Miller and Dinn: We can't answer the first part of this question, but we can answer the second part. Of the total multifamily stock in the CoStar database, some 44% of it are one- and two-star buildings. This is based on the five-star rating system developed by CoStar to reflect the overall quality of the units, and it is highly correlated with age along with sustainability, location and design, among other factors. The highest rating is five stars, with one star as the lowest rating. More than 60% of the US multifamily stock was built before 1960, and much of this inventory is rated as one or two stars. Much of what the government and most other vendors are missing seems to be the older stock, often one- and two-star-rated units. Sixty-two percent of the gap in counts can be explained by the high correlation with older and lower-quality units. Few institutional investors buy one- and two-star properties, and this could explain why the lack of interest in this lower-quality housing.

GlobeSt.com: Why does CoStar's database reflect this lower-quality housing?

Miller and Dinn: CoStar has the entire spectrum of multifamily housing—not because of catering to a specific investor profile, but rather as result of acquiring several of the leading websites (i.e., Apartments.com and others) on which rental units are promoted. Such data has been assembled, and it is remarkable that it exceeds the US government estimates by such a large margin. For 50 selected metros, we provide a graph below showing the CoStar estimates of units in five-plus-unit buildings compared with the US Census estimates after adding adjustments for construction to bring it to 2016 estimates.

|

GlobeSt.com: Why is this so important?

Miller and Dinn: First, it is important to know how much supply of housing we have prior to suggesting how much we need for policy reasons. Second, since much of the missing stock is rated on the lower end of the quality scale and rents at the lowest rent levels, it represents what the Urban Land Institute calls NOAH, for naturally occurring affordable housing. Investors might target such housing for upgrade opportunities, but governments might also want to preserve it and encourage simple maintenance so that more affordable housing continues to be available. In fact, it would not be that irrational to suggest property-tax abatements on this older stock if it meets building codes, is not rehabbed and if it rents to the lower 40% of the household-income spectrum in local markets.

GlobeSt.com: Where is NOAH concentrated?

Miller and Dinn: Based on using one- and two-star units as a proxy, we observe two key factors: 1. expensive, less-affordable-housing metro markets tend to have a large share of one- and two-star stock, with 48% in New York City, 58% in San Diego and 54% in San Francisco as examples; and 2. older metros with a slower pace of new-housing supply and lagged economic growth also have a high percentage of one- and two-star stock, with 54% in Pittsburgh, 52% in Detroit, 48% in Cincinnati and 46% in Cleveland.

One way the US government might save money the next decennial census is to start with the CoStar data base; otherwise, we will never know how much multifamily housing we really have.

|

SAN DIEGO—Comparing supply data from three different sources reveals disparities in the number of multifamily units in the US housing stock, Norm Miller and Michael Dinn tell GlobeSt.com. Miller is Hahn Chair of Real Estate Finance at the University of San Diego, affiliated with the Burnham-Moores Center for Real Estate, and is a principal at Homer Hoyt Advisory Services; Dinn is president of Dinn Focused Marketing Inc. We spoke to them exclusively about why government estimates of the stock may be so far off.

GlobeSt.com: What have you noticed about the multifamily-housing-stock numbers?

Miller and Dinn: In recent studies on the multifamily market, we had the opportunity to compare 2016 supply data from three sources, including the US Census ACS numbers adjusted for reported construction permits, a variety of well-known commercial real estate services companies and CoStar. The CRE services companies had, on average, just a slightly higher estimate of the stock by metro than the US Census, but both were significantly below the CoStar estimates. Focusing only on the rental stock in buildings with five or more units, and for the nation as a whole, the CoStar data base had 3,897,961 more rental units than the Census, or 17.2% more units in total. When compared for 50 selected metros, the result was 16.2% more units in total, suggesting that a disproportionate number of units that are missing from government estimates are from smaller cities and rural areas.

|

GlobeSt.com: How could the government estimates be so far off, and what are they missing?

Miller and Dinn: We can't answer the first part of this question, but we can answer the second part. Of the total multifamily stock in the CoStar database, some 44% of it are one- and two-star buildings. This is based on the five-star rating system developed by CoStar to reflect the overall quality of the units, and it is highly correlated with age along with sustainability, location and design, among other factors. The highest rating is five stars, with one star as the lowest rating. More than 60% of the US multifamily stock was built before 1960, and much of this inventory is rated as one or two stars. Much of what the government and most other vendors are missing seems to be the older stock, often one- and two-star-rated units. Sixty-two percent of the gap in counts can be explained by the high correlation with older and lower-quality units. Few institutional investors buy one- and two-star properties, and this could explain why the lack of interest in this lower-quality housing.

GlobeSt.com: Why does CoStar's database reflect this lower-quality housing?

Miller and Dinn: CoStar has the entire spectrum of multifamily housing—not because of catering to a specific investor profile, but rather as result of acquiring several of the leading websites (i.e., Apartments.com and others) on which rental units are promoted. Such data has been assembled, and it is remarkable that it exceeds the US government estimates by such a large margin. For 50 selected metros, we provide a graph below showing the CoStar estimates of units in five-plus-unit buildings compared with the US Census estimates after adding adjustments for construction to bring it to 2016 estimates.