1# RECORD ddfdca9d11843f81e9f91bacfe723acfa0d83bbb2# TITLE: Energy Retrofitting for the Modern Heritage Enhancement in Weak Real Estate Markets: The Olivetti Housing Stock in Ivrea3# YEAR: 2022.0 VENUE: Sustainability DOI: 10.3390/su140635074# AUTHORS: Alice Barreca; R. Curto; Giorgia Malavasi; D. Rolando5# PAGES: 25 CANDIDATE TABLE PAGES (1-based): [15, 16]6# ABSTRACT: The enhancement of Modern Heritage buildings is nowadays a challenging issue as they are often degraded or abandoned and their historical value is not well acknowledged by potential buyers and their owners. Moreover, they are usually energy inefficient and obsolete, but investments for energy retrofit interventions are not always convenient, especially in socio-economic contexts characterized by weak real-estate market dynamics. This paper aims to study the influence of intrinsic and extrinsic characteristics on housing prices and, in particular, to understand whether elements of building cultural connotations or some housing green features are monetized by the real estate market. The UNESCO site “Ivrea, industrial city of the 20th century”, located in the weak real-estate context of the Eporediese territorial area, was selected as an emblematic case study and two data samples of property listings were built to perform spatial regression analyses. The results showed that the green features of housing, such as the heating type and the EPC level, have a greater influence on property prices than those characteristics related to the cultural connotations of a building, such as the Olivettian context. Therefore, the current incentive-based policies for energy efficiency can represent great opportunities that can be exploited both to preserve and to improve the condition of this valuable Modern Heritage.789===== PAGE 1 (score 2, numeric-paren 0, table-words 0, var-words 2) =====101121 October 202612POLITECNICO DI TORINO13Repository ISTITUZIONALE14Energy Retrofitting for the Modern Heritage Enhancement in Weak Real Estate Markets: The Olivetti Housing Stock in15Ivrea / Barreca, A., Curto, R., Malavasi, G., Rolando, D.. - In: SUSTAINABILITY. - ISSN 2071-1050. - 14:6(2022), p.163507. [10.3390/su14063507]17Original18Energy Retrofitting for the Modern Heritage Enhancement in Weak Real Estate Markets: The Olivetti19Housing Stock in Ivrea20Publisher:21Published22DOI:10.3390/su1406350723Terms of use:24Publisher copyright25(Article begins on next page)26This article is made available under terms and conditions as specified in the corresponding bibliographic description in27the repository28Availability:29This version is available at: 11583/2959270 since: 2022-03-23T15:24:18Z30MDPI313233===== PAGE 15 (score 31, numeric-paren 0, table-words 5, var-words 15) =====3435Sustainability 2022, 14, 35073614 of 2437Table 4. DS SEM regression results. (Source: Authors’ elaboration).38Regression Model: Spatial Error Model (SEM)39Dependent Variable: Listing Prices Logarithm (logTLP)40Independent Variables41Coefficient42Std.Error43z-Value44Probability45LAMBDA460.321470.0456487.04349050CONSTANT5110.23520.07653134.07954055PT (Villa)560.381570.038589.90359060BC (Building Class-Economic)61−0.114620.03563−3.281640.00165BC (Building Class-Medium)66−0.065670.03168−2.079690.03770BC (Building Class-Noble)710.135720.047732.862740.00475FA (Floor area)760.004770.0017822.81679080TR (Terrace)810.105820.025834.12684085US (Unit Status-Partially renovated)860.271870.038887.10589090US (Unit Status-Renovated)910.466920.0429310.96294095EPC (Energy Performance Certification -B and over)960.162970.066982.45990.014100HT (Heating Type-independent)1010.1061020.0211035.0131040105E (Elevator)1060.1011070.0331083.0541090.002110BN (Bathrooms number)1110.9581120.0171135.7281140115HD (Distance from the nearest highway exit)116−1.42 × 10−0051175.628 × 10−005118−2.5171190.011120RE (Retail services)1210.2481220.0831232.9801240.002125Number of observations1261154127R squared1280.664129Log likelihood ratio test130−521.721310132Breush-Pagan test133103.8971340135AIC1361073.44137DIAGNOSTICS FOR SPATIAL DEPENDENCE FOR WEIGHT MATRIX138MI/DF139Value140Probability141Moran’s I1420.1161437.7231440145Lagrange Multiplier146114743.8851480149Robust LM15011515.7461520.016153Lagrange Multiplier154115553.5181560157Robust LM158115915.3791600.001161Lagrange Multiplier162216359.2641640165The analysis of the impact of the housing green features highlighted that high EPC166levels (EPC = B and over) and the heating type (HT = independent) significantly affect167LP, confirming that, as demonstrated in other studies [32,37,62], the energy efficiency is a168noteworthy feature with an increasing influence in the real estate market.169Finally, it is worth mentioning the influence of the extrinsic characteristics, such as the170presence of retail services (RE) and the distance from the nearest high-way exit (HD).1715.2. SS-Ivrea172Focusing on the city of Ivrea, the real estate market dynamics were analyzed in relation173to both some property features (including the “green” ones) and to the presence of the site174“Ivrea, industrial city of 20th century”. Therefore, the two variables “distance from the175UNESCO site (UD) and the “building Olivettian connotation” (OL) were considered in176the model since the UNESCO site will become a relevant point of interest for tourists and177citizens, even if the weak socioeconomic context may not foster this process.178Additionally, in this case, the logarithm of total listing price (logTLP) was used as179a dependent variable and a set of intrinsic and extrinsic features, selected by means of a180“foreword” stepwise method, were assumed as independent variables. Additionally, in this181case, the spatial autocorrelation tests indicated the Spatial Error Model (SEM) regression as182the most suitable model.183The regression results (Table 5) showed that this model is able to explain 62% of the184price variation (R squared = 0.620).185186187===== PAGE 16 (score 18, numeric-paren 0, table-words 3, var-words 7) =====188189Sustainability 2022, 14, 350719015 of 24191Table 5. SS SEM regression results. (Source: Authors’ elaboration).192Regression Model: Spatial Error Model (SEM)193Dependent Variable: Listing Prices Logarithm (logTLP)194Indipendent Variables195Coefficient196Std.Error197z-Value198Probability199Lambda (TLP)2000.2462010.12022.4512030.014204CONSTANT20510.5282060.10720798.022080209BC (Building Class-Noble)2100.352110.0782124.4542130214FA2150.0052160.00121712.8132180219US1 (Unit Status-Partially renovated)2200.2372210.0732223.2482230.001224US2 (Unit Status-Renovated)2250.3982260.0852274.682280229HT (Heating Type-independent)2300.1352310.0462322.9332330.003234PG (Private Garden)2350.1522360.0652372.3432380.019239OL (Building Olivettian Connotation)240−0.1742410.066242−2.6582430.008244UD (Distance from the UNESCO site)245−9.921 × 10−005246−3.49 × 10−005247−2.8452480.004249EPC (Energy Performance Certification-B and over)2500.3112510.142522.2192530.027254Number of observations255248256R squared2570.62258Log likelihood ratio test259−80.1692600261Breush-Pagan test262104.5242630264AIC265180.338266Diagnostics For Spatial Dependence For Weight Matrix267MI/DF268Value269Probability270Moran’s I2710.08312722.78922730.005274Lagrange Multiplier27512766.2062770.013278Robust LM27912801.4322810.231282Lagrange Multiplier28312845.4712850.019286Robust LM28712880.6972890.403290Lagrange Multiplier29122926.9032930.031294In line with the results obtained by the previous model, the building category (BC = noble),295the unit status (US = renovated), the heating type (HT = independent) and high EPC levels296(EPC = B and over) are characteristics that can considerably influence TLP.297Unlike the previous result, this regression analysis also highlighted that the presence298of a private garden (PG) positively affects TLP; in this regard, it is worth mentioning that299in Ivrea there are several villas and detached houses that include their own garden.300Focusing on the influence of the UNESCO site, it is interesting to notice that the301distance from the UNESCO site (UD) has a negative coefficient: thus, the greater the302distance from the site, the lower the property TLP. This is an important result which303confirms that this important and strategic point of interest could progressively increase its304positive influence on the Ivrea real estate market.305On the other hand, the variable “Olivettian connotation” (OL) of buildings presents306a negative coefficient, which means that the Olivettian connotation is not positively rec-307ognized by the real estate market. This result is in agreement with recent studies that308demonstrated that the architectural quality is not easily acknowledged by buyers and309sellers [8,63]. In Ivrea, in fact, even if the Olivettian housings often represent valuable310examples of Modern Heritage, they are classified as “economic” buildings and thus de-311valued. These buildings represent the legacy of the Olivetti company, but even if their312tangible and intangible value is relevant for architectural, social and political reasons, it313does not positively influence the property price variation. On the contrary, the building314category (BC), the unit status (UT) and some green features have a stronger influence on315property prices.316Therefore, it is possible to affirm that the green features (of which the EPC is a317proxy) are even more recognised by the real estate market than the architectural or318historical values.319320321===== PAGE 17 (score 12, numeric-paren 0, table-words 3, var-words 5) =====322323Sustainability 2022, 14, 350732416 of 243255.3. Prediction of Property Values after Energy Class Upgrade326In order to simulate a potential increase in the property price after a hypothetical327energy class upgrade deriving from possible energy retrofit intervention, the results of the328last regression were used to estimate new real estate market values of the asset collected329in the SS. This operation aimed not only to understand the potentiality of an energetic330optimization on the Modern Heritage, but also to evaluate if the costs of intervention could331be covered by the sale of the properties.332Firstly, only the listing price of the assets with the lowest EPC levels (E-F-G, 446 obser-333vations) were estimated; this operation was performed to compare the model predicted334values to the observed listing prices. The results highlighted an acceptable discrepancy in335the prediction of the listing price, which represents the error rate of the model.336The estimated property prices were calculated on the basis of the predicted values of337the SEM model calculated by GeoDa, assuming the simplified following formula [45,60,64]:338ELP (CEL) = (λTLP + 10.528)−1 (0.35 BC + 0.005 FA + 0.237 US1339+ 0.398 US2 + 0.152 PG −0.174 OL −9.921−005 UD).340where: ELP (CEL) is the predicted value of the Total Listing Price with the current EPC341level of the SEM model (Table 5) expressed by its logarithm, λ stands for added spatial342error coefficient; BC, FA, US1, US2, PG, OL, UD are the variables introduced for each of the343analyzed characteristics.344Subsequently, a second estimation was performed to understand whether an energy345retrofit intervention, which would change the lowest EPC levels (“E-F-G”) to the highest346ones (“B and over”), could lead to an increase in the listing prices. The energy upgraded347property price was estimated by changing the marginal coefficient related to the lowest EPC348levels with the one related to the highest EPC levels, according to the following formula:349ELP (ELU) = (λ TLP + 10.528)−1 (0.35 BC + 0.005 FA + 0.237 US1 + 0.398 US2350+ 0.152 PG −0.174 OL −9.921−005 UD + 0.311 EPC)351where: ELP (ELU) is the predicted value of the Total Listing Price with the EPC level352upgraded of the SEM model (Table 5) expressed by its logarithm, λ stands for added spatial353error coefficient; BC, FA, US1, US2, PG, OL, UD are the variables introduced for each of354the analyzed characteristics, with the addition of the calculated marginal coefficient of355upgraded EPC level (“B and over”).356Therefore, the results deriving from the two estimations were compared. Figure 6357shows the comparison between the first estimation with Total Listing Prices at the current358energy level and the same cases but with a hypothetical upgrade in energy class maybe due359to a general intervention such as walls and roof insulation and the replacement of windows.360The difference between those two estimated prices is positive for 93% of the observations,361thereby providing evidence of the cost-effectiveness of retrofitting houses, and it can be362considered as a starting point for the conservation of these residential buildings, especially363those located in the Olivetti’s residential districts (Figure 6).364Initial evidence can be found in Canton Vesco; by looking at the maps, it is possible to365see the retrofit intervention effects on the listing prices. In Figure 6a, values of this district366fall into the lowest Jenks range, while in the map in Figure 6b the estimated listing prices367fall into the highest price range. In the Historical Centre of the city and in the UNESCO368Site, prices rise from a mean of 110,000 Euro/sqm to 118,000 Euro/sqm and by observing369the map it is possible to see that prices increase not so much to the higher range but to the370intermediate one. In particular, the number of apartments with the lowest observed listing371prices is 191, while the number of those with an ECP level from B is 176. Furthermore,372there is also a considerable increase in the dwellings with the highest values, i.e., in the first373image only one case is observed, but in the second there are 34.374