1# RECORD 877cbdabc2aba0920a744f15d91bf843fa770f5c2# TITLE: The Road to Higher Prices: Will Improved Road Standards Lead to Higher Housing Prices?3# YEAR: 2020.0 VENUE: Journal of real estate finance and economics DOI: 10.1007/s11146-020-09751-y4# AUTHORS: Theis Theisen; A. Emblem5# PAGES: 25 CANDIDATE TABLE PAGES (1-based): [12, 13, 18, 19]6# ABSTRACT: We study the impact of improved road standards on housing prices in a region of Southern Norway, where towns are located along the coast, like pearls on a string, with sparsely populated areas between the towns. In this region towns have since long been linked together by a road, but in 2009 a new highway, roughly parallel to the old road, was opened. The new highway runs Eastwards from the main town in the region and reduced one-way commutes to the main town by 10–15 min. We examine how this change in transportation infrastructure has affected housing prices, using difference-in-difference regressions to identify the impact of the new highway. Three smaller towns located East of the main town constitute the treatment region, while four towns and communities West of the main town form the control region. We find that the new highway on average increased housing prices in the treatment region by about 5%, but the effect on price differed substantially between the towns. The full impact on housing prices materialized when the new highway was opened. We estimate that the increase in total housing capital in the treatment region amounts to almost 84% of the construction costs of the new highway.789===== PAGE 1 (score 6, numeric-paren 0, table-words 0, var-words 6) =====1011The Road to Higher Prices: Will Improved Road12Standards Lead to Higher Housing Prices?13Theis Theisen114& Anne Wenche Emblem115Published online: 14 March 202016Abstract17We study the impact of improved road standards on housing prices in a region of Southern18Norway, where towns are located along the coast, like pearls on a string, with sparsely19populated areas between the towns. In this region towns have since long been linked20together by a road, but in 2009 a new highway, roughly parallel to the old road, was opened.21The new highway runs Eastwards from the main town in the region and reduced one-way22commutes to the main town by 10–15 min. We examine how this change in transportation23infrastructure has affected housing prices, using difference-in-difference regressions to24identify the impact of the new highway. Three smaller towns located East of the main town25constitute the treatment region, while four towns and communities West of the main town26form the control region. We find that the new highway on average increased housing prices27in the treatment region by about 5%, but the effect on price differed substantially between the28towns. The full impact on housing prices materialized when the new highway was opened.29We estimate that the increase in total housing capital in the treatment region amounts to30almost 84% of the construction costs of the new highway.31Keywords Roadinvestments.Infrastructureinvestments.Housingprices.Difference-in-32difference33Introduction34Since the contributions of Mohring (1961), Alonso (1964), Muth (1969), Mills (1972),35Solow (1972) and others, the relationship between housing (land) prices and monetary36J Real Estate Finan Econ (2021) 62:258–28237https://doi.org/10.1007/s11146-020-09751-y38* Theis Theisen39Theis.Theisen@uia.no40Anne Wenche Emblem41Anne.Wenche.Emblem@uia.no42143Department of Economics and Finance, School of Business and Law, University of Agder, P.O.44box 422, 4604 Kristiansand, Norway45# The Author(s) 2020464748===== PAGE 12 (score 38, numeric-paren 28, table-words 2, var-words 2) =====4950Table 2 Descriptive statistics, entire sample, N = 37,21651Mean52Std. Dev.53Min54Max55Dependent variables:56Price571,966,910581,053,6145940,0006023,000,00061Mutual6246,671.863187,298.3640653,440,00066Dwelling characteristics and lot:67Floor68111.38696952.5146270167146772Age7336.892557435.752547507646177Lota78402.2755795104.21480081381,35482Dwelling types and ownership83Co-op84.164955985.371145886087188Detached89.390853490.487948291092193Twin-house94.101837995.302439196097198Row-house99.0940993100.291970810101021103Apartment104.4132094105.492416410601071108Treatment variables109TreatA110.036624111.187839511201131114TreatB115.082572116.275237911701181119TreatC120.1793046121.383612312201231124TreatE125.2985006126.457606412701281129TreatK130.5382094131.498544613201331134a. Lot size above 1000 sqm. For non-cooperative dwellings exclusive of apartments135Table 3 Descriptive statistics for subsamples136Mean values137P-values for test of different meansc138Variable139Treatment140Control141Core142Treatment143Treatment144Control145region146region147town148Control149Core150Core151Price1521,828,9691531,810,5501542,090,8531550.21071560.0000***1570.0000***158Mutuala159402,929.9160921,140.5161238,264.21620.0000***1630.0000***1640.0000***165Floor166119.9799167122.7199168103.18261690.0008***1700.0000***1710.0000***172Age17340.5166117429.732117537.055021760.0008***1770.0000***1780.0000***179Lotb180693.9936181466.2936182198.36911830.0339*1840.0061**1850.0022**186Co-op187.084436188.026987189.25147281900.0000***1910.0000***1920.0000***193Detached194.5859213195.4999177196.24957561970.0000***1980.0000***1990.0000***200Twin-house201.0864164202.1316439203.1013482040.0000***2050.0000***2060.0000***207Row-house208.0559006209.0929735210.11562662110.0000***2120.0000***2130.0000***214Apartment215.2717616216.2754649217.53344982180.60242190.0000***2200.0000***221N22211,109223607722420,030225a. means of mutual debt for dwellings that carry a positive mutual debt226b. Lot size above 1000 sqm. For non-cooperative dwellings exclusive of apartments227c. Stars indicate statistical significance, with p-values: * p < 0.05, ** p < 0.01, *** p < 0.001228The Road to Higher Prices: Will Improved Road Standards Lead to...229269230231232===== PAGE 13 (score 10, numeric-paren 7, table-words 1, var-words 0) =====233234Table 4 Correlation matrix with P-values in parentheses, N = 37,216a235Treat E10236Floor237Age238Lot239Co-op240Twin- house241Row-house242Apartment243TreatE102441.0000245–246Floor247.0499***2481.0000249(.0000)250–251Age252−.0379***253.0194***2541.0000255(.0000)256(.0002)257–258Lot259.0265***260.1049***261.0786***2621.0000263(.0000)264(.0000)265(.0000)266–267Co-op268−.0748***269−.3407***270−.0247***271−.0665***2721.0000273(.0000)274(.0000)275(.0000)276(.0000)277–278Twin-house279−.0212***280.0156**281−.0609***282−.0008283−.1348***2841.0000285(.0000)286(.0026)287(.0000)288(.8698)289(.0000)290–291Row-house292−.0536***293−.0190***294−.0323***295−.0209***296−.0304***297−.1085***2981.0000299(.0000)300(.0003)301(.0000)302(.0001)303(.0000)304(.0000)305–306Apartment307−.0775***308−.6391***309−.1070***310−.1256***311.4453***312−.2826***313−2705***3141.0000315(.0000)316(.0000)317(.0000)318(.0000)319(.0000)320(.0000)321(.0000)322–323a. Stars indicate statistical significance, with p-values: * p < 0.05, ** p < 0.01, *** p < 0.001324270325T. Theisen, A. W. Emblem326327328===== PAGE 14 (score 12, numeric-paren 0, table-words 5, var-words 2) =====329330correlation between most independent variables is low. The only exception is the331correlation between the apartment dummy and floor-space, but even this is not alarm-332ingly high.333Figure 3 shows box plots of housing price per square meter floor-space for detached334dwellings, which dominates the housing market in both the treatment and control335regions, and of which there are many also in the Core town. The figure reveals that336the median price as well as the change in prices over the entire study period is similar337for the treatment region and the control region. The Core town differs, however, in that338the price level is higher, and the price increase over the entire period is stronger than in339the treatment and control region. The higher price level in the Core town may inter alia340be related to the fact that its population is larger than in the other towns and commu-341nities. Stronger price growth in the Core town may be related to stronger population342growth in the Core town than in the treatment and control regions, cf. Fig. 2. The higher343housing price level and growth in the Core town are among the reasons why we have344chosen to exclude this town from both the treatment and the control region.345Empirical Results346Results from estimating various versions of Eq. (4) are shown in Table 5. In Regression347A all towns in the treatment region are subsumed under one category, TreatE. The348treatment period includes the years 2010–2013, the years 2003–2006 constitute the349non-treatment period, while data from the years 2007–2009 are excluded from the350estimation. Regression A explains 71% of the variation in housing prices. The Vif351statistics indicate that collinearity of regressors is not a problem, but since the Breusch-352Pagan statistics reveals that error terms are heteroscedastic we use throughout robust353standard errors estimated with Huber-White’s “sandwich” estimator.354The treatment effect in Regression A is affiliated with the variable TreatE10, where355E indicates that the treatment region lies East of the Core town and 10 indicates that the356first year in the treatment period, i.e. the first year when the full treatment effect kicks357in, is 2010. The impact parameter of TreatE10 in Table 5 is statistically significant and358amounts to 0.0488. Hence, we ascribe an increase in housing prices in the treatment359region of almost 5% to the new highway.6 Except for the Apartment dummy, all other360independent variables in Regression A that are explicitly shown in Table 5 have a361statistically significant impact on housing prices, with signs and magnitudes that are362reasonable and in good accordance with previous research. In estimating Regression A363we have controlled for the location of dwellings by including zip-codes. Many of these364have a strong and statistically significant impact on housing prices, but since they are365not of primary interest for the issues we address, they are not shown in Table 5. Dummy3666 We estimated also a variant of Regression A in which all terms entered linearly, as well as a semilogarithmic367function where the dependent variable was log-transformed while all independent variables entered untrans-368formed. In contrast to the double-logarithmic form, the residuals in the two other functional forms deviated369substantially from the normal distribution, and heteroskedasticity was also much more pronounced than in the370double-logarithmic form we have used throughout the paper. Hence, we do not pay more attention to the linear371and semilogarithmic functional forms, but to mention that in the semilogarithmic function the treatment372variable, TreatE10, indicated an increase in housing price by 5.4%, while the linear function implied that373treatment increased housing prices by 8.4%.374The Road to Higher Prices: Will Improved Road Standards Lead to...375271376377378===== PAGE 18 (score 17, numeric-paren 7, table-words 4, var-words 2) =====379380Regression A, while Rob-2 shows results from estimation Regression A using381bζ ¼ 0:70. Comparing Rob-1 and Rob-2 with Regression-A it is evident that the382estimated treatment parameter is robust towards even very large variations in the383mutual debt parameter. Moreover, except for the parameter affiliated with the Co-op-384variable the other coefficients in Rob-1 and Rob-2, including the treatment coefficients,385are very similar to the corresponding coefficients in Regression A. The fit of Regres-386sion A is also slightly better than that for Rob-1, and very similar to that of Rob-2.387Hence, we conclude that using bζ ¼ 0:86 as prior information in the estimation seems to388be unproblematic.389Rob-3 in Table 7 shows results from a placebo analysis, where the treatment effect is390assumed to materialize as early as in 2004, i.e. one year before the decision to build the391new highway was made. The estimated treatment parameter is close to zero and392statistically insignificant, indicating that housing prices in the treatment region did393not divert from those in the control region as early as in the year prior to the final394decision to build the new highway.395Table 6 Alternative starts of treatment period, OLS regressions, N = 17,186a396Year treatment period is assumed to start397200539820063992007400200840120094022010403Treatment coefficient404.0032405.0140406.0262**407.0398***408.0539***409.0549***410(.0110)411(.0096)412(.0084)413(.0079)414(.0076)415(.0076)416Ln (Floor)417.5845***418.5848***419.5850***420.5853***421.5853***422.5852***423(.0072)424(.0072)425(.0072)426(.0072)427(.0072)428(.0072)429Ln (Age)430−.0629***431−.0629***432−.0629***433−.0628***434−.0628***435−.0627***436(.0020)437(.0020)438(.0020)439(.0020)440(.0020)441(.0020)442Ln (Lot)443.0094***444.0094***445.0094***446.0094***447.0094***448.0094***449(.0011)450(.0011)451(.0011)452(.0011)453(.0011)454(.0011)455Co-op456−.0423***457−.0422***458−.0419***459−.0417***460−.0419***461−.0418***462(.0093)463(.0093)464(.0093)465(.0093)466(.0093)467(.0093)468Twin-house469−.0695***470−.0693***471−.0692***472−.0690***473−.0693***474−.0693***475(.0058)476(.0058)477(.0058)478(.0058)479(.0058)480(.0058)481Row-house482−.1131***483−.1131***484−.1129***485−.1128***486−.1126***487−.1128***488(.0064)489(.0064)490(.0064)491(.0064)492(.0064)493(.0064)494Apartment495.0031496.0032497.0032498.0032499.0029500.0034501(.0073)502(.0073)503(.0073)504(.0073)505(.0073)506(.0073)507Constant50811.53***50911.52***51011.52***51111.52***51211.51***51311.52***514(.0397)515(.0395)516(.0396)517(.0396)518(.0395)519(.0395)520Zip-codes521Yes522Yes523Yes524Yes525Yes526Yes527Years528Yes529Yes530Yes531Yes532Yes533Yes534Months535Yes536Yes537Yes538Yes539Yes540Yes541R-Sq542.6901543.6901544.6902545.6905546.6909547.6909548a. Dependent variable: Ln (Price+0.86*MUT). Robust (Huber-White) standard errors in parentheses. Stars549indicate statistical significance, with * denoting 5% level, ** 1% level and *** 0.1% level550The Road to Higher Prices: Will Improved Road Standards Lead to...551275552553554===== PAGE 19 (score 23, numeric-paren 7, table-words 7, var-words 2) =====555556Table 7 Robustness checks based on Regression A, OLSa557Regression A558Rob-1559Rob-2560Rob-3561Rob-4562Rob-5563Rob-6564ζ = 0.86565ζ = 1.0566ζ = 0.7567TreatE04568Detached569Small570Apartment571TreatE10572.0488***573.0496***574.0478***575.0446***576.0658***577.0391***578(.0092)579(.0092)580(.0092)581(.0138)582(.0150)583(.0173)584TreatE04585−.0078586(.0144)587Ln(Floor)588.5825***589.5820***590.5832***591.5844***592.5716***593.4651***594.6615***595(.0086)596(.0086)597(.0086)598(.0072)599(.0119)600(.0167)601(.0162)602Ln(Age)603−.0679***604−.0690***605−.0664***606−.0630***607−.0625***608−.0661***609−.0786***610(.0023)611(.0023)612(.0023)613(.0020)614(.0041)615(.0045)616(.0035)617Ln(Lot)618.0097***619.0098***620.0097***621.0094***622.0103***623.0016***624(.0013)625(.0013)626.0013)627(.0011)628(.0014)629(.0023)630Co-op631−.0382***632.0032633−.0896***634−.0423***635.0369***636−.0497***637(.0107)638(.0107)639(.0117)640(.0093)641(.0552)642(.0140)643Twin-644−.0740***645−.0752***646−.0725***647−.0695***648.0499***649house650(.0069)651(.0069)652(.0069)653(.0058)654(.0082)655Row-656−.1168***657−.1168***658−.1167***659−.1131***660house661(.0074)662(.0074)663(.0075)664(.0064)665Apartment666.0017667−.0006668.0048669.0031670(.0086)671(.0086)672(.0086)673(.0073)674Constant67511.54***67611.55***67711.53***67811.53***67911.63***68011.97***68111.02***682(.0473)683(.0472)684(.0474)685(.0399)686(.0712)687(.1243)688(.0745)689R-Sq690.7100691.7095692.7101693.6901694.6632695.7848696.7386697N69812,63069912,63070012,63070117,186702705370321227043455705a. Dependent variable: Ln(Price+ζ*MUT), with ζ = 0.86, except in Rob1 and Rob2, where ζ is specified in the head of the table. Rob1 and Rob2 are variants of Regression A. In Rob-3706the treatment effect is assumed to kick in in 2004. Rob-4, Rob-5 and Rob-6 are similar to Regression A, but estimated from the subsamples of respectively detached houses, small707houses and apartments. Robust (Huber-White) standard errors in parentheses. Stars indicate statistical significance, with * denoting 5% level, ** 1% level and *** 0.1% level708276709T. Theisen, A. W. Emblem710711712===== PAGE 20 (score 23, numeric-paren 0, table-words 10, var-words 3) =====713714In regressions Rob-4, Rob-5 and Rob-6 in Table 7 we have re-estimated Regression715A separately for detached houses, apartments, and small houses (row-houses and twin-716houses taken together). The estimated treatment parameters for the three types of717dwellings are all positive and statistically significantly different from zero. The718dwelling-type specific parameter estimates are not statistically significantly different719from the treatment parameter in Regression A, but the differences are sufficiently large720to warrant a closer examination.721In Table 8 we present results from estimating Regression A separately for three types722of dwellings, with alternative assumptions on when the treatment effect kicks in.723Hence, the estimated relationships are of the same type as those in Table 6, except that724they are dwelling-type specific in Table 8. We observe from Table 8 that the estimated725treatment coefficients for each of the three dwelling types are roughly of the same726magnitude for the years 2009 and 2010, while the estimates for earlier years are smaller.727Hence, for each of the three dwelling types, it seems that the full treatment effect728materializes towards the end of the construction period. Since the estimated coefficients729for detached houses, small houses and apartments are far from being statistically730significantly different from each other, this provides evidence that the full treatment731effect is similar for all types of dwellings.732Consider now the results obtained by assuming that the treatment effect kicks in in733alternative years before 2009. The regressions in column 2008 in Table 8 are based on734the presumption that the treatment effect kicks in in 2008, and show that the magnitude735of the estimated treatment coefficient is much larger for small houses and apartments736than for detached houses. Moreover, for small houses and apartments the estimated737treatment effects in column 2008 are almost as large as those in columns 2009 and7382010. Hence, from the results in Table 8 it seems warranted to conclude that the739treatment effect for apartments and small houses kicks in with considerable magnitude740already in the middle of the construction period. For detached houses by contrast, the741estimates in Table 8 indicate that the treatment effect in the middle of the treatment742period is much weaker. Finally, from the results in Tables 8 and 6 there are indications743of a small treatment effect also in 2007, the year when construction of the new road744started up. Hence, we conclude that the treatment effect seems to kick in somewhat745earlier for apartments and small houses, and that the early effects also seem to be746stronger for such dwellings than for detached houses.747The observations just made deserve some elaboration. An important reason why the748treatment effect does not kick in before the year construction of the new road started749may be that people do not react before they with their own eyes see that construction of750the new road has started. Next, there may be several reasons why the treatment effect in751the middle of the construction period according to our results seems to be stronger for752small houses and apartments: First, most apartments and small houses are more753standardized than detached houses, i.e. comparable apartments and small houses are754easily found. Market actors may then easily find comparable units recently sold.755Adjustment of asking price to a realistic level may then work smoother than for756detached houses, where it may be hard to find similar houses recently sold within the757same local environment. In addition, sellers of detached houses in the markets we758consider often may have to wait quite long before the “right” bidder turns up and hands759in his or her bid. Consequently, sellers of detached houses may often be uncertain why760he or she does not see a “sufficiently high” bid: Is it because the asking price is too high761The Road to Higher Prices: Will Improved Road Standards Lead to...762277763