Morningstar - Q3 2020 - 35

the U.S. or the U.K. from 1926 to 2020, 1871
to 2020, or even (as Will Goetzmann and others
have done3) several centuries of history-
is almost certain to be an overestimate. The
very existence of a long data series is evidence
of greater-than-average survivorship. The
right way to forecast, if you believe the
future will equal the observable past (itself a
contentious assumption), is to gather data
on all the investment opportunities that existed
in the observable past and measure their
aggregate return, including the returns on the
opportunities that disappeared because of
war, revolution, or any other reason.
This is difficult because gaps in the historical
record do not allow us to know exactly what the
opportunity set was at any given point-that
is, we do not know what proportion of the entirety
of the past is observable. For example, data
from Dimson, Marsh, and Staunton, or DMS, have
started to include Chinese equities starting
in 1900 but not Chinese bonds.4 Yet a colleague of
ours has Chinese railroad bonds from that time
framed and hanging on the wall of her office.
We assume these are a total loss, because they
appear to be worth more as wallpaper than
as securities, and they did not make it into the
DMS estimates of the historical global bond
return. This is no fault of theirs; nobody has the
data, which were almost certainly destroyed
in the 1911 revolution, the 1949 revolution, or the
1976 "evolution."
Looking Only at Modern History
The bias from looking only at modern data, even
if you could include all countries whose markets
did not survive or whose markets went away
and then came back, is different and possibly even
more severe. At least we know how to adjust,
albeit imperfectly, for country survival: We go back
and try to include as many countries at the
start of the period as we can, knowing there will
be some left out (and knowing, more or less,

which ones those are). But with the modern-period
bias, there is an infinite number of ways the
economy could have evolved, or not evolved, or
become extinct.
We can call this the sample-of-one bias, in
reference to Paul Samuelson's famous comment
that "we have only one sample of the past."5
His concern, like ours, is that we should not
be misled by thinking it is the only past that could
have occurred.

There was capitalism, and ownership and trading
of joint-stock companies, before 1695-
quite a bit of it, actually. Some of the returns were
probably very good. But there was very little
broad-based economic growth; McCloskey, Joel
Mokyr, Gregory Clark, and others have made
a powerful case that most people in 1700 were not
much better off than they were in 700.6 They
lived just above subsistence level and faced the
continual threat of famine and disease.
So, if we are forecasting returns under the
assumption that stock market performance since
the Industrial Revolution is not only repeatable
but repeatable because the economic growth
rates of that period are repeatable, it makes no
sense at all to use our sparse stock market
data from before 1695 as an input to any forecast.
The data can be used to infer something about the
experience of a few investors whose records
survived for us to read them but nothing about

The oldest continuous stock price series is
three centuries old. (It is the price series for the
Bank of England starting in 1695.) There
is no assurance that the next three centuries
will in any way resemble the past three. One
thing we will definitely not do is go from
worldwide extreme poverty to modest affluence.
The reason is that our starting point, this time,
is modest affluence.

EXHIBIT 1

From Emerging Market to Economic Superpower In 1900, the U.S. made up 15% of
the world equity market, on par with Germany and France.
End-1899

Start-2020
60%

54.5
40

25.0
20
15.0

11.5
5.1

U.S.

U.K.

3.2

France

13.0

2.6

Germany

3.5

2.2

Australia

Source: Dimson, E., Marsh, P., & Staunton, M. 2020. Credit Suisse Global Investment Returns Yearbook 2020 (Zürich: Credit Suisse).

3 Goetzmann, W.N. 1993. "Patterns in Three Centuries of Stock Market Prices." Journal of Business, 66(2): 249-270. Mirowski, P. 1987. "What Do Markets Do? Efficiency Tests of the 18th-Century
London Stock Market." Explorations in Economic History, 24 (2), 107.
4 Dimson, E., Marsh, P., & Staunton, M. 2020. Credit Suisse Global Investment Returns Yearbook 2020 (Zürich: Credit Suisse).
5 Samuelson, P.A. 1994. "The Long-Term Case for Equities (and How It Can Be Oversold)." Journal of Portfolio Management (Fall).
6 McCloskey, op. cit. Mokyr, J. 2016. A Culture of Growth: The Origins of the Modern Economy (Princeton, N.J.: Princeton University Press). Clark, G. 2007. A Farewell to Alms: A Brief Economic
History of the World (Princeton, N.J.: Princeton University Press).

morningstar.com/lp/magazine

189

http://www.morningstar.com/lp/magazine

Morningstar - Q3 2020

Table of Contents for the Digital Edition of Morningstar - Q3 2020