Description
This paper aims to examine the anatomy of a real estate bubble. In the process, the paper
identifies three phases of the market’s evolution: flips, flops and foreclosures. An examination of the
Las Vegas real estate market illustrates the three phases.
Journal of Financial Economic Policy
Flips, flops and foreclosures: anatomy of a real estate bubble
Craig A. Depken II Harris Hollans Steve Swidler
Article information:
To cite this document:
Craig A. Depken II Harris Hollans Steve Swidler, (2011),"Flips, flops and foreclosures: anatomy of a real
estate bubble", J ournal of Financial Economic Policy, Vol. 3 Iss 1 pp. 49 - 65
Permanent link to this document:http://dx.doi.org/10.1108/17576381111116759
Downloaded on: 24 January 2016, At: 21:40 (PT)
References: this document contains references to 16 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 434 times since 2011*
Users who downloaded this article also downloaded:
Dag Einar Sommervoll, Gavin Wood, (2011),"Home equity insurance", J ournal of Financial Economic
Policy, Vol. 3 Iss 1 pp. 66-85http://dx.doi.org/10.1108/17576381111116768
Beverley Searle, (2011),"Recession and housing wealth", J ournal of Financial Economic Policy, Vol. 3 Iss 1
pp. 33-48http://dx.doi.org/10.1108/17576381111116867
Steve Swidler, (2011),"Homeownership: yesterday, today and tomorrow", J ournal of Financial Economic
Policy, Vol. 3 Iss 1 pp. 5-11http://dx.doi.org/10.1108/17576381111116786
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Flips, ?ops and foreclosures:
anatomy of a real estate bubble
Craig A. Depken II
Department of Economics, Belk College of Business, UNC-Charlotte, Charlotte,
North Carolina, USA, and
Harris Hollans and Steve Swidler
Department of Finance, Auburn University, Auburn, Alabama, USA
Abstract
Purpose – This paper aims to examine the anatomy of a real estate bubble. In the process, the paper
identi?es three phases of the market’s evolution: ?ips, ?ops and foreclosures. An examination of the
Las Vegas real estate market illustrates the three phases.
Design/methodology/approach – The paper examines transaction data from the metropolitan
Las Vegas area (Clark County) from1994 to 2009. The ?rst part of the analysis identi?es the three phases
of the bubble and is descriptive in nature. This is followed by more formal tests of Granger causality.
Findings – In the early part of the sample, a large percentage of transactions are speculative or
“?ips” causing prices to rapidly increase. Eventually, ?ipping loses its pro?tability and over the last
three years, there is an increasing number of foreclosures leading to falling prices. The descriptive
analysis of the Las Vegas market is augmented with causality tests which show that prices were the
driving force behind all three phases in the market’s evolution.
Research limitations/implications – Future research might focus on underlying structural
inter-temporal relationships to augment the Granger causality tests.
Practical implications – Analysis shows that price is the driving force behind a bubble and that
loan modi?cation programs alone will not solve the current housing crisis.
Social implications – Government entities might expand neighborhood stabilization programs to
affect both demand and supply of homes. Moreover, it might be prudent to include information related
to ?ipping on multiple listing service agreements. Additionally, local governments should be
consistent in their record keeping.
Originality/value – To the best of the authors’ knowledge, this is the ?rst paper to examine the
housing bubble using an extensive set of transaction data.
Keywords Mortgage companies, Mortgage default, Real estate, United States of America
Paper type Research paper
1. Introduction
Conventional wisdom in the USA blames the housing market as the “?rst domino” that
fell in the lead-up to the recession that began in early 2007 (Shiller, 2009). Some have
alluded to a housing bubble that was unsustainable and which caused individuals to
have a false perception that housingprices would continue to increase, therebymakingit
pro?table to purchase more expensive homes or to speculate on residential real estate in
so-called “?ipping” (Wheaton and Nechayev, 2008). Indeed, the popularity of ?ipping
might be re?ected in popular culture in which television shows such as “Flip this House”
(A&E network) and “Flip that House” (TLC) were amongst the most popular television
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1757-6385.htm
JEL classi?cation – G11, G21, R31
Real estate
bubble
49
Journal of Financial Economic Policy
Vol. 3 No. 1, 2011
pp. 49-65
qEmerald Group Publishing Limited
1757-6385
DOI 10.1108/17576381111116759
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
shows in the early 2000s. As markets have cooled off or even collapsed, these shows have
since been replaced by less bullish shows such as late night television “infomercials”
selling foreclosed residential condominiums.
This paper examines the anatomy of a real estate bubble. In the process, we identify
three phases of the market’s evolution: in the ?rst phase, a large percentage of
transactions are speculative or “?ips” causing prices to rapidly increase; in phase 2,
?ipping loses its pro?tability and many individuals are caught “holding the bag,”
(i.e. cannot resell their house at a higher highprice); andinphase 3, there are anincreasing
number of foreclosures leading to falling prices. Eventually, properties held by banks
(shadowinventory) must be soldor destroyedbefore the market canrecover andstabilize.
To illustrate a real estate bubble, we investigate the evolution of the Las Vegas
metropolitan housing market from 1994 to 2009. We begin with positive economic
analysis that is mainly descriptive in nature and graphically captures the three phases of
the Las Vegas real estate bubble. Our subsequent analysis formally investigates the
extent to which ?ips, foreclosures and percentage change in price are related to each
other. Granger causality tests imply that percentage change in price is the driving force
behind ?ipping and foreclosure activity, but that ?ips and foreclosures are not directly
related to each other. Finally, normative analysis of a real estate bubble suggests a
number of policy propositions that might be considered by lawmakers and real estate
professionals.
While local housing markets follow idiosyncratic cycles, price trends include a
systematic component related to certain factors. For instance, the literature has
established that housing prices tend to increase as the local population increases, as new
housing stock replaces older homes, as local incomes increase, and as the supply of
developable land decreases (Brueckner, 1980; Capozza and Helsley, 1989). On the other
hand, (moderate) recessions are not necessarily associated with a fall in housing prices
but rather with a reduction in the number of houses sold in any particular time period
and an increase in the time-on-market for existing houses. In other words, house prices
tend to be sticky downward. Nevertheless, a suf?ciently severe recession might induce
price decreases. For instance, Case and Quigley (2008) ?nd that, in the last half of 2006,
sales activity slowed, but housing prices in Boston declined only moderately at the
beginning of the downturn. However, as the recession continued to deepen, Boston
housing prices at the end of 2009 were approximately 17 percent belowtheir peak, falling
to their 2003 levels (as measured by the Case-Shiller Index).
Although stable and increasing housing prices are frequently thought of as the norm,
it is possible for local housing markets to experience dramatic increases (and decreases)
in price. Whether such price volatility is generated by arti?cially restrained supply, for
instance, through overly restrictive zoning or land-use policies, or through arti?cially
enhanced demand, rapidly escalating prices might induce individuals to speculate on
residential properties in the formof “?ipping.” Flipping entails purchasing a residential
property, perhaps improving the property through cosmetic or structural changes, and
attempting to rapidly resell the property for a pro?t. House ?ipping contributes to an
increase in the demand for existing properties, thereby pushing up price. However,
house ?ipping might also be a rational response to other market signals such as a rapidly
increasing population or relatively easy credit for potential home buyers (Wheaton and
Nechayev, 2008). Estimating the relative in?uence of these possible factors is mainly an
empirical exercise.
JFEP
3,1
50
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Factors that lead to a dramatic escalation in housing prices cannot be expected to last
forever. Thus, a tapering-off period follows during which price increases moderate, the
pro?ts from“?ipping” fall, andthere is a decline inthe proportion of sales that are “?ips.”
This reduced exuberance might presage an actual, and potentially dramatic, decline in
price. If this is the case, those who attempted to participate in “?ipping” toward the end
of the exuberant period and many who purchased at the peak of the market will ?nd
themselves holding a depreciating asset. In this environment, the “?ipping” period is
followed by a period of “?ops” and ?nally a potential for foreclosures as some
individuals ?nd it in their best interest to default on their mortgage rather than trying to
sell the property for a loss. The subsequent analysis chronicles a cycle of ?ips, ?ops and
foreclosures in Clark County, NV, a district that essentially comprises the Las Vegas
metropolitan area.
2. Data and de?nitions of transaction type
The data sample used in this study describes 541,373 separate residential property
transactions from Clark County, NV from 1994:q1 to 2009:q4, obtained from the Clark
County tax assessor’s of?ce. The data describe, among other things, the transaction
price, the transaction type, the date of the transaction and a unique parcel identi?er.
We are able to examine up to nine separate transactions for each parcel, although the
vast majority of properties have less than four transactions during the sample period.
To facilitate the use of such a large data set and to provide a level of aggregation that
might inform policy discussion, the analysis translates each transaction date into the
appropriate quarter and year. The subsequent analysis is then undertaken on a
quarterly basis.
Each transfer of property is coded by the Clark County tax assessor’s of?ce according
to the transaction (sale) type. There are three categories that we examine:
(1) Recorded value. Denoting an arms-length transaction (coded with an “R”).
(2) Trustee’s deed. Trustee’s deed is the amount bid at foreclosure auction on the
trustee’s deed (coded with a “T”).
(3) Foreclosure. Foreclosure is a transfer indicating a resale after foreclosure (coded
with an “F”).
These three sale types constitute the bulk of all transactions ?led at the tax assessor’s
of?ce and are the most important categories for the purposes of our investigation.
Table I lists the distribution of residential transactions by sales type. For the entire
sample period, there are 464,093 “R” transactions, 47,320 “T” transactions and29,960 “F”
transactions recorded. Examining tax records on a quarterly basis, total transactions
trend upwards and reach a peak of more than 15,000 sales in the third quarter of 2005.
On a percentage basis, arms-length transactions (R) constitute more than 95 percent of
all sales through the fourth quarter of 2006. House prices in Las Vegas (as measured by
the Case-Shiller Index) re?ected the vigorous sales activity of this period, and after large
run-ups in 2004 and 2005, prices eventually crested in the second quarter of 2006.
To put the ?gures in perspective, Figure 1 shows the number of R, T and
F transactions for the period 2004:q1-2009:q4. As can be seen, arms-length transactions
dominate the distribution up until the end of 2006. In 2007, the F and T transactions
begin to increase in number and eventually foreclosure activity constitutes the leading
share of sales. The properties in foreclosure combined with additional new and used
Real estate
bubble
51
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Quarter Total R Total T Total F T(%) F(%) R(%)
1994q1 3,514 102 3 2.82 0.08 97.1
1994q2 4,422 77 4 1.71 0.09 98.2
1994q3 4,116 79 5 1.88 0.12 98
1994q4 3,868 83 4 2.1 0.1 97.8
1995q1 3,202 85 1 2.59 0.03 97.38
1995q2 3,935 72 7 1.79 0.17 98.04
1995q3 4,134 82 3 1.94 0.07 97.99
1995q4 3,977 84 4 2.07 0.1 97.83
1996q1 4,120 85 2 2.02 0.05 97.93
1996q2 4,730 68 5 1.42 0.1 98.48
1996q3 4,593 92 2 1.96 0.04 98
1996q4 4,755 92 7 1.9 0.14 97.96
1997q1 4,064 140 4 3.33 0.1 96.57
1997q2 4,867 116 20 2.32 0.4 97.28
1997q3 5,049 155 4 2.98 0.08 96.94
1997q4 4,923 146 8 2.88 0.16 96.96
1998q1 4,469 178 6 3.83 0.13 96.04
1998q2 5,669 197 4 3.36 0.07 96.57
1998q3 5,645 236 7 4.01 0.12 95.87
1998q4 5,727 231 3 3.88 0.05 96.07
1999q1 5,406 263 8 4.63 0.14 95.23
1999q2 6,656 253 3 3.66 0.04 96.3
1999q3 6,453 282 8 4.18 0.12 95.7
1999q4 6,048 233 6 3.71 0.1 96.19
2000q1 5,654 288 5 4.84 0.08 95.08
2000q2 6,850 254 9 3.57 0.13 96.3
2000q3 6,586 319 9 4.61 0.13 95.26
2000q4 6,756 276 6 3.92 0.09 95.99
2001q1 6,600 328 6 4.73 0.09 95.18
2001q2 8,134 300 5 3.55 0.06 96.39
2001q3 7,668 309 5 3.87 0.06 96.07
2001q4 8,086 356 9 4.21 0.11 95.68
2002q1 7,537 364 8 4.6 0.1 95.3
2002q2 8,724 447 7 4.87 0.08 95.05
2002q3 8,884 431 16 4.62 0.17 95.21
2002q4 9,180 392 6 4.09 0.06 95.85
2003q1 8,619 450 8 4.96 0.09 94.95
2003q2 10,762 466 10 4.15 0.09 95.76
2003q3 12,537 461 10 3.54 0.08 96.38
2003q4 12,354 363 7 2.85 0.06 97.09
2004q1 12,513 342 6 2.66 0.05 97.29
2004q2 15,432 99 14 0.64 0.09 99.27
2004q3 14,635 85 13 0.58 0.09 99.33
2004q4 12,907 39 8 0.3 0.06 99.64
2005q1 12,360 34 11 0.27 0.09 99.64
2005q2 14,970 25 3 0.17 0.02 99.81
2005q3 15,453 21 2 0.14 0.01 99.85
2005q4 14,302 42 6 0.29 0.04 99.67
2006q1 12,414 74 10 0.59 0.08 99.33
2006q2 13,267 128 3 0.96 0.02 99.02
(continued)
Table I.
Temporal distribution of
residential transaction
types in Las Vegas, NV
JFEP
3,1
52
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
properties on the market might be expected to exert downward pressure on price and
alter the expectations of potential buyers about future price changes. In fact, the last
three years of the sample period witnessed rapid price declines in the Las Vegas market.
The dispersionof foreclosure activityis not evenlydistributedacross the 104 different
tax districts of Clark County. Table II depicts the R, T and F transactions for the 21 tax
districts that constitute our sample and include all areas with more than 1,000 residential
transactions during the period of analysis. Of particular note is that the districts with the
largest foreclosure activity tend to have the lowest per capita income in the area.
In particular, foreclosures were more than 17 percent of total sales in the low-income
districts of North Las Vegas, Sunrise Manor and Whitney.
Figure 1.
Distribution of residential
property transactions by
type and quarter
(2004:q1-2009:q4)
2004q1
0
5,000
10,000
N
u
m
b
e
r
o
f
t
r
a
n
s
a
c
t
i
o
n
s
15,000
2005q3 2007q1 2008q3 2010q1
Date
Arms-length transactions (R)
Trustee's deeds (T)
Foreclosures (F)
Quarter Total R Total T Total F T(%) F(%) R(%)
2006q3 11,741 238 10 1.99 0.08 97.93
2006q4 10,511 314 18 2.9 0.17 96.93
2007q1 7,801 646 29 7.62 0.34 92.04
2007q2 7,451 980 76 11.52 0.89 87.59
2007q3 6,103 1,432 154 18.62 2 79.38
2007q4 5,521 2,231 213 28.01 2.67 69.32
2008q1 3,515 3,060 1,607 37.4 19.64 42.96
2008q2 3,828 4,586 3,095 39.85 26.89 33.26
2008q3 4,046 5,108 4,244 38.13 31.68 30.19
2008q4 3,755 4,473 3,861 37 31.94 31.06
2009q1 2,787 4,319 4,127 38.45 36.74 24.81
2009q2 3,461 3,591 5,643 28.29 44.45 27.26
2009q3 4,205 4,431 4,914 32.7 36.27 31.03
2009q4 1,842 1,787 1,639 33.92 31.11 34.97
Note: Quarterly data from 2009:q4 is not from the entire quarter Table I.
Real estate
bubble
53
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
T
a
x
d
i
s
t
r
i
c
t
n
u
m
b
e
r
T
a
x
d
i
s
t
r
i
c
t
n
a
m
e
T
o
t
a
l
t
r
a
n
s
a
c
t
i
o
n
s
R
t
r
a
n
s
a
c
t
i
o
n
s
T
t
r
a
n
s
a
c
t
i
o
n
s
F
t
r
a
n
s
a
c
t
i
o
n
s
T
a
n
d
F
t
r
a
n
s
a
c
t
i
o
n
s
(
%
)
5
0
B
o
u
l
d
e
r
C
i
t
y
L
i
b
r
a
r
y
2
,
5
5
6
2
,
4
3
7
8
2
3
7
4
.
6
6
1
0
7
L
a
u
g
h
l
i
n
T
o
w
n
B
i
g
B
e
n
d
C
o
l
o
r
a
d
o
R
i
v
e
r
1
,
1
5
4
1
,
0
6
0
7
1
2
3
8
.
1
5
1
2
5
A
r
t
e
s
i
a
n
B
a
s
i
n
F
i
r
e
9
1
1
1
,
1
9
8
9
9
5
1
1
4
8
9
1
6
.
9
4
2
0
0
L
a
s
V
e
g
a
s
C
i
t
y
1
6
4
,
1
2
3
1
4
1
,
0
6
3
1
4
,
5
2
3
8
,
5
3
7
1
4
.
0
5
2
0
3
L
a
s
V
e
g
a
s
C
i
t
y
R
e
d
e
v
e
l
o
p
m
e
n
t
1
,
1
1
8
9
3
7
1
2
5
5
6
1
6
.
1
9
2
5
0
N
o
r
t
h
L
a
s
V
e
g
a
s
C
i
t
y
5
2
,
1
6
9
4
3
,
0
5
5
5
,
6
5
2
3
,
4
6
2
1
7
.
4
7
2
5
4
N
o
r
t
h
L
a
s
V
e
g
a
s
L
i
b
r
a
r
y
3
1
,
8
1
5
2
5
,
9
1
1
3
,
3
9
9
2
,
5
0
5
1
8
.
5
6
3
4
0
S
u
n
r
i
s
e
M
a
n
o
r
4
3
,
0
0
9
3
5
,
2
5
1
4
,
9
7
8
2
,
7
8
0
1
8
.
0
4
4
1
0
W
i
n
c
h
e
s
t
e
r
T
o
w
n
2
,
1
6
9
1
,
8
8
5
1
9
3
9
1
1
3
.
0
9
4
1
7
S
p
r
i
n
g
V
a
l
l
e
y
T
o
w
n
4
6
,
2
8
5
4
0
,
2
6
2
3
,
5
4
1
2
,
4
8
2
1
3
.
0
1
4
2
0
S
u
m
m
e
r
l
i
n
T
o
w
n
A
r
t
e
s
i
a
n
B
a
s
i
n
1
0
,
4
6
8
9
,
6
2
3
4
7
8
3
6
7
8
.
0
7
4
7
0
P
a
r
a
d
i
s
e
T
o
w
n
2
6
,
2
4
9
2
3
,
0
1
5
2
,
1
1
0
1
,
1
2
4
1
2
.
3
2
5
0
0
H
e
n
d
e
r
s
o
n
C
i
t
y
3
,
3
9
0
3
,
0
2
7
2
2
2
1
4
1
1
0
.
7
1
5
0
3
H
e
n
d
e
r
s
o
n
C
i
t
y
R
e
d
e
v
e
l
o
p
m
e
n
t
1
,
8
3
9
1
,
5
5
6
1
8
2
1
0
1
1
5
.
3
9
5
0
5
H
e
n
d
e
r
s
o
n
A
r
t
e
s
i
a
n
B
a
s
i
n
4
9
,
8
3
2
4
5
,
2
5
5
2
,
8
3
8
1
,
7
3
9
9
.
1
8
5
1
4
H
e
n
d
e
r
s
o
n
C
i
t
y
L
i
b
r
a
r
y
D
e
b
t
4
,
1
4
4
3
,
8
7
8
1
5
3
1
1
3
6
.
4
2
5
1
6
H
e
n
d
e
r
s
o
n
L
i
b
r
a
r
y
D
e
b
t
/
A
r
t
e
s
i
a
n
B
a
s
i
n
1
8
,
8
6
8
1
7
,
1
3
7
1
,
0
2
0
7
1
1
9
.
1
7
5
2
1
H
e
n
d
e
r
s
o
n
C
i
t
y
R
e
d
e
v
e
l
o
p
m
e
n
t
5
2
1
4
,
5
2
2
3
,
7
6
5
5
1
4
2
4
3
1
6
.
7
4
5
7
0
W
h
i
t
n
e
y
A
r
t
e
s
i
a
n
B
a
s
i
n
1
2
,
4
1
9
1
0
,
1
9
6
1
,
3
4
1
8
8
2
1
7
.
9
0
6
3
5
E
n
t
e
r
p
r
i
s
e
F
i
r
e
A
r
t
e
s
i
a
n
L
i
b
r
a
r
y
9
1
1
M
a
n
p
o
w
e
r
5
9
,
3
7
9
4
9
,
3
5
5
5
,
6
2
6
4
,
3
9
8
1
6
.
8
8
9
0
1
M
e
s
q
u
i
t
e
C
i
t
y
4
,
6
6
7
4
,
4
3
0
1
5
8
7
9
5
.
0
8
T
o
t
a
l
s
5
4
1
,
3
7
3
4
6
4
,
0
9
3
4
7
,
3
2
0
2
9
,
9
6
0
Table II.
Tax districts and the
frequency of transaction
types (full sample)
JFEP
3,1
54
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
It is important that the county assessor accurately characterize each transaction as the
data serves as a foundation for mass-appraisal models used to determine “fair market
value” for ad valorem tax purposes. As such, arms-length transactions denoted as
R transactions in the data serve as the benchmark for market value estimation.
Frequently, arms-length transactions are thought of as a sale between a willing seller and
a willing buyer.
Atrustee’s deed transaction (coded T) denotes a foreclosure sale and signi?es that the
property either resides in the real estate-owned (REO) inventory of the lender or was
purchased at the foreclosure sale by an owner/investor. Typically, the lender is the
winning bidder at auction and the recorded sale price of the Ttransaction represents the
amount bid on the trustee’s deed. A T transaction may also represent a deed-in-lieu of
foreclosure, which entails the lender repossessing the house without pursuing a
foreclosure on the property, with the result that the homeowner loses whatever equity
they have in the house. Presumably there is little or, more likely, negative equity
precipitating the transference of the property. In a deed-in-lieu of foreclosure, the lender
often agrees to not pursue the individual home owner for recourse, which has arguably
become easier under the recently passed rules of the Trouble Asset Relief Program and
the American Recovery and Reinvestment Act.
Early in the sample period, an F code denoted a deed-in-lieu of foreclosure
transaction. With the recent increase in foreclosure activity, the county changed an
F transaction to mean that the transfer of a property is a resale after foreclosure. The
typical example of a recently coded F transaction would be the sale of the house by the
lender (who acquired the trustee’s deed through a T transaction) to a new homeowner.
If the sale price is thought to be different from market value, the county then codes this
as an F transaction.
Table III gives the ?avor of foreclosure activityinClarkCounty andhowthe codingof
deed-in-lieu of foreclosure changed over the sample period. In the early part of the
sample (1994:q1-2006:q4), homes going into foreclosure were predominantly coded
Ttransactions that signaled transference of a trustee’s deed. In a typical quarter, nearly
161 homes were T transactions, while ?ve were coded F by the county and primarily
denoted a deed-in-lieu of foreclosure sale. Looking at the last three columns of Table III,
whether designated a Tor an Ffor homes going into foreclosure, virtually all subsequent
sales were coded R by the county. This implies that almost all houses were sold to the
new homeowners at “market value,”, i.e. the county considered the transaction between
the lender and new homeowner as an arms-length sale.
Over the more recent sample period (2007:q1-2009:q4), Table III illustrates two
important changes that occurred. First, there were virtually no examples of a T(trustee’s
deed) followed by an F (deed-in-lieu of foreclosure) in the early sample period. However,
Tthen F becomes the predominant foreclosure sequence of transactions moving into the
latter sample period and implies that for most of 2007-2009, Ttransactions included both
trustee’s deed and deed-in-lieu of foreclosure transactions. Moreover, an F transaction in
2007-2009 refers to a resale after foreclosure. In the third column, the few100 examples of
F followed by R found in 2007-2009 presumably denote properties that sold in an earlier
period as a deed-in-lieu of foreclosure, but then were coded as an arms-length transaction
on the subsequent sale.
Of potentially more interest is the second change made in recent years. Whereas it has
already been noted that from 1994 to 2006 virtually all second sales in the foreclosure
Real estate
bubble
55
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
sequence were considered R transactions, by 2009 more than 90 percent of second sales
were coded F by the county. This has important implications for housing valuation.
In recent years, the county no longer considered the vast majority of the foreclosure
resales as an arms-length transaction, and therefore, they would not likely be used in any
tax valuation model or price index.
3. Property ?ipping
As documented earlier, the rapid rise in housing prices from 2003 to 2005 corresponded
to a period that experienced a high number of sales. The county designated virtually all
transactions as arms length (R). Many of these sales involved property ?ipping.
Depken et al. (2009) de?ne a ?ip as the purchase of a home with the intent of quickly
reselling the property at a higher price. They examine all cases that involve two
arms-length transactions for a property within a two-year window. Two years is a
relevant time frame as the Internal Revenue Service allows any capital gains to be
excluded from taxable income if the seller has used the home as his or her primary
residence for two of the previous ?ve years. This de?nition does not depend upon
?ipping motivation or economic pro?tability, but rather focuses on the short-term
investment horizon that is the epitome of house ?ipping.
House ?ipping is often depicted as purchasing property in poor condition at a
discount, renovating the house and then selling it at or near full market value. This is
sometimes referred to as a “?x and ?ip” and has been the basis for a number of reality
television shows. However, this is not the only situation in which a property might be
ripe for “?ipping.” Some properties might be purchased at a discount due to forced
circumstances such as relocation, divorce, or a pending foreclosure. Such situations
might provide for a nominally pro?table opportunity if the house can be resold at market
value in a relatively short amount of time.
Quarter
T
followed
by an R
T
followed
by an F
F
followed
by an R
F
followed
by a T
F or T
followed by
an R
F or T
followed by
an F
F or T
followed
by a T
Quarterly
average 1994:q1-
2006:q4 160.8 0.2 5.0 0.3 165.7 0.2 2.6
2007:q1 174 14 6 1 180 14 11
2007:q2 288 33 4 2 292 33 19
2007:q3 268 86 3 1 271 86 13
2007:q4 370 119 3 2 373 120 16
2008:q1 96 1,380 20 5 116 1,382 30
2008:q2 85 2,772 29 15 114 2,779 65
2008:q3 114 3,892 85 7 199 3,905 43
2008:q4 116 3,697 97 10 213 3,700 58
2009:q1 126 3,896 95 10 221 3,902 31
2009:q2 79 5,148 160 12 239 5,158 53
2009:q3 95 4,619 208 20 303 4,626 65
2009:q4 76 1,571 117 9 193 1,571 42
Totals 1994:q1-
2009:q4 10,246 27,237 1,086 107 11,332 27,287 581
Note: Quarterly data from 2009:q4 is not from the entire quarter
Table III.
Sales around foreclosures
JFEP
3,1
56
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Givenmultiple transactions for a givenproperty, it is possible toidentify, expost, which
transactions are the front-endof aneventual house ?ip, calledthe buy-side ?iptransaction,
and which are the back-end of an eventual house ?ip, called the sell-side ?ip transaction.
This, in turn, allows for the calculation of any economic pro?ts due to ?ipping. House
?ipping, similar to any speculative activity, is an inherently risky proposition, and higher
risksuggests higher expectedreturn. Depkenet al. (2009) ?nd that ?ippers earned positive
economic pro?ts in the Las Vegas residential housing market up through 2005, in large
part, re?ecting sell-side ?ip prices that tended to be higher than other similar arms-length
transactions.
The premiumon the sell-side transaction might indicate improvement to the property
not captured in the tax records. However, the housing stock in Las Vegas is relatively
young, and many ?ips were of new or nearly new houses. It is unlikely that sell-side
premiums were primarily a re?ection of home improvements.
Alternatively, a sales premium might obtain due to the circumstances of ?ipping.
If information is costly to obtain, ?ippers may have superior knowledge of the local real
estate market. Since there is no pressure to relocate, ?ippers can take time to search for
buyers offering the highest price for the property. Still another possibility due to
asymmetric information is that sell-side ?ips might carry a premium because of illegal
activity. For instance, appraisers might collude with mortgage originators and the
?ipper’s broker to provide an in?ated appraisal of the ?ipped property. In?ated prices
might then be the result of mortgage companies arti?cially stimulating demand by
qualifying buyers for more expensive homes. To limit “predatory ?ipping,” the
Department of Housing and Urban Development promulgated a set of guidelines in 2004
(amendedin2006) that prohibitedFederal HousingAuthority-insuredmortgage ?nancing
for properties re-sold within 90 days. These guidelines, however, did little to stop
mortgage fraud that has been the focus of, “Operation Stolen Dreams,” an FBI campaign
that targets illegal ?ipping activity, loan origination schemes, and equity skimming
(www.fbi.gov/page2/june10/mortgage_061710.html).
To get a better idea of Las Vegas ?ipping activity during the sample period, Figure 2
shows quarterlysell- andbuy-side ?iptransactions as a percentage of all arms-lengthsales.
Figure 2.
Percent buy-side and
sell-side ?ip transactions
(1994:q1-2009:q4)
1995q1
0.25
0.2
0.15
0.1
0.05
0
2000q1
P
e
r
c
e
n
t
o
f
a
l
l
t
r
a
n
s
a
c
t
i
o
n
s
2005q1
Date
Percent sell-side flip transactions
Percent buy-side flip transactions
2010q1
Real estate
bubble
57
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
As can be seen, buy- and sell-side transactions comprise a small percentage of all
transactions until sometime after 2002 when the buy-side ?ip transactions break the
10 percent barrier. Eventually, the buy-side ?ip transactions peaks at approximately
23 percent of all arms-length transactions in the ?rst quarter of 2004 after which the
percentage drops dramatically. The falling number of buy-side ?ip transactions coincides
with the declining pro?tably of ?ipping and the eventual fall in housing prices
(Depken et al., 2009).
Figure 2 also shows a pattern of seasonality in the ?ip transactions data. Over the
entire sample period, there is no statistically meaningful difference between the
percentage of buy-side transactions in quarters 1, 2 or 3. However, there is a statistically
signi?cant decline in the percentage of buy-side transactions in the fourth quarter (by
approximately 3.8 percent between q1 and q4, p-value ¼ 0.032). There exists a different
but distinct pattern in sell-side ?ip transactions, as well: there is a positive and slightly
signi?cant increase in sell-side ?ip transactions in quarter 4 (by approximately
2.7 percent between q1 and q4, p-value ¼ 0.098) although there is no statistically
meaningful difference between quarter 1 and quarter 2 or 3.
When focus is restricted to the period starting with 2004:q1, the seasonal patterns
disappear for both buy- and sell-side ?ip transactions. As can be seen in Figure 3, the
percentage of buy-side ?ip transactions falls below the percentage of sell-side ?ip
transactions by the third quarter of 2004, that is, buy-side ?ip transactions fall before the
Las Vegas market experienced dramatic decreases in prices and increases in foreclosure
activity. Moreover, Figure 3 shows that ?ipping activity never actually reached zero;
even during the dramatic downward adjustment of the market after 2006, there were still
some properties that speculators felt were potentially pro?table “?ips.”
4. Descriptive analysis of the Las Vegas housing bubble
Figure 4 shows an evolution of ?ips, foreclosures, housing prices and price changes in
Clark County over the entire sample period. The bar graph above the x-axis
Figure 3.
Percent buy-side and
sell-side ?ip transactions
(2004:q1-2009:q4)
0.25
0.2
0.15
0.1
0.05
0
2004q1 2005q3 2007q1 2008q3 2010q1
Date
P
e
r
c
e
n
t
o
f
a
l
l
t
r
a
n
s
a
c
t
i
o
n
s
Percent sell-side flip transactions
Note: Buy-side flips in 2008 and 2009 do not cover entire two year flipping window
Percent buy-side flip transactions
JFEP
3,1
58
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
represents the percentage of all transactions identi?ed ex post as one side of a ?ip
(both buy- and sell-side each quarter). The bar graph belowthe horizontal axis illustrates
the number of trustee deeds (T) and foreclosures (F) as a percentage of total transactions.
The percentage of ?ips and foreclosures are then drawn against the median price of all
arms-length transactions (solid line) and the annualized percentage change in median
price for each quarter (dashed line). In chronicling the boomto bust housing cycle in the
Las Vegas area, the graph can be neatly divided into three distinct regimes.
Flips – 1994:q1 through 2005:q4
Between 1994 and 2000, the great majority of sales were arms-length transactions, with
relatively few ?ips or foreclosures. During this period, quarterly price changes were
small (approximating the nominal in?ation rate), and the median transaction price in the
Las Vegas market remained relatively modest. After 2000, the percentage change in
prices increased and median prices followed accordingly. Seemingly in response to these
price changes, ?ipping activity also increased.
By 2004, ?ipping activity rose to roughly 40 percent of all housing transactions in
Las Vegas, and while perhaps not clear ex ante, ?ipping activity of this magnitude could
not be sustained forever. The arti?cial stimulus in demand from?ipping, in turn, helped
fuel anincrease in newhomes built. The housingstocknearly doubled in Las Vegas from
2000 to 2008; however, Clark County’s population grew only 33.9 percent (from 1.39 to
1.87 millionpeople) during the same period. Flippers, home builders and (non-?ip) sellers
eventually found it more dif?cult to ?nd buyers leading to the next stage of the bubble.
Flops – 2006:q1 through 2007:q4
Figure 5 replicates the previous graph, but centers on the transition period between ?ips
and foreclosures. As ?ippers found it more dif?cult to locate buyers for their properties,
price increases attenuated and prices eventually peaked in 2006. In many cases,
Figure 4.
Percentage of transactions
that were ?ips, percentage
of transactions
foreclosures, quarterly
median price and
percentage change in
median price
(1994:q1-2009:q4)
1
0.50
0
0.50
1
P
e
r
c
e
n
t
/
P
c
t
.
c
h
a
n
g
e
1994q1
Percent flips Percent foreclosures
Median price Pct. change med. price
1996q1 1998q1 2000q1 2002q1 2004q1 2006q1 2008q1 2010q1
1,00,000
1,50,000
M
e
d
p
r
i
c
e
M
e
d
i
a
n
p
r
i
c
e
2,00,000
2,50,000
3,00,000
Real estate
bubble
59
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
?ips were no longer pro?table leading to economic losses and a fall in ?ipping activity.
A reduction in ?ips occurred because fewer investors decided to buy properties for
?ipping, and those that did could not always sell them within two years. For many
investors, 2006-2007 was a period of ?ops, and by 2008, Figure 5 shows that ?ipping
activity had fallen to less than 5 percent of all transactions. This contributed to further
price declines, and starting in 2008, foreclosures constituted the majority of transactions
in the Las Vegas area.
Foreclosures – 2008:q1 through 2009:q4
In the ?nal phase, ?ipping was not economically pro?table as median prices continued
to decline, potential buyers were reluctant to purchase a home in anticipation of lower
future prices and more existing home owners found themselves underwater on their
mortgage. Coupled with an increasingly soft labor market and higher unemployment,
the number of foreclosures “snowballed” despite policy interventions such as mortgage
restructuring, ?rst-time homebuyer tax credits and the Fed’s purchase of
mortgage-backed securities leading to lower interest rates.
One more effect of the high foreclosure rate is a sharp increase in shadow inventory.
Figure 6 shows the stock of homes owned by the lender, where quarterly changes equal
the net ?ows of properties going into foreclosure minus REO inventory that has been
resold. During 2004, with prices increasing rapidly, REOinventory was being sold faster
than the (small) number of new foreclosures and shadow inventory fell to its local
minimum. However, as house prices began to fall in 2007, the negative equity position of
owners caused foreclosures to escalate. Initially, newforeclosures exceeded REOresales
and shadow inventory reached its peak at the beginning of 2009. In a situation where
asset prices are falling dramatically, lenders were trying to sell properties as quickly as
possible. Given the brisk turnover of REO property along with shadow inventory near
Figure 5.
Percentage of transactions
that were ?ips, percentage
of transactions
foreclosures, quarterly
median price and
percentage change in
median price
(2004:q1-2009:q4)
1
0.50
0
0.50
1
P
e
r
c
e
n
t
/
P
c
t
.
c
h
a
n
g
e
2004q1 2006q1 2008q1 2010q1
1,50,000
2,00,000
2,50,000
3,00,000
3,50,000
M
e
d
p
r
i
c
e
M
e
d
i
a
n
p
r
i
c
e
Percent flips Percent foreclosures
Median price Pct. change med. price
JFEP
3,1
60
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
historically high levels, there is further reason to believe that housing prices will
continue to decline in a soft Las Vegas housing market.
5. Establishing causal relationships
The fan graphs discussed in the previous section are suggestive of intertemporal
relationships between quarterly foreclosures (including deed-in-lieu of foreclosure),
median price, the quarterly percentage change in median price and quarterly ?ipping
activity (both buy- and sell-side transactions). To establish whether intertemporal
relationships actually exist, we consider Granger causality tests between any two
variables of interest. We do not consider the level of median price explicitly as it is
subsumed within the percentage change in median price variable. While Granger
causality methodology is a useful starting point to examine intertemporal relationships,
it is limited in its ability to identify structural relationships between more than two
variables at a time. Thus, with this methodology we simply seek to establish whether
there are unilateral, bilateral, or independent relationships between each variable dyad.
Granger’s (1969) de?nition of causality asserts that variable Xcauses variable Yif past
values of Xand Yhelp explain the variation in current Ybetter than previous values of Y
itself. Granger causalityis predicatedona rather simple concept. Abase-line is established
byusingprevious values of Yto explainthe current value of Y. Byaddinglaggedvalues of
another variable X, the subsequent model will explainat least as muchof the variationinY
as the base-line model. If the additional explanatory value of the lagged values of X more
than outweigh the lost degrees of freedom associated with the additional explanatory
variables, variable Xis said to Granger cause variable Y. It is possible that variable Ycan
also Granger cause variable X, suggesting bilateral feedback between the two variables or
a potential third variable that is causing both X and Y. If neither X nor Y Granger cause
each other, then the two variables can be considered independent in the Granger sense of
causation even if the correlation between the two variables is positive.
Figure 6.
Shadow inventory and
median prices
(2000:q1-2009:q4)
12,000
10,000
8,000
6,000
4,000
2,000
S
h
a
d
o
w
i
n
v
e
n
t
o
r
y
2000q1
2002q3
2005q1 2007q3 2010q1
1,50,000
2,00,000
2,50,000
3,00,000
3,50,000
M
e
d
i
a
n
p
r
i
c
e
Shadow inventory Median price
Real estate
bubble
61
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Before testing the interrelationships between the percentage of transactions that are
part of a ?ip, the percentage of transactions that are foreclosures, and the percentage
change in median price, it is important to establish the stationarity of each time series.
Non-stationarity of one or more of the variables could lead to spurious results in any
Granger causality test. To establish stationarity, the ?rst column of Table IV reports
Dickey-Fuller test statistics assuming no drift for each variable (qualitatively similar
results were obtained when allowing for drift). As can be seen in all cases, the null
hypothesis of a unit root, i.e. non-stationarity, cannot be rejected.
It is dif?cult to reconcile how percentage of transactions that are ?ips or foreclosures
can actually be non-stationary in the long run because the two variables are constrained
from above and below. One possibility is that the Dickey-Fuller tests are misleading
because of structural breaks inthe data whichmake the series appear to be non-stationary.
The methodology developed by Zivot and Andrews (1992) tests for non-stationarity after
controlling for any data-determined structural break the procedure discovers. The ?nal
two columns of Table IV report the results of Zivot-Andrews tests assuming a single
structural break in the data. The second column shows that after controlling for a
structural break, the three variables are all stationary. In other words, the Dickey-Fuller
results in column one are likely incorrect.
The ?nal column in Table IV reports the data-determined structural breaks for the
three variables. The structural breaks are remarkably aligned with anecdotal evidence
of when these characteristics of the Las Vegas housing market experienced fundamental
change. For instance, the percentage change in median price and the percentage of
transactions that were ?ips reveal a structural break during the ?rst quarter of 2004,
exactly when claims of many industry observers suggested ?ipping experienced a
fundamental increase in popularity. In addition, the percentage of transactions that were
foreclosures reveals a structural break in the second quarter of 2007, very close to when
industry observers suggest the housing market peaked. As median prices started to fall,
highly leveraged home buyers began to ?nd their mortgages underwater, leading to a
shift in the temporal pattern of foreclosures in the Las Vegas housing market.
The results presented in Table IV suggest that, after controlling for the structural
breaks in the data, all three variables are stationary. Thus, the standard Granger
causality regression, where the current value of the dependent variable is regressed on
the lagged values of the dependent and independent variables, is augmented with a
dummy variable that takes a value of zero before the structural break of the independent
variable and one thereafter.
The results of the Granger causality tests are reported in Table V. Each dyad between
the three variables involves a pair of Granger causalitytests. The ?rst number ina cell uses
onlyone-quarter laggedvalues of bothvariables, andwe label this as short-termcausation.
Variable Dickey-Fuller test Zivot-Andrews test Break point
Percentage ?ips 21.097 25.555
*
2004:q1
Percentage foreclosures 1.274 24.812
*
2007:q2
Percentage change in median price 21.887 24.635
* *
2004:q1
Notes: Signi?cance at:
*
p , 0.05 and
* *
p , 0.07; p-values indicate rejection of the null hypothesis of
a unit root (non-stationarity)
Table IV.
Stationarity tests
JFEP
3,1
62
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The second Granger causality statistic uses four quarters of lagged values of both
variables and we refer to this as longer term causation.
The Granger causality tests in Table V reveal an intriguing set of relationships
between the percentage change in median price, the percentage of transactions that are
?ips and the percentage of transactions that are foreclosures. The ?rst result is that there
seems to be independence between ?ips and foreclosures (i.e. there is no Granger causality
ineither direction). Onthe other hand, ?ips in?uence percentage change inmedianprice in
the longer term, while foreclosures in?uence the percentage change in median price in the
short term. Finally, the last rowin Table Vshows that percentage change in median price
in?uences the other two variables over both one and four quarters.
Thus, while there is some feedback from ?ips and foreclosures on the percentage
change in price, it is evident that the driving force among these three variables is
the percentage change in price. One explanation for this is that percentage change in
price in?uences the pro?tability (or loss) of ?ips and foreclosures. As prices increase,
regardless of their level, individuals attempt to reap pro?ts from house ?ipping. On the
other hand, as housing prices fall, more individuals ?nd their equity eroded and
eventually ?nd their mortgages underwater which might ultimately lead to foreclosure.
A somewhat surprising result from these bivariate tests is the lack of a causal
relationship (in either direction) between ?ips and foreclosures. Conventional wisdom
might suggest that ?ips Granger cause foreclosures, that is, those who bought in the
sell-side ?ip transaction might have overpaid for the property and be more likely to walk
away once underwater. On the other hand, foreclosures might be expected to in?uence
?ips as foreclosures lower the price on the buy-side of the ?ip and make ?ips potentially
more pro?table. However, there is no causal relationship between the two suggesting
that they are independent of each other in the Las Vegas market.
6. Policy discussion and conclusions
Looking backwards, it is easy to trace through the housing bubble in Las Vegas over the
last decade. The percentage change in price was the driving force behind a surge of
?ipping activity that arti?cially boosted demand for housing in the metropolitan area.
This, in turn, ignited further price increases, and home builders responded by
constructing more new homes. Ultimately, growth in the Las Vegas housing stock
outstripped population growth and the resulting moderation in price increases meant
Variable
Percentage
?ips
Percentage
foreclosures
Percentage change median
price
Percentage ?ips 1.47/0.87 2.29/4.22
*
Percentage foreclosures 3.94/1.08 3.97
*
/2.39
Percentage change median
price 5.10
*
/2.57
*
8.45
*
/3.14
*
Notes: Signi?cance at:
*
p , 0.10,
* *
p , 0.05, and
* * *
p , 0.01; the hypothesis tested is that the
column (action) variable Granger causes the row (reaction) variable; values represent F-statistics with
(1,60) and (4,51) degrees of freedom, respectively; in the tests involving the percentage change in
median price, the degrees of freedom are (1,59) and (4,50), respectively; the ?rst number reported uses
one lag to test for Granger causality; the second number reported uses four lags to test for Granger
causality; row variables are action variables and column variables are reaction variables
Table V.
Granger causality tests
Real estate
bubble
63
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
that many ?ips were no longer pro?table. As ?ipping activity slowed considerably,
house prices began to fall. Eventually some homeowners found their mortgages
underwater and defaulted on their notes. These foreclosures led to a further decline in
prices causing more foreclosures in the area.
That price changes drive foreclosures is consistent with Elul et al. (2010) and
Bhutta et al. (2010) who ?nd that negative equity is a primary reason for default.
Together, this work suggests that loan modi?cation programs will necessarily have
limited success in curbing new foreclosures, and steps must be taken to ?rm up prices.
One possibility is to expand resources like HUD’s Neighborhood Stabilization Program
(http://hudnsphelp.info/). This program shores up demand of foreclosed properties by
providing ?nancial assistance to ?rst-time home buyers. Still another Neighborhood
Stabilization Program in?uences supply of foreclosed homes by granting funds to
government entities for the purpose of demolishing blighted neighborhoods. Detroit, for
example, plans to knock down 3,000 homes by September 2010 using federal
government funds. Moreover, Mayor Dave Bing has promised to tear down 10,000
structures in his ?rst termin of?ce to “right-size” Detroit and align housing needs with a
shrinking city population (Kellogg, 2010).
Giventhat excess supplyis part of the foreclosure problem, it is perhaps surprisingthat
new home building continues in Las Vegas. As one Las Vegas builder noted (Streitfeld,
2010), “We’re buildingthembecause we’re sellingthem.” Yet newhome buildingcontinues
to add to the problem of excess supply, falling prices and foreclosed homes. In
circumstances like these, local government units might develop policies that encourage
renovationof properties andrehabilitationof neighborhoods. While declaringa temporary
moratoriumon newhomes may diminish local tax and permit revenues generated by new
housing construction in the short run, the offset is that city will not have to contend with
the costs associated with abandoned homes and blighted neighborhoods.
At least two other lessons can be derived from the Las Vegas housing market bubble.
First, ?ipping activity contributed to rising home prices, and given asymmetric
information, it might be prudent to alert potential homebuyers of legal ?ipping activity.
One way to do that is to require multiple listing service listings to include information on
when the current owner bought the property and whether the current owner lives in the
home. This solutiontothe asymmetric informationproblemis similar tothe requirement in
several states that sellers divulge information that they are an agent/owner of a property.
(See Levitt andSyverson(2008) for market distortions relatedto the agent/owner problem.)
The second lesson is that municipal governments must be consistent in their record
keeping. ClarkCountyinchanging the Ftransactioncode fromdeed-in-lieu of foreclosure
to foreclosure resale makes it dif?cult to do any meaningful time comparisons. More
importantly, up to the last three years, foreclosure resales were coded as arms-length
transactions (R). But in the recent downturn, 90 percent of foreclosure resales were coded
F. The upshot is that tax valuation models and many price indexes will not include these
F transactions. In the case of tax valuation models, excluding foreclosure resales may
seriously bias upward the county’s assessment of market value.
Finally, looking forward, it will be important to focus on underlying structural
inter-temporal relationships, perhaps with the use of vector autoregressive models. So,
for example, how did mortgage rates or easy credit in?uence prices, ?ipping activity
and foreclosures? Or, did pricing dynamics differ when prices were going up versus
down? Answers to questions such as these will be left for future research.
JFEP
3,1
64
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
References
Bhutta, N., Dokko, J. and Shan, H. (2010), “The depth of negative equity and mortgage default
decisions”, Finance and Economics Discussion Series No. 2010-35, working paper, Federal
Reserve Board, Washington, DC.
Brueckner, J.K. (1980), “A vintage model of urban growth”, Journal of Urban Economics, Vol. 8
No. 3, pp. 389-402.
Capozza, D.R. and Helsley, R.W. (1989), “The fundamentals of land prices and urban growth”,
Journal of Urban Economics, Vol. 26 No. 3, pp. 295-306.
Case, K.E. and Quigley, J.M. (2008), “How housing booms unwind: income effects, wealth effects,
and feedbacks through ?nancial markets”, European Journal of Housing Policy, Vol. 8
No. 2, pp. 161-80.
Depken, C.A. II, Hollans, H. and Swidler, S. (2009), “An empirical analysis of residential property
?ipping”, Journal of Real Estate Finance & Economics, Vol. 39 No. 3, pp. 248-63.
Elul, R., Souleles, N., Glennon, D. and Hunt, R. (2010), “What triggers mortgage default?”, Federal
Reserve Bank of Philadelphia Working Paper No. 10-13, Federal Reserve Bank of
Philadelphia, Philadelphia, PA.
Granger, C. (1969), “Investigating causal relations by econometric models and cross spectral
methods”, Econometrica, Vol. 37 No. 3, pp. 424-38.
Kellogg, A. (2010), “Detroit shrinks itself, historic homes and all”, available at:http://online.wsj.
com/article/SB10001424052748703950804575242433435338728.html (accessed24 July2010).
Levitt, S. and Syverson, C. (2008), “Market distortions when agents are better informed: the value
of information in real estate transactions”, The Review of Economics and Statistics, Vol. 90
No. 4, pp. 599-611.
Shiller, R. (2009), “Unlearned lessons from the housing bubble”, The Economists’ Voice, Vol. 6
No. 7, p. 1.
Streitfeld, D. (2010), “Building is booming in a city of empty houses”, available at: www.nytimes.
com/2010/05/16/business/16builder.html
Wheaton, W. andNechayev, G. (2008), “The 1998-2005 housing‘bubble’ andthe current ‘correction’:
what’s different this time?”, Journal of Real Estate Research, Vol. 30 No. 1, pp. 1-26.
Zivot, D. and Andrews, D. (1992), “Further evidence on the great crash, the oil price shock and the
unit root hypothesis”, Journal of Business and Economic Statistics, Vol. 10 No. 3, pp. 251-70.
Further reading
Dickey, D.A. and Fuller, W.A. (1979), “Distribution of the estimators for autoregressive time series
with a unit root”, Journal of the American Statistical Association, Vol. 74 No. 366, pp. 427-31.
Immergluck, D. and Smith, G. (2006), “The external costs of foreclosure: the impact of
single-family mortgage foreclosures on property values”, Housing Policy Debate, Vol. 17
No. 1, pp. 57-79.
Lin, Z., Rosenblatt, E. and Yao, V. (2009), “Spillover effects of foreclosure on neighborhood
property values”, Journal of Real Estate Finance & Economics, Vol. 38 No. 4, pp. 387-407.
Corresponding author
Steve Swidler can be contacted at: [email protected]
Real estate
bubble
65
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Katrin B. Anacker, Laurie A. Schintler. 2015. Flip that house: visualising and analysing potential real
estate property flipping transactions in a cold local housing market in the United States. International
Journal of Housing Policy 1-19. [CrossRef]
2. References 294-311. [CrossRef]
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_895891351.pdf
This paper aims to examine the anatomy of a real estate bubble. In the process, the paper
identifies three phases of the market’s evolution: flips, flops and foreclosures. An examination of the
Las Vegas real estate market illustrates the three phases.
Journal of Financial Economic Policy
Flips, flops and foreclosures: anatomy of a real estate bubble
Craig A. Depken II Harris Hollans Steve Swidler
Article information:
To cite this document:
Craig A. Depken II Harris Hollans Steve Swidler, (2011),"Flips, flops and foreclosures: anatomy of a real
estate bubble", J ournal of Financial Economic Policy, Vol. 3 Iss 1 pp. 49 - 65
Permanent link to this document:http://dx.doi.org/10.1108/17576381111116759
Downloaded on: 24 January 2016, At: 21:40 (PT)
References: this document contains references to 16 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 434 times since 2011*
Users who downloaded this article also downloaded:
Dag Einar Sommervoll, Gavin Wood, (2011),"Home equity insurance", J ournal of Financial Economic
Policy, Vol. 3 Iss 1 pp. 66-85http://dx.doi.org/10.1108/17576381111116768
Beverley Searle, (2011),"Recession and housing wealth", J ournal of Financial Economic Policy, Vol. 3 Iss 1
pp. 33-48http://dx.doi.org/10.1108/17576381111116867
Steve Swidler, (2011),"Homeownership: yesterday, today and tomorrow", J ournal of Financial Economic
Policy, Vol. 3 Iss 1 pp. 5-11http://dx.doi.org/10.1108/17576381111116786
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Flips, ?ops and foreclosures:
anatomy of a real estate bubble
Craig A. Depken II
Department of Economics, Belk College of Business, UNC-Charlotte, Charlotte,
North Carolina, USA, and
Harris Hollans and Steve Swidler
Department of Finance, Auburn University, Auburn, Alabama, USA
Abstract
Purpose – This paper aims to examine the anatomy of a real estate bubble. In the process, the paper
identi?es three phases of the market’s evolution: ?ips, ?ops and foreclosures. An examination of the
Las Vegas real estate market illustrates the three phases.
Design/methodology/approach – The paper examines transaction data from the metropolitan
Las Vegas area (Clark County) from1994 to 2009. The ?rst part of the analysis identi?es the three phases
of the bubble and is descriptive in nature. This is followed by more formal tests of Granger causality.
Findings – In the early part of the sample, a large percentage of transactions are speculative or
“?ips” causing prices to rapidly increase. Eventually, ?ipping loses its pro?tability and over the last
three years, there is an increasing number of foreclosures leading to falling prices. The descriptive
analysis of the Las Vegas market is augmented with causality tests which show that prices were the
driving force behind all three phases in the market’s evolution.
Research limitations/implications – Future research might focus on underlying structural
inter-temporal relationships to augment the Granger causality tests.
Practical implications – Analysis shows that price is the driving force behind a bubble and that
loan modi?cation programs alone will not solve the current housing crisis.
Social implications – Government entities might expand neighborhood stabilization programs to
affect both demand and supply of homes. Moreover, it might be prudent to include information related
to ?ipping on multiple listing service agreements. Additionally, local governments should be
consistent in their record keeping.
Originality/value – To the best of the authors’ knowledge, this is the ?rst paper to examine the
housing bubble using an extensive set of transaction data.
Keywords Mortgage companies, Mortgage default, Real estate, United States of America
Paper type Research paper
1. Introduction
Conventional wisdom in the USA blames the housing market as the “?rst domino” that
fell in the lead-up to the recession that began in early 2007 (Shiller, 2009). Some have
alluded to a housing bubble that was unsustainable and which caused individuals to
have a false perception that housingprices would continue to increase, therebymakingit
pro?table to purchase more expensive homes or to speculate on residential real estate in
so-called “?ipping” (Wheaton and Nechayev, 2008). Indeed, the popularity of ?ipping
might be re?ected in popular culture in which television shows such as “Flip this House”
(A&E network) and “Flip that House” (TLC) were amongst the most popular television
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1757-6385.htm
JEL classi?cation – G11, G21, R31
Real estate
bubble
49
Journal of Financial Economic Policy
Vol. 3 No. 1, 2011
pp. 49-65
qEmerald Group Publishing Limited
1757-6385
DOI 10.1108/17576381111116759
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
shows in the early 2000s. As markets have cooled off or even collapsed, these shows have
since been replaced by less bullish shows such as late night television “infomercials”
selling foreclosed residential condominiums.
This paper examines the anatomy of a real estate bubble. In the process, we identify
three phases of the market’s evolution: in the ?rst phase, a large percentage of
transactions are speculative or “?ips” causing prices to rapidly increase; in phase 2,
?ipping loses its pro?tability and many individuals are caught “holding the bag,”
(i.e. cannot resell their house at a higher highprice); andinphase 3, there are anincreasing
number of foreclosures leading to falling prices. Eventually, properties held by banks
(shadowinventory) must be soldor destroyedbefore the market canrecover andstabilize.
To illustrate a real estate bubble, we investigate the evolution of the Las Vegas
metropolitan housing market from 1994 to 2009. We begin with positive economic
analysis that is mainly descriptive in nature and graphically captures the three phases of
the Las Vegas real estate bubble. Our subsequent analysis formally investigates the
extent to which ?ips, foreclosures and percentage change in price are related to each
other. Granger causality tests imply that percentage change in price is the driving force
behind ?ipping and foreclosure activity, but that ?ips and foreclosures are not directly
related to each other. Finally, normative analysis of a real estate bubble suggests a
number of policy propositions that might be considered by lawmakers and real estate
professionals.
While local housing markets follow idiosyncratic cycles, price trends include a
systematic component related to certain factors. For instance, the literature has
established that housing prices tend to increase as the local population increases, as new
housing stock replaces older homes, as local incomes increase, and as the supply of
developable land decreases (Brueckner, 1980; Capozza and Helsley, 1989). On the other
hand, (moderate) recessions are not necessarily associated with a fall in housing prices
but rather with a reduction in the number of houses sold in any particular time period
and an increase in the time-on-market for existing houses. In other words, house prices
tend to be sticky downward. Nevertheless, a suf?ciently severe recession might induce
price decreases. For instance, Case and Quigley (2008) ?nd that, in the last half of 2006,
sales activity slowed, but housing prices in Boston declined only moderately at the
beginning of the downturn. However, as the recession continued to deepen, Boston
housing prices at the end of 2009 were approximately 17 percent belowtheir peak, falling
to their 2003 levels (as measured by the Case-Shiller Index).
Although stable and increasing housing prices are frequently thought of as the norm,
it is possible for local housing markets to experience dramatic increases (and decreases)
in price. Whether such price volatility is generated by arti?cially restrained supply, for
instance, through overly restrictive zoning or land-use policies, or through arti?cially
enhanced demand, rapidly escalating prices might induce individuals to speculate on
residential properties in the formof “?ipping.” Flipping entails purchasing a residential
property, perhaps improving the property through cosmetic or structural changes, and
attempting to rapidly resell the property for a pro?t. House ?ipping contributes to an
increase in the demand for existing properties, thereby pushing up price. However,
house ?ipping might also be a rational response to other market signals such as a rapidly
increasing population or relatively easy credit for potential home buyers (Wheaton and
Nechayev, 2008). Estimating the relative in?uence of these possible factors is mainly an
empirical exercise.
JFEP
3,1
50
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Factors that lead to a dramatic escalation in housing prices cannot be expected to last
forever. Thus, a tapering-off period follows during which price increases moderate, the
pro?ts from“?ipping” fall, andthere is a decline inthe proportion of sales that are “?ips.”
This reduced exuberance might presage an actual, and potentially dramatic, decline in
price. If this is the case, those who attempted to participate in “?ipping” toward the end
of the exuberant period and many who purchased at the peak of the market will ?nd
themselves holding a depreciating asset. In this environment, the “?ipping” period is
followed by a period of “?ops” and ?nally a potential for foreclosures as some
individuals ?nd it in their best interest to default on their mortgage rather than trying to
sell the property for a loss. The subsequent analysis chronicles a cycle of ?ips, ?ops and
foreclosures in Clark County, NV, a district that essentially comprises the Las Vegas
metropolitan area.
2. Data and de?nitions of transaction type
The data sample used in this study describes 541,373 separate residential property
transactions from Clark County, NV from 1994:q1 to 2009:q4, obtained from the Clark
County tax assessor’s of?ce. The data describe, among other things, the transaction
price, the transaction type, the date of the transaction and a unique parcel identi?er.
We are able to examine up to nine separate transactions for each parcel, although the
vast majority of properties have less than four transactions during the sample period.
To facilitate the use of such a large data set and to provide a level of aggregation that
might inform policy discussion, the analysis translates each transaction date into the
appropriate quarter and year. The subsequent analysis is then undertaken on a
quarterly basis.
Each transfer of property is coded by the Clark County tax assessor’s of?ce according
to the transaction (sale) type. There are three categories that we examine:
(1) Recorded value. Denoting an arms-length transaction (coded with an “R”).
(2) Trustee’s deed. Trustee’s deed is the amount bid at foreclosure auction on the
trustee’s deed (coded with a “T”).
(3) Foreclosure. Foreclosure is a transfer indicating a resale after foreclosure (coded
with an “F”).
These three sale types constitute the bulk of all transactions ?led at the tax assessor’s
of?ce and are the most important categories for the purposes of our investigation.
Table I lists the distribution of residential transactions by sales type. For the entire
sample period, there are 464,093 “R” transactions, 47,320 “T” transactions and29,960 “F”
transactions recorded. Examining tax records on a quarterly basis, total transactions
trend upwards and reach a peak of more than 15,000 sales in the third quarter of 2005.
On a percentage basis, arms-length transactions (R) constitute more than 95 percent of
all sales through the fourth quarter of 2006. House prices in Las Vegas (as measured by
the Case-Shiller Index) re?ected the vigorous sales activity of this period, and after large
run-ups in 2004 and 2005, prices eventually crested in the second quarter of 2006.
To put the ?gures in perspective, Figure 1 shows the number of R, T and
F transactions for the period 2004:q1-2009:q4. As can be seen, arms-length transactions
dominate the distribution up until the end of 2006. In 2007, the F and T transactions
begin to increase in number and eventually foreclosure activity constitutes the leading
share of sales. The properties in foreclosure combined with additional new and used
Real estate
bubble
51
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Quarter Total R Total T Total F T(%) F(%) R(%)
1994q1 3,514 102 3 2.82 0.08 97.1
1994q2 4,422 77 4 1.71 0.09 98.2
1994q3 4,116 79 5 1.88 0.12 98
1994q4 3,868 83 4 2.1 0.1 97.8
1995q1 3,202 85 1 2.59 0.03 97.38
1995q2 3,935 72 7 1.79 0.17 98.04
1995q3 4,134 82 3 1.94 0.07 97.99
1995q4 3,977 84 4 2.07 0.1 97.83
1996q1 4,120 85 2 2.02 0.05 97.93
1996q2 4,730 68 5 1.42 0.1 98.48
1996q3 4,593 92 2 1.96 0.04 98
1996q4 4,755 92 7 1.9 0.14 97.96
1997q1 4,064 140 4 3.33 0.1 96.57
1997q2 4,867 116 20 2.32 0.4 97.28
1997q3 5,049 155 4 2.98 0.08 96.94
1997q4 4,923 146 8 2.88 0.16 96.96
1998q1 4,469 178 6 3.83 0.13 96.04
1998q2 5,669 197 4 3.36 0.07 96.57
1998q3 5,645 236 7 4.01 0.12 95.87
1998q4 5,727 231 3 3.88 0.05 96.07
1999q1 5,406 263 8 4.63 0.14 95.23
1999q2 6,656 253 3 3.66 0.04 96.3
1999q3 6,453 282 8 4.18 0.12 95.7
1999q4 6,048 233 6 3.71 0.1 96.19
2000q1 5,654 288 5 4.84 0.08 95.08
2000q2 6,850 254 9 3.57 0.13 96.3
2000q3 6,586 319 9 4.61 0.13 95.26
2000q4 6,756 276 6 3.92 0.09 95.99
2001q1 6,600 328 6 4.73 0.09 95.18
2001q2 8,134 300 5 3.55 0.06 96.39
2001q3 7,668 309 5 3.87 0.06 96.07
2001q4 8,086 356 9 4.21 0.11 95.68
2002q1 7,537 364 8 4.6 0.1 95.3
2002q2 8,724 447 7 4.87 0.08 95.05
2002q3 8,884 431 16 4.62 0.17 95.21
2002q4 9,180 392 6 4.09 0.06 95.85
2003q1 8,619 450 8 4.96 0.09 94.95
2003q2 10,762 466 10 4.15 0.09 95.76
2003q3 12,537 461 10 3.54 0.08 96.38
2003q4 12,354 363 7 2.85 0.06 97.09
2004q1 12,513 342 6 2.66 0.05 97.29
2004q2 15,432 99 14 0.64 0.09 99.27
2004q3 14,635 85 13 0.58 0.09 99.33
2004q4 12,907 39 8 0.3 0.06 99.64
2005q1 12,360 34 11 0.27 0.09 99.64
2005q2 14,970 25 3 0.17 0.02 99.81
2005q3 15,453 21 2 0.14 0.01 99.85
2005q4 14,302 42 6 0.29 0.04 99.67
2006q1 12,414 74 10 0.59 0.08 99.33
2006q2 13,267 128 3 0.96 0.02 99.02
(continued)
Table I.
Temporal distribution of
residential transaction
types in Las Vegas, NV
JFEP
3,1
52
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
properties on the market might be expected to exert downward pressure on price and
alter the expectations of potential buyers about future price changes. In fact, the last
three years of the sample period witnessed rapid price declines in the Las Vegas market.
The dispersionof foreclosure activityis not evenlydistributedacross the 104 different
tax districts of Clark County. Table II depicts the R, T and F transactions for the 21 tax
districts that constitute our sample and include all areas with more than 1,000 residential
transactions during the period of analysis. Of particular note is that the districts with the
largest foreclosure activity tend to have the lowest per capita income in the area.
In particular, foreclosures were more than 17 percent of total sales in the low-income
districts of North Las Vegas, Sunrise Manor and Whitney.
Figure 1.
Distribution of residential
property transactions by
type and quarter
(2004:q1-2009:q4)
2004q1
0
5,000
10,000
N
u
m
b
e
r
o
f
t
r
a
n
s
a
c
t
i
o
n
s
15,000
2005q3 2007q1 2008q3 2010q1
Date
Arms-length transactions (R)
Trustee's deeds (T)
Foreclosures (F)
Quarter Total R Total T Total F T(%) F(%) R(%)
2006q3 11,741 238 10 1.99 0.08 97.93
2006q4 10,511 314 18 2.9 0.17 96.93
2007q1 7,801 646 29 7.62 0.34 92.04
2007q2 7,451 980 76 11.52 0.89 87.59
2007q3 6,103 1,432 154 18.62 2 79.38
2007q4 5,521 2,231 213 28.01 2.67 69.32
2008q1 3,515 3,060 1,607 37.4 19.64 42.96
2008q2 3,828 4,586 3,095 39.85 26.89 33.26
2008q3 4,046 5,108 4,244 38.13 31.68 30.19
2008q4 3,755 4,473 3,861 37 31.94 31.06
2009q1 2,787 4,319 4,127 38.45 36.74 24.81
2009q2 3,461 3,591 5,643 28.29 44.45 27.26
2009q3 4,205 4,431 4,914 32.7 36.27 31.03
2009q4 1,842 1,787 1,639 33.92 31.11 34.97
Note: Quarterly data from 2009:q4 is not from the entire quarter Table I.
Real estate
bubble
53
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
T
a
x
d
i
s
t
r
i
c
t
n
u
m
b
e
r
T
a
x
d
i
s
t
r
i
c
t
n
a
m
e
T
o
t
a
l
t
r
a
n
s
a
c
t
i
o
n
s
R
t
r
a
n
s
a
c
t
i
o
n
s
T
t
r
a
n
s
a
c
t
i
o
n
s
F
t
r
a
n
s
a
c
t
i
o
n
s
T
a
n
d
F
t
r
a
n
s
a
c
t
i
o
n
s
(
%
)
5
0
B
o
u
l
d
e
r
C
i
t
y
L
i
b
r
a
r
y
2
,
5
5
6
2
,
4
3
7
8
2
3
7
4
.
6
6
1
0
7
L
a
u
g
h
l
i
n
T
o
w
n
B
i
g
B
e
n
d
C
o
l
o
r
a
d
o
R
i
v
e
r
1
,
1
5
4
1
,
0
6
0
7
1
2
3
8
.
1
5
1
2
5
A
r
t
e
s
i
a
n
B
a
s
i
n
F
i
r
e
9
1
1
1
,
1
9
8
9
9
5
1
1
4
8
9
1
6
.
9
4
2
0
0
L
a
s
V
e
g
a
s
C
i
t
y
1
6
4
,
1
2
3
1
4
1
,
0
6
3
1
4
,
5
2
3
8
,
5
3
7
1
4
.
0
5
2
0
3
L
a
s
V
e
g
a
s
C
i
t
y
R
e
d
e
v
e
l
o
p
m
e
n
t
1
,
1
1
8
9
3
7
1
2
5
5
6
1
6
.
1
9
2
5
0
N
o
r
t
h
L
a
s
V
e
g
a
s
C
i
t
y
5
2
,
1
6
9
4
3
,
0
5
5
5
,
6
5
2
3
,
4
6
2
1
7
.
4
7
2
5
4
N
o
r
t
h
L
a
s
V
e
g
a
s
L
i
b
r
a
r
y
3
1
,
8
1
5
2
5
,
9
1
1
3
,
3
9
9
2
,
5
0
5
1
8
.
5
6
3
4
0
S
u
n
r
i
s
e
M
a
n
o
r
4
3
,
0
0
9
3
5
,
2
5
1
4
,
9
7
8
2
,
7
8
0
1
8
.
0
4
4
1
0
W
i
n
c
h
e
s
t
e
r
T
o
w
n
2
,
1
6
9
1
,
8
8
5
1
9
3
9
1
1
3
.
0
9
4
1
7
S
p
r
i
n
g
V
a
l
l
e
y
T
o
w
n
4
6
,
2
8
5
4
0
,
2
6
2
3
,
5
4
1
2
,
4
8
2
1
3
.
0
1
4
2
0
S
u
m
m
e
r
l
i
n
T
o
w
n
A
r
t
e
s
i
a
n
B
a
s
i
n
1
0
,
4
6
8
9
,
6
2
3
4
7
8
3
6
7
8
.
0
7
4
7
0
P
a
r
a
d
i
s
e
T
o
w
n
2
6
,
2
4
9
2
3
,
0
1
5
2
,
1
1
0
1
,
1
2
4
1
2
.
3
2
5
0
0
H
e
n
d
e
r
s
o
n
C
i
t
y
3
,
3
9
0
3
,
0
2
7
2
2
2
1
4
1
1
0
.
7
1
5
0
3
H
e
n
d
e
r
s
o
n
C
i
t
y
R
e
d
e
v
e
l
o
p
m
e
n
t
1
,
8
3
9
1
,
5
5
6
1
8
2
1
0
1
1
5
.
3
9
5
0
5
H
e
n
d
e
r
s
o
n
A
r
t
e
s
i
a
n
B
a
s
i
n
4
9
,
8
3
2
4
5
,
2
5
5
2
,
8
3
8
1
,
7
3
9
9
.
1
8
5
1
4
H
e
n
d
e
r
s
o
n
C
i
t
y
L
i
b
r
a
r
y
D
e
b
t
4
,
1
4
4
3
,
8
7
8
1
5
3
1
1
3
6
.
4
2
5
1
6
H
e
n
d
e
r
s
o
n
L
i
b
r
a
r
y
D
e
b
t
/
A
r
t
e
s
i
a
n
B
a
s
i
n
1
8
,
8
6
8
1
7
,
1
3
7
1
,
0
2
0
7
1
1
9
.
1
7
5
2
1
H
e
n
d
e
r
s
o
n
C
i
t
y
R
e
d
e
v
e
l
o
p
m
e
n
t
5
2
1
4
,
5
2
2
3
,
7
6
5
5
1
4
2
4
3
1
6
.
7
4
5
7
0
W
h
i
t
n
e
y
A
r
t
e
s
i
a
n
B
a
s
i
n
1
2
,
4
1
9
1
0
,
1
9
6
1
,
3
4
1
8
8
2
1
7
.
9
0
6
3
5
E
n
t
e
r
p
r
i
s
e
F
i
r
e
A
r
t
e
s
i
a
n
L
i
b
r
a
r
y
9
1
1
M
a
n
p
o
w
e
r
5
9
,
3
7
9
4
9
,
3
5
5
5
,
6
2
6
4
,
3
9
8
1
6
.
8
8
9
0
1
M
e
s
q
u
i
t
e
C
i
t
y
4
,
6
6
7
4
,
4
3
0
1
5
8
7
9
5
.
0
8
T
o
t
a
l
s
5
4
1
,
3
7
3
4
6
4
,
0
9
3
4
7
,
3
2
0
2
9
,
9
6
0
Table II.
Tax districts and the
frequency of transaction
types (full sample)
JFEP
3,1
54
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
It is important that the county assessor accurately characterize each transaction as the
data serves as a foundation for mass-appraisal models used to determine “fair market
value” for ad valorem tax purposes. As such, arms-length transactions denoted as
R transactions in the data serve as the benchmark for market value estimation.
Frequently, arms-length transactions are thought of as a sale between a willing seller and
a willing buyer.
Atrustee’s deed transaction (coded T) denotes a foreclosure sale and signi?es that the
property either resides in the real estate-owned (REO) inventory of the lender or was
purchased at the foreclosure sale by an owner/investor. Typically, the lender is the
winning bidder at auction and the recorded sale price of the Ttransaction represents the
amount bid on the trustee’s deed. A T transaction may also represent a deed-in-lieu of
foreclosure, which entails the lender repossessing the house without pursuing a
foreclosure on the property, with the result that the homeowner loses whatever equity
they have in the house. Presumably there is little or, more likely, negative equity
precipitating the transference of the property. In a deed-in-lieu of foreclosure, the lender
often agrees to not pursue the individual home owner for recourse, which has arguably
become easier under the recently passed rules of the Trouble Asset Relief Program and
the American Recovery and Reinvestment Act.
Early in the sample period, an F code denoted a deed-in-lieu of foreclosure
transaction. With the recent increase in foreclosure activity, the county changed an
F transaction to mean that the transfer of a property is a resale after foreclosure. The
typical example of a recently coded F transaction would be the sale of the house by the
lender (who acquired the trustee’s deed through a T transaction) to a new homeowner.
If the sale price is thought to be different from market value, the county then codes this
as an F transaction.
Table III gives the ?avor of foreclosure activityinClarkCounty andhowthe codingof
deed-in-lieu of foreclosure changed over the sample period. In the early part of the
sample (1994:q1-2006:q4), homes going into foreclosure were predominantly coded
Ttransactions that signaled transference of a trustee’s deed. In a typical quarter, nearly
161 homes were T transactions, while ?ve were coded F by the county and primarily
denoted a deed-in-lieu of foreclosure sale. Looking at the last three columns of Table III,
whether designated a Tor an Ffor homes going into foreclosure, virtually all subsequent
sales were coded R by the county. This implies that almost all houses were sold to the
new homeowners at “market value,”, i.e. the county considered the transaction between
the lender and new homeowner as an arms-length sale.
Over the more recent sample period (2007:q1-2009:q4), Table III illustrates two
important changes that occurred. First, there were virtually no examples of a T(trustee’s
deed) followed by an F (deed-in-lieu of foreclosure) in the early sample period. However,
Tthen F becomes the predominant foreclosure sequence of transactions moving into the
latter sample period and implies that for most of 2007-2009, Ttransactions included both
trustee’s deed and deed-in-lieu of foreclosure transactions. Moreover, an F transaction in
2007-2009 refers to a resale after foreclosure. In the third column, the few100 examples of
F followed by R found in 2007-2009 presumably denote properties that sold in an earlier
period as a deed-in-lieu of foreclosure, but then were coded as an arms-length transaction
on the subsequent sale.
Of potentially more interest is the second change made in recent years. Whereas it has
already been noted that from 1994 to 2006 virtually all second sales in the foreclosure
Real estate
bubble
55
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
sequence were considered R transactions, by 2009 more than 90 percent of second sales
were coded F by the county. This has important implications for housing valuation.
In recent years, the county no longer considered the vast majority of the foreclosure
resales as an arms-length transaction, and therefore, they would not likely be used in any
tax valuation model or price index.
3. Property ?ipping
As documented earlier, the rapid rise in housing prices from 2003 to 2005 corresponded
to a period that experienced a high number of sales. The county designated virtually all
transactions as arms length (R). Many of these sales involved property ?ipping.
Depken et al. (2009) de?ne a ?ip as the purchase of a home with the intent of quickly
reselling the property at a higher price. They examine all cases that involve two
arms-length transactions for a property within a two-year window. Two years is a
relevant time frame as the Internal Revenue Service allows any capital gains to be
excluded from taxable income if the seller has used the home as his or her primary
residence for two of the previous ?ve years. This de?nition does not depend upon
?ipping motivation or economic pro?tability, but rather focuses on the short-term
investment horizon that is the epitome of house ?ipping.
House ?ipping is often depicted as purchasing property in poor condition at a
discount, renovating the house and then selling it at or near full market value. This is
sometimes referred to as a “?x and ?ip” and has been the basis for a number of reality
television shows. However, this is not the only situation in which a property might be
ripe for “?ipping.” Some properties might be purchased at a discount due to forced
circumstances such as relocation, divorce, or a pending foreclosure. Such situations
might provide for a nominally pro?table opportunity if the house can be resold at market
value in a relatively short amount of time.
Quarter
T
followed
by an R
T
followed
by an F
F
followed
by an R
F
followed
by a T
F or T
followed by
an R
F or T
followed by
an F
F or T
followed
by a T
Quarterly
average 1994:q1-
2006:q4 160.8 0.2 5.0 0.3 165.7 0.2 2.6
2007:q1 174 14 6 1 180 14 11
2007:q2 288 33 4 2 292 33 19
2007:q3 268 86 3 1 271 86 13
2007:q4 370 119 3 2 373 120 16
2008:q1 96 1,380 20 5 116 1,382 30
2008:q2 85 2,772 29 15 114 2,779 65
2008:q3 114 3,892 85 7 199 3,905 43
2008:q4 116 3,697 97 10 213 3,700 58
2009:q1 126 3,896 95 10 221 3,902 31
2009:q2 79 5,148 160 12 239 5,158 53
2009:q3 95 4,619 208 20 303 4,626 65
2009:q4 76 1,571 117 9 193 1,571 42
Totals 1994:q1-
2009:q4 10,246 27,237 1,086 107 11,332 27,287 581
Note: Quarterly data from 2009:q4 is not from the entire quarter
Table III.
Sales around foreclosures
JFEP
3,1
56
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Givenmultiple transactions for a givenproperty, it is possible toidentify, expost, which
transactions are the front-endof aneventual house ?ip, calledthe buy-side ?iptransaction,
and which are the back-end of an eventual house ?ip, called the sell-side ?ip transaction.
This, in turn, allows for the calculation of any economic pro?ts due to ?ipping. House
?ipping, similar to any speculative activity, is an inherently risky proposition, and higher
risksuggests higher expectedreturn. Depkenet al. (2009) ?nd that ?ippers earned positive
economic pro?ts in the Las Vegas residential housing market up through 2005, in large
part, re?ecting sell-side ?ip prices that tended to be higher than other similar arms-length
transactions.
The premiumon the sell-side transaction might indicate improvement to the property
not captured in the tax records. However, the housing stock in Las Vegas is relatively
young, and many ?ips were of new or nearly new houses. It is unlikely that sell-side
premiums were primarily a re?ection of home improvements.
Alternatively, a sales premium might obtain due to the circumstances of ?ipping.
If information is costly to obtain, ?ippers may have superior knowledge of the local real
estate market. Since there is no pressure to relocate, ?ippers can take time to search for
buyers offering the highest price for the property. Still another possibility due to
asymmetric information is that sell-side ?ips might carry a premium because of illegal
activity. For instance, appraisers might collude with mortgage originators and the
?ipper’s broker to provide an in?ated appraisal of the ?ipped property. In?ated prices
might then be the result of mortgage companies arti?cially stimulating demand by
qualifying buyers for more expensive homes. To limit “predatory ?ipping,” the
Department of Housing and Urban Development promulgated a set of guidelines in 2004
(amendedin2006) that prohibitedFederal HousingAuthority-insuredmortgage ?nancing
for properties re-sold within 90 days. These guidelines, however, did little to stop
mortgage fraud that has been the focus of, “Operation Stolen Dreams,” an FBI campaign
that targets illegal ?ipping activity, loan origination schemes, and equity skimming
(www.fbi.gov/page2/june10/mortgage_061710.html).
To get a better idea of Las Vegas ?ipping activity during the sample period, Figure 2
shows quarterlysell- andbuy-side ?iptransactions as a percentage of all arms-lengthsales.
Figure 2.
Percent buy-side and
sell-side ?ip transactions
(1994:q1-2009:q4)
1995q1
0.25
0.2
0.15
0.1
0.05
0
2000q1
P
e
r
c
e
n
t
o
f
a
l
l
t
r
a
n
s
a
c
t
i
o
n
s
2005q1
Date
Percent sell-side flip transactions
Percent buy-side flip transactions
2010q1
Real estate
bubble
57
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
As can be seen, buy- and sell-side transactions comprise a small percentage of all
transactions until sometime after 2002 when the buy-side ?ip transactions break the
10 percent barrier. Eventually, the buy-side ?ip transactions peaks at approximately
23 percent of all arms-length transactions in the ?rst quarter of 2004 after which the
percentage drops dramatically. The falling number of buy-side ?ip transactions coincides
with the declining pro?tably of ?ipping and the eventual fall in housing prices
(Depken et al., 2009).
Figure 2 also shows a pattern of seasonality in the ?ip transactions data. Over the
entire sample period, there is no statistically meaningful difference between the
percentage of buy-side transactions in quarters 1, 2 or 3. However, there is a statistically
signi?cant decline in the percentage of buy-side transactions in the fourth quarter (by
approximately 3.8 percent between q1 and q4, p-value ¼ 0.032). There exists a different
but distinct pattern in sell-side ?ip transactions, as well: there is a positive and slightly
signi?cant increase in sell-side ?ip transactions in quarter 4 (by approximately
2.7 percent between q1 and q4, p-value ¼ 0.098) although there is no statistically
meaningful difference between quarter 1 and quarter 2 or 3.
When focus is restricted to the period starting with 2004:q1, the seasonal patterns
disappear for both buy- and sell-side ?ip transactions. As can be seen in Figure 3, the
percentage of buy-side ?ip transactions falls below the percentage of sell-side ?ip
transactions by the third quarter of 2004, that is, buy-side ?ip transactions fall before the
Las Vegas market experienced dramatic decreases in prices and increases in foreclosure
activity. Moreover, Figure 3 shows that ?ipping activity never actually reached zero;
even during the dramatic downward adjustment of the market after 2006, there were still
some properties that speculators felt were potentially pro?table “?ips.”
4. Descriptive analysis of the Las Vegas housing bubble
Figure 4 shows an evolution of ?ips, foreclosures, housing prices and price changes in
Clark County over the entire sample period. The bar graph above the x-axis
Figure 3.
Percent buy-side and
sell-side ?ip transactions
(2004:q1-2009:q4)
0.25
0.2
0.15
0.1
0.05
0
2004q1 2005q3 2007q1 2008q3 2010q1
Date
P
e
r
c
e
n
t
o
f
a
l
l
t
r
a
n
s
a
c
t
i
o
n
s
Percent sell-side flip transactions
Note: Buy-side flips in 2008 and 2009 do not cover entire two year flipping window
Percent buy-side flip transactions
JFEP
3,1
58
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
represents the percentage of all transactions identi?ed ex post as one side of a ?ip
(both buy- and sell-side each quarter). The bar graph belowthe horizontal axis illustrates
the number of trustee deeds (T) and foreclosures (F) as a percentage of total transactions.
The percentage of ?ips and foreclosures are then drawn against the median price of all
arms-length transactions (solid line) and the annualized percentage change in median
price for each quarter (dashed line). In chronicling the boomto bust housing cycle in the
Las Vegas area, the graph can be neatly divided into three distinct regimes.
Flips – 1994:q1 through 2005:q4
Between 1994 and 2000, the great majority of sales were arms-length transactions, with
relatively few ?ips or foreclosures. During this period, quarterly price changes were
small (approximating the nominal in?ation rate), and the median transaction price in the
Las Vegas market remained relatively modest. After 2000, the percentage change in
prices increased and median prices followed accordingly. Seemingly in response to these
price changes, ?ipping activity also increased.
By 2004, ?ipping activity rose to roughly 40 percent of all housing transactions in
Las Vegas, and while perhaps not clear ex ante, ?ipping activity of this magnitude could
not be sustained forever. The arti?cial stimulus in demand from?ipping, in turn, helped
fuel anincrease in newhomes built. The housingstocknearly doubled in Las Vegas from
2000 to 2008; however, Clark County’s population grew only 33.9 percent (from 1.39 to
1.87 millionpeople) during the same period. Flippers, home builders and (non-?ip) sellers
eventually found it more dif?cult to ?nd buyers leading to the next stage of the bubble.
Flops – 2006:q1 through 2007:q4
Figure 5 replicates the previous graph, but centers on the transition period between ?ips
and foreclosures. As ?ippers found it more dif?cult to locate buyers for their properties,
price increases attenuated and prices eventually peaked in 2006. In many cases,
Figure 4.
Percentage of transactions
that were ?ips, percentage
of transactions
foreclosures, quarterly
median price and
percentage change in
median price
(1994:q1-2009:q4)
1
0.50
0
0.50
1
P
e
r
c
e
n
t
/
P
c
t
.
c
h
a
n
g
e
1994q1
Percent flips Percent foreclosures
Median price Pct. change med. price
1996q1 1998q1 2000q1 2002q1 2004q1 2006q1 2008q1 2010q1
1,00,000
1,50,000
M
e
d
p
r
i
c
e
M
e
d
i
a
n
p
r
i
c
e
2,00,000
2,50,000
3,00,000
Real estate
bubble
59
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
?ips were no longer pro?table leading to economic losses and a fall in ?ipping activity.
A reduction in ?ips occurred because fewer investors decided to buy properties for
?ipping, and those that did could not always sell them within two years. For many
investors, 2006-2007 was a period of ?ops, and by 2008, Figure 5 shows that ?ipping
activity had fallen to less than 5 percent of all transactions. This contributed to further
price declines, and starting in 2008, foreclosures constituted the majority of transactions
in the Las Vegas area.
Foreclosures – 2008:q1 through 2009:q4
In the ?nal phase, ?ipping was not economically pro?table as median prices continued
to decline, potential buyers were reluctant to purchase a home in anticipation of lower
future prices and more existing home owners found themselves underwater on their
mortgage. Coupled with an increasingly soft labor market and higher unemployment,
the number of foreclosures “snowballed” despite policy interventions such as mortgage
restructuring, ?rst-time homebuyer tax credits and the Fed’s purchase of
mortgage-backed securities leading to lower interest rates.
One more effect of the high foreclosure rate is a sharp increase in shadow inventory.
Figure 6 shows the stock of homes owned by the lender, where quarterly changes equal
the net ?ows of properties going into foreclosure minus REO inventory that has been
resold. During 2004, with prices increasing rapidly, REOinventory was being sold faster
than the (small) number of new foreclosures and shadow inventory fell to its local
minimum. However, as house prices began to fall in 2007, the negative equity position of
owners caused foreclosures to escalate. Initially, newforeclosures exceeded REOresales
and shadow inventory reached its peak at the beginning of 2009. In a situation where
asset prices are falling dramatically, lenders were trying to sell properties as quickly as
possible. Given the brisk turnover of REO property along with shadow inventory near
Figure 5.
Percentage of transactions
that were ?ips, percentage
of transactions
foreclosures, quarterly
median price and
percentage change in
median price
(2004:q1-2009:q4)
1
0.50
0
0.50
1
P
e
r
c
e
n
t
/
P
c
t
.
c
h
a
n
g
e
2004q1 2006q1 2008q1 2010q1
1,50,000
2,00,000
2,50,000
3,00,000
3,50,000
M
e
d
p
r
i
c
e
M
e
d
i
a
n
p
r
i
c
e
Percent flips Percent foreclosures
Median price Pct. change med. price
JFEP
3,1
60
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
historically high levels, there is further reason to believe that housing prices will
continue to decline in a soft Las Vegas housing market.
5. Establishing causal relationships
The fan graphs discussed in the previous section are suggestive of intertemporal
relationships between quarterly foreclosures (including deed-in-lieu of foreclosure),
median price, the quarterly percentage change in median price and quarterly ?ipping
activity (both buy- and sell-side transactions). To establish whether intertemporal
relationships actually exist, we consider Granger causality tests between any two
variables of interest. We do not consider the level of median price explicitly as it is
subsumed within the percentage change in median price variable. While Granger
causality methodology is a useful starting point to examine intertemporal relationships,
it is limited in its ability to identify structural relationships between more than two
variables at a time. Thus, with this methodology we simply seek to establish whether
there are unilateral, bilateral, or independent relationships between each variable dyad.
Granger’s (1969) de?nition of causality asserts that variable Xcauses variable Yif past
values of Xand Yhelp explain the variation in current Ybetter than previous values of Y
itself. Granger causalityis predicatedona rather simple concept. Abase-line is established
byusingprevious values of Yto explainthe current value of Y. Byaddinglaggedvalues of
another variable X, the subsequent model will explainat least as muchof the variationinY
as the base-line model. If the additional explanatory value of the lagged values of X more
than outweigh the lost degrees of freedom associated with the additional explanatory
variables, variable Xis said to Granger cause variable Y. It is possible that variable Ycan
also Granger cause variable X, suggesting bilateral feedback between the two variables or
a potential third variable that is causing both X and Y. If neither X nor Y Granger cause
each other, then the two variables can be considered independent in the Granger sense of
causation even if the correlation between the two variables is positive.
Figure 6.
Shadow inventory and
median prices
(2000:q1-2009:q4)
12,000
10,000
8,000
6,000
4,000
2,000
S
h
a
d
o
w
i
n
v
e
n
t
o
r
y
2000q1
2002q3
2005q1 2007q3 2010q1
1,50,000
2,00,000
2,50,000
3,00,000
3,50,000
M
e
d
i
a
n
p
r
i
c
e
Shadow inventory Median price
Real estate
bubble
61
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Before testing the interrelationships between the percentage of transactions that are
part of a ?ip, the percentage of transactions that are foreclosures, and the percentage
change in median price, it is important to establish the stationarity of each time series.
Non-stationarity of one or more of the variables could lead to spurious results in any
Granger causality test. To establish stationarity, the ?rst column of Table IV reports
Dickey-Fuller test statistics assuming no drift for each variable (qualitatively similar
results were obtained when allowing for drift). As can be seen in all cases, the null
hypothesis of a unit root, i.e. non-stationarity, cannot be rejected.
It is dif?cult to reconcile how percentage of transactions that are ?ips or foreclosures
can actually be non-stationary in the long run because the two variables are constrained
from above and below. One possibility is that the Dickey-Fuller tests are misleading
because of structural breaks inthe data whichmake the series appear to be non-stationary.
The methodology developed by Zivot and Andrews (1992) tests for non-stationarity after
controlling for any data-determined structural break the procedure discovers. The ?nal
two columns of Table IV report the results of Zivot-Andrews tests assuming a single
structural break in the data. The second column shows that after controlling for a
structural break, the three variables are all stationary. In other words, the Dickey-Fuller
results in column one are likely incorrect.
The ?nal column in Table IV reports the data-determined structural breaks for the
three variables. The structural breaks are remarkably aligned with anecdotal evidence
of when these characteristics of the Las Vegas housing market experienced fundamental
change. For instance, the percentage change in median price and the percentage of
transactions that were ?ips reveal a structural break during the ?rst quarter of 2004,
exactly when claims of many industry observers suggested ?ipping experienced a
fundamental increase in popularity. In addition, the percentage of transactions that were
foreclosures reveals a structural break in the second quarter of 2007, very close to when
industry observers suggest the housing market peaked. As median prices started to fall,
highly leveraged home buyers began to ?nd their mortgages underwater, leading to a
shift in the temporal pattern of foreclosures in the Las Vegas housing market.
The results presented in Table IV suggest that, after controlling for the structural
breaks in the data, all three variables are stationary. Thus, the standard Granger
causality regression, where the current value of the dependent variable is regressed on
the lagged values of the dependent and independent variables, is augmented with a
dummy variable that takes a value of zero before the structural break of the independent
variable and one thereafter.
The results of the Granger causality tests are reported in Table V. Each dyad between
the three variables involves a pair of Granger causalitytests. The ?rst number ina cell uses
onlyone-quarter laggedvalues of bothvariables, andwe label this as short-termcausation.
Variable Dickey-Fuller test Zivot-Andrews test Break point
Percentage ?ips 21.097 25.555
*
2004:q1
Percentage foreclosures 1.274 24.812
*
2007:q2
Percentage change in median price 21.887 24.635
* *
2004:q1
Notes: Signi?cance at:
*
p , 0.05 and
* *
p , 0.07; p-values indicate rejection of the null hypothesis of
a unit root (non-stationarity)
Table IV.
Stationarity tests
JFEP
3,1
62
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The second Granger causality statistic uses four quarters of lagged values of both
variables and we refer to this as longer term causation.
The Granger causality tests in Table V reveal an intriguing set of relationships
between the percentage change in median price, the percentage of transactions that are
?ips and the percentage of transactions that are foreclosures. The ?rst result is that there
seems to be independence between ?ips and foreclosures (i.e. there is no Granger causality
ineither direction). Onthe other hand, ?ips in?uence percentage change inmedianprice in
the longer term, while foreclosures in?uence the percentage change in median price in the
short term. Finally, the last rowin Table Vshows that percentage change in median price
in?uences the other two variables over both one and four quarters.
Thus, while there is some feedback from ?ips and foreclosures on the percentage
change in price, it is evident that the driving force among these three variables is
the percentage change in price. One explanation for this is that percentage change in
price in?uences the pro?tability (or loss) of ?ips and foreclosures. As prices increase,
regardless of their level, individuals attempt to reap pro?ts from house ?ipping. On the
other hand, as housing prices fall, more individuals ?nd their equity eroded and
eventually ?nd their mortgages underwater which might ultimately lead to foreclosure.
A somewhat surprising result from these bivariate tests is the lack of a causal
relationship (in either direction) between ?ips and foreclosures. Conventional wisdom
might suggest that ?ips Granger cause foreclosures, that is, those who bought in the
sell-side ?ip transaction might have overpaid for the property and be more likely to walk
away once underwater. On the other hand, foreclosures might be expected to in?uence
?ips as foreclosures lower the price on the buy-side of the ?ip and make ?ips potentially
more pro?table. However, there is no causal relationship between the two suggesting
that they are independent of each other in the Las Vegas market.
6. Policy discussion and conclusions
Looking backwards, it is easy to trace through the housing bubble in Las Vegas over the
last decade. The percentage change in price was the driving force behind a surge of
?ipping activity that arti?cially boosted demand for housing in the metropolitan area.
This, in turn, ignited further price increases, and home builders responded by
constructing more new homes. Ultimately, growth in the Las Vegas housing stock
outstripped population growth and the resulting moderation in price increases meant
Variable
Percentage
?ips
Percentage
foreclosures
Percentage change median
price
Percentage ?ips 1.47/0.87 2.29/4.22
*
Percentage foreclosures 3.94/1.08 3.97
*
/2.39
Percentage change median
price 5.10
*
/2.57
*
8.45
*
/3.14
*
Notes: Signi?cance at:
*
p , 0.10,
* *
p , 0.05, and
* * *
p , 0.01; the hypothesis tested is that the
column (action) variable Granger causes the row (reaction) variable; values represent F-statistics with
(1,60) and (4,51) degrees of freedom, respectively; in the tests involving the percentage change in
median price, the degrees of freedom are (1,59) and (4,50), respectively; the ?rst number reported uses
one lag to test for Granger causality; the second number reported uses four lags to test for Granger
causality; row variables are action variables and column variables are reaction variables
Table V.
Granger causality tests
Real estate
bubble
63
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
that many ?ips were no longer pro?table. As ?ipping activity slowed considerably,
house prices began to fall. Eventually some homeowners found their mortgages
underwater and defaulted on their notes. These foreclosures led to a further decline in
prices causing more foreclosures in the area.
That price changes drive foreclosures is consistent with Elul et al. (2010) and
Bhutta et al. (2010) who ?nd that negative equity is a primary reason for default.
Together, this work suggests that loan modi?cation programs will necessarily have
limited success in curbing new foreclosures, and steps must be taken to ?rm up prices.
One possibility is to expand resources like HUD’s Neighborhood Stabilization Program
(http://hudnsphelp.info/). This program shores up demand of foreclosed properties by
providing ?nancial assistance to ?rst-time home buyers. Still another Neighborhood
Stabilization Program in?uences supply of foreclosed homes by granting funds to
government entities for the purpose of demolishing blighted neighborhoods. Detroit, for
example, plans to knock down 3,000 homes by September 2010 using federal
government funds. Moreover, Mayor Dave Bing has promised to tear down 10,000
structures in his ?rst termin of?ce to “right-size” Detroit and align housing needs with a
shrinking city population (Kellogg, 2010).
Giventhat excess supplyis part of the foreclosure problem, it is perhaps surprisingthat
new home building continues in Las Vegas. As one Las Vegas builder noted (Streitfeld,
2010), “We’re buildingthembecause we’re sellingthem.” Yet newhome buildingcontinues
to add to the problem of excess supply, falling prices and foreclosed homes. In
circumstances like these, local government units might develop policies that encourage
renovationof properties andrehabilitationof neighborhoods. While declaringa temporary
moratoriumon newhomes may diminish local tax and permit revenues generated by new
housing construction in the short run, the offset is that city will not have to contend with
the costs associated with abandoned homes and blighted neighborhoods.
At least two other lessons can be derived from the Las Vegas housing market bubble.
First, ?ipping activity contributed to rising home prices, and given asymmetric
information, it might be prudent to alert potential homebuyers of legal ?ipping activity.
One way to do that is to require multiple listing service listings to include information on
when the current owner bought the property and whether the current owner lives in the
home. This solutiontothe asymmetric informationproblemis similar tothe requirement in
several states that sellers divulge information that they are an agent/owner of a property.
(See Levitt andSyverson(2008) for market distortions relatedto the agent/owner problem.)
The second lesson is that municipal governments must be consistent in their record
keeping. ClarkCountyinchanging the Ftransactioncode fromdeed-in-lieu of foreclosure
to foreclosure resale makes it dif?cult to do any meaningful time comparisons. More
importantly, up to the last three years, foreclosure resales were coded as arms-length
transactions (R). But in the recent downturn, 90 percent of foreclosure resales were coded
F. The upshot is that tax valuation models and many price indexes will not include these
F transactions. In the case of tax valuation models, excluding foreclosure resales may
seriously bias upward the county’s assessment of market value.
Finally, looking forward, it will be important to focus on underlying structural
inter-temporal relationships, perhaps with the use of vector autoregressive models. So,
for example, how did mortgage rates or easy credit in?uence prices, ?ipping activity
and foreclosures? Or, did pricing dynamics differ when prices were going up versus
down? Answers to questions such as these will be left for future research.
JFEP
3,1
64
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
References
Bhutta, N., Dokko, J. and Shan, H. (2010), “The depth of negative equity and mortgage default
decisions”, Finance and Economics Discussion Series No. 2010-35, working paper, Federal
Reserve Board, Washington, DC.
Brueckner, J.K. (1980), “A vintage model of urban growth”, Journal of Urban Economics, Vol. 8
No. 3, pp. 389-402.
Capozza, D.R. and Helsley, R.W. (1989), “The fundamentals of land prices and urban growth”,
Journal of Urban Economics, Vol. 26 No. 3, pp. 295-306.
Case, K.E. and Quigley, J.M. (2008), “How housing booms unwind: income effects, wealth effects,
and feedbacks through ?nancial markets”, European Journal of Housing Policy, Vol. 8
No. 2, pp. 161-80.
Depken, C.A. II, Hollans, H. and Swidler, S. (2009), “An empirical analysis of residential property
?ipping”, Journal of Real Estate Finance & Economics, Vol. 39 No. 3, pp. 248-63.
Elul, R., Souleles, N., Glennon, D. and Hunt, R. (2010), “What triggers mortgage default?”, Federal
Reserve Bank of Philadelphia Working Paper No. 10-13, Federal Reserve Bank of
Philadelphia, Philadelphia, PA.
Granger, C. (1969), “Investigating causal relations by econometric models and cross spectral
methods”, Econometrica, Vol. 37 No. 3, pp. 424-38.
Kellogg, A. (2010), “Detroit shrinks itself, historic homes and all”, available at:http://online.wsj.
com/article/SB10001424052748703950804575242433435338728.html (accessed24 July2010).
Levitt, S. and Syverson, C. (2008), “Market distortions when agents are better informed: the value
of information in real estate transactions”, The Review of Economics and Statistics, Vol. 90
No. 4, pp. 599-611.
Shiller, R. (2009), “Unlearned lessons from the housing bubble”, The Economists’ Voice, Vol. 6
No. 7, p. 1.
Streitfeld, D. (2010), “Building is booming in a city of empty houses”, available at: www.nytimes.
com/2010/05/16/business/16builder.html
Wheaton, W. andNechayev, G. (2008), “The 1998-2005 housing‘bubble’ andthe current ‘correction’:
what’s different this time?”, Journal of Real Estate Research, Vol. 30 No. 1, pp. 1-26.
Zivot, D. and Andrews, D. (1992), “Further evidence on the great crash, the oil price shock and the
unit root hypothesis”, Journal of Business and Economic Statistics, Vol. 10 No. 3, pp. 251-70.
Further reading
Dickey, D.A. and Fuller, W.A. (1979), “Distribution of the estimators for autoregressive time series
with a unit root”, Journal of the American Statistical Association, Vol. 74 No. 366, pp. 427-31.
Immergluck, D. and Smith, G. (2006), “The external costs of foreclosure: the impact of
single-family mortgage foreclosures on property values”, Housing Policy Debate, Vol. 17
No. 1, pp. 57-79.
Lin, Z., Rosenblatt, E. and Yao, V. (2009), “Spillover effects of foreclosure on neighborhood
property values”, Journal of Real Estate Finance & Economics, Vol. 38 No. 4, pp. 387-407.
Corresponding author
Steve Swidler can be contacted at: [email protected]
Real estate
bubble
65
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Katrin B. Anacker, Laurie A. Schintler. 2015. Flip that house: visualising and analysing potential real
estate property flipping transactions in a cold local housing market in the United States. International
Journal of Housing Policy 1-19. [CrossRef]
2. References 294-311. [CrossRef]
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
4
0
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_895891351.pdf