During technological revolutions, stock prices of
innovative firms tend to exhibit bubble-like
patterns. After an initial surge, stock prices usually
fall in the presence of high volatility, as they did during the“biotech revolution” of the early 1980s and the internet craze
of the late 1990s.
While the bubble-like stock price behavior is commonly attributed to the irrationality of overenthusiastic investors, why would investors make the same mistake over and over again?
In the recent study “Technological Revolutions and Stock
Prices,” University of Chicago Graduate School of Business
professors Lubos Pastor and Pietro Veronesi propose the first
rational explanation for why stock prices should be expected
to exhibit a bubble during a technological revolution—a period
concluded by a large-scale adoption of a new technology.
“Our explanation for the bubbles is that the nature of
risk associated with new technologies changes over time,”
says Veronesi.
Uncertainty about productivity gains is a natural feature of
innovative technologies. At first this uncertainty, or risk, is
mostly “idiosyncratic,” because the new technology is initially
developed on a small scale and the probability of large-scale
adoption is low. For new technologies that become widely
adopted, the uncertainty gradually changes from idiosyncratic
to “systematic.” When systematic risk increases, prices decline.
These increases in systematic risk can be expected in hindsight, by researchers who look back knowing that the revolutions took place, but they are unexpected by real-time investors who do not know whether the new technology will eventually be adopted on a large scale.
According to Pastor and Veronesi, the “bubbles” should be most pronounced in revolutions characterized by high uncertainty about, and fast adoption of, the technology—such as the recent internet revolution.
The authors developed an economic model to provide a rational explanation for stock price movement during technological revolutions. To test their model, the authors examined stock prices in 1830–61 and 1992–2005, the respective periods when railroad and internet technologies spread in the United States.
“Bubbles are not merely possible in a rational world, but
should be expected during technological revolutions,” write
Pastor and Veronesi.
The Changing Nature of Risk
In order to explain how stock prices should behave during technological revolutions, Pastor and Veronesi developed
what economists call a “general equilibrium model.” In the
model, investors study the productivity of a new technology,
and must decide whether adopting this new technology on a
large scale would be worthwhile. Large-scale adoption would
constitute a technological revolution. The authors determine
the optimal time for adopting the new technology and show
that when the technology is optimally adopted, there should
be bubbles in stock prices.
As Pastor and Veronesi write: “The problem we solve
resembles the problem of making an irreversible marriage
decision. It is generally suboptimal to marry a new acquaintance
immediately because of substantial uncertainty regarding
the quality of the personality match. Instead, it seems
advisable to first develop the relationship on a small scale, by
dating without any commitment, and then to marry if we
learn that the relationship is likely to work in the long run.”
The model has two sectors: “new economy” and “old economy.”
In the new economy, small-scale production uses new
technology. In the old economy, large-scale production initially
uses old technology and switches to new technology if the
latter appears to provide additional productivity gains. When
this switch occurs, the inherent risk of new technology
becomes systematic in nature, since it affects mass-scale
production and aggregate economic growth. However, if the
new technology turns out to be faulty, it may stall aggregate
economic development.
Stock prices initially rise during a revolution because of
good news about the productivity of the new technology. As
the probability of large-scale adoption increases, the systematic
risk in the economy increases as well, because it becomes
more likely that the risk of the new technology will affect mass
production. As the systematic risk in the economy increases,
stock prices fall.
There are two measures of systematic risk that increase
during technological revolutions: the stock volatility of the old
economy and the “beta” of the new economy. A “beta” captures
the sensitivity of the stock price to the aggregate stock market. If a stock has a beta of 2, and the aggregate stock market goes up
by 1 percent, this stock’s price will typically go up by 2 percent.
Breaking it down into these two pieces, the authors show
that systematic risk increases in both the old and new
economies, thus it should be expected that stock prices fall
after a run-up in both economies. New economy stocks are
expected to fall more than old economy stocks shortly before
the adoption of new technology.
The model produces the following empirical predictions
that should apply to technological revolutions: 1) the bubble in
stock prices should be more pronounced in the new economy
than in the old economy; 2) stock prices in both economies
should reach their lowest levels at the end of the revolution;
3) the new economy’s market beta should increase sharply
before the end of the revolution;
4) the new economy’s volatility
should rise sharply and exceed the
old economy’s volatility; 5) the old
economy’s volatility should rise, but
less than the new economy’s volatility;
6) the new economy’s beta and both
volatilities should all peak at the
end of the revolution; and 7) the
old economy’s productivity should begin rising at the end of
the revolution.
The Internet and Railroads
Pastor and Veronesi found substantial support for their
empirical predictions in evidence from railroad and internet
technology stock prices. For both technological revolutions,
the authors considered the key quantities in the model, such
as the new economy’s market beta and the level and volatility
of stock prices.
Today, internet technology is an indelible part of the economic landscape. However, in the mid-1990s, it was not clear that the internet would play a dominant role in the economy. The predictions of the model, however, were all supported by empirical evidence from the internet revolution.
The authors used the technology-loaded NASDAQ stock
index to represent the new economy and the NYSE/AMEX to
represent the old economy. NASDAQ’s beta doubled between 1997 and 2002, which is highly statistically significant. There
was a clear “bubble” in the new economy, whose market value
increased five times over, and then fell by more than half. The
bubble pattern was much stronger in the NASDAQ index than
in the NYSE/AMEX, with stock prices in both reaching bottom
in 2002. The NYSE/AMEX’s return volatility doubled and
NASDAQ’s volatility tripled over the same period. NASDAQ’s
beta and both volatilities peaked in 2002. These patterns also
support other predictions of the model.
Patterns of NASDAQ’s beta and NYSE/AMEX’s volatility
also show that both sectors experienced large increases in
systematic risk between 1997 and 2002, supporting the key
prediction of the model. In addition, productivity increased
dramatically after the stock market bubble burst in 2002.
According to the model, all of this evidence suggests that the
internet technology was widely adopted by 2002.
In addition to studying the internet revolution, the authors
analyzed the first major technological revolution that took
place in the United States since the NYSE was organized in
1792: the introduction of steam-powered railroads. Nearly all
railroads began as corporations funded by private investors.
In the 1830s and 1840s, there was substantial uncertainty
about whether railroad technology would be adopted on a
large scale. The authors analyzed stock prices before the Civil
War and found that stock prices fell before and during 1857,
with railroad stocks falling more than nonrailroad stocks.
While the volatility of all stocks rose in 1857, the volatility
of railroad stocks consistently exceeded that of nonrailroad
stocks. The railroad stock beta increased sharply in the 1850s
before falling immediately after 1857. This evidence is consistent
with a large-scale adoption of railroad technology around
1857, just after railroads began expanding west of the Mississippi
River.
“Looking at stock prices can help us determine when a
technological revolution took place, pinpointing when new
technology was adopted on a large scale,” says Pastor.
Natural Developments
“During periods when stock prices are increasing rapidly,
many argue that the Federal Reserve should intervene to
deflate the bubble,” says Veronesi. “However, if investors are
rational and their behavior yields accurate market prices,
policy interventions would be a mistake.”
Interventions, such as regulation or increasing interest
rates, might distort the allocation of resources and discourage
entrepreneurs from investing in new technologies. This could
have negative implications on economic growth, in part
because new technology might not become available for
mass production.
“Intervention would be misguided if something that is a
natural development of the technological revolution is instead
taken as irrational exuberance,” says Veronesi.
Pastor and Veronesi’s model also reflects the lag in productivity
gains from new technology. For example, although
electricity first appeared around 1880, it was not until the
1920s that the productivity of the U.S. economy increased as
a result of large-scale adoption of electricity.
“Our findings also have implications for stock market investors,” says Pastor. “Investors living through technological revolutions should not try to anticipate the market peak, because they usually don’t know ahead of time whether the revolution will actually take place or when it will end.”
“Technological Revolutions and Stock Prices.” Lubos Pastor and Pietro Veronesi.