Morningstar - Q3 2020 - 38

Strategies

EXHIBIT 4

German Extremes Annual returns in Germany were wildly volatile, the result of
losing two world wars and experiencing periods of hyperinflation and depression.

Survivorship Bias in
Fund Data

Cumulative Index of Real Equity Returns of Germany, 1900-2019 (Dec. 31, 1899 = 1.00)
60%

U.S. Dollar

Survivorship bias is also a problem when
evaluating investment managers. Managers shut
down poor-performing funds, sometimes moving
40
their assets into larger or better-performing
funds where the bad record won't be seen.
Hedge funds do this even more than long-only
managers. Thus, unless survivorship is accounted
for, average active manager performance
20
looks better than it really is, as we have already
13.0 with countries.
seen

3.2

2.6

e

Germany

3.5

2.2

Local Currency
100.00

Logarithmic Scale

56.63
50.35

10.00

1.00

0

Australia
0.10

epoch had faced different evolutionary pressures,
a species that developed big brains and an
upright posture might have died out and we would
not be here. In a similar fashion, we are studying
the stock and bond markets as a way to
accumulate assets because we happen to live
in a time and place where that is possible.
There is no law of nature saying that is always
and everywhere true.

Portugal
any

35

Compound Annual Return
The Future Will Be Different
8%
Survivorship bias and sample-of-one bias are
separate but related. Survivorship bias
almost certainly overstates the return that was
earned over the whole opportunity set of
6
investments in our one "run" of the past. Therefore,
analysts using that run of the past to forecast
the future should consider adjusting their
4
estimates downward.
That is not an ironclad rule because the use
of past data to forecast the future mixes
2
several effects: (1) the continuity or stability of
basic economic relationships, such as the return
0
required as a reward for risk, and between the
40
45
50
past and the future; (2) the amount of risk in the
future compared with the past; (3) survivorship
bias, as discussed above; and (4) institutional
changes, such as the dramatic lowering of investor
costs due to innovations such as the issuance of
shares to the public, open-outcry trading, mutual

38

Morningstar Q3 2020

0.01
1899

1919

1939

1959

1979

1999

2019

Sources: Dimson, Marsh, and Staunton (op. cit.), Morningstar Direct. Data as 12/31/2019.

funds, and in the modern era, index funds and
exchange-traded funds.
Sample-of-one bias is harder to adjust for because
we cannot know the characteristics of the
"population" from which the sample of one was
drawn. It is like speculating what you would
be like if you had never been born and someone
else had been born in your place. Most
children ask this question at some point and get
a profoundly unsatisfying answer, something like,
"You wouldn't exist, so you wouldn't care.
You would have a brother or sister who would
care." That may be unkind, but it is correct.
So, the resolution to the sample-of-one bias is
not an adjustment upward or downward-
it is indeed possible that the future will be better
than the past 200 or 300 years, just as it might
be worse. Instead, it is having modesty about the
accuracy of the forecasts. While "future
equals past" may be a good first approximation,
it comes with wide error bars.

But we can be sure of one thing. While the future
may resemble the past in meaningful ways-
we will eat, sleep, and make merry the way we
always have, because we inhabit human bodies
and minds-it will also be very different. K
Paul D. Kaplan, Ph.D., CFA, is director of research with
Morningstar Canada. He is a member of the editorial board
of Morningstar magazine.
Laurence B. Siegel, the author of Fewer, Richer, Greener:
Prospects for Humanity in an Age of Abundance, is
the Gary P. Brinson director of research at the CFA Institute
Research Foundation and an independent consultant.
His website is http://www.larrysiegel.org.
http://www.larrysiegel.org
Morningstar - Q3 2020

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