I find each Casey Mulligan post more baffling than the last. Yesterday's post was on the failure of some studies praised by Krugman on stimulus spending that looked at cross-state variation.
He starts the post off by mentioning how moving Yankees stadium increased jobs in one area but decreased jobs in another. So far so good. He went on to say that sports stadiums displaces spending on other entertainment. Fair enough, he doesn't claim displacement is 100% and I'd agree there is a lot. I don't go to many sporting events, but even if I discover a really great restaurant (which is more my weakness) I probably only go out once or twice a year more because of it so I can buy displacement would be very high.
On the other hand, when I hear about a really great book I tend to go out and buy it over and above my normal book budget without checking my bank balance first, so here, displacement doesn't really exist (though my savings may be slightly less, but channeling this back towards demand is very, very indirect).
Back to where he's making sense, he points out that cross-state variation studies on the stimulus should take the national level into account. Fair enough, but if we're talking aggregates G as a whole dropped due to decreases in state spending. Since both state and government spending are very similar as a first approximation I'd expect the displacement effect to be very high (someone may have studied this, I'm just guessing). So I don't know what would have to be looked at at the national level, it seems like state level comparisons might be far more appropriate here because we can get cases where net G increased at the state level, which did not happen at the national. There may of course be some displacement between states but only to the extent where similar resources are being bid for, which may or may not be the case.
Of course, in some circumstances I would expect strong displacement effects from government spending. A newly industrializing nation would probably see a strong displacement effect from both government and the private sector placing demands on cement, steel, and other construction goods. If the employment rate were high, there would also be a large amount of displacement as wages get built up. But the amount of displacement should be tightly linked to the similarity of the real goods being demanded and supplied, the more different they are, the lower the displacement should be. Since the labor pool is differentiated, we shouldn't expect much competition at high levels of unemployment, the displacement would only occur when both are competing for workers that both find marginal and not particularly well suited for the need of either.
Looking back historically, I find a lot of evidence for demand driven rather than supply driven impacts. One of my favorite books is Jan de Vries' The Industrious Revolution (bah links still won't work) which looked at the household economy and how households massively increased their labor supply as demand for goods such as sugar, tea, and coffee rose. This drove both the supply of goods as well as changes in household structure that allowed for far more labor to be mobilized to obtain goods. There is far more friction today, few of us have many options to increase our individual labor supply for more goods, but the government has the ability to play a similar role by putting idle resources to work producing goods. Assuming that government is not bidding for scarce resources, there should be little displacement effect.
Also, historically, if displacement were a significant drag on employment and growth we should expect states with small states to be the big ones today. This isn't the case. Great Britain had one of the largest states in the 18th century before the industrial revolution with a tax take of about 20% compared to 10 to 13% for France, 4 to 8% in China, and 4.5% falling to 2 to 3% late in the century, in the Ottoman Empire. While distant from the main question, long run historical evidence doesn't fit the idea that state spending generally decreases growth and employment, even in the long run. The opposite seems to be the case.
Of course, the idea that government spending displaces growth must come from somewhere. I can assume that it does in fairly marginal and uncommon conditions, such as full employment or rapid industrialization where resources are near full utilization. I also assume the displacement objection is using money as a stand in for real resources. But this was never accurate, states can make money and when under pressure so do individuals. During currency shortages in the Ming era, before the transfer of silver got under way from the west, parts of China reverted to cowrie shells as a means of exchange. I am reading a book on Africa that mentions briefly how this happened in certain parts of Africa as well. This is quite consistent with the Keynesian position, when demand for the means of exchange are high, resources become idled unless money is created. Historically, even pre-modern societies realized this and responded, if very inefficiently (though the Song Chinese use of paper currency functioned quiet well at monetizing the economy for over a century and only fell apart due to military defeat and the loss of half their territory, I doubt the US dollar would hold value either if half the country was seized by foreign forces).
I'm not sure these digressions will convince anyone, but my point is there are multiple strands of evidence pointing to the importance of demand and how weak displacement effects are. Size of government just isn't well correlated with growth over long run, historical time frames. Displacement requires demand for similar resources, if government starts building roads as long as private industry isn't facing a shortage of steel and asphalt I see no reason for investment to be crowded out. Money is an abstract representation of real resources and not a very accurate one, the US government printing money is equivalent to Chinese traders using cowrie shells to meet a shortage in copper cash. If it gets otherwise idle resources working, there's no reason that displacement will occur.