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 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.

August 09, 2013

Recent Trends in Bank Construction Lending and Sentiment

One of the big questions for the economic recovery is the extent to which the improvement in the housing sector is sustainable. The statements from the Federal Open Market Committee (FOMC) over the past three years reveal an interesting evolution in the way the Committee views housing activity. Consider the subtle changes:

- 06/2009: Household spending has shown further signs of stabilizing but remains constrained by ongoing job losses, lower housing wealth, and tight credit.
- 09/2009: [A]ctivity in the housing sector has increased.
- 11/2009: Activity in the housing sector has increased over recent months.
- 12/2009: The housing sector has shown some signs of improvement over recent months.
- 03/2010: [I]nvestment in nonresidential structures is declining, housing starts have been flat at a depressed level.
- 04/2010: Housing starts have edged up but remain at a depressed level.
- 06/2010: Housing starts remain at a depressed level.
- 09/2010: Housing starts are at a depressed level.
- 11/2010: Housing starts continue to be depressed.
- 12/2010: The housing sector continues to be depressed.
- 08/2011: [T]he housing sector remains depressed.
- 04/2012: Despite some signs of improvement, the housing sector remains depressed.
- 08/2012: Despite some further signs of improvement, the housing sector remains depressed.
- 09/2012: The housing sector has shown some further signs of improvement, albeit from a depressed level.
- 12/2012: [A]nd the housing sector has shown further signs of improvement....
- 01/2013: [T]he housing sector has shown further improvement.
- 03/2013: [T]he housing sector has strengthened further....
- 07/2013: [T]he housing sector has been strengthening....

I'll leave the exact parsing of FOMC statements to private experts. What I want to address is the way banks are reacting, or perhaps contributing or being less of a barrier, to the strengthening housing sector.

In my July 10 posting, I discussed the correlation between bank construction lending and residential construction activity—larger changes in construction lending are associated with a higher level of construction put in place. Last week the Census Bureau reported that total construction spending fell. The good news is that residential construction, on a seasonally adjusted annual rate, was essentially flat from May to June and was up significantly compared to June 2012. So is bank behavior consistent with improving residential construction spending? Two sources help to shed light on what banks are thinking and doing: the Senior Loan Officer Opinion Survey (SLOOS) and bank call reports as of June 30, 2013.

The SLOOS asks how the respondent banks' credit standards for approving applications for commercial real estate loans (CRE) loans have changed over the past three months. Since 2011Q2, the net percentage of domestic banks tightening standards for CRE loans has been negative, which indicates loosening (see the chart). CRE, however, includes not only loans for construction and land development (C&D), but also loans secured by nonfarm, nonresidential properties and multifamily residential properties. The latter two loan types finance existing structures rather than construction activity, thus it is impossible to determine whether the loosening since 2011 applies to construction lending.


Fortunately, the most recent SLOOS had special questions that asked for the changes in standards and demand over the past 12 months for the three different types of CRE loans: C&D loans, loans secured by nonfarm nonresidential properties, and loans secured by multifamily residential properties.


Though on net the standards for all CRE loans type appear to be loosening, multifamily and nonresidential loans are likely to have been the drivers of the easing of overall CRE standards since 2011Q2. Unfortunately, because these were special questions, there is no time series to aid with putting the responses in context.

Do actions speak louder than words?
Ideally, to gauge bank lending activity, we would look at the volume of new construction loans—new loans enable new activity. What we observe, however, is total C&D loans outstanding—which net out loan payoffs, write-downs, and reclassifications—continue to be decreasing in aggregate. In contrast to the overall industry where the top 100 banks account for 80 percent of total bank assets, the top 100 C&D lenders, which are not necessarily the top 100 banks, account for only 60 percent of bank C&D lending. Given the available data and the fact that smaller banks are relatively more important in the C&D space, the percentage of banks that are increasing their construction lending may be a better indicator of changing sentiment (see table below).


With 48.1 percent of banks reporting year-over-year growth in C&D loans, activity cannot be classified as robust, but neither is it bouncing along the bottom. Not that we should aspire to the frothy levels of C&D lending that prevailed during the housing bubble, but compared to 2010, it is clear that more banks are reentering the C&D market, which bodes well for housing starts and construction spending going forward.

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

August 9, 2013 in Credit conditions, 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.

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.

June 19, 2013

Asset-Quality Misrepresentation as a Factor in the Financial Crisis

A provocative new paper by Tomasz Piskorski, Amit Seru, and James Witkin (2013, henceforth PSW) presents evidence in support of a popular theory of the financial crisis. That theory is that issuers of mortgage-backed securities (MBS) misrepresented the credit quality of the loans backing those bonds. Without the proper information about what they were buying, MBS investors lost so much money that the entire financial system nearly buckled as a result. A closer look at the evidence, however, shows that any deception in the marketing of MBS had little or no effect on investor decisions, so it is unlikely that MBS misrepresentation played a significant role in the financial crisis.

The basic approach of the PSW paper is to compare two sets of loan-level mortgage datasets in a search for issuer misrepresentation. One set of records is provided by the MBS issuers themselves (formally known as trustees), while the other is a set of loan-level records from a credit bureau. PSW assumes that if they find an MBS trustee reported one fact about a given loan while the credit bureau data shows something different, then the MBS trustee misrepresented the loan. For example, assume that a trustee reported that the borrower on a particular loan was an owner-occupier, not an absentee investor. PSW consulted the credit bureau data to see whether the bureau reported a borrower living at the same address as the property. If this was not the case, then PSW say the loan was misrepresented by the MBS trustee as a loan for an occupier when in reality it was a loan for an investor (and thus much riskier than advertised).

Loans backed by investor properties were likelier to default
The headline finding is that misrepresentation was present. According to PSW, about 7 percent of investor loans were misrepresented as backed by owner-occupied homes. PSW also find that loans backed by investor-owned properties were much more likely to default, and that MBS issuers systematically failed to report the existence of second liens.

The link between misrepresentation about MBS and the mortgage crisis may at first glance seem obvious. MBS investors constructed their forecasts of loan defaults using the loan characteristics reported by MBS trustees. When defaults turned out to be higher than the investors expected—because the MBS trustees had misrepresented the loan characteristics—massive investor losses and a financial crisis resulted.

Yet the historical record reveals something puzzling. Despite the ostensible misrepresentation by trustees, investor forecasts of MBS performance were exceptionally accurate. The table comes from a Lehman Brothers analyst report from 2005 and provides forecasts of the performance of securitized subprime loans under varying scenarios for house prices. The bottom row of this table gives the "meltdown" scenario: three years of house-price declines at an annual rate of 5 percent. Under this scenario, Lehman researchers expected subprime deals to lose about 17 percent. In reality, prices fell 10 percent per year—double the rate in the meltdown scenario—yet the actual losses on subprime deals from that vintage are expected to come in around 23 percent.

Distribution of Returns: Shiller Home Price Index

Source: Lehman Brothers, "HEL Bond Profile across HPA Scenarios," in U.S. ABS Weekly Outlook, August 15, 2005.

This table shows that investors knew that subprime investments would turn sour if housing prices fell. The "meltdown" scenario for housing prices implies cumulative losses of 17.1 percent on subprime-backed bonds. Such losses would be large enough to wipe out all but the highest-rated tranches of most subprime deals. The table also shows that investors placed small probabilities on these adverse price scenarios, a fact that explains why they were so willing to buy these bonds.

If issuers were lying about the quality of the loans, then why did the investors produce such reasonable forecasts of loan defaults?

There are two closely related answers. The first comes from information economics: rational investors were properly skeptical of any information they couldn't verify, so they rationally assumed that there was some misrepresentation going on. The logic here is exactly that of the celebrated "lemons" model of George Akerlof, who won a Nobel Prize for his insight about the effect of asymmetric information on markets. Assume, for example, that used-car buyers know that used-car sellers have private information about the quality of the cars that they bring to the market. As a result, potential buyers assume that they will be offered only low-quality used cars (lemons), so these buyers offer appropriately low prices for all cars in the used-car market. The same skepticism was likely to be present in the market for mortgages. In our case, pooling across mortgage loans and securitization deals means that investors will sometimes overestimate the share of misrepresented loans and sometimes underestimate it, but on average they get it right.

The second, related, answer to the question of why misrepresentation doesn't matter is that this lack of trust leads investors to base their forecasts on the historical performance of the loans. In other words, investors do not construct a theoretical model of how often an idealized owner-occupant should default. Rather, the investors simply measure the previous default probabilities of loans that were represented as going to owner-occupants. As long as misrepresentation doesn't significantly change over time, the forecasts based on the investors' reduced-form statistical models would not have underpredicted defaults. (Indeed, the table above shows that this method seems to have worked pretty well.) Of course, one might worry that the misrepresentation problem did get worse over time. However, although the authors claim in the abstract that the problem got worse, the results in the paper show that differences between the level of misreporting in 2005 and 2006 and 2007 were minimal. Moreover, in many cases, the differences have the wrong sign.

Looking elsewhere for an explanation
At the end of the day, the PWS paper—like many others written since the crisis—tries to explain a fact that isn't really a fact. Investors didn't really think of securitized subprime loans as less risky than they actually were. The documentary evidence repeatedly shows that investors understood the risk inherent in purchasing MBS that were backed by loans to people with bad credit histories. As the table shows, analysts expected 17.1 percent losses in the meltdown scenario. If we assume a 50-percent loss rate for each default, then overall losses of 17.1-percent imply that 34.2 percent of the loans—more than a third—would go to foreclosure. With the benefit of hindsight, we can see that the real problem for investors was not that they didn't think subprime borrowers would default if house prices fell. Rather, they didn't think house prices would fall in the first place.

Finally, it is important to stress that the quantitative effects of the level of misrepresentation found in the paper are economically insignificant. Remember that the harm of misrepresentation for an investor arises because misrepresentation leads investors to under-forecast defaults. Quantitatively, the largest finding of misrepresentation reported by PSW is that 15 percent of purchase mortgages had some form of misrepresentation, and that the misrepresented loans were 1.6 times more likely to default. So, if an investor assumes that none of the loans were misrepresented, then a simple calculation shows that actual defaults are likely to come in about 1.09 times higher than forecast:

(0.85 x 1) + (0.15 x 1.6) = 1.09

This calculation implies that if a naïve investor forecasted, say, 8 percent defaults for a pool of subprime loans, then the true number would actually be 8 percent x 1.09 = 8.72 percent. But even this figure overstates the effect of misrepresentation, because it assumes that the investors had access to a "pure" data set that was uncontaminated by misreporting. In any case, for subprime loans originated in 2006, actual defaults came in about 40 percentage points over what was expected. In other words, even if we assume investors were completely naïve, misrepresentation can, at most, explain 0.72 percentage points out of 40.

We feel researchers should look elsewhere for an explanation for why investors lost so much money during the housing crisis. A good place to start is the belief by all actors in the drama— borrowers, Wall Street intermediaries, and investors—that housing prices could only go up.

Photo of Paul WillenBy Paul Willen, senior economist and policy adviser at the Federal Reserve Bank of Boston, with help from

Photo of Chris Foote Chris Foote, senior economist and policy adviser at the Federal Reserve Bank of Boston, and

Photo of Kris Gerardi Kris Gerardi, financial economist and associate policy adviser at the Federal Reserve Bank of Atlanta

June 19, 2013 in Housing prices, Mortgage crisis, Subprime MBS | 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.

June 12, 2013

Is Investing in Housing Really a Losing Proposition?

In a recent article in the New York Times (here), Robert Shiller notes that a home may not be such a great investment after all. After adjustments are made for inflation, Shiller says that real home prices are more or less flat over the long term and that investors can make better returns by investing elsewhere. Bill McBride and Tom Lawler from Calculated Risk have chimed in on this debate several times over the past few years (here, here and here) by pointing out that there are several methodological issues with the way Shiller calculated home prices before 1986, and that using an alternative series results in a clear upward slope.

While we acknowledge that the gains over time are sensitive to the index you choose to use, we think it's also important to note that returns on investments in housing have not consistently increased regardless of which index you use. Even if you exclude the most recent bubble, there have been notable ups and downs, although none as severe. Shiller and Lawler's work conclude that long-run returns have averaged somewhere between 0.2 percent and 1.2 percent, depending on which series you use, but neither touches on the distribution of returns. This got us wondering—with average returns so close to zero, just how often has the housing market produced losers? And how does investing in housing compare to investing in equities, as Shiller seems to prefer?

As a first step toward answering these questions, we computed the average annual return of home prices across all possible combinations of start and stop points using the Shiller house price series from 1926 to 2012. The distribution depicts returns concentrated around zero with some skewness to the right. Eighty percent of all start-stop point observations experience some degree of positive return (see chart 1).

Distribution of Returns: Shiller Home Price Index

We acknowledge that this exercise alone is imperfect because it fails to take into account the duration of ownership. Based on analysis published (here) earlier this year by Paul Emrath at the National Association of Home Builders, we assume that the average homeowner lives in his or her home for 13.3 years. We applied this duration to our analysis and found that the volatility in the data series is significant enough to change the distribution of returns. The average annual returns for an asset held for a period of 13 or more years is substantially less volatile than for an asset held for fewer than 13 years, and those investing for the longer term were much more likely to have positive returns. Perhaps more important than the shape of each curve is that both are concentrated at or just above zero. We compute that 40 percent of homes owned for less than 13 years have negative average annual returns, compared to 12 percent of homes owned for 13 years or more (see chart 2). Interestingly, while a much greater portion of those owning for 13 or more years obtain positive returns, the average annual return was actually slightly higher for those owning fewer than 13 years (0.95 percent versus 1.03 percent).

Distribution of Returns: Shiller Home Price Index

Since it is pretty clear that the volatility in returns varies by length of ownership, we apply weights for average length of ownership using Emrath's Survival Table. Using the weights, we recomputed average annual returns across all possible combinations of start and stop points for average length of ownership. The distribution continues to show that returns are concentrated around zero with skewness to the right; two-thirds of all investors in this distribution experience some degree of positive return (see chart3).

Distribution of Returns: Shiller Home Price Index

After getting a better feel for average annual returns on homes purchased using Shiller's real home price index, we thought it would be interesting to run through this same exercise with the S&P 500 Index (which we used as a proxy for the stock market) to provide an apples-to-apples comparison of the average annual returns that one could expect from an alternative investment in stocks. The results depict a wider distribution, with longer, fatter tails and some skewness to the right. In other words, there is more volatility in terms of return, but with that volatility comes an opportunity for larger gains over time (see chart 4). In fact, the weighted average annual return of the S&P 500 is 4.55 percent, compared to 0.97 percent for the Shiller real home price index.

Distribution of Returns: S&P 500 Index

As a final exercise, we added a time dimension and charted the average annual return on assets for housing and the S&P 500 Index assuming that each asset is held for 13 years from its purchase (see chart 5).

Average Annual Return on Assets Held for 13 Years

It's important to note that the distributions of returns for housing in all these computations are not the distribution of returns for every possible house purchase. Likewise, the returns shown for the S&P 500 are not the entire universe of returns from buying and selling individual stocks. Instead, these returns are based on a pool of housing and a pool of stocks. Therefore, the chart speaks not to the distributions of returns to individual assets, but the group as a whole. Further, the returns to housing in the chart ignore the fact that homeowners might have additional gains from owning if their mortgage replaces rent. Indeed, according to some calculations, homeowners who buy a home today and hold it for seven years can expect to pay 44 percent less than people who choose to rent.

Depicting average annual returns in this format helps to demonstrate two points. First, Shiller's point that "real home prices rose only 0.2 percent a year, on average" was not far off the mark, as returns on investments in housing using our approach do appear to hover around zero for most of the time series. Second, Shiller's comment that "it's hard for homes to compete with the stock market in real appreciation" seems to be fair. If a home is purchased only as an investment and not as a place to live, this comparison of average annual returns clearly shows that investing in equities offers favorable returns more often than investing in housing.

Photo of Ellie TerryBy Ellyn Terry, an economic policy specialist, and

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

June 12, 2013 in Homeownership, Housing boom, Housing prices | Permalink


It is not clear in this report what, if any, costs were included for maintenance and repair by an owner.

Posted by: Michael D. Miller | June 13, 2013 at 11:18 AM

If the average duration is 13 years, most people who own a home for that long have probably remodeled or renovated. That would reduce the annual return on housing and make equities relatively more attractive.

Posted by: Art | June 17, 2013 at 12:02 PM

I was wondering if the analysis includes dividend returns and the rental incomes (for investment properties)

Posted by: Charles Ie | August 14, 2013 at 06:51 AM

Does the research include federal and state tax deductions on interest payments?

Posted by: Morris Hagerman | April 14, 2015 at 03:15 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.

November 17, 2011

Taking on the conventional wisdom about fixed rate mortgages

The long-term fixed rate mortgage (FRM) is a central part of the mortgage landscape in America. According to recent data, the FRM accounts for 81 percent of all outstanding mortgages and 85 percent of new originations.1 Why is it so common? The conventional wisdom is that the FRM is a great product created during the Great Depression to bring some stability to the housing market. Homeowners were defaulting in record numbers, the story goes, because their adjustable rate mortgages (ARMs) adjusted upward and caused payment shocks they could not absorb.

In a Senate Committee on Banking, Housing, and Urban Affairs hearing on October 20, some experts presented testimony that followed this conventional wisdom. As John Fenton, president and CEO, Affinity Federal Credit Union, who testified on behalf of the National Association of Federal Credit Unions, laid out in his written testimony:

Prior to the introduction of the 30-year FRM, U.S. homeowners were at the mercy of adjustable interest rates. After making payments on a loan at a fluctuating rate for a certain period, the borrower would be liable for the repayment of the remainder of the loan (balloon payment). Before the innovation of the 30-year FRM, borrowers could also be subject to the "call in" of the loan, meaning the lender could demand an immediate payment of the full remainder. The 30-year FRM was an innovative measure for the banking industry, with lasting significance that enabled mass home ownership through its predictability.

Of course, this picture of the 30-year FRM as bringing stability to the housing market has profound implications for recent history. Many critics attribute the problems in the mortgage market that started in 2007 to the proliferation of ARMs. According to the narrative, lenders, after 70 years of stability and success with FRMs, started experimenting with ARMs again in the 2000s, exposing borrowers to payment shocks that inevitably led to defaults and the housing crisis. Indeed, one of the other panelists at the hearing, Janis Bowdler, senior policy analyst for the National Council of La Raza, argued in her written testimony that "when the toxic mortgages began to reset and brokers and lenders could no longer maintain their refinance schemes, a recession ushered in record-high foreclosure rates."

I argue, on the other hand—both in my testimony at the hearing and in this post—that the narrative of the fixed rate mortgage as an inherently safe product invented during the Depression that would have mitigated the subprime crisis because it

eliminated payment shocks does not fit the facts.

Parsing the myths around the fixed rate mortgage
First, the FRM has been around far longer than most people realize. Most people attribute the FRM's introduction to the Federal Housing Administration (FHA) in the 1930s.2 But it was the building and loan societies (B&Ls), later known as savings and loans, that created them, and they created them a full hundred years earlier. Starting with the very first B&L—the Oxford Provident Building Society in Frankfort, Pennsylvania, in 1831—the FRM accounted for almost every mortgage B&Ls originated. By the time of the Depression, B&Ls were not a niche player in the U.S. housing market. They were, rather, the largest single source of funding for residential mortgages, and the FRM was central to their business model.

As Table 2 of my testimony shows, B&Ls made about 40 percent of new residential mortgage originations in 1929 and 95 percent of those loans were long-term, fixed-rate, fully amortized mortgages. Importantly, B&Ls suffered mightily during the Depression, so the facts simply do not support the idea that the widespread use of FRMs would have prevented the housing crisis of the 1930s.

Source: Grebler, Blank and Winnick (1956)
Note: Market percentage is dollar-weighted. Building and loan societies were the main source of funds for residential mortgages and almost exclusively used long-term, fixed-rate, fully amortizing instruments.

To be sure, at 15–20 years, the terms on the FRMs the FHA insured were somewhat longer than those of pre-Depression FRMs, which typically had 10–15 year maturities.3 The 30-year FRM did not emerge into widespread use until later. It must be stressed that none of the arguments that Fenton made hinge on the length of the contract. Furthermore, the argument that Bowdler made in her testimony—that by delaying amortization, a 30-year maturity lowers the monthly payment as compared to a loan with shorter maturity—applies as much to ARMs as it does to FRMs.

But even though the ARMs may not have caused the Depression, FRM supporters might ask, didn't the payment shocks from the exotic ARMs cause the most recent crisis? Again, the data say no. Table 1 of my Senate testimony shows that payment shocks actually played little role in the crisis.

Source: Lender Processing Services and author's calculations.
Note: Sample is all first-lien mortgages originated after 2005 on which lenders initiated foreclosure proceedings from 2007 to 2010.

Of the large sample of borrowers who lost their homes, only 12 percent had a payment amount at the time they defaulted that exceeded the amount of the first scheduled monthly payment on the loan. The reason there were so few is that almost 60 percent of the borrowers who lost their homes had, in fact, FRMs. But even the defaulters who did have ARMs typically had either the same or a lower payment amount due to policy-related cuts in short-term interest rates.

To be absolutely clear here, my discussion so far focuses entirely on the question of whether the design of the FRM is inherently safe and eliminates a major cause of foreclosures. The data say it does not, but that does not necessarily mean that the FRM does not have benefits. As I discussed in my testimony, all else being equal, ARMs do default more than FRMs, but since defaults occur even when the payments stay the same or fall, the higher rate is most likely connected to the type of borrower who chooses an ARM, not to the design of the mortgage itself.

The difficulty of measuring the systemic value of fixed rate mortgages
One common response to my claim that the payment shocks from ARMs did not cause the crisis is that ARMs caused the bubble and thus indirectly caused the foreclosure crisis. However, it is important to understand that this argument, which suggests that the FRM has some systemic benefit, is fundamentally different from the argument that the FRM is inherently safe. This difference is as significant as that between arguing that airbags reduce fatalities by preventing traumatic injuries and arguing that they somehow prevent car accidents.

Measuring the systemic contribution of the FRM is exceedingly difficult because the use of different mortgage products is endogenous. Theory predicts that home buyers in places where house price appreciation is high would try to get the biggest mortgage possible, conditional on their income, something that an ARM typically facilitates. When the yield-curve has a positive slope (in most cases) and short-term interest rates are lower than long-term interest rates, ARMs loans offer lower initial payments compared to FRMs. Thus, it is very difficult to disentangle the causal effect of the housing boom on mortgage choice from the effect of mortgage choice on the housing boom.

In addition, there is evidence from overseas that suggests that the FRM is not essential for price stability. As Anthony B. Sanders, professor of finance at the George Mason School of Management, points out in his written testimony, FRMs are rare outside the United States. A theory of the stabilizing properties of FRMs would have to explain why Canadian borrowers emerged more or less unscathed from the global property bubble of the 2000s, despite almost exclusively using ARMs.

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

1 First liens in LPS data for May 2011.

2 See the testimony of Susan Woodward for a discussion.

3 See the discussion in chapter XV of Leo Grebler, David M. Blank, and Louis Winnick (Princeton, NJ: Princeton University Press, 1956), 218–235; available on the website of the National Bureau of Economic Research.

November 17, 2011 in Homeownership, Housing prices, Mortgage crisis, Mortgage default, Subprime mortgages | Permalink


June 20, 2011

The housing wealth effect revisited

A lot of economic researchers have struggled to answer what seems like a simple question: how much does consumption rise when housing prices go up? Many economists believe that, by increasing the wealth of homeowners, rising house prices during the housing boom had a substantial positive effect on consumption (see this speech by former Fed Chairman Alan Greenspan). Along the same lines, the dramatic decline in home values during the Great Recession is thought by many to be a drag on consumption today.

Conventional wisdom among market participants, and to some extent academics, is that the marginal propensity to consume (MPC) out of housing wealth is somewhere between 3 and 5 cents per dollar for households in the United States (see this paper by Bostic, Gabriel, and Painter for a nice survey of the literature on housing wealth effects). In other words, on average, every $1 increase in housing wealth results in a 3–5 cent increase in consumption. However, many researchers are highly skeptical of these numbers for a number of reasons (see, for example, this 2009 critique). First, a basic life-cycle model that incorporates Friedman's permanent income hypothesis (PIH) would predict that modest changes in housing wealth (both anticipated and unanticipated) should result in changes in consumption that are smaller than the 3–5 cent estimate.

Second, most research that finds big MPCs out of housing wealth estimate these with aggregate consumption and wealth data (see, for example, this 2005 study by Case, Quigley, and Shiller). Aggregate regressions like these are highly susceptible to an omitted variables critique; it may be that both consumption and housing prices are driven by some other variable not accounted for in the regressions. One possibility is that as expected future income in a particular area goes up, the anticipated increase would cause residents in this area to consume more—because their future incomes are higher—at the same time house prices are rising—because more people would want to live in that particular area and enjoy these higher incomes. Ideally, one would use micro-level consumption and housing data to control for these confounding variables. Unfortunately, high-quality panel data on both housing values and household consumption have been hard to find.

An innovative new paper by Jie Gan of the Hong Kong University of Science and Technology recently published in the Review of Financial Studies may have partly solved these data issues. The paper uses property-transaction and credit-card data from Hong Kong to estimate the causal effect of changes in housing wealth on individual-level consumption behavior. For her mortgage and housing data, Gan uses a data set similar in quality and scope to those now being used in the United States (including some papers that we have written). Specifically, Gan obtained detailed mortgage data and some demographic information from a large Hong Kong bank for the period 1988–2004. She also obtained government data on the universe of all housing and mortgage transactions in Hong Kong from the early 1990s to the mid-2000s, which allowed her to construct district-level house price indices. For her consumption data, Gan used monthly credit card statements from the largest credit card issuers in Hong Kong, available for the early 2000s. While credit card spending is certainly not a perfect measure of consumption, Gan argues that credit

cards account for more than 20 percent of consumer spending in Hong Kong and are used to purchase a diverse, representative set of products.

 Merging these three data sets yields a panel of about 12,000 homeowners along with monthly information on credit expenditures and home values during the 2000–2 period. Gan then estimates a regression of quarterly consumption growth, measured with the individual-level credit-card data, on quarterly growth in lagged housing values, measured at the district level, and a set of individual fixed effects that control for time-invariant individual characteristics. The regression also includes a set of interaction variables between quarterly time dummies and occupational categories. These variables control for changes in income growth that may be correlated with both house price movements and consumption growth.

Gan finds that the elasticity of consumption growth to house price growth is about 0.17, which implies that a 10 percent increase in house price growth leads to a 1.7 percent increase in consumption growth. While this is a large effect, it translates into a slightly lower MPC out of housing wealth than the 3–5 cent effect common in previous studies. The reason is that the estimated sensitivity of consumption growth to house price growth that Gan estimates must be multiplied by the consumption-to-housing wealth ratio in Hong Kong in order to construct an estimated MPC out of housing wealth. The consumption-to-housing-wealth ratio is relatively small in Hong Kong (11.5 percent) because housing is extremely expensive. Consequently, the MPC out of housing wealth in Hong Kong is about 2 cents per dollar. While 2 cents seems low, especially given previous empirical evidence, Gan argues that this seemingly low MPC should not be construed as evidence that changes in housing wealth have only a small effect on economic activity. To the contrary, because housing is so expensive in Hong Kong, a small MPC implies substantial impacts of changes in house prices on the real economy.1

Given these large and statistically significant results, an obvious question is whether they truly correspond to causal effects, or whether omitted variables might still be causing problems. The answer is probably mixed. On one hand, Gan's data is a significant improvement over previous studies on multiple dimensions. The micro-level credit-card panel data allow her to use individual-level fixed effects, something not possible when using aggregate consumption data. As Gan points out, a fixed-effects specification controls for unobserved household heterogeneity that could lead to biased estimates. Other studies have used micro-level consumption data from the Panel Study of Income Dynamics (PSID), but that data set only measures expenditures on food and has been found to contain substantial measurement error (see Runkle 1991 for evidence of substantial measurement error in the PSID measure of food consumption). Furthermore, the district-level house price indexes used to calculate estimates of home values are more disaggregated than the state-level and even MSA-level house price indexes that previous studies have used. As a result, Gan's price indexes probably suffer less from measurement error than the price indexes in other work.

But even with these significant improvements in data quality, Gan may not have solved the endogeneity issue. As Gan writes in her paper, "[s]ince housing prices are available only at the aggregate level or at the level of the metropolitan statistical areas (MSAs), the observed consumption sensitivities may be driven by economic- or MSA-wide shocks that simultaneously affect housing prices and consumption." In other words, from an econometric standpoint, a regression of consumption growth at the individual level on house price growth at the district level is only identified from time-series changes in average consumption growth at the district level and house price growth at the district level.2 That means that other, unobserved district-level variables could potentially be driving the correlation between consumption growth and house price growth. As noted above, one potential omitted variable is expectation of higher future income at the district level.

Gan downplays this issue by noting that many Hong Kong residents tend to work and reside in different districts. As a result, district-level shocks are less likely to simultaneously influence both housing prices and consumption. But as an empirical matter, it is not clear from the paper how common it is to work and reside in separate districts. And from a theoretical perspective, it is not clear whether this argument effectively rules out the simultaneity issue. If a significant fraction of individuals who work in one district reside in the same outside district, then a shock to employment in a district could cause prices and consumption to co-move in the outside district. In addition, there are other types of district-level shocks that could create simultaneity regardless of the commuting patterns of its residents. For example, a public works project that improved existing infrastructure or that developed new park space (or created some other desirable public good) in a district would be expected to increase the attractiveness of the district—thus increasing housing demand, which would raise prices and possibly also average district-wide consumption, by changing the income/wealth composition of its population. To completely solve the simultaneity issue, one would need time-series variation in home values at the individual level. Unfortunately, that data is simply not available at this point—in Hong Kong or anywhere else.

Innovative tests of the housing wealth channel
Still, while Gan's data may not completely solve the simultaneity issue, some of her additional empirical tests go a long way toward alleviating these concerns. The first is a check on whether the consumption–housing wealth relationship is stronger for people who own multiple homes. A pure housing wealth effect would predict that the consumption behavior of individuals with higher levels of housing wealth would be more sensitive to changes in wealth. This is exactly what Gan finds in her data.

In a series of other tests, Gan tries to distinguish between the role of credit constraints and a precautionary savings motive in generating the positive housing-wealth effect. The credit constraint story is that increasing housing wealth relaxes borrowing constraints for individuals, which results in higher consumption. This effect is expected to be relevant only for households that are borrowing-constrained. The precautionary savings motive refers to the tendency for risk-averse individuals to accumulate wealth in order to self-insure against negative future income or wealth shocks. If individuals consider housing equity to be a component of precautionary savings, then an increase in housing wealth might increase consumption by reducing other components of precautionary savings. For example, if a household is saving a certain percentage of each paycheck for precautionary motives, then an increase in housing equity might be considered to be a viable substitute, and the household might be expected to decrease the percentage saved of each paycheck, and thus increase consumption.

Gan focuses on households that refinance as a first test in distinguishing between these two effects. She argues that borrowing-constrained households would need to refinance in order to access any equity increases, while precautionary savers would not need to refinance, since they could increase consumption by decreasing other forms of saving. Thus, if the relaxation of borrowing constraints is driving the positive elasticity of consumption growth to changes in housing wealth, then the elasticity estimate should be larger among households that refinance.3 Gan finds evidence of both effects: the estimated elasticity is significantly higher for households that refinance but is still positive and statistically significant for households that do not.

In a second test, she identifies households that are likely borrowing-constrained based on their use of credit card lines and separately estimates regressions for households that are close to their credit limit and those that are far from their limit. If the credit-constraint channel is present, the elasticity should be higher for the households that are close to their limit, while the precautionary savings channel, in contrast, predicts that the elasticity should be higher for less-constrained households and thus households that are far from their credit limit. The results of this exercise provide support for the precautionary savings motive, as less-leveraged households have a stronger consumption elasticity than more-leveraged households.

Gan performs a few more clever tests to try to distinguish between the credit-constraint and precautionary-savings channels. She finds strong evidence in favor of the precautionary savings channel and little evidence of an important role for credit constraints. It appears that in Hong Kong, households use housing wealth as an important component of an overall self-insurance strategy and view increases in housing wealth as a substitute for other types of savings and thus an opportunity to increase consumption. This is a very interesting and important finding on its own, but we also view it as strong evidence that Gan has truly identified a causal relationship between housing prices and consumption. The reason is that if simultaneity bias is truly responsible for the positive estimate of the consumption elasticity, then we wouldn't expect the estimate to be sensitive to different samples. The fact that it is, and in ways that are consistent with theory, suggests that Gan has really identified the impact of housing wealth on consumption behavior.

One final caveat is that Gan focuses solely on homeowners, leaving renters out of her analysis. We would expect renters to be hurt by increases in housing values. Thus, while Gan finds a significant positive effect of housing wealth on consumption for the population of homeowners in Hong Kong, we interpret her results as an upper bound of the effect of housing wealth on aggregate consumption in Hong Kong.

Photo of Kris GerardiKris Gerardi
Research economist and assistant policy adviser at the Federal Reserve Bank of Atlanta

1 I believe that what Gan means here, although it's not completely clear, is that since the housing stock in Hong Kong is so highly valued, a 1 percent change in house prices translates into a lot of additional consumption, even with a marginal propensity to consume out of housing wealth of 2 cents.

2 The reason is that any time the variation in a right-hand side variable is at a more aggregated level than the dependent variable, the coefficient estimate associated with that right-hand side variable is identified from the variation in both variables at the more aggregated level.

3 Another potential way that a borrowing-constrained household could access increases in housing equity is through home equity lines of credit (HELOC). HELOCs do not require households to refinance their mortgage. This possibility is not discussed in the paper, perhaps because HELOCs are not quite as popular in Hong Kong as they are in the United States.

June 20, 2011 in Housing boom, Housing prices | Permalink


April 18, 2011

What effect does negative equity have on mobility?

A debate has broken out in the housing literature over the effect of negative equity on geographic mobility. The key question is whether homeowners with negative equity—those who are "under water"—are more or less likely to move relative to homeowners with positive equity. In a paper published in the Journal of Urban Economics last year (available on the New York Fed website), Fernando Ferreira, Joseph Gyourko, and Joseph Tracy (hereafter FGT leaving out these categories) argue that underwater owners are far less mobile. Using data from 1985 to 2005, they find that negative equity reduces the two-year mobility rate of the average American household by approximately 50 percent. This is a very large effect and, if true, FGT's findings have important policy implications for both the housing market and the labor market today. For example, the economist and Nobel laureate Joseph Stiglitz, in testimony to the Joint Economic Committee of Congress on December 10, 2009, stated:

But the weak housing market will contribute to high unemployment and lower productivity in another way: a distinguishing feature of America's labor market is its high mobility. But if individuals' mortgages are underwater or if home equity is significantly eroded, they will be unable to reinvest in a new home.

The fear is that if people with negative equity can't move to new jobs, then the job-matching efficiency of the U.S. labor market will suffer, putting upward pressure on the unemployment rate. This type of "house lock" is exactly what the economy doesn't need as it emerges from the recent housing crisis and recession.

However, recent research by Sam Schulhofer-Wohl, an economist from the Minneapolis Fed, casts doubt on FGT's conclusions, as well as the economic intuition in Stiglitz's testimony. Schulhofer-Whol replicated the FGT analysis using the same data set (the American Housing Survey, or AHS) over the same sample period. But he found the exact opposite result: negative equity significantly increases geographic mobility.

What is the source of the discrepancy?
The difference in results stems from what at first blush seems like a small discrepancy in how the two papers identify household moves in the AHS. Here are the details: the AHS is conducted every two years by the U.S. Census Bureau as a panel survey of homes. That means that AHS interviewers go to the same homes every two years to record who lives there (among other pieces of information). For a home that is owner-occupied in one survey year, there are four possibilities regarding its status two years later. First, the home could still be owner-occupied by the same household as before. Second, the home could be owner-occupied by a different household. Third, it could be occupied by a different household that rents the home but doesn't own it. Finally, the home could be vacant.

In their paper, FGT treated the first category as a non-move and the second category as a move. FGT threw out of their analysis any observations that fell into the third and fourth categories.1 Dropping these last two categories, rather than coding them as moves, introduces significant bias into FGT's results. As Schulhofer-Wohl notes, it effectively assumes that households in negative equity positions are no more likely to rent out their homes, or leave them vacant when they move, than are households with positive equity. But it is relatively straightforward to show that this assumption is not borne out in the data. Specifically, Schulhofer-Wohl finds that positive-equity households who move sell their houses to new owner-occupiers two-thirds of the time. The other two possibilities (renting out the home or leaving it vacant) combine to occur only one-third of the time. In contrast, among negative-equity households who move, sales to new owner-occupant households occur half of the time, with the other two possibilities occurring the other half. Thus, by dropping the last two categories of transitions, FGT are artificially increasing the mobility rate of positive equity households relative to negative equity households.

Schulhofer-Wohl recodes the moving variable so that instances in which an owner-occupied home is rented or vacated also count as moves. He then re-estimates FGT's regressions. The coding change reverses the estimated relationship between negative equity and mobility. The new estimates show that negative equity raises the probability of moving by 10 to 18 percent, relative to the overall probability of moving in the AHS data. This of course is in marked contrast to FGT's results, where negative equity was found to significantly decrease the probability of moving.

What does theory tell us?
When thinking about what economic theory might say about the relationship between negative equity and mobility, it is important to distinguish how equity might affect selling versus how equity might affect moving. FGT write that their results suggest a role for what behavioral economists call "loss aversion." In this context, loss aversion can occur when owners are reluctant to turn paper losses into real ones by selling a home that has fallen in price. But, as Schulhofer-Wohl's analysis makes clear, it is possible and even common for households to move to different houses without selling their old ones. That means that loss aversion potentially affects the probability of selling a home without affecting the probability of moving.

Of course, while moving and selling are theoretically distinct, they often occur together in practice. One reason for the tight relationship between moving and selling involves liquidity constraints. Even short-distance moves entail nontrivial transaction costs, so households that do not have liquid wealth may not be able to move without selling their home. As a result, to the extent that negative equity decreases the probability of selling (via loss aversion), it may also decrease the probability of moving.

Besides loss aversion, there are at least two other channels through which liquidity constraints are relevant for the way that negative equity affects homeowner mobility. By definition, underwater households cannot retire their mortgages by selling their houses. Liquidity-constrained households that are also under water do not have the cash to make up the difference between the outstanding mortgage balance and sale price. As a result, negative equity could reduce selling (and, by extension, moving). On the other hand, liquidity-constrained households are more likely to simply default on their mortgages. Thus, negative equity might increase the probability of moving, though the moves that it facilitates are accompanied by foreclosures and not sales. Note that this "default channel" between negative equity and mobility depends importantly on expectations of future housing prices. Negative-equity households who do not think housing prices will rise any time soon are more likely to default on their mortgages, and thus move, than households who think that higher prices and restored housing equity are just around the corner.

The offsetting implications of liquidity constraints on mobility mean that theory doesn't provide a clean prediction for how negative equity should affect mobility. The question boils down to which implication is dominant in the data. The findings from the Schulhofer-Wohl paper suggest that the default channel may be relatively large, so concern about negative equity impeding homeowner mobility may be overblown.

Are these studies relevant to the current environment?
The sample period for both papers we have discussed ended in 2005. While we certainly believe that the issue addressed by both papers is very important, and that the Schulhofer-Wohl analysis corrects an important omission in the FGT study, we would offer a cautionary note to those who would extrapolate the findings of these studies to the current environment. The period 1985–2005 was a boom time in housing markets for most areas of the country. One way to see this is by noting the low number of negative equity observations in both the FGT and Schulhofer-Wohl papers. The majority of negative equity observations in the AHS data is likely from only a couple of areas in the country and from a narrow time period (most likely from the East and West coasts in the late 1980s and early 1990s). These places and time periods may be unrepresentative of the average negative-equity owner today.

Even more importantly, there were very few foreclosures from 1985 to 2005 relative to the past several years. This paucity of foreclosures was probably due not only to the low number of negative-equity households, but also to the low probability of foreclosure conditional on having negative equity. Recall that if housing prices are generally rising, households with negative equity will try hard to hang on to their homes and reap the benefits of future price appreciation, even if they are liquidity-constrained. It's probably safe to say that price expectations are lower today than they were in 1985–2005. Because low price expectations increase defaults, and because defaults and foreclosures increase the mobility of negative-equity owners through the default channel, it might be the case that the current effect of negative-equity on mobility is not only positive, but also even larger than the positive estimates in Schulhofer-Wohl's paper.

Photo of Kris GerardiKris Gerardi
Research economist and assistant policy adviser at the Federal Reserve Bank of Atlanta

1 This coding choice is not divulged in the FGT paper. The authors confirmed in private correspondence that it was a conscious decision to omit these categories and not a coding error, and that they are currently working on a revision of their original work that will address this issue.

April 18, 2011 in Housing prices, Mortgage default, Negative equity | Permalink


February 14, 2011

New study claims to solve the econometric problem of the link between foreclosure and house prices

Many policymakers are now concerned about how the next wave of foreclosures will affect the housing market. Analysts have cited a large "shadow inventory" of homes, referring to the mass of delinquent mortgages that have yet to make their way through the foreclosure process. When these foreclosures occur, they could raise the number of homes for sale and put downward pressure on house prices. They could also impose negative externalities to other homes in the same neighborhoods, sending house prices even lower. (We recently blogged about the so-called contagion effects of foreclosures on surrounding properties.)

These potential effects seem intuitive, but measuring them is not easy. The main problem is what economists call "simultaneity." Foreclosures lead to an increased supply of homes for sale, which can lower prices—but lower prices also increase the probability that borrowers have negative equity, which can lead to foreclosure. Thus, there is simultaneous causality: foreclosures can reduce prices, and lower prices can cause the negative equity that leads to foreclosure. As a result, simply showing a correlation between foreclosures and falling house prices is not sufficient to measure—or even establish—a causal effect of foreclosures on prices.

A new study by Atif Mian, Amir Sufi, and Francesco Trebbi claims to have solved this econometric problem. Their paper reports a substantial causal impact of foreclosures on not only house prices, but also residential investment and automobile purchases. However, the authors make a major data error that, in our opinion, invalidates a large part of their analysis. In addition, there are important conceptual issues that raise deep questions about their identification strategy, even if it is possible to correct the data error.

Can simultaneity be solved by classifying states as judicial, nonjudicial?
The authors attack the simultaneity problem with a classic method: they use differences in state laws as an instrumental variable. The essential idea is that states vary randomly as to whether they are judicial or nonjudicial. Judicial states are typically characterized by longer foreclosure durations, since the mortgage servicer must navigate through the legal system to get court approval, which usually entails a significant amount of time (see Pennington-Cross 2010 for a nice discussion). If the judicial/nonjudicial classification is random with respect to the health of state-level housing markets, then state laws will generate random variation in the number of foreclosures across states. Under these assumptions, using the classification as an instrument yields consistent estimates of the effect of foreclosures on house prices.

Of course, the classification of states into judicial and nonjudicial groups may not be random. It turns out that there is a strong regional component to this classification. Figure 3 in the Mian-Sufi-Trebbi paper shows that states in the Northeast and Midwest tend to be judicial, while the states in the South and West are mostly nonjudicial. It's no secret that problems in the U.S. housing market also have a strong regional character, with housing markets in Arizona, California, Florida, and Nevada (all located in the South and West) in particularly bad shape.

One way to check for the possibility of confounding effects across the two classifications of states is to compare their observable variables. The authors do this, and then claim that "states with a judicial foreclosure requirement are remarkably similar to other states in all attributes of interest except the propensity to foreclose" (p.3). But eyeballing their Figure 3 should give a reader pause. Nevada and Arizona, which are nonjudicial states, include the number one and two MSAs for new construction and for house price appreciation in the two years prior to the collapse of the mortgage market.1

Cross-state differences challenge regressions
Regional patterns in both state laws and housing markets cause problems for the authors' identification strategy. If we find that foreclosures tend to be more frequent in the nonjudicial states, this might be because foreclosing on delinquent homeowners is easier in those states, as the authors' identification strategy assumes. But high foreclosure rates in the nonjudicial states could also stem from negative shocks to housing demand in the parts of the country where the nonjudicial states happen to be located. Consequently, if we find that housing prices are lower and foreclosure rates are higher in nonjudicial states, then we can't be sure what's causing what. The high foreclosure rates could be causing the falling prices, as the authors' claim. But it could also be true that low regional demand and falling prices in the South and West are causing the high foreclosure rate—the very possibility that the authors were hoping to rule out.

The authors recognize that unobserved cross-state differences make the state-level experimental approach problematic so they propose an alternative set of regressions that are not subject to such criticism. In addition to estimating the first set of regressions—which, in the manner described above, uses all the states in the country—they estimate a second set that includes only ZIP codes adjacent to borders between judicial and nonjudicial states. The idea is that while unobserved heterogeneity across states could potentially invalidate the first set of regressions, this heterogeneity is less likely to be a problem in the second. In other words, the housing market in Arizona may differ markedly from the housing market in Maine and not just because Arizona is a nonjudicial state while Maine is judicial.

However, the ZIP codes just north of the Massachusetts-Rhode Island border are likely to have similar housing markets to the ZIP codes that are just south of this border. So, if the border ZIP codes in Massachusetts, which the authors label a judicial state, are experiencing higher foreclosures than the border ZIP codes in Rhode Island, a nonjudicial state, then differences in the two state's laws—and not unobserved differences in demand— are probably the reason why. And if the state laws are generating random variation in foreclosures, then the authors claim that this variation can be used to get a clean estimate of the causal effect of foreclosures on housing prices.

Problems in the data: Massachusetts, Wisconsin are misclassified
The authors find similar results in both sets of regressions. This similarity gives them some confidence that they have truly pinned down the direct effect of foreclosures on other economic outcomes. But here's where the data error comes in: the authors make a mistake in classifying at least two states as judicial or nonjudicial, which has major implications for their results. Specifically, they misclassify Massachusetts as judicial and Wisconsin as nonjudicial.2 Most sources, including the National Consumer Law Center (NCLC), reverse those classifications.

(For readers interested in the gory details, we show that for Massachusetts, there is no question that the NCLC is right.)

While the misclassification of two out of 50 states may seem minor, it turns out that Wisconsin and Massachusetts dominate the samples for the "border discontinuity" regressions. As the table shows, depending on the sample, using the alternative classification from the NCLC invalidates between 58 and 78 percent of the ZIP codes the authors use. Consider the sample that uses ZIP codes in 5-mile bands around state borders. Because it uses homes closest to state borders, this sample is least susceptible to unobservable differences between geographic areas, although we argue below that even 5-mile bands are inadequate to obtain clean identification. In this sample, classifying Massachusetts—correctly—as nonjudicial eliminates 70 percent of the comparisons.3

One response to this criticism would be to reclassify the states correctly and then reestimate both sets of regressions. The problem for the border regressions is that Massachusetts's and Wisconsin's borders with judicial and nonjudicial states respectively are sparsely populated and do not meet the authors' criteria for inclusion in the border sample. For example, farms and weekend homes comprise most of the properties in border ZIP codes between western Massachusetts and southern Vermont.

Misclassification proves detrimental to the identification strategy
As the authors have written the paper, they claim to find big differences in ZIP-code-level outcomes based on the judicial/nonjudicial classification. However, they use regressions with the wrong classification for most of the comparisons. If the identification strategy worked as the authors had hoped, their regressions would have implied that there are no important differences on either side of most judicial/nonjudicial borders because these borders in fact separated states with similar laws. However, because the regressions instead reported significant differences, some other important sources of heterogeneity across the state lines must exist—and if the authors can't control for heterogeneity across, say, the Massachusetts–Rhode Island border, the reader can't be expected to have confidence in their ability to control for unobserved differences between Massachusetts and Nevada.

Another way of putting this is that the authors have inadvertently performed and failed a falsification, or placebo, test on their data. They estimated their regressions on a sample of borders that are, for the most part, not characterized by differences in foreclosure laws, at least in terms of the judicial/nonjudicial classification, and found large effects where they should have found none. In our opinion, this is very strong evidence against their claim that judicial/nonjudicial foreclosure laws are a valid instrument for foreclosure rates. Even if the authors correctly reclassify the states and reestimate the IV regressions for the border sample, this failed falsification test still sheds doubt on the entire empirical strategy.

In addition to this primary critique, we also found some other important drawbacks in the analysis. For readers that are interested in learning more about these issues, here is a detailed discussion.

We remain unconvinced by the authors' claim that exogenous increases in foreclosures substantially reduce housing prices. This issue, of the link between foreclosure and house prices, is of first-order importance to policymakers, who struggle not only with the foreclosure problem itself but also with the potential effects of foreclosures on the economic recovery. However, the authors' research strategy is unlikely to be helpful in addressing these problems given the deep conceptual issues it did not deal with and the poor data on which it is based.

Photo of Kris GerardiKris Gerardi
Research economist and assistant policy adviser at the Federal Reserve Bank of Atlanta


Photo of Paul WillenPaul Willen
Research economist and policy adviser at the Boston Fed

1 Moreover, one of the main stylized demographic facts about the United States in the last 50 years has been the spread of population south and west across the country. Indeed, for the past 25 years, population has consistently and steadily grown twice as fast in the states the authors identify as nonjudicial compared to the states they identify as judicial.

2 Arguably, the authors misclassify as many as six states: the two listed plus Maryland, Nebraska, New Mexico, and Iowa. However, as we explain below, it's the misclassification of Massachusetts and Wisconsin that dramatically affects their results.

3 The authors are aware that there are alternative classifications but view the discrepancies as minimal, relegating the following comment to a footnote: "The only states that differ across these three classifications are Massachusetts, Nebraska, Oklahoma, Rhode Island, and Wisconsin." It is unclear whether they were aware that two of those states accounted for most of their border sample and that their border sample specification was not robust to the alternatives.

February 14, 2011 in Foreclosure contagion, Foreclosure laws, Housing prices, Mortgage crisis | Permalink


November 15, 2010

Mortgage relief in the Great Depression

Bemoaning the unwillingness of lenders to renegotiate loans in the current mortgage crisis, critics often point to the "old days" when, they argue, foreclosures were a rarity because of a different institutional setup. John Geanakoplos and Susan Koniak took this view in an op-ed they wrote for The New York Times.

In the old days, a mortgage loan involved only two parties, a borrower and a bank. If the borrower ran into difficulty, it was in the bank's interest to ease the homeowner's burden and adjust the terms of the loan. When housing prices fell drastically, bankers renegotiated, helping to stabilize the market.

Luigi Zingales uses almost the same language to make the same argument in an article in The Economist's Voice.

In the old days, when the mortgage was granted by your local bank, there was a simple solution to this tremendous inefficiency. The bank forgave part of your mortgage….

But what evidence do we have to back up these claims? The authors do not provide any direct evidence, do not provide a source for the evidence, and are not clear about what exactly they mean by "the old days." Until recently, there was little hard evidence on the subject. However, the simple existence of foreclosure crises in the past—in New England in the 1990s, for example, and across the nation during the Depression—is, at least on the surface, evidence that large numbers of mortgages escaped the seemingly win-win solution of modification even in the past.

In the last two years, two researchers, Andra Ghent of Baruch College and Jonathan D. Rose, an economist with the Federal Reserve System's Board of Governors, have gone to the records from the Depression, in one possible definition of "the old days," to see if, indeed, lenders renegotiated on a wide scale. Looking at the Depression gives us a good opportunity to test the theory that our current set of institutions are the problem because the institutional setup was different during the Depression—and if there was ever a profitable opportunity to modify loans, it was the period from 1932 to 1939. The extent of the crisis at the time minimized the information problems that we argue prevent profitable modifications now. Even if some borrowers who received modifications then could have afforded to repay their loans or more than their modified loans, there were so many deeply delinquent borrowers that the gains on the rest should have made up for it.

Comparing the 1930s Home Owners Loan Corporation to today's programs
In this post, we focus on the paper by Rose, who explores the Home Owners Loan Corporation (HOLC), a federal program aimed at transitioning troubled borrowers into new loans. (We will focus on Ghent's paper in the next post.) Many of today's critics have held up HOLC as an example of enlightened government policy; it was the model of the Hope for Homeowners (H4H) program enacted by Congress in 2008. Rose argues in his paper that, contrary to popular belief, HOLC actually did not offer particularly good terms to borrowers and instead focused mostly on assisting banks.

The last time that national housing prices crashed as low as they have over the past three to four years was during the Depression. As with today's crash, the 1930s fall coincided with a rash of mortgage defaults and foreclosures. According to Wheelock (2008), by 1933, 13.3 of every 1,000 mortgages in the United States was in foreclosure, and by the beginning of 1934, almost half of all outstanding urban home mortgages were delinquent. To stem the rising tide of foreclosures, the Roosevelt Administration created HOLC in 1933. Over the following three years, HOLC purchased and refinanced more than 1 million delinquent home loans. Although 1 million loans may not seem like that many, keep in mind that the U.S. mortgage market was significantly smaller 80 years ago and that current government programs have permanentlymodified or refinanced far fewer than 1 million loans during this crisis.

HOLC was a voluntary program that aimed to prevent foreclosures by refinancing troubled borrowers into mortgages that were more affordable. The program accepted applications from borrowers from June 1933 to November 1934, and then again from May to June 1935. Rather than paying cash for the mortgages, HOLC exchanged its own bonds for the lender's claim on the underlying house. The tax-exempt bonds were essentially equivalent to U.S. Treasury securities and thus could be considered very low-risk assets, especially relative to mortgage debt at the time. After HOLC purchased the loan from the lender, it would issue a new 15-year, fully amortizing mortgage at an interest rate of between 4.5 and 5 percent. The HOLC loans contained no prepayment penalties and had an interest-only option for the first three years. Thus, in most cases, a HOLC refinance gave the borrower a more affordable mortgage by lowering the interest rate and stretching out payments. Like modification programs today, HOLC tried to help borrowers in financial duress, discouraging applications from borrowers who just wanted a lower interest rate or borrowers for whom a refinance from a private lender was a viable alternative.

Did HOLC inflate home appraisals to encourage lenders to modify loans?
Until Rose did the research for his paper, a lack of data—other than a handful of aggregate statistics—prevented us from knowing much about HOLC activities. However, Rose was able to obtain loan-level data for a sample of HOLC loans from New York, New Jersey, and Connecticut.1 His goal was to analyze how HOLC encouraged lenders to part with their loans—after all, although HOLC bonds were much less risky than the mortgage debt that lenders held on their portfolios at the time, the bonds also carried lower interest rates. Thus, some 1930s lenders could have decided to take their chances with their old mortgages and refuse participation in the government's program. One key factor affecting the lender's decision was the amount of mortgage debt that HOLC was willing to refinance, which because of a combination of law and HOLC policy, was only 80 percent of the value of the property as estimated by a HOLC appraisal. If the amount of the new HOLC mortgage was lower than the old mortgage, then a participating lender would receive a "haircut" on the loan and the borrower would receive a principal reduction in addition to a lower interest rate and longer maturity schedule.2

Rose's main finding is that HOLC seems to have recognized that placing a low value on the house would make it more likely that the lender would have to offer a principal reduction, so that a low appraisal would reduce the chance that the lender would participate in the program. As a result, Rose argues, HOLC tended to place high values on properties in its appraisal process. This practice was good for lenders, who, in many cases, were paid in full for their mortgages. But high appraisals were bad for borrowers, because they made principal reductions less likely.3

A strength of the Rose paper is a careful explanation of how HOLC appraisals came to be relatively high. The HOLC appraisal formula consisted of three components. The first was the estimated present market value of the property, as in today's appraisals. The second was the  estimated cost of purchasing the lot and constructing a similar structure. The third component was capitalizing the estimated monthly rental value of the property over the past ten years over a ten-year period assuming no discount rate. HOLC averaged these three measures to determine the final appraised value.

Because of the dramatic decline in housing values at the beginning of the Depression, the second and third measures were typically higher than the first one, the market-based measure, which resulted in appraisals being higher than market values on average. According to Rose's data, which consists of loan applications that HOLC accepted and mortgages that they refinanced, the appraisals exceeded the market-value estimates almost 74 percent of the time and equaled the market-value estimates approximately 8 percent of the time. The value that came out of this process was not necessarily the actual value the organization used, however. HOLC performed two additional reviews (at the district and then the state level) on each application to guard against any obvious errors. These two reviews were highly subjective. HOLC's policy was that these reviews could lower the final appraisal without bound but could raise the appraisal only by 10 percent. According to Rose's analysis, the final appraisal exceeded the market value estimate in 58.5 percent and equaled it in 10.6 percent of the cases, showing that the review process was proactive in adjusting the values that came out of the three-component appraisal formula. Even more compelling is the fact that almost one-third of the HOLC refinances had amounts that exceeded 80 percent of the estimated market value of the property, while HOLC regulations meant that none had amounts that exceeded 80 percent of the final appraisal.

Inflated appraisals helped keep Depression-era banks solvent, at the expense of homeowners
We view these findings as convincing evidence that HOLC was inflating appraisals in order to increase lender participation rather than directly reducing principal or trying to make lenders take write-downs. Rose takes this reasoning a step further and concludes that the inflated appraisals were motivated by the desire to keep banks and other lending institutions solvent, at the expense of mortgage borrowers. While he cannot offer a straightforward way to confirm this interpretation, Rose does offer some tantalizing contemporaneous quotations to support it. For example, he includes this quote from one of the HOLC loan examiners:

There seems to be a deliberate effort made by the Connecticut officials to make high appraisals with the purpose of holding up real estate values. We have had this suspicion confirmed in a recent interview with the State Counsel, Mr. Tierney. This gentleman, during a call in our office last month, stated that they believed it necessary, to prevent depreciation of realty value as much as possible so as to maintain the soundness of the banks and other financial institutions which had made mortgage loans during the past 5 years, to make high appraisals. His opinion was that many of these financial institutions would be today in an unsound condition if their mortgage loans were appraised on a basis of today's realty values. This statement is illuminating when appraisals by our Connecticut offices are being analyzed. (p. 19)

One potential problem with Rose's interpretation is that it assumes HOLC didn't negotiate to the fullest possible extent with lenders. That may be true, but it's also possible that lenders were unwilling to substantially write down loans, which would have forced HOLC to maximize lender participation by paying high prices.

What does the HOLC experience teach us about the current foreclosure situation?
Lenders today still seem reluctant to modify large numbers of troubled loans. In the Depression, HOLC solved the problem of lender reluctance with high appraisals and by essentially transferring a large amount of mortgage credit risk from the private sector to the public sector. By contrast, in today's Home Affordable Mortgage Modification (HAMP) program, government payments encourage a modification only when the modification is determined to be a win-win proposition for both the borrower and the lender. The small number of modifications to date may suggest that the number of win-win modifications is low. In other words, just as in the Depression, today's lenders may be willing to take their chances with existing mortgages rather than offer generous concessionary modifications to borrowers.

We find the HOLC policy of refusing to directly reduce mortgage principal to be potentially informative to the current modification debate in another way. Principal reductions appear to have been as rare in the 1930s as they are today (more on this in our post about the Ghent paper). Many have blamed securitization by private institutions for this pattern today, but if securitization were the real culprit, how do we explain a similar lack of principal reduction in a period when securitization was basically nonexistent?

By Chris Foote, senior economist and policy adviser at the Federal Reserve Bank of Boston, and Kris Gerardi, research economist and assistant policy adviser at the Federal Reserve Bank of Atlanta

1 While these three states are not necessarily representative of the entire country, they were among the worst hit by the foreclosure crisis of the 1930s.

2 HOLC could purchase the loan at full value from the lender and extend the borrower a principal reduction, but Rose shows that HOLC never did this.

3 Of course, even borrowers who did not receive principal reductions were helped because they could swap their short-term balloon mortgages with longer-term HOLC loans. A longer amortization period tends to lower monthly payments, which make homeownership more affordable. Borrowers who could not roll over balloon mortgages when they came due no doubt found HOLC mortgages particularly helpful in preventing foreclosure.

November 15, 2010 in Housing prices, Loan modifications, Mortgage default | Permalink