The over $800 billion American fiscal stimulus package is enormously unpopular in many quarters. Yet for others, including some well-known economists, a second package is needed to make unemployment decline faster. It is very difficult to give a definite answer to the effectiveness of what has been spent so far, but I believe both theory and quite limited empirical evidence suggest that it is likely having a small “multiplier”.
According to simple Keynesian models, the rationale for increased government spending during a recession is straightforward. If governments employ underutilized labor and capital when they spend more, that provides an immediate boost to employment and output. A further stimulus comes from the spending by the owners of the unemployed labor and capital who benefit from the government spending. Their spending helps multiply the effects of the government spending still further, and so does spending by the recipients of this second round of spending. The total impact is the sum of all these separate boosts, and its ratio to the initial level of government spending is called the spending multiplier. A study by Dr. Christina Romer, Chair of the Council of Economic Advisers, published shortly before she took office claimed the spending multiplier during the recession at that time would be well over 1.0.
In reality, several major considerations neglected by this Keynesian analysis suggest a rather different multiplier. There is often a long delay between discussions of a fiscal stimulus, enactment of a law providing for the stimulus, and the actual spending. Dr. Romer discussed the broad dimensions of the current stimulus package even before she took her current position, and Congress passed the law providing for about an $800 fiscal stimulus shortly after President Obama took office. Yet it is now more than one year later, and only about a half of the stimulus package has been spent. In the meantime, the economy first hit bottom, and then has begun a strong recovery in GDP, and a modest recovery in unemployment. The long lag between discussion, enactment, and actual spending is an old criticism of fiscal stimulus even by very committed Keynesians.
Nor did the actual stimulus package voted by Congress correspond very closely to the theory behind the multiplier from government spending. Instead of being concentrated mainly on sectors that had suffered large increases in unemployment, such as construction, the stimulus spending was oriented toward many sectors that liberal members of the Democratic controlled Congress wanted to greatly expand. These include healthcare, schools, alternative energy sources, and medical and basic research. Some of this spending may have considerable social value, but it provides little stimulus since sectors like health, education, and wind power development were not hit hard by the recession, and hence have had low levels of unemployment. Government spending in these areas would tend mainly to bid labor away from either the private sector, or other government programs, or raise the earnings of those already employed. As a result, any multiplier from the stimulus package is likely to be less than one, not greater than one.
Even though the theory behind the multiplier has been around for a long time, few studies credibly estimate multipliers from different kinds of government spending. Part of the problem is that many other things are usually also changing when government spending changes, so it is difficult to find “natural” experiments where the effects of government spending are isolated from changes in other important variables. Some of the relevant factors that may be changing are prices, including wage rates and interest rates, consumption, and private investment.
Robert Barro of Harvard University has been estimating both spending and tax multipliers from past episodes, not from the current one. Using historical data obviously has many limitations from the point of view of evaluating the current stimulus, but such an evaluation is difficult because many aspects of the American economy have been changing during this recession along with stimulus spending, such as the monies spent on bank bailouts. Barro's estimated spending multipliers are based mainly on fluctuations in defense spending before, during, and after wars, such as World War II and the Korean War. He reports in a recent Wall Street Journal article (February 23) estimated spending multipliers over a two-year period following changes in defense spending. These multipliers equal about 0.4 following the first year, and 0.6 after two years.
They indicate sizable changes in overall GDP per $100 billion change in defense spending, but they also suggest that government spending significantly crowds out either personal consumption spending, private investment, or exports. Still, if these were the only effects of stimulus packages, multipliers of about 0.6 would indicate that larger government spending, such as the Obama stimulus package, can induce significant growth in GDP, presumably especially when unemployment rates are high.
However, greater spending has to be paid for eventually by higher taxes, including inflation taxes, since governments are subject to long term budget balance requirements, just as are households and businesses. Barro also estimates tax multipliers based on the effects on GDP of changes in average marginal tax rates from federal and state and local income taxes, and social security payroll taxes. These tax multipliers are about -1.1, which means that increases in marginal tax rates that raise tax revenue by say $300 billion would lower GDP by over $300 billion in the following year.
Putting these estimated spending and tax multipliers together, in recognition that government budgets have to be balanced eventually, indicates that the Obama stimulus package on balance would have a small, possibly even negative, overall effect on GDP. To be sure, these results have to be qualified since changes in nondefense spending, as in this stimulus package, may have larger multipliers than those estimated by Barro, although it is not obvious why that should be so. Moreover, if the spending occurs in 2009 and 2010, and raises GDP for a few years during a recession, perhaps that is worth any reductions in GDP that follows in later years when taxes are actually increased to produce the budget balance (although much of that effect might be anticipated earlier).
A few other discussions of the current stimulus package have much rosier conclusions about its effects on employment and GDP. But these estimates have little credibility because of the difficulties in disentangling the stimulus effects from those of Fed and Treasury spending on bailouts, many other factors that have been changing along with the stimulus spending, and the intrinsic strengths of the private sector of the American economy (see, for example, an article by Cogan, Taylor, and Wieland in the Wall Street Journal of September 17, 2009 for a discussion of some of these factors).
My conclusion is that the Obama-Congress stimulus was an attempt at reengineering government’s role in the economy that was badly designed and executed relative to the alleged goal of giving a short-term stimulus to the economy. Its design, and the general evidence on the effects of spending fiscal stimulus, should make Americans wary of any attempt to pass a second stimulus package, whatever form it takes.