Content uploaded by Adam Banai
Author content
All content in this area was uploaded by Adam Banai on Jul 04, 2017
Content may be subject to copyright.
Ádám Banai–Nikolett Vágó–Sándor Winkler
The MNB’s house price index
methodology
MNB Occasional Papers 127
2017
Ádám Banai–Nikolett Vágó–Sándor Winkler
The MNB’s house price index
methodology
Ádám Banai–Nikolett Vágó–Sándor Winkler
The MNB’s house price index
methodology
MNB Occasional Papers 127
2017
The views expressed are those of the authors and do not necessarily reflect the official view of the central bank of Hungary
(Magyar Nemzeti Bank).
MNB Occasional Papers 127
The MNB’s house price index methodology
(Az MNB lakásárindex módszertana)
Written by Ádám Banai, Nikolett Vágó, Sándor Winkler
Budapest, May 2017
Publisher in charge: Eszter Hergár
H-1054 Budapest, Szabadság tér 9.
www.mnb.hu
ISSN 1585-5678 (online)
MNB Occa siONal PaPers 127 • 2017 3
Contents
Abstract 5
1 Introduction, motivation 7
2 A brief review of the relevant literature 9
3 Data 13
4 The MNB’s house price index methodology 16
4.1 The backcasting of useful NIA 16
4.2 Methodology of outlier filtering 17
4.3 Methodology of regression estimation 20
4.4 Disaggregated indices 21
4.5 Selection of explanatory variables 22
5 Presentation of the results of the MNB’s house price index 23
5.1 Results of the MNB’s house price index 23
5.2 Regression results 27
6 Robustness analysis 33
7 Conclusions 40
References 41
Annex 43
Publisher in charge: Eszter Hergár
H-1054 Budapest, Szabadság tér 9.
www.mnb.hu
ISSN 1585-5678 (online)
MNB Occa siONal PaPers 127 • 2017 5
Abstract
This study presents the detailed method of the MNB’s house price index and the results of the new price
sheds light on the strong regional heterogeneity underlying Hungarian housing market developments.
JEL codes: C430, R210, R310
Keywords: housing market, house price index, hedonic regression
MNB Occa siONal PaPers 127 • 2017 7
1 Introduction, motivation
impact on the banking sector. Changes in the prices of the real estate collateral behind mortgage loans may
not only determine the performance of the loans, but the recovery through the sale of collateral in the event
new disbursements. Compared to corporate loans, the banking sector can earn larger spreads on mortgage
boosts demand for housing loans.
arising in the real estate market are of key importance for the central bank. The house price index is designed
aspects. Having been available since 1998, the FHB index is typically published with a considerable lag (5–8
Figure 1
Interaction between market participants and the housing market
Policy
environment: Non-financial corporates
Households Housing
market
Financial sector
Financial
markets
• Fiscal
• Monetary
• Macroprudenal
• Microprudenal
Construcon
House
supply
House prices,
transacons Housing loans
Banks Non-bank financial
intermediaries
Housing
demand
Other
corporaons
Source: ESRB, MNB.
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 20178
Compared to these two indices, the house price index family presented in this study represents a step forward
in several regards. (1) The index developed by the authors of this study has been constructed on the longest and
from 1990. This is important because both the modelling of house prices and the assessment of housing market
heterogeneity of house prices across regions and municipality types. By contrast, the previous indices provided
used for the purposes of our indices. In Chapter 5, we outline our regression results and present the derived
house price indices. In Chapter 6 we conduct a robustness analysis for the methodology of the indices. Finally,
Chapter 7 provides conclusions.
MNB Occa siONal PaPers 127 • 2017 9
2 A brief review of the relevant
literature
1
Consequently, house price changes
handbook issued by Eurostat (2013).
Stratified sample mean method
This approach compiles the index based on average price changes computed within homogeneous groups that
the method are that it is easy to apply and explain to users, and that the sub-indices can be interpreted
1 House price indices in New Zealand, Denmark, the Netherlands and Sweden are constructed from stock appraisals.
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201710
of individual sub-indices, which reduces the reliability of the method.
used as a point of reference in the academic literature (e.g. Mark and Goldberg (1984), Crone and Voith (1992),
Hedonic regression models
hedonic regression models
In the case of the ,2
log yi=
β
0+
β
1x1i+
β
2x2i+L+
β
kxki +
δ
tdti +
ε
i
t=2
T
∑,
where y is the price of the house, x dt
t, β δ
t
ε is
adjacent-period model,
T
means the number of all review periods, then a total of (T
t
log yi
t=
β
0
t+
β
1
tx1i
t+
β
2
tx2i
t+L+
β
k
txki
t+
δ
tdi
t+
ε
i
t, (t = 2,3,…,T) (1)
2 The model was originally developed by Court (1939).
MNB Occa siONal PaPers 127 • 2017 11
where y is the price of the house, x d β expresses
δ ε is the residual
In the ,3
t is the following:
log yi
t=
β
0
t+
β
1
tx1i
t+
β
2
tx2i
t+L+
β
k
txki
t+
ε
i
t, (t = 2,3,…,T)
where y is the price of the house, x β
the control variables, and ε is the residual value. This method should be preferred to the previously described
as possible should be considered, hence reducing the risk of bias arising from missing variables.
Sub-samples are separated according to some criteria relevant to the analysis (e.g. region, municipality type).
3
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201712
Repeated sales method
Another frequently used approach is the repeated sales method, which only considers price changes in those
and Nourse (1963). Besides the Federal Housing Finance Agency (FHFA), Standard & Poor’s (2009) compiles
Beyond the models presented above, there are combined methods that use hedonic regressions and the
so far (presented in Chapter 3) are primarily suitable for the purposes of hedonic regression models.
MNB Occa siONal PaPers 127 • 2017 13
3 Data
typically derived from real estate property appraisals and, as shown at the beginning of the previous chapter, at
in the modelling of house prices in the following.
were collected and stored in a uniform structure. With respect to the variables stored in the database, there
1. before 2008, it was unknown whether the person liable for duty payment in the transfer of property is
4
4 The database does not include the properties purchased by the National Asset Management Agency (NAMA).
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201714
around 3.1 million property transfers between 1990 and 2016 Q2. The variables of the database and the list
incomplete or zero. In order to prevent the loss of an inordinate amount of data, this incomplete NIA data must
be back-cast. The precise methodology of this exercise is explained in Sub-Chapter 4.1. We need to stress that
Budapest sales data have only been available in the NTCA duty database from 2001, which should be borne
in mind during the assessment of the pre-2001 house price index values.
Table 1
Definition of the dataset and the individual variables used in the estimation
Source Variable Description
NTCA duty database
ar_ln The price of the real estate is the dependent variable of the regressions. It is the larger
among National Tax Authority's valuation and the price in the transaction contract.
The variable is in a logarithmic form.
adnev Quarter of property acquisition duty.
The net internal area (NIA) is in the regressions by categories of the type of dwelling.
The category variable "type of dwelling" can have the following values: family house in
inner and outer districts (in the case of Bp.), family house in county seat and other
cities (in the case of cities), condomnium, panel block of flats and homestead. In case
of data before 2008 there are only two categories: family house and flat.
Category variable: new or used property.
Variables made
based on the HCSO
id of the settlement
bp_ker Category variable: districts of Budapest.
agglomerácio Category variable: 8 districts distinguished: agglomeration of Szeged, Pécs, Debrecen,
udulokorzet
Velence - Vértes.
megye Category variable: county of the settlement.
TS TAR
database
de02_ln Population at the end of the year. The variable is in a logarithmic form.
de66_ln Size of the municipality. The variable is in a logarithmic form.
on23_ln Amount of local housing subsidies. The variable is in a logarithmic form.
Geox
database
ido_p_bp_ln The shortest distance from Budapest expressed in minutes. The variable is in a
logarithmic form.
ido_p_msz_ln The shortest distance from the county seat expressed in minutes. The variable is in a
logarithmic form.
NTCA PIT
database
Net labour income per capita. The variable is in a logarithmic form.
Note: The Geox database contains the location of Hungarian municipalities relative to specific nodes and centres (e.g. distance from Budapest or
from the nearest highway node). The distances are expressed both in time and kilometres. The TSAR database is maintained by the HCSO and
contains comprehensive information on Hungarian municipalities (e.g. demography, institutional coverage, tourism, etc.). Inner districts in the
Budapest model: I, II, III, V, VI, VII, VIII, IX, XI, XII, XIII, XIV.
Apart from the NIA of the dwelling, the only known variables are type (detached house, semi-detached or row
the housing market of the area, driving up local house prices. Another important factor in the assessment
DATA
MNB Occa siONal PaPers 127 • 2017 15
of the house price indices are included in the Annex.
and the municipality-level variables included in the model create a unique opportunity for the modelling of
house prices.
MNB Occa siONal PaPers 127 • 201716
4 The MNB’s house price index
methodology
indicates the NIA of the structures. The database also includes a variable referred to as the “property area”
We supply the missing values of the “useful NIA” variable from the appropriate values of the “property area”
variable, and backcast the missing values with a regression method. In summary, we use the following method
5
1. We considered all useful NIA data under 15m2 and over 500m2 to be missing parameters.
2, we
consider the “property area” to be the “NIA”.
method to backcast the size of the “NIA”.
Table 2
Percentages of missing useful NIA information by year and by municipality type
(%)
Budapest Cities Municipalities Tot al
1990–2007 4.2 20.3 53.4 25
2008–2015 7.4 25.6 63.1 28.7
and 2007 included the following explanatory variables: property price, quarter dummy, property type (house
we were also able to control for a variable that indicates whether the property is newly built, for the district
and brick dwellings.
5
respect of the details of the regression estimate.
MNB Occa siONal PaPers 127 • 2017 17
Table 3
Percentiles of NIA by municipality type and by sub-sample
Percentile 1990–2007 2008–2016 Q1
Budapest Cities Municipalities Budapest Cities Municipalities
5% 27 33 39 27 34 40
10% 31 37 45 31 38 45
25% 39 49 56 40 50 60
50% 52 56 70 53 57 75
75% 67 70 90 68 75 96
90% 86 95 120 90 103 120
95% 105 116 137 111 126 143
6 in the database. Firstly, the database might
7
• the NIA of the dwelling was smaller than 15m2 or larger than 500m2
8
()
=−
−
ryy
MS
Eh
1
,
i
ii
ii
i
()
()
where yi is the observed price for the i
yi
i ,
∑
()
=−
=
MS
En
yy
1
ii
i
i
n
()
2
1
is the mean squared error excluding the ihii is the leverage.9
6 Zrínyi et al. (2012) cite numerous outlier definitions from the academic literature.
7
consumer price index.
8
9
the hat matrix that shows the leverage exerted by the ith observed price (
yi
) on the ith estimated value (
yi
).
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201718
which is addressed by the 1−hii term in the formula. The mean squared error included in the indicator is
derived from a regression that does not include the reviewed i
i
10
11
()
=−
−
−
CD yy
MSEh
h
h
MSE
MSEp
11
1
i
ii
iii
ii
ii
i()
()
2
()
where
yi
i
yi() i
i p is the
number of explanatory variables in the regression, MSE is the mean squared error, MSE(i) is the mean squared
ihii is the leverage of the i
i
12
()
()
=−
−
−
−
WD yy
MSEh
hn
h
1
1
1
i
ii
iii
ii
ii
()
()
where
yi
i
yi() i
in is the
number of elements in the sample, MSE(i) ith
hii is the leverage of the i
to be discarded.
13
DFBETAij =bj−bj(i)
MSE(i)′
X X
( )
j j
−1
where bj is the j bj(i) is the jth element of the regression that
results when the i MSE(i)
of the iX is the matrix of the explanatory variables included in the linear regression model.
10 It should be borne in mind that with several similar outliers in the database, the regression received after the removal of individual observations
may be very similar to the regression that includes all of the observations.
11
12
13
X'X
MNB Occa siONal PaPers 127 • 2017 19
without the i
literature, based on sample size and the number of the explanatory variables, each indicator can be used to
14
each other.
Table 4
Percentage of discarded observations in a multi-step filtering procedure
2001–2007 2008–2015
Tot al 1st step of
filtering
2nd step
of filtering
Tot al
Outliers Tot al 1st step of
filtering
2nd step
of filtering
Tot al
Outliers
Number of
%%%Number of
%%%
Budapest 364269 0.2 4.3 4.5 308734 0.2 5.4 5.6
Municipalities 317219 2.2 4.1 6.3 241947 2.5 5.1 7.6
Cities
Southern Great
Plains 138899 0.7 4.9 5.6 110286 1.4 6.0 7.4
South West
Hungary 88198 0.6 5.1 5.7 64286 0.5 5.8 6.3
Northern Great
Plains 148379 0.7 5.2 5.8 100227 0.5 5.8 6.3
Northern
Hungary 88907 0.7 4.5 5.2 62331 0.7 6.0 6.7
Central
Transdanubia 99829 0.6 4.9 5.4 77462 0.6 5.7 6.3
Central Hungary 108341 1.4 4.2 5.6 88414 0.9 5.7 6.6
Western
Hungary 73726 0.4 4.9 5.3 71439 1.0 5.5 6.5
Tot al 1427767 1125126
14 We performed a robustness analysis for this criterion, which is discussed in Chapter 6.
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201720
methodological reasons15 with the release of new data.
15
transaction data used for the calculations are made available with a significant lag.
Figure 2
National MNB house price index with various outlier filtering procedures
(quarterly changes)
–10
–8
–6
–4
–2
0
2
4
6
8
10
–10
–8
–6
–4
–2
0
2
4
6
8
10
Per cent Per cent
Aggregated MNB house price index
Aggregated house price index without outlier filtering
Aggregated house price index without the second step of outlier filtering
2008 Q1
2008 Q2
2008 Q3
2008 Q4
2009 Q1
2009 Q2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
2011 Q1
2011 Q2
2011 Q3
2011 Q4
2012 Q1
2012 Q2
2012 Q3
2012 Q4
2013 Q1
2013 Q2
2013 Q3
2013 Q4
2014 Q1
2014 Q2
2014 Q3
2014 Q4
2015 Q1
2015 Q2
2015 Q3
2015 Q4
2016 Q1
2016 Q2
Note: The black dot ted line indicates the 95 per cent confidence interval of the MNB’s aggregate house price index.
MNB Occa siONal PaPers 127 • 2017 21
each region, and consequently, we do not compile regional price indices for villages.
Table 5
Number of obser vations included in the estimate for each year of the review period
Budapest
Municipa-
lities
Cities
Tot al
Southern
Great
Plains
South
Hungary
Northern
Great
Plains
Northern
Hungary
Central
Trans-
danubia
Central
Hungary
Hungary
2001 27,348 34,703 18,340 10,224 18,648 5,202 9,698 11,620 9,190 144,971
2002 38,649 39,351 19,441 13,395 23,843 12,259 12,650 13,974 11,485 185,044
2003 62,771 45,160 21,242 14,301 24,969 16,121 15,709 17,121 12,205 229,597
2004 46,584 40,182 14,794 10,479 15,730 10,380 11,241 12,874 7,393 169,654
2005 45,247 44,447 15,615 10,135 16,990 12,066 12,401 12,792 7,763 177,455
2006 56,830 43,946 20,153 11,737 20,022 13,215 14,195 15,268 10,518 205,883
2007 54,117 45,586 19,967 11,790 18,574 12,951 16,185 16,831 10,297 206,296
2008 46,925 38,902 17,442 10,164 15,143 10,898 13,246 15,268 9,821 177, 8 0 6
2009 30,928 28,229 11,930 7,261 11,157 6,695 7,855 9,903 7,215 121,170
2010 30,484 24,077 10,672 6,549 9,458 6,207 7,289 8,678 6,806 110,218
2011 29,389 23,422 10,533 6,262 9,422 5,708 7,587 8,280 7,053 1 0 7, 6 5 4
2012 30,522 23,095 11,195 6,253 9,762 5,800 7,299 8,108 7,862 109,894
2013 29,693 23,189 10,716 6,335 9,838 5,979 7,484 7,966 7,720 108,917
2014 38,370 27,181 12,572 7,391 12,047 7,410 9,052 9,954 8,793 132,769
2015 46,272 28,326 13,406 8,074 13,134 7,709 10,265 12,073 8,977 148,233
2016 20,324 15,316 7,127 4,334 6,751 4,036 5,369 5,699 4,492 73,446
Note: values for 2016 refer to Q1 and Q2.
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201722
Finally, we construct the aggregate house price index by chaining the previously received quarterly indices.
We selected the explanatory variables based on two main criteria: on the one hand, we tried to include in the
municipality: per capita net income.
The TSTAR database maintained by the HCSO contains nearly 1,800 municipality-level variables. We selected
the amount of local subsidies granted for housing purposes).
from Budapest and from the county seat as the rest of the variables did not increase the model’s explanatory
power considerably.
16
16 The issue is only relevant in this particular case, as municipality-level variables are available at a yearly frequency.
MNB Occa siONal PaPers 127 • 2017 23
5 Presentation of the results of the
MNB’s house price index
of the historic developments of Hungarian house prices. In the second half of the chapter, we discuss the
from 1990 in an aggregated form, but in a disaggregated form we could only construct the indices from 2001
2001 is properly described by the aggregated MNB house price index without Budapest.
Figure 3
Nominal and real MNB house price index
(2001 Q1 = 100%)
0
25
50
75
100
125
150
175
200
225
0
25
50
75
100
125
150
175
200
225
Per cent Per cent
Aggregated nominal MNB house price index
Aggregated real MNB house price index
Aggregated nominal MNB house price index without Budapest
Aggregated real MNB house price index without Budapest
1990 Q1
1991 Q1
1992 Q1
1993 Q1
1994 Q1
1995 Q1
1996 Q1
1997 Q1
1998 Q1
1999 Q1
2000 Q1
2001 Q1
2002 Q1
2003 Q1
2004 Q1
2005 Q1
2006 Q1
2007 Q1
2008 Q1
2009 Q1
2010 Q1
2011 Q1
2012 Q1
2013 Q1
2014 Q1
2015 Q1
2016 Q1
Note: The real index is deflated with the consumer price index. Aggregated from the national estimates until 20 01 and from the sub-indices from 2001.
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201724
a half years, on average, prices have already exceeded the previous “peak” of 2007–2008. In real terms,
Hungarian housing market. Budapest has witnessed more dynamic price increases in recent years than those
as in the rest of the country, while house price levels in Northern Hungary, for example, fall far behind. One
Budapest house price changes. At present, the pick-up in the housing market is strongly Budapest-oriented,
in real terms (Figure 5).
Figure 4
The MNB’s nominal house price index by municipality type
(2010 = 100%)
40
50
60
70
80
90
100
110
120
130
140
150
160
40
50
60
70
80
90
100
110
120
130
140
150
160 Per cent Per cent
Budapest
Cies
Municipalies
2008 Q1
2008 Q3
2007 Q1
2007 Q3
2006 Q1
2006 Q3
2005 Q1
2005 Q3
2004 Q1
2004 Q3
2003 Q1
2003 Q3
2002 Q1
2002 Q3
2001 Q1
2001 Q3
2009 Q1
2009 Q3
2010 Q1
2010 Q3
2011 Q1
2011 Q3
2012 Q1
2012 Q3
2013 Q1
2013 Q3
2014 Q1
2014 Q3
2015 Q1
2015 Q3
2016 Q1
MNB Occa siONal PaPers 127 • 2017 25
Figure 5
The MNB’s real house price index by municipality type
(2010 = 100%)
60
70
80
90
100
110
120
130
140
150
60
70
80
90
100
110
120
130
140
150
Per cent Per cent
Budapest
Cies
Municipalies
2008 Q1
2008 Q3
2007 Q1
2007 Q3
2006 Q1
2006 Q3
2005 Q1
2005 Q3
2004 Q1
2004 Q3
2003 Q1
2003 Q3
2002 Q1
2002 Q3
2001 Q1
2001 Q3
2009 Q1
2009 Q3
2010 Q1
2010 Q3
2011 Q1
2011 Q3
2012 Q1
2012 Q3
2013 Q1
2013 Q3
2014 Q1
2014 Q3
2015 Q1
2015 Q3
2016 Q1
Note: Deflated by the consumer price index.
Figure 6
The MNB’s nominal house price index for cities by region
(2010 = 100%)
40
50
60
70
80
90
100
110
120
130
140
40
50
60
70
80
90
100
110
120
130
140 Per cent Per cent
Cies – South West Hungary
Cies – Northern Great Plains
Cies – Northern Hungary
Cies – Central Transdanubia
Cies – Central Hungary
Cies – Western Hungary
Cies – Southern Great Plains
2008 Q1
2008 Q3
2007 Q1
2007 Q3
2006 Q1
2006 Q3
2005 Q1
2005 Q3
2004 Q1
2004 Q3
2003 Q1
2003 Q3
2002 Q1
2002 Q3
2001 Q1
2001 Q3
2009 Q1
2009 Q3
2010 Q1
2010 Q3
2011 Q1
2011 Q3
2012 Q1
2012 Q3
2013 Q1
2013 Q3
2014 Q1
2014 Q3
2015 Q1
2015 Q3
2016 Q1
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201726
The MNB’s aggregate house price index is consistent with the house price indices constructed by the HCSO.
At present, the HCSO compiles a separate house price index for new and used houses, and also publishes
detail in the chapter describing the robustness analysis.
Figure 7
The MNB’s real house price index for cities by region
(2010 = 100%)
60
70
80
90
100
110
120
130
140
150
160
60
70
80
90
100
110
120
130
140
150
160 Per cent Per cent
Cies – South West Hungary
Cies – Northern Great Plains
Cies – Northern Hungary
Cies – Central Transdanubia
Cies – Central Hungary
Cies – Western Hungary
Cies – Southern Great Plains
2008 Q1
2008 Q3
2007 Q1
2007 Q3
2006 Q1
2006 Q3
2005 Q1
2005 Q3
2004 Q1
2004 Q3
2003 Q1
2003 Q3
2002 Q1
2002 Q3
2001 Q1
2001 Q3
2009 Q1
2009 Q3
2010 Q1
2010 Q3
2011 Q1
2011 Q3
2012 Q1
2012 Q3
2013 Q1
2013 Q3
2014 Q1
2014 Q3
2015 Q1
2015 Q3
2016 Q1
Figure 8
Quarterly change of the MNB’s aggregate house price index and the HCSO’s national house price index
–10
–8
–6
–4
–2
0
2
4
6
8
10
–10
–8
–6
–4
–2
0
2
4
6
8
10 Per cent Per cent
MNB House prices index
HCSO House prices index
2008 Q1
2008 Q2
2008 Q3
2008 Q4
2009 Q1
2009 Q2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
2011 Q1
2011 Q2
2011 Q3
2011 Q4
2012 Q1
2012 Q2
2012 Q3
2012 Q4
2013 Q1
2013 Q2
2013 Q3
2013 Q4
2014 Q1
2014 Q2
2014 Q3
2014 Q4
2015 Q1
2015 Q2
2015 Q3
2015 Q4
2016 Q1
2016 Q2
Note: The black dot ted line indicates the 95 per cent confidence interval of the MNB’s aggregate house index.
Source: HCSO, MNB.
MNB Occa siONal PaPers 127 • 2017 27
and villages are described separately. Due to the limited scope of this study, we only present the Southern
Transdanubian region for the city indices constructed for the individual regions, because this region aptly
On the other hand, since the regression model is run over and over again for each quarter pair, of all the results
Table 6 shows the regression outputs of the Budapest model. Run on a sample covering 2015 Q3 and 2015
Q4, the model has a 69 per cent explanatory power.17
ε0.0278=1.028
cent.
homes in the inner districts18
ceteris paribus, more expensive
on average.
17 It should be noted that backcasting the data improves the explanatory power of the regressions artificially.
18
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201728
Table 6
Regression results of the Budapest house price index model for 2015 Q4
Number of obs = 23386
F(32, 23353) = 1602,81
Prob > F = 0
R-squared = 0,6871
Adj R-squared = 0,6867
Root MSE = 0,3176
Price (ln) tP>|t| Interval]
Quarter (reference: 2015 Q3)
2015 Q4 0.0278 0.0042 6.6700 0.0000 0.0196 0.0360
Condonimium 1.2556 0.0767 16.3600 0.0000 1.1053 1.4060
Panel block of flats 1.4491 0.0819 17.7000 0.0000 1.2886 1.6096
Detached house (inner city) 1.0233 0.0718 14.2500 0.0000 0.8825 1.1640
Detached house (outer city) 1.1667 0.0692 16.8700 0.0000 1.0311 1.3023
Condonimium –0.0416 0.0097 –4.3000 0.0000 –0.0606 –0.0227
Panel block of flats –0.0954 0.0117 –8.1500 0.0000 –0.1183 –0.0725
Family house (inner city) 0.0124 0.0095 1.3000 0.1920 –0.0062 0.0311
Family house (outer city) –0.0265 0.0083 –3.2000 0.0010 –0.0427 –0.0102
Districts of Budapest (reference: 1)
2 0.0361 0.0197 1.8300 0.0670 –0.0026 0.0747
3 –0.3261 0.0187 –17.4200 0.0000 –0.3628 –0.2894
4 –0.4888 0.0194 –25.2400 0.0000 –0.5267 –0.4508
5 0.2110 0.0214 9.8700 0.0000 0.1691 0.2528
6 –0.1000 0.0198 –5.0600 0.0000 –0.1387 –0.0613
7 –0.2811 0.0190 –14.8000 0.0000 –0.3183 –0.2439
8 –0.4511 0.0188 –23.9800 0.0000 –0.4880 –0.4142
9 –0.2544 0.0196 –12.9500 0.0000 –0.2929 –0.2159
10 –0.6225 0.0197 –31.5400 0.0000 –0.6612 –0.5838
11 –0.1284 0.0185 –6.9500 0.0000 –0.1646 –0.0921
12 –0.0038 0.0205 –0.1900 0.8530 –0.0439 0.0363
13 –0.1747 0.0186 –9.4000 0.0000 –0.2111 –0.1382
14 –0.3210 0.0185 –17.3300 0.0000 –0.3574 –0.2847
15 –0.5841 0.0199 –29.3500 0.0000 –0.6231 –0.5451
16 –0.3297 0.0219 –15.0700 0.0000 –0.3725 –0.2868
17 –0.5175 0.0211 –24.5200 0.0000 –0.5588 –0.4761
18 –0.5632 0.0199 –28.2700 0.0000 –0.6023 –0.5242
19 –0.5846 0.0212 –27.5800 0.0000 –0.6261 –0.5430
20 –0.6932 0.0208 –33.2700 0.0000 –0.7340 –0.6523
21 –0.7378 0.0199 –37.0600 0.0000 –0.7768 –0.6988
22 –0.4066 0.0222 –18.3100 0.0000 –0.4501 –0.3631
23 –1.0009 0.0373 –26.8300 0.0000 –1.0741 –0.9278
New flat (reference: used) 0.3507 0.0162 21.6400 0.0000 0.3189 0.3824
Constant 12.5399 0.1524 82.3000 0.0000 12.2413 12.8386
MNB Occa siONal PaPers 127 • 2017 29
Table 7
Combined partial effect of the linear and squared terms of the NIA variable by average NIA
Mean (sq metre) of
sub-samples by
property type
Average partial effect
by property type
Mean (sq metre) of
total sample by
settlement type
Average partial effect
by settlement type
Budapest
Condonimium 55.6 0.9209 61.2 0.9129
Panel block of flats 52.0 0.6953 61.2 0.6639
Family house (inner
city) 119.5 1.1421 61.2 1.1255
Family house (outer
city) 102.1 0.9217 61.2 0.9488
Condonimium 58.0 0.8869 70.6 0.8683
Panel block of flats 52.7 0.9079 70.6 0.8856
Family house (county
seat) 92.4 2.1536 70.6 2.0193
Family house (other) 84.5 2.6631 70.6 2.5259
Municipalities
Condonimium 67.3 1.1013 80.1 0.5848
Homestead 75.0 2.7992 80.1 2.6731
Family house 80.8 3.0359 80.1 3.0512
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201730
Table 8
Regression results of the Southern Transdanubian house price index model for 2015 Q4
Number of obs = 4300
F(21, 4278) = 342.65
Prob > F = 0
R-squared = 0.6271
Adj R-squared = 0.6253
Root MSE = 0.4021
Price (ln) tP>|t| Interval]
Quarter (reference: 2015 Q3)
2015 Q4 –0.0270 0.0124 –2.1800 0.0290 –0.0512 –0.0028
Condonimium 1.2706 0.4071 3.1200 0.0020 0.4725 2.0686
Panel block of flats 1.2105 0.4295 2.8200 0.0050 0.3685 2.0526
Family house (inner city) –0.1041 0.3688 –0.2800 0.7780 –0.8272 0.6189
Family house (outer city) –0.7235 0.3734 –1.9400 0.0530 –1.4556 0.0086
Condonimium –0.0472 0.0503 –0.9400 0.3480 –0.1458 0.0513
Panel block of flats –0.0382 0.0570 –0.6700 0.5030 –0.1499 0.0736
Family house (inner city) 0.2494 0.0428 5.8200 0.0000 0.1654 0.3334
Family house (outer city) 0.3816 0.0436 8.7500 0.0000 0.2961 0.4671
County (reference: Baranya)
Somogy 0.0635 0.0308 2.0600 0.0390 0.0031 0.1238
Tolna 0.0123 0.0429 0.2900 0.7740 –0.0717 0.0964
New flat (reference: used) 0.3202 0.0417 7.6800 0.0000 0.2385 0.4019
Agglomeration (reference: not
agglomeration)
Agglomeration of Pécs 0.2244 0.0409 5.4900 0.0000 0.1442 0.3045
Recreational area (reference: not
recreational area)
0.8919 0.0645 13.8300 0.0000 0.7655 1.0182
–0.0044 0.0528 –0.0800 0.9330 –0.1079 0.0990
Total income per capita (ln) 1.0131 0.0605 16.7500 0.0000 0.8945 1.1317
Distance from Budapest(ln) –0.0978 0.0771 –1.2700 0.2050 –0.2490 0.0534
Distance from county seats (ln) –0.0801 0.0090 –8.8600 0.0000 –0.0979 –0.0624
Population(ln) –0.1172 0.0200 –5.8500 0.0000 –0.1565 –0.0780
Size of municipalities(ln) 0.1098 0.0251 4.3700 0.0000 0.0605 0.1591
–0.0105 0.0034 –3.0700 0.0020 –0.0172 –0.0038
Constant –1.0344 1.2943 –0.8000 0.4240 –3.5719 1.5031
MNB Occa siONal PaPers 127 • 2017 31
Balaton – which can be accessed more easily from Budapest than other areas in the Southern Transdanubian
however, proved to be an important factor. Moreover, an increase in net per capita income for the municipality
are, ceteris paribus,
that the inclusion of municipality-level variables contributes the largest share of value added in smaller, typically
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201732
Table 9
Regression results of the village house price index model for 2015 Q4
Number of obs = 16006
F(46, 15959) = 974,21
Prob > F = 0
R-squared = 0,7374
Adj R-squared = 0,7366
Root MSE = 0,5868
Price (ln) tP>|t| Interval]
Quarter (reference: 2015 Q3)
2015 Q4 –0.0140 0.0093 –1.5000 0.1340 –0.0323 0.0043
Condonimium 13.6006 0.3561 38.2000 0.0000 12.9027 14.2986
Municipality 11.0535 0.3412 32.4000 0.0000 10.3847 11.7223
Family House 10.8612 0.3288 33.0300 0.0000 10.2166 11.5057
Condonimium –1.4847 0.0442 –33.6300 0.0000 –1.5712 –1.3981
Municipality –0.9559 0.0453 –21.1000 0.0000 –1.0447 –0.8671
Family House –0.8908 0.0370 –24.0800 0.0000 –0.9634 –0.8183
County (reference: Pest)
Györ-Moson-Sopron 0.0581 0.0416 1.3900 0.1630 –0.0235 0.1397
Vas 0.0515 0.0475 1.0900 0.2780 –0.0415 0.1446
…
New flat (reference: used) 0.3895 0.0827 4.7100 0.0000 0.2274 0.5515
Agglomeration (reference: not agglomeration)
Szeged Agglomeration 0.4817 0.0542 8.8900 0.0000 0.3755 0.5880
Pécs Agglomeration 0.4383 0.0586 7.4800 0.0000 0.3235 0.5531
Debrecen Agglomeration 0.5747 0.0606 9.4800 0.0000 0.4559 0.6935
Miskolc Agglomeration 0.3117 0.0509 6.1200 0.0000 0.2119 0.4115
Székesfehérvár Agglomeration 0.0239 0.0430 0.5600 0.5780 –0.0603 0.1081
Budapest Agglomeration –0.0293 0.0303 –0.9700 0.3330 –0.0886 0.0300
Györ Agglomeration –0.3023 0.0399 –7.5800 0.0000 –0.3804 –0.2241
Sopron Agglomeration 0.6258 0.0845 7.4000 0.0000 0.4601 0.7915
Recreational area (reference: not
recreational area)
0.8194 0.0325 25.2200 0.0000 0.7557 0.8830
0.3928 0.0314 12.5100 0.0000 0.3312 0.4543
Dunakanyar 0.1991 0.0275 7.2400 0.0000 0.1452 0.2530
0.3469 0.0327 10.6200 0.0000 0.2829 0.4110
0.1432 0.0555 2.5800 0.0100 0.0344 0.2520
0.0100 0.0437 0.2300 0.8190 –0.0756 0.0956
0.0309 0.0424 0.7300 0.4660 –0.0523 0.1141
Total income per capita (ln) 0.5281 0.0234 22.6100 0.0000 0.4823 0.5739
Distance from Budapest(ln) –0.2668 0.0273 –9.7900 0.0000 –0.3202 –0.2134
Distance from county seats (ln) –0.1884 0.0123 –15.2800 0.0000 –0.2126 –0.1642
Population(ln) 0.1156 0.0089 12.9500 0.0000 0.0981 0.1330
Size of municipalities(ln) –0.0815 0.0101 –8.0300 0.0000 –0.1014 –0.0616
Constant –20.8978 0.8095 –25.8100 0.0000 –22.4845 –19.3110
Note: While the model includes all counties as a dummy variable, due to space constraints, only the counties with a positive sign were presented
in the table.
MNB Occa siONal PaPers 127 • 2017 33
6 Robustness analysis
ceteris paribus,
Figure 9
Robustness analysis for the filtering out of transactions by business organisations
(quarterly price changes)
–10
–8
–6
–4
–2
0
2
4
6
8
10
–10
–8
–6
–4
–2
0
2
4
6
8
10
Per cent Per cent
Aggregated MNB house price index
Without business organisaons
2008 Q1
2008 Q2
2008 Q3
2008 Q4
2009 Q1
2009 Q2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
2011 Q1
2011 Q2
2011 Q3
2011 Q4
2012 Q1
2012 Q2
2012 Q3
2012 Q4
2013 Q1
2013 Q2
2013 Q3
2013 Q4
2014 Q1
2014 Q2
2014 Q3
2014 Q4
2015 Q1
2015 Q2
2015 Q3
2015 Q4
2016 Q1
2016 Q2
Note: The black dot ted line indicates the 95 per cent confidence interval of the MNB’s aggregate house index.
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201734
where it was smaller than 150m22 limit within the range of 100m2 and 200m2 did
Figure 10
Robustness analysis for the backcasting of the useful NIA
(quarterly price changes)
–10
–8
–6
–4
–2
0
2
4
6
8
10
–10
–8
–6
–4
–2
0
2
4
6
8
10
Per cent Per cent
Aggregated MNB house price index
Without backcasng
2008 Q1
2008 Q2
2008 Q3
2008 Q4
2009 Q1
2009 Q2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
2011 Q1
2011 Q2
2011 Q3
2011 Q4
2012 Q1
2012 Q2
2012 Q3
2012 Q4
2013 Q1
2013 Q2
2013 Q3
2013 Q4
2014 Q1
2014 Q2
2014 Q3
2014 Q4
2015 Q1
2015 Q2
2015 Q3
2015 Q4
2016 Q1
2016 Q2
Note: The black dot ted line indicates the 95 per cent confidence interval of the MNB’s aggregate house index.
Figure 11
Robustness analysis for the first entry of the useful NIA
(quarterly price changes)
–10
–8
–6
–4
–2
0
2
4
6
8
10
–10
–8
–6
–4
–2
0
2
4
6
8
10
Per cent Per cent
Aggregated MNB house price index
First entry of the useful NIA (100)
Second entry of the useful NIA (200)
2008 Q1
2008 Q2
2008 Q3
2008 Q4
2009 Q1
2009 Q2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
2011 Q1
2011 Q2
2011 Q3
2011 Q4
2012 Q1
2012 Q2
2012 Q3
2012 Q4
2013 Q1
2013 Q2
2013 Q3
2013 Q4
2014 Q1
2014 Q2
2014 Q3
2014 Q4
2015 Q1
2015 Q2
2015 Q3
2015 Q4
2016 Q1
2016 Q2
Note: The black dot ted line indicates the 95 per cent confidence interval of the MNB’s aggregate house index.
MNB Occa siONal PaPers 127 • 2017 35
by regression models, with separate models constructed for each municipality type. The main reason for this
reference horizon.
Figure 12
Robustness analysis for the backcasting of the missing values of the NIA variable
(quarterly price changes)
–10
–8
–6
–4
–2
0
2
4
6
8
10
–10
–8
–6
–4
–2
0
2
4
6
8
10
Per cent Per cent
Aggregated MNB house price index
With the backcasng of HCSO
2008 Q1
2008 Q2
2008 Q3
2008 Q4
2009 Q1
2009 Q2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
2011 Q1
2011 Q2
2011 Q3
2011 Q4
2012 Q1
2012 Q2
2012 Q3
2012 Q4
2013 Q1
2013 Q2
2013 Q3
2013 Q4
2014 Q1
2014 Q2
2014 Q3
2014 Q4
2015 Q1
2015 Q2
2015 Q3
2015 Q4
2016 Q1
2016 Q2
Note: The black dot ted line indicates the 95 per cent confidence interval of the MNB’s aggregate house index.
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201736
Figure 13
Robustness analysis for first-step filtering
(quarterly price changes)
–10
–8
–6
–4
–2
0
2
4
6
8
10
–10
–8
–6
–4
–2
0
2
4
6
8
10
Per cent Per cent
Aggregated MNB house price index
Alternave first-step filtering (lower limit: 4000 /sqr metre)
2008 Q1
2008 Q2
2008 Q3
2008 Q4
2009 Q1
2009 Q2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
2011 Q1
2011 Q2
2011 Q3
2011 Q4
2012 Q1
2012 Q2
2012 Q3
2012 Q4
2013 Q1
2013 Q2
2013 Q3
2013 Q4
2014 Q1
2014 Q2
2014 Q3
2014 Q4
2015 Q1
2015 Q2
2015 Q3
2015 Q4
2016 Q1
2016 Q2
Note: The black dot ted line indicates the 95 per cent confidence interval of the MNB’s aggregate house index.
Figure 14
Robustness analysis for outlier filtering
(quarterly price changes)
–10
–8
–6
–4
–2
0
2
4
6
8
10
–10
–8
–6
–4
–2
0
2
4
6
8
10
Per cent Per cent
Aggregated MNB house price index
First-step and second-step filtering with the methodology of HCSO
2008 Q1
2008 Q2
2008 Q3
2008 Q4
2009 Q1
2009 Q2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
2011 Q1
2011 Q2
2011 Q3
2011 Q4
2012 Q1
2012 Q2
2012 Q3
2012 Q4
2013 Q1
2013 Q2
2013 Q3
2013 Q4
2014 Q1
2014 Q2
2014 Q3
2014 Q4
2015 Q1
2015 Q2
2015 Q3
2015 Q4
2016 Q1
2016 Q2
Note: The black dot ted line indicates the 95 per cent confidence interval of the MNB’s aggregate house index.
MNB Occa siONal PaPers 127 • 2017 37
19
19 For the purposes of the robustness analysis, we included the 2013 values of the municipality-level variables in the regression equations.
Figure 15
Robustness analysis for second-step filtering
(quarterly price changes)
–10
–8
–6
–4
–2
0
2
4
6
8
10
–10
–8
–6
–4
–2
0
2
4
6
8
10
Per cent Per cent
Aggregated MNB house price index
Tesng for second-step filtering with two indicators
Tesng for second-step filtering with four indicators
2008 Q1
2008 Q2
2008 Q3
2008 Q4
2009 Q1
2009 Q2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
2011 Q1
2011 Q2
2011 Q3
2011 Q4
2012 Q1
2012 Q2
2012 Q3
2012 Q4
2013 Q1
2013 Q2
2013 Q3
2013 Q4
2014 Q1
2014 Q2
2014 Q3
2014 Q4
2015 Q1
2015 Q2
2015 Q3
2015 Q4
2016 Q1
2016 Q2
Note: The black dot ted line indicates the 95 per cent confidence interval of the MNB’s aggregate house index.
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201738
20
20 It hinders the testing of the temporal stability of the parameters that even the constant is different over time, the effect of which on the
parameters of individual variables cannot be factored in properly.
Figure 16
Robustness analysis for the estimation methodology
(quarterly price changes)
–10
–8
–6
–4
–2
0
2
4
6
8
10
–10
–8
–6
–4
–2
0
2
4
6
8
10
Per cent Per cent
Aggregated MNB house price index
Total me interval esmate
2008 Q1
2008 Q2
2008 Q3
2008 Q4
2009 Q1
2009 Q2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
2011 Q1
2011 Q2
2011 Q3
2011 Q4
2012 Q1
2012 Q2
2012 Q3
2012 Q4
2013 Q1
2013 Q2
2013 Q3
2013 Q4
2014 Q1
2014 Q2
2014 Q3
2014 Q4
2015 Q1
2015 Q2
2015 Q3
2015 Q4
2016 Q1
2016 Q2
Note: The black dot ted line indicates the 95 per cent confidence interval of the MNB’s aggregate house index.
MNB Occa siONal PaPers 127 • 2017 39
Figure 17
Dynamic analysis of the parameters of selected municipality-level variables in the models constructed for villages
–0.6
–0.4
–0.2
0.0
0.2
0.4
0.6
0.8
Per cent Per cent
–0.6
–0.4
–0.2
0.0
0.2
0.4
0.6
0.8
Total income per capita of municipalies
Distance from County seat
Populaon
Size of Municipality
Distance from Budapest
2008 Q1
2008 Q2
2008 Q3
2008 Q4
2009 Q1
2009 Q2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
2011 Q1
2011 Q2
2011 Q3
2011 Q4
2012 Q1
2012 Q2
2012 Q3
2012 Q4
2013 Q1
2013 Q2
2013 Q3
2013 Q4
2014 Q1
2014 Q2
2014 Q3
2014 Q4
2015 Q1
2015 Q2
2015 Q3
2015 Q4
2016 Q1
2016 Q2
MNB Occa siONal PaPers 127 • 201740
7 Conclusions
Hungarian housing market developments are of key importance both for the banking sector and the real
economy and accordingly, it is also the central bank’s interest to gain insight into these processes. With that in
from the NTCA for the reference period 1990–2015, which resulted in the longest and broadest Hungarian
terms.
MNB Occa siONal PaPers 127 • 2017 41
References
A Guide to House Price Indexes. 6464
Journal
of the American Stascal Associaon, Vol. 58, pp. 933–942
Regression Diagnoscs: Idenfying Inuenal Data and Sources
of Collinearity. Wiley.
cases. In: Modern Methods of Data Analysis (eds.: 257–291. Sage. Newbury Park, CA
Stascal Science, Vol. 1, pp. 379–393
Cook, R. D. (1977): Technometrics, Vol. 19, pp. 15–18
Court, A. T. (1939): In: The Dynamics of Automobile Demand.
Crone, T. M. – Voith, R. P. (1992): A Comparison of Methods. Journal of
Housing Economics, 2(4), pp. 324–338
European Central Bank (2011): Financial Stability Review,
European Central Bank (2015): Financial Stability Review, November 2015
Eurostat (2013): Handbook on Residenal Property Price Indices (RPPIs),
Journal of Urban Economics, 35(2), pp. 221–244
Griliches, Z. (1961, 1971a): Hedonic Price Indexes for Automobiles: An Econometric Analysis for Quality Change.
In: Z. Griliches (ed.): Price Indexes and Quality Change: Studies in New Methods of Measurement. Harvard
Cambridge, MA
(1995): Mulvariate Data Analysis.
Horváth, Á. (2007): Ingatlanárindexek számításának módszertana (On the Methodology of Real Estate Price
Indexes). Stascal Review, Vol. 85, Issue 3
Stascal Review, Vol. 87, Issue 6
Indexes. Real Estate Economics, 23(2), pp. 187–205
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201742
(1966): A New Approach to Consumer Theory. Journal of Polical Economy, Vol. 74, pp. 132–157
American Real Estate
and Urban Economics Associaon Journal, 12(1), pp. 30–49
O’Brien, D. – Westermann, T. – Vidalis, N. (2014): Review of the ECB’s house price valuaons framework.
Oikarinen, E. (2012):
Journal of Housing Economics, 21(1), pp. 41–54
Rosen, S. (1974): Hedonic Prices and Implicit Markets: Journal of
Polical Economy, Vol. 82, pp. 34–55
Standard & Poor’s (2009): S&P Case-Shiller Home Price Indices Index Methodology. Standard & Poor’s.
A New Index of Average House Prices. Economic Trends, 348, pp.
134–138
Vellemen, P. F. – Welsch, R. E. (1981): The American Stascian,
Vol. 35, pp. 234–242
Wang, T. – Zorn, P. M. (1997): What’s the Aim of the
Game? Journal of Housing Economics, 6(2), pp. 93–118
Welsch, R. E. (1982): In: Modern Data Analysis (eds.:
and A. F. Siegel), pp. 149–169 Academic Press.
Stascal Review, Vol. 90, Issue 7–8
MNB Occa siONal PaPers 127 • 2017 43
Annex
Figure 18
Partial effect of the interaction between NIA and property type in a model specified for Budapest
0.0
0.3
0.6
0.9
1.2
1.5
0.0
0.3
0.6
0.9
1.2
1.5
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
105
110
115
120
125
130
135
140
145
150
Per cent Per cent
Panel block of flats
Condominium
Family house (inner city)
Family house (outer city)
Note: The horizontal axis indicates the size of the property expressed in square metres. The figure shows the price increase generated, ceteris
paribus, by a 1 per cent increase in the NIA of a given property based on an estimate run on a sample covering 2015 Q3 and 2015 Q4. If the model
only included linear terms, the f igure would present constant func tions.
Figure 19
Partial effect of the interaction between NIA and property type in a model specified for Southern Transdanubian cities
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
105
110
115
120
125
130
135
140
145
150
Per cent Per cent
Panel block of flats
Condominium
Family house (County seat)
Family house (other)
Note: The horizontal axis indicates the size of the property expressed in square metres. The figure shows the price increase generated, ceteris
paribus, by a 1 per cent increase in the NIA of a given property based on an estimate run on a sample covering 2015 Q3 and 2015 Q4. If the model
only included linear terms, the f igure would present constant func tions.
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201744
Figure 20
Partial effect of the interaction between NIA and property type in a model specified for municipalities
–2
–1
0
1
2
3
4
5
6
7
8
–2
–1
0
1
2
3
4
5
6
7
8
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
105
110
115
120
125
130
135
140
145
150
Per cent Per cent
Homestead
Condominium
Family house
Note: The horizontal axis indicates the size of the property expressed in square metres. The figure shows the price increase generated, ceteris
paribus, by a 1 per cent increase in the NIA of a given property based on an estimate run on a sample covering 2015 Q3 and 2015 Q4. If the model
only included linear terms, the f igure would present constant func tions.
MNB Occa siONal PaPers 127 • 2017 45
Table 10
Distribution of category variables for the period of 1990–2000
Budapest Cities Municipalities Tot al
number
%number
%number
%number
%
Property type
Family house 204,864 56.6 128,114 96.0 332,978 67.2
County seat
51,585 14.2 51,585 10.4
Other
153,279 42.3 153,279 30.9
Flat 157,172 43.4 5,311 4.0 162,483 32.8
Agglomeration
Not 237,585 65.6 113,917 85.4 351,502 70.9
Szeged 25,223 7.0 1,831 1.4 27,054 5.5
Pécs 4,051 1.1 482 0.4 4,533 0.9
Debrecen 25,786 7.1 1,649 1.2 27,435 5.5
Miskolc 678 0.2 103 0.1 781 0.2
Székesfehérvár 8,726 2.4 2,703 2.0 11,429 2.3
Budapest 48,819 13.5 9,344 7.0 58,163 11.7
Györ 7,880 2.2 2,937 2.2 10,817 2.2
Sopron 3,289 0.9 459 0.3 3,748 0.8
Recreational area
Not 289,246 79.9 106,851 80.1 396,097 79.9
8,963 2.5 2,903 2.2 11,866 2.4
2,580 0.7 3,860 2.9 6,440 1.3
Dunakanyar 20,604 5.7 6,265 4.7 26,869 5.4
27,918 7.7 7,201 5.4 35,119 7.1
6,079 1.7 1,270 1.0 7,349 1.5
1,865 0.5 2,626 2.0 4,491 0.9
4,782 1.3 2,449 1.8 7,231 1.5
County
Budapest
Baranya 6,475 1.8 2,315 1.7 8,790 1.8
8,909 2.5 3,402 2.5 12,311 2.5
Békés 18,487 5.1 3,476 2.6 21,963 4.4
1,221 0.3 700 0.5 1,921 0.4
Csongrád 39,342 10.9 5,658 4.2 45,000 9.1
18,483 5.1 8,244 6.2 26,727 5.4
Györ-Moson-Sopron 14,180 3.9 5,935 4.4 20,115 4.1
38,106 10.5 6,473 4.9 44,579 9.0
Heves 35,964 9.9 24,230 18.2 60,194 12.1
31,794 8.8 9,453 7.1 41,247 8.3
Nógrád 232 0.1 199 0.1 431 0.1
Pest 63,103 17.4 22,050 16.5 85,153 17.2
Somogy 20,525 5.7 13,288 10.0 33,813 6.8
Szabolcs-Szatmár-Bereg 6,903 1.9 3,228 2.4 10,131 2.0
21,571 6.0 6,729 5.0 28,300 5.7
Tolna 8,111 2.2 4,375 3.3 12,486 2.5
Vas 10,515 2.9 5,062 3.8 15,577 3.1
Veszprém 6,579 1.8 2,609 2.0 9,188 1.9
Zala 11,537 3.2 5,999 4.5 17,536 3.5
Note: Values received after first-step outlier filtering.
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201746
Table 11
Distribution of category variables for the period of 2001–2007
Budapest Cities Municipalities Tot al
number
%number
%number
%number
%
Property type
Family house 52,184 14.4 419,936 56.8 299,261 96.6 771,452 54.6
County seat
102,739 13.9 102,753 7.3
Other
317,197 42.9 317,240 22.5
Flat 310,364 85.6 319,631 43.2 10,561 3.4 640,685 45.4
Agglomeration
Not 284,580 100.0 484,678 65.4 267,583 86.3 1,037,006 77.6
Szeged 30,617 4.1 2,705 0.9 33,326 2.5
Pécs 28,627 3.9 3,407 1.1 32,038 2.4
Debrecen 39,596 5.3 2,817 0.9 42,418 3.2
Miskolc 29,698 4.0 3,857 1.2 33,559 2.5
Székesfehérvár 18,461 2.5 5,845 1.9 24,308 1.8
Budapest 82,900 11.2 15,652 5.0 98,563 7.4
Györ 17,232 2.3 7,157 2.3 24,391 1.8
Sopron 9,554 1.3 1,084 0.3 10,639 0.8
Recreational area
Not 363,415 100.0 615,302 83.0 266,038 85.8 1,244,938 88.0
17,325 2.3 5,570 1.8 22,897 1.6
4,357 0.6 6,556 2.1 10,914 0.8
Dunakanyar 32,111 4.3 11,160 3.6 43,275 3.1
46,747 6.3 8,902 2.9 55,655 3.9
15,634 2.1 2,916 0.9 18,552 1.3
3,226 0.4 4,507 1.5 7,733 0.5
6,661 0.9 4,458 1.4 11,120 0.8
County
Budapest 284,580 100.0 284,680 21.3
Baranya 42,728 5.8 15,933 5.1 58,667 4.4
46,256 6.2 17,676 5.7 63,938 4.8
Békés 41,502 5.6 12,348 4.0 53,856 4.0
52,973 7.1 27,196 8.8 80,176 6.0
Csongrád 50,211 6.8 8,939 2.9 59,157 4.4
41,415 5.6 17,895 5.8 59,316 4.4
Györ-Moson-Sopron 34,489 4.7 14,765 4.8 49,259 3.7
63,513 8.6 13,521 4.4 77,043 5.8
Heves 23,344 3.1 20,773 6.7 44,120 3.3
32,409 4.4 10,442 3.4 42,855 3.2
Nógrád 12,072 1.6 12,008 3.9 24,082 1.8
Pest 107,057 14.4 37,470 12.1 144,541 10.8
Somogy 27,249 3.7 20,255 6.5 47,508 3.6
Szabolcs-Szatmár-Bereg 40,785 5.5 23,949 7.7 64,740 4.8
43,153 5.8 14,470 4.7 57,629 4.3
Tolna 17,757 2.4 12,240 3.9 29,999 2.2
Vas 19,907 2.7 9,134 2.9 29,044 2.2
Veszprém 25,451 3.4 10,410 3.4 35,864 2.7
Zala 19,092 2.6 10,683 3.4 29,778 2.2
Note: Values received after first-step outlier filtering.
MNB Occa siONal PaPers 127 • 2017 47
Table 12
Distribution of category variables for the period after 2008
Budapest Cities Municipalities Tot al
number
%number
%number
%number
%
Property type
Family Houses 44,574 13.1 303,270 47.6 247,671 95.6 595,576 47.9
Inner city of Budapest
12,066 3.6 12,070 1.0
Outer city of Budapest
32,508 9.6 32,518 2.6
County seat
69,584 10.9 69,595 5.6
Other cities
233,686 36.7 233,723 18.8
Condominium 276,948 81.5 256,275 40.2 11,380 4.4 544,725 43.8
Panel block of flats 18,104 5.3 77,686 12.2 95,808 7.7
Homestead 7,023 2.7 7,023 0.6
New 7,710 2.3 14,868 2.3 2,045 0.8 24,623 2.0
332,481 97.7 622,543 97.7 264,106 99.2 1,219,130 98.0
Districts of Budapest
I 6,051 1.8 6,051 1.8
II 17,371 5.1 17,371 5.1
III 22,711 6.7 22,711 6.7
IV 16,716 4.9 16,716 4.9
V 8,638 2.5 8,638 2.5
VI 13,304 3.9 13,304 3.9
VII 18,202 5.4 18,202 5.4
VIII 21,258 6.3 21,258 6.3
15,915 4.7 15,915 4.7
14,444 4.3 14,444 4.3
30,253 8.9 30,253 8.9
11,361 3.3 11,361 3.3
29,072 8.6 29,072 8.6
28,358 8.3 28,358 8.3
12,060 3.6 12,060 3.6
9,463 2.8 9,463 2.8
10,558 3.1 10,558 3.1
14,397 4.2 14,397 4.2
9,378 2.8 9,378 2.8
10,390 3.1 10,390 3.1
10,663 3.1 10,663 3.1
6,906 2.0 6,906 2.0
2,199 0.6 2,199 0.6
The table is continued on the next page.
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201748
Budapest Cities Municipalities Tot al
number
%number
%number
%number
%
Agglomeration
Not 340,191 100.0 396,399 62.2 225,291 84.6 962,043 77.3
Szeged 28,978 4.5 3,199 1.2 32,182 2.6
Pécs 25,333 4.0 2,511 0.9 27,848 2.2
Debrecen 33,811 5.3 2,032 0.8 35,848 2.9
Miskolc 24,972 3.9 3,108 1.2 28,084 2.3
Székesfehérvár 15,790 2.5 5,129 1.9 20,921 1.7
Budapest 81,590 12.8 15,659 5.9 97,262 7.8
Györ 19,773 3.1 7,788 2.9 27,564 2.2
Sopron 10,765 1.7 1,434 0.5 12,201 1.0
Recreational Area
not 340,191 100.0 517,811 81.2 223,535 84.0 1,081,718 87.0
21,023 3.3 7,451 2.8 28,477 2.3
3,781 0.6 6,850 2.6 10,632 0.9
Dunakanyar 30,382 4.8 9,530 3.6 39,917 3.2
39,197 6.1 7,601 2.9 46,804 3.8
16,575 2.6 3,515 1.3 20,093 1.6
2,302 0.4 3,360 1.3 5,662 0.5
6,340 1.0 4,309 1.6 10,650 0.9
County
Budapest 340,191 100.0 340,291 27.4
Baranya 35,100 5.5 11,551 4.3 46,657 3.8
45,608 7.2 17,886 6.7 63,501 5.1
Békés 29,733 4.7 8,715 3.3 38,453 3.1
43,747 6.9 21,705 8.2 65,459 5.3
Csongrád 44,632 7.0 9,900 3.7 54,539 4.4
33,606 5.3 15,063 5.7 48,674 3.9
Györ-Moson-Sopron 39,028 6.1 17,105 6.4 56,139 4.5
50,998 8.0 9,984 3.8 60,990 4.9
Heves 18,433 2.9 15,531 5.8 33,967 2.7
25,922 4.1 8,619 3.2 34,545 2.8
Nógrád 8,126 1.3 9,182 3.4 17,309 1.4
Pest 100,800 15.8 33,258 12.5 134,074 10.8
Somogy 22,696 3.6 16,033 6.0 38,733 3.1
Szabolcs-Szatmár-Bereg 28,343 4.4 18,257 6.9 46,604 3.7
29,313 4.6 10,422 3.9 39,740 3.2
Tolna 14,635 2.3 9,242 3.5 23,879 1.9
Vas 18,536 2.9 8,735 3.3 27,274 2.2
Veszprém 28,525 4.5 12,823 4.8 41,352 3.3
Zala 19,630 3.1 12,140 4.6 31,773 2.6
Note: Values received after first-step outlier filtering.
MNB Occa siONal PaPers 127 • 2017 49
Table 13
Descriptive statistics of continuous variables until 1990–2000
Mean
Standard
deviation
Minimum 5th
percentile
25th
percentile Median 75th
percentile
95th
percentile Maximum
Budapest
Cities 362,037 3,253,945 4,177,062 10,000 449,000 1,300,000 2,350,000 4,000,000 8,769,000
400,000,000
Municipalities 133,425 1,959,699 3,050,159 11,000 150,000 500,000 1,100,000 2,400,000 6,000,000
185,000,000
Population
(capita)
Budapest and cities 362,803 62,429 102,095 0 5,421 15,266 33,203 73,648 203,648 2,018,035
Municipalities 133,425 2,204 1,489 0 375 1,090 1,906 3,004 5,104 8,615
Size (sqkm) Budapest and cities 362,803 15,658 12,674 0 2,586 6,163 10,477 20,997 46,165 52,516
Municipalities 133,425 3,468 2,398 0 779 1,811 2,871 4,491 7,918 28,458
Subsidies
(thousand
Budapest and cities 362,803 0 0 0 0 0 0 0 0 0
Municipalities 133,425 0 0 0 0 0 0 0 0 0
Distance
from Bp.
(minutes)
Budapest and cities 362,803 105 49 0 34 57 106 151 186 234
Municipalities 133,425 106 48 0 40 66 100 140 191 255
Distance
from county
seat
(minutes)
Budapest and cities 362,803 27 26 0 0 0 26 49 74 109
Municipalities 133,425 43 20 0 14 27 42 55 77 138
Income of
settlement
year)
Budapest
and cities 362,788 296,550 60,029 105,255 190,035 256,644 299,903 330,796 392,599 542,522
Municipalities 133,186 212,510 64,621 17,033 115,114 164,742 209,903 254,585 314,919 1,109,502
Net internal area of
residential property (sqm)
Budapest house
flat
Cities
house county
seat 51,585 69 36 15 36 53 64 76 118 499
house other 153,279 70 32 15 38 55 66 79 110 499
flat 157,172 55 16 15 35 47 53 60 78 498
Municipalities
house 128,114 70 23 15 45 59 69 79 100 498
flat 5,311 55 21 15 34 45 51 59 88 472
Note: Values received after first-step outlier filtering.
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201750
Table 14
Descriptive statistics of continuous variables until 2001–2007
Number of
Mean
Standard
deviation
Minimum 5th
percentile
25th
percentile Median 75th
percentile
95th
percentile Maximum
Budapest 363,415 14,013,435 15,189,257 69,360 3,000,000 7,400,000 10,500,000 16,460,000 33,250,000 665,000,000
Cities 741,363 8,109,460 8,674,177 57,000 800,000 3,900,000 6,500,000 10,000,000 20,300,000 763,440,000
Municipalities 310,107 4,519,724 6,750,351 57,000 250,000 1,000,000 2,800,000 5,585,000 15,000,000 602,844,032
Population
(capita)
Budapest and cities 1,104,778 601,809 775,145 1,085 6,074 21,291 81,818 1,697,343 1,719,342 1,739,569
Municipalities 310,107 2,144 1,564 0 330 980 1,833 2,900 5,201 9,983
Size (sqkm) Budapest and cities 1,104,778 27,496 19,889 575 3,106 9,145 21,673 52,512 52,516 52,516
Municipalities 310,107 3,417 2,405 0 784 1,701 2,778 4,488 8,020 28,458
Subsidies
(thousand
Budapest and cities 1,104,778 172,818 276,316 0 0 0 5,100 378,960 831,095 831,095
Municipalities 310,107 300 1,258 0 0 0 0 150 1,575 100,000
Distance
from Bp.
(minutes)
Budapest and cities 1,104,778 77 68 0 0 0 75 140 180 234
Municipalities 310,107 122 51 0 44 81 120 163 203 255
Distance
from county
seat
(minutes)
Budapest and cities 1,104,778 18 25 0 0 0 0 37 66 109
Municipalities 310,107 44 22 0 14 28 42 57 88 138
Income of
settlement
year)
Budapest
and cities 1,104,778 523,941 143,657 124,979 293,701 417,290 516,128 621,751 803,533 965,277
Municipalities 309,975 344,009 128,432 16,731 167,056 250,738 325,811 419,383 585,803 1,271,065
Net internal area of
residential property (sqm)
Budapest house 52,184 73 36 15 30 53 69 86 130 499
flat 310,364 55 25 15 27 40 51 65 98 498
Cities
house county
seat 102,739 69 30 15 35 53 65 78 120 495
house other 317,197 70 27 15 38 55 68 80 110 499
flat 319,631 56 19 15 34 48 54 61 85 498
Municipalities
house 299,261 73 22 15 45 60 72 83 103 496
flat 10,561 61 26 15 34 47 55 68 108 483
Note: Values received after first-step outlier filtering.
MNB Occa siONal PaPers 127 • 2017 51
Table 15
Descriptive statistics of continuous variables from 2008
Number
Mean Standard
deviation Minimum 5th
percentile
25th
percentile Median 75th
percentile
95th
percentile Maximum
Budapest 340,191 16,664,362 17,662,256 90,000 4,250,000 8,450,000 12,500,000 19,653,180 40,000,000 850,000,000
Cities 637,411 9,964,693 9,301,686 79,328 1,250,000 5,000,000 7,800,000 12,500,000 25,000,000 660,000,000
Municipalities 266,151 6,074,002 8,484,386 79,400 343,000 1,500,000 3,500,000 7,500,000 20,000,000 441,249,984
Population (capita) Budapest and cities 977,602 644,731 800,110 1,000 6,007 23,573 111,836 1,721,556 1,757,618 1,757,618
Municipalities 266,151 2,081 1,652 0 283 891 1,745 2,759 5,128 10,282
Size (sqkm) Budapest and cities 977,602 13,586 19,582 6 46 228 525 19,393 52,513 52,513
Municipalities 266,151 1,718 2,382 0 10 27 527 2,717 6,581 28,458
Subsidies (thousand
Budapest and cities 977,602 50,947 86,976 0 0 0 4,190 58,090 303,985 378,960
Municipalities 266,151 194 3,438 0 0 0 0 0 800 250,000
Distance from Bp.
(minutes)
Budapest and cities 977,602 71 66 0 0 0 63 128 178 236
Municipalities 266,151 118 50 0 38 78 117 157 200 259
Distance from county
seat (minutes)
Budapest and cities 977,602 17 25 0 0 0 0 34 67 113
Municipalities 266,151 44 23 0 14 28 42 57 87 149
Income of settlement
Budapest and cities 977,602 794,693 142,988 283,727 543,267 692,342 819,877 885,874 971,593 1,348,615
Municipalities 266,150 583,976 191,088 -47,703 318,639 449,726 564,139 691,716 926,838 2,531,264
Net internal area of residential property (sqm)
Budapest
Condominium 276,948 56 24 15 27 39 52 66 100 499
Panel block of flats 18,104 53 17 15 30 43 52 61 77 409
Family house inner
districts 12,066 112 69 15 39 70 100 128 250 499
Family house outer
districts 32,508 102 63 15 44 74 93 109 198 499
Cities
Condominium 256,275 58 21 15 33 47 55 65 95 499
Panel block of flats 77,686 53 13 15 35 47 53 59 73 471
Family house
county seat 69,584 89 42 15 45 69 83 97 156 498
Family house other 233,686 83 33 15 50 69 80 90 130 499
Municipalities
Condominium 11,380 66 30 15 34 49 60 76 119 485
Homestead 7,023 70 23 20 49 59 67 76 96 496
Family house 247,671 79 26 15 50 65 76 87 114 498
Note: Values received after first-step outlier filtering.
MNB Occa siONal PaPers 127 • 2017 53
MNB OCCASIONAL PAPERS 127
May 2017
6 Tartu u., Veszprém H-8200
MAGYAR NEMZETI BANK
MNB OccAsIONAl PAPERs 127 • 201754
