If the government increased its spending on infrastructure when the economy has full employment, its main impact would likely be to draw labor, capital, and raw materials away from various other activities. In effect, increased government spending under these employment conditions would "crowd out" private spending. Measured GDP would not be much affected, if at all. To be sure, the efficiency of the economy would rise if too little had previously been invested in this infrastructure, while efficiency would fall if this government spending were more wasteful than the private spending that was crowded out.
This analysis is a useful starting point to consider the effects of stimulus packages, such as the one proposed by soon-to-be President Obama. Of course, the present situation is not one of full employment but of underemployment and excess unemployment, and employment is still falling. How does one adjust the full employment analysis in the first paragraph to account for the presence of unemployed labor and capital? One extreme assumes no crowding out of other private spending when governments increase their spending with significant underemployment in the economy. Increased government spending through a stimulus package under these conditions might even have a "multiplier" effect that would greatly increase, not crowd out, other private spending. The reason is that the recipients of the government spending in turn would increase their spending, and thereby stimulate other activities. Intermediate assumptions assume partial crowding out of other private activities, so a stimulus package would still increase employment and GDP. However, the value, if any, of the increase would depend on how effectively governments spend the stimulus compared to the private spending that is crowded out.
Various assumptions about multipliers and crowding out, some implicit, are found in a recent "official" evaluation ("The Job Impact of the American Recovery and Reinvestment Plan") of the effects on GDP and jobs of President Elect Obama's stimulus package. The authors- Christina Romer (incoming Chair of the Council of Economic Advisers) and Jared Bernstein (of the incoming Vice-President's staff)- assume in their calculations a stimulus package that spends a little over $775 billion on energy, infrastructure, health care, tax cuts, and direct payments to the unemployed and other low income individuals. This stimulus is about 7% of the real GDP of about $12 trillion that they estimate for the 4th quarter of 2010 without any stimulus. After working through their analysis, they conclude that this stimulus package will raise real GDP by 3.7 percent in the 4th quarter of 2010 compared to the situation without a stimulus package (Table 1, p.4), so that there is some significant crowding out of private spending. They also assume that this 3.7 % increase in GDP would raise jobs at that time by about 31/2 million. According to their calculations, with the stimulus package, unemployment would be at about 7% in the 4th quarter of 2010 instead of about 9 % without the stimulus.
Are these estimates reasonable? Let me first admit that in recent years I have not followed either the academic macroeconomic literature that estimates multipliers of different kinds from various spending and tax programs, or the literature that explicitly estimates crowd out effects of increased government spending. Moreover, Romer and Bernstein claim that they assume basically the same multipliers used in the Federal Reserve's FRB/US model, and by a leading private forecaster.
Nevertheless, I believe that they overestimate the effects of this stimulus package on the economy, and that the same techniques would similarly overestimate the employment effects of other types of government spending and tax reduction policies. One strange assumption in the Romer and Bernstein analysis is their assumption that households treat temporary tax cuts as permanent, although they admit that temporary tax cuts are mainly saved and not spent (p.6). However, even without any stimulus from tax cuts to households and from business tax incentives, they still get an increase in 2.7 million jobs from this stimulus package (Table 2, p.6). This is because in their calculations direct spending programs, such as on infrastructure or education, have the biggest effects on jobs per dollar of stimulus.
Perhaps their estimates of the stimulus provided by direct government spending are in the right ballpark, but I tend to believe that they are excessive. For one thing, the true value of these government programs may be limited because they will be put together hastily, and are likely to contain a lot of political pork and other inefficiencies. For another thing, with unemployment at 7% to 8% of the labor force, it is impossible to target effective spending programs that primarily utilize unemployed workers, or underemployed capital. Spending on infrastructure, and especially on health, energy, and education, will mainly attract employed persons from other activities to the activities stimulated by the government spending. The net job creation from these and related spending is likely to be rather small. In addition, if the private activities crowded out are more valuable than the activities hastily stimulated by this plan, the value of the increase in employment and GDP could be very small, even negative.
As Posner and others have indicated, there appears to have been a huge conversion of economists toward Keynesian deficit spenders, but the evidence that produced such a "conversion" is not apparent (although maybe most economists were closet Keynesians all along). This is a serious recession, but Romer and Bernstein project a peak unemployment rate without the stimulus of about 9%. The 1981-82 recession had a peak unemployment rate of about 10.5%, but there was no apparent major "conversion" of economists at that time. What is so different about the present recession compared to that one, and to other recessions since then, that would greatly raise the estimated stimulating effects of government spending on various types of goods and services?
It is relevant in answering this question that the origins of this recession were in the financial sector, and especially in the excessive mortgage credit to sub prime and other borrowers. The widespread collapse of the financial sector, and the wholesale retreat from risky assets, clearly has called for a highly pro-active Fed. But it is not obvious why this should lead to greater confidence in the power of government spending stimulus packages. Of course, perhaps the prior emphasis on crowding out, and skepticism toward the stimulating effects of government spending, were wrong, or that recessions were too short and mild after the 1981-82 recession to call for Keynesian-type stimulus packages.
Time will tell whether I am right that a spending and tax package of the type analyzed by Romer and Bernstein may stimulate the economy as measured by GDP and employment, but that the stimulus will be smaller then they estimate, and its value to consumers and taxpayers could be even smaller.