5/20 update: HUD has posted a more detailed response to the WP article that includes an re-analysis of the entire data set. That response is available here: “Setting the Record Straight: What the Washington Post Got Wrong About the HOME Program”
The Washington Post article last weekend that discussed the Beaumont HOME development failure is part of the Post’s data-driven look at HUD’s HOME program nationwide. The research supporting the article identified 664 projects that appeared to be delayed or abandoned after being started, out of 5,080 currently in “open” status (i.e. not completed) in the program. On a dollar basis, this represents $371 million, or 12%, of the $3.2 billion in funding examined by the paper. In Texas, $9 million of $173 million in currently-open developments appear delayed (5%), with another $9 million presumed canceled prior to project start.
For the families that won’t be served by these developments, one failure is too many, but the presentation of this statistic got me thinking about the context and drivers for this number. What’s the corresponding statistic in the private real estate market over the last few years?
The 12% (or, as they put it, “1 out of 7”) delayed statistic that comes out of the Washington Post article is the percent of developments currently approved for funding for construction that appear delayed or canceled after the initial draw. This is NOT the same as the percent of total developments funded or available for funding that are delayed or canceled after the first draw, because once a development is completed, it is removed from the denominator. The exclusion of completed developments from the analysis makes it difficult to identify a comparable private-sector match for the Post’s data set, and inflates the number compared to a delay rate that includes completed developments.
Nevertheless, for context, here are some related statistics from the private market (The HOME program funds both multifamily and single family homes.)
- Trepp, a commercial real estate analytics company, expects the delinquency rate in construction loans held by U.S. banks to be 18.3% in the first quarter of 2011. This statistic is probably the most similar number to the WP delay statistic, as construction loans are typically converted to other types financing at construction completion. However, the TREPP number covers non-residential construction as well as residential.
- Trepp reports that the nation-wide delinquency rate for Commercial Mortgage Back-Securities backed by existing multifamily apartments was 16.85% in January.
- The Mortgage Bankers Association reports that 12.84% of mortgage loans were either one payment delinquent or in the foreclosure process in Q1 2011.
- In their 2010 10-K, homebuilder KB Homes reported their contract cancellation rate was 37% (i.e. 37% of orders for new homes were cancelled by buyers prior to closing).
Given that context, what else can we determine about the drivers of the delays identified by the Post? If you drill into the Washington Post’s dataset[1], you can see that the concentration of “delayed developments” varies greatly by state. In fact, five “states”: New York, California, Puerto Rico(?), Florida, and New Jersey account for 66% of the delayed projects.
While I don’t have foreclosure data on Puerto Rico, the Joint Center indicates foreclosure rates in NY, CA, FL, and NJ are all above the national average. On the other hand, the 24 states with the smallest delays in development funding account for only 3% of the delayed projects, and all have foreclosure rates below the national average.
What’s the take away? Regional markets with tougher housing markets appear to be more likely to have HOME deals fall apart. This makes sense, as HOME failures don’t happen in a vacuum. As the Washington Post mentions, HOME funding is usually part of a larger funding package for an affordable housing development, and those other funding sources may be subject to the local credit market for new housing. And when one piece of a package falls apart, the whole deal falls apart.
Nevertheless, some states did well despite a tough regional market. For example, Nevada, the state with the second highest foreclosure rate, performed significantly better than the national average, with a 6% delay rate. Among large states with high foreclosure rates, Illinois also seems to be doing well, with only one development (0.3% of funding) delayed. (Although better then average, Texas isn’t an over-performing outlier like Illinois.)
This suggests that even given local housing conditions, some of the lower performing states could learn something from the activities of the higher performing states. HOME is a block grant program with great local flexibility, but local jurisdictions should reach out to other, high performing, jurisdictions to find practices that can improve their program.
So, in summary:
1) The magnitude of problems in financing development construction in the HOME program should be considered in the context of problems in the housing market in general.
2) Comparisons of the ‘delayed and abandoned’ benchmark between areas or over time should take into account the state of the regional housing markets
3) HUD could probably improve low-performing states by identifying and encouraging the best-practices of high performing states.
Hopefully the light the Washington Post is shining on the HOME program will lead to improvements in this program, because every failed housing unit is an opportunity lost for a family in need of housing.
As a country, we can always do better.
Other Online Discussions on the WP report:
- HUD’s Response to the WP report
- Affordable Housing Report post
- Council of Large Public Housing Authorities post
- National League of Cities post
[1] Note for researchers: The summary WP data by state can be extracted from the HTML code used to generate the graphics at the WP website.
[…] The Congressional House Financial Services Committee is holding a hearing this morning on the HOME program, recently the focus of a Washington Post article previously discussed here at Texas Housers. […]