Presidential Address: Social Transmission Bias in Economics and Finance

• DOI: http://doi.org/10.1111/jofi.12906
• Date: 01 August 2020
• Author: DAVID HIRSHLEIFER
• Published date: 01 August 2020

The Journal of Finance R

David Hirshleifer

The President of the American Finance Association 2019

THE JOURNAL OF FINANCE •VOL. LXXV, NO. 4 •AUGUST 2020

Presidential Address:

Social Transmission Bias

in Economics and Finance

DAVID HIRSHLEIFER∗

ABSTRACT

I discuss a new intellectual paradigm, social economics and finance—the study of the

social processes that shape economic thinking and behavior.This emerging field recog-

nizes that people observe and talk to each other. A key, underexploited building block

of social economics and finance is social transmission bias: systematic directional shift

in signals or ideas induced by social transactions. I use five “fables” (models) to illus-

trate the novelty and scope of the transmission bias approach, and offer several emer-

gent themes. For example, social transmission bias compounds recursively,which can

help explain booms, bubbles, return anomalies, and swings in economic sentiment.

THIS ADDRESS DISCUSSES A NEW intellectual paradigm, which I call social eco-

nomics and finance—the study of how social interaction affects economic out-

comes. In standard analyses of economic behavior, people interact only imper-

sonally via trading orders and observation of market price. A missing chapter

in our understanding of finance consists of the social processes that shape

economic thinking and behavior.

Social economics and finance recognizes that people observe each other and

talk to each other, where talking includes written text and social media. A key

but underexploited intellectual building block of social economics and finance

is social transmission bias, the systematic directional modification of ideas or

signals as they pass from person to person.

∗David Hirshleifer is at Merage School of Business, University of California. This paper was

the Presidential Address of the American Finance Association at the 2020 Annual Meeting in San

Diego, CA. I thank Daniel Andrei, Julien Cujean, Darrell Duffie, Ben Golub, Cam Harvey,Michael

Hirshleifer, Chong Huang, Nan Ma, Robert Merton, Amit Seru, Robert Shiller, Avanidhar Sub-

rahmanyam, Siew Hong Teoh, Sheridan Titman, Johan Walden, Ivo Welch, Jeff Zwiebel, the 2019

Retreat on Information and Social Economics (RINSE), the FRILLS (Finance Research Informal

Lunchtime Seminar) at Stanford University, and the Editor, Stefan Nagel, for extremely helpful

comments, and Yushui Shi and Josh Thornton for outstanding research assistance. Presentation

slides for the talk are available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3513210.

A transcript with integrated slides is available at https://ssrn.com/abstract=3513201. Video of the

talk is available at https://youtu.be/KWlv2uFtsJY.

Correspondence: David Hirshleifer, Merage School of Business, University of California, Irvine,

CA 92697; email: david.h@uci.edu.

DOI: 10.1111/jofi.12906

C2020 the American Finance Association

1779

1780 The Journal of Finance R

By way of comparison, previous intellectual paradigm shifts include infor-

mation economics, which recognizes that some people know things that others

do not, and behavioral economics and finance, which recognizes that people

make systematic mistakes. Of course, scholars knew these obvious facts long

before each of these paradigm shifts. The problem was that these facts were

considered only informally and sporadically—they were not systematically, ex-

plicitly, and routinely incorporated in our thinking and models to generate test

hypotheses. Similarly, social interaction is only starting to join the standard

intellectual toolkit in finance research.

As things stand, in behavioral finance, for example, the path from assump-

tions to conclusions is often very direct. With respect to beliefs, when we observe

that investors trade too aggressively, we conclude that they must be overcon-

fident. When we observe that expectations become more optimistic after price

run-ups than after run-downs, we conclude that they must overextrapolate.

With regard to preferences, when we see individual investors tilted toward buy-

ing lottery stocks, selling winners more than losers, or saving too little—well,

we have behavioral models in which investors have preferences for skewness,

for realizing gains not losses, and for immediate consumption.

I am a fan of these direct approaches. They capture a large part of the

truth. However, crucially, there can be attraction to a behavior without any

preference for it. Moths are attracted to flame—a nearby light source. But

moths are not flame-loving. There is no cognitive reward system that pays off

more for approaching a flame. Instead, moths evolved under natural selection

to navigate based on a distant light source, the moon. Nearby light sources

fool their navigation systems. So attraction to flame is an indirect effect, not a

direct preference.

Another kind of indirect effect is social emergence—the phenomenon whereby

social outcomes are not just the sums of individual propensities. An example

of a socially emergent effect is the phenomenon of death spirals among army

ants, as seen in Figure 1.

These can be up to hundreds of feet wide. The ants continue to walk in circles

until they die.

Surely this could not happen to a smarter animal, such as a mammal. Except,

apparently, for man’s best friend, as seen in videos of jostling dogs rotating

about their shared food bowl. Are birds smarter? The most disturbing spectacle

of all is seen in a video of turkeys marching in a circle around a dead cat—

apparently some kind of religious ritual.

Should we conclude from these behaviors that animals have a rotative in-

stinct? A heuristic or bias for circular motion? Of course not. In the case of

ants, there are instincts for random search for food, and instincts that pro-

mote following other ants. As a result, if the ant at the front of the line by

chance starts following the ant at the back of the line, a dysfunctional so-

cial outcome results—somewhat reminiscent of a fad or bubble.1These animal

1This phenomenon is closely akin to information cascades (Banerjee (1992), Bikhchandani,

Hirshleifer, and Welch (1992)). In information cascades, even rational social learning results in

fad-like outcomes.