New York - We have talked at length about the impact that
analysts can have on stock prices and what particular remedies to prescribe for
this problem with respect to performance measurement of analyst's recommendations. Recently, we have seen an extreme example of the potential
for an analyst to affect stock prices. Performance measurement which denies, or
at least diminishes the analyst’s ability to affect prices is a key part of a
performance evaluation system. Still the old adage continues to ring true
"good research may not be consistent with good performance".
The most recent example of analyst impact is the case
Citigroup analyst Bhatia, who called for E*trade to experience a run on its
banking operations and potentially declare bankruptcy. Other analysts covering
the stock were either in the hold category (66%) or the buy category (33%) and
there was apparently no immanent crisis facing E*Trade. In general, analyst
impact is a function of the volatility (uncertainty) in the market and the
level of surprise that the recommendation provides.
From a performance standpoint, Mr. Bhatia was the only
analyst that got the call right. From an investor perspective, Mr. Bhatia
destroyed shareholder wealth. E*Trade's shares fell by 59% on Monday, even though the
recommendation was unfounded. But the topic of this blog is neither the integrity
nor the competence of Mr. Bhatia. Rather we are interested in this case as it
relates to performance measurement.
If the call were logged prior to the price move, the analyst
would have looked psychic in his ability to time his recommendations. Further,
we would have logged a 59% improvement in his overall performance. This assumes
that the overall performance is measured by the synthetic returns that would
have been achieved if an investor had invested in and shorted or sold all the
analyst's recommendations over some period of time.
One model of coping with analyst impact in performance
analysis is to lag the recommendation. Some rules lag the recommendation to the
close of business, unless it is after 3PM, after which it is awarded the
following day's closing price. Others use the following day's closing price
regardless. For Mr. Bhatia, a one day lag in the selection of the price used to
open the short position would have resulted in a 41% loss on that recommendation.
Of course, the loss is more consistent with the timing that
even a savvy retail account could have reasonably acted upon the recommendation. Indeed,
one could argue that it is the retail account that the performance measurement
systems were designed to protect in the first place.
In the final analysis, the object of performance measurement
is to find good research. However, in the case of the Citi analyst, performance
measurement and good research diverged radically. And while it is clearly not
feasible for a performance measurement system to be concerned with the
distinction of being "right for the wrong reason, or wrong for the right
reason" this is the Achilles’ heel of performance
measurement as it relates to analyst impact.
