"Goverment Spending > Tax Cuts," says the nonpartisan CBO.
February 4, 2009
Honorable Judd Gregg
Ranking Member
Committee on the Budget
United States Senate
Washington, DC 20510
Dear Senator:
At your request, the Congressional Budget Office (CBO) has conducted an
analysis of the macroeconomic impact of the Inouye-Baucus amendment in the
nature of a substitute to H.R. 1. CBO estimates that this Senate legislation would
raise output and lower unemployment for several years, with effects broadly
similar to those of H.R. 1 as introduced. In the longer run, the legislation would
result in a slight decrease in gross domestic product (GDP) compared with CBO’s
baseline economic forecast.
Effects on Output and Employment
The macroeconomic impacts of any economic stimulus program are very
uncertain. Economic theories differ in their predictions about the effectiveness of
stimulus. Furthermore, large fiscal stimulus is rarely attempted, so it is difficult to
distinguish among alternative estimates of how large the macroeconomic effects
would be. For those reasons, some economists remain skeptical that there would
be any significant effects, while others expect very large ones.
CBO has developed a range of estimates of the effects of the Senate legislation on
GDP and employment that encompasses a majority of economists’ views.
According to these estimates, implementing the Senate legislation would increase
GDP relative to the agency’s baseline forecast by between 1.2 percent and
3.6 percent by the fourth quarter of 2010. It would also increase employment at
that point in time by 1.3 million to 3.9 million jobs, as shown in Table 1. In that
quarter, the unemployment rate would be 0.7 percentage points to 2.1 percentage
points lower than the baseline forecast of 8.7 percent. The effects of the
legislation would diminish rapidly after 2010. By the end of 2011, the Senate
legislation would increase GDP by 0.4 percent to 1.2 percent, would raise
employment by 0.6 million to 1.9 million jobs, and would lower the
unemployment rate by 0.3 percentage points to 1.0 percentage point.
Those estimated effects differ modestly from CBO’s estimates for H.R. 1 as
introduced.1 In particular, the effects on output and employment are slightly
higher in 2009 and 2010, but slightly lower in 2011. The differences stem from
three main sources. First, the Senate legislation’s provisions regarding the
alternative minimum tax (AMT), which do not appear in the House bill, would
add stimulus to the economy, especially in 2010. Second, the Senate legislation
would allow faster spending from the State Fiscal Stabilization Fund, increasing
such spending by about $20 billion over the 2009-2010 period compared with that
under the House bill (and decreasing spending correspondingly in the following
years). And last, the estimated decrease in withholding (and thus the reduction in
revenues) associated with the Making Work Pay Credit would be greater in 2009
under the Senate legislation than under H.R. 1.
Effects of Various Types of Legislative Provisions on Output
Although the Senate legislation has numerous detailed provisions, the
macroeconomic effects can be illustrated by considering the provisions in seven
categories. Table 2 shows the range of estimated effects on the economy-the
multiplier effects-of a one-time increase of a dollar of additional spending or a
dollar reduction in taxes. For all of the categories that would be affected by the
Senate legislation, the resulting budgetary changes are estimated to raise output in
the short run, albeit by different amounts.
The numbers in Table 2 indicate the cumulative impact on GDP over several
quarters. For example, a one-time increase in federal purchases of goods and
services of $1.00 in the second quarter of this year would raise GDP by $1.00 to
$2.50 in total over several quarters, with most of that effect in the first two
quarters and little effect beyond a year.
As shown in the first two categories in the table, direct purchases of goods and
services by governments, including investment in infrastructure, tend to have
relatively large effects on GDP. Because infrastructure spending takes time to
occur, increased funding for that purpose would not boost outlays or GDP much
this year, but it would probably provide significant stimulus from 2010 through
2012.
Grants to state and local governments (such as increased assistance for education)
might not increase state spending for the programs designated in the grants but,
instead, might free up funds that the states would otherwise spend on those
programs. States could use those extra funds in a variety of ways: direct purchases
of goods and services (or smaller cuts in such purchases), tax cuts (or smaller tax
increases), transfer payments, or reduced borrowing. The impact of grants
therefore would depend on how states used them.
Transfers to persons (for example, unemployment insurance and nutrition
assistance) would also have a significant impact on GDP. Transfers have a
relatively strong effect on consumption because they tend to go to people, such as
the poor or unemployed, who are likely to spend much of any additional income.
For that reason and because transfers can be increased quickly, they are estimated
to have a significant impact on GDP by early 2010. Transfers also include
refundable tax credits, which have an impact similar to that of a temporary tax
cut.
A dollar’s worth of a temporary tax cut would have a smaller effect on GDP than
a dollar’s worth of direct purchases or transfers, because a significant share of the
tax cut would probably be saved. The amount saved, and therefore the size of the
effect on GDP, would depend on who received the tax cut and how temporary it
would be. Most households probably save most of a temporary tax cut, to keep
their purchases relatively smooth over time. However, the predominantly lower-
income households that spend all of their income and would like to borrow funds
to spend more if they could (that is, households that are “liquidity constrained”)
probably spend a large share of temporary boosts to income. In addition, the
longer a tax cut is expected to last, the greater the impact on total after-tax
income, and the larger the likely effect on consumption.
CBO’s analysis divides the temporary tax cuts in the Senate legislation into those
that would go primarily to higher-income households and last for only one year
(mostly the provisions affecting the AMT) and those that would go primarily to
lower- and middle-income households and last for two years (predominantly the
Making Work Pay Credit), with the former having a considerably lower range of
multipliers than the latter. Taken together, the temporary nonbusiness tax cuts in
the Senate legislation would reduce revenues much more in 2010 than in 2009
because much of the reduction in taxes would be realized by households when
they filed their returns in 2010.
The provision for greater tax-loss carrybacks would result in a large up-front cost
to the government, but the effect of that provision on business spending would
probably be small because it primarily would affect firms’ after-tax income rather
than their marginal incentives for new investment. Therefore, the effect of the
provision on revenues would be significantly greater than its effect on the
economy.
The Relationship Between Output and Employment
CBO derived its estimates of the effect of the Senate legislation on employment
from the estimated effect on GDP. Historical evidence suggests that GDP growth
that is 1 percentage point faster over a year (relative to a baseline forecast) will
cause the unemployment rate to decline by a little more than half a percentage
point (relative to a corresponding baseline forecast). The fall in the unemployment
rate leads more people to enter the labor force and seek jobs and fewer to drop
out. Therefore, employment rises both from a decline in the number of
unemployed workers and a decline in the number of people out of the labor force.
In addition, some workers otherwise working part time move to full-time status.
The change in employment relative to the change in GDP in CBO’s estimates is
small compared with that in most industry-based studies of stimulus. By the end
of 2010, CBO estimates, about $140,000 of additional GDP would lead to one
additional person employed. That relationship is similar to those indicated by
other macroeconomic studies of stimulus proposals.2 However, a number of other
sorts of studies imply more employment per dollar of additional GDP. Because
the macroeconomic studies use the historical relationship between changes in
economic growth and changes in jobs, they incorporate a number of broad
economic effects. For example, output per employee tends to fall in a recession
because employers try not to fire their best workers even as they cut production in
response to decreased demand. Therefore, as fiscal stimulus increases demand,
firms can ramp up production without increasing employment proportionally.
Historical evidence thus suggests that fiscal stimulus boosts both productivity and
hours of work as well as employment. Studies that ignore those effects are likely
to overstate the impact of fiscal stimulus on employment.
Long-Run Effects on Output
Most of the budgetary effects of the Senate legislation occur over the next few
years. Even if the fiscal stimulus persisted, however, the short-run effects on
output that operate by increasing demand for goods and services would eventually
fade away. In the long run, the economy produces close to its potential output on
average, and that potential level is determined by the stock of productive capital,
the supply of labor, and productivity. Short-run stimulative policies can affect
long-run output by influencing those three factors, although such effects would
generally be smaller than the short-run impact of those policies on demand.
In contrast to its positive near-term macroeconomic effects, the Senate legislation
would reduce output slightly in the long run, CBO estimates, as would other
similar proposals. The principal channel for this effect is that the legislation
would result in an increase in government debt. To the extent that people hold
their wealth as government bonds rather than in a form that can be used to finance
private investment, the increased debt would tend to reduce the stock of
productive capital. In economic parlance, the debt would “crowd out” private
investment. (Crowding out is unlikely to occur in the short run under current
conditions, because most firms are lowering investment in response to reduced
demand, which stimulus can offset in part.) CBO’s basic assumption is that, in the
long run, each dollar of additional debt crowds out about a third of a dollar’s
worth of private domestic capital (with the remainder of the rise in debt offset by
increases in private saving and inflows of foreign capital). Because of uncertainty
about the degree of crowding out, however, CBO has incorporated both more and
less crowding out into its range of estimates of the long-run effects of the Senate
legislation.
The crowding-out effect would be offset somewhat by other factors. Some of the
Senate legislation’s provisions, such as funding for improvements to roads and
highways, might add to the economy’s potential output in much the same way that
private capital investment does. Other provisions, such as funding for grants to
increase access to college education, could raise long-term productivity by
enhancing people’s skills. And some provisions would create incentives for
increased private investment. According to CBO’s estimates, provisions that
could add to long-term output account for roughly one-quarter of the legislation’s
budgetary cost.
The effect of individual provisions could vary greatly. For example, increased
spending for basic research and education might affect output only after a number
of years, but once those investments began to boost GDP, they might pay off over
more years than would the average investment in physical capital (in economic
terms, they have a low rate of depreciation). Therefore, in any one year, their
contribution to output might be less than that of the average private investment,
even if their overall contribution to productivity over their lifetime was just as
high. Moreover, while some carefully chosen government investments might be
as productive as private investment, other government projects would probably
fall well short of that benchmark, particularly in an environment in which rapid
spending is a significant goal. The response of state and local governments that
received federal stimulus grants would also affect their long-run impact; those
governments might apply some of that money to investments they would have
carried out anyway, thus freeing funds for noninvestment purposes and lowering
the long-run economic return to those grants. In order to encompass a wide range
of potential effects, CBO used two assumptions in developing its estimates: first,
that all of the relevant investments together would, on average, add as much to
output as would a comparable amount of private investment, and, second, that
they would, on average, not add to output at all.
In principle, the legislation’s long-run impact on output also would depend on
whether it permanently changed incentives to work or save. However, according
to CBO’s estimates, the legislation would not have any significant permanent
effects on those incentives.
Including the effects of both crowding out of private investment (which would
reduce output in the long run) and possibly productive government investment
(which could increase output), CBO estimates that by 2019 the Senate legislation
would reduce GDP by 0.1 percent to 0.3 percent on net. H.R. 1, as passed by the
House, would have similar long-run effects. CBO has not estimated the
macroeconomic effects of the stimulus proposals year by year beyond 2011.
Other Effects of Stimulus Proposals
It is important to note that effects on GDP, the aggregate domestic output of the
economy, do not necessarily translate into effects on people’s well-being. First,
the part of GDP that contributes directly to people’s welfare is consumption.
However, changes in GDP do not necessarily imply corresponding changes in
consumption. For example, if GDP rises because foreigners finance greater
investment, much of the additional income generated by the investment will flow
overseas as payments to foreigners and will not be available to support higher
consumption.
More fundamentally, many things that make people better off do not appear in
GDP at all. For example, healthier children or shorter commute times can improve
people’s welfare without necessarily increasing the nation’s measured output in
the long run (though spending in those areas would still provide short-run
stimulus). Even legislation explicitly intended to affect output may also seek to
accomplish other goals and can be evaluated accordingly.
I hope this information is helpful to you. If you have any further questions, I
would be glad to answer them. The staff contacts for the analysis are Ben Page
and Robert Arnold, who may be reached at (202) 226-2750.
Sincerely,
Douglas W. Elmendorf
Director
Identical letter sent to the Honorable Charles E. Grassley.
Source
"Including the effects of both crowding out of private investment (which would reduce output in the long run) and possibly productive government investment (which could increase output), CBO estimates that by 2019 the Senate legislation would reduce GDP by 0.1 percent to 0.3 percent on net. H.R. 1, as passed by the House, would have similar long-run effects. CBO has not estimated the macroeconomic effects of the stimulus proposals year by year beyond 2011."
Here's the thing: 0.1 to 0.3% of GDP, TEN YEARS FROM NOW, is a MINOR cost to stimulate employment and GDP growth from 2009 to 2011:
