Wednesday, August 27, 2014

US Housing Starts: Better? Yes. Good? No.

The initial estimate of July housing starts exceeded expectations as construction began on 1.093 million dwellings on a seasonally-adjusted, annualized basis. Additionally, June starts were revised higher from 893,000 to 945,000. While improvement is welcome, perspective is necessary. Despite dramatic improvement from the all-time lows reached in 2009, housing starts remain moribund when measured against long-term pre-crisis norms. The Census Bureau’s data on housing starts begins in January 1959. Comparing July starts with data from inception through 2007 paints a very somber portrait of the housing market. The worst month for starts prior to 2008 was January 1991, when construction began on just 798,000 dwellings on a seasonally-adjusted, annualized basis, 12.71% more than the median level since December 2007. The median number of starts from 1959 through 2007 was 1.525 million, nearly 40% greater than the number of starts in July. In fact, from 1959 through 2007, starts exceeded the 1.093 million reported in July in 540 of 588 months, or more than 90% of the time.

Additionally, the nature of starts changed significantly in the 1990s and the credit crisis appears to have wrought a significant change once again. From data inception through the 1980s, multi-family starts, which include townhouses, condominiums, and apartment complexes, typically accounted for approximately one third of monthly starts. However, from 1990 through 2009, multi-family starts very typically comprised 20% of monthly starts, although with significant noise in the data in 2008 and 2009 as credit markets seized and the housing market ground to a halt. However, by the second half of 2010, multi-family starts began to rise as a percentage of total starts. In July, multi-family projects comprised nearly 40% of all starts.
One reason housing is a bellwether economic indicator is that home purchases lead to many additional purchases. The significant increase in multi-family units as a percentage of starts suggests fewer such additional purchases. As such, if the rate of multi-family starts remains elevated, further improvement in the housing market may come with fewer “add-on” effects for the economy. Furthermore, real disposable income growth has been in decline for some time. While it grew about 4%, on average, throughout the 1960s and 1970s, it averaged just 2.59% from 2000 through 2009 and just 1.86% over the 10-years ended in 2013. From the mid-1990s through 2007, consumption was buoyed by home equity extraction, allowing many Americans to live above their income levels. Given the increased proportion of multi-family units, which again, includes apartment complexes, there is likely to be less support for consumption through home equity extraction (This also applies to existing home sales. Since the crisis, investors have been significant buyers of existing single-family properties).
This recovery has been unusual in many ways. Six years into the expansion, job growth, especially that of full-time jobs, and the housing market remain weak. While the Fed has taken extraordinary measures, increasing its balance sheet by approximately $3.5 trillion, banks have been slow to lend, leaving the vast majority of the money the Fed has pumped into the financial system on reserve with the central bank. Given the unusual nature of the recovery, maybe it shouldn’t surprise us that investors, financial reporters, and market professionals alike get excited by economic reports that in any other era would be consider weak. Yet perception and reality are not one and the same and the perception that we are seeing strength in the housing market simply doesn’t match the data. Simply put, housing starts remain weak and the mix of starts unfavorable.