Real Estate Research provides analysis of topical research and current issues in the fields of housing and real estate economics. Authors for the blog include the Atlanta Fed's Kristopher Gerardi, Carl Hudson, and analysts, as well as the Boston Fed's Christopher Foote and Paul Willen.

November 18, 2014

Can the Atlanta Fed Construction and Real Estate Survey Predict Home Sales?

The slow recovery in housing remains an item of note in statements from the Financial Open Market Committee. That it's still something of a concern means that many people pay attention to housing-related data releases, several of which are due out this week, because they can shed some light on the direction of housing and the economy. The builder confidence index, released today, got things off to a good start by showing a four-point rise, from 54 to 58 (values greater than 50 mean that more builders view conditions as good rather than as poor). House starts and existing sales are due Wednesday and Thursday, respectively.

At the Atlanta Fed, we conduct a monthly survey of regional builders and real estate brokers to get their perspectives of the market. In August, we began to look at the results a little differently to see if they could tell us anything about subsequent housing-market data releases. In that exercise, we investigated the correlation between the expectations of our homebuilder contacts for construction activity and subsequent housing starts. We found that our builders are on point, more or less, and we reported on that discovery in an August post. We recently repeated the exercise, this time to explore the predictive power of the outlook for home sales of our homebuilders and residential brokers for subsequent new and existing home sales data releases. We report on our findings in this post.

Brokers and builders expect new home sales to rise
The September home sales data showed us that existing single-family home sales increased by 1.9 percent from the year-earlier level and new home sales increased by 22.6 percent. This news is fairly consistent with the reports we received from our real estate business contacts about September sales activity; more brokers and builders noted an increase than noted a decrease in home sales activity from the year-ago level.

But what exactly did our survey respondents tell us about their outlook for home sales? Diving deeper into the data, we find that brokers' and builders' outlooks remain mildly positive and that the two groups have tracked each other fairly closely in recent years. (In the pre-2011 period, brokers and builders diverged more sharply.) Specifically:

  • Of builder respondents, 40.0 percent indicated that they expect new home sales to increase over the next three months, 32.0 percent expect activity to decline, and 28.0 percent expect home sales activity to remain about the same. The home sales outlook diffusion index value for builders was 0.08.
  • Of broker respondents, 22.5 percent indicated that they expect new home sales to increase over the next three months, 27.5 percent expect activity to decline, and 50.0 percent expect home sales activity to remain about the same. The home sales outlook diffusion index value for brokers was -0.05.


The chart below features two scatter plots of the diffusion index value for the broker and builder home sales outlook on the horizontal axis and the year-over-year change in the three-month moving average of single-family home sales (for Alabama, Florida, Georgia, Louisiana, Mississippi, and Tennessee) on the vertical axis. Given that we are asking contacts to be forward-looking, we lag the contact responses.


Do home sales expectations correlate with subsequent sales data?

Three things stand out on this chart. First, if builders and especially brokers (who tend to be an optimistic lot) predict a decline, the subsequent home sales data release will probably be poor. Only a modest bit of net optimism is of little comfort—some of the worst declines occurred in years with net positive (albeit small net positive) outlooks.

Second, for builders, if the index is greater than 0.3, we find that sales generally grow—except for between August 2012 and April 2013, when sales did not match builders' optimism. When the broker index is above 0.3, sales either grow or decline by a smaller amount than when the index is negative. Like the builders, the broker panel missed the sales declines from August 2012 and April 2013. The brokers also missed the declining real estate market in 2006 to early 2007 (see the green triangles in the chart above)—despite a declining market, the broker index remained lofty until May 2007.

Third, the official statistics on housing sales could go either way when index values are between ‑0.1 and 0.3. This shouldn't come as a complete surprise, particularly because a diffusion index value near zero (regardless of whether that value is positive or negative) indicates that responses from contacts were mixed. And as we can see in the scatter plot above, large declines were much more likely given the time period covered.

A simple regression indicates that the outlook could explain just under 50 percent of the variation in sales measure, which indicates that our poll does a decent job of predicting subsequent sales. Given this finding, what do we now expect home sales to look like? The most recent downward trend in respondents' outlook puts the diffusion index in the center, suggesting that declines in seasonally adjusted sales over the next several months are just as likely as increases in sales.

The poll was conducted October 6–15, 2014. Sixty-five business contacts across the Southeast participated (40 residential brokers and 25 builders). To explore the latest poll results in more detail, please visit our Construction and Real Estate Survey page.

Photo of Carl Hudson By Carl Hudson, director for the Center for Real Estate Analytics in the Atlanta Fed´s research department, and


Photo of Jessica DillJessica Dill, senior economic research analyst in the Atlanta Fed's research department

November 18, 2014 in House price indexes, Housing demand, Housing prices | Permalink


Post a comment

Comments are moderated and will not appear until the moderator has approved them.

This blog allows comments only from registered users. Be sure your comments are appropriate before submitting them. To comment, please sign in.

December 04, 2013

Part 2: What Caused Atlanta's House Prices to Drop Again in 2011?

Note: This post is a follow-up to the November 8 post, "What Caused Atlanta's House Prices to Drop Again in 2011?"

Case-Shiller recently released its September data. Once again, the data show that Atlanta's year-over-year performance outpaced Case-Shiller's 20 metro area index—18.7 percent versus 13.3 percent, and Atlanta's low tier experienced its fifth consecutive month of year-over-year returns in excess of 50 percent (53.2 percent).

In the November 8 post, which explored the Case-Shiller house price tiers, we noted that from July 2011 to March 2012, both Atlanta thresholds—the low–middle and the middle-high—had noticeable declines, which corresponded to the time when the low tier's year-over-year performance began to recover. Given the methodology, we said that either all prices declined or a greater proportion of transactions came from lower-valued houses. Data on mortgage closings provide evidence consistent with the idea that investor activity in Atlanta's low tier influenced the market.

We can distinguish investor activity from "typical" house purchases by looking at the form of financing. Most owner-occupiers use mortgage financing whereas investors usually purchase with cash. Thus, if there is substantial investor activity in one of the three tiers, then we would expect that tier to be underrepresented in the number of mortgage closings.

To dig further into the issue of the 2011 house price drop, we looked at the distribution of mortgage closings by tier in the Lender Processing Services (LPS) Applied Analytics database. Residential mortgage servicing data from that database contain records from the servicing portfolios of the largest residential mortgage servicers in the United States. It covers about two-thirds of installment-type loans in the residential mortgage servicing market.

The chart displays the fraction of mortgage closings by Case-Shiller by tier from third-quarter 2008 to third-quarter 2013. We created the data for the chart by using each mortgage's sales price and assigning it to a tier as defined by the Case-Shiller thresholds for the month the mortgage closed (we excluded refinances). Once we'd bucketed the data this way, we calculated each bucket's percentage of the month's closing. If cash purchases were evenly distributed and the set of servicers in the LPS database is representative of Atlanta's overall market, we would expect each bucket to be one-third of the total.

Atlanta Mortgage Closings

From September 2008 to October 2011, the closings appear to be evenly distributed among the three buckets, with the shares varying between 25 and 43 percent. The average share for the low, middle, and upper tiers were 36 percent, 30 percent, and 34 percent, respectively. After November 2011, the low tier's share fell to an average of 18 percent, with a low share of 12.8 percent in May 2012.

Although the chart does not conclusively prove that investors entered the market en masse to purchase houses in Atlanta's low tier, the timing of the noticeable decline in the low tier's share of mortgage closings does coincide with the fall in Atlanta's low–middle and middle–high thresholds and the bottoming of the low tier's year-over-year price declines. More recently, the low tier's share of mortgage closings has been at its highest since November 2011—perhaps a sign that investor interest has cooled and we are now looking at a more normal market.

But with year-over-year prices in the low tier rising rapidly, let's hope buyers aren't expecting 50 percent year-over-year gains to be normal.

Photo of Carl HudsonCarl Hudson, Director, Center for Real Estate Analytics in the Atlanta Fed's research department

December 4, 2013 in House price indexes, Housing crisis, Housing prices | Permalink


Post a comment

Comments are moderated and will not appear until the moderator has approved them.

This blog allows comments only from registered users. Be sure your comments are appropriate before submitting them. To comment, please sign in.

November 08, 2013

What Caused Atlanta's House Prices to Drop Again in 2011?

What happened in Atlanta real estate the second half of 2011 and the first half of 2012? I asked myself this question after looking at the recent release of the Case-Shiller Home Price Indices for August. Atlanta home prices have recently been increasing at a faster rate than the composite index. Last month, the United States saw a 12.8 percent year-over-year increase in house prices while the Atlanta index rose 18.4 percent.

It wasn't long ago that things were much worse in Atlanta. From the pre-bust peak, the city experienced a 37 percent decrease in its Case-Shiller index versus a 33.8 percent decrease for the Case-Shiller 20-metro composite. From July 2011 to March 2012, Atlanta home prices took a second nosedive, almost as large as the initial bust in 2009 (see Chart 1). So what happened?

S&P/Case Shiller Home Price Index: Year-over-year growth in seasonally adjusted index

The Case-Shiller index is a repeated sales index, which means it uses the price change between two arms-length sales of the same single-family home. One way to gauge the amount of activity in a market is to look at the number of "sales pairs" in a period. To get a sense for whether Atlanta is experiencing particularly high volumes, we can look at Atlanta's sales volume relative to the nation's. In the third quarter of 2011, Atlanta's sales began to grow substantially, and Atlanta's share of composite sales pairs peaked in March 2012 at 9.7 percent, which is a much greater percentage than the 5 percent to 6 percent range from 2000 to 2005.

Around this time, many of the Atlanta Fed's local contacts reported that some investors were buying up distressed home to convert into rental property. Case-Shiller breaks its index into three price tiers—low, middle, and high. Looking at the tiers in Atlanta for the most recent data, the high end was up 12.9 percent year over year; the middle tier, 27.7 percent; and the low tier was up 52.5 percent. Looking back, we see that the growth rate in Atlanta's low-tier index started to recover in July 2011 (see Chart 2). It was not until March 2012 when the year-over-year changes in the middle and high tiers started their recent upward trends.

S&P/Case Shiller Home Price Index: Year-over-year growth in tiered indexes for Atlanta

The price thresholds for the three tiers are computed using all sales for each period and are set so that each tier has the same number of sales. From July 2011 to March 2012, both thresholds (low–middle and middle–high) had noticeable declines (see Chart 3). Given the methodology, either all prices declined or a greater proportion of transactions came from lower-valued houses. Note that after March 2012, the breakpoints started to increase, which was the same time as when the year-over-year growth in the middle and high tiers started to improve.

S&P/Case Shiller Home Price Index: Atlanta Sales Pairs and Tier Break Prices

Further work is needed in order to determine whether there really was a ramp-up in activity in the low end of the market. If such activity did occur, it raises the question as to what was driving the activity—could it have been investors? If not, how was this activity financed? Was this a case of inventory being absorbed, prices adjusting, and momentum moving from investors to "normal" buyers?

The low tier warrants attention given the fact that it may have driven Atlanta's recent house price performance. Understanding the July 2011 to March 2012 period may shed light on the factors that could influence the market going forward.

Photo of Carl HudsonCarl Hudson, Director, Center for Real Estate Analytics in the Atlanta Fed's research department

November 8, 2013 in House price indexes, Housing crisis, Housing prices | Permalink


Good analysis. When I see volatility like this in home price data the answer is either distressed sales which distort the price index or it is a large increase in sales activity in either lower or higher priced homes. People need to understand that these indexes are distorted by these two factors and adjust their thinking accordingly. Good analysis.

Posted by: Jack | November 08, 2013 at 08:34 AM

Post a comment

Comments are moderated and will not appear until the moderator has approved them.

This blog allows comments only from registered users. Be sure your comments are appropriate before submitting them. To comment, please sign in.

July 24, 2013

The Shape of the Housing Recovery in Atlanta

As with politics, all real estate is local, and it seems that it is always someone else's neighborhood that's doing well. By all indications, housing is recovering in 2013. Nationally, starts and prices show promising growth. Even in Atlanta, Georgia—a hotbed of subprime lending and speculative construction before the crash—home prices have made a strong rebound, showing a 21 percent year-over-year increase in April 2013 (see the chart). In this post, we address the questions: What shape is this recovery taking? Will this rebound look a lot like the early 2000s? Or can we expect permanent changes to the urban landscape post-crisis?

Strong Recovery in 2013 for Atlanta Home Prices

Using zip code-level data from CoreLogic to take a closer look at home price recovery in Atlanta, we see that this rebound is not evenly distributed. Instead, variation in growth rates is much higher than it used to be. In general, Atlanta home prices have been very depressed, but at the same time, contacts have remarked that in hot, single-family markets, prices are up, multiple offers are made as soon as a property is listed, and there is a general shortage of homes "where people want to move." The data confirms and adds to this anecdotal evidence: Atlanta's strong price growth is concentrated in select intown markets, as well as in many of the areas hardest hit.

Table 1: Variation Increases Prior to the crisis, home prices appreciated at about the same rate throughout the Atlanta metropolitan region. Yes, some areas were expensive and others affordable, but prices grew everywhere at roughly the same pace. But when the real estate market crashed in 2007, Atlanta home price rates of change began to diverge. In the past, zip codes with the highest growth rates grew 20 percent faster than zip codes with the lowest rates. Now, that ratio has risen to 300 percent.

To illustrate, the standard deviation in home price growth increased sharply during the crisis and continued to widen during the last year of recovery (see table 1). This pattern also describes the nationwide trend, although the increases in variation in Atlanta are more dramatic.

The obvious explanation for this is that during the crash, although prices fell everywhere, areas with concentrations of distressed properties fell more steeply, generating this variation. Now that the recovery is under way, though, will areas with a high density of distressed properties rebound? Or are these areas reset permanently at a lower level?

The evidence from Atlanta suggests that we will see a bit of both. Fast growth is concentrated in some of the areas that were hardest hit, as well as in some of the choicest neighborhoods in town. This extremely fast growth is paired with slow growth in many markets that never saw steep declines, generating a higher standard deviation in home prices.

The first map below shows year-over-year home price change during the recovery. The second map depicts peak and trough change, showing the depth of the decline. We see that two recovery stories emerge. First, north of I-20, areas that were quite resilient during the crisis and did not see strong declines are experiencing strong growth. Second, in the areas southeast of the interstate 285 perimeter, we see exurbs that were devastated by devaluation experiencing a strong rebound, with growth rates over 16 percent.

Atlanta Home Prices, Recovery

Atlanta Home Prices, Peak to Trough

How can we best understand this pattern of recovery? We reviewed a few likely correlates with home price increase: household income, household size, and density of high-risk lending and speculative construction during the bubble. None of these factors was significantly correlated with the 2013 rebound. The peak-to-trough change is significantly correlated with the rate of recovery, suggesting that much of this recovery is a price correction. Longer commute times are also correlated with recovery, revealing that demand is increasing in places that are not close to job centers, though it's possible that commute times are simply a proxy for severity of the crash as these areas also experienced the strongest declines.

Contacts tell us that neighborhoods with better school districts are performing well and recent investor interest may also be playing a role. What is certain is that Atlanta's strong overall house price growth is driven by increases in areas hardest hit by the housing downturn and by a few centrally located markets, and that underneath the citywide average there is a lot more variation than we have experienced in the past.

We'd love to hear your thoughts about these trends!

By Elora Raymond, graduate research assistant, Center for Real Estate Analytics in the Atlanta Fed's research department/PhD student, School of City and Regional Planning, Georgia Institute of Technology, and

Photo of Carl HudsonCarl Hudson, Director, Center for Real Estate Analytics in the Atlanta Fed's research department

July 24, 2013 in House price indexes, Housing crisis, Housing prices | Permalink


What about looking at the recovery based on nominal prices. When comparing current prices to troughs, the growth rates looks great but if prices are just back to year 2000 level prices, the recovery still does not look that strong.

Posted by: Marlon | September 11, 2013 at 11:26 AM

Post a comment

Comments are moderated and will not appear until the moderator has approved them.

This blog allows comments only from registered users. Be sure your comments are appropriate before submitting them. To comment, please sign in.

July 10, 2013

Why Housing Rebound May Continue at a Slower Rate Than Hoped For

Perhaps it's because I've worked with bank examiners for many years, but I often question financial news that seems too optimistic. On July 1, 2013, the U.S. Census Bureau reported that overall construction spending increased in May. Private residential construction, which generally leads the economy, grew 24.4 percent from May 2012 to May 2013. Beyond being cautious with one data point, I think that there are several reasons why housing's rebound may be slower than hoped.

To be clear, residential real estate conditions have been improving, albeit from record low levels of activity. Sales of both new and existing houses have been trending up recently, but remain near historically low levels. Additionally, the quantity of new and existing homes readily available for sale is low. Homebuilders in the Sixth Federal Reserve District (which includes Alabama, Florida, and Georgia and parts of Louisiana, Mississippi, and Tennessee) recently reported that new home sales and construction have been ahead of year-earlier levels and that buyer traffic remains strong (see this SouthPoint post). Builders noted, however, that access to financing and a shortage of developed lots continued to constrain construction activity. In conjunction with the recent construction spending data release, it is this last point that I aim to dig into a bit deeper in this blog post.

Since the housing bust, construction and development (C&D) lending has been in sharp decline in terms of aggregate dollars and as a percent of total bank assets. Using year-end data, we find that C&D loans peaked in 2007 at $629.4 billion. As of 2012, they stood at $203.8 billion. As of March 2013, C&D loans accounted for 1.4 percent of bank assets, unchanged from December 2012 and the lowest level since at least 1991. The decline in C&D lending is broad based given that similar trends are seen for banks under and over $1 billion in total assets. With the recent reports of growing construction spending, will bank lending practices dampen construction growth going forward?

Banks represent a significant funding source for homebuilders, especially nonpublic homebuilders. Using data from 1991 to 2012, there appears to be a strong, positive relationship between bank construction lending and private residential construction put in place—see the chart. Of course, it's impossible to tell from this chart whether construction activity is responding to changes in credit supply or credit supply is responding to changes in construction demand. However, banks have been extremely tight with credit in the aftermath of the financial crisis, and there aren't many signs that banks plan to change course any time soon. So it may be reasonable to assume that a continued reduction in bank C&D lending is likely to limit future gains in construction activity.

Construction Activity versus Credit Supply

A case for optimism
In conversations with banks of various sizes, two things are often repeated. First, bankers indicate there is little appetite for growth in C&D lending and second, banks of various sizes want to increase commercial and industrial lending (C&I). For many banks, a move from C&D lending to C&I lending is easier said than done—the skillsets needed for C&I lending differ from those associated with C&D. Acquiring C&I expertise is a challenge particularly for smaller banks. So what's a community bank going to do?

An old adage is to do what you know best. For many community banks that would be C&D lending. Given the reports of lot shortages and house inventory being low, it would seem that profitable opportunities for C&D lending exist. There is nothing wrong with C&D loans appropriately underwritten and subject to reasonable risk management. A key question is when banks start moving back to C&D lending, will they be able to resist the shortcuts of the last cycle? Let's hope that banks can successfully navigate a return to C&D lending so that the housing market can continue to recover.

Photo of Carl HudsonBy Carl Hudson, Director, Center for Real Estate Analytics in the Atlanta Fed's research department

July 10, 2013 in Credit conditions, House price indexes, Housing boom | Permalink


Post a comment

Comments are moderated and will not appear until the moderator has approved them.

This blog allows comments only from registered users. Be sure your comments are appropriate before submitting them. To comment, please sign in.

May 05, 2010

Can we identify foreclosure contagion effects?

Negative externalities of foreclosures are the primary reason that policymakers focus on implementing policies to avert foreclosures and keep families in their homes. If the costs of foreclosures were completely internalized by the households experiencing them, then the focus would likely be on a different set of policies—for example, providing rental housing assistance or counseling on how to rebuild credit histories. Despite their importance, the empirical evidence of negative externalities is extremely tenuous, because they are so difficult to measure accurately. The papers that have tried have, for the most part, found huge effects. One of the most-cited papers in the literature was written by Dan Immergluck and Geoff Smith. They looked at the Chicago housing market in the late 1990s and found significant negative effects of foreclosures on nearby property values: *

Cumulatively, this means that, for the entire city of Chicago, the 3,750 foreclosures that occurred in 1997 and 1998 are estimated to have reduced nearby property values by more than $598 million, for an average of $159,000 per foreclosure. This does not include effects on the value of condominiums, multifamily rental properties, and commercial buildings.

This is an enormous effect, and right away it should make us slightly skeptical because, like most papers that have attempted to estimate these externalities, it is based on a hedonic regression model. For readers who are not close to the academic housing literature, a hedonic regression model is simply a cross-sectional regression of housing transaction prices on characteristics of the house and neighborhood. The methodology can be very useful for measuring the price of certain housing characteristics (for example, how much an extra bathroom or bedroom is worth), but it is not so useful in measuring the contagion effect of foreclosures because it does not solve two severe econometric issues: a reverse causality problem stemming from the fact that declines in housing prices result in higher foreclosure rates (the recent crisis for example!), and the problem that there are likely many unobserved neighborhood characteristics that are correlated with both housing prices and the number of foreclosures in a given neighborhood.

In a recent paper published in the Journal of Urban Economics, John Harding, Eric Rosenblatt, and Vincent Yao try to overcome these econometric issues by employing the repeat-sales methodology that is usually used to estimate house price indices. This model uses the difference in sale prices for repeat transactions of the same properties to estimate the average trend of house prices in a given area. By taking differences, all of the characteristics of a property and neighborhood that do not change between the sales drop out, and so we do not need to account for them. The only characteristics that we need to worry about are those that vary over time (between sales).

Harding et al. make a slight modification to the repeat-sales methodology by including as an additional covariate the number of foreclosures surrounding a property in the regression. In this respect, the model becomes a hybrid between a repeat-sales regression and a hedonic regression. Most importantly, this methodology can control for the average trend in prices in an area to at least partially address the reverse causality issue—price declines, through their effect on equity positions, are causing increased foreclosures. In addition, because time-invariant property and neighborhood characteristics fall out of the regression, omitted variable bias (the possibility that there are unobserved variables correlated with foreclosures and house prices) is less of an issue, although it could still be a problem if there are time-varying unobserved variables that are correlated with both foreclosures and house values.

Findings support significant but reduced negative externalities
Using data from seven markets—Los Angeles, Atlanta, St. Louis, Charlotte, Las Vegas, Columbus, and Memphis—the authors find significant negative effects of foreclosures on property values, but the effects are smaller than those previous studies have found. According to the authors' estimates, the peak discount of a property's value due to a nearby foreclosure is about 1 percent, and this effect diminishes quickly as the distance to the foreclosure increases. The authors interpret these results to be contagion effects that largely come from poor aesthetics resulting from the deferred maintenance and neglect of properties in the state of foreclosure. In the conclusion, they state (p. 178): "We interpret these different patterns as suggesting that the negative externality from immediate neighbors is attributable to property neglect and uncertainty about the future owner."

In our opinion, this paper is a significant improvement over the previous literature, as it includes a number of methodological improvements over and above its use of the repeat-sales method. With their data, the authors are able to pinpoint two aspects of foreclosed properties that the previous literature has not been able to identify. First, the authors can identify the particular legal phase of any foreclosure proceeding. That is, they know when a lender has filed the initial foreclosure documents, when it has taken legal possession of a house, and when it has sold a house to a new owner. Second and perhaps most important, the authors are able to geocode the location of each foreclosure relative to any house that contributes observations to the repeat-sales dataset. The authors then draw four concentric rings with different radii (0–300 feet, 300–500 feet, 500–1,000 feet, and 1,000–2,000 feet) around each repeat-sales transaction and count the number of foreclosures in each ring. Consequently, in their empirical regressions, the authors can control for the distance between a repeat-sales transaction and any surrounding foreclosures, as well as account for the particular phase of the foreclosure process for each house in each ring. Consistent with the intuitive concept of contagion, the authors find that the negative effect of a foreclosure on the prices of other homes diminishes with distance. Moreover, the negative effect is strongest around the time of the foreclosure auction/sale and the real-estate owned (REO) sale, as opposed to the time period before the auction/sale. This finding also makes sense because the time after the formal foreclosure and before the REO sale is when the property is most likely to be in a state of deferred maintenance.

True contagion effects may be even smaller
The paper's empirical findings that both distance and the phase of the foreclosure process matter are not only very intuitive, but they also provide quite a bit of evidence in support of the contagion hypothesis. But for reasons we describe below, we believe the true effects of contagion may be even smaller than the reduced effects that the authors find.

First, there could be significant measurement error causing an upward bias in their estimates of the contagion effect. The authors estimate separate regressions for each of the seven Metropolitan Statistical Areas (MSAs) in their sample and are thus able to control for average price appreciation at the level of the MSA. However, an MSA is a relatively large geographical area that includes many heterogeneous areas. For example, in the Boston/Cambridge metro-area there are wealthy areas like Brookline and very poor areas like Dorchester. House price trends were very different in these areas, and foreclosure levels were also extremely different. Not controlling for these different trends could bias the estimates of the contagion effect. For example, since Dorchester experienced significantly more house price depreciation than the average area in Boston, the residuals corresponding to properties in Dorchester in the regression will be mostly large and negative. In addition, Dorchester experienced significantly more foreclosures. If the larger price declines caused the increased foreclosures in Dorchester (and likewise the smaller price declines caused the lower foreclosure numbers in Brookline), then the residuals will be correlated with the number of foreclosures, and the contagion estimate will be biased upward. One way to try to address this problem would be to estimate the repeat-sale regressions at a more disaggregated level, such as the town/city level or even at the ZIP code level.

Another potential problem comes in the way the authors treat REO sales. REO sales are not used in the construction of the repeat-sales pairs and thus are not reflected in the independent variable in the regressions. This is a normal assumption to make when constructing repeat-sales price indices, with the rationale being that distressed sales may not reflect true market prices. This approach implies that the estimates of the average MSA price trends in the regressions do not reflect foreclosure sales. But, if foreclosure sales do lower sale prices of non-distressed properties through channels independent of contagion, such as by increasing the supply of houses on the market, and the price declines result in more foreclosures (through the channel discussed above), then the estimated contagion effect will be biased upward. Basically, this would introduce measurement error into the price trend, which would in turn be correlated with the foreclosure contagion variables in the regression. However, the authors could easily check for error by simply including REO sales in the repeat-sales sample to see how the contagion estimates are affected.

Finally, as the authors acknowledge, there could be some omitted time-varying property or neighborhood characteristic that is correlated with both the residuals and the number of foreclosures surrounding a property. The authors try to deal with this issue by placing restrictions on their sample of repeat-sale pairs to eliminate properties that have likely changed significantly over time, and find the results to be robust to such changes. This finding certainly takes care of property characteristics that may be changing significantly over time and adding (or subtracting) value from the property, but it does not control for neighborhood characteristics.

The authors also use an instrumental-variables (IV) strategy whereby they try to find variables that explain the number of foreclosures but that aren't correlated with unobserved variables explaining house values in a given area. For instruments, they use FICO scores (90th percentile of the distribution), loan-to-value (LTV) ratios (90th percentile of the distribution), homeowner income (median), property size, and the stock of housing. They estimate the IV regression for one MSA (Los Angeles) and find that their results do not substantially change. Based on the results of the IV estimation, the authors conclude that omitted variables are not a problem. However, this particular exercise isn't completely convincing, because if the first critique above is a problem (not having a disaggregated measure of average house price appreciation), then the instruments will likely be correlated with the regression residuals. To see this, think about our Dorchester/Brookline example from above. Properties in Dorchester will have large negative residuals in the regression. In addition, since Dorchester is a lower-income area, the credit score distribution of its homeowners is likely lower than the average area in the Boston metro-area, the LTV distribution likely higher, and median income likely lower. In contrast, Brookline is probably the opposite in terms of the credit score, LTV, and income distributions of its homeowners. Thus, the regressions residuals will be correlated with the instruments, and the IV estimation will not solve the underlying econometric issues.

Paper is a nice starting point
Despite these econometric issues, the pattern of the findings seems to imply a contagion effect, even if the quantitative magnitude might not be measured accurately. As we discussed above, the authors go to great lengths to control for the distance from foreclosed properties as well as for the different phases of the foreclosure process, and estimate a very flexible specification for these variables. For example, they find very little effect from properties that are a year away from foreclosure and a much larger effect between the time of foreclosure sale/auction and the eventual REO sale. In addition, they find that the effects from foreclosures near the property (within 300 feet) are much stronger than the effects from foreclosures farther away (beyond 500 feet).

As a whole, we think this paper is an important contribution to the literature, as its econometric specification is much more robust and flexible than prior externality studies. There are still important econometric issues that future research must address in order to really pin down the quantitative magnitude of the effect of nearby foreclosures on the value of a non-distressed property, but this paper provides a nice starting point.

By Kris Gerardi, research economist and assistant policy adviser at the Atlanta Fed (with Boston Fed economists Christopher Foote and Paul Willen)

*In addition to effects on surrounding property values, foreclosures have been found to have negative impacts on other neighborhood characteristics such as vacancy rates and crime rates.

May 5, 2010 in Foreclosure contagion, House price indexes | Permalink


March 31, 2010

Anti-foreclosure policy and aggregate house price indexes

A new paper by researchers at the New York Fed and New York University argues that the Federal Housing Authority (FHA), the government's insurer of relatively high-risk loans, is seriously understating the amount of risk in its portfolio. The paper makes a number of different points, but we want to comment on one claim in particular that has policy relevance beyond the issue of FHA risk. In fact, if this claim is correct, then any policy designed to reduce foreclosures by eliminating negative home equity could face significant problems when put into effect.

Repeat sales indexes a poor predictor of individual home price
The specific issue we want to address is how well an aggregate house price index can predict the price of an individual home. A number of aggregate indexes measure average house prices for a particular area, from the national level to the ZIP-code level. Often, these indexes are based on repeat sales, meaning that they combine the price changes of individual homes over time. If a house sold for $200,000 in 1997 and $220,000 in 2001, this repeat sale provides a data point indicating that house prices rose by 10 percent from 1997 to 2001. It is true that the 2001 buyer might have gotten a great deal in that the house really should have sold for more than $220,000 at the second sale. However, the assumption is that the influence of good and bad deals washes out when data from many repeat sales are aggregated together. If they do, then researchers can infer the average, overall path of house prices.

The problem occurs, the authors of the paper say, when one uses the resulting aggregate index to predict the price of an individual home. Consider someone who purchased a home for $200,000 in 2007. Now assume that over the next two years the aggregate house price index for that particular area declined by 10 percent. The authors point out that the decline in the index does not necessarily mean that this particular homeowner would have sold the house for $180,000 in 2009. The owner may have taken extremely poor care of the house, or a beautiful park that was across the street from the house at the time of purchase may have become a strip mall. In either case, the homeowner was likely to have sold for less than $180,000. On the other hand, the homeowner may have made some improvements to the home that would have resulted in a sale of more than $180,000.

Research paper provides careful analysis of valuation errors in aggregate indexes
Potential problems with repeat-sales indexes were well known before the FHA paper was written. What the new paper contributes is a careful analysis of how large these so-called valuation errors can be and how they might relate to the probability of having negative equity. Using residential sales from Los Angeles County, the authors compare the actual sales prices of houses with predictions generated by different aggregate price indexes. The authors make two important findings.

First, the repeat-sales indexes are often biased, in the sense that the mean of the predictions does not match the mean of the recorded prices. For 2008 and 2009, repeat-sales indexes tended to overpredict house prices by 7 to 18 percent. In 2007, the indexes underpredicted house prices by about 4 percent. Second, dispersion in individual valuation errors is large—the standard deviation of valuation errors is about 20 to 25 percent, depending on the aggregate index used. Putting these two facts together gives a clear message: Using standard methods, it is difficult to predict what any individual house will sell for at any particular time.

Valuation errors undermine mortgage balance reduction policies
On a general level, this observation is not an indictment of the FHA, since a lot of other people also use aggregate indexes to infer prices of individual homes—including us. Moreover, without knowing the ins and outs of the FHA's default-prediction model, it is hard to know the quantitative importance of valuation errors in the calculation of FHA risk. But moving beyond this issue, it is not hard to see how large valuation errors could undermine the effectiveness of a policy that attempted to ease foreclosures by reducing mortgage balances for individual negative-equity homeowners. As we have blogged recently, some observers have claimed that many, if not most, foreclosures occur because owners with large amounts of negative equity simply walk away from their homes. The ostensible policy implication is to reduce these homeowners' mortgage balances to give them more of an incentive to stay.

If valuation errors are large, however, it is very difficult to know who has severe negative equity and who doesn't. This problem undermines the effectiveness of balance-reduction policies. Effective policymakers must know how to price individual homes to assess the depth of negative equity for those homes. Consider two homes that, according to an aggregate price index, have 30 percent negative equity. That amount may or may not be severe enough to get an owner thinking about walking away. If it is, then an appropriate policy might reduce both homes' mortgage balances by 20 percent, thereby reducing the negative equity to about 10 percent. (Leaving a little bit of negative equity is probably a good idea in practice because it may prevent homeowners from selling the moment that a balance reduction is made.)

Effective foreclosure prevention would consider both job loss and negative equity
If in reality one of these homes actually has 50 percent negative equity and the other has 10 percent negative equity, then the balance-reduction policy is likely to prevent no foreclosures. The owner with 50 percent negative equity remains underwater, to the tune of 30 percent, so is probably still thinking about walking away—according to the theory of default that motivated the balance-reduction policy in the first place. On the other hand, the owner with 10 percent negative equity was not going to walk, unless perhaps a job loss went along with the negative equity. But if the combination of job loss and negative equity is the real problem in the housing market rather than severe amounts of negative equity alone then we can devise much more cost-effective policies to reduce foreclosures than large-scale balance reductions.

The authors of the paper do not discuss balance reductions. However, in other papers, they argue that anti-foreclosure policy should consider balance reductions. We believe that the valuation-error results uncovered in the FHA paper indicate that balance-reduction policies face substantial hurdles in actual practice.

By Chris Foote, senior economist and policy adviser at the Boston Fed (with Atlanta Fed economist Kris Gerardi and Boston Fed economist Paul Willen)

March 31, 2010 in Federal Housing Authority, House price indexes, Loan modifications, Mortgage default, Negative equity | Permalink