You are on page 1of 22

Mejia-Dorantes, L., Heddebaut, O., Jayet, H.

1 Analyzing house-price evolution to understand a deprived area: The Northern


2 part of France case study
3
4 Submission date: August 1st 2013
5
6 Number of words: 5,982 + 2 figures × 250 + 4 tables × 250 = 7,482
7
8 First and corresponding author: Lucia Mejia-Dorantes, Ph.D.
9 Affiliation: Researcher
10 Address: Fraunhofer Institute for Systems and Innovation Research ISI
11 Sustainability and Infrastructure Systems
12 Breslauer Straße 48
13 76139 Karlsruhe, Germany
14 Phone: (+49) 721 6809-653
15 e-mail: lucia.mejia-dorantes@isi.fraunhofer.de
16
17
18 Second author: Odile Heddebaut, Ph.D.
19 Affiliation: Researcher
20 IFSTTAR - The French Institute of Science and Technology for Transport,
21 Development and Networks.
22 Department of Transport Economics and Sociology
23 Address: 20 rue Elisée Reclus BP 70317
24 59666 Villeneuve d’Ascq Cedex France
25 Phone: (+33) 3 20 43 83 57
26 e-mail: odile.heddebaut@ifsttar.fr
27
28
29 Third author: Hubert Jayet
30 Affiliation: Professor of Economics
31 Director of EQUIPPE
32 University of Science and Technology of Lille
33 Address: Faculty of Economics and Social Sciences
34 59655 Villeneuve d’Ascq Cedex France
35 Phone: (+33) 3 20 43 65 98
36 e-mail: hubert.jayet@univ-lille1.fr
37
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 2

1 ABSTRACT
2
3 It is widely perceived that large scale transportation projects will have several positive impacts
4 both on the areas around its stations as well as at regional level. Thus, it is expected that a new
5 transport infrastructure will produce changes on the real estate market. Generally, these changes
6 are analyzed using cross-sectional hedonic models and once the infrastructure is operating. This
7 study measures the actual impact of urban transportation on the real estate market in an area
8 where a future large-scale transportation project will run. In this analysis, we evaluate the effects
9 of transportation while accounting for a number of structural, locational and neighborhood
10 variables making use of a location-detailed dataset over a period of time. To complement our
11 research, qualitative analysis is also employed by means of interviews with real estate agents.
12 We explore if there are significant land price changes due to the current public/private
13 transportation characteristics of the area and discuss what may have happened due to the so-
14 called “announcement or anticipation effect”. Finally, the results are better explained when the
15 socioeconomic characteristics of the study area are deeply examined, hence some more issues
16 arise. As a case study, we analyze the Bassin Minier of Nord-Pas-de-Calais region in France. It is
17 a deprived area which was the main pole of mining activities; mines stopped several decades
18 ago, and the area was recently converted into patrimony of humanity. The Transport Authority
19 that serves this area has planned to build two new tramway lines in the near future, that will
20 improve the accessibility in the area and which are not exempt of polemic.
21
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 3
1
2 INTRODUCTION
3
4 It is generally perceived that a new transport infrastructure will trigger different changes in urban
5 areas in many different ways and that these changes, are in general supposed to be positive, and
6 noticed, over land-use, jobs and business activities, among others. However, negative
7 externalities may also appear, most of them coming from noise, barrier and landscape effects or
8 even a rise in criminality (1). Nevertheless, as commented by Loo (2), little is known about the
9 perception and response of people about a future change in their living environment, for example
10 due to a new transport infrastructure. Sell et al. (3) mention that the availability, quality and
11 timing of information are important for the perception of a change. They also conclude that
12 temporal factors affect awareness: people are more concerned about a project when it is under
13 execution rather than when it is under planning. Moreover, the process of forecasting change is
14 directly influenced by the level of information and agents’ socio-economic characteristics. For
15 example, Loo (4) states that people’s subjective feelings and evaluation on the impact of an
16 infrastructure were significantly associated with their knowledge about its spatial location.
17 Existing literature also shows that personal characteristics are significant determinants of
18 people’s perceptions about their living environment (2).
19
20 There is an extensive literature accounting for the impacts of the operation of a new
21 transport infrastructure over the real estate market by means of hedonic models, which refer to
22 indivisible goods characterized by several attributes (5). In the case of land values, the idea is
23 that property prices are determined by the attributes characterizing the property, which may be
24 structural, linked to the neighborhood or to accessibility characteristics. The identification of the
25 price function allows measuring the implicit price of each attribute. It is important to notice that
26 a house unit is fixed in its location and at the same time, it is influenced by the characteristics of
27 the neighborhood, which are basically a set of geographically distributed data, which means that
28 there may be spatial autocorrelation (6). To assess positive and negative externalities produced
29 by a transport infrastructure, cross-sectional datasets are generally used. In comparison, there is a
30 lack of studies focused on the benefits produced by current or future transport infrastructure
31 using time-series datasets.
32
33 On the other hand, little is known about deprived territories and their regeneration
34 strategies to overcome this issue. Recently, three national agencies in the USA have come
35 together to document the policies aiming at helping communities to support neighborhoods in the
36 field of transportation, housing and economic development (7). In the case of France, most of the
37 research in the field of transport infrastructure deals with the case of large and widely known
38 cities like Paris, Strasbourg, Nantes or Lyon. However, herein we present as an empirical
39 analysis a completely dissimilar case, known as the “Bassin Minier” which is the former coal
40 mining area in the Nord Pas-de-Calais region, at the northern part of France, next to Belgium,
41 known as a socioeconomically deprived area. The aim of this paper is threefold. First to analyze
42 residential property values from 2005 to 2010 within the seven most important towns
43 (Communes) of our study area, namely Béthune, Bruay-la-Buissière, Carvin, Hénin-Beaumont,
44 Lens, Liévin, and Noeux-les-Mines in order to understand the territory and to assess whether or
45 not capitalization of transport supply has impacted during the last years in the real estate market.
46 Secondly, we discuss about the so called anticipation effect and analyze if there has been a rise
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 4

1 due to the announcement effect of two new tramway lines. Finally, we discuss the policy
2 implications of having insignificant transport related effects in house prices. The methodology is
3 based on a hedonic approach. In order to obtain more accurate results, we use the real network
4 distance to different neighborhood amenities, and we used OLS and spatio-temporal models. We
5 complemented our study by means of qualitative research. The qualitative analysis was
6 undertaken by means of semi-structured interviews with 9 Real Estate agents. Using interviews
7 as a means of eliciting information about the real estate market has not really been used to
8 complement and verify information of hedonic models. This approach may be useful to tackle
9 certain problems such as co-linearity, significance or the absence of other variables. To our
10 knowledge, this approach has been rarely used in the assessment of house prices.
11
12 Even though there is an extensive literature on house price assessment, our final intention
13 is to contribute to the analysis of urban regeneration and transport policy initiatives through the
14 analysis of the evolution of house prices by means of different precise econometric models and
15 to contrast the results with quantitative techniques.
16
17 This paper is divided into 5 sections. After the introduction, section 2 portrays the study
18 area. Then, section 3 presents the theoretical background. Afterwards, section 4 describes the
19 methodology and dataset of our study. Section 5 presents the results and finally, section 6
20 exposes the concluding remarks and the final discussion.
21
22 STUDY AREA
23
24 As said before, this study is carried out in the former coal mining area, Bassin Minier, which is a
25 deprived area of the Nord-Pas-de-Calais region (see Figure 1 grey left part). This area is well
26 served by different highways: the A1, A21, and A26. For example, the city of Lille is around 40
27 kilometers from this area. Within the area, public transport is managed by a public authority, the
28 SMT Artois-Gohelle (Syndicat Mixte des Transports Artois-Gohelle), that operates in 115
29 communes. The seven most important towns account for 30% of the population living in the
30 area. Next paragraphs describe the socioeconomic characteristics of this area and the different
31 policies aimed at improving their situation.
32
33 Socioeconomic characteristics
34 The former coal mining area is commonly known by its multiple socio-economic problems.
35 Different countries have also experienced the same regeneration problems regarding to the
36 closure of mining and industry activities, as documented by Waddington and Parry (8). In this
37 case the unemployment rate is still among the highest in France, nowadays 13.1% compared to
38 an average of 9.6%. It also has the highest rate of mortality with 8.9 for 1,000 in 2005 against 8.6
39 in France, along with the lowest life expectancy (82 years for women and 74 years for men
40 against respectively 84 and 77 years in metropolitan France). In addition, the region also counts
41 with the highest rate of obesity within the country (9).
42
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 5

1
2 Figure 1 – Lille Metropolitan Area (Source: SCOT Lens Liévin Hénin Carvin (10))
3
4
5 Different socioeconomic conditions of this area are shown in Table 1 which were
6 obtained from the French Institute of Statistics and Economic Studies (INSEE) (11). In the main
7 towns of the study area, population diminished at the end of the 80’s due to the extinction of the
8 coal mining activity and the lack of industrial jobs. However, a slight increase –particularly in
9 the towns near the urban conurbation of Lille– is noticed in the recent years.
10
11 Table 1 also shows the type of jobs in the region. Most of the jobs in this area are focused
12 on commerce and services. Moreover, the rate of unemployment in the area has not importantly
13 diminished in a decade, even though, according to the INSEE, more firms have localized in the
14 area over the last years. There is also a problem related to poorly academic performances and
15 failures, for example, there is a high rate of population older than fifteen years that quitted the
16 educational system with no certificate or diploma of mandatory studies.
17
18 It is a region where private transportation plays a vital role for daily activities. The non-
19 motorization rate is higher in this area than in the rest of the region or France, despite the higher
20 rate of households owning only one car. The former may be explained by the fact that they earn
21 less and have a higher rate of unemployment than in the rest of the Region or France.
22
23
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 6

1 Table 1 – Different characteristics of the region under study (2009) (11)


Bruay- Noeux- Nord-Pas-
Béthune Hénin-
la- Carvin Lens Liévin les- de-Calais France
Beaumont
Buissière Mines Region
Surface (km²) 9.5 16.4 21.0 20.7 11.7 12.8 8.8 12,414.1 632,734.9
Density (inhab/km²) 2,723.7 1,444.7 821.1 1,241.8 3,062.4 2,494.9 1,377.5 324.9 101.6
2011 (inhabitants) 26,623 24,073 17,396 26,164 36,540 32,461 12,289 4,038,280 65,001,181
2009 (inhabitants) 25,766 23,261 17,267 25,731 35,830 32,009 12,177 4,033,197 64,304,500
Variation population 1999
-0.8 -0.2 -03. +0.2 -0.1 -0.4 +0.2 +0.1 +0.7
and 2009 in (%)
Variation pop 2009 and 2011
+3.3 +3.4 +0.7 +1.6 +1.9 +1.4 +0.9 +0.1 +1.0
(%)
Men in (%) 46.9 46.9 48.0 47.8 46.4 47.6 47.4 48.4 48.2
Women in (%) 53.1 53.1 52.0 52.1 53.6 52.4 52.6 51.6 51.7
Men 0-14 years in (%) 18.7 21.3 20.1 21.4 20.2 22.2 21.3 19.6 21.4
Women 0-14 years (%) 15.9 18.4 18.3 18.3 18.4 19.5 16.0 17.5 18.9
Men 15- 59 years (%) 65.9 51.9 63.4 62.6 62.9 62.8 60.6 60.6 40.6
Women 15-59 years (%) 58.7 52.6 60.2 58.0 55.1 58.2 54.3 57.9 58.7
Men > 60 years (%) 15.5 18.5 16.4 16.1 16.8 15.1 18.2 20.0 16.7
Women > 60 years (%) 25.4 29.0 21.4 23.8 26.4 22.3 29.6 24.6 22.5
Households 11,809 10,258 6,705 10,500 15,563 12,768 5,161 1,631,166 27,533,813
Owned-lodgment 4,535 3,294 3,664 5,178 4,090 4,711 2,252 918,735 15,503,017
(%) (38.4) (32.1) (54.6) (49.3) (26.3) (36.9) (43.6) (56.3) (57.7)
Rented-lodgment 7,125 5,928 2,658 5,009 9,875 7,306 2,631 673,174 10,656,267
(%) (60.3) (57.8) (39.6) (47.7) (63.4) (57.2) (51.0) (41.3) (39.7)
3,056 3,347 1,698 2368 4,050 3,531 1,547 311,022 3,916,233
From which HLM (%)
(25.9) (32.6) (25.3) (22.6) (26.0) (27.7) (30) (19.1) (14.6)
At least parking space 6,162 5,417 4,225 6,223 8,592 8,000 3,712 991,906 17,240,042
(%) (52.2) (52.8) (63.0) (59.3) (55.2) (62.7) (71.9) (60.8) (64.2)
No car at all 3,549 3,118 1,296 2,595 4,585 3,370 2,333 352,508 5,828,685
(%) (29.3) (30.4) (19.4) (24.8) (29.4) (26.4) (25.9) (21.6) (19.7)
Only 1 car 6,277 5,037 3,493 5,075 8,089 6,271 2,464 769,145 12,619,020
(%) (53.2) (49.1) (52.1) (48.3) (52.0) (49.1) (47.7) (47.2) (47.0)
2 cars or more 2,068 2,103 1,916 2,830 2,889 3,127 1,364 509,513 9,086,108
(%) (17.5) (20.5) (28.6) (26.9) (18.6) (24.5) (26.4) (31.2) (33.8)
Net income med. (€) 18,569 15,011 17,817 17,306 15,249 14,938 15,844 20,157 23,230
Hh exempt of taxes (%) 42.4 33.9 41.6 40.9 36.4 34.5 36.7 46.7 53.6
Jobs 18,663 8,591 5,971 12,336 23,460 11,444 3,718 1,479,135 26,278,722
Independent/entrepreneurs(%) 6.6 7.9 7.8 5.5 5.2 6.7 8.9 8.9 12.2
Econ.activ. age 15-64 (%) 62.7 64.6 66.8 65.9 63.9 62.7 65.2 67.5 71.7
Unemployed, 15-64 (%) 18.7 19.0 15.2 17.5 24.0 22.2 17.5 15.0 11.7
No-diploma population (%) 23.6 22.9 23.7 22.6 26.2 27.9 22.5 20.5 18.9
No. firms 2,412 1,113 807 1,316 2,327 1,223 481 223,465 5,423,058
Agriculture (%) 0.8 0.8 2.0 1.7 0.0 0.6 1.2 7.8 11.9
Manufacturing (%) 5.2 4.4 6.2 5.1 4.0 5.7 6.0 5.8 5.8
Construction (%) 4.9 5.7 7.4 6.9 6.9 7.3 7.1 7.8 9.3
Transp.,services, comm.(%) 69.3 65.0 66.4 65.9 70.0 64.8 65.3 60.5 58.9
From which commerce 22.2 27.0 23.2 23.3 23.8 22.6 27.0 19.3 16.9
Admin.,education, health (%) 19.9 24.1 18.0 20.7 19.1 21.6 20.4 18.1 14.0
Firms 1-9 employees (%) 33.7 38.1 32.5 33.3 36.4 31.7 35.3 31.1 28.2
Firms 10+ employees (%) 10.4 11.0 13.9 11.9 12.1 15.8 12.9 9.7 6.9
TGV station Yes No No No Yes No No
TER station Yes No No Yes Yes Yes Yes
“Bulle” Bus lines Yes Yes No Yes Yes Yes No
Bus lines Yes Yes Yes Yes Yes Yes Yes
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 7

1 The housing market in this study area used to be specifically regulated, which proves the
2 peculiarities of this coalfield area. Those dwellings were built by the coal mining companies for
3 miners’ families. At the end of the mining exploitation (1990) these housing were controlled by a
4 subsidiary company of the “Houillères du Nord Pas de Calais” (the main coal mining company).
5 Later, it was bought by an industrial and commercial public establishment (EPIC) created by the
6 2000 solidarity and urban renewal law (SRU), known as Soginorpa. It belongs to the region Nord
7 Pas de Calais (12). In 2009, they counted with 62,651 dwellings, mainly individual houses with
8 garden which are the typical miners’ detached or semi-detached housing, known as “Corons”.
9 These houses are placed on the coal mining area from Valenciennes to westward Bruay-la-
10 Buissière and are mainly occupied by either old miners or their widows, under a scheme of really
11 moderated rents.
12
13 There is also a very large share of public housing, known as HLM (Habitation à Loyer
14 Modéré in French), which are managed by public bodies or by private sector companies
15 supported by public loans. This type of housing is occupied by people in deprived social
16 conditions. Table 1 shows the rate of HLM for each commune, the Nord Pas de Calais region
17 and the country. It is interesting to note the rate of HLM within these communes may get to 32%
18 while in the rest of France is less than 15%.
19
20 Urban regeneration
21 As we said before, this region suffers from multiple disadvantages that are definitely related: A
22 very high unemployment rate, low wages, poor skills, health risks, etc. These disadvantages suit
23 perfectly with the definition of social exclusion made by Levitas (13). As highlighted by Lucas
24 (14), transport and land-use policies play an important role in social exclusion. Research has
25 proved as well that there is a strong relation among the dependence on automobile, the built
26 environment and obesity (15). Even though many urban regeneration policies have been
27 implemented over the last decades to palliate the situation of the Nord-Pas-de-Calais region, well
28 documented by Couch et al. (16), there is still a huge gap between poor and wealthy areas. More
29 recently new strategies have been planned aiming at the regeneration of the socioeconomic
30 activities in this area, such as the new Museum of Louvre in Lens, the inclusion of the Bassin
31 Minier as Unesco patrimony of humanity, and the improvement of public transportation.
32
33 In 2003, to celebrate the century of the Louvre museum, it was decided to let this
34 museum going “out of its walls” with the creation of a new site. Among six towns from the Nord
35 Pas de Calais region, Lens was chosen to host this new museum (17). The characteristics of this
36 town were ideal for the new museum, a strong symbolic connotation, and a region that deserved
37 to be recognized by its suffering in the past. The Louvre Lens is located in a former mine shaft
38 with 28,000 m² exhibition surface. It presents continuous temporary exhibitions of the Louvre
39 collection and also hosts temporary exhibitions. It opened to public in December 2012.
40
41 Over the last ten years people living in the area of the Nord-Pas de Calais coalfields have
42 felt the necessity of an official recognition of this area. In 2010, the French government officially
43 proposed the candidature of the Bassin Minier to the Unesco as an area patrimony of humanity
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 8

1 and it was recently accepted due to, among other issues, “a Continuing, Organically Evolved
2 Cultural Landscape” and “a combined work of man and nature” (19; 18).
3
4 Public transport
5 With regards to public transportation, and as shown in Table 1, the communes are served by
6 different public transport systems:
7 • “Bulle” lines: rapid buses that run every 15-20 minutes
8 • Regular buses: with different frequencies depending on time and route
9 • “Allobus” lines: Bus service on-demand. Users need to book in advance to make
10 the bus stop in a designated place
11 • TER station (“Transport Express Régional”): commuter regional trains which
12 connect the different towns
13 • TGV (Train à Grande Vitesse): High Speed Rail that connects the commune to
14 other cities like Paris, Lille
15
16 The Transport Authority (SMT Artois-Gohelle) planned to build two new tramway lines,
17 which are shown in Figure 2. They will serve different areas; hence they are not connected to
18 each other.
19
20 The project was announced at the end of 2008. At the time this research took place, the
21 final route of the tramway could still suffer slight changes and there were still some discussions
22 about its feasibility (21), however, in general terms, it would follow the routes of the Bulle lines.
23 Different media publications show that people living or working in the area are afraid of the
24 negative effects of these new lines. For example, Caratini (22) reported that the owners of retail
25 stores believed that the new tramway will allow their customers to go away more easily to other
26 areas for shopping; therefore in towns like Bruay-la-Bruissière, the center will be even less full
27 of people than nowadays. “Anti-tram” meetings were also taking place.
28
29 THEORETICAL BACKGROUND

30 Quantitative analysis
31 Hedonic regression analyses are very popular for land market studies. Literature using this
32 technique is extremely extensive, especially when using cross-sectional datasets and once the
33 infrastructure is operating. Nevertheless, literature regarding the announcement effect is more
34 limited and proportionally, not too many studies exist. The same happens when using time-series
35 datasets. The next paragraphs briefly mention the approaches used by different authors regarding
36 anticipation and effects over the long term.
37
38 McDonald and Osuji (23) analyzed the impact in Chicago’s Midway Line on residential
39 land values in 1990, three years before the opening, using a generalized before-and-after method,
40 with estimations for 1980 and 1990. They found an increase of 17% within one-half mile of the
41 stations.
42
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 9

2 Figure 2 – Tramway project: At the top from Liévin to Hénin Beaumont and at the bottom from
3 Béthune to Bruay-la-Buissière (Source: SMT Artois Gohelle (20))
4
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 10

1 Later, Henneberry (24) used the hedonic analysis to examine the impact of the tramway
2 line on house prices. He found that the infrastructure initially depressed the prices of nearby
3 houses which he relates to the announcement of the project and the nuisance of civil works.
4 However he mentions that a long term impact could be carried out to conclude about the actual
5 effect of it over house prices.
6
7 Next, McMillen and McDonald (25), by using hedonic and repeated sales methods, found
8 that in the Chicago’s midway line there was an anticipation effect in 1987, six years before the
9 opening. The infrastructure had decades under discussion, but finally, in middle 1984 the final
10 alignment decision was made and the first part of the budget was received.
11
12 Yiu and Wong (26) analyzed the effect of expected transport improvements on housing
13 prices for a tunnel by price index construction and price gradient analysis. They suggest that the
14 timing of the effects depends on the expected benefits of the transport improvements; the costs of
15 bearing the nuisance of construction and the information cost of improvements. Although they
16 did not use a hedonic regression, through different methods they exploited a times-series dataset,
17 and found an increase in prices in different sections although it was not necessarily at the
18 commencement of the works.
19
20 Tsutsumi and Seya (27) also studied the dynamical changes and announcement effect of a
21 commuter railway in Japan using interpolated land price maps. They used a data collection of
22 house prices that included the periods of zero-state, formulation of the project, construction and
23 one year after operations started. They found an increase in land prices, which they called the
24 “announcement effect” although in fact, it happens at the time construction of the line started.
25 Afterwards, they compared the spatial econometrics and geostatistical approach using cross
26 sectional data.
27
28 More recently Boucq and Papon (28) analyzed the accessibility gains or residential
29 dwelling in the Hauts-de-Seine department of France between 1996-2003 using ordinary least
30 squares methods. In their analysis they found no anticipation effects, which could be related to
31 different nuisances produced by civil works.
32
33 In general spatial analysis is used for cross-sectional data. Authors, such as Anselin et al.
34 (29), mention that space-time models are not commonly used due to its complexity. Pace et al.
35 (30), (31) have tackled the issues of spatial and temporal dependence with spatio-temporal
36 models. They have documented the substantial benefits of this approach for the case of Fairfax
37 County Virginia between 1969-1991; and Baton Rouge between 1984-1992, respectively. On the
38 other hand, Sun, Tu and Yu (32) make use of Bayesian estimation method to produce more
39 efficient estimates than the conventional OLS estimation and in order to deal with the spatio-
40 temporal autocorrelations and the heteroscedasticity problem for the case of the multi-unit
41 residential market of Singapur from 1990 to 1999. They also found that the incorporation of
42 spatial information improved the analysis.
43
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 11

1 Qualitative analysis
2 Nowadays, and as mentioned by different authors (33; 34) we are moving from researches based
3 solely on quantitative work to a dialogue on both the importance of quantitative and qualitative
4 methods. Researchers find qualitative analysis an important tool that complements quantitative
5 analysis. As stated by Delyser (35), quantitative research may answer important questions that
6 may not be addressed by a qualitative approach, but so does qualitative studies when bringing
7 about different ideas not accessible quantitatively.
8
9 To complement and contrast the quantitative study, personal interviews were employed to
10 elicit information from real-estate agents. This technique was the optimal solution for people
11 with a very complicated schedule and with high mobility around different towns. It would have
12 been almost impossible to plan a focus group. As mentioned by Clifton and Handy (36), personal
13 interviews may provide the same rich information as other qualitative analysis. They cite
14 interesting examples of this technique related to transportation issues. The main advantage of
15 qualitative methods is that they show the causality of certain facts.
16
17 METHODOLOGY & DATA
18
19 Since our sample does not cover the same lodgments over a period of time, we are not able to use
20 a repeated-sales model and therefore we have preferred to use both a typical OLS and spatio-
21 temporal model based on Pace (31) analysis.
22
23 The dataset, from 2005 to 2010, was obtained through the Regional Council of Nord-Pas
24 de Calais. It has all the selling prices registered at notary offices for the whole region and for all
25 types of buildings and land plots. Unfortunately the dataset used does not provide more
26 information about the characteristics of the lodgment. In this case, we analyzed the seven most
27 important towns in the Bassin-Minier. In fact, there were more houses than apartments in the
28 database, which is coherent since it is an area where many semi-detached and detached houses
29 were built for miners and this area accounts for few apartment buildings. We have developed
30 many different models, but for space limitations we will just discuss the best models found. For
31 example, after different trials we decided to concentrate only in houses, particularly because they
32 are more representative in the territory and in some towns the data on apartments was extremely
33 limited to be used.
34
35 As noticed by different authors, many times hedonic models exclude important location
36 variables (for example, distance to relevant points such as CBD) which has proven to increase
37 the probability of spatial error autocorrelation or heterogeneity (37). In order to minimize those
38 problems, we tested several location variables, of which only the significant ones or those that
39 not presented severe correlation were included in the final models and are described in Table 2,
40 the rest are not presented herein. A Geographic Information System was used to calculate real
41 network distances through the street network to eliminate the overestimation with respect to an
42 Euclidean approach (38).
43
44 As we explained before in the theoretical background section, time-series datasets present
45 both time and space autocorrelation. They are limitedly treated in literature because of their
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 12

1 complexity. We decided to tackle this issue with the Pace and colleagues approach, due to its
2 good results to obtain more efficient estimates. To carry out such analysis we make use of the
3 Spatial Statistics Toolbox developed by Pace et al. (39) to evaluate the STAR (Spatiotemporal
4 Autoregressive) model.
5
6 The models presented herein are based on the following equations:
Y = xit β + ε it (1)

7 Which is the typical hedonic model. Whereas the generalized STAR model proposed by
8 Pace (31) is:
Y = Zθ + Xβ1 + TXβ 2 + SXβ 3 + STXβ 4 + TSXβ 5 + φT TY + φ S SY + φST STY + φTS TSY + ε (2)

9 Z n x k matrix of observations which is not associated to spatial, temporal or spatiotemporal


10 lags
11 θ Associated k1 x 1 vector of parameters
12 X n x k2 matrix of observations of independent variables with spatial, temporal or spatial-
13 temporal lags
14 βi k2 x 1 parameters associated with spatial, temporal and spatial-temporal lagged variables
15 T Temporal weight matrix (standardized). It computes a number of 20 previous nearest
16 neighbors for each observation.
17 S Spatial weight matrix (standardized). It computes the 10 nearest neighbors
18 ϕ Autorregresive parameters in space (S) or time (T)
19
20 The spatial and temporal weight matrixes are standardizes and many trials are needed to
21 get the optimal values (37). Apart from the direct accessibility measures (such as the closest
22 amenity through the street network), we tested some variables which gave an overall measure of
23 accessibility, gravity based variables, which have the following form:
− β cij
Ai = ∑ e (1)

24 To get the β coefficient, we used the only household survey available from years
25 2005/2006 from the two areas of study. Values were similar, we tested both values but at the end
26 an intermediate value between both surveys was used (-0.303). Unfortunately, it was not possible
27 to calculate the employment gravity variable based on the number of employments in a certain
28 area because the information was not complete.
29
30 The variable Z900 looks for houses located in the real service area of 0-900 m the area of
31 impact of the future tramway, because slight changes regarding the final route may take place,
32 route which is by now covered by the Bulle-buses.
33
34 Due to the limited amount of house sales in 2009 and 2010 and the effect of
35 announcement, we had to gather those years together.
36
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 13

1 As it was mentioned before, the analysis of house prices was complemented by means of
2 a qualitative approach, using thematic analysis by means of semi-structured interviews. Semi-
3 structured interviews were used because they are the middle ground of interviews, and are
4 helpful to provide detail, depth and perspective information while allowing testing different
5 hypothesis which is also in the line with Leech (40). This approach was helpful to understand the
6 particularities of the region, to shed some light in some variables that were necessary to include
7 in the models, and to understand results.
8
9 Interviews were carried out during 2012. They took place at the interviewees’ office and
10 lasted around 45 minutes. To get in contact with them, a web research was carried out to obtain
11 the location of all the real-estate agencies for housing in Lens (because is the largest towns of the
12 area of study), which were then visited, and all of them accepted to be interviewed. The town
13 counts with 9 agencies that focus on housing, not only for the town of Lens but also for the
14 surroundings. The interview covered different topics. Each participant went deeper into any of
15 these topics as they felt like. The main questions covered were related to:
16
17 • Contact information
18 • Expertise
19 • The current situation of the market
20 • The Louvre museum factor
21 • The tramway factor
22 • Point of view of the project
23 • Clients’ requests
24 • Clients’ mobility profile: number of cars, where do they commute, how, etc.
25 • The proximity to public transportation: train stations, buses, “Bulle-lines”
26 • How important is for their clients parking availability
27 • Rate of people renting /buying a home in the area
28 • The importance of local downtown or other towns around the area
29 • The importance of the distance to the city of Lille
30 • Reasons for people wanting to live here
31 • Information of land availability in the area for building up new housing
32 • Other relevant information
33
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 14

1 Table 2 – Variables’ description & descriptive statistics

Variable Obs Mean Std. Dev. Min Max Description


Houseprice 3971 11.6555 0.4388 10.3138 13.0637 Official registered sale price, natural logarithm (LN)
Surface 3971 4.4497 0.3334 2.7081 5.7137 House real surface area, LN
House's total registered Area(i.e.with garden or
Totalsurf 3971 5.6089 0.7856 0.6931 9.8657 parking place); LN
Number of rooms (if the house has more than 7 rooms
Roomqual 3971 1.4058 0.2928 0 1.9459 it gets equal to 7), LN
Bus 3971 -1.7311 0.9241 -7.9576 0.5061 distance in km to the closest bus stop, LN
Rail 3971 0.8536 0.7364 -3.9776 2.0119 distance in km to the closest rail station, LN
Parking 3971 -0.3562 0.8721 -6.5929 1.6884 distance in km to the closest parking lot, LN
S_pool 3971 0.6604 0.8406 -4.0157 2.4264 distance in km to the closest swimming pool, LN
Medical 3971 0.0909 0.6822 -3.6208 1.6170 distance in km to the closest medical center, LN
Recreation 3971 -0.6136 0.7497 -6.0076 1.1624 distance in km to the closest recreational space, LN
R_access 3971 0.3524 0.7037 -3.6636 1.6801 distance in km to the closest access to the roadway, LN
L_downtown 3971 0.1198 0.6658 -7.8494 1.4074 distance in km to the closest local downtown, LN
indicates the gravity model for zones of economic
G_employ 3971 5.2119 0.2430 4.3476 5.6087 activity, LN
G_shops 3971 0.4018 0.8005 -1.2745 1.6301 indicates the gravity model for shopping areas, LN
G_edu 3971 3.7095 0.2701 2.8513 4.2155 indicates the gravity model for education facilities, LN
indicates if the property sold is within a 900 m service
Z900 3971 0.4193 0.4935 0 1 area of the future line
Communes Dummy variables, name of the town
Bethune 3971 0.1685 0.3743 0 1 Commune of reference
Bruay 3971 0.1204 0.3254 0 1 Bruay-la-Buissiere
Carvin 3971 0.1393 0.3463 0 1
HB 3971 0.2035 0.4026 0 1 Henin-Beaumont
Lens 3971 0.1219 0.3272 0 1
Lievin 3971 0.1717 0.3772 0 1
Noux 3971 0.0748 0.2631 0 1 Noeux-les-Mines
Year
indicates transactions occurred on the year 2005 (year
2006 3971 0.3118 0.4633 0 1 of reference)
2007 3971 0.2896 0.4536 0 1 indicates transactions occurred on the year 2007
2008 3971 0.0592 0.2360 0 1 indicates transactions occurred on the year 2008
2009 3971 0.0232 0.1505 0 1 indicates transactions occurred on the year 2009
2010 3971 0.0076 0.0866 0 1 indicates transactions occurred on the year 2010
indicates transactions occurred on the year 2009 or
2009/10 3971 0.0307 0.1726 0 1 2010
2
3
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 15

1 RESULTS
2
3 Tables 3 and 4 present the main results. We tried many different regressions, such as including
4 and separating the communes in the model, taking into account apartments and houses,
5 producing other variables for accessibility, and so on, but results were not conclusive. Hence, we
6 only present the best results we obtained by two models for housing using a classical and a
7 spatiotemporal approach. The spatial correlation of both analyses is evaluated with the Moran’s I
8 test, which results are presented at the end of both tables. Certainly, the spatio-temporal analysis
9 reduces the problems associated with the temporality and spatiality of the dataset as shown by
10 the Moran’s I results, which implies that this model provides more efficient estimates.
11
12 As table 3 and 4 show, the price is principally explained by the structural characteristics
13 of the lodgment and by the year when it was sold. One of the interviewees mentioned: “people
14 are always looking for a three bedroom house unit with garden and garage. That’s the typical
15 case”. Unfortunately our data set has no clear information regarding garden or garage, but only
16 indirectly through the variable Totalsurf. The number of rooms is also a positive asset.
17
18 Access to buses, to train stations, and/or parking lots is not always significant or positive
19 for real-estate prices (negative sign). The same happens with swimming pool. In general terms,
20 house prices increase in areas close to the local downtown, medical and recreational facilities,
21 along with education facilities. Although the impact of being close to the 900 m service area
22 (variable Z900) is positive, in table 4 is not significant.
23
24 The variable “Year” shows that the real-estate market is decaying, which is in line with
25 the information elicited from interviews. All the interviewees agreed that the market was
26 stagnating or even in decay. Citing one of them: “Nowadays, there is more offer than demand of
27 lodgments”. Most of them also noticed the importance of rent and social-rent in the area which,
28 according to the interviewees, would be around 50%. Table 1 also shows that social housing
29 (HLM) in this area is between 25-33% much higher than the rest of France. It also means that the
30 anticipation effect was not noticed since 2009.
31
32 Accessibility variables have a positive effect for poles of employment in both models
33 (being closer to more employment opportunities), whereas the other two (education and
34 shopping areas) are not that evident in both models. Shopping seems to be only relevant in the
35 second model, although not as important as employment. Moreover, people seems not to
36 positively valuate being close to educational centers (primary and secondary).
37
38 It can also be noticed that the town variables (Commune) do have an effect on prices.
39 Places like Lens and Lievin are more valuable than others like Noux-les-Mines (when compared
40 to Bethune). Which was actually our original idea when carrying out the interviews, if we were
41 to find a positive anticipation effect, it should be notice in places like Lens.
42
43
44
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 16

1 Table 3 – Hedonic models by commune with time dummy variables

Houseprice Coef. Std. Err. t


Surface 0.511 0.021 0.000
Totalsurf 0.157 0.007 0.000
Roomqual 0.123 0.023 0.000
Bus 0.015 0.005 0.006
Rail 0.051 0.012 0.000
Parking 0.005 0.009 0.545
S_pool 0.029 0.012 0.013
Medical -0.009 0.009 0.306
Recreation 0.000 0.008 0.973
R_access 0.027 0.011 0.016
L_downtown -0.036 0.012 0.004
G_employ 0.365 0.075 0.000
G_shops -0.037 0.038 0.337
G_edu -0.307 0.062 0.000
Z900 0.056 0.013 0.000
Years
2006 0.099 0.011 0.000
2007 0.188 0.011 0.000
2008 0.228 0.020 0.000
2009/10 0.199 0.044 0.000
Commune
Bruay -0.013 0.037 0.734
Carvin 0.132 0.057 0.020
HB 0.127 0.070 0.068
Lens 0.271 0.071 0.000
Lievin 0.282 0.089 0.002
Noux -0.072 0.026 0.006
Constant 7.308 0.399 0.000
Number of obs 3523
F( 25, 3497) 165.78 Moran's I: 0.05
Prob > F 0 istat: 5.02
R-squared 0.5424 imean: 0.00
Adj R-squared 0.5391 i.var: 0.00
Root MSE (Mean
Square Error) 0.2591 prob: 0.00
2
3
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 17

1 Table 4 – Spatial-Time model


Houseprice Space-time
Variables Coef. Space-time SRD P>t
Temporal parameters: (X-TX)
Surface 0.5086 49.6639 0.0000
Totalsurf 0.1583 46.9446 0.0000
Roomqual 0.1063 11.5009 0.0000
Bus 0.0109 5.037 0.0000
Rail 0.0581 12.4423 0.0000
Parking 0.0081 2.2994 0.0215
S_pool 0.0355 7.7804 0.0000
Medical -0.0208 -6.0701 0.0000
Recreation -0.0137 -4.3794 0.0000
R_access 0.0053 1.181 0.2376
L_downtown -0.0538 -10.9196 0.0000
G_employ 0.2453 9.2544 0.0000
G_shops 0.0535 3.9522 0.0001
G_edu -0.2039 -8.6251 0.0000
Z900 0.0057 1.1077 0.2680
Bruay -0.0534 -3.956 0.0001
Carvin -0.0099 -0.4943 0.6211
HB 0.0127 0.5306 0.5957
Lens 0.0695 2.8331 0.0046
Lievin 0.0604 1.9955 0.0460
Noux -0.0507 -5.0048 0.0000
Spatial parameters: S(X-TX)
Surface 0.0496 1.8926 0.0584
TotalSurf 0.0017 0.1828 0.8550
Roomqual -0.0138 -0.5874 0.5569
Bus -0.0122 -2.1655 0.0304
Rail 0.0074 0.6 0.5485
Parking -0.0133 -1.4383 0.1503
S_pool 0.0182 1.5951 0.1107
Medical -0.0012 -0.1375 0.8907
Recreation -0.0007 -0.091 0.9275
R_access 0.002 0.1789 0.8580
L_downtown -0.0094 -0.7762 0.4377
G_employ -0.0338 -0.6877 0.4916
G_shops 0.0107 0.4266 0.6696
G_edu 0.0156 0.3005 0.7638
Z900 -0.0232 -1.8777 0.0604
Bruay 0.009 0.3173 0.7510
Carvin -0.0401 -1.0576 0.2902
HB -0.0126 -0.3044 0.7608
Lens -0.0057 -0.1333 0.8940
Lievin -0.0009 -0.0177 0.9859
Noux -0.0208 -0.8369 0.4027

(continue in next page)


Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 18

SpatioTemporal parameters
S Houseprice -0.0281 -0.89 0.3734
ST Houseprice 0.0163 0.4808 0.6307
TS Houseprice -0.323 -18.562 0.0000
Intercept 3.9167 35.3973 0.0000

mt 20 Moran's I -0.01
rho 0.75 istat -0.76
n 3487 imean 0.00
k 46 ivar 0.00
Loglikelihood -6355.98 prob 0.45
R-squared 0.87
SSE (sum of Squared Error) 38.3
Likelihood ratio test* 6299.47 *Contrast model not reported
1
2 There are different reasons that might explain the lack of importance of public
3 transportation. See for example, the number of households with one car in this area (see table 1).
4 In this respect, all the interviewees agreed with the profile of clients regarding car-ownership is a
5 couple with at least one car. Moreover, one of the interviewees commented that she had noticed
6 that even if the lodgments she sold had no parking space, it was not really a problem because her
7 clients usually live in the early morning to work out of the commune and they come back in the
8 evening, where there are no parking restrictions and there is plenty of space to park on the
9 streets. Therefore, it is not really a problem if the house/apartment has or not a garage.
10
11 Many agents mentioned that being close to the train station might be more interesting
12 than other public transport service: “Buses are not really an option; it takes more than the double
13 amount of time to get to a place by bus than by car, that’s why buses are not crowded”. Another
14 agent mentioned: “of course Lens and Béthune have the TGV (High Speed Rail-HSR) which is
15 an asset, although it is not an option for commuting to Paris for example, as it would be very
16 expensive!” Another one also mentioned that it is interesting to locate next to the train station,
17 but because it is in the middle of the town, hence it is next to shops and everything is close.
18 “Elderly people are specially looking for small apartments or houses in downtown, where shops
19 and offices are reachable by foot”.
20
21 In fact, all the participants in the interviews stated that they had never discussed the
22 project of the tramway with their clients, nor have they noticed a change due to the future
23 tramway line. On the contrary, one of the interviewees pointed out that she has two clients that
24 have raised the price of their property because it is not far from the new Louvre-Lens expecting
25 that the proximity to the museum would let them charge a higher price. However, time has
26 passed and no one is interested in their properties. Other participants hesitated when being asked
27 if they thought that the Louvre-Lens would have an impact on property prices. But around 40%
28 said that there were some expectations around this area. Two clients mentioned that a couple of
29 years ago they had civil works for the sidewalks of the city of Lens and it was very annoying:
30 “Maybe this souvenir explains why we are not really looking forward for the project”.
31
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 19

1 CONCLUDING REMARKS AND DISCUSSION


2
3 This region has not achieved to lessen their major disadvantages produced by the reconversion of
4 industrial and coal mining activity. Hence the Bassin Minier urges policy measures to overcome
5 these problems. An efficient public transportation along with other policy measures should help
6 to tackle these issues.
7
8 This paper shows that further analysis, using other approaches, should be carried out in
9 this territory. With regards to our analysis, public housing, by means of social rent is an
10 important issue in this area. For this reason, even if we have the complete data base of property
11 sales, we count with a relative limited number of transactions for sale and registered at notary
12 offices in the study area. Moreover, our database lacks of relevant information such as most of
13 the house and building’s amenities (bathrooms, elevators, etc.) which would definitely improve
14 the models.
15
16 We tested many different accessibility and locational variables. Unfortunately, there are
17 differences among or within communes that cannot be captured by the dummy or categorical
18 variables we introduced. As noticed by different authors (42; 41) capitalization effects may even
19 vary among small scale sub-groups or property types. We also used different OLS and different
20 spatio-temporal models, in order to get more accurate results. We found no major impact of
21 transportation. We considered carrying out other spatial analyses, however at this point the
22 results provide no hint that there is a link with public transportation.
23
24 In general terms, the communes most accessible by public transport are Lens and
25 Béthune. They both benefit from four different types of resources: a bus system, “Bulle-lines”,
26 regional train stations, and TGV (high speed train). However in our analyses the most important
27 communes to live in are Liévin and Lens. In other words, transportation services like high speed
28 trains do not always match the needs of local population. In fact there might be spatial and
29 transportation mismatch in the area, jobs are far and the transportation infrastructure is not the
30 adequate to solve commuting needs, especially for more disadvantaged groups. The former needs
31 further analysis.
32
33 The results show that other methods should be employed to elicit more information
34 regarding accessibility. For example, the importance of private transport is not reflected in the
35 models of this study; they only explain the low impact of public transportation supply. In fact,
36 the usage rate of cars in the Lens area is 63% and 71% in the Bethune area whereas the overall
37 public transport use is only 2% (Household transport survey carried out in 2006 for the Lens area
38 and 2005 for the Béthune area).
39
40 As pointed out by literature, perception of change is not positive at the beginning. The
41 former is also in the line with the interviews carried out. Yet, this paper shows that there has not
42 been a positive announcement effect for the tramway project in the area of study, and the impact
43 of the variable regarding service area is very small while at the same time not conclusive. It
44 seems that in general terms the anticipation effect can be noticed once the infrastructure is
45 operating or at least civil works are being carried out. An example of the former is the difference
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 20

1 with the Louvre-Lens museum which recently opened. In this case, people were sure that the
2 opening was going to happen in the near future.
3
4 If no positive anticipation effects are identified by any other approach or study, and
5 giving the particularities of this region, there is an important risk that the tramway project will
6 not achieve its expectations. Hence, strong policy measures, such as toll parking and other
7 restrictions, public transport special fares, public transport transit priority, cycling lanes, better
8 information campaigns, etc., should accompany the operation of this infrastructure.
9
10 Finally, it seems clear that in case that the tramway project comes to reality, monitor long
11 term effects of non-transport benefits should be associated with the project, especially in the case
12 of this peculiar region. It could even be plausible to interview the same people in the real estate
13 sector over a certain period of time.
14
15 AKNOWLEDGMENTS
16
17 This paper is part of the SUIM (Services urbains innovants pour une mobilité durable) project.
18 Authors would like to thank the ADEME agency and Regional Council of Nord-Pas-Calais for
19 their joint financial support to carry out this research and the provision of the dataset.
20
21 REFERENCES
22
23 1. Cohen, J. P., and C. C. Coughlin. Spatial hedonic models of airport noise, proximity, and
24 housing prices. Journal of Regional Science, Vol. 48, No. 5, 2008, pp. 859–878.
25 2. Loo, B. P. Y. The potential impacts of strategic highways on new town development: a case
26 study of Route 3 in Hong Kong. Transportation Research Part A: Policy and Practice,
27 Vol. 36, No. 1, 2002, pp. 41–63.
28 3. Sell, J. L., E. H. Zube, and C. L. Kennedy. Perception of land use change in a desert city.
29 Journal of Architectural and Planning Research, 1988.
30 4. Loo, B. P. Y. How would people respond to a new railway extension? The value of
31 questionnaire surveys. Habitat International, Vol. 33, No. 1, 2009, pp. 1–9.
32 5. Rosen, S. Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition.
33 Journal of political economy, Vol. 82, No. 1, 1974, p. 34.
34 6. Anselin, L. Spatial econometrics: methods and models. Springer, 1988.
35 7. Partnership for Sustainable Communities.
36 http://www.sustainablecommunities.gov/studies.html. Accessed July 22, 2013.
37 8. Waddington, D., and D. Parry. Managing industrial decline: the lessons of a decade of
38 research on industrial contraction and regeneration in Britain and other EU coal producing
39 countries. Mining Technology: Transactions of the Institute of Mining and Metallurgy,
40 Section A, Vol. 112, No. 1, 2003, pp. 47–56.
41 9. Eschwège, E., M. A. Charles, and A. Basdevant, eds. ObÉpi: Enquête épidémiologique
42 nationale sur le surpoids et l'obésité. Une enquête INSERM / TNS HELATHCARE
43 (KANTAREHEALTH) / ROCHE. Roche, France, 2009.
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 21

1 10. SCOT Lens-Liévin et Hénin-Carvin. http://www.scot-llhc.fr/.


2 11. INSEE. Institut national de la statistique et des études économiques:. http://insee.fr/fr/.
3 Accessed 07.19.2012.
4 12. Froger, V., J. F. Ghékière, and V. Houillon. Vieillissement, changement social et
5 paupérisation: Le parc de logements face au renouvellement de la population du bassin
6 minier du Nord-Pas-de Calais (France). Espace populations sociétés, 2010, pp. 17 july 2012.
7 13. Levitas, R., C. Pantazis, E. Fahmy, D. Gordon, E. Lloyd, and D. Patsios. The multi-
8 dimensional analysis of social exclusion. Citeseer, 2007.
9 14. Lucas, K. Providing transport for social inclusion within a framework for environmental
10 justice in the UK. Transportation Research Part A: Policy and Practice, Vol. 40, No. 10,
11 2006, pp. 801–809.
12 15. Lopez-Zetina, J., H. Lee, and R. Friis. The link between obesity and the built environment.
13 Evidence from an ecological analysis of obesity and vehicle miles of travel in California.
14 Health & place, Vol. 12, No. 4, 2006, pp. 656–664.
15 16. Couch, C., O. Sykes, and W. Börstinghaus. Thirty years of urban regeneration in Britain,
16 Germany and France: The importance of context and path dependency. Thirty years of urban
17 regeneration in Britain, Germany and France: The importance of context and path
18 dependency, Vol. 75, No. 1, 2011, pp. 1–52.
19 17. Bodéré, A. Louvre-Lens: le Louvre autrement. CeROArt, 2010.
20 18. Mission Bassin Minier UNESCO. Mission Bassin Minier UNESCO. http://www.atlas-
21 patrimoines-bassin-minier.org/EN/UNESCO-17.html.
22 19. Le Monde. fr. L'Unesco distingue les terrils et les cités des bassins miniers du Nord.
23 www.lemonde.fr, Vol. 2012, 07/16, 2012.
24 20. SMT Artois-Gohelle. Le projet du tramway. http://www.smt-artois-
25 gohelle.fr/racine/accueil/montram/le_projet.
26 21. Cerdeiro, C. Le projet continue d'avancer. L'avenir de l'Artois, 2012, p. 17.
27 22. Caratini, D. Pourquoi le tramway va dérailler. L'avenir de l'Artois, Vol. 2012, 02/10, 2012.
28 23. McDonald, J. F., and C. I. Osuji. The effect of anticipated transportation improvement on
29 residential land values. Regional science and urban economics, Vol. 25, No. 3, 1995,
30 pp. 261–278.
31 24. Henneberry, J. Transport investment and house prices. Journal of Property Valuation and
32 Investment, Vol. 16, No. 2, 1998, pp. 144–158.
33 25. McMillen, D. P., and J. McDonald. Reaction of house prices to a new rapid transit line:
34 Chicago's midway line, 1983–1999. Real Estate Economics, Vol. 32, No. 3, 2004, pp. 463–
35 486.
36 26. Yiu, C. Y., and S. K. Wong. The effects of expected transport improvements on housing
37 prices. Urban Studies, Vol. 42, No. 1, 2005, p. 113.
38 27. Tsutsumi, M., and H. Seya. Measuring the impact of large-scale transportation projects on
39 land price using spatial statistical models*. Papers in Regional Science, Vol. 87, No. 3,
40 2008, pp. 385–401.
Mejia-Dorantes, L., Heddebaut, O., Jayet, H. 22

1 28. Boucq, E., and F. Papon. Assessment of the real estate benefits due to accessibility gains
2 brought by a transport project: the impacts of a light rail infrastructure improvement in the
3 Hauts-de-Seine department. European Transport, Vol. 40, 2008, pp. 51–68.
4 29. Anselin, L., J. L. Gallo, and H. Jayet. Spatial panel econometrics. The econometrics of panel
5 data, 2008, pp. 625–660.
6 30. Pace, R. K., R. Barry, J. Clapp, and M. Rodriquez. Spatiotemporal Autoregressive Models of
7 Neighborhood Effects. The Journal of Real Estate Finance and Economics, Vol. 17, No. 1,
8 1998, pp. 15-33.
9 31. Pace, R., R. Barry, O. W. Gilley, and C. Sirmans. A method for spatial–temporal forecasting
10 with an application to real estate prices. International Journal of Forecasting, Vol. 16, No. 2,
11 2000, pp. 229–246.
12 32. Sun, H., Y. Tu, and S.-M. Yu. A Spatio-Temporal Autoregressive Model for Multi-Unit
13 Residential Market Analysis*. The Journal of Real Estate Finance and Economics, Vol. 31,
14 No. 2, 2005, pp. 155-187.
15 33. Crang, M. Qualitative methods: the new orthodoxy? Progress in human geography, Vol. 26,
16 No. 5, 2002, pp. 647–655.
17 34. Zolnik, E. J. Context in human geography: a multilevel approach to study human–
18 environment interactions. The Professional Geographer, Vol. 61, No. 3, 2009, pp. 336–349.
19 35. Delyser, D. Teaching qualitative research. Journal of Geography in Higher Education,
20 Vol. 32, No. 2, 2008, pp. 233–244.
21 36. Clifton, K., and S. Handy. Qualitative methods in travel behaviour research. Transport
22 survey quality and innovation, 2003, pp. 283–302.
23 37. Mejia-Dorantes, L., A. Paez, and J. M. Vassallo. Analysis of House Prices to Assess
24 Economic Impacts of New Public Transport Infrastructure. In Transportation Research
25 Record: Journal of the Transportation Research Board, No. -1, 2011, pp. 131–139.
26 38. Gutiérrez, J., and J. C. García-Palomares. Distance-measure impacts on the calculation of
27 transport service areas using GIS. Environment and Planning B: Planning and Design,
28 Vol. 35, No. 3, 2008, pp. 480–503.
29 39. Pace, K. Spatial Statistics Software for Matlab and Fortran. http://www.spatial-
30 statistics.com/software_index.htm. Accessed July 12, 2012.
31 40. Leech, B. L. Asking questions: techniques for semistructured interviews. Political Science &
32 Politics, Vol. 35, No. 04, 2002, pp. 665–668.
33 41. Gerkman, L. Empirical spatial econometric modelling of small scale neighbourhood. Journal
34 of Geographical Systems, 2011, pp. 1–16.
35 42. Chatman, D. G., N. K. Tulach, and K. Kim. Evaluating the Economic Impacts of Light Rail
36 by Measuring Home Appreciation A First Look at New Jersey’s River Line. Urban Studies,
37 Vol. 49, No. 3, 2012, pp. 467–487.
38

You might also like