Investment banks and stockbrokers employ an army of analysts to research equities and issue “buy” or “sell” tips on individual stocks or sectors. In the past, this “sell-side research” (as opposed to “buy-side research”) has been given “free” to investors who use the services of the bank or broker (in other words, the cost is bundled up with the other services the client uses). But new rules are forcing banks to separate (“unbundle”) research costs from trading costs. Critics claim this will reduce the funds available for research, which will, they argue, make markets less efficient.
However, would a collapse in the volume of sell-side research matter? The evidence doesn’t look good for the analysts. For a start, they are poor stock pickers. Between 1993 and 2002, the average stock tipped by the largest 15 brokers in the US underperformed the S&P 500 index, according to J Randall Woolridge of Pennsylvania State University. And a more recent, UK-focused study by AJ Bell found that last year the ten FTSE 350 stocks that were most highly rated by analysts returned just 2.1% on average, far less than the 12.5% return on the index. Worse still, the ten most-hated stocks made an average of 56.2%.
Even the studies that found evidence of stock-tipping ability had other problems. In 2001, Brad Barber of the University of California, Davis found that a portfolio of the most heavily tipped shares between 1986 and 1996 would have beaten the market. But to keep up with the ever-changing recommendations, you would have had to trade so often that your costs would have wiped out any gains over and above investing in an index fund.
Can analysts do anything right? Some studies suggest that they are reasonably good at forecasting future earnings. Looking at the period from 2000 to 2010, Kenneth Lorek of Northern Arizona University found that analysts’ earnings forecasts were significantly better than those produced by a “random walk” model (ie, by chance). A 2014 study by Pawel Bilinski of Cass Business School had a similar finding for forecasts of dividend payouts.
Unfortunately for the analysts, this forecasting ability only holds for larger companies. Mark Bradshaw of Boston College found that a random-walk model beat the average analyst forecast for smaller firms. More to the point, Amy Hutton of Boston College suggests that management earnings forecasts are just as accurate as those of analysts, rendering the latter redundant. In short, if they didn’t give sell-side analysis away, it’s hard to see why you would pay for it. Bad news for analysts – but for investors, it just means fewer distractions.