The Most Important Residential Construction Metric Never Discussed

There are many data points that are referenced by executives, investors and reporters to interpret the direction and position of the housing market, including existing home sales, new home sales, housing starts, the homeownership rate, home prices, for-sale inventory, household formations, etc. However, receiving little attention are national occupancy rates. Sure, occupancy rates are often cited as it relates to rental properties, but few focus on the comparable ratio for all types of housing. And here is the kicker, it might be the most important metric to define the stage of the new construction cycle.

The national occupancy rate is simply households (demand) divided by housing units (supply). The complement is obviously the vacancy rate. When occupancy is high and vacancy is low, existing structures are insufficient to absorb incremental demand and new construction is more necessary. When occupancy is low and vacancy is high, builders’ services are not as necessary.

In addition to being overlooked, we believe it is difficult to even find a reliable source for housing’s occupancy level. In the short-run, the American Community Survey (ACS) and Housing Vacancy Survey (HVS) are conducted by the Census Bureau, but they not only conflict with each other but also differ substantially with the Decennial Census that is administered in the most robust fashion.

For example, in 2010 when all three surveys have a data point, the Decennial Census vacancy rate was at 11.4%, the ACS stood at 13.1% and the HVS was the highest at 14.3%. All of these measures were far above historical norms as the country was still dealing with the Great Recession and the foreclosure crisis. The variance might not appear extreme, but it is actually massive. According to our math, excess vacant units at this point would have ranged from 2.7 million to 7.0 million depending on which source you chose.

The output of the HVS is particularly troubling since it is intended to measure vacancies but the survey has had a material deviation from the more robust Decennial Census ever since 1990 when its methodology was revised.

To better understand national occupancy rates outside of the government conclusion, we triangulate new construction, demolitions and our estimates of household formation based on electric utility customers to extend the Decennial Census data points to the present year. According to our analysis, the national vacancy rate ended 2017 at 9.2%, nearly matching the decade end points from 1960 through 2000, which we interpret as a more normal level. In other words, excess housing caused by overbuilding before the recession and weak employment during the recession has been completely absorbed.

This should mean that new construction supply has returned to mid-cycle levels, essentially matching household formation plus stock lost for demolitions or obsolescence. Unfortunately, that is not the case because supply bottlenecks around labor capacity and land have governed the pace of new construction. More specifically, in 2018, we estimate new residential supply will total roughly 1.25 million but we believe sustainable demand is closer to 1.55 million. Until this differential collapses, occupancy rates will decline further and price appreciation will continue to run at an above-average rate.