Optimal Listing Strategy in Selling
Residential Real Estate1
Christopher Lako, Zheng Liu, and Charles McKinney2
Working Paper
March 3, 2014
Abstract
The sale price of a home is determined by agent bidding around the listing price.
Setting too high of an initial listing price will discourage market participants
from bidding, and setting too low of an initial listing price will encourage greater
market participation, but agents might not bid far above (if at all) the initial
listing price. Using a multiple-listing service dataset encompassing two years of
sale and listing data for the United States, we analyze the impact of the initial
listing strategy on the realized sale price. We introduce a convexity measure to
determine the rate of sale price degradation based on the degree of over or
under-listing. We find that the dominant listing strategy varies by market with a
lower listing relative to value being preferred in fast-appreciating or hot markets
and a higher listing relative to value being preferred in cold markets.
1
The views expressed in this paper are those of the authors and are not necessarily those of
Freddie Mac, Freddie Mac’s Board of Directors, or any of Freddie Mac’s regulators.
2
Christopher Lako is a senior economist, Zheng Liu is a manager, and Charles McKinney is a
senior director in Freddie Mac’s Single Family Servicing and REO division. Please direct all
correspondence to Christopher Lako at Freddie Mac, 8200 Jones Branch Drive, McLean, VA
22102, or at [email protected]. We are grateful to Hanqing Zhou, Yuan Yuan,
Alexey Serednyakov, and Mingxuan Zuo for feedback and advice on our model and
methodology. We would also like to thank Eric Will, Dave Wendling, Rhonda Montgomery,
and Tammy Vincent for sharing business insights that informed the research and development
of the topics discusses in the paper.
Page 1
Introduction
When selling a home, a key problem for most sellers is deciding the initial
listing price of their property. Listing too low will generate interest in a property
however buyers may not bid above the listing price. Conversely, if the list price
is set too high, a higher price may be realized but a lower price may also be
realized due to fewer visits from prospective buyers. Furthermore, the level of
increased or decreased interest in a property based on the listing price may not
be constant across disparate geographic areas or across properties with different
characteristics or values. There is a finite but dynamic pool of buyers in a given
market with a distinct set of preferences and willingness to pay.
In financial markets, convexity is a common barometer of risk in a
security. As rates rise or fall so does return. A peak is observed, with differing
degrees of risk on the low and high side of the peak. For some securities, return
falls at a slower rate to the right of the peak, indicating there is less risk on the
high side, while for other securities return falls at a decreased rate on the left side
of the peak. The return for selling a home can be thought of similarly, based on
the level of over- or under-listing different causing different returns to be
realized. Using metrics to measure the degree of over or under-listing and the
realized return, a “convexity” curve can be constructed using a pool of
geospatially controlled properties. The location of the peak of the curve informs
us if it is optimal to over or under list a property and to what degree. This is the
first paper that borrows a concept from financial markets to explain a
phenomenon in real estate finance for housing market microstructure.
Page 2
As shown by Case and Shiller (2012), there exist hot and cold markets,
which have different characteristics and dynamics. We find that in a hot market,
sellers should list lower to capture increased market efficiency. A low listing in a
hot market leads to increased demand from buyers and buyers will bid over the
list price, and as shown in this paper, will lead to a maximal return for the seller.
In a cold market, a seller needs to list higher due to decreased demand and to
avoid being taken advantage of by low bids.
This paper starts with an overview of previous research on the role of the
listing price in the sale of residential real estate. We show that unlike most of the
previous research we use a longer and richer time series across multiple
geographical areas. We then describe our methodology for constructing a
“convexity” curve followed by a description of our data. Finally, we provide our
findings from performing multiple analyses with different types of data
stratification.
Background
Much of the previous research on the effects of listing price on sale price
have made conclusions using thin datasets that cover one geographic area. For
example, Knight (2002) looked at sales in Stockton, CA; Yavas and Yang (1995),
sales in State College, PA; Haurin et al. (2010), sales in Columbus, OH; Anglin et
al. (2003), sales in Arlington, TX; Green and Vandell (1995) covered Madison, WI;
Miller and Skalrz (1987) looked at Honolulu, HI; and Han and Strange (2012)
Page 3
looked at one anonymous metropolitan area. While the conclusions drawn from
these studies inform a body of knowledge, the conclusions may not generalize
across all geographic areas or across time for the housing markets studied. Case
and Shiller (1988) studied four geographic areas, with two areas in a boom, one a
cold market, and the other a stable market, and they suggested that macro
variables only partially explain housing dynamics and that local market
conditions play a major factor.
As has been shown by papers such as Case and Shiller (1988) and Han and
Strange (2013), the list price is not a ceiling, sales do occur, and with great
frequency in some markets, with a sale price above the initial list price. Previous
work by Haurin et al (2010) and Yavas and Yang (1995) employed models that
assumed the list price is a ceiling. Case and Shiller (1988) showed that in cooler
markets, such as Boston following its housing bubble and Milwaukee, WI, very
few properties sell above the listing price (0.5 and 3.3% respectively). However,
for warmer markets, such as Anaheim and San Francisco, 6.3 and 9.8%
respectively of sales occur above the listing price. We also show that in our cool
market, Cleveland, OH, very few properties sell above the listing price, however
in our warm market, Los Angeles, a significant share of properties sell above the
listing price.
Model and Methodology
Measuring the degree of over or under-listing and the return for a
property is a non-trivial task. If markets are efficient, one can assume that the
sale price converges to the “true” value of the property. Thus, the degree of over
Page 4
or under-listing can be represented by the initial list price to sale price ratio. The
realized return can be thought of as the sale price relative to the estimated value
of the home. An automated valuation model (AVM) that predicts what the home
should sell for can be used as a benchmark for the estimated value of the home.
For an extensive analysis on a property’s return using a sale price to AVM ratio
please reference Zhou et al (2013).
Another intuitive approach would be to look at the relationship between
initial listing price to AVM and sale price to AVM. Miller and Sklarz (1987)
studied the relationship between initial listing price to AVM and sale price to
AVM and found a relatively linear relationship. Without seeing curvature in the
relationship, an optimal list price strategy cannot be inferred. We know that the
relationship cannot be linear otherwise it would imply that every home should
be listed well above the value of the home. We do not study why the relationship
found between list price to AVM and sale price to AVM is linear in this paper.
In order to construct a smooth curve, we first remove outliers from the
data. We start with tuples of the following form:
𝐿𝑖𝑠𝑡 𝑃𝑟𝑖𝑐𝑒𝑖𝑡
𝑆𝑎𝑙𝑒 𝑃𝑟𝑖𝑐𝑒𝑖𝑡+𝑘
�
�
�
�
𝑡+𝑘 ,
𝐴𝑉𝑀𝑖𝑡
𝑆𝑎𝑙𝑒 𝑃𝑟𝑖𝑐𝑒𝑖
where 𝑖 is a property, 𝑡 is the time of listing, and 𝑡 + 𝑘 is the time of sale, where
𝑘 > 0. The tuples are partitioned into bins in increments of 500bps by the list to
sale ratio. Within each bin, studentized residuals are calculated for each datapoint. If the data-point’s studentized residual is greater than two in absolute
value the record is removed. Next, using the bins in 500bps increments we
Page 5
remove bins from the left and right tails that contain less than 50 data-points, to
remove noise effects in the construction of the “convexity” curve. Once the
outliers and thin data buckets are removed, we apply the non-parametric
regression method of LOESS, with a smoothing parameter of 0.50. As described
by Cleveland and Devlin (1988), the LOESS method with a smoothing parameter
of 0.50 will at each data-point use 50% of the total data-points with a weight
assigned based on the inverse of the distance from the sample data-point based
on a tri-cube method. Then weighted least squares is performed at each datapoint.
Data
Using multiple geographic areas and different time periods, differences in
optimal listing price strategy can be observed. This study uses MLS data from
the Los Angeles and Cleveland CBSAs for sales from January 2012 to December
2013. The MLS data contains information such as initial listing price, final listing
price, sale price, initial list date, sale date, and identification describing if the
property is a REO or short-sale. For Cleveland we observe 25,252 sales and for
Los Angeles we observe 68,973 sales.
Tables 1 and 2 present descriptive statistics for the two samples, with
overall statistics, and statistics broken out by if the listing had a list price
adjustment, property condition, distressed status (normal and REO), and the
year of the sale. As expected, Los Angeles and Cleveland have different market
dynamics and a different mix of properties. In Los Angeles, 31% of listings incur
a list price change, with an average list price change of -4%, while in Cleveland
Page 6
46% of listings incur a list price change, with an average change of -10%. In
Cleveland, the average AVM (labeled as HVE - see below for a description) at
time of listing is $167,000 and in Los Angeles the average AVM at listing is
$385,000. Properties in Los Angeles sell much quicker at 109 days on average
compared to 144 days in Cleveland. Homes in Cleveland tend to be newer and
larger (by both average living area and average number of rooms). Interestingly,
for the populations for which condition is known, both Cleveland and Los
Angeles have similar distributions of condition.
The data are merged with Freddie Mac’s Appraisal Portal to obtain
information about property condition. The Appraisal Portal sample only
includes properties where the buyer applied for a conforming loan.
Consequently, the sample that includes condition is biased. Furthermore, not all
the data that is in the Appraisal Portal matches to the MLS because of coverage
differences in the MLS and data differences for address information and listing
date. In Los Angeles 26% of the sales in our data have a match to the Appraisal
Portal and in Cleveland 31% of sales are matched with a condition in the
Appraisal Portal. Properties with a condition of C1 are new construction, and
thus are not in our sample, C2 and C3 represent good condition, and C4, C5, and
C6 represent worse than average condition. In Los Angeles, 12% have a
condition of C2, 60% are assigned a condition of C3, and 26% have a condition of
C4. Similarly, in Cleveland 11% have a condition of C2, 63% have a condition of
C3, and 27% have a condition of C4. The remaining properties have a condition
of C5 or C6.
To obtain a property value at the time of sale, we use Freddie Mac’s Home
Value Explorer (HVE), which is an automated valuation model (AVM). HVE
Page 7
uses Freddie Mac's unique proprietary algorithm that blends model estimates
returned by our repeat sales model and hedonic model, which is considered our
combining process. HVE provides extensive coverage of all 50 states and more
than 3100 counties with its database of approximately 81 million property
records (http://www.freddiemac.com/hve/hve.html).
Miller and Skalrz (1987) found very few properties listed above the list
price equals sale price line, indicating that almost all properties in their sample
from Honolulu sold below the initial listing price. We looked at data for sales
from January to June 2013 for the Cleveland, OH and Los Angeles, CA CBSAs
and with properties with HVE values in the middle 30% of the HVE value
distribution, and constructed two scatterplots of the initial list price to HVE and
sale price to HVE relationship (Charts 1 and 2). Chart 1 shows that for Cleveland,
there are indeed very few properties above the list price equals sale price line,
similar to the Miller and Skalrz 1987 study. However, Chart 2 shows that in Los
Angeles we see roughly an even mix above and below the list price equals sale
price line, indicating that in Los Angeles it is very common for properties to have
a sale price above the initial listing price.
Results
Comparing two disparate markets, Los Angeles and Cleveland, we find
that in a strong and dynamic market, a seller can correct a low listing due to
market demand, however in a slow and tepid market a seller has to take a risk
and set a higher listing price and accept a slightly longer days on market to avoid
being taken advantage of by low bidders. In a slow market, a discount below the
Page 8
listing price is expected by buyers while in a hot markets, buyers enter into
competition and bid above the list price if the list price is advantageous. We
compare the overall markets to subgroups defined by distressed status, property
value buckets, list price adjusted and non-list price adjusted, and sale date
groups.
By Distressed Status
Using the curve construction technique described in the Model and
Methodology section, we start by observing convexity curves for Los Angeles
and Cleveland for all sales in 2012 and 2013 and broken out by normal sales (not
a foreclosure, short-sale, or REO) and REO sales (Charts 3 and 4). Chart 3 shows
that for Cleveland it is optimal to have an initial listing price to sale price ratio of
l.04, with less risk to the right side (since the slope of the line to the right of the
peak has a shallower slope compared to the line to the left side of the peak),
while for Los Angeles the optimal ratio is around 1.00, with equal risk to the high
and low side. For Los Angeles we observe that there are very minimal
differences for the return and listing strategy between REO sales and normal
sales, while in Cleveland REO sales have a much lower return at the peak and
the peak is to the right of the peak for normal sales. This indicates the REO
properties should be listed higher relative to value compared to normal
properties in Cleveland.
The greater spread in return between REO and normal properties in
Cleveland is due to many factors. From Tables 1 and 2 we observe that for the
condition controlled population, Cleveland REO properties are in worse
condition relative to normal properties, while in Los Angeles REO properties
Page 9
have roughly the same distribution of condition as normal properties. As noted
by Zhou et al (2013), for REO properties the sale price to AVM ratio improves as
condition improves.
By Property Value
Charts 5 and 6 show the convexity curves for Cleveland and Los Angeles
where the data are grouped into buckets by median HVE at time of listing
($146,000 for Cleveland and $365,000 for Los Angeles). For Cleveland we observe
that the peak moves slightly to the left for higher value properties and for Los
Angeles we observe that the peak is unaffected by stratifying the data by
property value. This can be explained by Los Angeles being an active market
where regardless of property value there will be an active pool of buyers
whereas for Cleveland, higher value properties may attract a larger number of
buyers compared to lower value properties due to the overall thinness of
demand.
Cleveland having a higher peak for higher value properties is consistent
with Zhou et al (2013) wherein the authors showed that REO properties with a
higher property value have a higher sale price to AVM ratio before controlling
for property condition. In Los Angeles, however, we see the opposite effect, in
that lower value properties have a higher peak relative to higher value
properties. This indicates that lower value properties, if listed optimally, will
have a higher return relative to AVM than will be observed for higher value
properties.
Page 10
By List Price Adjustment
If a property’s price is set such that suitable offers are not received and the
price has to be adjusted the return on the property will be affected due to
possible longer days on market and other factors. Knight (2002) found that
properties with high markups and vacant homes were most likely to have a list
price revision. Charts 7 and 8 show the data for Cleveland and Los Angeles
bucketed by properties which did and did not have list price revisions.
Interesting, list price revision curves are at the same level of return or higher
than the curves for properties that did not have list price adjustments. Another,
interesting finding is that there is a population of homes that had a list price
revision and had an initial list price to sale price ratio less than one.
Tables 3 and 4 show the distribution of the final list price to initial list
price ratio for homes that had a list price revision and that had an initial list price
to sale price ratio less than one. In approximately 75% of the sales for both
Cleveland and Los Angeles, the initial list prices were adjusted upwards. Tables
5 and 6 show the distribution of final list price to initial list price ratio for homes
that had a list price revision and that had an initial list price to sale price ratio
greater than one. We observe in Cleveland, a less active market, that if the list
price has to be adjusted downwards, the adjustment is slightly greater than in
Los Angeles, which is a more dynamic market.
Constructing the convexity curves using the final list price to sale price
ratio as the independent variable (Charts 9 and 10), we observe that the curve for
properties that had a list price adjusted is pushed inwards and below the curve
for properties that are not list price adjusted. We observe that the optimal list
Page 11
price to sale price ratio is largely unaffected by the list price adjustment decision.
However, we observe that the return is inferior compared to properties that are
not list price adjusted. Furthermore, the gap in return between homes which are
and are not list price adjusted is greatest in Cleveland. Sellers are penalized more
for mispricing their homes in a market that is less dynamic. We do not offer
perspective on the optimal list price adjustment strategy or if it is optimal to even
have a list price adjustment.
By Sale Date
Chart 11 shows that in the first half of 2012, Cleveland and Los Angeles
had similar 12 month growth rates in HPI, however between the 2nd half of 2012
and the end of 2013, the 12 month change in HPI grew much faster in Los
Angeles, reaching a maximum of almost 20%, compared to less than 5% in
Cleveland. As the market improved at an increased rate in Los Angeles we
would expect to see a greater effect over time in the convexity curve in Los
Angeles.
Charts 12 and 13 show the convexity curves broken into half year
increments. We observe that for Cleveland, the curve for the first half of 2012 had
a peak to the right of the curves for the second half of 2012 and the two halves of
2013. After the middle of 2012, 12 month HPI growth became positive in
Cleveland and remained stagnant thereafter. It appears that the convexity curve
shifted in response to a change in the level of HPI growth. For Los Angeles we
observe a similar phenomenon. We observe shifts to the left in the curves as 12
month HPI change was growing, with the largest shift observed when the largest
increase in 12 month HPI growth rate was observed. Once the HPI growth rate
Page 12
started to stabilize and fall slightly in the second half of 2013 we observe a slight
leftward shift in the peak of the convexity curve.
From these charts we observe that the health of a market, as measured by
house price change, affects the optimal listing strategy. As a market is improving,
one can list a property slightly lower to allow the market to correct for
underpricing by the seller.
Conclusion
Building on past research on optimal listing price setting, we introduce a
convexity measure to analyze the risk profile of over and under listing.
Analyzing two disparate markets, one hot and one cold, we find that the optimal
listing strategy varies by market. In a hot market a seller should list slightly
lower relative to value compared to a seller in a cold market. A hot market is
more dynamic and efficient, allowing a seller to achieve a better price by enticing
a bidding war. In a cold market, sellers need to set a higher price as offers below
list price are more commonplace.
We find that the value of a property has at most a minimal effect on the
initial listing strategy. The largest driver that we found for the optimal initial
listing strategy is the health of the local market. As HPI is growing quickly, the
optimal listing strategy converges to a lower listing price being optimal.
Furthermore, the optimal initial listing price strategy is unaffected by having a
list price adjustment and that the list price adjusted curve has a maximal return
that is below the non-list price adjusted population. We leave an analysis of
optimal list price adjustment process as a topic for a future paper.
Page 13
References
Anglin, Paul M., Ronald Rutherford, and Thomas M. Springer. "The trade-off
between the selling price of residential properties and time-on-the-market: The
impact of price setting." The Journal of Real Estate Finance and Economics 26.1
(2003): 95-111.
Case, Karl E., and Robert J. Shiller. "The behavior of home buyers in boom and
post-boom markets." (1989).
Case, Karl E., Robert J. Shiller, and Anne Thompson. What have they been thinking?
Home buyer behavior in hot and cold markets. No. w18400. National Bureau of
Economic Research, 2012.
Cleveland, William S., and Susan J. Devlin. "Locally weighted regression: an
approach to regression analysis by local fitting." Journal of the American Statistical
Association 83.403 (1988): 596-610.
Green, Richard K., and Kerry D. Vandell. Optimal asking price and bid acceptance
strategies for residential sales. University of Wisconsin--Madison, School of
Business, 1995.
Han, Lu, and William Strange. What is the Role of the Asking Price for a House?.
working paper, 2012.
Han, Lu, and William C. Strange. "Bidding Wars for Houses." Real Estate
Economics (2013).
Haurin, Donald R., et al. "List prices, sale prices and marketing time: an
application to US housing markets." Real Estate Economics 38.4 (2010): 659-685.
Knight, John R. "Listing price, time on market, and ultimate selling price: Causes
and effects of listing price changes." Real estate economics 30.2 (2002): 213-237.
Merlo, Antonio, and Francois Ortalo-Magne. "Bargaining over residential real
estate: evidence from England." Journal of Urban Economics 56.2 (2004): 192-216.
Miller, Norman G., and Michael A. Sklarz. "Pricing strategies and residential
property selling prices." Journal of Real Estate Research 2.1 (1987): 31-40.
Page 14
Yavas, Abdullah, and Shiawee Yang. "The strategic role of listing price in
marketing real estate: theory and evidence." Real Estate Economics 23.3 (1995): 347368.
Zhou, Hanqing, Yuan Yuan, Christopher Lako, Michael Sklarz, and Charles
McKinney. Foreclosure Discount: Definition and Dynamic Patterns. working paper,
2013.
Page 15
Cleveland CBSA Jan-Jun 2013 Normal Sales with Listing HVE in [124k,180k]
Sale to HVE at Listing
2.00
1.75
1.50
1.25
1.00
0.75
0.50
0.50
0.75
1.00
1.25
1.50
1.75
2.00
Initial Listing Price to HVE at Listing
Chart 1: An empirical relationship for the Cleveland CBSA of the initial listing price to HVE and the sale price to HVE,
with both HVEs constructed at the time of listing. The data is from the MLS with sales from January to June 2013 and with
a HVE value between $124,000 and $180,000.
Page 16
Los Angeles Jan-Jun 2013 Normal Sales with Listing HVE in [327k,442k]
Sale to HVE at Listing
2.00
1.75
1.50
1.25
1.00
0.75
0.50
0.50
0.75
1.00
1.25
1.50
1.75
2.00
Initial Listing Price to HVE at Listing
Chart 2: An empirical relationship for the Los Angeles CBSA of the initial listing price to HVE and the sale price to HVE,
with both HVEs constructed at the time of listing. The data is from the MLS with sales from January to June 2013 and with
a HVE value between $327,000 and $442,000.
Page 17
Cleveland CBSA - Market Data
Sale to HVE at Sale
1.50000
1.25000
1.00000
0.75000
0.50000
0.50
0.75
1.00
1.25
1.50
1.75
2.00
Initial Listing Price to Sale Price
type
All Sales
Normal Sales
REO Sales
Chart 3: Convexity curves for the Cleveland CBSA with sales from the MLS between January 2012 and December 2013.
The data are shown for all sales, normal sales, and REO sales.
Page 18
Los Angeles CBSA - Market Data
Sale to HVE at Sale
1.50000
1.25000
1.00000
0.75000
0.50000
0.50
0.75
1.00
1.25
1.50
1.75
2.00
Initial Listing Price to Sale Price
type
All Sales
Normal Sales
REO Sales
Chart 4: Convexity curves for the Los Angeles CBSA with sales from the MLS between January 2012 and December 2013.
The data are shown for all sales, normal sales, and REO sales.
Page 19
Cleveland CBSA - Market Data - Normal Sales Stratified by Listing HVE Bucket Median ($146,000)
Sale to HVE at Sale
1.25000
1.00000
0.75000
0.50000
0.50
0.75
1.00
1.25
1.50
1.75
2.00
Initial Listing Price to Sale Price
type
Normal Sales
Normal Sales Above Median
Normal Sales Below Median
Chart 5: Convexity curves for the Cleveland CBSA with sales from the MLS between January 2012 and December 2013.
The data are shown for normal sales and normal sales above and below the median HVE at time of listing ($146,000).
Page 20
Los Angeles CBSA - Market Data - Normal Sales Stratified by Listing HVE Bucket Median ($365,000)
Sale to HVE at Sale
1.25000
1.00000
0.75000
0.50000
0.50
0.75
1.00
1.25
1.50
1.75
2.00
Initial Listing Price to Sale Price
type
Normal Sales
Normal Sales Above Median
Normal Sales Below Median
Chart 6: Convexity curves for the Los Angeles CBSA with sales from the MLS between January 2012 and December 2013.
The data are shown for normal sales and normal sales above and below the median HVE at time of listing ($365,000).
Page 21
Cleveland CBSA - Market Data - Normal Sales Bucketed By List Price Change versus No Change
Sale to HVE at Sale
1.25000
1.00000
0.75000
0.50000
0.50
0.75
1.00
1.25
1.50
1.75
2.00
Initial Listing Price to Sale Price
type
Normal Sales
Normal with List Price Change
Normal without List Price Change
Chart 7: Convexity curves for the Cleveland CBSA with sales from the MLS between January 2012 and December 2013.
The data are shown for normal sales and normal sales with and without a list price change.
Page 22
Los Angeles CBSA - Market Data - Normal Sales Bucketed By List Price Change versus No Change
Sale to HVE at Sale
1.25000
1.00000
0.75000
0.50000
0.50
0.75
1.00
1.25
1.50
1.75
2.00
Initial Listing Price to Sale Price
type
Normal Sales
Normal with List Price Change
Normal without List Price Change
Chart 8: Convexity curves for the Los Angeles CBSA with sales from the MLS between January 2012 and December 2013.
The data are shown for normal sales and normal sales with and without a list price change.
Page 23
Cleveland CBSA - Market Data - Normal Sales Bucketed By List Price Change versus No Change
Sale to HVE at Sale
1.25000
1.00000
0.75000
0.50000
0.50
0.75
1.00
1.25
1.50
1.75
2.00
Final Listing Price to Sale Price
type
Normal Sales
Normal with List Price Change
Normal without List Price Change
Chart 9: Convexity curves for the Cleveland CBSA with sales from the MLS between January 2012 and December 2013.
The convexity curve uses the final listing price in the numerator of the independent variable in this chart in lieu of the
initial listing price. The data are shown for normal sales and normal sales with and without a list price change.
Page 24
Los Angeles CBSA - Market Data - Normal Sales Bucketed By List Price Change versus No Change
Sale to HVE at Sale
1.25000
1.00000
0.75000
0.50000
0.50
0.75
1.00
1.25
1.50
1.75
2.00
Final Listing Price to Sale Price
type
Normal Sales
Normal with List Price Change
Normal without List Price Change
Chart 10: Convexity curves for the Los Angeles CBSA with sales from the MLS between January 2012 and December 2013.
The convexity curve uses the final listing price in the numerator of the independent variable in this chart in lieu of the
initial listing price. The data are shown for normal sales and normal sales with and without a list price change.
Page 25
Cleveland CBSA
Los Angeles CBSA
20.00%
15.00%
10.00%
5.00%
0.00%
-5.00%
-10.00%
1/2012
7/2012
1/2013
7/2013
Chart 11: Single-Family distressed excluded 12-month change in the Corelogic home
price index for the Cleveland and Los Angeles CBSAs
Page 26
Cleveland CBSA - Market Data - Normal Sales Stratified by Sale Date
Sale to HVE at Sale
1.25000
1.00000
0.75000
0.50000
0.50
0.75
1.00
1.25
1.50
1.75
2.00
Initial Listing Price to Sale Price
type
Normal Sales
Normal Sales in 2012H1
Normal Sales in 2012H2
Normal Sales in 2013H1
Normal Sales in 2013H2
Chart 12: Convexity curves for the Cleveland CBSA with sales from the MLS between January 2012 and December 2013.
The data are shown for normal sales and normal sales stratified by sale date in half year buckets.
Page 27
Los Angeles CBSA - Market Data - Normal Sales Stratified by Sale Date
Sale to HVE at Sale
1.25000
1.00000
0.75000
0.50000
0.50
0.75
1.00
1.25
1.50
1.75
2.00
Initial Listing Price to Sale Price
type
Normal Sales
Normal Sales in 2012H1
Normal Sales in 2012H2
Normal Sales in 2013H1
Normal Sales in 2013H2
Chart 13: Convexity curves for the Los Angeles CBSA with sales from the MLS between January 2012 and December 2013.
The data are shown for normal sales and normal sales stratified by sale date in half year buckets.
Page 28
All
Number of Sales
Percent of Sales with List Price Change
Percent of Sales which are REO
Percent of Sales which are Normal
Percent of Sales with Appraisal Portal Match
Percent with C2 Condition
Percent with C3 Condition
Percent with C4 Condition
Percent Single-Family
Change Between Initial and Final List Price
Average Number of Rooms
Average Living Area
Percent Built Before 1940
Percent Bult Between 1940 and 1970
Percent Built Between 1970 and 2000
Percent Built After 2000
Average Days on Market
Average Initial Listing Price
Average Sale Price
Average HVE at Listing
Average HVE at Sale
Average Initial List Price to Sale Price
Average Initial List Price to HVE at Listing
Average Sale Price to HVE at Sale
25,252
46%
9%
85%
31%
11%
63%
27%
95%
3.24
1,870
14%
41%
32%
13%
144
$183,891
$167,833
$167,588
$169,576
1.11
1.08
0.97
Cleveland CBSA
List Price Change Appraisal Condition
No
Yes
C2, C3
C4
13,674
11,578
5,749
2,079
43%
48%
8%
9%
5%
9%
87%
83%
91%
85%
32%
30%
11%
10%
63%
61%
25%
29%
95%
94%
99%
100%
-10%
-7%
-10%
3.25
3.24
3.36
3.27
1,874
1,866
2,073
1,765
14%
14%
14%
15%
40%
42%
32%
54%
32%
32%
35%
27%
14%
13%
19%
4%
90
209
127
137
$181,574 $186,629 $226,188 $160,265
$173,906 $160,660 $209,952 $144,354
$169,878 $164,883 $197,579 $153,768
$171,739 $167,021 $200,466 $155,333
1.05
1.19
1.08
1.12
1.05
1.12
1.14
1.03
0.99
0.94
1.05
0.91
Property Type
Normal
REO
21,480
2,172
45%
47%
33%
11%
63%
25%
95%
-9%
3.25
1,895
14%
40%
33%
13%
144
$192,373
$176,228
$171,548
$173,732
1.11
1.10
1.00
21%
6%
51%
43%
95%
-15%
3.21
1,772
12%
45%
31%
12%
110
$127,534
$115,512
$147,172
$147,990
1.13
0.85
0.76
Year Sold
2012
2013
11,618
13,634
50%
43%
9%
8%
84%
86%
28%
34%
12%
10%
61%
63%
27%
26%
95%
94%
-11%
-9%
3.23
3.26
1,865
1,876
13%
15%
41%
41%
33%
31%
13%
14%
155
135
$182,572 $185,015
$164,677 $170,522
$164,542 $170,183
$166,065 $172,568
1.13
1.10
1.09
1.07
0.97
0.97
Table 1: Descriptive statistics for the Cleveland CBSA with data from the MLS for sales between January 2012 and
December 2013. The data are stratified by list price change status, appraisal condition (C2 and C3 represent average and
above average condition and C4 represents below average condition), property type (normal and REO), and year sold.
Page 29
All
Number of Sales
Percent of Sales with List Price Change
Percent of Sales which are REO
Percent of Sales which are Normal
Percent of Sales with Appraisal Portal Match
Percent with C2 Condition
Percent with C3 Condition
Percent with C4 Condition
Percent Single-Family
Change Between Initial and Final List Price
Average Number of Rooms
Average Living Area
Percent Built Before 1940
Percent Bult Between 1940 and 1970
Percent Built Between 1970 and 2000
Percent Built After 2000
Average Days on Market
Average Initial Listing Price
Average Sale Price
Average HVE at Listing
Average HVE at Sale
Average Initial List Price to Sale Price
Average Initial List Price to HVE at Listing
Average Sale Price to HVE at Sale
68,973
31%
9%
81%
26%
12%
60%
26%
89%
3.00
1,580
18%
56%
20%
6%
109
$401,556
$393,687
$385,314
$396,625
1.02
1.04
1.00
Los Angeles CBSA
List Price Change Appraisal Condition
No
Yes
C2, C3
C4
47,913
21,060
12,896
4,693
28%
28%
9%
10%
8%
7%
82%
79%
91%
90%
27%
23%
12%
12%
60%
62%
26%
27%
89%
90%
96%
97%
-4%
-3%
-4%
2.95
3.11
3.02
2.96
1,554
1,639
1,628
1,560
19%
17%
19%
17%
56%
55%
58%
66%
19%
22%
18%
16%
5%
6%
5%
1%
86
163
73
84
$395,255 $415,891 $479,713 $438,634
$396,377 $387,569 $476,101 $433,710
$384,944 $386,155 $436,854 $428,503
$395,373 $399,473 $447,640 $440,142
1.00
1.08
1.01
1.01
1.03
1.08
1.11
1.03
1.01
0.98
1.08
0.99
Property Type
Normal
REO
56,050
6,446
30%
33%
28%
12%
63%
27%
90%
-4%
2.97
1,577
19%
56%
19%
5%
102
$415,988
$407,509
$393,530
$403,647
1.02
1.06
1.02
20%
14%
63%
24%
90%
-6%
3.09
1,543
17%
57%
20%
7%
86
$349,991
$344,944
$345,009
$354,958
1.02
1.02
0.98
Year Sold
2012
2013
35,115
33,858
35%
26%
9%
10%
85%
77%
21%
30%
12%
12%
59%
64%
29%
25%
90%
89%
-6%
-2%
3.14
2.86
1,587
1,572
15%
22%
59%
53%
21%
19%
6%
6%
122
96
$380,834 $423,046
$367,318 $421,035
$366,867 $404,445
$372,446 $421,702
1.04
1.00
1.03
1.05
0.99
1.01
Table 2: Descriptive statistics for the Los Angeles CBSA with data from the MLS for sales between January 2012 and
December 2013. The data are stratified by list price change status, appraisal condition (C2 and C3 represent average and
above average condition and C4 represents below average condition), property type (normal and REO), and year sold.
Page 30
Percentile
Final List Price
to Initial List
Price Ratio
Percentile
100%
1.81
100%
2.22
99%
1.41
99%
1.63
95%
1.32
95%
1.30
90%
Final List Price
to Initial List
Price Ratio
1.21
90%
1.19
75%
1.11
75%
1.09
50%
1.05
50%
1.05
25%
1.02
25%
1.02
10%
0.97
10%
1.00
5%
0.90
5%
0.97
1%
0.75
1%
0.92
0%
0.00
0%
0.74
Table 3: Distribution of the final list price to initial list
price ratio for properties sold in the Cleveland CBSA
between January 2012 and December 2013 and that
had a list price revision and had an initial list price to
sale price ratio less than 1.
Table 4 Distribution of the final list price to initial list
price ratio for properties sold in the Los Angeles
CBSA between January 2012 and December 2013 and
that had a list price revision and had an initial list
price to sale price ratio less than 1.
Percentile
Final List Price
to Initial List
Price Ratio
Percentile
Final List Price
to Initial List
Price Ratio
100%
1.44
100%
1.44
99%
0.99
99%
1.03
0.99
95%
0.98
95%
90%
0.97
90%
0.98
75%
0.96
75%
0.97
50%
0.93
50%
0.94
0.90
25%
0.88
25%
10%
0.81
10%
0.84
5%
0.75
5%
0.80
1%
0.63
1%
0.69
0.41
0%
0.43
0%
Table 6: Distribution of the final list price to initial list
price ratio for properties sold in the Los Angeles
CBSA between January 2012 and December 2013 and
that had a list price revision and had an initial list
price to sale price ratio greater than 1.
Table 5: Distribution of the final list price to initial list
price ratio for properties sold in the Cleveland CBSA
between January 2012 and December 2013 and that
had a list price revision and had an initial list price to
sale price ratio greater than 1.
Page 31