«reNtal HousiNg Policy iN tHe uNited states Volume 13, Number 2 • 2011 U.S. Department of Housing and Urban Development | Office of Policy ...»
The exhibit focuses on two levels of incomes and rents, households earning at or below (and units affordable at) 30 percent of AMI, and households earning between 30 and 50 percent of AMI (and units affordable at 50 percent of AMI). As with the national level analysis, this metropolitan level analysis reveals that in every housing market too few rental housing units exist that have rents that are affordable to extremely low-income households. This is not to suggest that every housing market needs more housing—a look at the vacancy rates confirms that some markets have more than adequate supply—rather it indicates that the number of households with incomes too low to affordably support local rents far exceeds the few rental housing units with rents affordable to this group. If a goal of housing policy is to ensure that low-income households do not expend enormous shares of their budget towards housing, then additional rental assistance is needed to affordably house these families.
For renter households that are very low-income (30 to 50 percent of AMI), but not likely to be living below the poverty line, the picture is more nuanced. Some housing markets exhibit significant rental housing affordability stresses for households at 50 percent of AMI. Miami, Florida; San Diego, California; Boston, Massachusetts; New York, New York; Los Angeles, California; Tampa, Florida; and Honolulu, Hawaii are extremely tough places to find affordable units at 50 percent of AMI with fewer than 50 units per 100 renters. Several of these cities have significant rental housing supply constraints because the amount of developable land is limited or because of the presence of regulatory barriers, which keep vacancy rates low and create upward pressure on rents (Glaeser and Gyourko, 2008). In several Midwest cities where housing values are low, land is cheap, and the population has shrunk, an apparent surplus of rental housing units that are affordable to households at 50 percent of AMI exists. In these more affordable metropolitan areas, the estimated vacancy rate for units with rents affordable at 50 percent of AMI exceeds 11 percent in each of the metropolitan areas (except Minneapolis) suggesting an adequate supply of affordable rental housing for households earning 50 percent of AMI. These housing markets are where rehabilitation and tenant-based rental assistance may be a more market sensitive intervention than additional new construction.
It is important to note that this analysis does not parse out substandard rental housing units, and does not incorporate the geographic distribution of these units within a metropolitan area. For instance, an affordable rental housing unit in a high-poverty central city neighborhood would be classified as available and affordable even if the renter is in a distant suburb.
Rent Burdens The supply gap approach to describing affordability dynamics is useful because it directly accounts for changes in the rental housing stock, however, it may fail to capture nuances of low-income households’ housing options. Important considerations such as geography or the physical adequacy of housing may further limit options to the poor. Therefore it is also critical to examine observed rent burdens. Exhibit 21 shows the trend of rent burdens from 1990 through 2009. In 2009 the median rent burden was 16.6 percent higher than in 1990. More startling, the percentage of renters paying more than one-half of their income for housing increased nearly 38 percent since 1990.
6,000 5,379 5,198 5,203 5,175 5,104 4,856 4,842 5,000 4,535 4,000 3,000 2,000 1,000
Exhibit 25 Percent of Unassisted Very Low-Income Renters With Severe Rent Burden or in Severely Inadequate Housing Worst Case Housing Needs 30.00 Percent of All Renters 20.1 20.00 17.6 16.8 15.8 15.5 15.2 15.4 15.0 14.5 14.3 13.9 10.00
Summary and Findings This article attempted to provide an update of rental housing affordability trends and levels since 1990, including an exploration of how rental housing affordability has been affected by the recession of 2007 through 2009. In the first half of the 1990s real rents declined or were stagnant, when the economy took off in the second half of the decade real rents rose in most housing markets, but these increases were married with increasing real renter incomes, leaving rent burdens slightly lower by the end of the decade. The 2000s have been a far more trying time period for renters.
Real renter incomes declined in nearly every housing market in the first half of the 2000s, and ended the decade below 2000 levels in 25 of 25 markets in large part due to a historic recession.
Even with downward rent pressure through the recession, real rent levels ended with 2009 above 2000 levels in 19 of 26 housing markets. These trends combined to drive rent burdens to historic highs. For renters at 50 percent of AMI, rental housing affordability varies significantly across metropolitan areas.
In 2009, the number of extremely low-income renter households paying more than 50 percent of their income for housing increased to more than 6.7 million households, and only 41 affordable and available units existed per 100 extremely low-income renters.
During the recession, the moderately affordable rental housing stock with rents affordable to households at 80 percent of AMI increased sizably, but the data is less clear on the changes at the bottom of the rental housing stock.
Future rent pressures will depend on the extent to which the current rental housing stock can absorb expected future demand increases for rental housing. Multifamily rental production slowed significantly during the past 2.5 years, meaning that few new rental housing completions will come online in 2011 and 2012. Given the incredible heterogeneity in rental housing affordability dynamics presented in this article, different housing markets are likely to absorb these demand changes very differently. Communities with persistently high vacancy rates, whether from elastic supply or historic population loss, should be able to accommodate future rental demand fairly easily, although communities with low vacancy rates and inelastic supply may face steep upward pressure on rents as demand increases.
Finally, although the sources of rental housing data have improved over the years, development of detailed, timely, representative data still lags the collection of ownership data. This study demonstrates that additional data is needed to understand time series dynamics in rental housing affordability. Particularly, the collection of data on rent concessions, which are the primary method of nominal rent cuts, but are rarely collected in public surveys.
1.02 1.00 0.98 0.96
1.02 1.00 0.98
1.05 1.00 0.95
Appendix C. Exploring Rental Housing Affordability Declines AHS 2007 to 2009 To examine the extent to which falling local incomes may be driving the reduction in rental housing affordability, a decomposition of the rent distributions was constructed using a model similar to
one used by Quigley and Raphael (2004). The basic formula for the composition follows:
CDF Rents 2009− CDF Rents 2007 = [CDF(Rents 2009)Incomes 2009 – CDF(Rents 2007)Incomes 2009] – [CDF(Rents 2007)Incomes 2007 – CDF(Rents 2007)Incomes 2009] where the overall change in the percentage of rental housing units that are affordable to renters at the income threshold can be decomposed into the first term: the cumulative distribution function (CDF) of the 2009 rent distribution evaluated at the 2009 income limits less the 2007 rent distribution also evaluated at the 2009 income limits provides the percentage change in rental housing affordability due to changing rents, and the second term is the CDF of 2007 rent distribution evaluated at the 2007 income levels less the CDF of 2007 rents evaluated at the 2009 income level—interpreted as the change in rental housing affordability due to changing income. The first term is operationalized with an estimate of the portion of rental housing units in 2009 with rents below the 2009 income threshold, minus the portion of rental housing units in 2009 with rents below the 2007 income threshold. All rents and the rental housing affordability income thresholds are adjusted to be in constant 2009 dollars. This analysis requires having a specified income limit for each rental housing unit, so only units that existed in both 2007 and 2009 could be included.
The slightly narrowed samples represent roughly 34 million rental housing units in 2007 and 32 million units in 2009 (this difference arises primarily to sample adjustment in 2009). HUD is required to publish income limits annually by a specified date, which does not allow for use of the most up-to-date data. The 2009 income limits are in fact based on 2007 data with adjustments for inflation, so income limits have not fallen commensurate with the real income declines. This is apparent in the data, with 71 percent of the sample rental housing units that were rented in either 2007 or 2009 being located in local jurisdictions where the AMI increased in real terms from 2007 to2009. Exhibit C-1 displays the results of the decomposition above.
The decomposition reveals that for low-income renter households, rental housing affordability worsened between 2007 and 2009, and the reduction in affordability appears driven by increases in reported gross rents. Exhibit C-2 shows the percentage of rental housing units that were affordable in 2007, but changed affordability status in 2009 and experienced either a real rent increase or decrease from 2007 to 2009. About 93 percent of the sample units were affordable at 30 percent of AMI in 2007. These same units were not affordable at 30 percent of AMI in 2009 because of increases in real rents. These increases also suggest that changes in rental housing affordability shown in the AHS between 2007 and 2009 were driven primarily by rent increases.
Longitudinal Estimates of Rental Housing Stock Changes, AHS 2007 to 2009 This appendix features estimates of the movements of rental housing units between affordability categories and to and from other rental housing stock segments (that is, owned, vacant, and so on) using estimated longitudinal weights for 2007 through 2009 following an approach taken in the HUD Components of Inventory Change (CINCH) reports. This analysis suggests that 1.1 million extremely low-rent units in 2007 moved to a less affordable category, which was only partially offset by 700,000 rental housing units moving into the extremely low-rent category from higher rent categories. The other primary source of this loss is the movement of rental units either to ownership or a permanent loss category. Roughly 400,000 extremely low-rent units in 2007 were either lost or moved to owner-occupied units, which was only partially recouped through 240,000 new rental units.
Exhibit C-3 presents estimates of the change in the rental housing stock by affordability category.
The changes are separated into a forward-looking component and backward-looking component.
A description of the weighting methodology is provided in text that follows. Note that when adjusting the weights, the 2007 and 2009 data no longer perfectly mirror the cross sectional analysis.
See description of the weighting methodology following exhibit C-3.
Weighting Methodology The AHS provides cross-sectional weights with each sample that are applied to provide a snapshot of the nation’s housing market at a given time. To understand how a trend among unweighted sample rental housing units corresponds to a weighted total for the nation, it is necessary to construct longitudinal weights. This article takes an approach that parallels the methods in HUD’s CINCH reports. Due to sample adjustments, new construction, losses, tenure changes, and noninterviews, the weights for particular AHS sampled rental housing unit change between survey years. Following the approach taken in the CINCH reports, two separate sets of weights are developed: one for a forward-looking analysis to describe how the status of units in the 2007 rental housing stock changed in 2009, and one for a backward-looking analysis to describe the status of the 2009 rental housing stock in 2007.
For the forward-looking analysis, rental housing units are categorized as either existing in both 2007 and 2009 or existing in 2007 and lost in 2009. The basic weighting approach is to estimate the weighted count of losses in 2009 from the 2007 rental housing stock, and then adjust the pure weights
for the units existing in both 2007 and 2009 so that they sum to the base 2007 count, net of losses:
Additional adjustments are made so that the revised 2007 weights sum to the corresponding published totals, distinguishing between tenure and occupancy status, and between mobile homes and all other rental housing units.
For the backward-looking analysis, rental housing units are categorized as existing in both 2007 and 2009, new construction in 2009 or other additions in 2009 (added from nonresidential use, made habitable from correction of deficiency, or added through merger or conversion). Similar to the forward-looking analysis, the pure weights are adjusted based on the changes to the rental housing stock. Estimates of new construction and other additions from the 2009 AHS are used to adjust the 2009 pure weights for rental housing units that existed in both 2007 and 2009 so that
they sum to 2009 count net of new additions:
Acknowledgments The author thanks Ben Winter, Erika Poethig, and Todd Richardson of the U.S. Department of Housing and Urban Development for their invaluable comments and suggestions. The author also thanks Ingrid Gould Ellen and an anonymous referee for their tremendous input and review of this article. Any errors or omissions made, and opinions expressed herein, are those of the author and do not necessarily represent the position of HUD.
Author Rob Collinson is an analyst at the U.S. Department of Housing and Urban Development.