1# RECORD a855c4aba93669899f6285712fd4b23cfb3500bb2# TITLE: Estimation of a Hedonic House Price Model with Bargaining: Evidence from the Italian Housing Market3# YEAR: 2013.0 VENUE: Aestimum DOI: 10.13128/aestimum-131234# AUTHORS: Mauro Iacobini; G. Lisi5# PAGES: 14 CANDIDATE TABLE PAGES (1-based): [11]6# ABSTRACT: This empirical paper tests the role of bargaining in the formation process of housing prices in Italy. Housing markets are “thin”, local and decentralized, and thus buyers and sellers may have some market power. Hence, the selling price is influenced both by the characteristics of the product as well as by the bargaining power of the buyers and sellers. Furthermore, the bargaining power of the seller (buyer) can also be viewed as the cost of incomplete information imposed on the buyer (seller). The empirical results derived from multiple regression analysis support our theoretical assumptions. In fact, the variables created as proxies of bargaining power of the parties, and incorporated into the hedonic price function, are statistically significant and help to improve the performance of the hedonic model, thus reducing the differences between predicted and observed selling prices.789===== PAGE 1 (score 2, numeric-paren 0, table-words 0, var-words 2) =====1011© Firenze University Press 12www.fupress.com/ceset13XLI Incontro di Studio del Ce.S.E.T.: 41-5414Estimation of a Hedonic House Price 15Model with Bargaining: Evidence from 16the Italian Housing Market*17This empirical paper tests the role of bargaining in the formation pro-18cess of housing prices in Italy. Housing markets are “thin”, local and de-19centralized, and thus buyers and sellers may have some market power. 20Hence, the selling price is influenced both by the characteristics of the 21product as well as by the bargaining power of the buyers and sellers. 22Furthermore, the bargaining power of the seller (buyer) can also be 23viewed as the cost of incomplete information imposed on the buyer 24(seller). The empirical results derived from multiple regression analysis 25support our theoretical assumptions. In fact, the variables created as proxies 26of bargaining power of the parties, and incorporated into the hedonic price 27function, are statistically significant and help to improve the performance of 28the hedonic model, thus reducing the differences between predicted and ob-29served selling prices.30Mauro Iacobini 31Gaetano Lisi32Territorial Agency, Rome33E-mail: mauro.iacobini@34agenziaterritorio.it 35E-mail: gaetano.lisi@36agenziaterritorio.it 37Key words: Hedonic price 38theory, bargaining power, 39asymmetric and incomplete 40information, multiple 41regression analysis421. Introduction43The hedonic price theory, traced back to the papers of Lancaster (1966), Rosen 44(1974), and Epple (1987), is the theory of reference for the housing market. Ac-45cording to this theory, the price of a composite good, such as a home, depends on 46its intrinsic and extrinsic characteristics, whose prices (being implicit or hedonic) 47are not readily observable but can be revealed to economic agents through the ob-48served selling prices of houses and the specific amounts of attributes associated 49with them (for an exhaustive overview see Sheppard, 1999 and Malpezzi, 2003). 50Hence, in the standard hedonic price model, only the heterogeneous nature of 51real estate goods is taken into account, since the selling price is a function of the 52housing characteristics.53According to the price dispersion literature (for a review see Leung, Leong 54and Wong, 2006), an important part of price volatility cannot be attributed to the 55heterogeneous nature of real estate goods. In fact, two similar houses (namely, 56two houses with the same attributes and located close to each other) can be val-57ued differently at the same time (Maury and Tripier, 2010). Thus, remaining price 58differentials may be caused by the heterogeneity of the parties (tastes, preferences, 59patience, search costs, asymmetric information, etc.).160* This work is connected to the experimentation activity (year 2011) conducted by the Territorial 61Agency, Head Office of Real Estate Market Observatory and Appraisal Services, Real Estate 626364===== PAGE 11 (score 20, numeric-paren 0, table-words 7, var-words 3) =====6566Estimation of a Hedonic House Price Model with Bargaining: Evidence from the Italian… 6751685.2. Marginal prices69In equation [2a] and [2b], the marginal prices implicitly reflect the effects of 70bargaining, in the sense that the estimated coefficients are different (or equal by co-71incidence) to those predicted in the absence of Γ and Ω. This consideration is very 72important for property valuation where estimation of the marginal value of an at-73tribute is commonly used to adjust comparable sales data. Therefore, if the estimat-74ed regression coefficients are used as marginal prices in real estate appraisals (for 75example, by the Sales Comparison Approach), the price correction will be different us-76ing the standard hedonic model rather than the hedonic model with bargaining.77Indeed, as regards the variables “location” and “state of building” and with ref-78erence to the city of Taranto, a somewhat surprising result emerges: the hedonic 79prices are systematically overestimated in the standard hedonic model (see Table 3).80Table 3. 81a) “Location”82City – zone OMI83Model with bargaining84Model without bargaining85Marginal price86p-value87Marginal price88p-value89Taranto – B190€ 9.603,85 910.001 92€ 12.766,47 930.002 94Taranto – C195€ 11.463,41 960.000 97€ 17.416,52980.00099Taranto – D1100€ 20.538,28 1010.000 102€ 29.548,34 1030.000 104Taranto – E2105€ 5.311,70 1060.113 (*)107€ 7.352,90 1080.060 109b) “State of building”110City – zone OMI111Model with bargaining112Model without bargaining113Marginal price114p-value115Marginal price116p-value117Taranto – B1118€ 8.340,49 1190.000 120€ 10.216,55 1210.002122Taranto – C1123€ 6.959,62 1240.248 (*)125€ 12.507,79 1260.108 (*)127Taranto – D1128€ 15.764,46 1290.000 130€ 17.403,33 1310.000 132Taranto – E2 133€ 14.109,49 1340.008 135€ 28.117,67 1360.000 137(*) It means that the marginal price is not significant at the confidence level of 10%.138As regards the estimated marginal prices, with the exception of the floor area, 139the very significant characteristics vary depending on the city and area consid-140ered, thus confirming the “local” character of housing markets (see Table 4). Fur-141thermore, note that the prevailing orientation is the only variable that is never sta-142tistically significant.143144145===== PAGE 12 (score 15, numeric-paren 0, table-words 1, var-words 13) =====14614752 148M. Iacobini, G. Lisi149Table 4. Hedonic prices.150City – zone OMI151Characteristics statistically significant (level of confidence 10%)152Alessandria – B11535: Floor area, Number of balconies, New construction, Architectural style, 154Quality of view155Cosenza – B1 1562: Floor area, Number of bathrooms 157Cosenza – D11583: Floor area, Architectural style, Quality of view159Crotone – D11604: Floor area, Location, Architectural style, Number of balconies161Genova – D431624: Floor area, Location, Floor level with elevator, Number of balconies163Taranto – B11646: Floor area, Location, New construction, Quality of view, State of 165building, Floor level with elevator166Taranto – C11675: Floor area, Number of bathrooms, Number of balconies, Location, 168Quality of view169Taranto – D11704: Floor area, Location, New construction, State of building171Taranto – E21726: Floor area, Number of bathrooms, Number of balconies, New 173construction, Architectural style, State of building174Venezia Mestre – E23 3: Floor area, Quality of view, Number of bathrooms175Vercelli B11767: Floor area, Location, Floor level with elevator, Number of balconies, 177Architectural style, State of building, Location178Vercelli B21792: Floor area, State of building1806. Conclusions181Bargaining is an important part of the price formation process for heteroge-182neous goods, such as a home. In fact, housing markets are not perfectly com-183petitive and consequently the “true” market price is not readily observable. Un-184der these conditions, the selling price is influenced both by the characteristics of 185the product as well as by the heterogeneity of the parties. Indeed, the bargaining 186power of the buyers and sellers is a good proxy of the heterogeneity of the par-187ties. Furthermore, the bargaining power of the seller (buyer) can also be viewed as 188the “tax” or cost imposed on the buyer (seller) for having incomplete information.189In this paper we find empirical evidence for our theoretical assumptions in the 190Italian housing market. This result has two main consequences:191• 192First, ignoring bargaining may lead to omitted variable bias on estimated implicit 193prices in hedonic pricing models. This question is significant for property valua-194tion where estimation of the marginal value of an attribute is commonly used to 195adjust comparable sales data.196• 197Second, bargaining helps improve the performance of the hedonic model, thus 198reducing the differences between predicted and observed selling prices. This is-199sue is significant in making correct predictions regarding housing sale prices.200Furthermore, this model may be useful for selecting the comparables most 201suitable for the Sales Comparison Approach. In fact, it is quite clear that the resi-202dential property affected by the bargaining power should be eliminated.203