INTERVIEW
Disproving The Efficient Market Theory
Samuel Eisenstadt Of Value Line
by David Penn and John Sweeney
For more than 50 years, Samuel Eisenstadt has been one of the major
reasons for the phenomenal success of Value Line, Inc., developers of the
Value Line Ranking System. Eisenstadt is research chairman, senior vice
president, and a director of the company, and his primary areas of responsibility
include the application of various quantitative methods to securities valuation
and forecasting. A frequent lecturer on Value Line techniques to business
school audiences and financial analyst societies, Eisenstadt has successfully
debated prominent efficient market proponents and was a pioneer in challenging
their thesis that the markets could not be beaten.
Staff Writer David Penn interviewed Eisenstadt on October 16, 2000.
This interview first ran in the January/February 2001 issue of Working
Money, The Investors' Magazine, our companion publication. On February
14, 2001, STOCKS & COMMODITIES Interim Editor John Sweeney updated
the interview.
ILLUSTRATION BY CARMELO BLANDINO
How long have you been at Value Line?
Oh, forever. I've been with Value Line since May 1946.
Value Line was founded in 1931, when the attitude toward the market
was quite a bit different than it is today. How would you say Value Line
has changed since then?
It has become a lot more quantitative. We've introduced statistical
methods that were not used in the early years, such as multiple regressions
and cross-sectional analysis. As a whole, it's much more mathematical than
it ever was.
Was your background in mathematics before you started at Value Line?
My degree was in statistics. At Value Line, I had the opportunity to apply
the mathematics I learned in college. The data was voluminous, and the
problem such that it looked like mathematics could help.
Were your statistical studies intended to prepare you for a career
in finance?
Oh, no. I majored in statistics because I had an affinity for mathematics.
What I was going to do with that specialty, I was not even sure myself.
When I got out of school, I went into the service. I got out at the end
of 1945 and was employed by Value Line in the middle of 1946. I got into
the finance field by pure happenstance. I literally started as a proofreader;
that was my initial exposure to this field. As I got exposed to it, I saw
the abundance of numbers and the applicability of statistical techniques
to analyze those numbers. We started to apply schoolbook techniques. By
"schoolbook," I mean regression analysis, which had not been
used before then in the company.
I've always been curious about the name "Value Line."
The name goes back to the 1930s, when Arnold Bernhard started the company.
In the early years, he was making a visual attempt to measure value by
constructing a line and comparing it with the price history of the stock.
When the price history was below the value line, which was constructed
from earnings, that indicated undervalued or, conversely, overvalued.
What was the goal?
It was an attempt to establish a discipline. A very early attempt. Almost
a pioneering attempt. The problem was that in the fitting process, it had
to be more judgmental. There were quantitative ways of fitting that so
that two people would come to the same conclusion. This is really what
is known as regression analysis. So we started to do that. Instead
of a visual value line, we developed a mathematical value line. And we
lived with that procedure from 1946 until about 1965.
What happened in 1965?
We made a radical departure, largely at my suggestion. Instead of looking
at individual stocks and doing each one as a separate analysis with a separate
formula and a separate fitting, we thought maybe we ought to develop one
regression analysis that covered all stocks. Then apply the same formula
to all stocks and, instead of looking at them over time, look at them at
a single point in time. This is known in statistics as cross-sectional
analysis.
What had you been doing up to then?
Up to then, we had been doing time series analysis, over a long period.
This new way gave us relative values. It told us which stocks were attractive
and which ones were less attractive at a point in time. That's really what
we were trying to do. As soon as we switched over to this new procedure,
the results took a sharp turn for the better. And we have essentially been
living with that approach since. We've added bells and whistles as we've
gone along, but the cross-sectional procedure was the radical departure.
Before 1965 you were regressing what against what? And after 1965,
what against what?
Prior to 1965 our formulas were based on time series analysis, with variables
such as annual earnings, dividends, book values, lagged prices, and so
on. Multiple regression analysis was used and each stock had its own formula.
These regressions were designed to predict next year's average price of
a stock, based on projections of the above variables.
After 1965, cross-sectional analysis was introduced. The variables were
10-year growth in relative annual earnings, prices, earnings momentum,
earnings surprise, and price volatility.
Stocks are ranked sequentially (no longer absolute price forecasts)
and grouped from one to five, with one being the best. Of the 1,700 stocks
in the Value Line Investment Survey, 100 are ranked in group 1; 300 in
group 2; 900 in group 3; 300 in group 4; and 100 in group 5. Rankings are
based on known information -- no forecasts required.
What got you thinking about moving away from time series to cross-sectional
analysis?
Mostly the desire to get better results. We were getting decent results
with the time series, but when you're looking across a long period and
doing an analysis, you're picking years. You're picking years of overvaluation
and undervaluation. You're going to say, "Oh, maybe in 1946 it was
overpriced, but in 1953 it was underpriced." That wasn't our
main problem. We're looking for stocks that are overvalued and undervalued.
So the observations ought to be over stocks, rather than over years. And
if your observations are over stocks, you're talking cross-sectional analysis.
That solves the problem. Then it says, the best stocks are these and the
worst stocks are these.
...Continued in the May 2001 issue of Technical
Analysis of STOCKS & COMMODITIES
I'm sure you're acquainted with the efficient market
hypothesis, which argues that it's impossible to make meaningful predictions.
Our ranking system has proved otherwise. -- Samuel Eisenstadt
Excerpted from an article originally published
in the May 2001 issue of Technical Analysis of STOCKS & COMMODITIES
magazine. All rights reserved. © Copyright 2001, Technical Analysis,
Inc.
