You are on page 1of 32

No.

10-5

Reasonable People Did Disagree: Optimism and Pessimism


About the U.S. Housing Market Before the Crash

Kristopher S. Gerardi, Christopher L. Foote, and Paul S. Willen

Abstract:
Understanding the evolution of real-time beliefs about house price appreciation is central to
understanding the U.S. housing crisis. At the peak of the recent housing cycle, both borrowers and
lenders appealed to optimistic house price forecasts to justify undertaking increasingly risky loans. Many
observers have argued that these rosy forecasts ignored basic theoretical and empirical evidence that
pointed to a massive overvaluation of housing and thus to an inevitable and severe price decline. We
revisit the boom years and show that the economics profession provided little such countervailing
evidence at the time. Many economists, skeptical that a bubble existed, attempted to justify the historic
run-up in housing prices based on housing fundamentals. Other economists were more uncertain,
pointing to some evidence of bubble-like behavior in certain regional housing markets. Even these more
skeptical economists, however, refused to take a conclusive position on whether a bubble existed. The
small number of economists who argued forcefully for a bubble often did so years before the housing
market peak, and thus lost a fair amount of credibility, or they make arguments fundamentally at odds
with the data even ex post. For example, some economists suggested that cities where new construction
was limited by zoning regulations or geography were particularly “bubble-prone,” yet the data shows
that the cities with the biggest gyrations in house prices were often those at the epicenter of the new
construction boom. We conclude by arguing that economic theory provides little guidance as to what
should be the “correct” level of asset prices —including housing prices. Thus, while optimistic forecasts
held by many market participants in 2005 turned out to be inaccurate, they were not ex ante
unreasonable.

JEL Classifications: G12, G14, G21, B23

Kristopher S. Gerardi is a research economist and assistant policy advisor in the research department at the Federal
Reserve Bank of Atlanta. His e-mail address is kristopher.gerardi@atl.frb.org. Christopher L. Foote is a senior
economist and policy advisor at the Federal Reserve Bank of Boston. His e-mail address is chris.foote@bos.frb.org.
Paul S. Willen is a senior economist and policy advisor at the Federal Reserve Bank of Boston. His e-mail address is
paul.willen@bos.frb.org.

This paper, which may be revised, is available on the web site of the Federal Reserve Bank of Boston at
http://www.bos.frb.org/economic/wp/index.htm.

We thank Susan Wachter, Jeff Fuhrer, Elizabeth Murry, and two anonymous reviewers for helpful comments.

The views expressed in this paper are those of the authors and do not necessarily represent those of the Federal
Reserve Bank of Atlanta, the Federal Reserve Bank of Boston, or the Federal Reserve System.

This version: August 12, 2010


I. Introduction
The home-price bubble feels like the stock-market mania in the fall of 1999, just
before the stock bubble burst in early 2000, with all the hype, herd investing
and absolute confidence in the inevitability of continuing price appreciation.

Robert Shiller
Quoted in “The Bubble’s New Home”
by Jonathan R. Laing, Barron’s (2005)

Optimism about future house price growth shoulders much of the blame for the mortgage
crisis that swamped the U.S. and world economies starting in 2007. Had borrowers and
investors in 2006 known that the nearly decade-long expansion of the U.S. housing sector
had ended and that house prices would decline 30 percent over the next three years, few
would have made the decisions they did.
But to many, the fact that investors and borrowers were optimistic about house prices in
2006 seems inexplicable or even inexcusable. The experience up to that point showed all the
hallmarks of an asset-pricing bubble. From 1997:Q1 to 2005:Q1, real home prices went up 72
percent according to the Case–Shiller repeat-sales index, and by 41 percent according to the
OFHEO (now FHFA) repeat-sales index. To put these numbers in perspective, according
to Shiller (2006), real house prices in the United States were basically flat between the late
1940s and the mid-1990s.1 Even more striking is the fact that average real house prices
in the mid-1990s were at essentially the same level as they were in the late 1890s.2 The
divergence of house price appreciation from long-run trends appears not only when one looks
at the level of house prices but also if one considers the ratio of house prices to income or
rents.
The view that optimism was unjustified in 2005 is not just hindsight. Some observers
argued at the time that housing was overvalued and likely to crash. Among the pessimists
were economists, like Robert Shiller, quoted above. Some journalists, most notably at The
Economist, also argued for the inevitable collapse of the U.S. residential real estate market.3
1
OFHEO is an acronym for the Office of Federal Housing Enterprise Oversight, which regulated the
large government-sponsored housing enterprises (Fannie Mae and Freddie Mac) before they were placed
into conservatorship in September 2008. OFHEO’s regulatory role has been taken over by the Federal
Housing Financing Agency (FHFA), which has also continued to produce OFHEO’s repeat-sales house
price index.
2
According to Shiller (2006), there was a large decline in real prices after World War I that was com-
pletely offset by a large increase in prices from 1942 to 1947.
3
In addition, John Cassidy, a journalist at The New Yorker, wrote “The Next Crash,” a prescient
discussion of housing-market valuations relatively early in the boom (Cassidy 2002). Many blogs also
followed housing-market developments in real time. “Calculated Risk” was one of the notable blogs that
took a pessimistic view of the market.

1
The main question that motivates this paper is why market participants were so optimistic—
in other words, why they largely ignored the pessimists. Our answer is somewhat surprising:
the pessimistic case was a distinctly minority view, especially among professional economists.
We review the academic literature written during the housing boom, focusing closely on the
period 2004–2006, which turned out to be the peak of the boom. This review indicates that
there were widely dispersed opinions on whether the market accurately valued housing. As
noted above, some economists were truly pessimistic and warned of an impending collapse
in real estate markets with severe consequences. But other economists were truly optimistic
and dismissed such warnings. Granted, some of these optimists were real estate industry
boosters such as David Lereah, an economist with the National Association of Realtors, who
in 2006 published Why the Real Estate Boom Will Not Bust—And How You Can Profit from
It. But many in the optimistic camp were serious researchers with established reputations
who made convincing arguments in respected academic forums. These optimists cannot
be dismissed as purveyors of self-serving industry propaganda, nor as scholars who would
dismiss counterarguments without giving such opposing viewpoints due consideration.
The most prevalent opinion among economists who studied house pricing during the
bubble was essentially no judgment at all—we call it an “agnostic” view. Agnostics were
unwilling to make anything more than guarded, highly qualified statements on the future
path of U.S. housing prices. The discussion by Gallin (2008) is typical of this genre:

Because a low rent-price ratio has been a harbinger of sluggish price growth since
1970, it seems reasonable to treat the rent-price ratio as a measure of valuation
in the housing market. Indeed, one might be tempted to cite the currently low
level of the rent-price ratio as a sign that we are in a house price “bubble.”
However, several important caveats argue against such a strong conclusion and
in favor of future research. (p. 19)

In a sense, this reluctance to commit should not surprise anyone familiar with modern asset-
pricing theory. The “Fundamental Theorem of Asset Pricing” implies that the evolution
of asset prices is, to a first approximation, unpredictable. If housing was so obviously
overvalued, as the pessimists suggested, then investors stood to make huge profits by betting
against housing. By doing so, investors would have ensured that house prices would have
fallen immediately. Regardless of whether the theory of the unpredictability of asset prices
is correct, the theory is part of the basic training of almost every economist. Consequently,
any economist who suggests to his or her peers that an asset is over- or under-valued faces
a heavy burden of proof.4 This hurdle may explain why the arguments of some of the
4
We argue that the fundamental theory of asset pricing is one reason why the agnostic economists were
reluctant to predict falling housing prices, but it may not have been the only reason. Economists at policy
institutions may have shied away from making pessimistic predictions for fear of spooking the markets. And

2
pessimists early in the boom were largely dismissed at the time.
It is instructive to read the logic of non-economists who looked at house price data in the
same period. Paolo Pellegrini and John Paulson, whose wildly successful 2006 bet against
subprime mortgages is now the stuff of Wall Street legend, made the following argument,
as chronicled in Zuckerman (2009). First, they noted that house prices had deviated from
trend:

Suddenly, the answer was as plain as the paper in front of him: Housing prices
had climbed a puny 1.4 percent annually between 1975 and 2000, after inflation
was taken into consideration. But they had soared over 7 percent in the following
five years, until 2005. The upshot: U.S. home prices would have to drop by
almost 40 percent to return to their historic trend line. (p. 107)

Those facts are indisputable, but the logic that followed would have earned the two investors
a zero on an undergraduate finance exam:

To Paulson and Pellegrini, their discovery meant that housing prices were bound
to fall, at least at some point, no matter what the moves in unemployment,
interest rates, or the economy. (p. 108)

The “fundamental theorem of asset prices,” in its pure form, largely rejects the idea that
prices mean-revert, which is what Paulson and Pellegrini assumed. Researchers have found
evidence of mean reversion, but Lo and MacKinlay (2001) argue that it operates on a
relatively small scale, and that research has not “uncovered tremendous untapped profit
opportunities” (p. xxii). In other words, academic finance provided little support for mean
reversion on the scale that Paulson and Pellegrini expected.
Economists who argued that house prices were going to fall made two closely related
arguments. The first was that key relationships in the data had deviated from long-run
averages. The second was that there was a bubble in house prices, so that buyers were
willing to pay high prices today because of their expectations of even higher prices tomorrow.
Proving either claim was a challenge.
Regarding the first argument, the key data relationships most often analyzed were the
ratios of house prices to either incomes or rents. Making a convincing argument that these
relationships had deviated from long-run or equilibrium values was far harder than it sounds.
Simply measuring these ratios, for example, was an enormous challenge, because different
repeat-sales indices of house prices often yielded different results. Moreover, results based
on repeat-sales measures of prices differed sharply from results generated by hedonic pricing
models. An additional practical complication with using the rent-price ratio to evaluate
economists both inside and outside academia may have been reticent to make any sort of predictions, for
fear of damaging their reputations if they were wrong.

3
housing markets is that for residential housing, renters’ and buyers’ markets are qualitatively
distinct, so finding comparable properties proved challenging. Theoretical challenges also
confronted work with price–income or price–rent ratios. Economic theory predicts that the
level of interest rates should affect price-income and price–rent ratios, but researchers faced
questions over the proper interest rates to use in their calculations. Finally, researchers using
price-income or price–rent ratios were forced to take a stand on whether a deviation from
a long-term trend reflected disequilibrium behavior, or simply a change in the equilibrium
ratio itself. All of these issues made it difficult to mount a convincing counterargument at
the time.
What about asset-price bubbles? While there is great interest in bubbles among research
economists, the science of analyzing them remains primitive. The starting point for a formal
analysis of bubbles is the standard idea taken from partial equilibrium analysis of individual
choice that the amount someone should pay for something depends only on the market price
and not on any link between that price and the notion of its fundamental underlying value.
If the something is a financial asset, then the expected return is the key determinant for the
investor and the investor correctly ignores the question of why that return is as high or low
as it is when making decisions. In most contexts in economics, equilibrium—the invisible
hand—ensures that the value of the asset is tied to its fundamentals, typically because a
deviation from fundamentals would allow someone to make infinite profits. But in recent
years, economists have constructed theoretical examples in which investors cannot, for one
reason or another, bring prices into line with fundamentals and thus bubbles can persist.
Matching such models with the real world, however, has proved problematic.
In Section II.B, we review the arguments of a prominent pessimist, Paul Krugman.
Although his arguments were made in his widely read New York Times column rather
than in a formal academic paper, Krugman, now a Nobel Prize-winning economist, has
substantial credibility. He argued that because it is difficult to build in coastal areas of the
United States, those areas are more “bubble-prone.” Consequently, the rapid price increases
on the coasts but not elsewhere were prima facie evidence that there was a bubble and that
prices would eventually collapse.
It is tempting to call Krugman prescient because beginning in late 2006 prices did indeed
crash. But his arguments were problematic both ex ante and ex post. Ex ante, it is unclear
why Krugman thinks the coasts are more “bubble-prone.” The models we have of asset-price
bubbles do a reasonable job explaining why they can persist but have little to say about
where we might expect them to start. As one prominent researcher in the field writes, “we
do not have many convincing models that explain when and why bubbles start.”5 Krugman’s
thesis seems to hinge on the idea that scarce coastal land is valuable and bubbles can only
5
This quotation comes from the concluding paragraph of Brunnermeier (2008).

4
happen when assets are in short supply, but the whole point about bubbles is that the
fundamentals of supply and demand do not matter. Thus, there is no reason why land in
places where it is easy to build could not experience bubbles. Ex post, as we will explore at
length, the places in the United States where the housing market most resembled a bubble
were Phoenix and Las Vegas. According to recent research, both locations are characterized
by relatively high housing-supply elasticities; unlike certain coastal areas, the two cities have
an abundance of surrounding land on which to accommodate new construction.
Ultimately, our paper argues that the academic research available in 2006 was basically
inconclusive and could not convincingly support or refute any hypothesis about the future
path of asset prices. Thus, investors who believed that house prices were going to fall could
find evidence to support their position, while those who wanted to believe that house prices
would continue to rise could not be dissuaded either. There were reasonable arguments on
both sides.
We view a retrospective understanding of the real-time evolution of beliefs about the
path of house prices as central to understanding the financial crisis of 2007–2009. Many
lending decisions made during the housing boom seem daft in light of what happened.
Why would lenders extend credit to borrowers with troubled credit histories, low (or no)
down payments, or poorly documented incomes? Loans to these borrowers make sense for
investors who are optimistic about the future trajectory of house prices, because even small
amounts of positive appreciation protect mortgage investors from large losses. If house prices
fall, however, then borrowers are likely to have negative equity, leaving them vulnerable to
adverse life events like job loss that reduce their ability to make regular mortgage payments.
In particular, owners with negative equity are unable to pay off their mortgages by selling
their homes, so foreclosures often follow adverse life events.6 Gerardi et al. (2008) shows that
the negative relationship between house price appreciation and mortgage losses was well-
understood by market participants in 2006. Specifically, the paper shows that investors
could and did understand that subprime loans in particular carried a high degree of risk if
house prices fell. Other work, such as Foote et al. (2009), shows that prime loans, while
generally less risky than subprime loans, were also adversely affected by the unexpected
decline in housing prices. Thus, understanding the sources of optimism about house prices
in 2006 in light of the catastrophe that followed is more than an exercise in the history of
economic thought. Rather, it is a starting point toward an understanding of the role that
economic theory and the pronouncements of economists might have played in the biggest
financial crisis since the Great Depression.
The remainder of the paper proceeds as follows. Sections II and III review the argument
6
Falling house prices can also cause foreclosures if negative-equity borrowers believe that house prices
will not recover in any reasonable length of time—the so-called “ruthless” or “strategic” default.

5
made by housing-market pessimists and optimists, respectively. Section IV reviews the
research of the agnostic economists. We conclude in Section V with a discussion of the
challenges of forecasting house prices and of acting on those forecasts.

II. The Housing Pessimists


In this section, we focus on economists who took a strong position on the unsustainability
of future house price growth. We start with some research that pointed to the existence of
a housing bubble as early as 2002 and 2003, several years before housing markets peaked in
the United States. We then review arguments advanced by Krugman and others that geo-
graphical differences in residential building restrictions were important in getting a bubble
started.

A. Early warnings on the U.S. housing market


One of the first prominent housing pessimists was Dean Baker of the Center for Economic
and Policy Research in Washington, who in 2002 wrote that:

In the absence of any other credible theory, the only plausible explanation for
the sudden surge in home prices is the existence of a housing bubble. This means
that a major factor driving housing sales is the expectation that housing prices
will be higher in the future. While this process can sustain rising prices for a
period of time, it must eventually come to an end. (Baker 2002, p. 116)

Baker (2002) focused on the price–rent ratio, the housing equivalent of the price–dividend
ratio used to assess valuation in equity markets.7 According to standard theory, the price
of an asset should be equal to the present value of the sum of expected future dividends. A
housing asset’s dividend is essentially the service flow that it provides (that is, the flow value
of shelter), which is in turn roughly equal to the house’s rental value. By this reasoning,
the price–rent ratio (or its inverse) can be used as a measure of how close housing prices
are to fundamentals.
Baker showed that from 1995 to 2002, U.S. housing prices, as measured by the OFHEO
index, rose by almost 50 percent in nominal terms, an increase that was substantially greater
than the increase in overall prices (that is, inflation) during the same period. Akin to Shiller’s
findings, Baker found that prior to 1995, changes in nominal house prices moved in tandem
with overall prices, so that real house prices displayed no upward trend. The rise in the price
7
Leamer (2002) conducted a price–rent ratio analysis for a few markets in California and came to mixed
conclusions. He found the San Francisco Bay Area to be overvalued in the early 2000s, but concluded that
prices in the Los Angeles market were more consistent with fundamentals.

6
index in the seven-year period analyzed by Baker was also significantly greater than the rise
in rents; the Bureau of Labor Statistics (BLS) rental index rose by roughly 10 percent in real
terms during this seven-year period, while the house price index rose by almost 30 percent.
This difference is consistent with the time-series pattern of the price–rent ratio we present
in figure 1. Baker noted that the rental and house price indexes had diverged in the late
1970s and again in the late 1980s, when growth in housing prices outpaced growth in rents,
but that both episodes were followed by periods in which housing prices declined relative
to rents. In addition, he noted that the rental vacancy rate in 2002 was 9.1 percent—the
highest rate on record since the Census Bureau had begun collecting the data. The large
number of vacancies suggested some downward pressure on rents going forward.
In addition to the divergence between rents and prices in the U.S. housing market, Baker
also called attention to changes in demographic trends that could put additional downward
pressure on house prices. He noted that during the 1970s and early 1980s, housing grew from
about 17 percent of consumption to more than 25 percent, in large part due to increased
demand for housing from the first baby boom cohorts, who were then entering adulthood.
From the early 1980s to the mid-1990s, the housing share of consumption remained relatively
constant, consistent with the modest demographic changes taking place in the United States
at that time. In the future, Baker argued, as the baby boomers entered retirement, housing
demand—and hence prices—would likely fall.8
Finally, Baker discussed the role of interest rates in moving prices higher. As we will
discuss in more detail below, many economists pointed to low interest rates as justifying
higher housing prices, but Baker was skeptical of this claim. Nominal interest rates were
indeed low in the early 2000s, as the Federal Reserve had adopted a loose monetary policy
to combat the effects of the 2001 recession. However, Baker pointed out that nominal rates
could not explain the divergence of housing prices from fundamentals, as it is the real interest
rate (the difference between the nominal rate and expected inflation) that should influence
prices. During the boom period, real mortgage interest rates did not seem to be significantly
lower than their levels in the mid-1990s before the run-up in house prices began. To further
illustrate the point, Baker noted that from the early- to late-1980s there was a large drop in
nominal rates (from almost 15 percent in 1981 to approximately 9 percent in 1988) without
a large divergence between the rental and price indexes.
Another important early study in the pessimist camp was the 2003 Brookings paper
by Karl Case and Robert Shiller (2003). Its analysis can be separated into two parts.
The first part consists of a state panel-data analysis of the fundamentals that should have
driven the trajectory of U.S. housing prices from 1985 through 2002. The main focus is on
the relationship between prices and income per capita, although population, employment,
8
See also Mankiw and Weil (1989) on this point.

7
unemployment rates, housing starts, interest rates, and debt-to-income ratios (DTIs) also
enter into the regression models. The empirical findings did not provide conclusive evidence
one way or the other regarding the stability of housing prices, relative to what the authors
perceived to be housing-market fundamentals. In particular, the results were sensitive to the
particular geographic regions considered. For example, ratios of house prices to income per
capita were stable in the vast majority of states over the 18-year period, but in eight states,
the price-to-income ratios displayed pronounced cyclical patterns.9 In those eight states,
the other variables that proxy for housing-market fundamentals (besides income) added
significant explanatory power to the house price panel regressions. However, in an out-of-
sample forecasting exercise for those states, Case and Shiller found that the fundamentals
significantly underpredicted the actual movements in housing prices. Thus, while there was
some evidence of a potential bubble in these eight coastal states, the authors concluded
from their empirical analysis that overall, U.S. housing prices tracked market fundamentals
fairly well.
However, the authors came to a different conclusion in the second part of their paper,
which consisted of a survey of recent homebuyers in Orange County (Los Angeles, CA),
Alameda County (San Francisco, CA), Middlesex County (Boston, MA), and Milwaukee
County (Milwaukee, WI) in 2003. The survey was virtually identical to a 1988 survey that
Case and Shiller had conducted in the same areas, in which they found what they believed
to be fairly strong evidence of a housing bubble at that time. After the 1988 survey, housing
prices in many coastal areas fell dramatically, partially reversing the price gains of the mid-
to-late 1980s. Relative to the 1988 survey, the results of the 2003 survey were somewhat
mixed. The majority of homebuyers surveyed in 2003 reported that investment was a
“major consideration” in their purchase, or that they at least “in part” thought of their
housing purchase as an investment. Far fewer said that they were buying a house “strictly
for investment purposes” as compared to 1988. In addition, while only a small fraction of
borrowers believed that purchasing a house involved a great deal of risk in the 2003 survey,
the perception of risk in 1988 was even smaller in three out of the four areas (Milwaukee
being the one exception). The most striking result of the survey involved the respondents’
expectations of future price appreciation. In both the 1988 and 2003 surveys, more than 90
percent of respondents expected an increase in prices over several years, with the average
expected increase over the next year being extremely high (between 7.2 and 10.5 percent in
2003 and 6.1 and 15.3 percent in 1988). Expectations over a longer-term horizon (10 years)
were even higher in all four cities, varying from 11.7 percent per year in Milwaukee to 15.7
9
The volatile states, ordered alphabetically by region, were Connecticut, Massachusetts, New Hamp-
shire, Rhode Island, New Jersey, New York, California, and Hawaii.

8
percent in San Francisco in 2003.10 Given these survey findings, Case and Shiller concluded
that while the indicators of a bubble were perhaps not as strong in 2003 as they were in
1988 (except in Milwaukee), they were still present. In their concluding remarks, Case and
Shiller acknowledged that in the majority of states, home prices seemed to track well with
housing-market fundamentals, yet they took the position that a bubble had likely formed:

Nonetheless, our analysis indicates that elements of a speculative bubble in


single-family home prices—the strong investment motive, the high expectations
of future price increases, and the strong influence of word-of-mouth discussion—
exist in some cities. For the three glamour cities we studied, the indicators of
bubble sentiment that we documented ... remain, in general, nearly as strong in
2003 as they were in 1988. (p. 341)

B. The role of building restrictions in housing pessimism


Paul Krugman, among others, claimed that the geographic distribution of price increases
in the boom years proved the existence of a bubble and the inevitability of a crash. In
an August 8, 2005, column for the New York Times, Krugman wrote that with respect to
housing supply, the United States was really two countries — “Flatland” and the “Zoned
Zone”:

In Flatland, which occupies the middle of the country, it’s easy to build houses.
When the demand for houses rises, Flatland metropolitan areas, which don’t
really have traditional downtowns, just sprawl some more. As a result, hous-
ing prices are basically determined by the cost of construction. In Flatland, a
housing bubble can’t even get started. But in the Zoned Zone, which lies along
the coasts, a combination of high population density and land-use restrictions
— hence “zoned” — makes it hard to build new houses. So when people become
willing to spend more on houses, say because of a fall in mortgage rates, some
houses get built, but the prices of existing houses also go up. And if people
think that prices will continue to rise, they become willing to spend even more,
driving prices still higher, and so on. In other words, the Zoned Zone is prone to
housing bubbles. And Zoned Zone housing prices, which have risen much faster
than the national average, clearly point to a bubble. (Krugman, “That Hissing
Sound,” New York Times)
10
In an accompanying comment to the Case and Shiller paper, Quigley raises the possibility that the
respondents did not understand the concept of compounding interest, and may have answered differently if
the question had been framed in a slightly different manner.

9
As the Krugman excerpt illustrates, it is easy to find individual cities (like New York and
San Diego) that support a Flatland/Zoned Zone dichotomy. But a more systematic look
at the data reveals a more complicated story. Many cities with highly elastic supply curves
for housing also experienced rapid increases in prices, especially in the last few years of
the housing boom (2002–2005). Moreover, there does not seem to be a negative correlation
between the price increase of new homes and the number of new homes built during the
boom, as a simple dichotomy would predict.
A recent paper by Saiz (2010) allows a precise look at the correlations between housing-
supply elasticities and price increases during the recent U.S. housing boom. Saiz uses the
unique geographic features of different metropolitan areas to construct exogenous housing-
supply elasticities. Topographical features that limit new building projects include coast-
lines, large bodies of water near city centers, or land that is steeply sloped. By calculating
the prevalence of these features for each large city in the United States, Saiz constructs
housing-supply elasticities that do not rely on the (potentially endogenous) use of legal
restrictions that limit the construction of new homes.
Saiz finds that his elasticity measure correlates strongly with changes in housing prices
at the level of the metropolitan statistical area (MSA), a finding that we replicate in fig-
ure 2. On the vertical axes of each of the three panels in the figure is the percentage
change in OFHEO-measured prices from a given time period (1995–2005, 1999–2002, and
2002–2005).11 The horizontal axes feature the Saiz elasticities for 92 U.S. cities that had
populations greater than 500,000 as of 2000.12 As we might expect, cities with very high
supply elasticities saw only small increases in housing prices during the three time periods.
The two cities on the far right of each panel, with supply elasticities above 5, are Fort
Wayne, IN (elasticity = 5.36) and Wichita, KS (elasticity = 5.45). With elasticities so high,
whatever demand increase these cities might have experienced resulted in more supply via
new construction, not higher prices.
As we move to the left from these two points, price increases become more common.
However, we do not have to move all of the way to the left to find large price increases;
many cities with relatively high elasticities (above 1.5) also experienced substantial price
increases during the housing boom. Figure 3 presents the same data as the previous figure,
but restricts the sample to cities with elasticities of 3.0 or less.13 The fitted regressions lines
in these panels also illustrate a negative correlation between supply elasticities and price
11
The periods are measured from the fourth quarters of the given years, so the 1995–2005 span covers
1995:Q4–2005:Q4, etc.
12
There are actually 95 U.S. cities that meet this population criterion. However, there are no OFHEO
price indexes for Ventura, CA; Sarasota, FL; and Jersey City, NJ. Thus, the panels of figure 2 have only 92
data points each.
13
This restriction generates a legible graph that includes the names of the remaining cities. Eleven cities
are eliminated by this restriction, so each panel of figure 3 has 81 cities.

10
increases. But these lines also leave a lot of variation unexplained. For example, in panel
A of figure 3, which uses data from 1995:Q4 through 2005:Q4, the R2 of the regression line
is only 0.42. The corresponding R2 for panel B (1999:Q4–2002:Q4) is 0.29, while that for
Panel C (2002:Q4–2005:Q4) is 0.30.14
The takeaway point from figure 3 is that house price increases occurred in many cities
where new construction could have occurred with relative ease. We can press this point
further by noting that new construction did in fact occur in areas that also had large price
increases. For this analysis, we switch to annual state-level data, due to inconsistences
in the reporting of building permits at the city level.15 Figure 4 presents a scatterplot of
state-level increases in OFHEO house prices from the three periods against increases in
new building permits.16 Krugman’s Flatland/Zoned-Zone theory, combined with a uniform
increase in demand across the different states, would predict a negative relationship between
price increases and construction. Yet there is little evidence of a negative correlation in this
graph. In unreported regressions, we found that the raw correlations between price increases
and new construction were actually often positive, though these correlations often declined
to insignificance in regressions where state-level unemployment rates were also included.
Of course, raw correlations between highly endogenous variables like house prices and
new construction do not provide airtight evidence for any particular theory of the housing
market. But such correlations can provide evidence against theories that predict different
correlations. Simply put, it is hard to tell a story that during the housing boom, the main
driver of differential price increases across states was varying supply elasticities combined
with a uniform increase in demand. It appears that some states with abundant land, like
Arizona, Nevada, and Florida, experienced much larger increases in demand than other
states, including those located on the coasts.17
Do our findings suggest that Krugman was mistaken to point to differential price in-
creases in support of the bubble story? Not at all. For one thing, it is true that as a general
matter, price-inelastic cities like New York and Boston saw higher house price increases than
did price-elastic cities like Fort Wayne and Wichita. More to the point, however, one could
argue that big price increases in elastic cities like Las Vegas and Phoenix provided even
14
Including the square of the Saiz elasticity raises the R2 by less than 0.04 in all cases. Linear regressions
that include all 92 cities in figure 2 have R2 s of 0.39 (1995–2005), 0.30 (1999–2002), and 0.32 (2002–2005).
15
We found that tabulations of city-level building permits were less consistent than the price data from
OFHEO, due to definitional changes for metropolitan areas that occurred in the mid-2000s (which is our
main period of interest). Nevertheless, we investigated the relationship between new construction and price-
increases for about 70 cities that we could match. Our results were consistent with the state-level evidence
presented here.
16
The absolute changes in building permits are divided through by the state’s initial population.
17
Davidoff (2010) makes a similar point, but does not use the Saiz elasticities. He shows that coastal
markets saw greater average price growth during the boom, but did not experience greater crashes or greater
volatility than non-coastal areas, where supply elasticities are assumed to be larger.

11
stronger evidence that bubble psychology had gripped the market. Due to abundant sur-
rounding land, Las Vegas and Phoenix had been long able to accommodate large numbers
of new migrants without big increases in house prices. The fact that some people became
willing to buy houses in these cities at prices far above construction costs, especially after
2002, is perhaps a signal that price expectations had become irrational. Unfortunately,
the uneven pattern of price increases we find across U.S. cities was interpreted differently
by the anti-bubble camp. This group sometimes conceded that prices in some cities had
gotten out of line, but argued that a few overheated markets do not necessarily reflect a
national housing bubble. Indeed, as a general matter, geographic differences in house price
appreciation played a large role in the writings of many anti-bubble economists. We turn
to a detailed analysis of these arguments in the next section.

III. The Housing Optimists


Some economists expressed deep skepticism about the possibility of a bubble and went on
the record in stating that they did not expect house prices to fall:

As of the end of 2004, our analysis reveals little evidence of a housing bubble.
In high-appreciation markets like San Francisco, Boston and New York, current
housing prices are not cheap, but our calculations do not reveal large price
increases in excess of fundamentals. For such cities, expectations of outsized
capital gains appear to play, at best, a very small role in single-family house
prices. Rather, recent price growth is supported by basic economic factors such
as low real long-term interest rates, high income growth and housing price levels
that had fallen to unusually low levels during the mid-1990s. (Himmelberg,
Mayer, and Sinai 2005, p. 68)

This quotation comes from the introduction of Himmelberg, Mayer, and Sinai (2005), which
became perhaps the most widely cited evidence against a housing bubble upon its publi-
cation. The authors offered a persuasive critique of the usual measures marshalled by
pro-bubble advocates, including the price–rent ratio and price–income ratio. Himmelberg,
Mayer, and Sinai based their analysis on the more formal “user cost of housing” concept,
which recognizes that the cost of owning a house reflects more than the purchase price of
the property. For example, other variables that affect the cost of ownership include tax
benefits (like the mortgage interest deduction), property taxes, maintenance costs, antici-
pated capital gains, opportunity costs, and the risk of large capital losses. Thus, studying
the ratio of rental costs to ownership costs is a more theoretically sound way to investigate
whether a bubble may exist than by investigating the ratio of rents to house prices. We can

12
write such a relationship as follows:

Rt = Pt · µt , (1)

where µt , the user cost of housing, is defined as

µt = rtrf + ωt − τt (rtm + ωt ) + δt − gt+1 + γt . (2)

In this expression, rtrf is the one-year, risk-free interest rate, ωt is the one-year cost of
property taxes, the term τt (rtm + ωt ) represents the tax deductibility of mortgage interest
and property taxes (for those that itemize their returns), δt reflects maintenance costs, and
gt+1 is the expected capital gain over a one-year horizon.18 Finally, γt is a risk premium,
reflecting the higher risk of owning versus renting.
Equation 1 tells us that the price–rent ratio equals the inverse of the user cost ( µ1t ),
so variation in the price–rent ratio should reflect variation in the user cost of housing.
Himmelberg, Mayer, and Sinai (2005) made the correct point that simply looking at time-
series variation in price–rent ratios, or cross-sectional differences in these ratios, without
considering possible variation in user costs is misleading and could not shed light on the
bubble question.
The right thing to do is to work with careful estimates of user costs in various cities, so
the authors constructed measures of user costs for 46 metropolitan areas from 1980 through
2004. Himmelberg, Mayer, and Sinai compared differences in user costs across cities, as
well as changes in user costs over time within a city. Their first observation was that user
costs can vary greatly across U.S. cities. The highest user-cost market (Pittsburgh) had
user costs that were more than double that of the lowest user-cost market (San Jose). In
addition, user costs were falling over time, and by 2004 were below their long-run averages
in most cities. According to equation 1, declining user costs imply rising price–rent ratios.
Consequently, the increasing price–rent ratios used as evidence to support a bubble actually
reflected, in large part, changes in user costs and thus in fundamentals.19
To make their analysis more precise, the authors then compared the ratio of imputed
rent (calculated by multiplying their user cost estimates by the OFHEO house price index)
to actual rent (the BLS index) in 2004 to its 25-year average in each city (figure 2 in their
paper). In the vast majority of cities they found the ratio in 2004 to be very close to or
18
The gt+1 term is negative if capital losses are expected.
19
In this exercise the authors were not comparing user costs across cities, but were only looking at trends
within cities over a 25-year period. In this exercise variation in user costs does not come from assumptions
about cross-city variation in expected capital gains (the gt+1 term), as the authors set expected capital
gains in each city to the one-year average realized return calculated over the entire 25-year period. Thus,
by assumption, there was no variation in expected capital gains across cities.

13
below its 25-year average. Moreover, the cities in which the ratio in 2004 was substantially
above its average were not places that had experienced rapid price appreciation (Detroit,
Milwaukee, Minneapolis, and so on). The authors also constructed the ratio of imputed
rent to per capita income as an alternative measure of housing valuations, and came to a
similar set of conclusions: the ratio in 2004 was below its 25-year average in most cities in
their sample.
Another study that took a skeptical view of the housing bubble was McCarthy and Peach
(2004). The paper analyzed the arguments in support of a housing bubble, and concluded
that there was very little credible evidence:

Our main conclusion is that the most widely cited evidence of a bubble is not
persuasive because it fails to account for developments in the housing market
over the past decade. In particular, significant declines in nominal mortgage
interest rates and demographic forces have supported housing demand, home
construction, and home values during this period. Taking these factors into
account, we argue that market fundamentals are sufficiently strong to explain
the recent path of home prices and support our view that a bubble does not
exist. (p. 2)

The authors’ first argument was that the most popular types of house price indices used at
the time—repeat-sales indices and median price indices—were not very accurate in measur-
ing the value of the U.S. housing stock. Median price indices had the most severe problems.
According to McCarthy and Peach, median price indices are too volatile in the short run
because the composition of sales fluctuates substantially from month to month. Moreover,
the pricing data used to construct the indices reflect only recent sales transactions, so they
are likely not very representative of the nation’s entire housing stock. The repeat-sales in-
dices largely solve these issues, but even these measures cannot control for changes in the
quality of the housing stock over time due to depreciation or renovations.
As an alternative price index, the authors advocated the constant-quality new-home
price index that is published by the U.S. Bureau of the Census. This is a hedonic index
that accounts for changes in the quality of homes over time.20 Compared to the other price
indices, the Census Bureau series showed significantly less home price appreciation over
the authors’ sample period (1977–2003). The implication was that a significant portion of
the house price appreciation measured using the median price and repeat-sales indices was
attributable to quality increases. When the authors used the hedonic constant-quality index
instead of the repeat-sales index to calculate a time series of the aggregate price–rent ratio,
20
A hedonic index is constructed by regressing sale prices on various housing and neighborhood charac-
teristics.

14
they found much less support for a housing price bubble.21
The paper then pointed out a list of shortcomings with respect to the most common
statistics used by proponents of the housing bubble perspective: the price-to-income ratio
and the price–rent ratio. According to McCarthy and Peach, the first significant drawback
of the two statistical series is that they ignore the effect of nominal interest rates. As we
discuss below, decreases in nominal interest rates make homeownership more affordable for
borrowing-constrained households. Lower nominal rates also lower the opportunity cost of
homeownership, as these rates imply lower yields on competing assets.22 To support their
argument, the authors pointed out that from 1990 to 2003, the decline in nominal interest
rates resulted in a 130-percent increase in the maximum mortgage debt for which a family
earning the median U.S. household income could qualify, using a then-standard maximum
debt-to-income ratio of 27 percent.23
Finally, McCarthy and Peach used a stylized structural model of the housing market
in order to compare predicted equilibrium prices from the model with actual prices in the
data. They then decomposed price changes into contributions from growth factors such
as income and consumption, as well as from user-cost declines largely due to interest rate
movements. According to the model, housing prices were expected to appreciate in the late
1990s and early 2000s, in large part because of a substantial decline in user costs induced
by declining interest rates over the period. But more significant is their finding that the
model’s predicted prices were greater than actual prices at the end of the sample period,
leading the authors to conclude that U.S. house prices during this period were actually lower
than what the fundamentals would justify.
John Quigley is another well-respected real estate economist who was highly skeptical
that a housing bubble was present. Along with Christopher Mayer, he was tasked with
providing formal comments to the Brookings paper written by Case and Shiller (2003). In
those comments (Quigley 2003), he discussed numerous reasons for skepticism regarding
how they interpreted their survey results. In addition, Quigley listed eight specific reasons
for questioning the existence of a housing bubble in 2003. First, he pointed to Case and
Shiller’s finding of a stable co-movement of income and house prices for the large majority of
U.S. states. Second, in an argument similar to that of Himmelberg, Mayer, and Sinai (2005),
Quigley pointed out that the user cost of housing in the United States had been trending
downward since the early 1980s. Third, Quigley noted that demographic trends would
21
See chart 7 in McCarthy and Peach (2004).
22
In standard versions of asset-pricing theory, which do not include borrowing constraints, the real
interest rate, not the nominal rate, should matter for housing valuations. Whether the real or the nominal
interest rate matters more in practice is a topic we return to below.
23
This is the same argument that Mayer (2003) discussed in his effort to rationalize a role for nominal
rates in house price increases (although he was skeptical of this channel). We discuss this argument in more
detail below.

15
probably increase the demand for housing in the United States, a claim that stood in contrast
to the discussion in Baker (2002).24 Fourth, Quigley wrote that the continued desirability of
the amenities available in many coastal cities, along with the inelasticity of housing supply
in those areas, would put continued upward pressure on prices in many cities.25 Fifth,
Quigley noted the severe downward rigidity in nominal housing prices, which would make
large nominal price declines unlikely. Sixth, Quigley claimed that high transactions costs
in housing markets would tend to decrease the amount of speculation and trading volume.
Seventh, like McCarthy and Peach (2004), Quigley noted the inability of popular house price
indices to capture the increasing quality of American homes. Finally, Quigley emphasized
the point that U.S. housing markets are local, so the imperfect correlation across local
housing markets would make a large national price decline unlikely.
Of all the pre-crisis papers that were skeptical of the housing bubble, perhaps Smith and
Smith (2006) was the most adamant. The paper used a net present value (NPV) framework
to calculate the fundamental value of housing in various areas across the United States. The
valuation framework employed was based on the ideas that underpin much of the price–rent
analysis discussed above. The starting point is that a house, like any other traded asset,
can be valued by calculating the projected stream of future service flows, discounted by an
appropriate rate of return. However, unlike a stock, which has only future cash dividends,
the future service flows associated with homeownership come from many sources. Some of
these flows are largely observable financial variables. Examples include the imputed rent net
of mortgage payments, as well as expenses such as maintenance costs and property taxes.
Yet the service flows to homeownership also include unobservable nonfinancial benefits,
such as privacy and the discretion to modify the property. According to this framework,
a homebuyer should be able to use projected cash flows and an assumed required rate of
return to determine whether the NPV of homeownership is positive or negative. If it is
positive, then purchasing the house is a worthwhile investment; if it is negative, renting is
the more attractive option. Smith and Smith focus most of their analysis on calculating
internal rates of return (IRRs), which are defined as the rates of return that would make
NPVs zero. Equivalently, these IRRs are the rates of return that would make potential
buyers indifferent between owning and renting.
Smith and Smith’s innovation was to obtain estimates of the imputed rent of owner-
occupied properties, which are by definition unobservable. This calculation was made
possible using matched data on rental properties in areas with owner-occupied houses of
24
Unlike Baker, who argued that the baby boomers would likely downsize their homes as they crossed
the retirement threshold, Quigley argued that they would “continue to resist downsizing” (p. 357).
25
Note the contrast with Krugman on this point. Quigley saw higher prices in coastal areas as evidence
that prices were in line with fundamentals, while Krugman argued that coastal areas were more bubble-
prone.

16
similar quality. Thus, the authors could obtain relatively precise estimates of the rental
income that the owner-occupied houses would command if they were placed on the rental
market. Their data included 10 U.S. markets representative of both the Flatland and the
Zoned Zone: Atlanta, Boston, Chicago, Dallas, Indianapolis, Los Angeles County, New Or-
leans, Orange County, San Bernardino County, and San Mateo County. The results of their
IRR calculations were striking. Only one of the markets (San Mateo County) appeared to
be significantly overpriced at the time the paper was written. Moreover, even some widely
perceived bubble areas—such as Boston, Chicago, and San Bernardino County—were sig-
nificantly underpriced relative to the estimated fundamental values.

IV. The Housing Agnostics


As we explained in the introduction, the most prevalent view among economists in the pre-
crisis period was an unwillingness to make strong predictions about the future path of U.S.
housing prices. We first turn to Mayer (2003), who investigated the role of real and nominal
interest rates in the determination of housing prices. Two important empirical findings in
this paper are, 1) that real mortgage rates were actually higher in the early 2000s than they
were in the 1970s, and not significantly lower compared to their levels in the 1990s (figure
3 in Mayer 2003), and 2) that estimates from historical data do not show large correlations
between real interest rates and housing prices. Given these results, Mayer pointed toward
low nominal rates as perhaps the only rational explanation for the increase in housing prices
during the 2001 recession, though he recognized that standard theory predicts that real and
not nominal rates should influence housing demand. One way in which nominal prices could
matter, Mayer wrote, is if target payment-to-income ratios were important in the housing
market. If there was some maximum payment-to-income ratio that lenders refused to exceed
for a significant fraction of potential borrowers, then this constraint would essentially be
relaxed if the nominal mortgage rate fell. The demand for housing would therefore increase,
and prices would rise. However, Mayer was skeptical about this channel, stating:

There are many reasons to be skeptical of a model in which demand for housing
is generated by a fixed payment-to-income ratio. For example, this model clearly
does not hold when one examines cross-sectional data for U.S. metropolitan ar-
eas, which show that consumers do not have a fixed ratio of housing costs to
income. Evidence suggests that homeowners and renters spend a higher percent-
age of income in high-priced areas like San Francisco than in low-priced ones like
Milwaukee. So, at least cross-sectionally, this theory does not hold. Whether it
is true within metropolitan areas, one could make some arguments. Although it
is hard to believe in a target payment-to-income ratio completely, it is the only

17
model that I can come up with that predicts that lower nominal interest rates
will lead to higher home prices. (p. 352)

Haines and Rosen (2007) found evidence that a measure of housing affordability, driven
in large part by changes in nominal mortgage interest rates, was, in fact, an important
determinant of U.S. housing prices in the period 1980–2006. The paper estimated a reduced-
form regression of housing prices on a number of housing supply and demand determinants
including an “affordability index,” defined as the ratio of median household income to
the annual payment on a fixed-rate, 30-year, $100,000 mortgage with a 20 percent down
payment. The affordability index rises with either an increase in household income or a
decrease in the nominal mortgage rate, as these movements make it easier for a household
to buy a home at any given price. The authors pooled data across the largest 43 MSAs and
found that the affordability index was a significant determinant of movements in housing
prices over time and price differences across markets.26
In addition to their focus on the relationship between affordability and house prices,
Haines and Rosen (2007) also attempted to shed light on whether a bubble existed in the
national housing market. On the one hand, they noted that most of the academic literature
had taken a skeptical view of bubble claims, writing:

The general consensus of the academic literature is that home prices are largely
in line with fundamentals. Overpriced markets, if any, are limited in number
and in the scope of overpricing. (p. 18)

However, the authors also pointed out that many non-academic studies had firmly concluded
a bubble existed; a key example is a study by Global Insight and National City Corporation
(2006). Haines and Rosen (2007) hypothesized that the difference was due to the use of more
recent data in non-academic research. The publication lag in scholarly journals meant that
many recent academic papers had not made use of data from 2005 and 2006, a period which
was beginning to show a softening in the national housing market. For their part, Haines and
Rosen (2007) ran house-price regressions on data through 2006, using the predicted values
from these regressions as proxies for fundamental housing prices. They then compared these
predictions to actual prices at the end of the sample period. Results were mixed. In many
markets, actual 2006 prices were below predicted prices, while in some other markets, actual
prices were 10 to 20 percent above their predicted values. The authors noted that the MSAs
that seemed to be the most overvalued according to the model (for example, New York and
San Francisco) had generally been more volatile than the majority of U.S. markets, so the
authors downplayed the significance of these findings somewhat. At the end of the day,
26
The R2 of a univariate regression was 0.76.

18
the authors did not interpret their empirical work as strong evidence for either side of the
bubble debate.
Krainer and Wei (2004) was one of the pre-crisis studies that focused on the price–
rent ratio. A key contribution of this paper was to study the price–rent ratio for housing
using statistical techniques that had been developed to model the price–dividend ratio for
stocks.27 Previous stock-market research had pointed out that the price–dividend ratio
could be decomposed into two terms, one that reflected expected growth in dividends and
another that reflected expected future discount rates (or, equivalently, the rates of return
that future investors would require). If dividends were expected to rise sharply in the future,
or if we expected future discount rates to be low, then we would predict that the today’s
price–dividend ratio would be high. Conversely, if dividends were expected to fall, or if we
thought that future investors would discount income highly, then investors today would be
unwilling to pay high prices for stocks. The current price–dividend ratio would therefore
be low. One crucial issue in this line of research is that discount rates are unobservable.
Consequently, researchers often proxied for required rates of return with the actual rates of
return observed in the stock market. In particular, the required rate of return from period
t to t + 1 was set equal to the sum of the dividend yield in period t and the capital gain in
the stock price between periods t and t + 1.
To apply this methodology to the housing market, Krainer and Wei constructed a time-
series of the price–rent ratio at the national level going back to the early 1980s. The
numerator of this ratio was the repeat-sales existing house price series from OFHEO. The
owner’s equivalent rent series from BLS was used for the denominator.28 The authors then
decomposed their price–rent series into movements in future returns (that is, movements in
the proxy for future discount rates) and movements in rent growth. Because this exercise
used actual future values of dividend growth and rates of return to proxy for expected future
values of these two variables, they could not decompose movements in the price–rent ratio at
the end of the sample, because not enough future values of rents and returns were available.
Krainer and Wei reported two main findings. First, they found that future rents and
returns did a reasonable job of explaining the total variation in the price–rent ratio. In
other words, the price–dividend decomposition originally developed to explain the stock
market did a good job of explaining patterns in the housing market as well. Second, the
authors found found that most of the movements in the price–rent ratio could be explained
by movements in required rates of return, rather than movements in rents.29
27
See, for example, Campbell and Shiller (1988) and Cochrane (1992).
28
Krainer and Wei found that this ratio had substantially increased in the early 2000s after having been
roughly flat over the previous 15 years. As we saw in figure 1, there was a dramatic rise in the price–rent
ratio from around 1.0 in the late 1990s to nearly 1.4 by 2006.
29
This result is similar to what Cochrane (1992) found in a decomposition of the price-to-earnings ratio
in equity markets.

19
What did these results imply about the possibility of a bubble in the housing market? On
one level, the authors were comforted by the fact that a rational model based on fundamental
factors like future rents and discount factors explained much of the historical variation in
the price–rent ratio. “Put another way,” the authors wrote, “other factors, such as bubbles,
do not appear to be empirically important for explaining the behavior of the aggregate
price–rent ratio” (p. 3). Yet the authors had mixed feelings about whether a bubble existed
in the current market. Like other researchers, Krainer and Wei used actual returns to proxy
for the unobserved discount rates. These returns included capital gains, which are simply
changes in prices. Thus, the statement that future discount rates were likely to explain the
high price–rent ratio during the housing boom was really a statement that future prices
should be expected to move in ways that were unrelated to rents. Specifically, house-
price appreciation should slow down, reflecting the lower discount rates required by future
investors. These lower discount rates, in turn, could justify the high level of the price–rent
ratio at the time the article was written. Of course, there would be no way to know for sure
whether movements in future prices had anything to do with future discount rates, because
the discount rates are unobservable. All in all, while the model could conceivably explain
movements in housing’s price–rent ratio in the past, there were no observable factors that
conclusively proved that the current price–rent ratio did not reflect a bubble. The model
could only point out how future prices would be expected to move if a bubble did not exist.
Gallin (2008) also provided a technical analysis of house prices and rents, focusing on
the rent-price ratio (which, of course, is simply the inverse of the price–rent ratio discussed
above). To construct this ratio, Gallin used a repeat-sales index developed by Freddie Mac
and a price index for tenants’ rent from the BLS.30 The sample period was 1970 through
2003; while Gallin’s paper was published in 2008, the first draft of the paper was completed
in 2004 and the data was not updated in the intervening time period. The paper focused
on determining how well the rent-price ratio predicts future changes in real rents and house
prices. Gallin’s first finding was that there is a stable long-run relationship between house
prices and rents (or, more formally, that rents and prices are cointegrated over the sample
period). Given this finding, he estimated an error-correction model (ECM) to determine
how rents and prices fluctuate in the short-run around their stable long-run relationship.
The ECM model estimates suggested that rents and prices both correct back toward each
other, but the findings were inconclusive because the estimates were imprecise. Gallin then
turned to an analysis of long-horizon regressions (as opposed to the higher frequency ECM).
Specifically, he tested the hypothesis that the rent-price ratio predicts future movements
of rents and house prices over a three-year horizon. The data showed that periods of low
30
Gallin used the tenant’s rent index as opposed to the owner’s equivalent rent index because of the
former’s superior time series coverage.

20
rent-price ratios were followed by periods where real rent growth was faster than usual, and
real house price growth was slower than usual.31 This observation is consistent with the
finding of mean-reversion from the ECM estimates. In addition, the response of house prices
to the rent-price ratio in returning to the long-run relationship was found to dominate the
response of rents.
Gallin concluded that it was reasonable to use the rent-price ratio as a measure of
valuation and a gauge for housing price fundamentals. But he was disinclined to interpret
the rapid decline in the ratio at the end of his sample period as conclusive evidence of a
bubble:

Because a low rent-price ratio has been a harbinger of sluggish price growth since
1970, it seems reasonable to treat the rent-price ratio as a measure of valuation
in the housing market. Indeed, one might be tempted to cite the currently low
level of the rent-price ratio as a sign that we are in a house price “bubble.”
However, several important caveats argue against such a strong conclusion and
in favor of future research. (p. 19)

The caveats included potential measurement error in rents and house prices, concerns that
the theory motivating the empirical work was too simplistic and ignored important issues
like risk and transactions costs, and misgivings over relying too much on any particular
statistic to provide out-of-sample forecasts of a volatile and notoriously unpredictable asset
price.
Davis, Lehnert, and Martin (2008) constructed the longest time series of rent-price ratios
(to our knowledge), using the Decennial Census of Housing (DCH) to obtain data going all
the way back to 1960. The authors used the DCH to construct five benchmark estimates
of the average rent-price ratio for the stock of owner-occupied housing.32 They then used
quarterly rent and price indexes (BLS tenants’ rent index and the Freddie Mac repeat-sales
index) to interpolate rent and price data between the DCH years and to extrapolate the
data beyond 2000. They found that between 1960 and 1995, the rent-price ratio fluctuated
in a relatively narrow range of between 5.0 and 5.5 percent (except for a brief period in the
early 1970s when it rose to approximately 6.0 percent). In the period 1995–2006, however,
the ratio declined sharply, to a historical low of 3.5 percent. The authors speculated that
this dramatic decline could have occurred because of a drop in the discount rate on housing
dividends (rents), because of an increase in expected future capital gains, or both. They
concluded that a return of the rent-price ratio to its historical average would likely require
a modest decline in housing prices:
31
In these regressions, Gallin controlled for the user cost of housing.
32
They estimated imputed rents using data on rental properties and a hedonic regression technique to
map the rental property data to owner-occupied houses.

21
If the rent-price ratio were to rise from its level at the end of 2006 up to about
its historical average value of 5 percent by mid-2012, house prices might fall by
3 percent per year, depending on rent growth over the period. (p. 280)

This quote summarizes the views of the agnostics quite nicely. Despite finding fairly
convincing evidence that the U.S. housing market was overvalued in 2004–2006, they were
unwilling to take a strong position on the existence of a housing bubble for various rea-
sons, including a lack of confidence in the data and ambiguity regarding the predictions of
economic theory.

V. House Price Forecasts and Public Policy


From our review of the pre-crisis housing literature from the early-to-mid-2000s, it is ap-
parent that well-trained and well-respected economists with the best of motives could and
did look at the same data and come to vastly different conclusions about the future tra-
jectory of U.S. housing prices. This is not such a surprising observation once one realizes
that the state-of-the-art tools of economic science were not capable of predicting with any
degree of certainty the collapse of U.S. house prices that started in 2006. The asset-pricing
literature does not yet have a firm grasp on when and why prices can deviate from market
fundamentals for long periods of time. Even the best models have a difficult time explaining
many of the extremely large movements in asset prices that have characterized our financial
markets in the past, including the stock-market crashes of 1929, 1987, and 2000–2002. One
wonders how market participants might have responded to a consensus in the economics
profession that housing prices were unsustainable, meaning that highly leveraged bets on
housing would be likely to end disastrously. But as this paper shows, market participants
were never forced to respond to such a consensus because consensus did not exist.
While the assumption that asset markets are efficient in an informational sense has un-
doubtedly improved our understanding of financial markets in many ways, it appears that
large and systematic departures from efficiency can and do take place. Concerns that real-
world asset markets can be subject to irrational exuberance has been heightened by the
results of laboratory experiments, where bubbles often form in artificial trading environ-
ments. Smith, Suchanek, and Williams (1988), a seminal paper in this field, documented
that bubbles arose in controlled experiments in which participants had full knowledge of
the environment, including the intrinsic dividend value of the asset. In a more recent study,
Haruvy, Lahav, and Noussair (2007) followed the same methodology, and not only veri-
fied the existence of bubbles in controlled settings, but also found substantial evidence of
adaptive expectations:

22
We find that individuals beliefs about prices are adaptive, and primarily based
on past trends in the current and previous markets in which they have partic-
ipated. Most traders do not anticipate market downturns the first time they
participate in a market, and, when experienced, they typically overestimate the
time remaining before market peaks and downturns occur. (p. 1901)

One of the more positive aspects of their study was the finding that bubbles are much less
likely to occur as traders become more experienced. However, in a follow-up study, Hussam,
Porter, and Smith (2008) showed that this experience effect is not robust to changes in the
underlying environment:

[I]n order for price bubbles to be extinguished, the environment in which the
participants engage in exchange must be stationary and bounded by a range of
parameters. Experience, including possible “error” elimination, is not robust
to major new environment changes in determining the characteristics of a price
bubble. (p. 924)

These findings seem especially relevant to the experience of housing markets both in the
United States and around the world over the past decade. The expansion of housing credit
to previously underserved segments of the U.S. population (that is, to subprime borrowers)
meant an influx of new, and potentially less financially sophisticated housing-market par-
ticipants. In addition, the introduction and widespread use of more complicated structured
finance products, like the many types of collateralized debt obligations (CDOs), implied a
rapidly changing menu of financial products.
The evolving landscape of mortgage lending is also relevant to an ongoing debate in
the literature about the direction of causality between reduced underwriting standards and
higher house prices. Did lax lending standards shift out the demand curve for new homes
and raise house prices, or did higher house prices reduce the chance of future loan losses,
thereby encouraging lenders to relax their standards? Economists will debate this issue for
some time. For our part, we simply point out that an in-depth study of lending standards
would have been of little help to an economist trying to learn whether the early-to-mid
2000s increase in house prices was sustainable. If one economist argued that lax standards
were fueling an unsustainable surge in house prices, another could have responded that
reducing credit constraints generally brings asset prices closer to fundamental values, not
farther away.
In our view, understanding the ways in which economists try to understand asset prices
in real time is critically important with regard to the direction of future policy. First, it
should give substantial pause to those who believe we can eliminate the formation of bubbles
by simply curbing corruption and by imposing more regulation on markets. Since bubbles

23
arise even in the most controlled environments, we may need to acknowledge that we do
not currently have the ability to prevent a bubble from forming or the ability to identify
a bubble in real time. Instead, we could focus on ensuring that potential homeowners
and investors understand the risks associated with their investments, and take necessary
precautions against such declines. If we have learned anything from this crisis, it is that
large declines in house prices are always a possibility, so regulators and policymakers must
take them into account when making decisions. A 30 percent fall in house prices over
three years may be very difficult, if not impossible, to generate in any plausible econometric
model, but a truly robust financial institution must be able to withstand one. The fact that
so many professional investors as well as individual households ignored this possibility, even
in 2006, suggests that we cannot allow investors to try to time market collapses.
Finally, individual homeownerships should also be robust to significant declines in hous-
ing prices. A standard way to do this is by requiring a substantial downpayment, which
reduces the probability of negative equity when prices fall. In the future, robustness might
also be enhanced by requiring the purchase of insurance against house price declines, a
financial product made at least theoretically possible by the development of new markets
in house price risk. At the very least, any policy that encourages buyers to take on loans
with low downpayments deserves to be evaluated carefully, given the sharp decline in house
prices we have recently experienced.

24
References
Baker, Dean. 2002. “The Run-Up in Home Prices: A Bubble.” Challenge 45 (6): 93–119.
Brunnermeier, Markus K. 2008. “Bubbles.” In The New Palgrave Dictionary of Economics,
edited by Steven N. Durlauf and Lawrence E. Blume, Second Edition. Basingstoke:
Palgrave Macmillan. Available at http://www.dictionaryofeconomics.com/article?id=
pde2008 S000278.
Campbell, John Y., and Robert J. Shiller. 1988. “The Dividend-Price Ratio and Expecta-
tions of Future Dividends and Discount Factors.” The Review of Financial Studies 1
(3): 195–228.
Case, Karl E., and Robert J. Shiller. 2003. “Is There a Bubble in the Housing Market?”
Brookings Papers on Economic Activity 2003 (2): 299–342.
Cassidy, John. 2002. “The Next Crash.” The New Yorker, November 11. Available
at http://www.newyorker.com/archive/2002/11/11/021111fa fact cassidy.
Cochrane, John H. 1992. “Explaining the Variance of Price-Dividend Ratios.” The Review
of Financial Studies 5 (2): 243–280.
Davidoff, Thomas. 2010. “Supply Elasticity and the Housing Cycle of the 2000s.” Working
Paper. Sauder School of Business, University of British Columbia. Available at http:
//strategy.sauder.ubc.ca/davidoff/shark.pdf.
Davis, Morris, Andreas Lehnert, and Robert Martin. 2008. “The Rent-Price Ratio for the
Aggregate Stock of Owner-Occupied Housing.” Review of Income and Wealth 54 (2):
279–284.
Foote, Christopher L., Kristopher Gerardi, Lorenz Goette, and Paul S. Willen. 2009.
“Reducing Foreclosures: No Easy Answers.” NBER Macroeconomics Annual 22:89–
138.
Gallin, Joshua. 2008. “The Long-Run Relationship Between House Prices and Rents.”
Real Estate Economics 36 (4): 635–658.
Gerardi, Kristopher, Andreas Lehnert, Shane M. Sherlund, and Paul Willen. 2008. “Mak-
ing Sense of the Subprime Crisis.” Brookings Papers on Economic Activity 2008 (Fall):
69–145.
Global Insight and National City Corporation. 2006. “House Prices in America: Valuation
Update for the 2nd Quarter of 2006.” Available at www.globalinsight.com/gcpath/
Q22006report.pdf.

25
Haines, Cabray L., and Richard J. Rosen. 2007. “Bubble, Bubble, Toil, and Trouble?”
FRB Chicago Economic Perspectives. Available at http://www.chicagofed.org/digital
assets/publications/economic perspectives/2007/ep 1qtr2007 part2 haines rosen.pdf.
Haruvy, Ernan, Yaron Lahav, and Charles N. Noussair. 2007. “Traders’ Expectations in
Asset Markets: Experimental Evidence.” American Economic Review 97 (5): 1901–
1920.
Himmelberg, Charles, Christopher Mayer, and Todd Sinai. 2005. “Assessing High House
Prices: Bubbles, Fundamentals, and Misperceptions.” Journal of Economic Perspec-
tives 19 (4): 67–92.
Hussam, Reshmaan N., David Porter, and Vernon L. Smith. 2008. “Thar She Blows: Can
Bubbles Be Rekindled with Experienced Subjects?” American Economic Review 98
(3): 924–937.
Krainer, John, and Chischen Wei. 2004. “House Prices and Fundamental Value.” FRBSF
Economic Letter, no. 2004–27. San Francisco: Federal Reserve Bank of San Francisco.
Available at http://www.frbsf.org/publications/economics/letter/2004/el2004-27.pdf.
Krugman, Paul. 2005. “That Hissing Sound.” New York Times, August 8. Available
at http://www.nytimes.com/2005/08/08/opinion/08krugman.html.
Laing, Jonathan R. 2005. “The Bubble’s New Home.” Barron’s Online, June 20. Available
at http://4sarasotahomes.com/images/6-20-05Barrons-Bubble.pdf.
Leamer, Edward E. 2002. “Bubble Trouble? Your Home Has a P/E Ratio Too.” UCLA
Anderson Forecast, June. Available at http://www.anderson.ucla.edu/faculty/edward.
leamer/pdf files/LeamerForecasts/20022006BubbleTrouble.pdf.
Lo, Andrew W., and A. Craig MacKinlay. 2001. A Non-Random Walk Down Wall Street.
Princeton, N.J.: Princeton University Press.
Mankiw, N. Gregory, and David N. Weil. 1989. “The Baby Boom, the Baby Bust, and the
Housing Market.” Regional Science and Urban Economics 19 (2): 235–258.
Mayer, Christopher. 2003. “[Is There a Bubble in the Housing Market?]. Comments and
Discussion.” Brookings Papers on Economic Activity 2003 (2): 343–354.
McCarthy, Jonathan, and Richard W. Peach. 2004. “Are Home Prices the ‘Next Bubble’ ?”
Economic Policy Review. New York: Federal Reserve Bank of New York. Available at
http://www.newyorkfed.org/research/epr/04v10n3/0412mcca.pdf.
Quigley, John. 2003. “[Is There a Bubble in the Housing Market?]. Comments and Dis-
cussion.” Brookings Papers on Economic Activity 2003 (2): 354–358.

26
Saiz, Albert. 2010. “The Geographic Determinants of Housing Supply.” Working
Paper. Wharton School of Business, University of Pennsylvania. Available at http:
//real.wharton.upenn.edu/∼saiz/GEOGRAPHIC\%20DETERMINANTS.pdf.
Shiller, Robert J. 2006. “Long-Term Perspectives on the Current Boom in Home Prices.”
The Economists’ Voice 3 (4): 1–4. Available at http://www.bepress.com/ev/vol3/iss4/
art4/.
Smith, Margaret Hwang, and Gary Smith. 2006. “Bubble, Bubble, Where’s the Housing
Bubble?” Brookings Papers on Economic Activity 2006 (1): 1–50.
Smith, Vernon L., Gerry L. Suchanek, and Arlington W. Williams. 1988. “Bubbles,
Crashes, and Endogenous Expectations in Experimental Spot Asset Markets.” Econo-
metrica 56 (5): 1119–1151.
Zuckerman, Gregory. 2009. The Greatest Trade Ever: The Behind-the-Scenes Story of
How John Paulson Defied Wall Street and Made Financial History. New York, N.Y.:
Broadway Books.

27
Figure 1. The Housing Price–Rent Ratio in the United States: 1983:Q1–2010:Q1
1.4
1.3
Index (1983q1 = 1.00)
1.1 1
.9 1.2

1985q1 1990q1 1995q1 2000q1 2005q1 2010q1

Sources: National house price series from the Office of Housing Enterprise Oversight (OFHEO). Owner’s
equivalent rent series from the Bureau of Labor Statistics (BLS).

28
Figure 2. House Price Growth and Supply Elasticities in 92 Metropolitan Areas

Panel A: 1995:Q4−2005:Q4
200
10−yr % Change in Prices
0

0 1 2 3 4 5 6
Saiz Elasticity

Panel B: 1999:Q4−2002:Q4
50
3−yr % Change in Prices
20 30 40 10

0 1 2 3 4 5 6
Saiz Elasticity

Panel C: 2002:Q4−2005:Q4
80 100
3−yr % Change in Prices
40 60 20
0

0 1 2 3 4 5 6
Saiz Elasticity

Sources: Supply elasticities are from Saiz (2010). House prices are from the Office of Federal Housing Enterprise Oversight
(OFHEO).
29
Figure 3. House Price Growth and Supply Elasticities in Metropolitan Areas with Saiz
Elasticities below 3

Panel A: 1995:Q4−2005:Q4
San Diego
Oakland
200

Los Angeles Vallejo


San Jose Riverside Stockton
San
Ft. Francisco
Lauderdale
West Palm Beach
Miami
Fresno
Boston Phoenix
Washington
150

New York Tampa Bakersfield


Orlando Providence
Charleston Las Vegas
Jacksonville
Newark
Virginia Beach Baltimore
Seattle Minneapolis
Tucson
100

New Haven Philadelphia


Tacoma Springfield Wilmington
Portland
Chicago Denver
Hartford Richmond
New Orleans Milwaukee Detroit Albany Allentown Ann Arbor
St. Louis
Colorado Springs Atlanta
Baton Rouge HoustonGrand Rapids Austin
Salt Lake City Knoxville Mobile Birmingham
Nashville
Toledo Louisville Columbia
50

Pittsburgh Harrisburg
Scranton Dallas
Albuquerque
Syracuse Cincinnati Columbus
Little
Greenville Rock San Antonio
Cleveland Memphis
Gary Raleigh Akron
YoungstownFort Worth
El Paso
Rochester Buffalo
0

.5 1 1.5 2 2.5 3
Saiz Elasticity

10−yr % Change in Prices Fitted values

Panel B: 1999:Q4−2002:Q4
Vallejo
San Diego
50

Oakland Stockton
Boston Providence
Ft. Lauderdale
40

Miami Washington
Los New York
Angeles
San Francisco
San Jose Riverside
West Palm Beach
Newark Fresno
Minneapolis
Tampa
New Haven
30

Springfield
Denver Philadelphia
OrlandoBaltimore
Jacksonville Hartford Bakersfield
Chicago Wilmington
Charleston Colorado Springs
Albany Austin
20

Virginia BeachMilwaukee Tacoma Phoenix Allentown St. Louis


Ann Arbor
Seattle Detroit Tucson
Las Vegas Atlanta
Richmond
New Orleans Pittsburgh Scranton Dallas Houston
Syracuse Grand Rapids Fort Worth
Toledo Columbia
Cincinnati Youngstown
Portland
Cleveland Knoxville Buffalo Birmingham Louisville Columbus San Antonio
Baton Rouge Mobile Akron Greenville
Gary
Harrisburg El Paso
Nashville Little Rock
Rochester
10

Memphis Raleigh
Albuquerque
Salt Lake City

.5 1 1.5 2 2.5 3
Saiz Elasticity

3−yr % Change in Prices Fitted values

Panel C: 2002:Q4−2005:Q4
100

Bakersfield

Riverside Fresno
Los Angeles West Palm Beach
Las Vegas
80

Ft. Lauderdale
Miami
Phoenix Stockton
Orlando Washington
Virginia Beach Vallejo
San Diego Tampa Baltimore
60

Tucson
Oakland
New York Jacksonville Providence
San Francisco Tacoma
Newark Albany
Philadelphia Wilmington
San Jose Allentown
40

Portland Charleston
New Haven Springfield Richmond
Seattle
Boston
Chicago Hartford
Albuquerque
New Orleans Milwaukee Minneapolis Harrisburg
SyracuseEl St.Paso
Louis
20

Salt Lake City Knoxville ScrantonBaton Rouge San Antonio


Birmingham
Pittsburgh Rochester Colorado Buffalo
Gary Springs Mobile Nashville Columbia Little Rock
Raleigh Louisville
Houston
Ann Arbor
Grand Atlanta
Cincinnati
Rapids Columbus
Cleveland Denver Memphis Toledo Akron Greenville
YoungstownFort Worth Austin
Detroit Dallas
0

.5 1 1.5 2 2.5 3
Saiz Elasticity

3−yr % Change in Prices Fitted values

Sources: Supply elasticities are from Saiz (2010). House prices are from the Office of Federal Housing Enterprise Oversight
(OFHEO).
30
Figure 4. House Price Appreciation and New Construction on the State Level

Panel A: 1995−2005
200
DC
CA
10−yr % Change in House Prices
150

RI MA FL
NH
NJ AZ NV
MD NY MNVA ME
100

CT
VT ORCO WA DE
MI WIPA AK IL MT HI WY
MO GA
KSKY LA SD SC ID
NEND
WV NMIATNAL
OK NC TX
UT
50

OH MS AR
IN
0 .25 .5 .75 1 1.25
10−yr Change in Permits / Initial Population

Panel B: 1999−2002
60

DC
50
3−yr % Change in House Prices

MA NH
RI CA
40

NJ MN
NY ME FL
30

COCT
MD VT VA DE
20

AZ IL HI
MI MO
PAWY WI TXGA NV
KS SCWA OK LA IA
MT SD
MSTNOHOR
NCKY IN AL AR
NE
WV IDND
AK
10

NM
UT
0

−.1 −.05 0 .05 .1 .15 .2


3−yr Change in Permits / Initial Population

Panel C: 2002−2005
80

NV
CA HI
DC
3−yr % Change in House Prices

FL
60

MD
RI AZ
VA NJ
NY DE
40

VT CT ME
NH PA MAMT WA OR
AK WY ID
MN IL NM
WI WV ND
20

MO
LASD AR SC
KYIA
NE
KS MS GAOK TN
NC AL UT
CO
MI
OHIN TX
0

−.1 0 .1 .2 .3 .4 .5 .6 .7
3−yr Change in Permits / Initial Population

Sources: House prices are from OFHEO. State-level building permits and initial population levels are from the Bureau of the
Census.

31

You might also like