The Microeconomics of "Stimulus" Policy
But couldn’t economic productivity be increased by targeting federal spending on hiring the unemployed either directly to work for government or by subsidizing private firms to hire them? Such an approach makes sense only if it produces more value than it costs, and there are several reasons for doubting that it does. First, with the federal government spending well over 20 percent of GDP, and most of this spending reducing economic productivity (spending additional dollars creates less value than it costs), it is unlikely that there are many government jobs left in which additional workers would add to the net productivity of the economy.
Second, assuming that there are government jobs in which the right people could create more value than their opportunity costs, without reliable market prices and wages guiding political decisions, it is very unlikely that political authorities would identify those jobs and match them with the right workers. This would be a problem even if the information were available to place government workers in jobs where they would be most productive. Political influence is far more important than economic productivity when officials decide what government jobs to create and on how much to pay those who are hired. This political influence is also dominant when private firms are subsidized to reduce unemployment by hiring more workers. Those subsidies are more likely to go to firms in politically favored industries that have been generous campaign contributors. Also, workers hired for federally funded or assisted construction projects are required by the 1931 Davis-Bacon Act to be paid the prevailing union wage, which is invariably higher than the market wage.
This is from this Dwight Lee, “Reducing Real Output by Increasing Federal Spending,” this month’s Featured Article on Econlib. The whole thing is worth reading.