By New Deal Democrat
With one exception, I don't find construction spending a particularly useful metric.
The overall measure gives us almost no information about where the economy is going. Further, government construction spending has generally been steady over the long term, once we adjust for inflation. And nonresidential spending (for things like office buildings and plants) typically is only cut back once it dawns on the executives of the company involved that we might actually already be in a recession!
The one useful measure is residential construction spending.
A house gets built after the permit is taken out and/or after the buyer signs a construction contract, so while it is a leading indicator, it is not as leading as either of those two measures:
Typically, the statistic has risen and fallen at the same time as the metric of "dwellings under construction" from the survey of permits and starts, and diverges depending on the mix of single-family vs. multi-unit residences, and the cost mix of those residences (but note the recent divergence):
The special value of the residential construction metric is that there is very little "noise" in the number - even less than that of single-family permits. Its trajectory, whether at turning points or on trend, has typically been smooth for the 25-year duration of the series:
So we can use residential construction to verify, with a few months' delay, what we are seeing in permits and in particular single-family permits. Here's what that looks like over the past few years:
The longer-term trend has still been rising, but since June, residential construction spending has been basically flat. In the next several months, it should confirm (or not) if the peak we saw in single-family permits and new home sales from late winter is real.
New Deal Democrat, XE.com