Read an Excerpt
1
Who Could Design
a Brain . . .
Alfred Marshall, the great Victorian economist, opens his Principles
of Economics with these words:
Economics . . . examines that part of individual and social action
which is most closely connected with the attainment and with the
use of the material requisites of wellbeing. Thus it is on the one side
a study of wealth; and, on the other, and more important side, a part
of the study of man.
Marshall's Principles were to set the tone of economics for the next
half century. In this work, despite his noble words in the quotation
above, he made the study of man secondary to the study of wealth.
Under all conditions, man in classical economics is an automaton capable
of objective reasoning. Furthermore, disagreement about the future—
a fundamental feature of the study of man—has no place in this
particular study of wealth. Marshall's approach was finally dislodged,
with great difficulty and after many years of dispute, by the publication
in 1936 of his student John Maynard Keynes's masterwork, The General
Theory of Employment, Interest, and Money.
The bundle of ideas, models, concepts, and systems embodied in the
theoretical structure of modern finance—what I describe as Capital
Ideas—appeared between 1952 and 1973. They owe little to Keynes
and almost everything to Marshall. The entire underlying structure of
Capital Ideas rests on one overriding assumption: Investors have no difficulty
in making optimal choices in the bewildering jumble of facts,
rumors, discontinuities, vagueness, and black uncertainty that make up
the real world around us.
Over time, this tension between an ideal concept of human rationality
and the coarse reality of our daily lives has become an increasingly
contentious issue. How much do we know about how people in the real
world arrive at decisions and make choices? How great are the differences
between the theoretical assumptions and the real world? And do
those differences matter?
Although these questions have always been central to understanding
the way investors behave and how their responses affect the performance
of financial markets, no one made any systematic effort to provide
the answers until the mid-1960s. The most significant and
inf luential effort to approach these problems, a field of study that has
come to be known as Behavioral Finance, began to take shape quite by
accident when two junior psychology professors at Hebrew University
in Jerusalem, Daniel Kahneman and Amos Tversky, happened to compare
notes one day about their work and their life experiences. The
hugely productive result of their friendship and subsequent collaboration
has created a competing vision to the rational model of how people
make choices and reach decisions under conditions of uncertainty.**
The essence of this work is the study of man—of human behavior.
As Kahneman and Tversky wrote in 1992: "Theories of choice are
at best approximate and incomplete. . . . Choice is a constructive and
contingent process. When faced with a complex problem, people . . .
use computational shortcuts and editing operations."1 The result is a
decision-making process differing in many aspects from the assumptions
of Capital Ideas.
It would be a mistake to accuse Kahneman and Tversky of tarring
all humanity with the black brush of irrationality. That was never the
case, as Kahneman's autobiography makes clear: "The interpretation of
our work as a broad attack on human rationality rather than a critique
of the rational-agent model attracted much opposition [to our efforts],
some quite harsh and dismissive."2 As Kahneman put the point to me,
"The failure in the rational model is . . . in the human brain it requires.
Who could design a brain that could perform in the way this model
mandates? Every single one of us would have to know and understand
everything, completely, and at once."† He expresses this position even
more precisely in writing:
I am now quick to reject any description of our work as demonstrating
human irrationality. When the occasion arises, I carefully explain
that research on heuristics and biases only refutes an unrealistic
conception of rationality, which identifies it as comprehensive coherence.
. . . In my current view, the study of judgment biases requires
attention to the interplay between intuitive and ref lective
thinking, which sometimes allows biased judgments and sometimes
overrides or corrects them.3
* * *
Kahneman's and Tversky's published papers, both individually and
jointly, constitute an imposing compendium of evidence, ideas, and axioms
of human behavior in the process of decision making. One of the
most interesting features of Kahneman's and Tversky's work is the innovative
nature of their discoveries. The patterns of human nature they
discuss must have existed since the beginning of time, but no one before
them had caught their vision. They unleashed a far larger flood of
research from other academics and, over time, from the practitioner
side as well.
In highly compressed fashion, the rest of this chapter conducts a
survey of Behavioral Finance based on a small but characteristic sample
of these investigations. The implications of this survey for investment
are fascinating, but along the way the material also provides a mirror in
which we see ourselves probably more often than we would like.
The real issue is this: How much damage has this attack inf licted
on the standard theories and models of f inance? Do the critique of the
rational-agent model and the demonstrations of its empirical failures
render my book, Capital Ideas, useless and at best obsolete? Or, in a
more practical mode, do the teachings of Behavioral Finance lead us to
alpha—to an excess return on our investments after adjustment for risk?
Final judgment must await the presentation of the evidence. But
final judgment will be rendered.
Before moving on, a separate point is worth making. The focus of
the discussion so far has been on how the findings of Behavioral Finance
relate to each of us as an investor. But a deeper issue is also involved,
set forth by John Campbell of the Economics Department at
Harvard in his presidential address to the American Finance Association
in January 2006:
Even if asset prices are set efficiently, investment mistakes can
have large welfare costs for households. . . . They may greatly reduce
the welfare gains that can be realized from the current period
of financial innovation. . . . If household finance can achieve good
understanding of the sources of investment mistakes, it may be
possible for the field to contribute ideas to limit the costs of these
mistakes.‡
* * *
A story that Kahneman recounted in the course of his address accepting
the Nobel Prize provides a typical example of the "computational
shortcuts and editing operations" we use in our attempts to make
choices in complex problems. Kahneman had conducted an experiment
with two different audiences. Although he offered both audiences an
identical set of choices, he presented these choices in different settings
that led to strikingly different results.
He asked each audience to imagine a community preparing for the
outbreak of a dreaded disease. The experts have predicted the disease
will kill 600 people if nothing is done, but they offer two different programs
to deal with the contingencies.
Under Program A, 200 people will be saved. Under Program B,
there is a one-third possibility that all 600 people will be saved and a
two-thirds' probability that everybody will die. Kahneman found that
the audience presented with these choices overwhelmingly favored Program
A, on the basis that the gamble in Program B was too risky. The
certainty that 200 people would be saved was preferable to a two thirds'
chance that everybody will die.
Then Kahneman presented the identical choices to the other audience,
but in a revised setting. Under Plan C, 400 people will die. Under
Plan D, there is a one-third chance that nobody will die and a two thirds'
probability that 600 people will die. Now the audience's choice
was for Plan D. The gamble, in its Plan D garb, now seems preferable
to Plan C, in which it is certain 400 people will die.
How can we account for these opposing sets of responses to what
are identical choices and probabilities? As Kahneman explains it, nobody
has ever figured out a perfect technique for dealing with uncertainty.
Consequently, in making choices and decisions, we tend to
overweight certain outcomes relative to uncertain outcomes, even when
the uncertain outcomes have a high probability. In the case of the first
audience, the certainty of saving 200 out of 600 people is "disproportionately
attractive." In the case of the second audience, accepting the
certain death of 400 out of 600 people is "disproportionately aversive."
Kahneman and Tversky have defined these kinds of inconsistencies
in decision making as "failure of invariance." The failure of invariance
comes in many colors, with endless variations of the theme.* Invariance
means that if A is preferred to B and B is preferred to C, rational people
should prefer A to C. In the case above, if the rational decision in
the first set is 200 lives saved for certain, saving 200 lives for certain
should be the rational decision in the second set as well.
Kahneman and Tversky use the expression, "framing," to describe
these kinds of failures of invariance, which are widely prevalent. In the
example of the outbreak of the dreaded disease, the audience in the first
case framed their responses around how many people would live, while
the second audience framed their responses around how many people
might die. Kahneman's Nobel address defines framing as "the passive
acceptance of the formulation given." And then he adds, "Invariance
cannot be achieved by a finite mind."4
* * *
Richard Thaler of the University of Chicago, one of Kahneman's
and Tversky's earliest and most articulate disciples, describes an amusing
example of the failure of invariance involving money. Thaler proposed
to students in one of his classes that they had just won $30. Now
they could choose between two outcomes: a coin f lip where the individual
would win $9 on heads or lose $9 on tails, or no f lip of the coin
at all. The coin f lip was the choice of 70 percent of the students. When
his next class came along, Thaler asked the students to assume that they
had a starting wealth of zero. Now they could choose between these
two options. The first was a coin f lip where the individual wins $39 on
heads and $21 on tails. The second was $30 for certain. Only 43 percent
of the students chose the coin f lip; the majority preferred the $30 for
certain.
When you study the options offered to both classes, you will find
that the payoffs are identical. Whether the starting wealth is $30 or zero,
the students in both cases are going to end up with either $39 or $21
versus ending up with $30 for sure. Yet the majorities of the two classes
made entirely different choices, resulting in a failure of invariance.
Thaler ascribes this inconsistency to what he calls "the house
money effect." If you have money in your pocket, you will choose the
gamble. If you have no money in your pocket, you would rather have
the $30 for certain than take the risk of ending up with $21.5
In the real world, the house money effect matters. Investors who
are already wealthy are willing to take significant risks because they can
absorb the losses, while investors with limited means will invest conservatively
because of fear they cannot afford to lose the little they have.
This is precisely the opposite of how people with different wealth levels
should arrive at decisions. The wealthy investor is already wealthy and
does not need to take the gamble. If investors with only a small amount
of savings lose it all, this would probably make little difference, but a
killing on the small accumulation could change their lives.
Another investment-oriented version of the distortions caused by
framing resulted in an experiment conducted in 2001 by Thaler and his
frequent coauthor Shlomo Benartzi of UCLA.6 Participants were divided
into three separate groups with no contact among the groups.
Each group was given a choice of two fund offerings for their retirement
plans. One group was offered a fund holding just stocks and a
fund holding just bonds. The second group was offered a fund holding
just stocks and a balanced fund that includes stocks and bonds. The third
group was offered a bond fund and a balanced fund.
Even though these choices were for retirement funds that should
have had roughly the same asset allocation decisions, the three groups
ended up with wide differences in portfolio structures. The differences
arose because the 50-50 choice is always popular: It seems like common
sense; it looks like diversification; and it avoids the complex decision
about how assets should be allocated in a retirement fund. The consequences
were dramatic. The first group, choosing between a stock fund
and a bond fund, ended up with an average allocation of 54 percent to
equities. The second group, offered a stock fund and a balanced fund,
also leaned in the 50-50 direction between the two funds, but ended
up with an average allocation of 73 percent to equities and only 27 percent
to bonds, because half the balanced fund was already invested in
equities. The third group, offered a bond fund and a balanced fund,
ended up with an average of 65 percent in bonds and only 35 percent in
equities.
The experiment demonstrates that framing determined the decision
making among the three groups. The proper approach should have been
to consider the different expected rates of return and risks of each asset
class and to see through to the underlying structure of the balanced
fund in making the final choice. Fifty percent to each asset class might
not have been optimal, but it would have been a sensible choice for
someone with no experience or no understanding of the different risk return
trade-offs between stocks and bonds. In fact, however, the design
of the offering dominated. Most of the participants were unwilling
to make the intellectual effort to see through the 50-50 allocation of
the balanced fund and recognize that the true asset allocation was a long
way from 50-50.
This experiment was not just an artificial effort to find out how
people make choices where framing is likely to dominate. The 50-50
choice tends to dominate at TIAA-CREF, the huge retirement fund for
university faculties. Here, at least, there is professional advice available
to help participants avoid the simplifications of framing and, instead, to
understand the structure that would best suit their needs. But I must
also report that one of the famous developers of the theory of finance,
whose current activities receive an entire chapter in this book, has confessed
he has also made the 50-50 choice at TIAA-CREF.
Continues …
* Tversky died at the age of 59 in 1996. Kahneman, now at Princeton University, was
awarded the Nobel Prize in Economic Sciences in 2002.
**Unless otherwise specified, all quotations come from personal interviews or personal
correspondence.
†Campbell (2006).
‡See, in particular, Thaler (1991), which describes many examples of the failure of invariance
and framing.
