Monday, October 16, 2017

Far From Static, Rural Home Prices are Dynamic and Growing in Most of the Country

by Alexander Hermann
Research Assistant
While many believe that home prices in rural areas are largely stagnant, this is not the case, according to a new Joint Center analysis from the Federal Housing Finance Agency (FHFA). Rather, non-metro home prices are dynamic, highly variable, and growing—much like home prices for the nation as a whole.

Nationally, home prices grew significantly over the last half-decade, following years of decline in the aftermath of the housing crisis. But these broad indicators mask significant variation by region, market, and even neighborhood. While these trends were discussed extensively in this year's State of the Nation's Housing report and on our blog, those analyses focused almost exclusively on the roughly 62 million homeowners living within metropolitan areas, where 83 percent of the country's owner-occupied units are located.

But what about home prices in the largely rural, non-metro areas that are home to about 15 percent of the population? Due to limited data availability, these areas often are ignored in discussion about trends in home prices. A new Joint Center analysis of the FHFA House Price Index, which measures price changes from the sale, refinancing, and appraisal of the same properties, seeks to fill this gap. In particular, the analysis examined house prices for all counties outside of Metropolitan Statistical Areas (MSAs). (Metro-area counties are those that contain an urbanized area of at lease 50,000 persons plus any adjoining counties that commuting patterns show are economically integrated with their metropolitan neighbors.)

Several key findings emerge from this analysis, most notably:

  • Rather than stagnating, home prices outside the metropolitan areas grew considerably between 2000 and 2016. In nominal terms, non-metro home prices grew 58 percent and real non-metro home prices grew nearly 15 percent. Moreover, by the fourth quarter of 2016, nominal non-metro and rural home prices were two percent above their pre-recession peak—the same as national home prices at the same point. However, when adjusted for inflation, home prices in non-metro areas were still 11 percent below their peak, which again, is somewhat similar to national patterns (Figure 1).

Note: The US non-metro index is a weighted-average of state non-metro HPIs, with each state's value weighted by its share of non-metro detached single family housing units. Real home prices are adjusted for inflation using the CPI-U for All Items less shelter. 
Source: JCHS tabulations of the Federal Housing Finance Agency, All-Transactions House Price Index.

  • While significant, these increases were more modest than the gains experienced by the nation as a whole. Nationally, real home prices grew 23 percent in 2000-2016, about 8 percentage points more than prices in non-metro areas (Figure 2). The difference is largely due to the more modest cyclicality of non-metro home prices movements during and after the recession. In the immediate aftermath of the housing crisis, national home prices fell severely for several years before starting to rise steadily in early 2012. In contrast, home prices in non-metro areas were mostly stagnant from 2011 to 2014, and, compared to metro areas, have grown less quickly since. As a result, between 2012 and 2016, the percentage-point increase in non-metro home prices was greater than the percentage-point increase in statewide prices only in Alaska, Hawaii, Mississippi, Montana, and Nebraska.

Note: The US non-metro index is a weighted-average of state non-metro HPIs, with each state's value weighted by its share of non-metro detached single family housing units. Real home prices are adjusted for inflation using the CPI-U for All Items less shelter. 
Source: JCHS tabulations of the Federal Housing Finance Agency, All-Transactions House Price Index.

  • Rural home price trends by state vary widely. While non-metro home prices within states generally change in ways that are similar to the broader state-wide trends, increases in rural areas did not always trail overall increases in their states (Figure 3). Rather, from 2000 to 2016, the increase in non-metro house prices actually exceeded statewide prices increases in 24 of the 47 states where at least some part of the state was not in an MSA. (Three states—Delaware, New Jersey, and Rhode Island—do not have non-metro areas.)

    The gaps were largest in North Dakota (13 percentage points greater), Nebraska (12 percentage points greater), South Dakota (11 percentage points greater), New Mexico (7 percentage points greater), and Louisiana (6 percentage points greater). In contrast, the increases in statewide prices most exceeded rural prices in coastal states, where prices in metropolitan areas have grown significantly. The gaps were greatest in California (26 percentage points), Hawaii (23 percentage points), Virginia (22 percentage points), Oregon (19 percentage points), and Maryland (18 percentage points).

  • Non-metro home prices rose more slowly in the run-up to the housing crisis, and rarely fell as far in the aftermath. Between 2000 and 2007, real non-metro home prices increased by 28 percent, far less than the 41 percent increase for all single-family homes. However, in about half of the 47 states with non-metro areas, the percent increase for all single-family homes. However, in about half of the 47 states with non-metro areas, the percent increase in non-metro home prices exceeded states-wide home-price gains in the run-up to peak. After the crash, real national home prices fell below 2000 levels briefly in 2012, while non-metro prices remained about three percent above their 2000 levels. Moreover, this pattern held true in most states. In only 10 of 47 states were the recessionary-low home prices (relative to 2000) in non-metro areas lower than for the state overall.

  • Unclear signals for the future. While this analysis shows that non-metro house prices generally follow national patterns, since the recession price growth in rural areas has been slower than in the nation as a whole. The homeownership rate in non-metro areas was also about 71 percent in 2015, nearly 9 percentage points higher than in metro areas. These differences held across racial and ethnic groups, as well as for low- and moderate-income households. Collectively, this indicates that these markets merit close attention in the coming years.

Monday, October 2, 2017

The Negative (and Positive) Spillovers of Concentrated Foreclosure Activity in New York City

by Kristin L. Perkins
Postdoctoral Fellow
Foreclosures have negative effects not only for the people who lose their homes, but also for the neighborhoods where they lived.

In an article that recently appeared in Urban Affairs Review, Michael J. Lear, Elyzabeth Gaumer, and I conclude that, at least in New York City, neighborhood foreclosure activity during the peak of the Great Recession was associated with an individual property's risk of foreclosure, but not in the way that most previous research has assumed. These findings suggest that financial institutions' practices during the foreclosure process may have contributed to how quickly neighborhoods recovered from that crisis above and beyond the institutions' roles in causing it. The research focused on two key phases in the foreclosure process, an early phase when a lender filed a foreclosure notice after a property owner missed some mortgage payments, and a later phase when properties were actually scheduled to be auctioned. (In New York, the latter process often occurs more than a year after the former one.)

Although New York City's housing market fared better than many other markets during the recession, it was not immune to the problems associated with that downturn. Illustratively, in 2007 and 2008, there were nearly 14,000 foreclosure filings annually, double the number in 2004. In 2009, the number increased to over 20,000. The number of foreclosure auctions, however, was much smaller, ranging from 3,000 to 4,500 a year between 2007 and 2009.

This foreclosure activity was also highly concentrated. Between 2007 and 2009, over half of the city's foreclosure filings occurred in just nine of the city's 55 sub-borough areas (SBAs), all of them in Brooklyn or Queens. Moreover, over half of the city's auctions took place in just six SBAs. Not all specific areas with the highest concentration of foreclosure auctions, however, were among the areas with the highest concentration of foreclosure filings. Rather, the share of foreclosure filings that result in scheduled foreclosure auction varied considerably across boroughs, from less than 10 percent being scheduled for auction in parts of Brooklyn to more than 40 percent in parts of Queens.

My coauthors and I hypothesized that the number of scheduled foreclosure auctions surrounding a property that had received a foreclosure filing is positively associated with the likelihood of that property itself reaching auction, net of other factors. Since foreclosure filings do not necessarily involved a transfer of ownership, however, we thought they might not have as strong an association as auctions may have.

Drawing on data about individual property and neighborhood characteristics in addition to foreclosure activity, we found that levels of neighborhood foreclosure activity in both phases were, in fact, associated with an individual property's outcome, but in different directions. Holding constant individual property and neighborhood characteristics, as the number of nearby properties with a foreclosure filing increases, the probability that an individual financially-distressed property will be scheduled for foreclosure auction decreases. This pattern is reversed for properties in the later phase of the foreclosure process. As the number of nearby properties scheduled for auction increases, the probability that an individual financially-distressed property will be scheduled for foreclosure auction also increases (See Figure).

These findings also suggest that at least in New York, banks and loan servicers may have delayed the processing of foreclosures in areas with larger numbers of properties with foreclosure filings. They may have done so because foreclosed properties may not sell as quickly or as profitably as in more desirable areas where there are fewer distressed properties and/or where foreclosed properties sell right away. Future research should examine these practices in more detail, not only in New York, but in other states as well and, in doing so, underscore the importance of financial institution practices not just in the lead up to the Great Recession, but throughout the recovery process as well.

Thursday, September 28, 2017

Successful Collaboration in Community Development: Easier Said Than Done

by Alexander Von Hoffman
Senior Research Fellow
What are the keys to successful collaborations of nonprofit housing organizations? A remarkable attempt to form a novel alliance by five such groups in western New York State reveals several keys to an effective collaboration. Each of the five organizations -- NeighborWorks® Rochester, West Side (Buffalo) Neighborhood Housing Services, Niagara Falls Neighborhood Housing Services, Arbor Housing and Development, and NeighborWorks® Home Resources -- were long-established in their geographic areas. Moreover, each belonged to the network of NeighborWorks® America, a congressionally chartered nonprofit corporation that provides grants, technical assistance, training, and organizational assessment to housing and community development organizations. Their experiences, which are documented in a recent Joint Center case study, shows both the problems and the possibilities of putting the idea of collaboration into action.

Buffalo, New York

Before going into details, it bears mention that it has become almost an axiom in the community development field that nonprofit organizations must "collaborate" if they are going to survive, much less transform low-income communities. And the idea of collaboration is appealing: two or more organizations agree to coordinate activities in a systematic way -- as opposed to carrying out a one-time joint venture. Such collaborations can range from a temporary partnership to outright mergers (or anything in between). But many practitioners and scholars believe such initiatives can address a host of serious problems. For most community development organizations, money is always short, and especially so in recent years as the federal government has reduced funding for the Community Development Block Grant (CDBG) program. In addition, many nonprofit groups appear to be financially weak, undersized, relatively unproductive, organizationally stagnant, or some combination of the above. By sharing business lines, programs, and administrative functions, smaller and financially weaker groups could become more efficient and possibly tap the resources and knowledge of stronger organizations. If so, they could stabilize their finances and begin to grow, which would allow them to devote more time and attention to serving their low- and moderate-income constituents effectively.

But as the new case study documents, putting these ideas into practice can be difficult. After extensive discussions, in 2012 the leaders of five western New York State groups devised the concept of a "collaborative merger." In this structure, each organization would become a subsidiary division of a new central organization. As subsidiaries, the five groups would maintain their separate identities, offices, and geographic service areas while increasing their capabilities and expanding the types and volume of business. The central organizations, which would be overseen by board members from each of the participating groups, would provide core administrative functions, and, in doing so, bring the efficiency and resources of a large organization to the work of what had been separate smaller groups.

Just as the groups were about to create the new entity, however, the alliance came to an abrupt halt. Many factors contributed to the breakdown of the process. The biggest obstacle was the difficulty of bringing five disparate groups together under a common structure. The organizations covered starkly different kinds of geographic territories. Three of the organizations (NeighborWorks® Rochester, West Side Neighborhood Housing Services (Buffalo), and Niagara Falls Neighborhood Housing Services) were rooted in cities. The other two (Arbor Housing and Development, and NeighborWorks® Home Resources) were rural entities with geographically extensive service areas.

Moreover, there were significant differences in the organizations' programmatic offerings. Arbor, the largest of the five groups not only provided residential services for people with special needs and victims of domestic violence, it also developed and managed low-income housing. The other four groups were traditional "neighborhood housing service" groups that emphasized homeownership counseling, lending, and community engagement. Over time, key staff and board members of the housing service organizations became increasingly concerned that if the alliance went forward, their organizations would lose their identities and be less able to perform their core functions. Ultimately, the concerns became so great that the groups' leaders decided not to proceed with the planned alliance.

That was not the end of the story, however. Following the original idea, albeit on a smaller scale, the three urban-oriented neighborhood housing service groups (NeighborWorks® Rochester, West Side Neighborhood Services, and Niagara Falls Neighborhood Housing Services) merged to form a new organization called NeighborWorks® Community Partners. Meanwhile, Arbor, which continued to be a NeighborWorks® member, has not only thrived, but has also expanded its service area as far as Pennsylvania and Albany, NY. The fifth organization, NeighborWorks® Home Resources, remained in business under the name Rural Revitalization Corporation, but has left the NeighborWorks® network.

The experiences of these five organizations not only underscores the importance of building trust among partners in any collaboration, it also offers several lessons for those interested in collaborating with other entities. First, prospective collaborators might do well to begin by collaborating on actual programs before they start building a grand organizational structure. Second, collaborators should take time to develop a common vision, which means wrestling honestly with with the differences that separate the participating groups. Third, and related to the above, communication - open and constant - is essential, as is the full and committed participation of all of the involved parties. The leaders of such efforts must go to great lengths to ensure that everyone - including staff and board members from all the organizations - understand and support the collaborative effort.

Finally, everyone must understand that bringing existing groups into a new organizational arrangement is not business as usual. It is an act of creation that will change the status quo. Such a collaboration requires extraordinary care to ensure that the participants recognize the process and the outcome as legitimate. And this in turn means it is essential to tackle difficult questions about management, sharing leadership, and the roles and responsibilities of staff and board members sooner rather than later.

Tuesday, September 19, 2017

Are Integrated Neighborhoods Becoming More Common in the United States?

by Jonathan Spader and
Shannon Rieger
The share of the population living in racially and ethnically integrated neighborhoods in the US has increased since 2000, according to our new Joint Center research brief. However, most Americans continue to live in non-integrated neighborhoods, and evidence suggests that some of the recent increases in integration may be the temporary byproduct of gentrification and displacement.

The new brief, "Patterns and Trends of Residential Integration in the United States Since 2000," assesses whether the nation's increasingly diverse population is fostering the growth of integrated neighborhoods or whether the choices people make about where to live are reinforcing existing lines of segregation and exclusion. Specifically, we use data from the 2000 Census and the 2011-15 American Community Survey (the most recent data available at the census tract level) to describe the number, stability, and characteristics of integrated neighborhoods.

Because there is no single measure for identifying integrated neighborhoods, our analyses applies two commonly-used definitions to define integration. The first approach—which we refer to as "no-majority neighborhoods"—defines integrated neighborhoods as those where no racial or ethnic group accounts for 50 percent or more of the population. While this definition identifies neighborhoods with a plurality of races and ethnicities, it may exclude some neighborhoods with relatively high levels of integration relative to the median neighborhood in the United States. For example, under this definition, a census tract that is 49 percent black and 51 percent white would be classified as non-integrated.

The second definition—which we refer to as "shared neighborhoods"—uses a broader definition of integration, identifying neighborhoods as integrated if any community of color accounts for at least 20 percent of the tract population AND if the tract is at least 20 percent white. While this definition might be expanded to include neighborhoods in which any two groups account for at least 20 percent of the tract's population, this definition requires that the neighborhood population be at least 20 percent white because of whites' long history of exclusionary practices as well as attitudinal surveys suggesting that, on average, whites are less willing that other groups to live in integrated neighborhoods.

Both definitions suggest that the number of integrated neighborhoods—and the share of the US population living in integrated neighborhoods—increased between 2000 and 2011-15 (a time when the white, non-Hispanic share of the population fell from 69.1 percent to 62.3 percent). The number of "no-majority neighborhoods" increased from 5,423 census tract in 2000 to 8,378 in 2011-15, and the share of the US population residing in such tracts increased from 8.0 percent in 2000 to 12.6 percent in 2011-15 (Figure 1).

Similarly, the number of "shared neighborhoods" increased from 16,862 tracts in 2000 to 21,104 tracts in 2011-15, and the share of the US population residing in "shared" tracts increased from 23.9 percent in 2000 to 30.3 percent in 2011-15. These figures are higher than the estimates for "no-majority" tracts, reflecting the broader definition of integration used to define "shared neighborhoods." Nonetheless, both definitions show increases in integration between the 2000 Census and the 2011-15 ACS.

Notes: "No-majority neighborhoods" are census tracts in which no racial or ethnic group accounts for 50 percent or more of the population. "Share d neighborhoods" are census tracts in which whites account for 20 percent or more of the population and any community of color accounts for 20 percent or more of the population. 
N=71,806 Census tracts.

While the share of the population living in integrated neighborhoods has increased since 2000, most Americans continue to live in non-integrated areas, with white individuals the least likely to live in integrated areas. While 12.6 percent of the total US population lives in one of the 8,378 "no-majority" tracts, these neighborhoods include just 7.2 percent of the nation's whites, compared to 20.3 percent of blacks, 20.3 percent of Hispanics, 30.9 percent of Asians, and 19.5 percent of individuals of other races and ethnicities (Figure 2).

A similar pattern is present within "shared neighborhoods." While 30.3 percent of the total US population lives in one of the 21,104 "shared neighborhoods," these neighborhoods include just 22.9 percent of the nation's whites, compared to 43.0 percent of blacks, 42.8 percent of Hispanics, 44.8 percent of Asians, and 36.5 percent of individuals of other races and ethnicities.

Note: Estimates show the percent of each group that live in integrated neighborhoods. White, black, Asian, and Other are non-Hispanic.

The research brief provides more specific details about the relative composition of integrated and non-integrated neighborhoods by race, ethnicity, and other demographic characteristics. Additionally, it describes the stability of integrated neighborhoods between 2000 and 2011-15, as well as the geographic distribution of integrated neighborhoods across central cities, suburbs, and non-metropolitan areas.

Taken together, this evidence offers support for the conclusion that the number of integrated neighborhoods has increased in recent years. However, it also highlights that this conclusion is subject to two important caveats. First, some portion of the increase in integration reflects neighborhood change processes associated with the gentrification and displacement pressures affecting the central areas of many cities. While some of these neighborhoods may become stably integrated areas, it is not yet clear how many of the newly integrated neighborhoods will become stably integrated and how many will eventually become non-integrated areas.

Second, integrated neighborhoods remain the exception rather than the rule in the United States. The 2011-15 ACS shows that fewer than one in three Americans lives in a shared neighborhood, with just 12.6 percent living in "no-majority neighborhoods." As the country moves toward a population in which people of color are projected to be a majority by the middle of the century, further growth will be necessary for such changes to produce a more inclusive society.

Wednesday, September 6, 2017

Rebuilding Housing in Harvey’s Aftermath: Two Lessons from Hurricanes Katrina and Rita

by Jonathan Spader
Senior Research Associate
As floodwaters finally subside in Houston, and as Florida residents prepare for Irma, residents, civic leaders, and policymakers can glean two important lessons from the intensive efforts to rebuild homes and communities after Hurricanes Katrina and Rita, two devastating storms that hit the U.S. in back-to-back succession in 2005. 

First, rebuilding residential properties is a lengthy process likely to take several years. Second, the rebuilding process will be especially lengthy for rental properties (as compared to owner-occupied homes), which could greatly affect the 950,000 renters (who account for 41 percent of households) in the greater Houston metropolitan area, as well as additional renters affected by Hurricane Harvey in elsewhere in Texas and in other states. The slower pace of rental rebuilding is due to several factors including both renters’ dependence on property owners to rebuild rental housing units and historical differences in the availability and terms of federal aid for rental property owners as compared to homeowners.

To be sure, the need for emergency assistance and shelter for displaced residents will continue for weeks to come. Nevertheless, Congress is already starting to discuss an aid package. Moreover, the extensive damage (and the need to reauthorize the National Flood Insurance Program before September 30) may spur new efforts to develop policies and programs to support housing recovery in the wake of future natural disasters. As policymakers, civic leaders, and local residents begin to focus on the rebuilding process, they might want to keep the following in mind.

Extensive flooding from Hurricane Harvey in Southeast Texas. Air National Guard photo by Staff Sgt. Daniel J. Martinez

1. Rebuilding residential properties takes time.

An initial lesson from Hurricanes Katrina and Rita is that the rebuilding process takes time, with many properties continuing to show observable damage several years after the storms had passed. In early 2010—almost five years after both hurricanes made landfall—a HUD-commissioned study that I worked on surveyed the exterior conditions of properties damaged by those storms. The survey produced representative estimates of the rebuilding outcomes of properties that experienced “major” or “severe” damage—defined by FEMA as $5,200 or more in storm-related damage—that were located on significantly-affected blocks—defined as a city block on which three or more properties experienced “major” or “severe” damage.

The survey found that 17 percent of hurricane-damaged properties in Louisiana and Mississippi still showed substantial repair needs as of early 2010, almost five years after the storms had hit. Almost half these properties did not meet the U.S. Census Bureau’s definition of a “habitable structure,” a housing unit that is closed to the elements with an intact roof, windows, and doors and does not show any positive evidence (e.g. a sign on the house) stating that the unit was condemned or was going to be demolished. Only 70 percent of hurricane-damaged properties in Louisiana and Mississippi were rebuilt by early 2010, and 13 percent contained cleared lots in which the damaged property had been removed from the parcel (Figure 1).

In the case of Hurricanes Katrina and Rita, the properties that still were damaged included some whose owners had received rebuilding grants through federal programs designed to aid housing recovery. The largest source of assistance following the 2005 hurricanes was the $18.9 billion special Community Development Block Grant (CDBG) appropriations passed by Congress between 2005 and 2008. Some portion of the properties with remaining damage likely also reflect abandonment by owners who moved elsewhere in the wake of the hurricanes. For such properties, funding for demolition, rehabilitation, and land banking may be necessary to transition the properties to a new use, and potentially to support efforts to encourage residents to rebuild in areas with lower flood risks.

Notes: Sample is representative of properties in Louisiana and Mississippi that experienced major or severe hurricane damage and that were located on significantly-affected blocks. Rebuilt structures are residential structures that do not show substantial repair needs as defined in Turnham (2010). Cleared lots contain an empty lot or a foundation with no standing structure. Damaged structures are residential structures that show substantial repair needs—and include all uninhabitable structures. Uninhabitable structures are residential structures that do not meet the Census definition of habitability. 

2. Rental properties were rebuilt more slowly than homeowner properties.

A second lesson from the rebuilding process following Hurricanes Katrina and Rita is that rental properties were rebuilt more slowly than owner-occupied homes. This likely was due to several factors. While homeowners directly control the rebuilding progress of their home, renters are dependent on landlords’ rebuilding decisions. Smaller “mom-and-pop” landlords may also be slower to rebuild investment properties if their own home is also damaged. And policymakers have been wary of providing rebuilding assistance to rental property owners who did not purchase sufficient insurance.

Following Hurricanes Katrina and Rita, both Louisiana and Mississippi used the CDBG special appropriations for disaster recovery to create rebuilding assistance programs for homeowners and small rental property owners. (Texas, which faced less damage from Hurricanes Katrina and Rita, created only a homeowner program.) In both Louisiana and Mississippi, the homeowner programs covered much of the difference between the estimated cost to rebuild and the amount available to the homeowner from insurance and other rebuilding-assistance programs. Conversely, the grant programs for 1-4 unit small rental properties included a more complex set of eligibility requirements that included commitments for the rebuilt units to be rented to qualifying low- and moderate-income tenants. The result was that few rental property owners applied for and received rebuilding assistance, compared to widespread take-up of the homeowner assistance programs. While concerns about the incentive effects associated with bailing out under-insured investors are reasonable, a secondary effect was to reduce the number of rebuilt properties available to renters.

Figure 2 displays the share of hurricane-damaged properties on significantly-affected blocks that received a rebuilding grant through the CDBG-funded homeowner and small rental programs, along with the share of homeowner and small rental properties that were rebuilt by early 2010. The results show that 58 percent of hurricane-damaged homeowner properties in Louisiana and Mississippi received a rebuilding grant, compared to 10 percent of small rental properties. While this rental figure is limited to 1-4 unit small rental properties, a GAO report similarly found that federal assistance through CDBG, the Individual and Households Program, and the Home Disaster Loan Program together reached only 18 percent of all damaged rental units (including units in larger multi-family buildings), compared to 62 percent of damaged homeowner units. The rebuilding outcomes documented in the HUD-commissioned survey also showed sizable gaps, with 74 percent of homeowner properties rebuilt by early 2010 compared to 60 percent of rental properties.

A final question for policymakers is whether to use this opportunity to create a permanent program to support housing recovery following natural disasters. While Congress has relied on the CDBG program for this purpose since the early 1990s, its role is currently defined by the special appropriations legislation drafted following each individual disaster. Making disaster recovery a permanent function of the CDBG program (or creating some other permanent program for housing recovery) would allow HUD to develop permanent regulations and program guidance in anticipation of future disasters. While it is too late for this change to benefit victims of Hurricane Harvey, it might improve preparedness for the next disaster.