The vacancy rate for single-family homes increased in 2013 and remains well above bubble and pre-bubble levels.
What? Too much new single-family construction? It sounds hard to believe, with only 618,000 single-family housing starts in 2013, heading toward 622,000 in 2014 - far below the pre-bubble average of 1.1 million per year in the 1990s. Even when adding in multi-unit building, which is booming, construction remains a laggard in the housing recovery and is contributing less than it should to employment and economic growth.
Of course, the historical norm doesn't tell us what the just-right level of construction is now. That depends on the rate at which new households are formed. If new construction runs ahead of household formation, more homes sit empty and the vacancy rate rises. In 2004 and 2005, during the bubble, construction of single-family homes soared to over 1.5 million units. Then, during the bust, household formation slowed, in part because more young people lived with parents. Too much housing and too few households were a dangerous cocktail during the housing bust and recession, causing the vacancy rate to climb until 2010. Since then, the vacancy rate has fallen, but single-family construction has continued to wallow near all-time lows.
Newly released data from the Census Bureau's American Community Survey (ACS) show that the vacancy rate for single-family homes actually ticked up a bit in 2013. That's a big surprise. It suggests even today's low level of single-family construction might still be too much, too soon. To determine whether we're building too many homes, we need first to understand household formation, and then the vacancy rate.
Single-Family Rentals Increased Despite Low Household Formation Rate
To understand what's happening with vacancy rates, let's start by looking at changes in households and housing units in the past year broken down by owner-occupied and rented, and single-family and multi-unit:
Since the bubble, including in the most recent year from 2012 to 2013, the largest increase of any household type was for single-family rentals. In contrast, the number of single-family homes occupied by owners declined in absolute terms, not just as a share of households, both in the past year and since the bubble.
Strikingly, household formation -- the year-to-year change in total households -- was a meager 321,000 between 2012 and 2013. That's way below the 1.2 million household formation rate we'd expect given the growth in the adult population. (A separate estimate of household formation from other Census data released this week was similarly low.) Low household formation means less demand for new construction.
Multi-Unit Vacancies Down, But Single-Family Vacancies Up
The vacancy rates help us see whether construction is falling behind, keeping up with, or running too far ahead of household formation. For any housing type, the vacancy rate equals the percentage of units that are unoccupied. Unlike the "renter vacancy rate" and "homeowner vacancy rate" reported by the Census Bureau, our vacancy measure for each type -- single-family and multi-unit -- includes vacant homes held off the market, which is essential for understanding today's housing market. If you ignore vacant homes held off the market, it looks like there's a shortage of homes -- but in fact there's only an inventory shortage, not a national housing shortage. (See note.)
The multi-unit vacancy rate peaked at 17.2% in 2010 and has dropped steadily since then to 15.1% in 2013. Multi-unit buildings are leading the construction recovery, but demand for apartment rentals has been so strong that the multi-unit housing stock is filling up. Thus, the multi-unit vacancy rate has fallen, returning in 2013 almost to its 2005 level. That rate might start to level off as apartment buildings broke ground in 2013 reach completion in 2014 and add to the housing supply.
The single-family market tells a very different story. Even though single-family construction remains depressed compared with historical norms, the single-family vacancy rate is still high. It was 10.7% in 2013, ticking up from 10.6% in 2012, and close to its peak of 11.0% in 2011. That compares with 7.4% in the pre-bubble days of 2000.
For the single-family vacancy rate to move back down to bubble or pre-bubble levels, the growth in households occupying single-family homes needs to exceed the growth in the single-family housing stock. In other words, to get back to normal, we gotta fill up homes faster than we build them. That didn't happen in 2013. Even though relatively few single-family homes were built, even fewer were filled, despite the increase in single-family rentals.
It's worth noting that it matters where the vacant homes are. A high vacancy rate might not imply less need for construction provided that the vacancy rate is low in areas with strong population growth and housing demand. But that's in theory. In fact, it turns out that it isn't only economically struggling areas that have stubbornly high single-family vacancy rates. Many faster growing metros overbuilt during the bubble still have lots of vacant homes.
Wrapping up: Even though single-family construction is low by historical standards, it may still be running ahead of demand. (Note: corrected from original, which said "not running ahead.") If anything, we built slightly more single-family homes in 2013 than we filled. And on the flip side, apartments and condos are being snapped up, even though multi-unit construction has been chugging ahead of normal levels. For the housing market to get back in balance, multi-unit must continue to lead the construction recovery.
Notes: this post is based on American Community Survey tables B25024 and B25032 via FactFinder, 2005-2013.
The vacancy rates for single-family and multi-unit homes include seasonally vacant homes because, unfortunately ,the ACS does not break out vacancy type by "units in structure" (that is, single-family or multi-unit). However, the Quarterly Homeownership and Vacancy Survey (HVS) confirms that the share of vacant homes that are seasonally unoccupied remained relatively constant over this period.
Vacancy by "units in structure" is available in the ACS and in the 2000 Decennial Census, but not in the 2010 Decennial Census. So there's no way to compare Decennial and ACS estimates for these measures for the same year.
The "renter vacancy rate" and "homeowner vacancy rate," defined here, are used in many Census reports; however, they omit the high (40%) and growing share of vacant units held off the market. The HVS reports renter vacancy rate and homeowner vacancy rate by "units in structure," but excludes vacant homes held off the market.