After a sharp decline in the spring, home building has staged a dramatic rebound. In fact, supported by low mortgage rates, an evolving geography of housing preferences, and favorable demographic tailwinds, housing demand has improved so quickly that the current difference between the pace of newly-built single-family home sales and for-sale single-family construction starts has reached a historic level. The fact that sales are outpacing construction starts to this degree indicates additional home building lies ahead.
The following graph shows the benchmark Census measures of home construction, the seasonally adjusted annual rates of single-family starts and single-family home sales. Starts account for the beginning of construction of homes, whether that construction is for a home already under a sales contract, being built for-sale, being built for-rent, or undertaken for a construction contract (a custom build on an owner’s lot). New home sales are signed sales contracts for new builds, whether that home has started or completed construction (new home sales is thus a counterpart to the NAR pending sales index, rather than existing home sales, which account for closed contracts).
The reason that starts outpace sales on the chart below is because, as noted above, starts represent all home builds. Thus, while the measure of new home sales represents only the more narrow for-sale class, housing starts also include custom builds and built-for-rent construction.
The consequences of the virus-induced 2020 downturn (the Great Disruption?) are seen clearly at the end of the graph, particularly the V-shaped nature of the impact on housing. While both measures have staged impressive rebounds, the sales measure has completely closed the gap between the two series.
However, the actual effect is larger. For an apples-to-apples comparison of the rates of for-sale construction and new home sales, we need to filter the starts series to remove custom builds and built-for-rent single-family construction. Using Census quarterly data of these construction types, I interpolated the quarterly data into monthly, seasonally adjusted data and then subtracted this new series from the existing Census data series. The new data series, single-family for-sale starts, allows for a precise comparison of the pace of home building in the for-sale sector relative to sales, as graphed below.
The new series is much closer to the sales data, with just occasional periods of notable difference amid the statistical noise. The adjusted data makes clear how great the current difference is between sales (red) and starts (blue). This gap is unprecedented in the 20 years of data presented here, and there is no comparable period in the data going back to 1963. Plotting the difference between the monthly rates of for-sale starts and sales yields the following series, which peaks in the most recent data.
The degree to which current new home sales are outpacing starts is clear in the above graph. These data are consistent with Census estimates and NAHB surveys that indicate builders are selling homes that have not begun construction in greater numbers. Indeed, the count of such home sales is up 69% compared to a year ago, an incredible jump. To place the current data into context, I smoothed the data using 6-month moving averages. While this dilutes somewhat the scale of the current gap, it shows three relevant periods over the last two decades where sales and for-sale starts disconnected.
As seen above, the first period occurred during the housing boom when for-sale starts exceeded sales, leading to an inventory overhang that was part of the housing crisis preceding the Great Recession. The second separation occurred as single-family starts plummeted during the Great Recession and sales, helped by the three stages of the federal home buyer tax credit, reduced excess inventory (new home months’ supply peaked at 11.1 during the spring of 2008).
These two prior periods happened in an overbuilt environment. The script is flipped today. Months’ supply for new homes is down to a lean 3.3, and existing home months’ supply (per NAR data) is at a very tight 2.8. Thus, the third period of separation between for-sale starts and new home sales occurring now is a signal of the degree to which home building will need to play catch-up with current demand.
As with other economic impacts related to the virus, prior trends have been accelerated. With home building, the last decade (the Long Recovery) was characterized by underbuilding due to supply-side limitations such as labor availability and law/regulatory cost impacts. The lagging pace of construction, relative to current sales, is an intense, compressed version of these general economic trends, with builders citing lumber and material issues as delaying some, current construction projects.
Because builders do not want to contract home sales that they will not be able to deliver effectively, the current, historic gap between elevated sales volume and improving if relatively lower construction rates means that the pace of growth for new home sales will need to slow and/or the rate of home building will need to increase to balance the market. Strong levels of the NAHB/Wells Fargo HMI measure of home builder confidence are consistent with this expectation, for starts at least.
In the meantime, the current gap between for-sale starts and sales is unprecedented. Moreover, this gap is not the only historic, current data reading of the housing industry. The NAHB/Wells Fargo HMI reached a data series high in September.
And the gap between median newly-built and existing home prices, which peaked near $95,000 three years ago, has closed and inverted. According to NAR data, the median price for a resale single-family home was $315,000, higher than the Census reported median of $312,800 for newly-built homes. That inversion has only occurred one other time over the last two decades (June 2005).
In sum, low levels of existing inventory, rising resale prices relative to new construction, strong builder confidence, and sales exceeding for-sale starts point to solid levels of home construction in the months ahead.
Home building sustains jobs (approximately 2.9 per home built and 0.75 per $100,000 in remodeling), which means more residential construction employment gains in the near-term. In fact, over the next two to three months residential construction will likely post a year-over-year gain for employment, a notable sign of strength for housing in the recovering economy.