«reNtal HousiNg Policy iN tHe uNited states Volume 13, Number 2 • 2011 U.S. Department of Housing and Urban Development | Office of Policy ...»
Exhibit 3 shows the percent of single-family home sales that fell into the directly affected category by year. In 2004, when prices were at (Cleveland) or near (suburbs) their peak, the percentage of affected sales hovered near 35 and 10 percent, respectively. As the percentage of affected houses climbed, house prices declined, and, by the first quarter of 2009, 85 percent of the sales in Cleveland and two-thirds of the sales in the suburbs were affected. These data reflect a dramatic change in the market—arms-length sales numbered fewer than one in five in the city and fewer than one in three in the suburbs. The median sale was now an affected sale and the median prices, shown in exhibit 4, clearly reflect that.
Prices in the affected market have been substantially lower since 2004; houses are selling at roughly 50 percent of the price in the nonaffected market. Because affected sales constitute a larger component of the market over time, the median price of all sales continues to drift closer and closer to the affected price. Through 2008, however, houses selling in the arms-length market had seen only modest declines. Compared with the data in exhibit 1, the data in exhibit 4 are much more illustrative of actual market conditions.
The distinction between the affected and arms-length market also plays out at the neighborhood level. Exhibits 5 and 6 show the sales counts and median prices for each segment in Cleveland’s Ward 12, which includes Slavic Village, a neighborhood that received much attention for its role in the housing crisis, both nationally and in Cleveland. From the beginning of 2004 to the end of 2007, affected sales increased five fold. At the same time, arms-length sales, which were increasing from 2003 to 2005, began to slide, and they have yet to recover. Prices in the arms-length market Exhibit 3 Percent of Single-Family Home Sales Identified as “Affected,” Cuyahoga County, Ohio, by Quarter, 2004–2010 Percent
Exhibit 5 Single-Family Home Sales by Market Segment, Ward 12, Cleveland, Ohio, by Quarter, 2003–2010 Number of single-family home sales Single-Family Home Median Sales Prices, Ward 12, Cleveland, Ohio, by Quarter, 2003–2010 100,000 90,000 80,000 70,000
remained strong (hovering around $80,000) until early 2008, when affected sales began flooding that market. Sales prices since then have been erratic (partly because of the sparse number of sales).
Using this sales dichotomy at the neighborhood level can identify locations where arms-length activity is still occurring, or, at the other end of the spectrum, identify when neighborhoods hit a turning point in which the market fails to distinguish between affected and nonaffected properties.
In Ward 12, the arms-length market seems to have held on in volume until early 2005 and in sales price until 2008.
Exhibits 3 through 6 relied on combining sales data with foreclosure filing data, linking the two sets of data by parcel number. For each sale, we checked to see if there had been a sale in the last 2 years with a deed type that indicated a sheriff’s sale. The analysis also cross-referenced the foreclosure filing data to determine if the sold property had a foreclosure filing in the previous 2 years.
Lesson #3: What Is Selling Now Is Not the Same as What Was Selling Previously Tracking the median home sales price over time implicitly assumes measuring the same market from one year to the next, so that a 5-percent price increase represents a change in the value of the market, not a change in its composition. For example, if in one year all the houses that sold in a
neighborhood were below average size and in the next year they were all above average size, one would anticipate a price increase from one year to the next simply because the larger houses were selling.
Exhibit 7 shows a change in value-composition occurring gradually since 1999, but accelerating since 2004, in the eastern suburbs of Cuyahoga County. For each year, the county auditor provides the estimated market value of each property (used for taxation purposes). Analysts at the Center then divide the entire housing stock into quartiles based on those values. For the transactions in each year, they run a frequency distribution on the value quartile and calculate the percentage of all houses that sold from each of the four value quartiles. If sales were balanced, they would expect 25 percent of all sales to come from each value quartile, as was roughly the case, for example, in 1993.
By 2008, however, the top 25 percent of all valued homes constituted only 17 percent of all sales.
Similarly, one-third of all houses sold came from the lowest 25 percent value quartile. Thus, part of the downward pressure on median prices came from the fact that lower valued houses made up a larger portion of the overall market. This finding builds on the information displayed in exhibit 4.
Not only are different market segments at work, but also larger numbers of lower valued houses are entering into each of those markets.
Exhibit 7 Single-Family Home Sales by Value Quartile, Eastern Suburbs, Cuyahoga County, Ohio, 1993–2009 Percent of all sales
Lesson #4: “Flipping” Went From a Red Flag to Worse Before the housing crisis, the region’s housing market experienced its fair share of “flipping,” which is when a homebuyer purchases a house and then resells it very quickly (within 90 days, by the Center’s definition) at a substantially higher price (25 percent higher, by the Center’s definition).
The analysts tracked and reported on these quick resales because the large price changes over such a short period of time were a concern. Under some flipping scams, very little, if any, renovation was done, but the house would be appraised and resold to an unsuspecting buyer for much more than its true value. This scenario was ripe for subsequent mortgage default. The Center created lists of those individuals involved with these transactions, and the analysts profiled the activities of those individual actors, over time, by price level and location.
As the local market worsened, these quick resale analyses unearthed a dramatic change in the quick resale market, as shown in exhibit 8. The median resale price was $90,000 in 2006, but it plummeted to less than $4,000 in 2008, and the volume of sales doubled from 2007. Although sales volume has since cooled, the 2010 median resale price is still less than one-fourth of its 2006 level.
Within a short period of time, the quick resale market focused on an entirely different type of structure. It went from what was likely questionable flipping of inhabitable housing to desperation sales—property-churning of dilapidated houses that are unlikely to be occupied again.
182 Data Shop Separating the Good From the Bad From the Ugly: Indicators for Housing Market Analysis Conclusions Any single indicator presented likely generates several questions regarding why it was selected over another. Although analysts at the Urban Center engaged in much learning-by-doing, they were guided by a handful of data goals in their analyses. They used data that were available monthly and at the parcel level—primarily sales and foreclosure filings. Because the market changed quickly in our region, they wanted the most disaggregate, recent, and helpful data available.
The analysts kept it simple, both methodologically and graphically. They were producing these analyses for busy people who were not housing experts. The consumers of our research would not have all day to mull over its methodological nuances. They made maps, but often only to investigate findings produced via the indicators they have shared here. On the other hand, the analysts produced most of their work at a variety of aggregation levels. A Geographic Information System (GIS) is not necessary to conduct these analyses. Spreadsheet (such as Microsoft Excel) skills are necessary, however, and it is essential to have a way to join data on a common field (typically, a parcel number), which can be accomplished with SAS, SPSS, or Microsoft Access. This opens up the possibility of linking different data sets that contain information on a common object, such as a house, parcel, or neighborhood.
Many of the analysts’ data decisions appear, and, in fact, were arbitrary. Their analyses were exploratory, and they had little past research to guide them. It is, of course, possible to test these data decisions for robustness and to experiment with different indicators over shorter or longer periods of time. Finally, but by no means trivially, the analysts cleaned the data that they used, making their analyses slightly more involved than what they have shown in this article. They did not delve into these details because they anticipate that other analysts will have different data issues than the Urban Center analysts had. The Urban Center will provide these details to interested analysts.
At a time when many thought the entire Cleveland market was “the ugly,” the Center’s research goal was to identify the good (nonaffected sales), the bad (affected sales and changed value composition), and the ugly (the balance between affected/nonaffected sales and the change in nature of the quick-resale market) in a way that was useful for local policymakers. The goal in this article is to relate that process and its results in a way that can be useful for other housing analysts in markets facing similar challenges.
Acknowledgments Tom Bier and Ivan Maric provided useful input on the indicators reported in the article. Cuyahoga County’s Don’t Borrow Trouble foreclosure prevention program, the Cleveland City Council, and Cleveland’s Department of Community Development provided funding for the research in which we developed these indicators.
Authors Brian A. Mikelbank is an associate professor of urban studies at Cleveland State University.
Charlie Post is a research associate at the Urban Center at Cleveland State University.
184 Data Shop Impact A regulatory impact analysis must accompany every economically significant federal rule or regulation. The Office of Policy Development and Research performs this analysis for all U.S. Department of Housing and Urban Development rules. An impact analysis is a forecast of the annual benefits and costs accruing to all parties, including the taxpayers, from a given regulation. Modeling these benefits and costs involves use of past research findings, application of economic principles, empirical investigation, and professional judgment.
Regulatory Impact Analysis:
Emergency Homeowners’ Loan Program Michael K. Hollar U.S. Department of Housing and Urban Development Program Summary The Emergency Homeowners’ Loan Program (EHLP), as enacted in the Dodd-Frank Wall Street Reform and Consumer Protection Act, allows the U.S. Department of Housing and Urban Development (HUD) to provide a maximum of $50,000 to homeowners who are 90 or more days delinquent on their mortgages due to a 15-percent or greater reduction in household income and face the threat of foreclosure. Reasons for the reduction of income are limited to involuntary unemployment, involuntary underemployment, and medical conditions. EHLP participants must come from households that earned no more than 120 percent of Area Median Income (AMI) before the decrease in income. EHLP provides assistance through a 5-year, no-interest loan, with loan repayment beginning after program assistance ends. Payments cease after 24 months or $50,000, whichever comes first, which allows up to 7 years from loan disbursement to full repayment. Finally, EHLP assistance is limited to homeowners in Puerto Rico and in the 32 states that are not assisted by the Department of the Treasury’s Innovation Fund for Hardest Hit Housing Markets program.
Cost-Benefit Analysis EHLP is intended to assist a segment of delinquent homeowners who face a high probability of foreclosure and have become delinquent because of a temporary loss of income. Assisted households are expected to recover financially within 24 months. The benefits of this program’s rules include the avoidance of costs associated for (1) owners of foreclosed properties, (2) lenders holding mortgages Cityscape 185 Cityscape: A Journal of Policy Development and Research • Volume 13, Number 2 • 2011 U.S. Department of Housing and Urban Development • Office of Policy Development and Research Hollar on foreclosed properties, (3) homeowners living near the foreclosed properties, and (4) local governments.
Overall, the benefits of this rule are estimated to be between $928 million and $1.9 billion, offset by administration costs, namely participant selection ($87.3 million) and servicing the EHLP loans ($7.4 to $11.3 million), and up to $29.5 million of incremental costs of foreclosure to lenders caused by borrowers assisted by EHLP who subsequently default anyway. In addition, participants in this program receive a transfer ranging from $28.32 to $43.3 million, which is equal to the government’s cost of borrowing the funds. Lenders also receive a transfer totaling $105 to $213 million, which includes costs related to the mortgage, such as interest payments, from the homeowner.
Demand for EHLP Loans The amount of EHLP assistance for which a homeowner qualifies depends on the monthly mortgage payment and current income. Under program rules, homeowners are required to pay their monthly mortgage payments equaling up to 31 percent of their current monthly income. EHLP assistance can cover the remaining mortgage amount, for a period of up to 24 months. EHLP assistance can also be used to pay delinquent mortgage payments (principal and interest), taxes, insurance, and certain other related fees.