THE VALUE OF BUILDING SAFETY: A HEDONIC PRICE APPROACH

Transcription

1 THE VALUE OF BUILDING SAFETY: A HEDONIC PRICE APPROACH by Yung YAU * Department of Public Policy City Univerity of Hong Kong The Hong Kong Special Adminitrative Region People Republic of China Paper ubmitted to ARES 19 th International Conference Gold Coat, Autralia, July 2014 Abtract Theoretical and empirical tudie on how building quality or performance i valued by the property market abound in the literature. While ome of them invetigate the change in property price after building renovation, little ha been done on the pricing of afety performance of building. In thi regard, thi preliminary reearch aim to explore if reidential propertie in afer building command higher market value in Hong Kong. Hong Kong provide a good laboratory for thi tudy becaue it i a denely-populated city where high-rie building are common. In uch an environment, building failure can poe a eriou threat. A pate of building-related accident in the city in recent year ha demontrated the painful conequence of neglect of building afety. For the purpoe of thi tudy, the afety performance of a building i meaured by the weighted number of unauthorized building work (UBW) preent on the external wall of the building. By their nature, UBW are building work that are contructed without prior approval and conent from the government. A hedonic price model i developed for aeing the market value of building afety. For the model etimation, apart from the property tranaction data, the number of unauthorized appendage (i.e., UBW attached to the building facade) in each building under tudy i obtained through a building urvey. Baed on the analyi reult, everal hypothee built upon the theorie of elf-protection and elf-inurance put forward by Ehrlich and Becker (1972) are teted. Keyword: Building afety; hedonic pricing model; elf-protection; elf-inurance; unauthorized building work * For comment and enquirie, pleae contact Dr. Yung Yau at y.yau@cityu.edu.hk or Department of Public Policy, City Univerity of Hong Kong, Tat Chee Avenue, Kowloon Tong, Kowloon, Hong Kong

2 THE VALUE OF BUILDING SAFETY: A HEDONIC PRICE APPROACH INTRODUCTION A afe home i a good home, a the aying goe. Safety i defined by the World Health Organization (1998: 6) a a tate in which hazard and condition leading to phyical, pychological or material harm are controlled in order to preerve the health and well-being of individual and the community. From an etymological perpective, afety i about the wholene of phyical life. It i becaue alvu and olwo, the Latin and Proto-Indo-European origin of word afe, mean uninjured and healthy and whole repectively (Nilen et al., 2004). Therefore, it i enible for cholar like Thygeron (1977) and Hunter (1992) to define afety a a tate free from hazard or danger. With reference to thee definition, Yau et al. (2008: 503) interpreted building afety a the achievement of a building in afeguarding it occupant and the general public from the harm originated from the built environment, which in turn reduce injurie and death. To perform it function, a building hould be afe in many ene. For example, the building hould be tructurally table and fire-reitant, and the ecape route are free from any obtruction. In order to enure the afety of the building tock, government uually adopt two approache. The firt approach i building control which make ure the afety level of all building in a city meet the minimum acceptable tandard through trict monitoring and enforcement of building code or regulation (Crook & Hughe, 2001). For intance, building authoritie enforce againt propertie in direpair and illegal building work (Hatti, 1981; Yiu & Yau, 2005). On the other hand, ubidization i the econd mean of tate intervention. Grant and loan have been offered to property owner for the improvement (including maintenance and rehabilitation) of their propertie around the world (e.g. Boyne et al., 1991, Scanlon, 2010; Whalley, 1988; Yau et al., 2013). A a matter of fact, apart from law enforcement and ubidization, market force may help motivate property owner to keep their building afe (Yau, 2009). In theory, afer propertie hould have higher value, keeping other thing contant. There i a large volume of empirical literature on how an overall change in the building performance or condition (e.g. by mean of refurbihment and renovation) i reflected in property price or rental level (Chau et al. 2003; Forte & McCarthy, 2010; Hui et al., 2008) but little ha been done pecifically on the pricing of afety performance of building. In thi regard, thi preliminary reearch aim to explore if reidential propertie in afer building command higher market value in Hong Kong. For the purpoe of thi tudy, the afety performance of a building i meaured by the weighted number of unauthorized appendage preent on the external wall of the building. A hedonic price analyi i conducted for aeing the market value of building afety. Baed on the analyi reult, three hypothee built upon the theorie of elf-protection and elf-inurance put forward by Ehrlich and Becker (1972) are teted. Hong Kong provide a good laboratory for thi tudy becaue it i a denely-populated city where high-rie building are common. In uch an environment, building failure can poe a eriou threat. A number of terrifying building-related accident in the city in recent year, including the udden collape of a 55-year-old apartment building in To Kwa Wan, Hong Kong which claimed four live in January 2010 (Building Department, 2010). Thee accident have vividly demontrated the painful conequence of neglect of building afety. A for the afety performance of a building, it i proxied by the number of unauthorized appendage which are - 2 -

3 eentially illegal or unauthorized building work (UBW) for two major reaon. Firt, the problem of UBW proliferation ha attracted much public attention in Hong Kong particularly ince many top government official and lawmaker found themelve embroiled in candal of illegal tructure in 2011 and 2012 (Foo, 2012; Luk, 2012; Ma, 2011). Second, unauthorized appendage are the mot eaily obervable among all type of UBW. Thi article i organized a follow. Literature on the quality-price link in the property market and elf-protection and elf-inurance theorie will be firt reviewed, followed by the outline of hypothee and development of analytical model. The data ource will then be detailed. What come next are the analyi reult. The implication of the analyi reult will be dicued before the preliminary tudy i concluded and an agenda for further reearch i uggeted. LITERATURE REVIEW Building Quality a a Determinant of Property Value Building quality or performance ha been regarded a an important determinant of property value. Baum (1991, 1994) and Ho (2000) howed that commercial building of better quality brought higher return to the owner. In the reidential property ector, Feijtem and Mulder (2005) aerted that quality of a dwelling can be ideally reflected in it value or price. In the broadet ene, the quality of a building embrace all the attribute related to the building. A urvey of the empirical literature (e.g. Chau, et al., 2001; Mok, 1995; Mok et al., 1995; So et al., 1997; Te & Love, 2000; Te et al., 1997; Yau et al., 2008, 2009) ugget that apart from the time factor, the value of a reidential property i determined by it tructural characteritic (e.g. age, ize and floor level), locational characteritic (e.g. acceibility to public tranport) and external environment (e.g. view and proximity to park). Nonethele, previou tudie inclined to reearch the impact of building deign and environmental factor on property price. Thee factor are motly intrinic or not eaily adjutable by the property owner, particularly after a property i in ue. Building quality attribute like building condition and preence of illegal tructure, which are more eaily adjutable or manageable by the property owner, have been have not attracted much attention from the academic. In point of fact, imilar to the deign and environmental characteritic, building condition have often been thought of being influential to property price. Without any empirical upport, Aren (1997) analytically argued that defective propertie (e.g. thoe with ignificant phyical problem like cracked lab or uboil ubidence) hould be valued at a dicount in view of the potential cot of remediation, higher vacancie and extra inurance premium. Kain and Quigley (1970) empirically evidenced that propertie in better condition (in term of interior and external phyical environment) were old at higher price. Similar finding were returned from other tudie like Bouraa and Peng (1999), Bulter (1982) and Jimenez (1983) and which found that value wa added to propertie with better building condition. On the other hand, Murdoch et al. (1993) found that the preence of ubtandard tructural item in a building uppreed property price. In another tream of reearch, property value i evidenced to change with building condition a a reult of an improvement project. For example, Chau et al. (2003) carried out a hedonic price analyi to tudy how a refurbihment project affected property value in Hong Kong. Their reearch howed an approximately 9% rie in property value brought about by the refurbihment, implying that there wa a ignificant poitive relationhip between building quality and property value. Another local tudy by Hui et al. (2008) uncovered that building - 3 -

4 rehabilitation reulted in an average of 35.6% enhancement of property value. In New Zealand, property price wa alo found to increae by 1-8% after home improvement (Forte & McCarthy, 2010). A a whole, although empirical tudie on relationhip between building quality and property value abound, mot of them are not pecifically related to the afety performance of building. The improvement project invetigated by Chau et al. (2003) and Hui et al. (2008) were aociated with upgrading of afety, hygienic and aethetical quality of the building at the ame time. Thee previou tudie do not help u know whether the houing market value propertie in a afer building with a premium. Theorie of Self-inurance and Self-protection The price-afety connexion of houing can be conceptualized baed on the theorie of elfinurance and elf-protection put forward by Ehrlich and Becker (1972). Thee two theorie frame how rational individual make choice among different action when facing rik. Very often, individual participate in an array of riky activitie that may jeopardize their own afety (Blomquit, 2004). Driving a car and cycling are typical example of thee riky activitie. Engaging in thee activitie, an individual may rik injury or property damage o he can take out inurance againt the potential loe. Alternatively, the individual can chooe to drive hi car or ride on a bicycle lowly and carefully in order to reduce the chance of accident, or recoure can be made in the form of protective meaure like eat belt and afety helmet to reduce loe in cae of accident. Choice among market inurance, elf-inurance and elf-protection In their claic paper, Ehrlich and Becker (1972) devied a tate-preference approach to explain an individual choice and behaviour under uncertainty by combining the indifference curve and expected utility analye. In their premie, there are two tate of the world good tate (or well-endowed tate) and bad tate (or le well-endowed tate). Example of the bad tate include outbreak of fire, earthquake and many other man-made or natural diater. Facing a propective lo in a bad tate, an individual can either inure againt the lo, or take tep to lower the likelihood that the lo will occur. In thi ene, individual are required to determine their optimal expenditure on a et of alternative intrument, namely market inurance, elfinurance, and elf-protection. In Ehrlich and Becker language, elf-inurance refer to effort to reduce the ize of propective loe from bad tate, given the probability of ditribution of the correponding bad tate. Self-protection, in contrat, refer to effort to reduce the probabilitie of bad tate given the magnitude of the correponding propective loe. 1 In the characterization by Ehrlich and Becker (1972), market-inurance and elf-inurance are imilar. They aim to both lower ize of lo in bad tate by tranferring an individual income from good tate to bad tate. By chooing not to inure through a market or elf-inure, the 1 Ehrlich and Becker (1972: 634) admitted that it wa omewhat artificial to ditinguih behaviour that reduce the probability of the lo from behaviour that reduce the ize of a lo ince many action do both. Thi ditinction i often blurred in actual practice. However, it remain a ueful theoretical ditinction, given the contrating reult in the model of elf-inurance and elf-protection

5 individual ha to bear by himelf the loe from any bad tate that realize. Sometime, thi i the only option available becaue the intrument for market inurance and elf-inurance are not acceible. Market inurance, if available, can be purchaed at a price which i uually called the premium. What make elf-inurance different from market inurance i the abence of an inurance market for elf-inurance. An explicit price for elf-inurance hence doe not exit. Nonethele, the price of elf-inurance can be imputed to the cot incurred in elf-inuring by the individual. Other than pricing, the mechanim of rik pooling alo mark a ignificant difference between the two type of inurance. Rik i pooled acro different individual in market inurance but not in elf-inurance. Unlike the two inurance option, elf-protection doe not involve any income tranfer from good to bad tate. It target for the reduction of the probabilitie of bad tate rather than lowering the ize of lo in bad tate. In a world where inurance market do not exit, individual can only reort to elf-inurance and elf-protection to mitigate any propective loe. If market inurance i available, it i a cloe ubtitute for elf-inurance a long a the premium of market inurance i independent of the extent of elf-inurance taken (Ehrlich & Becker, 1972). The take-up of elf-inurance will drop with an increaing availability of actuarially-fair priced market inurance. 2 For mitigating a rare or low-probability lo, market inurance i preferred to elf-inurance becaue the premium of market inurance decreae with the probability of lo while the cot incurred or the implicit price of elf-inurance do not (Ehrlich & Becker, 1972). In other word, to achieve the ame level of reduction in the probability of a rare lo, the cot or the amount of effort paid for elfinurance generally exceed market inurance premium. Therefore, the low cot-effectivene will dicourage people from elf-inuring againt rare loe. In other word, elf-inurance i ubject to crowding out (Simmon et al., 2002). Thi view i empirically upported by Frontin and Holtmann (1994). Contrariwie, market inurance and elf-protection complement each other provided that the former i available in the market at an actuarially fair rate (Ehrlich & Becker, 1972). The rationale for thi complementary relationhip i traightforward. In order to reduce the rik of moral hazard, the premium of market inurance hould reflect the amount of elfprotection effort taken by individual to reduce the probability of lo. Any effort of elfprotection perceived to be effective i recompened by the market in the form of lower premium. Following thi line of thought, the coverage of market inurance i not necearily negatively correlated with the amount of effort paid for elf-protection by individual. 3 Neverthele, other literature illutrated that the premium of market inurance might not repond to the amount of elf-protection effort due to information aymmetrie (Arnott & Stiglitz, 1988; Ligon & Thitle, 1996) or market failure (Arnould & Grabowki, 1981; Kunreuther & Kleffner, 1992). Beide, the analytical model of Ehrlich and Becker (1972) doe not thoroughly explain the interaction between elf-inurance and elf-protection under different condition regarding the availability of market inurance (Boyer & Dionne, 1983). Boyer and Dionne (1983) argued that in the abence of market inurance, elf-inurance hould be favoured over elf-protection by rik-avere individual. However, thi propoition wa devied baed on the aumption that the 2 The price or premium of market inurance i conidered actuarially fair if the probabilitie of the loe concerned are fully conidered in the pricing proce. 3 However, a alerted by Ehrlich and Becker (1972), the complementary relationhip between market inurance and elf-protection depend critically on the reponivene of the premium of market inurance to the amount of elfprotection undertaken by individual. In addition, the effort paid for elf-protection hould be obervable to market participant

6 two option brought equal reduction in the expected loe and were equally cotly. A correctly pointed out by Chang and Ehrlich (1985: 577), thee two mitigation option might be varied continuouly and choen optimally. By relaxing the aumption aforementioned, Chang and Ehrlich (1985) derived that elf-inurance wa not alway preferred to elf-protection in the abence of market inurance. Moreover, they analytically howed that the ratio of the former to the latter did not necearily exceed unity. Self-inurance, elf-protection and houe value A large volume of empirical reearch ha attempted to evaluate the market value of variou elfinurance and elf-protection meaure. Literature uch a Anderon (2005) and Boulding and Purohit (1996) tudied how elf-inurance meaure affected the price of a wide range of ervice and product like automobile. In the real etate market, elf-inurance uually appear in the form of mitigation meaure which reduce the loe of life and property in natural hazard. For example, houing in earthquake-prone area can be deigned and contructed to become earthquake-reitant. Uing the contingent valuation method, Willi and Agary (1997) found that the price of earthquake-reitant houe in Iran were ignificantly higher than thoe for nonreitant houe. Furthermore, Simmon et al. (2002) etimated the market value for hurricane blind for beachfront building on the Gulf Coat in the United State. On average, a premium of US$4,000 wa added to the value of a houe equipped with hurricane blind. Simmon and Sutter (2007) alo unveiled that a tornado helter increaed the ale price of a home by approximately US$4,200 in Oklahoma City, United State. A for elf-protection, everal hedonic tudie examined the value of reducing the probability of a lo from a natural hazard by relocating out of harm way. Brookhire et al. (1985) tudied the houing market of Lo Angele and San Francico and potted that home located outide earthquake-prone area were old at a premium. Shilling et al. (1985) found that home located outide a floodplain in Baton Rouge in the US were old at higher price than thoe within the floodplain. Donnelly (1989), MacDonald et al. (1987) and Speyrer and Raga (1991) conducted imilar tudie in other part of the US. Their reult confirmed the finding of Shilling et al. (1985). Finding in other tudie alo manifeted that elf-protection mitigation were priced in property tranaction. Price of houe were higher when they were ited beyond the influence of potentially hazardou facilitie uch a nuclear plant (Gamble &Downing, 1982), chemical plant (Carrol et al., 1996), landfill (Nelon et al., 1992) and ga pipe (Kak and Maani 1992). 4 Sometime the hazard aociated with the living environment are not known until relevant information i dicloed by ome other partie. It wa indicated by Bernknopf et al. (1990) that the announcement of hazard regarding earthquake and volcanic activitie in the Mammoth Lake area in the United State depreed houe price in that area. Comparable reult were obtained by Montz (1993) who focued on the dicloure of flooding rik in New Zealand. The houe price differential detected in thee tudie provide additional evidence for the value of elf-protection mitigation in property market. 4 One hould be mindful that locating a reidence in the face of hazard i not alway elf-protective behaviour. For example, locating a reidence farther away from the horeline i a mean of elf-inurance on account of the fact that the ize of the flood and torm damage i larger in the coatal area. The chance of a flood or torm i not reduced by uch a location-choice deciion. On the other hand, it i a elf-protection if a houe i ited outide an earthquake-prone zone becaue the probability of an earthquake in the region where the houe i located i ignificantly lower. In many cae, however, the ditinction between thee two alternative i ubtle

7 While empirical reearch applying the theorie of elf-inurance and elf-protection to property market abound, nearly all focued on external hazard including natural hazard (e.g. flood, hurricane and earthquake) and technological hazard (e.g. nuclear plant and wate torage or treatment facilitie). Internal hazard aociated with a property like the afety performance of a building uch a fire hazard and tructural failure have been largely ignored. Moreover, previou tudie predominately focued on low-rie, particularly ingle-family houe; little reearch worked on high-rie like apartment building. HYPOTHESES AND ANALYTICAL MODEL Proliferation of Unauthorized Appendage a Building Safety Performance To tudy how afety performance of a building i valued by property market, thi reearch take the degree of proliferation of unauthorized appendage in the building a the meaurement of the afety performance of the building. The unauthorized appendage are commonly found in Hong Kong and include unauthorized cage, drying rack, flower rack, lightweight canopie, and airconditioner upporting frame that are attached to the external wall of building. Figure 1 3 how ome example of unauthorized appendage in apartment building in the city. Thee building work are unauthorized or illegal becaue they were contructed without any prior approval of the building plan or conent to work commencement a per the requirement tipulated under the Building Ordinance (Ho et al., 2008). The degree of proliferation of authorized appendage i ued to proxy building afety performance for both practical and academic reaon. Practically peaking, a attached on the external wall of building, unauthorized appendage are very eaily identified compared with other type of UBW. Beide, unlike tructural tability and fire afety the evaluation of which uually involve ophiticated teting or aement, the appraial of authorized appendage proliferation in a building i traightforward and relatively le cotly. Academically peaking, employing the degree of UBW proliferation a a meaure of building afety performance can facilitate the valuation of elfprotection mitigation and elf-inurance mitigation in afeguarding a building afety in the abence of any influence by market inurance. It i imply becaue property loe, peronal injurie and death caued by UBW are not covered by the property-all-rik or third-partyliability inurance policie for building in Hong Kong. Therefore, the effect of market inurance can be ignored a long a unauthorized appendage are tudied. For increaing the amount of uable pace or amenitie for the building uer, unauthorized appendage are relatively eaier to contruct compared with other type of UBW (Ho et al., 2008). Yet, uch UBW jeopardize building afety becaue their tructural oundne i uncertain and they affect approved building work on or near which they are built (Choy, 1998). Proliferation of unauthorized appendage may alo affect fire afety of a building adverely becaue the UBW block firemen acce to the building facade. From the perpective of health rik, unauthorized appendage often block natural light and ventilation from entering the building. A added by Chan (2000), the protruding unauthorized tructure might indirectly aggravate the problem of building decay becaue they make repair and maintenance of external wall more difficult. When an unauthorized appendage fail, it may bring about caualtie, property loe, and ocial cot (e.g. hopitalization and legal cot). In fact, fatal accident involving UBW are not rare in Hong Kong. A reported by Leung and Yiu (2004), there were 21 death and 135 injurie inflicted by UBW-related accident in Hong Kong during the period - 7 -

8 between January 1990 and December Several court judgment have etablihed that all coowner of a multi-owned building are liable for the caualtie and property loe caued by the failure of unauthorized appendage in the building.5 Figure 1: Solid extenion Figure 2: Light-weight canopie Figure 3: Metal cage Unauthorized appendage poe afety hazard of variou degree on building occupant and the public. Some are more harmful than other. From the public authority perpective, thoe 5 For example, Aberdeen Winner Invetment Co. Ltd. v. The Incorporated Owner of Albert Houe and Another HCA3408/

9 unauthorized appendage that are more hazardou hould be accorded more attention. The Building Department (2005) categorized unauthorized appendage under two head actionable and non-actionable UBW. The former were thought to bear an imminent danger to occupant and the public o they hould warrant priority removal. According to the Building Department (2005), thee high-rik unauthorized appendage included: (i) (ii) (iii) (iv) (v) (vi) (vii) cage, flower rack, tructure, or canopie of olid contruction on the external wall, re-entrant, and lightwell, irrepective of the extent of their projection; dilapidated canopie and advertiement ign on the external wall, re-entrant, and lightwell, irrepective of the extent of their projection; dilapidated or abandoned air-conditioning unit upporting frame, metal frame, or chimney cage, flower rack, tructure, or canopie of olid contruction on the external wall, re-entrant, and lightwell, irrepective of the extent of their projection; lightweight canopie projecting more than 500mm from the external wall; air-conditioning unit upporting frame projecting more than 600mm from the external wall; tructure on or attached to approved canopie, or attached to approved balconie; rooftop and flat roof tructure with projection; (viii) UBW of two torey or more; and (ix) UBW built on top of another UBW. On the other hand, the non-actionable unauthorized appendage were thoe poing a relatively limited degree of hazard on occupant and the public. Common example of thee are drying rack and lightweight canopie that project le than 500mm from the external wall. Owing to the comparatively lower rik aociated and the amenity offered to building uer, thee UBW have been tolerated by the Building Department. Following the Building Department (2005) dichotomy of unauthorized appendage, thi tudy meaure the degree of proliferation of unauthorized appendage by taking a weighted average of the number of actionable and nonactionable unauthorized appendage per dwelling unit in a building. Mathematically, Degree of proliferation = 1 N non-actionable + 5 N actionable (1) where N non-actionable and N actionable denote the number of non-actionable and actionable unauthorized appendage repectively. A heavier weighting i placed on the number of actionable unauthorized appendage becaue of the higher rik aociated with thi type of UBW. Hypothee for Empirical Teting Baed on Ehrlich and Becker (1972) theory of elf-protection, in cae afety rik created by UBW are not covered by market inurance, a rational individual hould pay le to buy a property in a building with a higher degree of proliferation of unauthorized appendage becaue the probability of a building failure aociated with unauthorized appendage i lower. Given the ame degree of proliferation, according to Ehrlich and Becker (1972) theory of elf-inurance, - 9 -

10 a rational individual hould pay le to buy a property in a building abutting on one or more buy treet. It i becaue when an unauthorized appendage fail and fall onto the treet, the damage will be maller. Founded on thee theoretical prediction, two hypothee are developed for empirical teting in thi tudy:- H1: Propertie in a building with a higher degree of proliferation of unauthorized appendage are old at a dicount, keeping other thing contant. H2: Propertie in a building with unauthorized appendage and abutting on one or more buy treet are old at a dicount, keeping other thing contant. Model for Analytical Analyi To tet the two hypothee above, an analytical model i broadly pecified a follow: PRICE = f (S, L, T, U). (2) In Equation (2), the ale price of a reidential property, PRICE, i taken a a function (f) of four vector of determinant, namely S, L, T and U. The vector S contain tructural characteritic of the property, including building age and floor area of the property. A for the vector L, it repreent the locational factor of property, including the vertical location of the property in a building and the ditance of the building from the nearet ma tranit tation. T i a vector of time dummie which indicate the date of tranaction (in month) of the property, for controlling the time effect on property price. The focu of thi tudy i placed on the vector U which encompae factor related to rik aociated with the proliferation of unauthorized appendage. The operationalized independent variable under different vector are reported in Table 1. Since the pecification of the analytical model i not known a priori in the abence of theoretical upport, a emi-log functional form i ued. Thi functional form i choen for the practical reaon becaue when there i the poibility of an omitted variable bia, emi-log model outperform other functional form (Cropper et al., 1988). Quadratic term are alo included in the model to cater for the non-monotonou effect of the continuou control variable on the dependent variable. The reultant hedonic price model etablihed for thi tudy to etimate the value of afety performance of a building i ln PRICE t = α + α AGE 0 + α AGE α AREA + α AREA 5 + α FLOOR + α FLOOR + β UNAPP + β UNAPP BSTREET T t=1 t 2 + α MTRD + α MTRD γ TIME where PRICE t denote the tranaction price of property at time t; TIME t i a vector of monthly time dummie; α, β and γ are coefficient to be etimated; and ε i the tochatic term. t + ε (3)

11 Table 1: Decription of the independent variable ued in the hedonic price model Symbol Unit Decription AGE month The age of the building, which equal the difference between the date of the iue of the occupation permit and the date of the tranaction FLOOR - The floor level of the tranacted property AREA quare feet The uable floor area of the tranacted property MTRD metre The ditance between the tranacted property and the nearet Ma Tranit Railway (MTR) tation UNAPP - The weighted average of the number of actionable and nonactionable unauthorized appendage per dwelling unit in the building in which the tranacted property i located BSTREET - A dummy variable that equal 1 when the property i located in a building abutting on one or more buy treet and zero if otherwie For the purpoe of thi tudy, a buy treet i defined a a treet with a high pedetrian flow or high traffic flow. More pecifically, if a treet i claified a a buy treet if it ha four or more traffic lane or a treet market or imilar feature i found on the treet. Thi characteriation delineate building that are more hazardou in cae of failure of any unauthorized appendage. DATA: SOURCES AND DESCRIPTIONS To etimate the empirical model, tranaction data of reidential propertie in 64 development in the old ditrict of Tai Po, Hong Kong were employed. Thi geographical area wa elected becaue the building tock in the area wa rather table. There were not extenive development or redevelopment project undergoing in the area in recent three year. Beide, building with different age and configuration (e.g. ingle tenement block and high-rie apartment building) were available in the area. The data for property tranaction dated between 1 July 2013 and 31 December 2013 were obtained from the Economic Property Reearch Centre. 6 A tudy window of ix month wa choen a a reult of a balance truck between two different force. On one hand, a wider tudy window entail more data point for more meaningful model etimation. On the other hand, the degree of unauthorized appendage proliferation in the ubject building vary over time, which make the empirical invetigation more complication. The longer i the time horizon being conidered, the more likely will be the change in the degree. It i aumed that the evaluated degree of unauthorized appendage proliferation in the ubject building held for the whole tudy period between 1 July 2013 and 31 December 2013, 6 Etablihed in 1991 by the Hong Kong Economic Time Limited with the upport of the leading real etate agencie in Hong Kong, the Economic Property Reearch Centre wa provide a databae of mot property tranaction in Hong Kong ince For each tranaction record, the EPRC provide detail about tranaction price, date of tranaction, addre of ubject property, the date of the occupation permit of a building, gro floor area, floor level, etc

12 regardle of the date of building urvey or inpection. Moreover, within the tudy window, there were no ignificant change in the condition of all the elected building. No large-cale refurbihment or rehabilitation work were undertaken to the building. Approved building plan of the 64 development, retrieved from the Building Department, were tudied. Inpection were carried out on thee building for the identification and counting of unauthorized appendage. Table 2 ummarize the tatitic of unauthorized appendage in the 64 development. 26 development out of 64 (40.6%) abutted on at leat one buy treet. Within the tudy period, there were altogether 412 tranaction. The decriptive tatitic of the continuou independent variable are ummarized in Table 3. Moreover, no ignificant correlation between the independent variable were potted upon the crutiny of the correlation matrix of the dataet. Table 2: Summary of unauthorized appendage proliferation in the 64 development Type of Unauthorized Appendage Maximum Mean Minimum σ Actionable Unauthorized Appendage Non-actionable Unauthorized Appendage Solid canopy Light-weight canopy (projecting more than 500mm from external wall) Air-conditioner upport frame (projecting more than 600mm from external wall) Metal frame Metal cage Solid extenion Flower rack Light-weight canopy (projecting not more than 500mm from external wall) Air-conditioner upport frame (projecting not more than 600mm from external wall) Drying rack Overall

13 Table 3: Decriptive tatitic of the continuou independent variable (n=412) Variable Maximum Mean Minimum σ Tranaction price (HK$ million) Flat ize (quare feet) 1, Floor level Age (month) Ditance from MTR tation (metre) Weighted average of the number of actionable and non-actionable unauthorized appendage per dwelling unit ANALYSIS RESULTS AND DISCUSSION The etimation reult of the hedonic price model, expreed in Equation (3), are hown in Table 4. The adjuted R 2 of the etimation wa The coefficient of AGE and AGE 2 were tatitically ignificant at leat at the 10% level, though their ign were different. With a negative coefficient for the firt-order term and a poitive coefficient for the econd-order term, the effect of the variable AGE on property price decreaed at a diminihing rate but the rate of change wa negligibly mall. Converely, oppoite reult for the effect of floor area on property price were returned from the hedonic price analyi. The coefficient of AREA wa found to be poitive wherea negative for AREA 2, and both coefficient were tatitically ignificant at the 1% level. Similarly, the diminihing trend in the poitive relationhip between floor area and property price wa trivial. Regarding the coefficient of FLOOR and FLOOR 2, the former wa found poitive and tatitically ignificant at the 5% level, with the latter being inignificant even at the 10% level. Thee finding confirmed the finding of my other hedonic price analye (e.g. Mok, 1995; So et al., 1997; Yau et al., 2008) in which a ignificant poitive relationhip between floor level and property price wa evidenced. A for the variable concerning the ditance between the ubject property and the nearet MTR tation, only the firt-order term wa found to be tatitically ignificant at the 10% level. The coefficient of MTRD wa found to be negative, implying that propertie with better acceibility were old at a higher price. With an eye to the aim of thi tudy, vat interet hould be placed on the etimated coefficient of the variable UNAPP and interaction term UNAPP BSTREET. The coefficient of the variable UNAPP (i.e., β 1 ) give an indication about the impact of the degree of proliferation of unauthorized appendage in a building on the elling price of dwelling unit in the building. In other word, the etimated coefficient of the variable meaure the value of elf-inurance mitigation with regard to building afety buying a property in a building with le rik created by unauthorized appendage. A for the coefficient of the interaction term UNAPP BSTREET (i.e., β 2 ), it meaure the price differential between propertie in a building bearing unauthorized appendage and facing one or more buy treet and thoe in a building bearing unauthorized appendage but not facing any buy treet. The coefficient aee the value of elf-protection mitigation with regard to building afety buying a property in a building with maller probable

14 loe (e.g. loe of propertie or damage for the injured) in cae of an unauthorized appendage failure. Table 4: Etimation reult of the hedonic price model Independent variable Coefficient t-tatitic CONSTANT *** AGE 3.13 x ** AGE x * FLOOR ** FLOOR x AREA 2.11 x *** AREA x *** MTRD 3.30 x * MTRD x UNAPP ** UNAPP BSTREET * Adjuted R Durbin-Waton tatitic 1.49 F-tatitic *** No. of obervation 412 Dependent variable ln PRICE Note: 1. (***), (**) and (*) denote the etimated coefficient of the variable and tet tatitic to be ignificant at the 1% level, 5% level and 10% level, repectively. 2. The reult for the time dummie were omitted, but are available upon requet. 3. All coefficient were etimated with White Heterokedaticity-conitent tandard error. From the analyi reult hown in Table 4, thee two element were found to have ignificant and negative effect on property price (at the 5% level at leat). That mean that, keeping other thing contant, dwelling unit in a building with a higher degree of proliferation of unauthorized appendage are old at a dicount and dwelling unit in a building with unauthorized appendage and abutting on one or more buy treet are old at a dicount. In other word, both hypothee of the reearch were not rejected by the empirical finding. The reult of the hedonic price analye indicated that dwelling in building with a lower probability of lo were old at a higher price, ceteri paribu. At the ame time, the lo reduction feature of the building under invetigation, i.e., locating away from a buy treet, wa found to have a poitive market value. Thee finding concurred with Ehrlich and Becker (1972) prediction that in the abence of market inurance, lo prevention and lo reduction mitigation were valued poitively by the market. In pite of it preliminary nature, thi reearch ha important finding that entail far-reaching policy implication. The analyi reult uggeted that lo prevention and lo reduction

15 meaure with repect to failure of unauthorized appendage were rewarded by the houing market in the abence of market inurance. Thee finding baically upport the ue of market force to motivate building owner to keep their building afe. Propertie in afer building command a high value o a value league on building afety performance can be etablihed. With an eye for higher property value, homeowner are incentivized to remove UBW in their building and keep them UBW-free. Intead of coercion and ubidie, government can intitutionalize meaure to facilitate a mooth exchange of information about afety performance of building among different player within the houing market. For example, government can make the information about UBW preent in each building available to the public. 7 By doing o, the market player can benchmark the afety performance among building more readily and confidently. The benefit of reduction in the probability of lo and prevent of lo will be more likely fully priced in property tranaction. On the other hand, the option of market inurance for lo reduction wa omitted in the current reearch. When market inurance i available, the price differential between propertie in afer and not-o-afe building may depend on homebuyer or building owner deciion to take out a building-related inurance policy (e.g. property-all-rik and/or third-party liability inurance). If a homebuyer plan to take out inurance for a property after it acquiition, he or he i willing to pay more for a afer property (or, more preciely, a property with a lower likelihood of building-related accident) in view of the lower inurance premium. In thi regard, the monetary return from lo prevention meaure regarding building afety will be more evident, or even amplified, if building-related inurance i made compulory for all reidential building in a city. CONCLUSION AND AGENDA FOR FURTHER RESEARCH Thi tudy wa motivated by urgent need to find way for utainably managing our building tock apart from law enforcement and ubidization. Market approach can be a probable reort given that market player put a value on the afety performance of building. Nonethele, economic of building afety ha been largely ignored in the literature. In order to examine whether a market approach can motivate building owner to keep their building afe, thi tudy evaluate the value of building afety performance in Hong Kong. Tetable hypothee were developed baed on Ehrlich and Becker (1972) theorie of elf-protection and elf-inurance. For the reearch purpoe, the afety performance of a reidential building wa meaured by the degree or extent of proliferation of unauthorized appendage in the building. For hypothei teting, hedonic price analyi wa conducted on a et of panel data which conit of property tranaction in building with different degree of unauthorized appendage proliferation and locational characteritic. The analyi reult howed that propertie in more unafe building were tranacted at a dicount, compared with thoe in relatively afer building. Thi exemplified the poitive market value of lo prevention effort. Moreover, given the ame level of building afety, propertie in a building abutting on one or more buy treet were old at a dicount, compared with thoe in a building not abutting on any buy treet. Lo prevention mitigation i thu valued poitively by the market in the abence of market inurance. A a whole, better afety performance of a building wa found to command a poitive value in Hong Kong houing market. 7 For example, report about illegal building tructure in Taipei, Taiwan have been made available for retrieval by the public through the online portal (

16 Frankly peaking, thi tudy i rather preliminary one on account of the reearch limitation like the mall number of obervation. Yet, the author hope that thi tudy can timulate more empirical tudie on the economic of building afety. The finding of uch further reearch could offer very valuable inight to the public adminitrator in formulating better-informed policie with regard to the utainable management of building tock in different part of the world. For example, the teting of the elf-protection and elf-inurance theorie can be extended by invetigating more type of UBW. Some UBW like erection of unauthorized appendage and alteration of load-bearing tructure increae the chance of building failure while ome other UBW like intallation of gate which obtruct ecape route amplify the caualtie (i.e., potential loe) in cae of fire and other emergencie. Furthermore, other apect of building afety like fire afety and freedom from external afety hazard (e.g. by locating further away from a petrol filling tation) can be reearched. Baed on different mitigation meaure embodied in either better building deign and proper building management and maintenance, a lo prevention index and lo reduction indicator index can be developed. With the two indice or indicator, market inurance can be taken into account and variou premie of Ehrlich and Becker (1972) can be tetified. For intance, when market inurance for building afety (ay third-party liability inurance) i available, the market value of lo prevention (or elf-protection) mitigation i theoretically expected to capture the capitalized value of the aving in market inurance premium. In addition, it i expected that elf-inurance i ubject to crowding out by market inurance available at actuarially fair price. ACKNOWLEDGEMENT The work decribed in thi paper wa fully upported by a grant from City Univerity of Hong Kong (Project No ). REFERENCES Anderon, H. (2005) The value of afety a revealed in the Swedih car market: an application of the hedonic pricing approach. Journal of Rik and Uncertainty, 30(3): Aren, S.B. (1997) The valuation of defective propertie: a common ene approach. The Appraial Journal, 65(2): Arnott, R.J. & Stiglitz, J.E. (1988) The baic analytic of moral hazard. Scandinavian Journal of Economic, 90(3): Arnould, R. and Grabowki, H. (1981) Auto afety regulation: an analyi of market failure. Bell Journal of Economic, 12(1): Baum, A. (1991) Property Invetment, Depreciation and Obolecence. London: Routledge. Baum, A. (1994) Quality and property performance. Journal of Property Valuation and Invetment, 12(1): Bernknopf, R.L., Brookhire, D.S. & Thayer, M.A. (1990) Earthquake and volcano hazard notice: an economic evaluation of change in rik perception. Journal of Environmental Economic and Management, 18(1):

17 Blomquit, G.C. (2004) Self-protection and averting behaviour, value of tatitical live, and benefit cot analyi of environment policy. Review of Economic of the Houehold, 2(1): Boulding, W. & Purohit, D. (1996) The price of afety. The Journal of Conumer Reearch, 23(1): Bouraa, S.C. & Peng, V.S. (1999) Hedonic price and houe number: the information of feng hui. International Real Etate Review, 2(1): Boyer, M. & Dionne, G. (1983) Variation in the probability and magnitude of lo: their impact on rik. The Canadian Journal of Economic, 16(3): Boyne, G.A., Millington, A.I. & Hunter, J.A. (1991) Government by grant: the cae of houing renovation. Public Policy and Adminitration, 6(3): Brookhire, D.S., Thayer, M.A. and Tchirhart, J. and Schulze, W.D. (1985) A tet of the expected utility model: evidence from earthquake rik. The Journal of Political Economy, 93(2): Building Department (2005) Blitz UBW Clearance Hong Kong: Building Department. Building Department (2010) Report on the Collape of the Building at 45J Ma Tau Wai Road, To Kwa Wan, Kowloon - K.I.L on 29 January Hong Kong: Building Department. Butler, R.V. (1982) The pecification of hedonic indexe for urban houing. Land Economic, 58(1): Carroll, T.M., Clauretie, T.M., Jenen, J. & Waddoup, M. (1996) The economic impact of a tranient hazard on property value: the 1988 PEPCON Exploion in Henderon, Nevada. Journal of Real Etate Finance and Economic, 13(2): Chan, K.J.K. (2000) Maintenance of old building. The Hong Kong Surveyor, 11(2): 4 7. Chang, Y.M. & Ehrlich, I. (1985) Inurance, protection from rik, and rik-bearing. The Canadian Journal of Economic, 18(3): Chau, K.W., Leung, A.Y.T., Yiu, C.Y. & Wong, S.K. (2003) Etimating the value enhancement effect of refurbihment. Facilitie, 21(1/2): Chau, K.W., Ma, S.M. & Ho, D.C.W. (2001) The pricing of luckine in the apartment market. Journal of Real Etate Literature, 9(1): Choy, Y.L. (1998) Building afety iue. In Building Department (ed.) Building Contruction in Hong Kong. Hong Kong: Building Department, pp Crook, A.D.H. & Hughe, J.E.T. (2001) Market ignal and direpair in privately rented houing. Journal of Property Reearch, 18(1): Cropper, M.L., Deck, L.B. & McConnell, K.E. (1988) On the choice of functional form for hedonic price function. Review of Economic and Statitic, 70(4): Donnelly, W. (1989) Hedonic price analyi of the effect of a floodplain on property value. Water Reource Bulletin, 24(3): Ehrlich, I. & Becker, G.S. (1972) Market inurance, elf-inurance, and elf-protection. The Journal of Political Economy, 80(4):

18 Feijtem, P. & Mulder, C.H. (2005) Life-coure experience and houing quality. Houing Studie, 20(4): Foo, K. (2012) Illegal tructure hunt at flat of health chief. The Standard, 15 May 2012, P08. Forte, R. & McCarthy, I. (2010) Renovate for profit: New Zealand reidential cae tudy. Property Management, 28(5): Frontin, P. & Holtmann, A.G. (1994) The determinant of reidential property damage caued by Hurricane Andrew. Southern Economic Journal, 61(2): Gamble, H.B. & Downing, R.H. (1982) Effect of nuclear power plant on reidential property value. Journal of Regional Studie, 26(4): Hatti, D.B. (1981) How exiting building and building rehabilitation are regulated? Bulletin of the Aociation for Preervation Technology, 13(2) Ho, D.C.W. (2000) An Analyi of Property-pecific Quality Attribute for Office Building. Unpublihed PhD thei, Department of Real Etate and Contruction, The Univerity of Hong Kong. Ho, D.C.W., Chau, K.W. & Yau, Y. (2008) Evaluating unauthorized appendage in private apartment building. Building Reearch and Information, 36(6): Hui, E.C.H., Wong, J.T.Y. & Wan, J.K.M. (2008) The evidence of value enhancement reulting from rehabilitation. Facilitie, 26(1/2): Hunter, T. A. (1992) Engineering Deign for Safety. New York: McGraw-Hill. Jimenez, E. (1983) The magnitude and determinant of home improvement in elf-help houing: Manila Tondo Project. Land Economic, 59(1): Kain, J.F. & Quigley, J.M. (1970) Meauring the value of houing quality. Journal of the American Statitical Aociation, 65(330): Kak, S.B. & Maani, S.A. (1992) Uncertainty, information and hedonic pricing, Land Economic, 68(2): Kunreuther, H. & Kleffner, A.E. (1992) Should earthquake mitigation meaure be voluntary or required. Journal of Regulatory Economic, 4(4): Leung, A.Y.T. & Yiu, C.Y. (2004) A review of building condition in Hong Kong. In Leung, Leung & Yiu, C.Y. (ed.) Building Dilapidation and Rejuvenation in Hong Kong. Hong Kong: City Univerity of Hong Kong & Hong Kong Intitute of Surveyor, pp Ligon, J.A. & Thitle, P.D. (1996) Information aymmetrie and informational incentive in monopolitic inurance market. The Journal of Rik and Inurance, 63(3): Luk, E. (2012) Illegal canopie found at home of houing official. The Standard, 27 November 2012, P07. Ma, M. (2011) Lawmaker mut get houe in order. The Standard, 17 May 2011, P04. MacDonald, D.N., Murdoch, J.C. & White, H.L. (1987) Uncertain hazard, inurance, and cutomer choice: evidence from houing market. Land Economic, 63(4):