To ensure that our results did not depend on a few unusual
observations, or "outliers," we re-ran our regression, omitting the
days with the most extreme or unusual market moves. This more
conservative analysis led to a smaller estimate of the effect of
policy actions on the stock market, a stock price multiplier of about
2.6 rather than 4.7. However, the effect remains quite sharp in
statistical terms.8
We considered other variations as well. For example, we investigated
whether the magnitude of the effect on the stock market of a surprise
policy tightening (that is, an increase in interest rates) differs
from that of a surprise easing of comparable size. It does not. Yet
another experiment consisted of asking whether an unanticipated
policy change has a larger effect if it is thought by the market to
signal a longer-lasting change in policy. We measured the perceived
permanence of policy changes by observing the effects of
unanticipated policy changes on the expected federal funds rate three
months in the future, as measured by the futures market. The stock
market multiplier associated with unanticipated policy moves that are
perceived to be more permanent is a bit higher, as would be expected;
its value is about 6.9
In short, the statistical evidence is strong for a stock price
multiplier of monetary policy of something between 3 and 6, the
higher values corresponding to policy changes that investors perceive
to be relatively more permanent. That is, according to our findings,
a surprise easing by the Fed of 25 basis points will typically lead
broad stock indexes to rise from between 3/4 percentage point and 1-
1/2 percentage points. Incidentally, similar results obtain for stock
values of industry groups: We find almost all industry stock
portfolios respond significantly to changes in monetary policy, with
telecommunications, high-tech, and durables goods industry stocks
being the most sensitive to monetary policy news, and energy,
utilities, and health care stocks being the least sensitive.10 These
results can be broadly explained by the tendency of each industry
group to move with the broad market, or (to use the language of the
standard capital asset pricing theory), by their industry "betas."
Why Does Monetary Policy Affect Stock Prices?
It is interesting, though perhaps not terribly surprising, to know
that Federal Reserve policy actions affect stock prices. An even more
interesting question, though, is, why does this effect occur?
Answering this question will give us some insight into how monetary
policy affects the economy, as well as the role that the stock market
should play in policy decisions.
A share of stock is a claim on the current and future dividends (or
other cash flows, such as stock buybacks) to be paid by a company.
Suppose, for just a moment, that financial investors do not care
about risk. Then only two types of news ought to affect current stock
values: news that affects investor forecasts of current or future
(after-tax) dividends or news that affects forecasts of current or
future short-term interest rates. News that current or future
dividends (which I want to think of here as being measured in real,
or inflation-adjusted, terms) are likely to be higher than previously
expected--say, because the company is expecting to be more profitable-
-should raise the current stock price. News that current or future
short-term interest rates (also measured in real, or inflation-
adjusted, terms) are likely to be higher than previously expected
should depress the stock price. There are two essentially equivalent
ways of understanding why expectations of higher short-term real
interest rates should lower stock prices. First, to value future
dividends, an investor must discount them back to the present; as
higher interest rates make a given future dividend less valuable in
today's dollars, higher interest rates reduce the value of a share of
stock. Second, higher real interest rates make investments other than
stocks, such as bonds, more attractive, raising the required return
on stocks and reducing what investors are willing to pay for them.
Under either interpretation, expectations of higher real interest
rates are bad news for stocks.
So, to reiterate, in a world in which investors do not care about
risk, stock prices should change only with news about current or
future dividends or about current or future real interest rates.
However, investors do care about risk, of course. Because investors
care about risk, and because stocks are viewed as relatively risky
investments, investors generally demand a higher average return,
relative to other assets perceived to be safer, to hold stocks. Using
long historical averages, one finds that, in the United States, a
diversified portfolio of stocks has paid 5 to 6 percentage points
more per year, on average, than has a portfolio of government bonds.
This extra return, known as the risk premium on stocks, or the equity
premium, presumably reflects, in part, the extra compensation that
investors demand to be willing to hold relatively more risky stocks.
11
Like news about dividends and real interest rates, news that affects
the risk premium on stocks also affects stock prices. For example,
news of an impending recession could raise the risk premium on stocks
in two ways. First, the macroeconomic environment is more volatile
than usual during a recession, so stocks themselves may become
riskier investments. Second, the incomes and wealth of financial
investors tend to fall during a downturn, giving them a smaller
cushion to support the lifestyles to which they are accustomed (that
is, to make house payments and meet other obligations). With less
discretionary income and wealth to absorb potential losses, people
may become less willing to bear the risks of more volatile financial
investments (Campbell and Cochrane, 1999). For both reasons, the
extra return that investors demand to hold stocks is likely to rise
when bad times loom. With expected dividends and the real interest
rate on alternative assets held constant, the expected yield on
stocks can rise only through a decline in the current stock price.12
We now have a list of three key factors that should affect stock
prices. First, news that current or future dividends will be higher
should raise stock prices. Second, news that current or future real
short-term interest rates will be higher should lower stock prices.
And third, news that leads investors to demand a higher risk premium
on stocks should lower stock prices.
How does all this relate to the effects of monetary policy on stock
prices? According to our analysis, Fed actions should affect stock
prices only to the extent that they affect investor expectations
about dividends, short-term real interest rates, or the riskiness of
stocks. The trick is to determine quantitatively which of these sets
of investor expectations is likely to be most affected when the Fed
unexpectedly changes the federal funds rate.
To make this determination, we used a methodology first applied by
the financial economist John Campbell, of Harvard University, and by
Campbell and John Ammer of the Federal Reserve Board staff (Campbell,
1991; Campbell and Ammer, 1993). Putting the details aside, we can
describe the basic idea as follows. Imagine that the expectations of
stock market investors can be mimicked by a statistical forecasting
model that takes relevant current data as inputs and projects
estimated future values of aggregate dividends, real interest rates,
and equity risk premiums as outputs. In principle, investors could
use such a model to make forecasts of these key variables and hence
to estimate what they are willing to pay for stocks. Besides a number
of standard variables that have been shown to be helpful in making
forecasts of such financial variables, suppose we include in the
forecasting model our measure of unanticipated changes in the federal
funds rate.13 That is, we use the information contained in these
unanticipated changes in making our forecasts of future dividends,
interest rates, and risk premiums.
Now we can consider the following thought experiment. Suppose we have
run our computer model, made our forecasts, and inferred the
appropriate values for stocks. But then we receive news that the Fed
has unexpectedly raised the federal funds rate by 25 basis points.
Based on our forecasting model, by how much would that information
change our previous forecasts of future dividends, interest rates,
and risk premiums? The answer to this question clarifies the channel
by which monetary policy affects stock prices. If we were to find,
for example, that the news of an unexpected increase in the funds
rate significantly changed the forecast of future dividends but did
not much affect the forecasts of interest rates or risk premiums,
then we could conclude that monetary policy affects stock prices
primarily by affecting investor expectations of future dividends. By
contrast, if news of the policy action changed the model forecasts
for real interest rates but did not change our forecasts for the
other two variables, we would decide that unanticipated policy
actions affect stock prices primarily by influencing the interest
rates expected by stock investors.
What we actually found when conducting this statistical experiment
was quite interesting. It appears that, for example, an unanticipated
tightening of monetary policy leads to only a modest change in
forecasts of future dividends and to still less of a change in
forecasts of future real interest rates (beyond a few quarters).
Quantitatively, according to our methodology, the most important
effect of a policy tightening is on the forecasted risk premium.
Specifically, an unanticipated tightening of monetary policy raises
expected risk premiums on stocks for a protracted period. For a given
expected stream of dividend payouts and real interest rates, the risk
premium and hence the return to holding stocks can only rise if the
current stock price falls.
In short, our analysis suggests that an unanticipated monetary
tightening lowers stock prices only to a small extent by lowering
investor expectations about future dividend payouts, and by still
less by raising expected real interest rates. The most powerful
effect of an unanticipated monetary tightening is to increase the
perceived risk premium on stocks, either by increasing the riskiness
of stocks, by reducing people's willingness to bear risk, or both.
Reduced willingness of investors to hold relatively more risky stocks
drives down stock prices.
Our analysis does not explain precisely how monetary policy affects
risk, but we can make reasonable conjectures. For example, tighter
monetary policy may raise the riskiness of shares themselves by
raising the interest costs and weakening the balance sheets of
publicly owned firms (Bernanke and Gertler, 1995). In the
macroeconomy more generally, by reducing spending and economic
activity, tighter money raises the risks of unemployment or
bankruptcy faced by individual households or firms. In each case,
tighter monetary policy increases risk by reducing financial buffers
or otherwise increasing the vulnerability of individuals or firms to
future shocks to the economy.
Implications of the Results for Monetary Policy
So far I have discussed two principal conclusions from the empirical
analysis: First, the stock price multiplier of monetary policy is
between 3 and 6--in other words, an unexpected change in the federal
funds rate of 25 basis points leads, on average, to a movement of
stock prices in the opposite direction of between 3/4 percentage
point and 1-1/2 percentage points. Second, the main reason that
unanticipated changes in monetary policy affect stock prices is that
they affect the risk premium on stocks. In particular, a surprise
tightening of policy raises the risk premium, lowering current stock
prices, and a surprise easing lowers the risk premium, raising
current stock prices.
What implications do these results have for our broader understanding
and for the practice of monetary policy? I will briefly discuss two
issues: first, the role of the stock market in the transmission of
monetary policy changes to the economy; and second, the efficacy of
monetary policy as a tool for controlling stock market "bubbles."
A long-held element of the conventional wisdom is that the stock
market is an important part of the transmission mechanism for
monetary policy. The logic goes as follows: Easier monetary policy,
for example, raises stock prices. Higher stock prices increase the
wealth of households, prompting consumers to spend more--a result
known as the wealth effect. Moreover, high stock prices effectively
reduce the cost of capital for firms, stimulating increased capital
investment. Increases in both types of spending--consumer spending
and business spending--tend to stimulate the economy.
This simple story can be elaborated somewhat in light of our results.
It is true, as I have discussed, that an easier monetary policy
raises stock prices, whereas a tighter policy lowers them. However,
easier monetary policy not only raises stock prices; as we have seen,
it also lowers risk premiums, presumably reflecting both a reduction
in economic and financial volatility and an increase in the capacity
of financial investors to bear risk. Thus, our results suggest that
easier monetary policy not only allows consumers to enjoy a capital
gain in their stock portfolios today, but it also reduces the
effective amount of economic and financial risk they must face. This
reduction in risk may cause consumers to trim their precautionary
saving, that is, to reduce the amount of income that they put aside
to protect themselves against unforeseen contingencies. Reduced
precautionary saving in turn implies more spending by households.
Thus, the reduction in risk associated with an easing of monetary
policy and the resulting reduction in precautionary saving may
amplify the short-run impact of policy operating through the
traditional channel based on increased asset values. Likewise,
reduced risk and volatility may provide an extra kick to capital
expenditure in the short run, as firms are more likely to undertake
investments in new structures or equipment in a more stable
macroeconomic environment.14
A second issue concerns the role of monetary policy in the management
of large swings in stock values, or "bubbles." In an earlier speech
(Bernanke, 2002), I gave a number of reasons why I believe that using
monetary policy--as opposed to microeconomic, prudential policies--is
not a good way to address the problem of asset-market bubbles. These
included the difficulty of identifying bubbles in advance; the
questionable wisdom, in the context of a free-market economy, of
setting up the central bank as the arbiter of asset values; the
problem that arises when a bubble occurs in only one asset class
rather than in all asset classes; and other reasons. A major concern
that I have about the bubble-popping strategy, however, is that
attempts to bring down stock prices by a significant amount using
monetary policy are likely to have highly deleterious and unwanted
side effects on the broader economy. The research I have described
today allows me to address this issue more concretely. Here I will
make just two points. |
