1# RECORD 21199d311319fd0034523954fb23eaaf347394282# TITLE: Assessing the Impact of Urban Improvement on Housing Values: A Hedonic Pricing and Multi-Attribute Analysis Model for the Historic Centre of Venice3# YEAR: 2017.0 VENUE: nan DOI: 10.3390/buildings70401124# AUTHORS: P. Rosato; M. Breil; C. Giupponi; Raul Berto5# PAGES: 24 CANDIDATE TABLE PAGES (1-based): [13, 14]6# ABSTRACT: The Hedonic Pricing Method is one of the principal assessment methods for evaluating services and resources not normally exchanged on the market. However, the method is often unable to account for the great variety of qualities in an urban context and faces scarce and heterogeneous market data. This paper presents a model for the valuation of benefits generated by environmental and urban improvement investments adopting a mixed hedonic-multi-attribute procedure for modeling a value function of urban real estate values. The peculiarity of the model is that the independent variables are aggregated indicators, which synthetize more detailed characteristics. Using the expertise of real estate agents, all relevant variables influencing real estate values were weighted and synthetized in a set of cardinal indicators. Next, market prices were used to calibrate a hedonic function that transforms the cardinal indicators into real estate values. The valuation model was integrated into a GIS for mapping the housing value, and its variation induced by urban investment. The proposed model pointed out plausible and robust results, in particular, the possibility to use any available information, such as location, position, technical and economic characteristics of buildings, and organize it in a flexible and transparent way, and to keep evident the role of each characteristic through the hierarchical structure of the model. The model was applied to the real estate market of Venice to test the effects of the MOSE project (Electromechanical Experimental Module) for the protection of Venice from high tides. The results of the application showed a relevant increase in real estate values in the center of Venice, especially related to property in ground floor units, of about 1.4 billion €.789===== PAGE 1 (score 3, numeric-paren 0, table-words 0, var-words 3) =====1011buildings12Article13Assessing the Impact of Urban Improvement14on Housing Values: A Hedonic Pricing and15Multi-Attribute Analysis Model for the Historic16Centre of Venice17Paolo Rosato 1,*18ID , Margaretha Breil 219ID , Carlo Giupponi 320ID and Raul Berto 121ID22123Department of Engineering and Architecture, University of Trieste, 34127 Trieste, Italy; rberto@units.it24225Fondazione Eni Enrico Mattei, 30124 Venice, Italy; margaretha.breil@feem.it26327Department of Economics, Ca’ Foscari University, 30121 Venice, Italy; cgiupponi@unive.it28*29Correspondence: paolo.rosato@dia.units.it; Tel.: +39-040-558-356930Received: 31 July 2017; Accepted: 23 November 2017; Published: 30 November 201731Abstract: The Hedonic Pricing Method is one of the principal assessment methods for evaluating32services and resources not normally exchanged on the market. However, the method is often unable33to account for the great variety of qualities in an urban context and faces scarce and heterogeneous34market data. This paper presents a model for the valuation of benefits generated by environmental and35urban improvement investments adopting a mixed hedonic-multi-attribute procedure for modeling a36value function of urban real estate values. The peculiarity of the model is that the independent37variables are aggregated indicators, which synthetize more detailed characteristics. Using the38expertise of real estate agents, all relevant variables influencing real estate values were weighted39and synthetized in a set of cardinal indicators. Next, market prices were used to calibrate a hedonic40function that transforms the cardinal indicators into real estate values. The valuation model was41integrated into a GIS for mapping the housing value, and its variation induced by urban investment.42The proposed model pointed out plausible and robust results, in particular, the possibility to use any43available information, such as location, position, technical and economic characteristics of buildings,44and organize it in a flexible and transparent way, and to keep evident the role of each characteristic45through the hierarchical structure of the model. The model was applied to the real estate market46of Venice to test the effects of the MOSE project (Electromechanical Experimental Module) for the47protection of Venice from high tides. The results of the application showed a relevant increase in real48estate values in the center of Venice, especially related to property in ground floor units, of about491.4 billion €.50Keywords: hedonic pricing method; multi-attribute analysis; real estate market; urban improvement;51MOSE project521. Introduction53The evaluation of urban improvements is a relevant issue for urban planning and public54decision making. Urban improvements are usually evaluated using either multiple-criteria analysis55methods, which allow for the coherent assessment of quantitative and qualitative data and produce56qualitative outputs in terms of preferred solutions, or cost–benefit analysis, which base judgments on57quantitative results in terms of benefits generated by public investments. The application of cost-benefit58approaches offers the potential advantage of assessing and comparing public benefits and co-benefits59of investments in urban infrastructures, especially in “soft” or “green” infrastructures with more60traditional assessments of infrastructure.61Buildings 2017, 7, 112; doi:10.3390/buildings704011262www.mdpi.com/journal/buildings636465===== PAGE 13 (score 12, numeric-paren 3, table-words 3, var-words 3) =====6667Buildings 2017, 7, 1126813 of 2469Table 1. Summary of cardinal variables of the sample.70Variable71Min72Max73Average74Std. Dev.75Price (€)7670,000772,500,00078402,95679271.51880Price (€ m−2)811679829000834528841.29885Residential area (m2)8627874008892895690Height above sea level (cm)91859221093145942895Table 2. Summary of nominal/ordinal variables of the sample.96Variable97Freq.98%99Cum. (%)100Urban Areas101Cannaregio1025710328.910428.9105Castello, San Pietro, via Garibaldi10661073.010832.0109Sant’Elena11021111.011233.0113Dorsoduro1142611513.211646.2117Giudecca/Sacca Fisola1182511912.712058.9121San Marco/Canal Grande122131236.612465.5125S. Polo1262412712.212877.7129S. Croce130171318.613286.3133Castello San Lio13451352.513688.8137Castello S. Martino138191399.614098.5141Case Muner, Baia del Re14211430.514499.0145Santa Marta14621471.0148100.0149Distance from Venice access150>20 min15110715254.315354.315410–20 min1554815624.415778.7158<10 min1594216021.3161100.0162Entrance163Shared16414516573.616673.6167Independent1685216926.4170100.0171Lift/Floor172Lift or ground floor1736317432.017532.0176No lift/2nd floor17710217851.817983.8180No lift/3rd floor1812418212.218395.9184No lift above 3rd floor18581864.1187100.0188Brightness189Very bright1905919129.919229.9193Bright1949619548.719678.7197Little bright1984219921.3200100.0201View202Unique203102045.12055.1206Great2072420812.220917.3210Good2118321242.121359.4214Mediocre2157021635.521794.9218Poor219102205.1221100.0222Maintenance223Excellent2246222531.522631.5227Good2286922935.023066.5231Mediocre2325123325.923492.4235Poor236152377.6238100.0239Finishes240Fine2416324232.024332.5244Ordinary2459024645.724777.7248Cheap2494425022.3251100.0252Table 3. Pearson correlations test of the touristic sample (85 cases; Sig. 2-code in brackets).253Variables254Characteristics255Position256Technical257Characteristics258Location259−0.052260−0.007261(0.637)262(0.952)263Position264–2650.283266–267(0.09)268269270===== PAGE 14 (score 24, numeric-paren 3, table-words 9, var-words 3) =====271272Buildings 2017, 7, 11227314 of 24274Table 4. The value function of touristic market segment, robust standard errors with respect to275heteroskedasticity, HC1 variant.276Variables277Coefficient278Std Error279Ratio280p-Value281Constant (β0)282753.3052831169.052840.64442850.5212286Location characteristics (Cl)2874963.652881300.152893.8182900.0003291Position characteristics (Cp)2922057.322931044.432941.9702950.0523296Technical characteristics (Ct)2973430.752981195.612992.8693000.0052301Note: Dependent variable market price (€ m−2); N. cases: 8430285; R-square3030.259846; Adjusted R-square3040.232433;305F(3, 81)3069.886302; P-value(F) 1.3 × 10−5.307Table 5. Pearson correlations test of the resident samples (112 cases; Sig. 2-code in brackets).308Variables309Characteristics310Position311Technical312Characteristics313Location314−0.2243150.164316(0.18)317(0.085)318Position319–3200.244321–322(0.10)323Table 6. The value function of residents’ market segment, robust standard errors with respect to324heteroskedasticity, HC1 variant.325Variables326Coefficient327Std Error328Ratio329p-Value330Constant (β0)3311211.81332323.0943333.7513340.0003335Location characteristics (Cl)336794.464337360.1943382.2063390.0295340Position characteristics (Cp)3411254.08342381.9143433.2843440.0014345Technical characteristics (Ct)3461950.08347350.8883485.558349<0.0001350Note: Dependent variable market price (€ m−2); N. cases351112; R-square3520.398214; Adjusted R-square3530.381498;354F(3, 81)35519.30655; p-value(F) 4.27 × 10−10.356According to the OLS stepwise analysis, economic characteristics did not significantly influence357the market price in the sample. Real estate values in the most attractive touristic areas were strongly358influenced by location (district) and technical characteristics (finishes and state of repair). In the areas359of prevalent local residential demand, a high and significant constant was observed. Where those360wanting to live in Venice had a significant “base” they were willing to pay, to which the effects of the361other characteristics were added. In these cases, positional and technical characteristics also appeared362to be significant, but with different priorities. Technical characteristics had the highest coefficient,363followed by position and location characteristics. For residents, the building characteristics appeared364to be much more important than its location. Buyers seemed to be willing to sacrifice the prestige365of a good location (higher presence of tourists) for good technical characteristics (maintenance and366finishing); however, this was dependent on the high refurbishment costs and the conflicts between the367resident’s life quality and tourist congestion.3683.3. The Geographic Model369The value functions developed in the previous paragraph were implemented in a GIS model370to represent the spatial distribution of present property values and simulate changes induced by371hypothetical urban improvements. In this study, GIS was used exclusively to represent the spatial372distribution of features and values. No spatial statistical analysis was performed that could be373an interesting development of the evaluation model. The GIS organized and geo-referenced the374information to represent the state of each building in Venice in relation to the characteristics (xi)375implemented in the valuation model. The layers adopted a raster grid of 1 m2 and were implemented376using data collected for preparing the master plan for the historic center of Venice. Assuming the quality377378379===== PAGE 15 (score 17, numeric-paren 0, table-words 0, var-words 17) =====380381Buildings 2017, 7, 11238215 of 24383of the available information, some simplifications and elaborations were made. First, the available384information referred to entire buildings rather than single flats, which are the most common property385type purchased. For missing data, an interpolation between nearby points was necessary. Regarding386the technical characteristics of the buildings, poor information was available to precisely define the387state of indicator, so a uniform (most probable) state of repair was assumed for those buildings lacking388specific information. The value functions applied to layers mapping the Venice property characteristics389allowed us to map the aggregate indicators (cl and cp) and, finally, the real estate value. In the following390paragraphs, some examples are illustrated.3913.3.1. The Distribution of Location Characteristics392The noise exposure of buildings noise exposure was assessed using the acoustic classification393plan developed by the municipality of Venice. The document shows that noise was essentially due to394boat traffic in the most important canals, and the presence of tourists and recreation facilities (mainly395bars) in the most congested areas. Plan mapping provided a classification in zones with similar noise.396Therefore, buildings in the various zones were assigned a relative noise class.397Flooding in Venice is an occurrence that happens as the result of a complex interplay of398metrological and oceanographic phenomena. High tides are favored especially where strong winds399from the south/south-east, such as the sirocco, blow for a long time. These winds push additional sea400water into the lagoon through the three lagoon outlets, which add up to water flowing into the lagoon401as a result of tidal flows and rising water levels flood Venice’s streets and squares. The phenomenon of402high tides has been being systematically monitored in the municipality of Venice since the 1870s and403allows for modeling frequencies of different flood levels.404The frequency of flooding was estimated with an interpolation between the level quotas given by405the Municipality of Venice. The classification was made according to the frequency with which the406areas flooded. According to the flood frequency, six classes were set, from the worst case of three or407more floods per year, up to the best situation of less than one flooding in 100 years [60]. Figure 7 shows408the annual distribution of tides higher than 110 cm for the period between 1872 and 2016.409According to the experts’ judgments, properties in the historic center of Venice were first classified410according to the six urban districts (Sestieri). This partition had to be further differentiated into 15.411During the panel discussion, the touristic/residential vocation of the various areas of the center of412Venice was also explored.413To qualify each building regarding its accessibility from public transport, a spatial-temporal414relation was created assuming a speed of 4 km h−1 for pedestrians. Measuring the minimum distance415from each building to the public boat stops along public walkways (using the network of canals as a416barrier), a map of times needed to reach the nearest boat stop was created. This map was integrated417with the travelling times of public transport from each boat stop towards the main access points to418Venice from the mainland (the Piazzale Roma car park and S. Lucia railway station) at a speed of4197 km h−1. The result consisted of a layer of travel time from each building to the access point from420the mainland.421The judgment on the density of services available was based on the linear distance towards public422services (hospital and other sanitary services, schools and public administrations) in the historic center,423and on the density of private services (using a classification of density of commercial services made by424the local authority).425