Behavioral Economics: Past, Present, and Future

1589
Thaler: behavioral economics: pasT, presenT, and fuTure
vol. 106 no. 7
only to fall back to the long-term trend. Because of the various forms of leverage 
involved, this rise and fall in prices helped create the global Great Recession.
The difference between the CUBA example and these much larger bubbles is that 
it is impossible to prove that prices in the latter were ever wrong. There is no clear 
smoking gun. But it certainly feels like asset prices can diverge significantly from 
fundamental value. Perhaps we should adopt the definition of market efficiency pro-
posed by Fischer Black 
(1986) in his presidential address to the American Finance 
Association, which had the intriguing one word title “Noise.” Black 
(1986, p. 553) 
says “we might define an efficient market as one in which price is within a factor of 
two of value, i.e., the price is more than half of value and less than twice value. The 
factor of two is arbitrary, of course. Intuitively, though, it seems reasonable to me, 
in light of sources of uncertainty about value and the strength of the forces tending 
to cause price to return to value. By this definition, I think almost all markets are 
efficient almost all the time. ‘Almost all’ means at least 90 percent.”
One can quibble over various aspects of Black’s definition but it seems about 
right to me, and had Black lived to see the tech bubble of the 90s he might have 
revised his number up to three. I would like to make two points about this. The first 
is that the efficient market hypothesis has been a highly useful, indeed essential 
concept in the history of research on financial markets. In fact, without the EMH 
there would have been no benchmark with which to compare anomalous findings. 
The only danger created by the concept of the EMH is if people, especially poli-
cymakers, consider it to be true. If policymakers think that bubbles are impossible, 
then they may fail to take appropriate steps to dampen them. For example, I think it 
would have been appropriate to raise mortgage-lending requirements in cities where 
price to rental ratios seemed most frothy. Instead, this was a period in which lending 
requirements were unusually lax.
$0.00 
$2.00 
$4.00 
$6.00 
$8.00 
$10.00 
$12.00 
$14.00 
$16.00 
Closing price 
($
)
Price
NAV
+70%
−10%
Dec 18 
Ma
y 20
14
Jul 20
14
Sep 20
14
No
v 20
14
Jan 20
15
Mar 20
15
Ma
y 20
15
Jul 20
15
Sep 20
15
No
v 20
15
Jan 20
16
Mar 20
16
Figure 1. Price and Net Asset Value for CUBA Fund
Note: 
On December 18, 2014, President Obama announced he was going to lift several restrictions against Cuba.
Source: 
Bloomberg

---

1590
THE AMERICAN ECONOMIC REVIEW
july 2016
There is a broader point to make. For lots of reasons we might expect that finan-
cial markets are the most efficient of all markets. They are the only markets where 
it is generally possible to cheaply sell short, an essential feature if we expect prices 
to be “right.” Yet if financial markets can be off by a factor of two, how much con-
fidence should we have that prices in other markets are good measures of value, 
where there are no realistic arbitrage opportunities?
To give just one example, consider labor markets. There has been considerable 
attention paid in recent years to the growing inequality in incomes and wealth 
around the world 
(Piketty 2014; Atkinson, Piketty, and Saez 2011). Although there 
has been much debate about the cause of this trend, most of the discussion within 
economics is based on the presumption that differences in income reflect differences 
in productivity. Is that presumption warranted? If stock prices can be off by a factor 
of two, might not that be true for workers, from hamburger flippers to CEOs?
There is reason for skepticism about that presumption from the bottom to the 
top of the income ladder. At the lower end of the wage distribution there has been 
a long literature begun by Slichter 
(1950) documenting odd inter-industry wage 
differentials. Simply put, some industries pay more than others, and this applies to 
clerical workers and janitors as well as higher paid executives. Important papers by 
Krueger and Summers 
(1988) and Dickens and Katz (1986) reignited this literature 
summarized in Thaler 
(1989). Card, Heining, and Kline (2013) have recently docu-
mented similar findings in Germany using panel data that allow for individual fixed 
effects. They find that when workers move from a bottom quartile paying industry 
to a top quartile industry their wages jump, and the opposite thing happens when 
workers move from a high paying industry to a low one. It seems implausible that 
these workers become significantly more or less productive simply by changing 
industries.
At the other end of the spectrum, the ratio of CEO pay to that of the average 
worker has skyrocketed in the past few decades. In 1965 for large firms based in the 
United States this ratio was 20; by 2014 it was over 300, more than twice the ratio 
in any other country 
(Mishel and Davis 2015). Of course some economists argue 
that this rise simply reflects the growing productivity of the CEOs 
(e.g., Kaplan, 
Klebanov, and Sorensen 2012
) but how confident should we be in this assessment? 
CEO pay is usually set by the compensation committees of boards of directors that 
rely on consultants who base their recommendations in part on the pay of other 
CEOs. This kind of recursive, self-fulfilling process is not one that generates high 
confidence that pay and performance are highly correlated. Of course there is no 
way to settle this argument. Rather, I just want to repeat my question. If stock prices 
can be off by a factor of two, why should we be confident that other markets do not 
diverge by that much, or more?