The Census Bureau released its latest estimates of construction activity for the various sectors of the economy on Tuesday. Those data suggest that more spending in one sector tends to reduce spending in other sectors, contrary to the multiplier hypothesis put forward by Keynesian economists.
Last week I examined measures of the profitability of capital in the housing and business sectors and noted how, during the housing boom, profitability in the two sectors moved in opposite directions: housing profitability fell while business sector profitability rose.
The market appears to have built so many homes during those years because of an expectation that houses would be profitable in the future, and, in part, because home buyers were encouraged by easy credit. As a result, an ever-growing housing inventory was competing for the same demand for shelter, which kept rents low and vacancies high.
At the same time, it was more difficult for business to obtain capital, because so much of it was going into housing.
The battle for resources between the business and housing can easily be seen in the data for each sector’s structures investment (that is, construction activity), at least before 2009. The chart below shows quarterly data back to the beginning of the year 2000. A value of, say, 90 for either series means that structures investment, adjusted for inflation, was 90 percent of what it was in the first quarter of 2000 (for a related series, see this post).
The housing boom and bust are readily visible in the housing sector’s green line, which rose sharply through the end of 2005, and then fell even more sharply through 2009.
Nonresidential building normally increases over time with growth in the population and the economy, but it was low during the housing boom (see the red line in the chart), probably because so much capital was going into home building. Both residential and nonresidential investment turned at almost exactly the same time, in opposite directions. Nonresidential investment increased throughout 2006, 2007 and 2008, while residential investment was collapsing.
When President Obama was proposing his stimulus law in early 2009, he and his advisers contended that government spending would stimulate private spending.
Yet a number of economists warned that sectors compete with each other for resources, so government spending tends to displace – “crowd out,” in economics jargon – private spending. For example, some people employed by stimulus projects are people who quit their private-sector jobs to accept better positions funded by the new law.
Even the Obama administration, in its first Economic Report of the President (click here and scroll to page 124), pointed to the same data and noted how this seemed to be crowding out, at least before the recession. Admittedly, looking at the economy as residential versus nonresidential is not the same as looking at it as public versus private, but the former gives us some information about how economic activity spills from one sector to another.
Crowding out comes from scarcity of supply, and Keynesian economists assert that supply is not at all scarce during recessions. Perhaps they would point to the parallel movement of the residential and nonresidential building series after 2008 as evidence that crowding out stopped once the recession worsened.
However, even if the availability of former home builders was by itself encouraging nonresidential building, nonresidential building could fall during the recession because of declining demand. The large reduction in the workforce that became apparent by 2009, not to mention tight credit, is likely to have reduced the desired number of nonresidential buildings, and this, by itself, would cut nonresidential investment activity. So determining whether crowding out stopped after 2008 requires separating the effect of increased supply of resources for nonresidential investment from reduced demand.
In doing so I looked at the relationship before 2008 between building in the two sectors and the business cycle. Not surprisingly, nonresidential building follows a business cycle, going up and down with national employment.
As you might guess from Chart 1, more residential building is associated with less nonresidential building at a given point in the business cycle: each $3 billion of home building seems to crowd out nonresidential building by at least $1 billion. Now that’s crowding out: something that stimulated home building by $3 billion would raise total building activity, but only by $2 billion.
I used the pre-2008 relationships to predict nonresidential building during this recession and through the third quarter of last year, under two assumptions: that crowding out continued as before, and that crowding out ceased once the recession began.
Chart 2 shows the results, together with the same red actual investment data from Chart 1. The black line shows what would have happened to investment if it continued to move with employment, but crowding out disappeared. The blue line shows what would have happened to nonresidential investment if, in addition to moving with employment, crowding out continued at the same rate it did before the recession.
The predictions based on continued crowding out (the blue line) correctly anticipated this pattern, as well as the actual sharp drop to begin 2009. Nonresidential building did fall after 2008 but, thanks to the movement of resources away from housing, it fell less than it would have in a recession this deep.
The Keynesian model ignores the crowding-out effect, and thereby consistently underpredicts nonresidential building during the recession. Thus, all else the same, it appears that less spending in one sector is partly replaced by more spending in another.
Sometimes there are good reasons for the government to spend more, but Chart 2 shows how one of the costs of government spending is that it displaces spending in other sectors, even during a recession.