Flood Hazards and Urban Housing Markets: The Effects of Katrina on New Orleans

Flood Hazards and Urban Housing Markets:
The Effects of Katrina on New Orleans

Russell McKenzie & John Levendis
Published online: 7 August 2008
# Springer Science + Business Media, LLC 2008
Abstract This study examines the impacts on consumers’ willingness to pay for
certain characteristics of housing in greater New Orleans before and after the
flooding of Hurricane Katrina. Single-family home sales from January 2004 to
August 2006 are collected and used in a hedonic price function to estimate the
changes in the value of amenities, and structural, neighborhood and geographic
characteristics, including the mean elevation of each property. Elevation, which
buyers did not know for certain prior to the storm, but may now be inferred from
water level marks in most neighborhoods, is found to have a positive relationship
with selling prices. Results indicate that pre-Katrina, there was a premium of only
1.4% per foot in flood-prone areas, and was insignificant in areas not subject to
flooding. This increased to 4.6% for flooded areas after Katrina. These findings are
attributed to not only the perceived risk of flooding, but also to the potential of
higher compliance costs associated with rebuilding under more stringent National
Flood Insurance Program (NFIP) guidelines.
Keywords Housing markets . Hedonic estimation . Flooding . Elevation .
Housing supply . Katrina


Introduction

The landfall of Hurricane Katrina on August 29, 2005 and the flooding that followed
devastated the city of New Orleans. With the breaching of the levees protecting New
J Real Estate Finance Econ (2010) 40:62–76
DOI 10.1007/s11146-008-9141-3
R. McKenzie (*)
College of Business, Southeastern Louisiana University, SLU 10813, Hammond, LA 70402, USA
e-mail: russell.mckenzie@selu.edu
J. Levendis
Department of Economics, Loyola University, 6363 St. Charles Ave., Box 15, New Orleans, LA,
USA
e-mail: jlevendi@loyno.edu

——————————————————————————————————————————–

Orleans from Lake Ponchartrain, approximately 80% of the city was flooded. While
the city’s population remains at roughly 60% of its pre-storm level (Brookings
2007), its housing stock may be less than that. Estimates suggest that, in some areas,
more than one-half of the housing units were either severely damaged or destroyed.
According to a U.S. Housing and Urban Development (HUD) study (2006),
approximately 34,300 (19.5% of the pre-storm stock) of the units in Jefferson Parish
suffered major or severe damage. In Orleans Parish, the number experiencing this
level of damage was 105,300, or 56% of the pre-storm units. Since the storm, many
of these have been demolished, while others stand in ruin.
With slightly more than half of its original population and as much as 38% of its
original housing stock destroyed, significant market adjustments are occurring to
ensure market clearing outcomes. According to a Brookings Institute study (2006),
the number of Orleans Parish houses on the market has risen from roughly 2,800 in
February 2006 to nearly 5,000 in October. During this same period, housing sales
increased only from 190 to 291. After a sharp spike immediately following the
storm, average housing prices retreated to approximately $222,000 in February
2006, significantly more than the pre-storm average of $203,000. Between February
2006 and February 2007, average prices fell to slightly more than $197,000, well
below the pre-storm level (MLS). These price changes may well reflect changes in
housing quality after the storm. If a substantial portion of the stock of housing was
damaged and not repaired, lower average prices would be expected.
From an economic perspective, these events provide an opportunity to examine
the responses of individuals in the local housing market to the realization of a major
flood. This paper estimates a hedonic price equation for a large sample of homes in
New Orleans to determine the effect of Katrina on the willingness-to-pay for various
amenities and housing characteristics. The uniqueness of this study is the treatment
of flood hazard and the incorporation of Geographic Information Systems (GIS) data
to extract information about individual property and neighborhood characteristics
and flooding in New Orleans. Previous studies using floodzone classifications have
yielded step-wise relationships between home prices and distinct classifications. The
current research produces a continuous gradient of home prices relative to elevation.
Two sets of regressions are conducted to determine these values both before and
after Katrina. This allows an examination of how consumer’s willingness-to-pay has
changed as a result of the storm.
Review of Hedonic Estimation and Flood Risk
It is commonly accepted that housing is a composite good, in that it is made up of
many distinct characteristics. Through a process known as hedonic estimation
[Wallace 1926, Court 1939, and Rosen 1974, among others], economists are able to
estimate the contribution to the total value of real estate attributable to each
characteristic possessed by that property. In general, the characteristics that have been
shown to have significant impacts on the price of a house include size, number of
bedrooms, number of bathrooms, and car storage, among others. While the structural
features of the house are certainly important, other factors such as neighborhood
characteristics have also been shown to have an impact on the price of property.

——————————————————————————————————————–

Many recent studies have examined the impact of flood risk exposure on housing
values. Most of these attempt to describe the discount associated with location
within a defined floodplain. Bin and Polasky (2004) examine the effects of location
in a floodplain in their analysis of housing prices in North Carolina. In addition to
numerous structural and locational characteristics, they include a binary variable for
flood risk, as determined by whether the property is located in a 100-year flood
zone. Their findings suggest that location in a floodplain is significant, and reduces
housing prices by 5.8%, or $7,529 at the mean. They also show that the occurrence
of Hurricane Floyd impacted the influence of location in a floodplain by conducting
a separate analysis of the data before and after the storm. This shows that prior to
Floyd, the discount for location in a floodplain was 3.8%, or $4,933 at the mean,
while after the storm it rose to 8.3%, or $10,774.
In their study of the effects of disclosure of floodplain status on the value of
residential real estate in California, Troy and Romm (2004) find that there is a 4.2%
discount associated with a property’s location in a floodplain. MacDonald et al.
(1990) also find a negative and significant influence of location in a floodplain on
housing prices. Their estimates varied with the predicted sales price, with an average
priced home ($65,000 in 1990) subject to a discount of 3.7%.
Speyrer and Ragas (1991) examine the effects of floodplain classification on
property values in New Orleans, and find results consistent with those mentioned
and significant. Their estimates indicate a maximum discount for location in a
floodplain of 8.3%. Finally, Shilling et al. (1985) find that houses located in a
floodplain sell at a 6.4% discount relative to comparable houses outside floodplain
zones.
Several of these studies were conducted for time periods following major storm
events, while others were not. Those conducted within a relatively short time of a
flood (Speyrer and Ragas 1991 and Bin and Polasky 2004) find a higher discount for
location in a floodplain. Those with no recent flooding show lower discounts. The
difference in findings of these studies suggests that consumer willingness-to-pay for
a reduction in flood risks increases with recent flooding.
Some studies have found location in a floodplain to be insignificant in
determining the price of real estate, such as Zimmerman (1979), Babcock and
Mitchell (1980), Skantz and Strickland (1987) and Bialaszewski and Newsome
(1990). While the structural and locational variables used in these studies are often
different, the consistent aspect is that each uses a binary variable of location in a
100-year floodplain to proxy for flood risk. One possible explanation for the
insignificance of the flood risk variable noted in some of the studies is the implicit
assumption that the flood risk is constant across the entire floodplain, and thus the
discount associated is also constant. In this study, the authors do not make such an
assumption.
The study at hand recognizes the likelihood that flood risks and the associated
discounts are inversely related to the elevation of the property within the floodplain.
Some studies (see Bartosova et al. 1999) have attempted to capture a similar measure
of risk by including the distance from the potential flood source as a variable in the
hedonic price equation. While this approach may well capture more of the risk of
flooding, it does not reflect the risk-extent of flooding interaction possible with the
use of elevation measures, especially in a flood event such as that which struck New

————————————————————————————————————

Orleans. In this case, the flooding occurred as a result of equalization of water levels
between a large body of water surrounding a bowl shaped city. Distance from the
lake had no bearing on whether or not a certain house flooded. More importantly, the
flooding from Katrina was unique in that most floods are either flash-floods, or
caused by inadequate drainage; they are moving-water floods. The flooding that
followed Katrina was a standing-water flood, so that elevation was of primary
importance.
In earlier studies, the justification for the price differentials was the capitalized
difference in flood insurance premiums. Therefore, the step-wise function is likely a
good representation of prices. However, it assumes that because all homes in a given
flood zone pay the same premia, the associated discounts are equal. The approach
used in this study avoids this step-wise pricing effect by incorporating a continuous
measure of elevation, which for all but flashflood events, better reflects the risks
associated from flooding. The result of this approach is a flood-risk discount
gradient based on elevation, and irrespective of flood zone classification. In addition,
the recent flooding in New Orleans may provide home buyers with a visual cue to
the possibility and extent of flood damages. This is due to the water level marks left
on many of the homes and buildings. Even now, two years after the storm, water
marks on New Orleans homes are ubiquitous.
Data and Study Area
The data used in this study include information regarding terms-of-sale, locational
and structural characteristics, and amenities for more than 33,700 single-family
homes sold in southeastern Louisiana between January 2004 and August 2006. The
data were collected from the Multiple Listing Service (MLS), a database used by
local realtors for listing properties. As the application of hedonic estimation implies
a homogenous market, where there are no systematic differences in the goods
exchanged, only those properties located in Jefferson and Orleans Parishes are
considered. Since greater New Orleans is primarily made up of these two parishes,
this delineation is used to reduce the degree of heterogeneity in the defined market.
The resulting number of observations is 16,258. The time horizon of the data
includes 20 months prior to Hurricane Katrina and 18 months after. The authors
believe this to be sufficient time for market adjustments to reflect the views of
market participants.
The choice of variables to include in the price equation was driven by a review of
the hedonic literature. The characteristics that have been shown to significantly
influence the price of residential properties are many. While there are differences
among studies, some characteristics are consistently shown to be significant. A
relatively complete review of the literature is presented by Sirmans et al. (2005), as
well as Malzeppi (2003). The variables used in this study are presented in Table 1,
with their expected signs and source. For clarity, these are divided into two
classifications. While several classification schemes have been offered for grouping
these characteristics, Wilkinson (1973) offered possibly the most basic, and useful
distinction between the types of factors involved—dwelling-specific vs. locationspecific
characteristics. This is the distinction used in this study.

————————————————————————————————————–

The dwelling-specific characteristics used in this study reflect the type, size,
condition, and amenities associated with each property. They include the number of
bedrooms, full bathrooms, half bathrooms, and total living area. Other dwellingspecific
variables include an assessed condition of the property, car parking
arrangements, type of roof, age of property, lot size, the presence of a pool, and
type of air conditioning. Finally, for convenience, a market condition variable, dayson-
market, is also included in dwelling-specific characteristics.
The included assessed condition variable is especially important to this exercise,
as the condition of much of the housing in post-Katrina New Orleans has changed
relative to that of pre-Katrina. The state of repair (disrepair) of the housing stock is
reflected in this variable. The source of this data is MLS. Listing agents assign a
condition classification to each listed property, based on an established set of criteria.
The meanings and implications of the assessed condition are given below:
0. poor—excessive abuse evident, abandonment or major reconstruction needed
1. fair—badly worn, in need of major overhaul, or post-Katrina gutted
2. average—major components still functional, minor repairs needed
3. very good—no obvious maintenance required, appearance and utility above
average
4. excellent—no functional inadequacies, like-new condition
5. new—new construction, never occupied
By controlling for condition, we ensure that the lower price of the houses at lower
elevations is due to the fact that they are damaged. That is, we control for condition
so that the coefficient on elevation better captures future flood risk, rather than
present flood damage.

————————————————————————————————————–

As a result of widespread flooding, and the damages incurred, many properties
have been partially or completely gutted prior to sale. Gutted houses are stripped of all
interior features, including drywall, fixtures, and insulation, often leaving only the
skeleton (the two-by-fours) in tact. This gutting significantly changes the characteristics
of the property, and is reflected in the assessed condition variable by a lower
score, as compared to an otherwise usable house. A gutted home that was not flooded
takes on a lower assessed value; however, gutting can be an improvement over a
moldy flooded home. This is due to the fact that gutting is the first step in
rehabilitating these structures, so some of the required work has already been done.
Gutting allows the skeleton of the house to dry, and helps prevent the growth of mold.
With the exception of waterfront location, the locational-specific characteristics
used in this study were extracted from multiple sources using Geographic
Information Systems (GIS) tools. Each property was geo-referenced, or coded for
mapping, using ArcView GIS software. This process uses a digital map of the area
including streets and roads to locate a property based on its address. While some
properties can be coded, others cannot, due to incorrect or misspelled addresses. As
each property is located, it is assigned a score reflecting the likelihood the location is
accurate, with 100 being a perfect score. Of the properties in the data, 2,144 could
not be coded. From those that were successfully coded, only those with scores above
75 were used. While the default minimum match score in ArcInfo is 60%, setting a
higher threshold increases the likelihood of accurate matches. The resulting sample
size was 11,617 properties, including omissions for incomplete data. Nine homes
(one in Orleans and eight in Jefferson Parish) closed in the month immediately
following the storm, on terms that would have been agreed to prior to the storm.
These nine observations are thus assigned to the pre-Katrina subset.
Once the observations were coded, different map layers were used to extract
additional data for each location. Parish-level mosaics of the LiDAR data from U.S.
Geological Survey (USGS) were used to extract the elevation of each property,
relative to mean sea level (msl). U.S. Census Bureau Tiger files were used for census
tract-level demographics, including median household income (MHI) and racial
composition. Finally, ArcView geospatial tools were used to calculate the Euclidean
distance to the centroid of the central business district (CBD).
Clearly the housing market in greater New Orleans (GNO; which we define here
to include Jefferson and Orleans parishes) is different pre-and post-Katrina. Thus, we
estimate each sample separately. Special care should be made when describing the
post-Katrina real estate market in GNO. A question of heterogeneity arises between
flooded and non-flooded houses. Should flooded and non-flooded homes be lumped
together into a single post-Katrina regression, or should they be separated?
To separate the samples, those houses within the post-Katrina data that were
flooded due to Katrina and those that were not must be identified. This was
accomplished by delineating the areas that were flooded using flood-depth maps
from the Federal Emergency Management Agency (FEMA). Since the flooding was
the result of breaches in levee protection and the resultant equalization of surface
water levels, elevation was the determining factor. Only those areas above sea level
and those for which levees did not fail were spared.
All houses located in areas where flooding was indicated by FEMA maps and
built on concrete slab foundations were assigned to the flooded subset. For homes on

—————————————————————————————————————-

Holding all else constant, a one percentage point increase in white residents
within the census tract raises selling price by 0.1% to 0.5%, or $253 to $883 at the
mean. Meanwhile, an increase in MHI of 1% results in an increase in price ranging
from 0.06% to 0.22%, or $111 to $507, when evaluated at the mean. Also, the

presence of a fire place is expected to demand a premium. The estimates for a
fireplace range from 3.3% to 5.7%, or $7,600 to $11,173.
The age and age squared variables are used to capture the impact of time on the
value of housing. It is expected that as a home ages, it will experience a discount
relative to other similar homes. At some point this process reverses as old homes
become “historic”, and prices reflect a premium. Therefore, the expected sign of age
is negative, while the expected sign of age squared is positive. In all four models,
this is shown to hold true.
The influence of the other variables changes between models. This suggests that
their impacts on selling price differ between the subsets (flood or no-flood) or were
altered by the storm (pre-and post-Katrina). In the first two regressions, the distance
from the Central Business District (CBD), central air conditioning, and number of
parking spaces were all significant, with expected signs. Estimates suggest that each
additional mile from the CBD results in a discount of 1.1%, or from $1,890 to
$2,113 at the mean in the pre-Katrina market, while in both post-Katrina models its
effect is not significant. Pre-Katrina, each additional parking space commands a
premium ranging from 1.7% to 4.3%. In both post-Katrina regressions, the estimated
coefficient on parking is significantly smaller, ranging from 1.2% to 2.1%. Likewise,
central air conditioning in the pre-Katrina market commands a premium of from
8.5% to 9.7%, or $15,016 to $19,516 at the mean. While still significant in the post-
Katrina market, its coefficients are significantly smaller, with estimates ranging from
5.6% to 6.3%. All of these results likely reflect a change in consumers’ priorities.
After the storm, residents give priority to living space, or shelter in general, while
certain amenities and characteristics do not command the premium they once did.
The assessed condition of the property, not found in most hedonic price models, is
critical to this analysis. In this study, it is realized that within the housing market in
post-Katrina New Orleans there is a very distinct difference between flooded and
non-flooded houses, and between those that are ready for occupants and those
consisting of only a shell. These differences are reflected in the scores assigned to
each property. The influence of this assessed condition is positive and highly
significant in all four models. It is also very consistent across all models.
Prior to Katrina, a one point increase in the agent-assessed condition (on a 5 point
scale) results in a premium of 12.7% in areas that did not flood, and 13.2% in areas
that flooded. This translates into a premium of $23,319 to $25,471, at the mean. After
Katrina, the same increase resulted in an 11.2% premium ($25,680) if the property was
not flooded and a 14.8% premium ($29,011) if the property was flooded. The size of
the coefficient, and its stability across regressions, shows that the condition of the
structure remains of utmost importance, regardless of the location of the property.
The most revealing results of this analysis are the impacts of those variables
related to the risk and extent of flooding, including the elevation of the property’s
centroid. Homes at higher elevations are expected to be less likely to flood and/or
less likely to suffer significant flooding. Even within a designated flood zone, homes
that are located on higher land are considered to be of less risk. Therefore, it is
expected that homes at higher elevations will command a premium relative to homes
at lower elevations, regardless of flood zone classification.
Prior to Katrina, an additional foot of elevation in flood-prone areas commanded a
premium of only 0.9% per foot of elevation, or $1,590 at the mean. Put differently,

———————————————————————————————————-

each additional foot below sea-level resulted in a discount of 0.9%. This is consistent
with other empirical studies that find a discount associated with locating in a flood
zone. However, these earlier studies tend to estimate a singular discount, and ignore
the elevation of the property. The discounts found by earlier studies are shown to be
roughly equal to the capitalized insurance premiums. This study’s estimates seem
small, and may illustrate the seemingly worry-free attitude of the residents of New
Orleans prior to Katrina, and their confidence in the levy system that protected them.
However, it is important to realize that these premia are for each foot of elevation.
For example, if an additional 4 feet of elevation would raise an otherwise identical
house above that of the other, that difference would command a premium of
approximately 4%, which is consistent with the findings of many earlier studies.
After the storm, elevation was at the forefront of many homeowners’ minds,
which is revealed by the changes in the influence of elevation on property values.
For those at-risk (flooded) houses, each foot of elevation commands a 4.5%
premium ($8,821 at the mean), which is consistent with the higher estimates of
earlier studies. This is a significant increase from the pre-storm premium. Such an
increase is not surprising, as the value of elevation became apparent to everyone in
the area.
There are several plausible reasons for this response. One reason may be the
difficulties faced by many homeowners in obtaining insurance. In the aftermath of
Hurricane Katrina, many insurers stopped writing policies on homes viewed as even
slightly risky. Another may simply be a desire to avoid the risk of flooding and the
associated costs. Furthermore, the cost of rebuilding is significantly higher in
flooded areas due to stricter building codes. Dehring (2006) finds that vacant land
values decline with implementation of new, more stringent building codes. Not only
are new structures subject to these codes, but any existing structure to which
“substantial improvements” are made also fall under these new guidelines. Defined
as suffering damage totaling 50% of its appraised value, many of the most heavily
damaged homes in New Orleans fall under this requirement. This study finds that
houses that were flooded in Katrina displayed a higher sensitivity to future flood
risk. That is, once a house was shown to be a flood risk, the question becomes one of
extent. For homes that were flooded, the extent of flooding and the associated costs
of compliant repairs could be inferred from the high water marks on the buildings.
This visible cue likely led to a higher demand and prices for those properties that
suffered the least damage, and therefore required lower repair/rebuilding costs.
Houses that were not flooded did not display the same type of sensitivity to
elevation. That is, in the eyes of many residents, Katrina seems to have been a once
in a lifetime event. If a house did not flood during Katrina, it is unlikely to flood in
the future. As a result, there is less incentive to pay for additional elevation. That is
not to say that residents do not appreciate the added safety, but additional elevation
in non-flooded areas has no impact on selling prices. This perception may have
existed prior to Katrina as well. Both before and after the storm, the elevation
variable is not significant as a determinant of housing prices for homes already
adequately elevated. A possible explanation for this lack of a major impact is that for
non-flooded homes, the visual cue of flood-related water marks is missing in areas
that did not flood. Another is the diminishing marginal benefit of elevation; homes
that did not flood in Katrina were and continue to be unlikely to flood in the future.

——————————————————————————————————–

Conclusion
The findings of this study highlight two important points about the New Orleans
housing market and its responses to flood risks. The first is that residents are aware
of, and willing to pay a premium for increased elevation, even if the higher elevation
property remains in the same flood zone classification. In effect, the use of flood
zone classifications with a range of elevations included assumes that the risks across
all homes in that zone are the same. This study shows that a more revealing
approach would be to incorporate the actual elevation of each property. While the
risk of flooding is important, so is the extent of flooding. Since all properties within
a flood zone are not situated at the same elevation, it is reasonable to assume that not
all will flood to the same extent. Including the elevations helps illustrate this point.
The second major insight of this study is how the influence of those flood-related
characteristics changes in response to a major storm event. Prior to Katrina, the
contribution of property elevation was small. However, after Katrina the impact of
these variables increased significantly. This indicates that residents began to
recognize the value of these characteristics. Several studies (see Speyrer and Ragas
1991 and Bin and Polasky 2004) have shown this to be a common occurrence that
may wane in time. But given the media exposure of the plight of New Orleans, this
may take much longer than usual.
This study gives rise to several related questions. The current paper and most
other studies in the hedonic pricing literature make a parametric assumption
regarding the functional form of the relationship between price and flood-risk. Some
notable exceptions are Speyrer and Ragas (1991), Pace (1993), and Pace (1995). The
most common parametric assumption is that elevation affects price in a log–log or
semi-log fashion. Given the large sample size in the current dataset, semi-parametric
methods may be used to more carefully examine the relationship between elevation
and price. A related issue is whether NFIP guidelines and national flood-insurance
rates are properly priced. If this is the case, one would expect monotinicity in the
price-elevation function. A useful extension to the present study is to supplement the
current hedonic equation with controls for NFIP and a non-parametric elevation
function.
References
Babcock, M., & Mitchell, B. (1980). Impact of flood hazard on residential property values in Galt,
Ontario. Water Resources Bulletin, 16(3), 532–537.
Bartosova, A., Clark, D. E., Novotny, V., & Taylor, K. (1999). Using GIS to Evaluate the Effects of Flood
Risk on Residential Property Values. Proc. Environmental Problem Solving with Geographical
Information Systems: A National Conference, U.S. EPA, September 22–24, 1999, Cincinnati, Ohio.
Bialaszewski, D., & Newsome, B. A. (1990). Adjusting comparable sales for floodplain location: the case
of Homewood, Alabama. Appraisal Journal, 58(1), 114–118.
Bin, O., & Polasky, S. (2004). Effects of flood hazards on property values: evidence before and after
hurricane Floyd. Land Economics, 80(4), 490–500. doi:10.2307/3655805.
Brookings Institute (2007). A review of key indicators of recovery two years after Katrina: Second
anniversary special. http://www.brookings.edu/reports/2007/katrinarecovery.
Court, A. T. (1939). Hedonic price indexes with automotive examples. The Dynamics of Automobile
Demand, New York, NY: General Motors.

—————————————————————————————————————————————-

Cropper, M. L., Deck, L. B., & McConnell, K. E. (1988). On the choice of functional form for hedonic
price functions. Review of Economics and Statistics, 70(4), 668–675. doi:10.2307/1935831.
Dehring, C. A. (2006). Building codes and land values in high hazard areas. Land Economics, 82(4), 513–
528.
Early, J. F., & Sinclair, J. H. (1983). Quality adjustment in the producer price indexes. In F. F. Murray
(Ed.), Studies in Income and Wealth 47 (pp. 107–145). Chicago: University of Chicago Press.
Halvorsen, R., & Pollakowski, H. O. (1981). Choice of functional form for hedonic price equations.
Journal of Urban Economics, 10(1), 37–49.
Housing and Urban Development (2006). Current housing unit damage estimates: Hurricanes Katrina,
Rita, and Wilma. U. S. Department of Housing and Urban Development’s Office of Policy
Development and Research.
MacDonald, D. N., White, H. L., Taube, P. M., & Huth, W. L. (1990). Flood hazard pricing and insurance
premium differentials: evidence from the housing market. The Journal of Risk and Insurance, 57(4),
654–663.
Malpezzi, S. (2003). Hedonic pricing models: A selective and applied review. In T. O. Sullivan, & K.
Gibbs (Eds.), Housing Economics and Public Policy: Essays in Honor of Duncan Maclennan.
Oxford, UK: Blackwell.
Pace, R. K. (1993). Nonparametric methods with applications to hedonic models. Journal of Real Estate
Finance and Economics, 7, 185–204.
Pace, R. K. (1995). Parametric, semiparametric, and nonparametric estimation of characteristic values
within mass assessment and hedonic pricing models. Journal of Real Estate Finance and Economics,
11, 195–217.
Rosen, S. (1974). Hedonic prices and implicit markets: product differentiation in pure competition.
Journal of Political Economy, 82, 34–55.
Shilling, J. D., Benjamin, J. D., & Sirmans, C. F. (1985). Adjusting comparable sales for floodplain
location. The Appraisal Journal, 53, (3)429–437.
Sirmans, S. G., Macpherson, D. A., & Zietz, E. N. (2005). The composition of hedonic pricing models.
Journal of Real Estate Literature, 13(1), 3–43.
Skantz, T., & Strickland, T. (1987). House prices and a flood event: an empirical investigation of market
efficiency. The Journal of Real Estate Research, 2(2), 75–83.
Speyrer, J. F., & Ragas, W. (1991). Housing prices and flood risk: an examination using spline regression.
Journal of Real Estate Finance and Economics, 4, 395–407.
Troy, A., & Romm, J. (2004). Assessing the price effects of flood hazard disclosure under the california
natural hazard disclosure law. Journal of Environmental Planning and Management, 47(1), 137–162.
Wallace, H. A. (1926). Comparative formland values in Iowa. Journal of Land and Public Utility
Economics, 2, 385–392.
Wilkinson, R. K. (1973). House prices and the measurement of externalities. The Economic Journal, 83
(329), 72–86.
Zimmerman, R. (1979). The effect of floodplain location on property values: Three Towns in Northeastern
New Jersey. Water Resources Bulletin, 15(6), 1653–1661.

Souce:  www.springerlink.com



Leave a Reply

You must be logged in to post a comment.