The Federal Reserve has a “dual mandate” to
maintain stable prices and maximum sustain-
able employment. It does so primarily by con-
trolling its target interest rate (the federal funds
rate), which inuences short-term market rates.
When unemployment is elevated, the Fed loos-
ens monetary policy (lowers its target rate) to
stimulate economic activity and boost output.
When ination is rising, the Fed tightens policy
(raises its target rate) to slow economic activity
and counteract inationary pressure.
But are both types of policy responses equally
eective? Since the Great Depression, econo-
mists have suspected that tight policy may
have a stronger eect on output than loose
policy because the Fed’s response to the market
crash of 1929 failed to avert the Great Depres-
sion. Milton Friedman and Anna Schwartz later
argued in their 1963 book, A Monetary History of
the United States, that this was because the Fed’s
policy stance during the early 1930s actually
was contractionary rather than expansionary,
but other examples have reinforced the view
that expansionary monetary policy may be more
limited than contractionary policy. Most recently,
the Fed was forced to turn to unconventional
March 2017, EB17-03
Economic Brief
EB17-03 - Federal Reserve Bank of Richmond
Are the Eects of Monetary Policy Asymmetric?
By Regis Barnichon, Christian Matthes, and Tim Sablik
The Federal Reserve uses monetary policy to stimulate the economy when
unemployment is high and to rein in inationary pressures when the econ-
omy is overheating. However, evidence suggests that these policy stances
have unequal eects. Contractionary monetary shocks raise unemployment
more strongly than expansionary shocks lower it.
Page 1
policies during the Great Recession after reduc-
ing its target rate to nearly zero seemed to have
little eect on unemployment.
Monetary policy asymmetry is best described
using a metaphor. Imagine a string with mon-
etary policy at one end and the economy at the
other. Employing tight monetary policy when
ination is rising is like pulling on the string to
keep the economy in check — it works fairly
well. But attempting to stimulate the economy
with loose policy during a downturn is like try-
ing to push on the string to move the economy
— not very eective.
In addition to monetary changes having asym-
metric eects based on their direction, the
strength of monetary policy may also vary with
the state of the economy. Previous tests for mon-
etary policy asymmetries have had somewhat
mixed results, but this Economic Brief presents
new evidence to conrm the asymmetric eects
of monetary policy.
Why Might Monetary Policy Be Asymmetric?
There are several theoretical reasons why mon-
etary policy could have asymmetric eects on
Page 2
economic output.
1
The rst relates to the behavior
of lenders and borrowers under dierent monetary
conditions. When the Fed raises its policy rates, mar-
ket rates tend to rise accordingly. One might expect
that banks would simply pass these higher rates on to
their borrowers. While this is true to an extent, raising
loan rates too high could increase the likelihood that
risky borrowers default. As a result, banks may choose
to ration credit during a period of high interest rates,
constraining credit for some consumers and leading
to a bigger decline in output, thus amplifying the
impact of contractionary monetary policy. On the
other hand, expansionary policy will not necessarily
increase borrowing and spending if economic condi-
tions have reduced demand. Unlike tight monetary
policy, it is not a binding constraint on consumers (as
expressed in the old adage, you can lead a horse to
water but you can’t make it drink).
Another reason why expansionary monetary policy
might be less eective than contractionary policy
is because prices seem less likely to adjust down-
ward — that is, they are “sticky.” Firms also may be
reluctant to lower wages for fear of damaging worker
morale. Because of such downward price and wage
rigidity, rms will tend to respond to contraction-
ary monetary policy by reducing output rather than
prices. Prices and wages are less upwardly sticky,
however. Firms are accustomed to raising prices and
wages gradually due to ination, for example. As a
result, expansionary monetary policy is more likely
to prompt a change in prices rather than output.
Finally, monetary policy may have asymmetric eects
during dierent points in the business cycle due to
changes in consumer outlook. Similar to the credit-
constraint argument, if consumers are pessimistic
about economic conditions, then lowering rates may
not do much to stimulate borrowing and spending.
This explanation is not entirely compelling, however,
since consumer optimism during a boom period
should also weaken the eect of tight monetary
policy. For contractionary policy to have a stronger
eect than expansionary policy, consumers and rms
would have to be more pessimistic during economic
downturns than they are optimistic during booms.
This is certainly possible but perhaps not realistic.
Testing for Asymmetry
If monetary policy does have asymmetric eects on
output, that nding would have important implica-
tions for how the Fed conducts policy. Conclusive
evidence one way or the other has proven some-
what elusive, however. A number of studies do nd
evidence that contractionary policy has a stronger
eect on output than expansionary policy, as the
theory predicts.
2
But other studies nd that what
matters is not the direction of the monetary change
but rather its size.
3
And still other studies nd evi-
dence that the impact of monetary policy depends
chiey on the state of the economy.
4
One problem facing economists trying to nd evi-
dence of asymmetry is that the standard models
used for measuring the eects of shocks, such as
changes in monetary policy, have diculty identify-
ing asymmetric eects. Economists have attempted
to get around this problem in two ways. The rst
involves looking at unanticipated increases and
decreases in the money supply and testing whether
these changes have asymmetric eects. One chal-
lenge with this approach is correctly identifying
unanticipated monetary shocks. Additionally, while
these models may be able to detect asymmetry
based on the direction of a monetary change,
they struggle to measure other potential causes of
asymmetry. Another approach makes use of regime-
switching models that allow for the impact of one
variable (monetary policy) to depend upon changes
in another variable (the state of the economy). But
while these models can identify whether the eects
of monetary policy change with the business cycle,
they are not able to determine if the eects of con-
tractionary policy are inherently different from
those of expansionary policy.
Two of the authors of this brief, Barnichon and Mat-
thes, have developed an alternative approach for
addressing these issues.
5
They start with a model
of the economy in which the behavior of a system
of macroeconomic variables is determined by its
(possibly asymmetric) response to past and present
shocks. They then use Gaussian functions to para-
meterize the dynamic eects of structural shocks
on the economy. The advantage of this approach
Page 3
hand, a 0.7 percentage point decrease in the federal
funds rate produces only a 0.04 percentage point de-
crease in unemployment — an effect that is not
statistically dierent from zero. This implies that
contractionary monetary policy has a signicantly
stronger eect on unemployment than expansion-
ary policy.
Barnichon and Matthes also nd some, albeit incon-
clusive, evidence that prices respond asymmetrically
to monetary changes. Prices appear stickier follow-
ing monetary contractions than following monetary
expansions. This provides some supporting evidence
for the theory that monetary policy has asymmetric
eects because rms are more reluctant to lower
prices and wages than to raise them.
Next, Barnichon and Matthes expand their model
to allow the eects of monetary policy to depend
on both the direction of the change and the state
of the economy. Again, they nd that expansionary
monetary policy has a weaker eect on unemploy-
ment than contractionary policy. Additionally, they
nd that the eect of expansionary policy depends
on the state of the economy. When unemployment
— dubbed Gaussian Mixture Approximation — is
that it uses only a small set of free parameters, which
then allows for a much more ecient estimation of
models with asymmetric responses. In simulations,
Barnichon and Matthes’ approach not only performs
as well as benchmark models in estimating linear or
symmetric responses, it can also detect asymmetric
responses in nonlinear models. Barnichon and Mat-
thes use their new methodology to estimate wheth-
er monetary shocks generate asymmetric responses
depending on the direction of the shock as well as
the state of the economy.
Results and Implications
Barnichon and Matthes rst test whether the direc-
tion of a monetary shock alone results in dierent
economic responses. Applying data from 1959
through 2007 to their model, they nd strong evi-
dence of an asymmetric response in unemployment
depending on the direction of the monetary change.
They estimate that an increase in the federal funds
rate of 0.7 percentage points results in an increase in
unemployment of 0.15 percentage points, a larger
eect than the 0.10 percentage points estimated by
a standard linear model.
6
(See Figure 1.) On the other
Figure 1: Asymmetric Unemployment Rate Responses to Monetary Policy Changes
Barnichon and Matthes Standard Linear Model
55 1313 2929 1717 2121 2525
Quarters since ShockQuarters since Shock
11 99
-0.05
0
0.05
0.1
0.15
0.2
1 5 9 13 17 21 25 29
Percentage Point Decrease in Unemployment
Quarters Since Shock
Expansionary Monetary Shock
Barnichon and Mahes
Standard model
Source: Regis Barnichon and Christian Matthes, “Gaussian Mixture Approximations of Impulse Responses and the Nonlinear Eects of Monetary
Shocks,” Federal Reserve Bank of Richmond Working Paper No. 16-08, June 2016.
Notes: Shaded areas represent margins of error above and below Barnichon and Matthes’ estimates. These areas account for 90 percent of all cases.
0.20
0.15
0.10
0.05
0
-0.05
0.20
0.15
0.10
0.05
0
-0.05
Percentage Point Decrease in Unemployment Rate
Percentage Point Increase in Unemployment Rate
Expansionary Monetary ShockContractionary Monetary Shock
-0.05
0
0.05
0.1
0.15
0.2
1 5 9 13 17 21 25 29
Percentage Point Increase in Unemployment
Quarters Since Shock
Contraconary Monetary Shock
Barnichon and Mahes
Standard model
Page 4
3
For example, Morten O. Ravn and Martin Sola, “A Reconsidera-
tion of the Empirical Evidence on the Asymmetric Eects of
Money-Supply Shocks: Positive vs. Negative or Big vs. Small?”
Birkbeck, University of London, Archive Discussion Paper No.
9606, February 1996.
4
For example, see Silvana Tenreyro and Gregory Thwaites,
“Pushing on a String: US Monetary Policy Is Less Powerful in
Recessions,” American Economic Journal: Macroeconomics,
October 2016, vol. 8, no. 4, pp. 43–74. A working paper ver-
sion is available online. Also, see Ming Chien Lo and Jeremy
Piger, “Is the Response of Output to Monetary Policy Asym-
metric? Evidence from a Regime-Switching Coecients Model,”
Journal of Money, Credit and Banking, October 2005, vol. 37,
no. 5, pp. 865–886. A working paper version is available online.
Also, see Charles L. Weise, “The Asymmetric Eects of Monetary
Policy: A Nonlinear Vector Autoregression Approach,” Journal
of Money, Credit and Banking, February 1999, vol. 31, no. 1,
pp. 85–108.
5
Regis Barnichon and Christian Matthes, “Gaussian Mixture Ap-
proximations of Impulse Responses and the Nonlinear Eects
of Monetary Shocks,” Federal Reserve Bank of Richmond Work-
ing Paper No. 16-08, June 2016.
6
In this context, “linear” implies symmetry.
This article may be photocopied or reprinted in its
entirety. Please credit the authors, source, and the
Federal Reserve Bank of Richmond and include the
italicized statement below.
Views expressed in this article are those of the authors
and not necessarily those of the Federal Reserve Bank
of Richmond, the Federal Reserve Bank of San Francisco,
or the Federal Reserve System.
is low, expansionary policy generates a substantial
increase in ination but little change in unemploy-
ment. When unemployment is high, expansionary
monetary policy has little eect on ination and
some positive eect on employment. This is consis-
tent with standard macroeconomic theory and the
Fed’s experience during the Great Ination.
The ndings from Barnichon and Matthes’ model
have a number of implications for monetary policy-
makers. They suggest that monetary policy asym-
metries may be larger than previous estimates have
found. The Fed’s ability to stimulate the economy
through expansionary policy appears less potent
than the negative eect contractionary policy has
on employment. Additionally, as theory and other
studies have suggested, attempting to use mone-
tary policy to stimulate the economy beyond full
employment is likely to only increase ination with
no signicant reduction in unemployment.
Regis Barnichon is a research advisor at the Federal
Reserve Bank of San Francisco. Christian Matthes is
a senior economist and Tim Sablik is an economics
writer in the Research Department at the Federal
Reserve Bank of Richmond.
Endnotes
1
Donald P. Morgan, “Asymmetric Eects of Monetary Policy,”
Federal Reserve Bank of Kansas City Economic Review, Second
Quarter 1993, vol. 78, no. 2, pp. 22–33.
2
For example, see James Peery Cover, “Asymmetric Eects of
Positive and Negative Money-Supply Shocks,” Quarterly Journal
of Economics, November 1992, vol. 107, no. 4, pp. 1261–1282;
and Emiliano Santoro, Ivan Petrella, Damjan Pfajfar, and Edo-
ardo Gaeo, “Loss Aversion and the Asymmetric Transmission
of Monetary Policy,” Journal of Monetary Economics, November
2014, vol. 68, pp. 19-36. A working paper version is available
online.
Richmond Baltimore Charlotte
FEDERAL RESERVE
BANK
OF RICHMOND
