There are two issues on the spending side to be considered. First, Ramey (2011)
argues that the VAR analysis of Blanchard and Perotti (2002) tends to overestimate
spending multipliers, because they do not distinguish between expected and unex-
pected government spending. They treat all government spending as exogenous
without taking into account that some of it is anticipated and some of it is not, which
should have a different effect on aggregate demand. By not distinguishing between
the two, the spending multiplier is overestimated. Thus, the Blanchard and Perotti
(2002) spending multiplier of around 1.0 is too high because they overestimated it.
On the other hand, in a recent paper, Perotti (2012) argues that estimating the effect
of government spending using military spending is not a good idea because it is
driven by only a very few years with major wars, such as World War II, the Korean
War, and the Vietnam War. If the analysis relies on military spending, it has too few
observations. More importantly, during periods of war there are a lot of other things
taking place, like price controls and nationalistic feelings, that cause people to work
harder. The war period, they argue, is not a good or normal period to start measuring
the effect of fiscal policy. Validation is difficult because during a war there are big
changes in government spending.
Therefore, the debate is still going on. But again, if the spending multiplier is not
much larger than 1.0, then how valuable is it to increase discretionary government
spending during a recession? Obviously, one major application in the discussion has
to do with the effect of the expansi onary government spending program that recently
occurred in the U.S. I think it will be close to impossible to give an answer about
whether the expansionary fiscal policy in the U.S. has been valuable in reducing the
size of the recession. First of all, we do not even agree on how large the amount of
government spending was. For example, some of the spending on the federal level
compensated for cuts in state level spending. Therefore, it is not clear how large the
stimulus actually was. Second, we will never have a counterfactual, namely, what
would have happened without it.
Leaving that aside, one can look at the unemployment rates that the administration
predicted in 2008 without the recovery plan and the lower unemployment rate it
predicted with the stimulus package. In March 2010, the unemployment rate was
9.7 %, which was way above what the administration had predicted it would be
without the recovery plan. So, the recovery plan seemed to have no effect. In March
2011, the unemployment rate was 8.8 %, which is exactly what it was predicted to be
without the recovery plan. There are several reasons why the administration’s pre-
dictions about the unemployment lev els with the stimulus plan failed. First, the
projections might have been wrong. Second, perhaps the unemployment rate during
the recession went up more than one could have predicted relative to the reduction in
GDP growth. The unemployment rate went up and stayed up longer than one would
have anticipated. This does not mean that the recovery plan did nothing, but it simply
means that the prediction was wrong. Nevertheless, there was a stimulus package, and
the unemployment rate was high and stayed high.
There is a similar debate going on about taxes. Romer and Romer (2010) find that
tax multipliers are very large. However, recent resear ch shows that they are probably
actually much lower. Another issue is that tax, spending and multipliers can be quite
different depending on the level of debt of the country, and depending on whether or
not the country is in a deep recession. There is a view that when you are in a deep
432 A. Alesina