Bill McBride, aka Calculated Risk, looks at vacancy rates and concludes

This suggests there are still close to 1.4 million excess housing units.

One of the difficult things in making economic forecasts is trying to figure out how far to drill down. Most forecasts revolve around the notion that there are fundamental behavioral regularities in markets, firms and individuals.

Right now Bill is looking at how many homes are empty. There are more empty homes than usual and this suggests that there are more homes than folks looking for a home. This in turn suggests that the impetus for building new homes will be mild and we should expect mild residential construction over the near term.

This is what I would think of as a medium drill down and it has certain forecast risks. The most obvious is that something odd may be affect the number of people looking for a new home. If that changes suddenly and more people start looking for a home, then you could have a shocking fall in vacancy rates, which then produces a rise in the demand for new construction.

I have been using a slightly deeper drill down that looks at population and the flow of housing. Under that comparison people per home is starting to rise above its historical average, and at the current pace of construction is set to make records over the medium term. Thus I see a strong demand for new construction building.

This drill down has risks as well, and they are basically the inverse. What if there is a “new normal” in household size. In that case I will be waiting for a boom that never comes. What changes instead is the way people live.

In either case the forecast risk is that you have picked the wrong regularity to hang your hat on. This runs across virtually all forecasting domains that I can think off. In every case you have to assume that certain things are attractors to which the economy will always be brought back and certain things bounce around from shocks.