1# RECORD 07441798ac5873fe53746864fd3cf3231a87b0562# TITLE: Impact of infill development on prices of existing apartments in Finnish urban neighbourhoods3# YEAR: 2018.0 VENUE: nan DOI: 10.3846/ijspm.2018.15404# AUTHORS: Hannele Ahvenniemi; K. Pennanen; Antti Knuuti; A. Arvola; Kauko Viitanen5# PAGES: 11 CANDIDATE TABLE PAGES (1-based): [6, 7]6# ABSTRACT: Popularity of infill development is increasing because of the environmental benefits and cost saving potential it provides, which relate to the possibility to use existing infrastructure and services. However, the impact of infill development on value of existing properties has not been studied to a sufficient extent. Therefore, the aim of our study is to analyse whether infill development affects the prices of existing apartments. We carried out statistical analysis based on data from seven case neighbourhoods, and prices of more than 6000 housing transactions from one decade. The results of our analysis do not support the hypothesis of infill development affecting positively existing housing prices, but neither did the study show a significant negative effect. Both amenity effect and negative externalities may provide explanations as to why property values do not change due to infill development.789===== PAGE 1 (score 9, numeric-paren 1, table-words 0, var-words 8) =====1011Copyright © 2018 The Author(s). Published by VGTU Press12This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unre-13stricted use, distribution, and reproduction in any medium, provided the original author and source are credited.14International Journal of Strategic Property Management15ISSN 1648-715X / eISSN 1648-9179162018 Volume 22 Issue 3: 157–16717https://doi.org/10.3846/ijspm.2018.154018*Corresponding author. E-mail: hannele.ahvenniemi@aalto.fi19IMPACT OF INFILL DEVELOPMENT ON PRICES OF EXISTING 20APARTMENTS IN FINNISH URBAN NEIGHBOURHOODS21Hannele AHVENNIEMI1,*, Kyösti PENNANEN1, Antti KNUUTI1, 22Anne ARVOLA1, Kauko VIITANEN2231 VTT Technical Research Centre of Finland, Espoo, Finland242 Department of Real Estate, Planning and Geoinformatics, Aalto University School of Engineering, Espoo, Finland25Received 21 October 2015; accepted 27 June 201726Abstract. Popularity of infill development is increasing because of the environmental benefits and cost saving potential it 27provides, which relate to the possibility to use existing infrastructure and services. However, the impact of infill develop-28ment on value of existing properties has not been studied to a sufficient extent. Therefore, the aim of our study is to ana-29lyse whether infill development affects the prices of existing apartments. We carried out statistical analysis based on data 30from seven case neighbourhoods, and prices of more than 6000 housing transactions from one decade. The results of our 31analysis do not support the hypothesis of infill development affecting positively existing housing prices, but neither did the 32study show a significant negative effect. Both amenity effect and negative externalities may provide explanations as to why 33property values do not change due to infill development.34Keywords: infill development, housing prices, urban neighbourhood, property value, price development.35Supplementary material associated with this article can be found, in the online version, at https://doi.org/10.3846/ijspm.2018.154036Introduction37Because of the challenges posed by growing cities more 38focus today is put on developing existing neighbourhoods 39with infill development (for example Ooi & Le, 2013; Haa-40land & Bosch, 2015). Several studies have shown the envi-41ronmental benefits of infill construction or transit orient-42ed development, such as decreasing energy consumption 43and related smaller carbon footprint from transport (for 44example Clark, 2013; Nahlik & Chester, 2014; Glaeser & 45Kahn, 2010). Another observed benefit of infill develop-46ment is the cost-effectiveness, as the already existing in-47frastructure allows developing a neighbourhood with only 48moderate investment (Nykänen et al., 2012; Biddle, Ber-49toia, Greaves, & Stopher, 2006).50Even if a large number of economic and environmen-51tal benefits of infill development have been proven, and 52promoting infill development has been presented as one 53of the targets of the housing policy of the national gov-54ernment of Finland (Ministry of the Environment, 2016), 55new housing development is often, however, opposed by 56residents of the neighbourhood. Resistance to infill devel-57opment might occur when residents fear that the com-58munity does not benefit from the development or, even 59worse, the new development might have a negative impact 60by worsening the amenities of the neighbourhood or by 61resulting in negative value development of existing dwell-62ings (for example Matthews, Bramley, & Hastings, 2015). 63Even if fear of decreasing property values is a common 64reason behind the claims in planning practice, the impact 65of new development on existing housing has not been suf-66ficiently studied (Seppälä, 2013).67Our contribution to this research field is to examine, 68by using case studies, whether new housing developments 69affect the value of existing apartments in Finnish neigh-70bourhoods. Earlier studies, which are presented in sec-71tion 1, mainly focus on different type of real estate mar-72ket (metropolitan cities in the U.S. and Asia) or different 73housing types (such as single-family houses) and therefore 74our research is among the first ones to study the impact of 75infill development on value of dwellings in Nordic urban 76neighbourhoods.777879===== PAGE 6 (score 28, numeric-paren 0, table-words 5, var-words 15) =====808116282H. Ahvenniemi et al. Impact of infill development on prices of existing apartments in Finnish urban neighbourhoods83where the dependent variable P is sales price (1000€) per 84m². The independent variables are explained in Table 2. 85Epsilon (ε) stands for normally distributed error term.86The analysis was carried out separately for the seven 87cases. The studied point in time varied, depending on the 88timing of the infill development: we used transactions up 89to five years before and after infill development. A long 90enough time period before and after was chosen to ensure 91that 1) the expectations due to new developments would 92not yet have affected the prices in the “before” groups and 932) the potential effects could be observed in the “after” 94groups. As the number of transactions per year remained 95rather small, in most cases we used transactions from a 96two-year period (in few cases it was necessary to study a 97three-year period). The sale prices were adjusted for infla-98tion using the Consumer Price Index, and with 2012 as 99the base year.100Table 2. Independent variables and mean values for the neighborhoods101Variable102Description103Mean104(N = 2762)105Min106Max107SD108CONDITION109Condition of the dwelling (0 = bad, 1 = average, 1103 = good)1111.754112011331141.13115ELEVATOR116Dummy for the elevator (0 = no, 1 = yes)1170.485118011911200.499121ROOMS122Number of rooms, excl. kitchen1232.373124112561260.962127FLOOR128Floor number/ number of total floors1290.6451300.113111320.281133SEA134Distance to the sea cost (m)1353 1201362213711 3401383 718139CBD140Distance to the city centre (railway station) (m)1419 8811422 95114321 1241444 894145GREEN146Distance to parks or green areas (m)147174.51480149609150142151TREAT152Dummy for the treatment group (0 = control group, 1 = treatment group)153AFTER154Dummy for the “after infill development” cases (0 = before, 1 = after)155Table 3. Mean values of prices and on-sale days for treatment and control groups156PRICE/m2 (thousand €)157On-sale time158N159Before160After161Before162After163Before164After165Case 1166Treatment1671.9191682.38416959.417062.817161172100173Control1741.8791752.43017668.817770.817858179109180Case 2181Treatment1822.1041832.94018443.218553.2186117187221188Control1892.0271902.91019139.119244.519354194120195Case 3196Treatment1971.6311982.26019947.620041.32017920267203Control2041.8352052.29620635.920769.92084020970210Case 4211Treatment2121.6122132.39621466.221535.82165121741218Control2191.7412202.60522147.222246.922314922468225Case 5226Treatment2272.8022284.43622930.623072.323145232138233Control2342.7472354.67623633.023742.0238115239230240Case 6241Treatment2421.6742432.59324447.524547.9246146247228248Control2492.1402502.94925144.725243.72532825444255Case 7256Treatment2571.1842581.75825945.626059.326136262116263Control2641.2502652.07626673.026741.9268117269942703. Result271Price development of both case neighbourhoods and ref-272erence neighbourhoods was analysed by examining the 273average sale prices before and after infill development. Av-274erage sale prices and average duration of apartments being 275on sale are shown in Table 3. A general price increase can 276be observed in all treatment and control groups indicating 277that both groups follow the rising price trend of housing 278transactions in the capital area of Finland. Also, changes 279in on-sale times give indication on the development of the 280housing market in each treatment and control neighbour-281hood. Longer on-sale time might lead into larger discount 282and hence, lower sale price. However, as the prices are 283consistently higher in the “after” cases but on-sale times 284vary largely (taking both larger and smaller values than in 285the “before” cases), no consistent logic between the sale 286prices and on-sale time can be observed.287288289===== PAGE 7 (score 134, numeric-paren 103, table-words 5, var-words 6) =====290291International Journal of Strategic Property Management, 2018, 22(3): 157–167292163293Figure 1 illustrates the development of average non-de-294flated sale prices in case neighbourhood 1 (Karakallio), in 295the corresponding reference neighbourhood (Viherlaakso) 296and the sale prices of new apartments in the case neigh-297bourhood. For comparison, also the price index of a larger 298surrounding area is presented. The graph illustrates how the 299price development in the case and the reference neighbour-300hood is very similar during the whole studied time frame, 301and the development of prices in both neighbourhoods also 302follow the development of the price index of the wider area. 303Similar graphs for the seven case neighbourhoods are pre-304sented in the Supplementary Appendix 1.305The results from the difference-in-difference hedonic 306regression are reported in Table 4. The following obser-307vations can be drawn from the regression results: Being 308located in a neighbourhood in which infill development 309takes place (TREAT) does not seem to provide higher 310prices compared to the control neighbourhood. In most 311cases the effect is insignificant, and in one of the cases with 312significant effect, the effect is the opposite, and the prices 313are higher in the control neighbourhood. Furthermore, as 314was observed also in Table 3, prices are constantly higher 315after infill development (AFTER) which is in line with the 316price trend of the whole capital area. The most interesting 31703185031910032015032120032225032330032435032540032603275003281 0003291 5003302 0003312 5003323 0003333 5003344 0003354 5003365 0003372000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012338Price index339Price, €/m² 340Year341Case 1 (Karakallio ‐ Espoo)342Karakallio (CN)343New apartments (CN)344Viherlaakso (RN)345Price index for the area346Infill development347Figure 1. Case 1: Mean values of sale prices (by m2) of old apartments in the case 348neighbourhood (CN), the reference neighbourhood (RN), new apartments in the 349case neighbourhood and the price index in the area350Table 4. Results of the difference-in-difference hedonic regression analyses of the seven cases351Case 1352Case 2353Case 3354Case 4355Case 5356Case 6357Case 7358CONDITION3590.0991***360(0.0149)3610.120***362(0.0181)3630.0750**364(0.0236)3650.050**366(0.0187)3670.1360***368(0.0187)3690.0541**370(0.0180)3710.0758***372(0.0011)373ELEVATOR3740.0695 375(0.0401)3760.0079377(0.0422)3780.1287379(0.1104)3800.0161381(0.0446)382–0.1336**383(0.0492)384–0.0249385(0.0436)3860.0253387(0.0256)388ROOMS389–0.2708***3900.0161391–0.2565***392(0.0189)393–0.2929***394(0.0250)395–0.2427*396(0.0216)397–0.091***398(0.0019)399–0.2384***400(0.0187)401–0.2798***402(0.0157)403FLOOR404–0.0617405(0.05654)4060.1484*407(0.0633)4080.0852409(0.0709)4100.0438411(0.0657)4120.2984***413(0.0074)414–0.0107415(0.0675)4160.1189**417(0.0399)418SEA419–0.00031 420(0.00075)421–0.00070***422(0.00019)4230.0017**424(0.005519425–0.00019426(0.00021)427–0.00076***428(0.00019)429–0.0003**430(–0.00010)4314.677e–5432(8.573e–5)433CBD4340.00051435(0.00069)4360.000132*437(0.000629)438–0.0016**439(0.0005349)4400.00023441(0.0002)4420.00017**443(6.175e–5)4440.00010445(9.64e–5)446–0.0002**447(7.082e–5)448GREEN4490.00049***450(0.00012)451–0.000495**452(0.000185)4531.089e–6454(0.6365e–3)4550.00004456(0.00016)457–0.00018458(0.00023)4597.77e–5460(0.0001)461–5.349e–5462(0.00015)463TREAT4640.2943**465(0.1074)4660.1873467(0.1094)4680.7863*469(0.3022)470–0.0244471(0.1102)4720.2010*473(0.100)474–0.7073**475(0.2201)4760.4244*477(0.1981)478AFTER4790.5759***4800.05084810.756***482(0.0751)4830.4324***484(0.0757)4850.8114***486(0.0504)4871.847***488(0.0574)4890.8183***490(0.1171)4910.8264***492(0.0321)493TREAT*AFTER494–0.0953495(0.0692)496–0.0298497(0.0905)4980.0641499(0.0852)500–0.1545501(0.0857)502–0.2703**503(0.1018)5040.1426505(0.1310)506–0.1597*507(0.0770)508Multiple R25090.68855100.60125110.73025120.74995130.78555140.6645150.8112516The dependent variable is price (thousand €) per m2. The significance levels are ***<0.001, **<0.01, *<0.05. The standard errors of coefficients are 517reported in brackets.518519520===== PAGE 8 (score 14, numeric-paren 0, table-words 0, var-words 14) =====521522164523H. Ahvenniemi et al. Impact of infill development on prices of existing apartments in Finnish urban neighbourhoods524information is provided by the variable TREAT*AFTER. 525This variable shows the effect of being located in both the 526treatment group and the “after” group. However, the ef-527fect remains insignificant in most cases. Only in case 5, 528the variable shows a more significant effect, but the effect 529is opposite to our assumption, suggesting lower prices for 530apartments being located in the treatment neighbourhood 531(after treatment). The specific nature of the case 5 neigh-532bourhoods however suggests that the results do not neces-533sarily indicate a negative price effect of infill development; 534In case 5, the treatment and control neighbourhoods are 535located on an island, which implies that attributes such as 536location close to the sea and high floor number are highly 537valued by the residents (as is also shown by the results 538of the regression analysis). Even more importantly, the 539on-going construction of a new metro line with a metro 540station in the central area of the island (which is also the 541location of the control neighbourhood) has supposedly, 542already since several years, provided a price premium for 543apartments located nearby.544Regarding other independent variables of our model, 545the impact of structural characteristics of the apartments/546buildings seems as expected: Larger room number leads to 547lower prices per m2 and the good condition of the apart-548ment provides a significant price premium. Higher floor 549number in relation to total floor number has a positive 550significant effect on prices in a few cases but not in all. 551Also surprisingly, access to elevator does not have a signif-552icant effect implying that apartments in low-rise buildings 553might be more preferred than apartments in tall buildings.554The effect shown by the environmental variables are 555somewhat as expected. Decreasing distance to sea has a 556significant positive effect on prices in those neighbour-557hoods which are located close to the coastline, whereas in 558others the impact is insignificant. Distance to CBD has a 559similar effect in two cases but the opposite effect in one 560case (case 5). In case 5 this abnormality can be explained 561by the specific characteristics of the neighbourhood (al-562ready explained earlier), which are likely to affect prices 563more than distance to CBD. Although earlier research has 564suggested a positive price impact of parks and green ar-565eas, our analysis does not show a price premium provided 566by easy access to green areas. In case 1, the effect is even 567against the assumption suggesting that prices are higher 568in locations with longer distances to green areas. This can 569however be explained by the geography of the area; as the 570whole area is surrounded by woods, it is not surprising 571that a location closer to the centre of the area, and closer 572to services (and also further away from the woods), might 573be more preferable.574Conclusions and discussion575Only a rather limited number of studies have examined 576whether infill development affects the prices of existing 577housing. These studies differ from each other regarding 578the type of buildings considered (single-family houses or 579apartments), and most of the studies focus on the U.S. 580housing market covering a rather limited geographic area. 581Thus our study sheds light on a geographic area and a type 582of housing market which has so far been rather undiscov-583ered, regarding the impact that infill development has on 584existing properties. All the selected case neighbourhoods 585are typical urban areas in the capital region of Finland, 586with a large number of residential apartment buildings 587built in the 60s, 70s and 80s. Even though our study is 588based on individual cases, we however do cover the capi-589tal area fairly well with case neighbourhoods spread over 590three cities (Helsinki, Espoo and Vantaa). During those 591years, a similar construction boom took place also in oth-592er European countries, and these buildings are currently 593facing challenges because of increasing renovation need, 594some also being in need of a “facelift”. It has been sug-595gested that infill development can have a crucial role re-596garding these issues, both by contributing to the upgrade 597of the neighbourhood as well as providing housing com-598panies with a solution to finance their renovation. Hence, 599the results of our study are undoubtedly of interest to a 600number of different actors, who might be involved with 601processes related to infill development, ranging from in-602dividual apartment owners to city planners.603Our study examined the effect of infill development on 604the prices of existing apartments by using difference-in-605difference hedonic regression. We carried out the analysis 606for seven case neighbourhoods and appointed reference 607neighbourhoods and analysed whether prices before infill 608development were statistically different from prices after 609in the two neighbourhood groups. Overall, the results 610showed no significant difference in price development. 611Hence the results of our analysis indicate that infill de-612velopment has no significant impact on prices of existing 613apartments. Only in one case a more significant effect of 614being located in the treatment group was observed, but 615in this case the regression analysis suggested a negative 616impact. However, given the specific nature of this neigh-617bourhood (being located on an island and the new metro 618line being built in the control neighbourhood), this result 619should not be considered as indication of a negative price 620effect of infill development.621Our results are in line with most of the previous studies 622which also did not find strong or any correlation between 623infill development and changes in housing prices (Blan-624chard et al., 2008; Zahirovich-Herbert & Gibler, 2014; Pol-625lakowski et al., 2005). Only two studies are contradicting 626with our results suggesting a significant impact (Ooi & 627Le, 2013; Mathur & Ferrell, 2013) but these studies differ 628greatly from our research by the location of the studies 629(Singapore and California) and by the type of housing 630studied (single-family houses in the study by Mathur & 631Ferrell, 2013).632To find explanations for the results, it is important to 633understand the specific nature of the property market. 634Attributes such as purchasing power, size and features of 635the neighbourhood affect the value of residential prop-636erties. Also, the general economic situation and develop-637ment (such as economic recession or depression) affect the 638