Go to Brad De Long's Home
Expectational Traps and Discretion
Comments on V.V. Chari, Lawrence J. Christiano, and Martin Eichenbaum,
"Expectational Traps and Discretion"
for the 1996 Stanford CEPR and Federal Reserve Bank of San Francisco
conference on Monetary Policy: Measurement and Management
Brad De Long
Associate Professor of Economics, 601 Evans
University of California at Berkeley
Berkeley, CA 94720-3880
(510) 643-4027 phone; (510) 642-6615 fax
The model is nicely done: monopolistically-competitive producers of intermediate
goods who exercise their monopoly power to set prices "too high"
for social welfare maximization. But they can--in theory, but not on average
in equilibrium--be fooled by unexpected monetary expansions into producing
at closer to the social optimum. Offsetting these benefits to unanticipated
inflation is inflation's action as a distorting tax on employment: anticipated
inflation benefits nobody.
As a result, it is painful, employment-reducing, and welfare-decreasing
for a government to ever let money grow at less than its anticipated value--except
to the extent that lowered money growth reduces expectations of future money
and inflation growth. And it is employment-enhancing and welfare-increasing
for a government to boost money growth to higher than its anticipated value--except
to the extent that higher money growth today creates expectations that such
higher growth will continue.
Suppose that positive deviations of money growth from anticipated values
trigger permanent expectations of vastly higher inflation, while
negative deviations of money growth from anticipated values do not reduce
expectations of future inflation. Then we have a central banker's worst
nightmare. A period-by-period welfare-maximizing government will find itself
straining every muscle to hit the monetary growth mark set by whatever private-sector
expectations happen to be. The government desperately tries to avoid exceeding
inflationary expectations because of the terrible consequences for future
expectations, and desperately tries to avoid undershooting on monetary growth
because it causes unemployment yet does no good at lowering expectations
of future inflation.
These equilibria, in which the benevolent government finds itself forced
to satisfy private-sector expectations of higher inflation are the ones
that Chari, Christiano, and Eichenbaum focus on.
They suggest that these particular expectational trap equilibria provide
a stylized look at one of the powerful mechanisms that gave the U.S. a decade
of higher inflation in the 1970s. Is this accurate?
- To some degree yes: no one in the 1970s ever wanted to
fight inflation by running a high risk of recession--at least not until
Paul Volcker. Herbert Stein recalls how he and his colleagues at the CEA
were "surprised and unhappy" when they learned that President
Nixon had authorized Labor Secretary George Shultz to tell the AFL-CIO
that the Nixon administration would "control inflation without a rise
of unemployment." This is a policy rule that money growth will always
equal or exceed expectations--just as in Chari, Christiano, and Eichenbaum's
- To some degree no: throughout the 1970 the actual inflation
outturn for each year ran ahead of the average forecast from the Survey
of Professsional Forecasters: thirteen years in a row, from the survey's
beginnings in 1969 through 1981. Getting thirteen heads in a row starting
from the moment you begin flipping a coin--that's a p-value of .00012.
Something strange was going on in expectations of inflation (at least the
expectations of professional economic forecasters) which the model cannot
- To some degree no: tightening monetary policy did, when finally
undertaken under Paul Volcker, have large and favorable effects on future
expectations. Certaintly previously unanticipated monetary tightening produced
unemployment now. But it also appears to have greatly reduced expectations
of future inflation. It seems highly probable that a truly benevolent
government would have tightened monetary policy much earlier than Paul
Volcker. Certainly the German and Japanese governments did:
Medium Term Financial Strategy
So can we see Chari, Christiano, and Eichenbaum's model in operation?
I think we can. Think of the British Conservative government, and the
Medium Term Financial Strategy it had adopted at the end of the 1970s. The
MTFS was a set of money stock growth rate targets for the next n
years, gradually declining from rates of growth consistent with the high
British inflation of the late 1970s to price stability. The idea was that
the government would hit the first monetary growth target or two--and that
would convince private-sector economic actors that the government meant
it. Their expectations of future money growth would change, and their expectations
of future money growth would fall.
Since the disinflation would be gradual, you would not have the unpleasant
situation in which money growth in year t was very low when a lot of economic
contracts were still based on expectations of year t money growth formed
in year t-3 when year t growth was expected to be high. You would get very
close to costless disinflation, and you would do so even in a John Taylor
overlapping-contracts world. Your firm commitment to a gradualist disinflationary
policy would ensure that there was never any large negative gap between
actual inflation and the "expected" rate of inflation that had
been frozen into place by the temporal contracting structure of the economy.
But the MTFS did not unfold according to plan. You see, one of the major
issues that the Conservative Party had used to defeat Labour in the elections
of the late 1980s was unemployment. Billboards all over the country
showing lots of people standing in an unemployment benefit line. Text underneath:
"Labour isn't working." Implied promise: more jobs--and lower
unemployment--under a Conservative government.
Disinflation--especially in its early stages, when it runs into an economy
in which lots of prices and contracts in force had been made under the assumption
that disinflation was not going to happen--is not a big employment-generating
policy. So the Conservative Party began to drop in the polls. And as it
dropped in the polls, people began to think "these guys--the current
government--mean it, but they are not going to be around for very long;
someone more left-leaning is almost certain to win the next election and
restoke the engine of inflation. And maybe a coup within the Conservative
Party will transfer power to the 'wet' faction." So the longer the
MTFS continued, the less credible it became--and the less people thought
that current low money growth heralded future low money growth.
Tom Sargent wrote a very good paper on this in the early 1980s, comparing
the disinflationary methods of Margaret Thatcher unfavorably to those of
pre-World War II French Premier Raymond Poincare. Its point was that what
was done gradually can be undone gradually--and hence is unlikely to be
credible. But I think that Thatcher's difficulties were caused much more
by the peculiar shape of British politics than by any theoretical
defect in a long-term and fully credible version of policies like the MTFS.
In the end Thatcher was rescued by Argentinian generals, and by suicidal
divisions among her political opponents who hated each other more than they
cared about winning elections. And the Conservative governments of the 1980s
did have time to convince British economic decision makers that, yes, low
money growth does herald low money growth in the future. But between the
Conservative election victories of the late 1970s and the Falklands War
we can see Britain's macroeconomy unfolding along a path eerily similar
to that of the model in Chari, Christiano, and Eichenbaum.
And this brings me to my final point. Chari, Christiano, and Eichenbaum
use their model to argue--as was first expressed in these terms by Kydland
and Prescott--that a commitment technology is a wonderful thing.
The entire expectational dance of the private sector forming views and the
monetary authority not daring to do anything other than validate them could
be avoided, if only there could be commitment to a stable monetary policy
This point is very true, but it is also dangerous--as Britain before the
Falklands War shows.
- The danger emerges for the somewhat subtle reason that a noncredible
commitment technology is an awful thing. A government can cling to
a policy in the hope of enhancing its credibility for the future--as Herbert
Hoover, say, clung to the gold standard and to peacetime budget-balance
as policy goals from 1929-1932. But when private sector economic actors
do not believe that the commitment will be sustained, such policies inflict
a lot of short-run harm for no long-run good.
- Moreover, our true inflation control commitment technology
is the commitment to macroeconmic laissez-faire: that the government
would not respond to unemployment-driven economic distress with every tool
at its disposal, including inflation.
- This "technology" has been gone since at least 1925. Perhaps
before 1925 governments could claim to be bystanders at the business cycle,
and a credible commitment to a nominal price anchor at all cost was possible
(although consider the U.S. experience under the threat of free coinage
of silver; Friedman and Schwartz (1963) argued that it would have been
better to throw in the towel as far as the gold standard was concerned
once free silver became a political issue). But even Herbert Hoover boasted
of his use of the government's powers to manage the macroeconomy, and drew
a sharp contrast between his views and policies and those of his Treasury
Secretary, Andrew Mellon.
So I find myself both more and less optimistic than Chari, Christiano, and
Eichenbaum. More optimistic, in that the kind of expectational trap they
assume seems to me to be found only rarely in our world. Less optimistic,
in that the credible commitment technology they hope to use as a
cure does not seem to me to be found at all.
Go to Brad De Long's Home
Professor of Economics J. Bradford De Long, 601 Evans
University of California at Berkeley; Berkeley, CA 94720-3880
(510) 643-4027 phone (510) 642-6615 fax