Economic Essay
1
Rethinking the Role of Fiscal Policy
Martin Feldstein∗
As recently as two years ago there was a widespread consensus among economists
that fiscal policy is not useful as a countercyclical instrument. Now governments in
Washington and around the world are developing massive fiscal stimulus packages,
supported by a wide range of economists in universities, governments, and businesses.
Why has this change occurred? What are the principles for designing a potentially
useful fiscal stimulus? And what will happen if the current fiscal stimulus fails?
I. The Rise and Fall of Fiscal Activism
Despite a wide array of government programs introduced by the Roosevelt
administration during the depression of the 1930s, the unemployment rate remained at
double digit levels until 1941 when the government began massive military spending for
the lend lease program and the start of World War II. Economists saw this favorable effect
of the military spending on employment and economic activity as a clear example of the
power of Keynesian fiscal policy.
After the war, most American macroeconomists focused on the potential
contribution of Keynesian fiscal policy to preventing unemployment. For some, the new
∗ Professor of Economics, Harvard University, and President Emeritus of the National Bureau of Economic Research. This paper was prepared for presentation at the January 2009 meeting of the American Economic Association.
2
econometric models held out the hope of eliminating or at least significantly damping the
business cycle.
But further analysis and experience soon raised doubts about the efficacy of these
new tools. Empirical research indicated that the Keynesian multiplier was much smaller
than earlier analyses had assumed, reduced by a crowding out of interest sensitive
spending caused by an induced increase in the demand for money and by the effect of the
larger national debt on long‐term interest rates. The leakage of demand through imports
and the effect of the fiscal expansion on the exchange rate further reduced the multiplier.
Despite improvements in data and in econometric techniques, it remained difficult
for economists to assess the current state of the business cycle and even harder to discern
where the economy was heading and how much it would be affected by a fiscal stimulus.
The resulting uncertainty implied that an activist fiscal policy could actually increase
cyclical volatility. Moreover, the long lags between decisions to raise spending or cut taxes
and the subsequent fiscal flows often meant that the stimulus occurred after the trough in
activity, adding undesirable increases in demand to a rapidly expanding economy. The
simultaneous rise in both inflation and unemployment in the 1960s made it clear that the
Keynesian fiscal strategy was not working.
The focus of countercyclical policy therefore shifted from fiscal policy to monetary
policy. Economists recognized that monetary policy could be adjusted more rapidly and
that changes in the interest rate could be effective in modulating aggregate demand
through a variety of channels. The low inflation rate since the early 1980s and the
decreased cyclical volatility both reinforced the case for relying on monetary policy.
3
II. The Recent Revival of Fiscal Policy
Why then the recent revival of interest in fiscal stimulus? By the fall of 2007 it
became clear to many economists that the current downturn is different from previous
recessions and that monetary policy would not be effective in bringing us back to full
employment.
Past recessions generally began after the Federal Reserve had raised interest rates
sharply to counter excess inflation. When the Fed felt that it had succeeded, it reversed
policy and lowered the interest rate. That was enough to trigger a recovery, driven in large
part by the responsiveness of housing starts to lower interest rates.
In contrast, the current downturn was not due to high interest rates and therefore
could not be fixed by a reversal of Fed policy. The seeds of this recession were sown in the
underpricing of risk and the resulting excessive leverage. A sharp rise in defaults on
subprime mortgages alerted market participants that risk had been mismeasured and
therefore underpriced. The resulting general repricing of risk caused sharp falls in the
prices of mortgage‐backed securities, of share prices, and of the values of homes. The
massive destruction of household wealth led to a sharp decline in consumer spending,
followed by falls in business investment and commercial real estate values.
The continuing declines in the value of mortgage backed securities and of the
derivatives based on them reflected the fear of an increasing future volume of mortgage
defaults. This eroded the capital of financial institutions, undermining their willingness to
make loans. The result was a dysfunctional credit market that no longer provided credit or
responded to changes in interest rates.
4
The Federal Reserve therefore could not reverse the downturn by lowering interest
rates. Although it reduced the federal funds rate substantially, mortgage rates and
corporate bond rates remained high. Even if mortgage rates had come down, the
continuing sharp decline in house prices would have prevented a rise in housing starts.
By the end of 2007, an anticipation of the Fed’s inability to prevent a serious
recession caused some of us to advocate a fiscal stimulus. A temporary one‐time tax rebate
was chosen as something that could get bipartisan support and be implemented quickly.
We recognized that both economic theory and much past experience implied that most of a
one‐time tax cut would be saved rather than contributing to consumer spending. But we
were encouraged by the stronger response to the 2001 tax cut even though we recognized
that that was technically more than a one time tax change.
The Congress quickly passed a tax rebate of $80 billion and the money was in the
hands of taxpayers by May and June. Unfortunately, consumer spending responded only
very weakly. I presented evidence in the Wall Street Journal (August 6, 2008) that
consumer spending in the second quarter rose by only $12 billion. The monthly personal
income and expenditure data confirmed this picture, with consumer spending in May and
June rising a total of only $11 billion.
More recently, Stephen Miran and I estimated a consumer expenditure equation
using monthly data from January 1980 through November 2008. While the marginal
propensity to consume out of real per capita disposable income is estimated to be 0.70, the
estimated MPC from the corresponding rebate variable is only 0.13 (standard error 0.05).
(The other variables in the equation are the unemployment rate, the 10‐year interest rate,
5
and a quadratic time trend.) A variety of short distributed lag specifications confirms this
result and indicate that there is no delayed impact of the rebate; all of the monthly lag
coefficients are completely insignificant and their sum is negative.
In recent months, the Federal Reserve and the Treasury have taken a variety of
steps to help the credit markets. These policies have succeeded in preventing a further
meltdown of credit availability in banks, money market mutual funds, and in the ability of
firms to issue commercial paper. But these measures have neither expanded total credit
nor dealt with the fundamental problem of a dysfunctional credit market caused by the
foreclosures that result from the rising loan to value ratios on non‐recourse mortgage
loans. Although fixing the credit market is necessary for long‐run sustained growth, would
not be sufficient to reverse the downward spiral of aggregate demand.
III. Designing the Current Fiscal Package
This brings us to the current situation and the perceived need for a large fiscal
package. The fall in the stock market and in the value of owner occupied real estate has
depressed household wealth by about $10 trillion. The estimated wealth effects imply a
decline of annual consumer spending by $400 billion or more. That reduction in consumer
spending implies reduced production, lower incomes, and therefore further reductions in
consumer spending. This could reduce aggregate demand by an additional $200 billion a
year or more. Automatic stabilizers – i.e., the reductions in personal and corporate taxes
and the increases in unemployment insurance and other transfers – probably offsets about
one‐third of this, leaving a net GDP gap of about $400 billion.
6
The current decline in demand is different from typical past business cycles in
which demand recovers as inventories and excess capacity is absorbed.
So that is the challenge: how to increase domestic spending by some $400 billion a
year in 2009 and 2010, and perhaps further into the future. Some of that may come from a
more competitive dollar, although the increased competitiveness of the dollar may only be
enough to offset the decline in export demand that results from the reduced level of foreign
incomes. So it falls to fiscal policy to support the increase in aggregate spending.
Some of the past problems in using fiscal policy to stimulate demand may be less of
an impediment in the current circumstances. Government borrowing to finance fiscal
deficits will not be offset by higher interest rates since the current environment is
characterized by very easy money and a dysfunctional credit market. The delays in starting
infrastructure projects and the long tail in that spending are not likely to be as much of a
problem now because the current downturn is likely to last much longer than previous
ones. In the past, the average recession lasted only 12 months from peak to trough. This
recession has already lasted 12 months and probably will last a good deal longer. I believe
we will be lucky if we see the recession end in 2009. Once the recovery begins, the upturn
will be very slow because households need to increase their saving – i.e., to consume less ‐‐
to rebuild their wealth for retirement and other purposes. So fiscal policy is likely to be
useful even if it is not strongly effective in 2009. It is not likely to overheat the economy if it
continues to add significantly to demand in 2010 and 2011.
Although a one‐time tax cut may not be effective, other forms of tax cutting can
increase aggregate demand. During his campaign, candidate Obama promised a permanent
7
tax cut of $500 per employed person. That would generate an annual tax cut of about $70
billion and would probably raise annual consumer spending by about $50 billion.
Experience confirms that some form of investment tax credit could stimulate
business investment, especially if it is not recaptured later. A larger R&D tax credit could
help to offset the currently predicted decline in private R&D spending. And lowering the
corporate tax rate to that of other industrial countries would encourage more business
investment and job creation in the United States.
The president elect announced that he would postpone increasing the tax rate on
high‐income individuals until 2011. But taxpayers, especially higher income ones, look
ahead. The future tax rise reduces the present value of their lifetime income and that can
be expected to reduce current spending. A statement now by the president‐elect that he
will postpone those tax increases indefinitely would raise aggregate spending now.
Finally, the taxes on dividends and capital gains are also scheduled to rise in the
near future. A promise to leave those tax rates unchanged would raise share prices,
offsetting some of the fall in the stock market, which would lead to more consumer
spending and increased business investment.
But while good tax policy can contribute to ending the recession, the heavy lifting
will have to be done by increased government spending. To be effective, that spending
should be big, quick, and targeted at increasing aggregate activity and employment. How
big depends on the form of the spending and the timing. But with low multipliers and some
relatively long spending tails, annual outlays of $300 billion to $400 billion seem like a
reasonable target for government spending in 2009 and 2010.
8
The speed of the outlays is an important consideration. A project that begins in
2009 but continues to spend at a high level in 2011 and 2012 is not likely to be as useful as
a countercyclical instrument as one that spends quickly and is then finished.
Bottlenecks are also a potential problem that could reduce the effectiveness of a
spending program. While there is no doubt a need to rebuild bridges and other
infrastructure, there are limited numbers of design engineers and other bridge builders.
IV. Spending Priorities
The Obama campaign has identified five priority areas for increased spending:
health, energy, education, infrastructure , and support for the poor. Some of that spending
would be by the federal governments but much of it would be delegated to the states and
local governments. Although these are important areas that can benefit from increased
spending, there are other parts of the budget that could also be useful as part of the
stimulus package.
Since the defense budget is as large as all of the other discretionary spending
combined, it is surprising that defense is not proposed as a part of the overall stimulus
package. It is surprising also to read in the press that there will be reductions in military
spending because, according to those stories, of the weakness of the economy. That logic is
exactly backwards. The overall weakness of demand in the economy implies that the next
two years are a time when military spending should rise.
The actions of the military in Iraq and Afghanistan have depleted supplies and
increased the wear and tear on equipment. Both supplies and equipment will eventually
9
need to be replaced. Now is the right time to do that. More generally, maintenance and
replacement schedules in both the military and civilian departments should be revised to
increase early maintenance and replacement while there is substantial slack capacity,.
Military recruiting and training could be expanded in response to the larger than
usual numbers of unemployed young men and women. Raising the military’s annual
recruitment goal by 15 percent would provide jobs for an additional 30,000 young men and
women in the first year. It would also be possible to depart from the military’s traditional
enlistment rules and bring in recruits for a short two‐year period of training followed by a
return to the civilian economy. As a minimum, this would provide education in a variety of
technical skills – electronics, equipment maintenance, computer programming, nuclear
facility operations, etc – that would lead to better civilian careers for this group. It would
also provide a larger reserve force that could be called upon if needed by the military in the
future.
A 10 percent increase in defense outlays for procurement and for research would
contribute about $20 billion a year to the overall stimulus budget. A 5 percent rise in
spending on operations and maintenance would add an additional $10 billion. That
spending could create about 300,000 additional jobs. And raising the military’s annual
recruitment goal by 15 percent would provide jobs for an additional 30,000 young men and
women in the first year.
The intelligence community and the FBI are also apparently facing potential budget
cuts at a time of increasing terrorism and greater crime rates. A temporary increase in
funding for these agencies could fill important gaps in training and facilities.
10
Another important omission in the current stimulus plan is funding for research.
Government spending for research is projected to fall in 2009 even though additional
research grants from the NIH and NSF could allow universities and hospitals to expand a
wide range of useful research activities that are now unfunded because of limited grant
budgets.
No doubt there are other important areas of government spending in which outlays
can be raised rapidly on useful activities that would also raise incomes and employment. In
each area, government budgeting must go beyond business as usual if it is to respond
appropriately to the opportunity for a short‐term spending surge.
V. What if it fails?
It is of course possible that the planned surge in government spending will fail. Two
three years from now we could be facing a level of unemployment that is higher than today
and that shows no sign of coming down.
While it is too soon to examine in detail what might then be done, it is useful to
consider the three possibilities. First, the level of government spending could be increased
even more. To know whether this would help, it is important to study in detail the
effectiveness of each of the different components of the spending surge. Second, the fiscal
stimulus could shift from increased spending to a substantial permanent reduction in
personal and corporate taxes. If this strategy is chosen, changes in spending policies have
to be adopted to limit the growth of the national debt. The third way out would be a fall in
the value of the dollar, either spontaneous or planned, that is large enough to eliminate
11
today’s large trade deficit, thus boosting exports and substituting American made goods
and services for imports from the rest of the world.
While these possibilities should be kept in mind, we can only hope that the new
program of tax changes and government spending in combination with mortgage market
reforms will be sufficient to return the economy to full employment.
Cambridge MA