# RECORD 4e94c698b9b6217c36e2e8842148142f5bccfb36
# TITLE: A RE-ASSESSMENT OF HOUSE PRICE INDICES: EVIDENCE FROM THE SPANISH MARKET
# YEAR: 2018.0  VENUE: International Journal of Strategic Property Management  DOI: 10.3846/ijspm.2019.6366
# AUTHORS: Paloma Taltavull de La Paz; Stanley McGreal
# PAGES: 13  CANDIDATE TABLE PAGES (1-based): [9]
# ABSTRACT: This paper using evidence from the Spanish housing market contributes significantly to the debate concerning the different results obtained from house price indices depending on the method used to build the index. Utilising a large database over the period 1994 to 2012, the paper constructs a time dummy hedonic index (HD) and an imputed hedonic index using a Laspeyres approach (HI), and compares the different effect on the price index evolution. The paper discusses control by quality changes and identifies those attributes experiencing structural changes over the analysis period, identified by the HI index but not by the HD index. Results indicate that changes in quality stem from socio-demographic conditions rather than changes to housing quality (other than size). The paper also shows that improvements in neighbourhood quality rather than change in a ‘typical house’ affects house price and argues that these considerations are important in both the method selected to calculate house price indices and the application of the methodology to estimate price changes.


===== PAGE 1 (score 7, numeric-paren 0, table-words 0, var-words 7) =====

Copyright © 2019 The Author(s). Published by VGTU Press
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unre-
stricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
*Corresponding author. E-mail: paloma@ua.es
International Journal of Strategic Property Management
ISSN: 1648-715X / eISSN: 1648-9179
2019 Volume 23 Issue 1: 23–35
https://doi.org/10.3846/ijspm.2019.6366 
A RE-ASSESSMENT OF HOUSE PRICE INDICES:  
EVIDENCE FROM THE SPANISH MARKET
Paloma TALTAVULL DE LA PAZ1,*, Stanley McGREAL2,3
1 Department of Applied Economy Analysis, International Economy Institute, University of Alicante,  
Campus de San Vicente del Raspeig, 03080 Alicante, Spain
2 Built Environment Research Institute, School of the Built Environment, University of Ulster, UK
3 School of Commerce, Division of Business, University of South Australia, Australia
Received 11 September 2015; accepted 10 January 2018
Abstract. This paper using evidence from the Spanish housing market contributes significantly to the debate concerning 
the different results obtained from house price indices depending on the method used to build the index. Utilising a large 
database over the period 1994 to 2012, the paper constructs a time dummy hedonic index (HD) and an imputed hedonic 
index using a Laspeyres approach (HI), and compares the different effect on the price index evolution. The paper discusses 
control by quality changes and identifies those attributes experiencing structural changes over the analysis period, identi-
fied by the HI index but not by the HD index. Results indicate that changes in quality stem from socio-demographic condi-
tions rather than changes to housing quality (other than size). The paper also shows that improvements in neighbourhood 
quality rather than change in a ‘typical house’ affects house price and argues that these considerations are important in both 
the method selected to calculate house price indices and the application of the methodology to estimate price changes.
Keywords: price index, housing market, hedonic models, Spain.
Supplementary material associated with this article can be found, in the online version, at https://doi.org/10.3846/ijspm.2019.6366
Introduction
In the post Global Financial Crisis era there has been in-
creasing attention given to the construction of house price 
indices globally and particularly in European countries 
with a growing number of public institutions and govern-
ments publishing house price indices in an open-access 
format. This stems from the perception that house prices 
have macro-economic effects and that continuous moni-
toring in support of the identification of economic imbal-
ances (European Economy [EU], 2012) facilitates trans-
parency in the market. Given this context, a number of 
studies have focussed on the macro effects of house price 
deviation from their long term trend, the role of monetary 
channels of transmission influences, and the effect of the 
‘housing channel’ on consumption prices (Mishkin, 2007; 
Bank for International Settlements [BIS], 2008; Taltavull & 
White, 2016; R. Grover & C. Grover, 2014), demonstrating 
the relevance of housing price indices. Furthermore, at the 
European level, the EU is advocating the building of price 
indices for member countries using the same methodolo-
gy (Eurostat, 2013) and the requirement for central banks 
to capture market data (also for non-residential real es-
tate markets1) in order to observe patterns in house prices. 
Furthermore, financial and investment institutions need 
reliable indices in updating their balance sheets in accord-
ance with the mandatory regulations (Nikolov, 2014).
Most studies are based on the current price index 
published in each country but discussion has intensified 
concerning to what extent indices correctly reflect the true 
housing values. Consequently, the debate on house price 
index methodology has had a renewed interest in different 
forums. However, complexity arises from the heterogene-
ous nature of housing in terms of location, structure, com-
position and behaviour with price considered to represent 
how a household evaluates attributes according to income, 
education, quality and other characteristics. Such multi-
plicity of attributes makes it difficult to determine the 
best approach to the construction of house price indices. 
The consensus is that to capture the contribution of such 
1	 See the Greece Central Bank or the British Treasury as exam-
ples of different indices.


===== PAGE 9 (score 69, numeric-paren 35, table-words 9, var-words 6) =====

International Journal of Strategic Property Management, 2019, 23(1): 23–35
31
Table 1. Urban area house prices growth rate in different periods (accumulate %)
Base 1998
Urban level (Imputed hedonic − HI)
‘Imputed-
hedonic (HI)᾽
(a)
‘Dummy 
hedonic (HD)᾽
(b)
Differences 
(%)
[(a/b) – 1] · 100
1
2
3
4
Alicante
1998−2012
89.5
117.7
84.8
114.3
108.6
146.1
−25.7
1998−2004
93.3
95.4
88.3
75.3
88.9
167.8
−47.0
2004−2007
38.5
16.5
21.5
1.2
16.2
29.9
−45.8
2007−2012
−20.3
−6.0
−16.1
20.8
−5.3
−28.6
−81.4
Madrid
1998−2012
54.0
157.5
96.9
179.7
−46.1
1998−2004
−2.1
30.6
16.1
39.4
31.1
152.2
−79.6
2004−2007
−39.2
55.8
24.3
66.8
56.9
25.7
121.4
2007−2012
−8.3
−8.5
50.7
19.0
−12.2
−255.7
Murcia
1998−2012
79.7
69.5
34.4
83.1
70.9
130.0
−45.4
1998−2004
80.4
67.7
18.0
58.5
67.2
134.6
−50.1
2004−2007
26.2
15.5
21.7
26.3
19.7
32.0
−38.6
2007−2012
−18.9
−11.9
8.2
−11.9
−12.1
−23.8
−49.4
Valencia
1998−2012
118.1
111.0
22.5
109.0
100.4
103.5
−3.1
1998−2004
101.4
104.6
37.0
89.4
93.4
130.3
−28.3
2004−2007
15.2
21.3
22.1
18.1
19.2
47.1
−59.3
2007−2012
−11.8
−12.2
−26.3
−9.4
−11.5
−38.6
−70.3
Table 2. Hedonic HI model for Alicante province
Alicante
var dep: Log of asking prices
b
Standardized b
Stand. error
constant
6.47
−
0.017
***
City (7)
PD
0.044
0.041
0.001
***
Pdev
−0.004
−0.006
0.001
***
PGR
0.039
0.026
0.001
***
P (logs)
0.035
0.115
0.000
***
Eco
0.035
0.057
0.000
***
UENV
0.158
0.025
0.004
***
Neighborhood (12)
Inc
0.054
0.057
0.001
***
Resarea
0.092
0.072
0.001
***
Con
0.001
0.017
0.000
***
URR
−0.001
−0.030
0.000
***
Qroad
0.020
0.005
0.003
***
Qw-pipe
−0.634
−0.562
0.001
***
Qlig
−0.009
−0.003
0.002
***
Qsh
0.006
0.008
0.001
***
Qschool
0.012
0.011
0.002
***
Qchurch
0.049
0.044
0.002
***
Qleis
0.080
0.088
0.001
***
Qhealth
−0.113
−0.115
0.001
***
Accessibility (3)
Bus
0.007
0.011
0.000
***
Train
−0.002
−0.002
0.001
***
Und-Tram
0.029
0.069
0.000
***
House (12)
TB
0.101
0.098
0.001
***
dwel
−0.001
−0.027
0.000
***
lift
0.058
0.075
0.001
***
age
−0.008
−0.194
0.000
***
Qshop_n
0.118
0.211
0.000
***
Inc_t
0.096
0.098
0.001
***
PD_t
−0.020
−0.018
0.001
***
Ori
0.003
0.013
0.000
***
Vi
0.019
0.021
0.001
***
ConQ
0.033
0.055
0.001
***
m2 (log)
0.811
0.400
0.001
***
m2_O (log)
0.034
0.055
0.000
***
Adj R2
0.768
Se2
4357.188
F
58081.547***
N
596170
Tolerance and VIF test indicates the absence of colineality
+ Categorical variable
*** p-value < 0.001, ** p-value < 0.01, * p-value < 0.05
Note: This Table 2 contains the average value of four urban level betas estimated for Alicante in 2010 (4 quarters). Total number of usable observations 
for Alicante (for the whole period) are 478418, and 11390 on average for every quarter of 2010.


===== PAGE 10 (score 11, numeric-paren 0, table-words 0, var-words 11) =====

32
P. Taltavull de la Paz, S. McGreal. A re-assessment of house price indices: evidence from the Spanish market
housing characteristics have smaller coefficients and in 
essence a lower ranking concerning the effect on prices 
(quality of construction, burb_env = 0.158, orientation, bo-
rient = 0.003, view, bview = 0.019, additional surface area 
bm2_other = 0.034, number of lifts blift = 0.058, and qual-
ity of construction bqconstr = 0.033). This suggests that the 
main determinant for house prices are neighbourhood 
attributes. Hence, indices failing to control by location/
characteristics of the surrounding area could be biased. 
The effects of missing neighbourhood quality change thus 
limit the ability of the index to control by quality.
5. Discussion
As the HD estimations do not allow the attribute param-
eters to vary with time, the results suggest that the model 
structure reflects the weights given to each attribute dur-
ing the first part of the time series period but differ from 
the pattern in the second period. The test for breaking 
points in attributes25 supports this argument and suggests 
that a change in quality accounts for part of the observed 
house price growth and that remaining differences reflect 
other price impulses. HI seems better placed at capturing 
the changes in quality while the HD model reflects ex-
ogenous influences26 such as the market influx on house 
prices from the demand side. One of these components 
could also be due to autocorrelation (spatial and time) in 
the data27.
Models from the HI Hedonic Imputed suggest that 
there is a pattern of initial growth characterised by rising 
house prices until 2005 followed by stable house prices 
suggesting that the share of price explained by property 
quality was increasing until 2005 but then stabilized. These 
results contrast with those from the HD model, notably 
over the period 2005–2010, illustrating that differences in 
house price growth and the extent of the level of growth is 
dependent on the methodology used. One interpretation28 
is that the hedonic dummy method does not take into ac-
count variability in the price of attributes (assumed to be 
constant over the period by the method) hence the index 
loses its capacity to capture quality changes over time. By 
contrast HI allows the estimated betas to vary. It seems that 
the type of dwellings sold over the boom and bust periods 
25	Performed in the typical house’s section 5.1.
26	When the Ph_hat is estimate in a ‘HD model, the values 
are equal to the observed asking prices. This is why it is not 
necessary to calculate an index using the traditional method 
(Laspeyres or Paasche), in a model controlled by attributes 
plus time dummies. The time dummies capture any external 
influence on prices.
27	Note that papers explaining imputed price methods based on 
traditional indices (Laspeyres, Paasche or Fisher) using the es-
timated prices obtained directly from a hedonic model do not 
allow intercept and attribute parameters to change over time 
and then capture the time effect.
28	We want to thanks to an anonymous referee for provide this 
highlight.
have varied to the point that either the relative prices of 
attributes changed or that comparing the reference house 
at the end of the period to the same house at the beginning 
does not make sense. Such reaction could be capturing 
changes in the type of house sold over time and may also 
reflect changes in market liquidity. For example, during the 
boom practically any type of unit was easily sold, even if of 
low quality, hence the rising quality adjusted prices while 
during the bust, the number of transactions declined and 
only certain types of dwellings were sold. One would have 
expected that only the very good houses got sold, so that 
the quality increase was such that the price would decline 
even more when adjusting for quality.
The house price index evolution compared in Figure 2 
suggests that HD and HI are measuring different concepts. 
The HI construction estimated price of the total attributes, 
letting the errors capture the increase in price due to other 
factors (as defined by Rosen). In essence, the HI models 
essentially provide an index of quality rather than an index 
controlled by quality, whereas the HD models isolate house 
prices from their components (time and attributes) obtain-
ing an index adjusted by quality. The different methods of 
construction mean that the interpretation of HI indices is 
not the same as that for HD indices; the former are indi-
ces of quality prices and the latter indices of house price 
controlled by quality. Thus both are complementary rather 
than substitutive, and both partially capture true quality 
controlled house price. As a consequence, in the case of 
HI the price growth rate could be calculated based on 
the equation residuals rather than on the imputed prices 
whereas in the case of a HD index, the equation to estimate 
dummy parameters includes a sufficient number of char-
acteristics to control by quality though the effect on prices 
of the unobserved or omitted key attributes may produce 
an index not fully controlled by quality.
5.1. The ‘typical house’
A further point of discussion relates to estimation of the 
‘typical house’ attributes structure to guarantee that the 
index is controlling by changes in quality and the assump-
tion that it can change over time, explaining the results 
found in the two estimated indices. In this respect, some 
evidence about changes in quality of the whole stock is 
needed to identify whether this type of change occurs 
and when. In the case of no statistically significant chang-
es in housing attributes (which determines the ‘type of 
house’ in the long term), a house price index measuring 
the same initial attribute structure should be precise in 
capturing price changes. However, a change in attributes 
could introduce a bias in the estimated house price index 
capturing ‘inexistent shifts’ in prices and in fact actually 
measuring changes in the type of house on the market. To 
illustrate this point further, the current database is used 
to identify shifts in quality (attributes) over the time. Ta-
ble S3 (supplementary material) shows the significance of 
Zivot-Andrews test of breaking points and its application 
to the density distribution for categorical variables and ex-