to two dollars, but all the rest of them will go to zero. So if I pick one of them randomly, then the returns will be low on average because I have not observed them going to two dollars yet.

“But people are willing to take this poor bet because it gives them a small chance of a big gain. This actually destroys a person’s wealth. If you’re really so lucky as to choose the right stock and it goes to two dollars, and you manage sell at two dollars, then you will make a lot of money. But the probability of that happening is quite low, so you should actually avoid this profile of stocks,” Professor Loh says.

Synergy with social sciences

Professor Loh intends to further his research by looking at other factors that influence investor decisions. His preliminary data indicates that investors should randomise their decisions if two choices are very similar, but in practice they do just the opposite.

“For example when choosing between telecom company A and B, which have almost identical plans, you try to figure out which one is better, down to how many minutes you use, etc. when you should actually randomise because the difference is very small. But when it comes to whether I should invest $50,000 in SingTel, people say: ‘My friend told me to invest, so I do it’ and don’t do any research on the company,” he says.

Many of his existing collaborations are with colleagues at the Lee Kong Chian School of Business. However, Professor Loh notes that his work shares some overlap with the social sciences. “I think there is definitely synergy there. A lot of the behaviour at the economic agent level sounds like social psychology, except that we look at the impact on actual stock prices rather than in an experimental setting. If you have a purely experimental paper in finance, it is going to be quite difficult to publicise or sell to journals,” he says. “But I think the field is becoming more accepting to this type of research.”

Finally, he intends to do further work on analyst behaviour, going beyond the numbers and looking at other types of information. “I think there are a lot of unanswered questions that we can address with new information. We are doing a textual analysis, looking at the tone of the reports and seeing if it has an impact. Of course the reports also have hard information that is already coded in the database, but there are things that are not recorded that also have implications on stock prices,” he says.