According to data from Zacks Investment Research about analysts’
recommendations on some 6,000 companies, only 1.0% of recommendations
were “sells” in late 1999 (while 69.5% were “buys” and
29.9% were “holds”). This situation stands in striking contrast to
that indicated by previous data. Ten years earlier, the fraction of
sells, at 9.1%, was nine times higher.16
Analysts are now reluctant to recommend that investors sell
anything. One reason often given for this reluctance is that a sell
recommendation might incur the wrath of the company involved.
Companies can retaliate by refusing to talk with analysts whom
they view as submitting negative reports, excluding them from
information sessions, and not offering them access to key executives
as they prepare earnings forecasts. This situation represents
a change in the fundamental culture of the investment industry,
and in the tacit understanding that recommendations are as objective
as the analyst can make them.
Another reason that many analysts are reluctant to issue sell recommendations
is that an increasing number of them are employed
by firms that underwrite securities, and these firms do not want
their analysts to do anything that might jeopardize this lucrative
side of the business. Analysts affiliated with investment banks give
significantly more favorable recommendations on firms for which
their employer is the co- or lead underwriter than do unaffiliated
analysts, even though their earnings forecasts are not usually
stronger.17
Those who know the ropes realize that today’s hold recommendation
is more like the sell recommendation of yesteryear. According
to James Grant, a well-known market commentator, “Honesty
was never a profit center on Wall Street, but the brokers used to
keep up appearances. Now they have stopped pretending. More
than ever, securities research, as it is called, is a branch of sales.
Investor, beware.”18
Analysts’ recommendations have been transformed by something
analogous to grade inflation in our schools: C used to be an average
grade, yet now it is considered as bordering on failure. Many of us know that such inflation happens, and we try to correct for
it in interpreting our children’s grades. Similarly, in the market we
factor inflation into analysts’ recommendations. But not everyone
is going to make adequate corrections for analysts’ newly hyperbolic
language, and so the general effect of their changed standards
will be to encourage the higher valuation of stocks.
Moreover, it is not just a change in the units of measurement that
infects analysts’ reports. Even their quantitative forecasts of earnings
growth show an upward bias. According to a study by
Steven Sharpe of the Federal Reserve Board, analysts’ expectations
of growth in the S&P 500 earnings per share exceeded actual growth
in sixteen of the eighteen years between 1979 and 1996. The average
difference between the projected and actual growth rate of
earnings was 9 percentage points. The analysts breezed through
both the steep recession of 1980–81 and the recession of 1990–91
making forecasts of earnings growth in the 10% range.19
This bias in analysts’ forecasts is a characteristic of their one-year
forecasts; they are usually more sober in predicting the next earnings
announcement just before it is released. Analysts tend to
comply with firms’ wishes to see positive earnings surprises each
quarter, by issuing estimates that fall slightly short of the actual
number. Firms may, just before making earnings announcements,
talk with analysts whose forecasts are on the high side, urging them
down, while neglecting to talk with analysts whose forecasts are
on the low side, thereby creating a downward bias in the average
earnings forecast without being blatantly untruthful.20 Casual
evaluation of analysts’ forecasts by clients would most naturally
take the form of comparing the latest earnings announcement with
the latest forecast, and therefore analysts do not sharply overestimate
earnings just before they are announced, which would be
an obvious embarrassment to them.
Analysts’ upward bias comes to the fore in predicting the vague,
undifferentiated future, not immediate quarterly or yearly outcomes.
And it is expectations for the vague, undifferentiated future, even
far beyond one-year forecasts, that lie behind the high market
valuations we see. Analysts have few worries about being uniformly
optimistic regarding the distant future; they have concluded that such generalized optimism is simply good for business. Certainly
they perceive that their fellow analysts are demonstrating such longrun
optimism, and there is, after all, safety in numbers. Glibly and
routinely offering “great-outlook-for-the-U.S.” patter to the investing
public, they perhaps give little thought to its accuracy.