UBC Faculty Research and Publications

Not In My Neighbor’s Back Yard : External Effects of Density Davidoff, Tom; Pavlov, Andrey; Somerville, Tsur Mar 31, 2018

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-Davidoff_T_et_al_Not_neighbor's_back_yard.pdf [ 2.21MB ]
JSON: 52383-1.0365257.json
JSON-LD: 52383-1.0365257-ld.json
RDF/XML (Pretty): 52383-1.0365257-rdf.xml
RDF/JSON: 52383-1.0365257-rdf.json
Turtle: 52383-1.0365257-turtle.txt
N-Triples: 52383-1.0365257-rdf-ntriples.txt
Original Record: 52383-1.0365257-source.json
Full Text

Full Text

Not In My Neighbor’s Back Yard: External Effects of DensityTom DavidoffA Andrey PavlovB Tsur SomervilleAAUniversity of British ColumbiaBSimon Fraser UniversityAbstractIn Vancouver, as in many “Superstar Cities,” low density zoning in residential neigh-borhooods is both a sacred cow of amenity preservation and a scapegoat for unaf-fordability. In 2008, the City Council allowed homeowners to build small “lanewayhomes” behind the main residential structure on qualifying lots in all of the city’ssingle family zones. We exploit this regulatory change to study the externalitiesthat infill housing imposes on neighboring property values. Laneway homes aremost commonly built when the main home is substantially rebuilt and generallyreduce the size of rear garages. Neighbors who build laneway homes thus add littlephysical density but impose a household of renters on neighbors. Overall we find asmall negative spillover that is not statistically different from zero, suggesting thatthe negative externalities of population density may be small. We do not find astronger negative effect in more expensive neighborhoods or on fancier homes, butthe negative spillover is statistically different from zero and larger in magnitude onproperties with larger lots (-11.2%).Key words: Real Estate Externalities, Real Estate Density1 We gratefully acknoledge BC Assessment’s assistance in providing transaction and property characteristicsdata for this paper.1 IntroductionNeighbors’ opposition to increasing density in existing residential neighborhoods haslong been studied by urban economists (Hamilton 1975, Fischel 2001). Recently, thisopposition has been seen as the root cause of rising prices in highly regulated metropoli-tan areas (Glaeser, Gyourko and Saks 2005) and has even been blamed for enormousmacroeconomic damage Moretti (2017). Foregoing the right to add density on one’s ownlot can be costly, and highly sought after cities such as San Francisco, Vancouver, andNew York tend to be highly regulated, suggesting that there are amenity benefits torestricting neighbors’ rights to build.While neighbors’ opposition to density seems important, there is little evidence con-cerning either (a) the negative externalities that development really imposes on nearbyresidents or (b) the price benefits from increases in development potential. In this paperwe examine the impact of the right to add a particular form of infill housing on propertyowners’ property value and the impact on value when neighbors exercise this option.The contribution of the paper is in using highly detailed property data in combinationwith a policy change on allowed densities that allows us to separate out the effect of anincrease in a property’s redevelopment option value from the effect of increased densityon an adjacent parcel.Our analysis takes advantage of a policy change by the municipal government in Van-couver, Canada, that was applied to all areas of the city zoned for single family detachedhousing. The policy allowed property owners whose lots and existing units met certaincriteria to construct a separate infill unit adjacent to the lane running at the rear of theproperty. These structures had to meet a variety of rigorous design and size criteria andcould not be sold, but allowed for a stand-alone rental property. From 2010 to 2016,following the July 2008 zoning change, approximately 2,356 laneway units were built insingle family areas, representing approximately 3.5 percent of the housing stock. In com-parison, over the same period 5,448 new single family properties were built, almost allreplacing existing properties. We both test the effect of the zoning change that allowed2for the ability to build these laneway units, the effect on property sales price of addinga laneway structure to a property with a separate single family structure property, andcritically for our test of the effect of density, the effect on a property’s price if a neighborconstructs a laneway.For the impact of the rezoning to allow laneway homes on own property value, we use adifference in differences approach, comparing prices of single family properties eligiblefor a laneway and those not satisfying the criteria before and after the regulatory changein 2008 that enabled some but not others to build. The key identifying assumption isthat the criteria that allowed laneway homes was not correlated with demand for thosehomes except through the new rights. To estimate the impact of a neighbor exercisingtheir post-2008 option to build a laneway on neighbors’ value, identification comes fromour assumed randomness of whether a new property built next to an existing propertyhas a laneway built or not. Supporting that assumption, we show that while there arebroad geographic differences in the propensity for newly built homes to add a lanewaythat are associated with prices, whether an adjacent new property has a laneway isindependent of the existing unit’s value or characteristics within neighborhoods. Thusthe treatment effect (that a newly built adjacent unit has a laneway) is apparently aconditionally random treatment. This randomness provides us with a cleaner test of theeffects of density than existing work, which looks at redevelopment at higher densities,which is unlikely to be random.There are two critical assumptions underlying our analysis of laneway home construc-tion’s impact on neighbors’ property value. First, we assume that conditional on neigh-borhood times year dummy variables, there are no systematic differences in unobservedvalue factors between properties adjacent to new homes with versus without lanewayhomes built. We show that there were no such differences in the pre-2008 period. Second,we assume that those who build new homes do not observe highly location-specific ex-ternalities associated with laneway homes: conceivably, one owner’s choice not to builda laneway home on Alma Street and another owner’s choice to build one on BlenheimStreet might reflect a greater negative externality of a laneway home on one’s own prop-3erty (and hence presumably on neighbors’ property) on Alma than on Blenheim. Rulingout the second possibility empirically is not possible, but we do show that characteristicsof homes that might be expected to be associated with a more negative externality oflaneway homes are not associated with a lower probability of a new home adding a newlaneway. The effective assumption then is that unobserved characteristics that affectlaneway externalities similarly do not affect laneway propensity (possibly because theyare also associated with higher laneway home rental value).The most important findings in this paper are about the magnitude and sign of densityspillovers. In the full sample of homes, a neighbor with an infill unit lowers the valueof the immediate neighbors by 1.8 percent, but the estimated effect is not statiscallydifferent from zero. While this implies on average no notable spillovers, this may notapply to higher densities: if the effect is linear a ten unit adjacent structure would have apoint estimate indicating a lowering of each neighboring house’s value by 15 percent. Thesmall magnitude and statistically insignificant result hides variation by property type.A natural conjecture is that willingness to pay to avoid living near renters might rise inlogs with property value or with lot size. We do not find that higher-end homes sufferfrom a notably larger negative externality. However, for a property on a lot of aboveaverage size, the reduction in value associated with a new laneway on a neighboringproperty is over ten percent.Section 2 of this paper reviews the extant research on density externalities and highlightsthe contribution of this study. Section 3 goes in depth into the data sources and iden-tification strategy. The results are shown in Section 4 and we conclude with commentson policy implications in Section 5.2 LiteratureEconomists generally look askance at zoning restrictions because they are seen to in-crease the cost of housing. Studies of housing markets in land-constrained environments(Saiz 2010) and those on land use regulation in general finger differences across commu-4nities in constraints on residential density though zoning as an important contributorto the variation in the price of housing (Gyourko and Glaeser, 2017). Models of ur-ban size and structure include a treatment of the negative externalities of residentialdensity, typically as a congestion externality assumed to apply to travel, which are off-set against the benefits of density through agglomeration externalities (Ahlfeldt, et. al.2015, Brinkmann 2016, Lucas and Rossi-Hansburg 2002).In these models the effects of density are at a fairly aggregated level. This literaturetreats density as a feature of neighborhoods or the city as a whole and ignores its mostimmediate local effects. Our contribution is to instead look at the extent to which densitycan have more localized negative effects. These localized effects are important as theyreflect the source of opposition by local residents to increased density, that while welfareincreasing on the whole, may result in more immediate negative effects for themselves(Fischel 2001).That density would have effects on neighboring property values is intuitive. Turnerand Haughwout (2014) disaggreate the different ways zoning restrictions effect propertyvalues; own-lot, external, and supply effects. Our focus is on the second of these, whichis sparse compared with the analysis of the larger supply question. Strange (1991),presents a theoretical model that categorizes the different ways density effects propertiesboth within and across neighborhoods. Critically, he allows for direct effects on nearbyproperties and indirect effects as increases in density in one area can induce wholesalezoning changes in individual neighborhoods or across a city. One of the contributions ofour work is that our empirical analysis is able to exclude the indirect effects he identifiesthat could bias estimates of the externality.An older empirical literature studies the effects of zoning and density changes on neigh-bors. Paper such as Sagelyn and Sternlieb (1972) and Stull (1975) find that multi-familyproperties in single family neighborhoods lower values of the nearby single family units.In contrast work by Crecine, et. al. (1967), Reuter (1973), Maser, et. al. (1977), andMark and Goldberg (1986) do not find that proximity, from a sample of properties a5given distance from the location of density, has an effect on house prices. The problemswith these empirical studies are those identified by Strange: that higher density at onelocation is likely to affect the possible redevelopment opportunities of nearby units,essentially positive redevelopment spillovers. Alternatively, areas that are conducive tohigher density may have other aspects that affect their prices: the presence of the higherdensity treatment buildings is non-random. Our contribution is to help separate the ef-fects of density from increased population, to look at newly built properties, and toavoid some of the sample selection problems that effect these other works by looking atchanges in density that occur within the existing zoning for newly built structures.Our identification comes from the teardown of older single family homes and theirreplacement with newer larger single family homes, some of which include a separate“laneway” infill which can be rented out. Studying the teardown phenomenon is notunique to our work. Wheaton (1982) among others included redevelopment formally intheir models of urban form and change. Helsley and Rosenthal (1994) were the firstto explicitly study the phenomenon as an empirical event, using these units to extractestimates of urban land values. Menace (1996), Helms (2003), and Dye and McMillen(2007) all estimate models to predict which units are torn down, looking at both unitand neighborhood characteristics. This form of redevelopment option exercise is studiedby Clapp and Salavi (2010), Clapp, et. al. (2012), and McMillen and O’Sulliven (2013),who attempt to estimate both the option exercise decision and the option value. Ourwork is different in that we do not attempt to model the redevelopment choice, butinstead use these properties, taking advantage of the outcome where some look to whatthe redevelopment mean for the values of adjacent properties.Density changes can also accompany changes in neighborhood composition, and it ispossible that effects observed reflect not the density, but the occupants. For instance, anumber of studies identify neighborhood effects of increased concentration of immigrantor low-income families in a neighborhood. For instance, Saiz and Wachter (2011) andSa (2015) document a negative price impact from increased immigrant concentration.Diamond and McQuade (2016) find a negative price impact of low-income housing in6high-priced neighborhoods. Our setting is distinct from this literature. The lanewayhouses in our sample are typically of very high quality and command high rents orare occupied by family members of the property owners. This allows us to investigatethe direct effect of density rather than the combined effect of density and changingdemographics.3 Data and IdentificationOur analysis focuses on the construction of laneway infill units in the backyards onsingle family detached properties. These units cannot be sold separately from the mainproperty as they do not have a distinct property title. They were intended to providerental housing in single family neighborhoods without dramatically changing the feel ofthe areas. By law laneway homes are either one- or two-bedroom units. Median size is663 sq ft (61.6 sq m), with the range from the 10th to 90th percentile of 489 to 865 sqft (45.4 to 80.4 sq m). There are strict rules on the design, spacing and layout of theseunits. 2 They are fairly high quality, with construction costs of $C200,000 to $C300,000,which excludes land costs. 3 High planning and building code requirements mean theirper square foot cost is more than that for an actual house. Between the first approvalsfor this form in July 2008 and the end of 2016, 2,361 such units were built in the Cityof Vancouver, BC, Canada.We wish to identify the effecs of a property having a lanewayunit on the values of theneighboring property. However, building a laneway for an existing property is likely tobe a non-random event. The nature of the building code is that building a lanewaycommonly means foregoing a two-car garage. Figures 1 through 3 suggest that theclientele for homes in fancier neighborhoods (generally those in the western half ofVancouver) care relatively more about garages and is less tempted by rental income2 Details on rules and regulations can be found at http://vancouver.ca/home-property-development/building-your-laneway-house.aspx .3 The per square foot cost for the median size unit is roughly $C 350 per squarefoot or $US 260. This is at the high end of single family construction costs:https://www.vancouverhome.builders/how-much-cost-to-build-house/ .7or less inclined to tolerate neighbors in close quarters. We thus confine the sample oflaneway homes in regressions to new units, comparing the price effects for new unitswith and without laneways. Given the greater propensity for laneway homes in EastVancouver, we include neighorhood times year dummy variables and rely on evidencethat within neighborhoods, the choice of whether or not to build a laneway home israndom.Another reason to study new homes is that among older homes that have not yetbuilt a laneway or been rebuilt, construction of a laneway in the near future may be avaluable option that is likely to be exercised in the not too distant (and not too heavilydiscounted) future. Thus the impact of exercising (versus retaining) that option mayhave a small impact on neighbors. By contrast, building a new two-car garage witha new home (by far the most common alternative to building a laneway among newhomes), involves a significant reduction in the option value of building a laneway home.Renovations are essentially the only source of new single family construction in Vancou-ver. With a few exceptions of subdivided lots, the overwhelming number of new singlefamily properties in Vancouver are a replacement of an existing single family unit. Sup-ply constraints and redevelopment in the city and metro area mean that the share ofhouseholds in single family units has been dropping steadily since 1981, and that inthe city and the metro area, the aggregate number of single detached houses dropped,despite a growing population. In this analysis we draw on data from nearly the wholecity (the Housing and Data Book from Metro Vancouver shows a decline from 50% to29% in the share of single detached homes of all private dwellings between 1991 and2016).We use transactions from 23 of the city’s 30 neighborhoods, limiting inclusion to thosewith single family transactions on property zoned for single family use drops 7 neigh-borhoods that in aggregate have 138 properties zoned for single family. Of these 5neighborhoods are in the downtown core and the single family units are unusual her-itage preservation outliers. Our transaction count for older properties (three years and8older) runs between 1,886 and 11,282 per neighborhood, with a median of 5,950. For newproperties (two years and fewer), the count per neighborhood runs from 33 to 850, witha median of 266 (just one neighborhood with fewer than 115 and one with more than626). The correlation between to these two counts is 0.66. So while the distributionis not perfectly uniform, the redevelopment phenomenon occurs throughout the city.Figure 1 shows the distribution of new units throughout the city. In Figure 2 we showthe distribution of transactions of properties that have a laneway. These are primarilynew houses, as most existing properties whose owners have invested in a laneway havenot sold their property.We combine transaction data from Vancouver for single family houses with propertyroll data that provides the standard set of property characteristics. Because we arecombining the transactions set with the universe of all properties we are able to identifythe characteristics of immediately adjacent properties, even if they do not transact. Forthe period of 2012-2016 from a full sample of single family transactions of 38,750, weget a reduced sample of 17,375 transactions of single family units, windsorized at the0.5th percentile on price, floor area, and lot size, where we limit the properties to thosein areas with single family zoning in 2017 and where the adjacent properties are allsingle family units. In total, 8.6 percent (1,494 transactions) have a neighbor that is lessthan three years old. The adjacent units have a median size of 1,987 square feet and astructure density (floor area or space ratio) of 0.39, compared to 1,803 and 0.35 for thetransacting units.3.1 Laneway AnnouncementOver our period of analysis, there were no major citywide changes in zoning that changedthe rules regarding the teardown and replacement of an existing single family homewith a second single family home. 4 In addition our analysis will include neighborhood- year dummies along with a citywide set of year-month dummies to control for both4 There are some changes along major arterials, but we limit our analysis to propertiesthat were in single family zones in 2017, thus excluding all previous sales of propertieswhose zoning subsequently changed.9high frequency general and moderate-frequency neighborhood variation in price levels.The decision to redevelop becomes an individual property owner or buyer’s decision asfunction of unit characteristics and personal preferences. The identification comes fromthe difference between properties adjacent to those that are redeveloped versus thosethat are not. In the data we define new as any property less than three years old toallow for a large enough window to have transactions.We exploit two announcements related to the ability of homeowners to add a lanewayhouse to their property to identify the value of the option to build. The first announce-ment was in July, 2008, when properties in the primary single-family zoning, SR-1,became eligible for a laneway house, subject to certain restrictions discussed below.Four years later, in July, 2013, the City of Vancouver extended this eligibility to allremaining single-family zoning designations. In addition to being with the appropriatezoning designation, a property needs to satisfy the following conditions to be eligiblefor a laneway house:(1) The property needs to back on a lane or another street. Properties that have nolane or street separating them from the property behind are not eligible to build alaneway house. This restriction applies even for corner lots, which in theory havethe necessary access for fire and other services. For this reason, we identify allproperties for which the lot polygon border is NOT within 4 meters of a lanewayas ineligible.(2) The requirements for cite coverage and access imply that properties with either ofthese characteristics are not eligible for a laneway house:• Lot is less than 110 ft deep OR narrower than 25 ft• Lot is BOTH less than 33 ft wide and less than 122 ft deep(3) he total site coverage of the main house and the laneway cannot exceed 40% ofthe property area. This restriction is particularly binding for properties known as“Vancouver Special” because of their size and location within the lot. To identifythese properties, we apply the following filters:10• Property built between 1963 and 1986, and• One story with full basement, and• Floor area exceeds 1500 sq. ft., and• Floor area to lot size greater than .5, and• Lot size is less than 9000 sq. ft.,• But allow laneways for large lots which exceed 148 x 36 feet.The above criteria for selecting properties eligible for laneway homes appears to beaccurate. All laneway homes observed in our data are located on properties that satisfythe above criteria. There are no laneway homes on properties that our mechanism labelsas ineligible. However, it is conceivable that properties labeled as eligible in our analysisare in fact not eligible. This could potentially bias down any impact of own lanewayhome eligibility that we find.Our identification strategy is based on the difference in price appreciation for propertiesthat are eligible and not eligible for a laneway house around each of the announcementdates. In efficient markets, this difference captures the option value to build a lanewayand thus substantially increase the FSR of the property.Specifically, we estimate the following difference-in-differences equation for all transac-tions subject to the zoning change:p = β0 + β1Characteristics+ β2∑I(Property in neighborhood i at time t)+β3Eligible+ β4postAnnouncement+ β5Eligible ∗ postAnnouncement(1)where p is the log - transaction price of property, “Characteristics” captures availableproperty characteristics, “Eligible” is an indicator variable whether the property meetsthe laneway requirements listed above, and “postAnnouncemnet” is an indicator vari-able that takes the value of one for transactions after the announcement date. Weestimate Equation (1) for properties within the zoning that changed: SR-1 to SR-7.11This methodology has two main threats to identification. First, it is conceivable thatsome other event differentially affected properties that are eligible for a laneway house.For instance, the value of backing onto a lane may have changed for a different reason.However, such a change would have to be perfectly contemporaneous with the lanewayannouncements.Second, the changes were not a total surprise, it is conceivable that they were anticipatedand reflected in prices before the announcement. However, both changes were highlycontested in passionate debates, and in our view the outcomes of the final votes werenot at all certain. Moreover, if the price impact took effect before the announcement,then our results would be biased downward, below the true effect of the announcement,thus making it possible that we do not detect an effect that was real, but never detectan effect by random chance.3.2 Laneway Option Exercise EffectIn this section we investigate the impact of actually building a laneway house on aproperty that is already eligible. This is the impact of exercising the option to build.Specifically, we estimate the following model:p = β0 + β1Characteristics+ β2∑I(Property in neighborhood i at time t)+β3(hasLaneway)(2)where “has laneway” is an indicator variable if the property has a laneway at the time oftransaction and all other variables are defined as above. We restrict this estimation onlyto properties eligible for a laneway house at that time. This includes all RS-1 propertiesup to 2013, and then properties in all RS zoning after 2013.While the above equation is rather straightforward, its estimation presents a difficulty.About 55% of all laneway houses were build as part of re-developing the entire property.12Therefore, we need to carefully separate the effects of a laneway from the effects ofproperty redevelopment. We do this by splitting the sample to new builds and existinghouses.3.3 Effect on Adjacent PropertyBuilding a laneway house represent a significant increase in density both in terms of sitecoverage and in terms of additional residents and vehicles. With this in mind, we inves-tigate the effect of building a laneway house on the neighbors. For each laneway built,we identify the immediate neighbors on each side. This identifies the properties mostimpacted by the newly built laneway house. To estimate the impact on the neighbors,we estimate the following model:p = β0 + β1Characteristics + β2∑I(Property in neighborhood i month t)+β3(Neighbour has laneway) + β4(Neighbour characteristics)(3)The neighbor characteristics we consider are floor area and lot area. The parameter β4estimates the effects of a neighboring laneway above and beyond the total floor areaand age of the neighbors.3.4 Randomness Check: Laneway House AssignmentOne potential concern with the estimation of the impact of own laneway or neighboringlaneway (Equations (2) and (3) above) is that laneway houses are not located at randomwithin neighborhoods but are systematically placed on more or less desirable lots. Sucha possibility would undermine the identification assumptions of Equations (2) and (3)and potentially generate misleading results.To address this possibilty, we first estimate the probability of having a laneway as afunction of the predicted property price and as a function of all available propertycharacteristis.13To further address the issue of random assignemt, we perform the following test forrandom assignment of laneway houses within neighborhoods. First, using data for the2011 - 2016 period we compute the share of new houses out of all new houses by censustract and 5-digit postal code (the first five characters in the Canadian alpha numericX0X 0X0 postal code):Share of New with Laneway =Number of New with LanewayNumber of all New(4)We then use the share variable to explain transaction prices in the pre-2010 sample.Specifically, we estimate the following hedonic model:p = β0 + β1Characteristics + β2∑I(Property in neighborhood i month t)+β3(Share of New w Laneway)(5)A significantly positive (negative) coefficient β3 would suggest that the 2011 - 2016laneway houses were concentrated in more (less) desirable census tracts or 5-digit postalcodes. Such a finding would undermine the identification of Equations (2) and (3).4 Results4.1 Announcement EffectWe offer summary statistics of the transaction data we use in Table 1. The averagetransaction price of over 1.1 million dollars over the entire period reflects Vancouver’sissues with housing affordability, which prompted the introduction of laneway housesas discussed above. However, the uptake of laneway houses is relatively low, penetrat-ing only 2.8 percent of the market. However, when considering only new construction,laneway houses are very popular, reaching nearly 50 percent in the first half of 2017.14Next we estimate the laneway announcement effect as given by Equation (1). Specifi-cally, we use July, 2008 as the first announcement event. At that time, properties zonedin the primary single-family zoning, SR-1 and SR-5, became eligible for a laneway,provided that they met the minimum requirements described in Section 4 above.Table 2 reports the estimation results for the July 2008 announcement for +/-6 and+/-9 - month estimation windows. The variable of interest is the interaction betweenthe post-July, 2008 indicator and the indicator for property eligibility “laneOK.” Whilethis interaction is positive for the two event windows we consider, it is only marginallysignificant for +/- 6 months, and not significant for the longer +/- 9 months window.Significance aside, the estimated effects are very large. For the +/- 6-month windowthe estimated impact is 15 percent of property value. In other words, properties thatbecame eligible in July, 2008 enjoyed a 15 percent increase in value relative to propertiesthat were not eligible. The fact that such a large impact is only marginally significantis likely a reflection that the vast majority of properties became eligible for a laneway.In fact, over 80 percent of all properties met the requirements for a laneway.Another consideration is the fact that the 2008 announcement was not a completesurprise. The City of Vancouver engaged in several community events and request forfeedback before the formal announcement. While the feedback was largely mixed andthe outcome was not easily predictable, it is possible that many homeowners startedpricing in the potential laneway announcement ahead of July, 2008.We also estimate Equation (1) for the 2013 announcement which made all remainingsingle-family zoning designations eligible for laneway houses. This estimation does notproduce a significant announcement effect at either horizon. This could be because ofa small sample issues or because the announcement was anticipated and already pricedin.154.2 Impact of Own Laneway HouseNext we report the estimation of Equation (2) which considers the own property valueimpact of building a laneway house. The results reported in Table 3 split the sample bynew/old house, big/small house, high/low price, and big/small lot. Houses built within4 years of the transaction date are defined as new. The remaining splits are at themedian values for the variable defining the split.We further report the results for the full sample and a sample restricted to propertiesthat meet the laneway requirements and have lot width between 25 and 48 feet andlength less than 148 feet.Building a laneway house increases property value for a new house by 10 to 14.5 percent,depending on whether we use the full or restricted samples. These estimates are highlysignificant and robust. This magnitude is not particularly surprising, as the cost ofbuilding a laneway house is likely has similar magnitude (15 percent of $2 million is$300,000, not far from the incremental cost of building a laneway home rather than agaragehome rather than a garage). Therefore, adding a laneway to a new project coversthe costs, but does not appear to generate large value increases beyond that.Adding a laneway to an existing house does not appear to increase its value. In fact, thiscoefficient, as reported in Model (3), is negative. This is likely because homeowners addlaneway houses to existing properties mostly for personal reasons - the need to housein-laws or generate some rental revenue. Owners who do that are unlikely to sell theirproperty. Therefore, the transactions for old houses with new laneways that we do seeare likely highly unusual and do not capture the true value enhancement of a lanewayhouse. Adding a laneway to an existing home may also indicate low redevelopmentoption value, as it would likely not make sense to build a laneway home soon beforerebuilding a property.The sub-samples for big lot or high price do not generate significant coefficients onthe impact of own laneway. The coefficients are actually quite small for high-priced16properties, between 0.3 and 1.7 percent, depending on the exact sample used. Thisfinding suggests that high-valued properties do not benefit from a laneway house. Mostlikely, the owners of high-value properties prefer the open space or larger garage andwould rather not share their property with other tenants.4.3 Impact of Neighbouring Laneway HouseNext we focus on the impact of a neighboring laneway house as specified by Equation(3). In order to isolate the impact of a neighboring laneway house from the impact of anewly built neighboring property we limit the sample to properties with a newly builtneighbor. We also limit the sample to properties with neighbors on both sides, thuseliminating corner lots or other unusual locations. Tables 4 and 5 provide the summarystatistics of the sample of properties with a new neighbor. Table 4 reports the summaryfor properties with a newly built neighbor who built a laneway house, and Table 5 forproperties with a newly built neighbor who did not. The two samples are very similarthe characteristics of the properties that transacted.Before we estimate Equation (3), it is important to establish that laneway homes areconstructed at random among all newly built houses that are eligible for a lanewayhome. If laneway houses are systematically built in superior or inferior areas, then theparameter on neighboring laneway house may potentially be capturing this system-atic difference in neighborhood quality rather than the impact of a laneway. We verifyrandom assignemnt of laneway homes in two ways.First, we estimate a predicted price for all new properties. This prediction model uses allavailable characteristics listed in the summary statistics tables, except for the garage-related variables. Since a 2-car garage and a laneway home are mutually exclusive, thegarage variables almost perfectly predict the existance of a laneway home. The age-related variables are also excluded because we are focusing only on newly constructedproperties. The predictive regression also include neighborhood effects, montly effects,and an interaction between neighborhood and year effects.17The predicted price is then used in a probit model to estimate the probability of having alaneway home. The predicted property price has no effect on the probability of having alaneway home, as reported in Table 6. We further estimate the probability of a lanewayas a direct funciton of all physical characteristics. All variables that have predictivepower are related to density of the property - numebr of bedrooms, availabilty of a rentalsuite, and number of stories. None of the other variables are significant in predictingthe probability of a lanway, including the lot size.Second, using data for the 2011 - 2016 period, we calculate the share of properties withlaneway houses out of all newly built properties by census tract and by 5-digit postalcode. We then include this share of laneway houses variable in a standard hedonic re-gression on all transactions prior to 2008, when laneway houses were not even discussed.If laneway houses are more likely to be built in a superior or inferior neighborhoods,than the post-2010 concentration variable would influence pre-2008 prices. We focuson the post-2010 concentrations because most laneways in our sample were built after2010, reflecting the time it takes for property owners to take advantage of the newlypermitted use (proposal development, approval, construction, and sale).Table 7 reports the result of this random assignemnt check. Specifically, the concentra-tion of laneway houses is small and not significant in explaining pre-2008 transactionprices for any of the definitions or sub-samples we use. In fact, the sign switches betweenpositive and negative between specifications, just as one might expect if the assignmentis random.Given the above two results, random assignemnt of laneway homes among all newlyconstructed houses appears to be very likely. With this in mind, we estimate the impactof a neigboring laneway as proposed in Equation (3). Table 8 reports this result. Thevariable of primary interest is “New neighbor has a laneway.” It captures the impact ofa laneway that is built as part of a newly re-developed neighboring property. For the fullsample the effect of a neighboring laneway house is small (1.8%) and not significant. Themagnitude of the coefficient is double in specification (2), where the sample is confined18to homes with hedonically fitted values greater than the mean. 5 This difference is notsignificant, and with the smaller sample size, we cannot reject the null hypothesis thatthere is zero impact even on more desirable homes. We find in specification (4), thathomes on large lots are highly negatively and significantly impacted (-.112) by lanewayhomes.These results suggest that in most cases building a laneway house does not generate asignificant negative externality for the neighbors. However, it does generate significantand substantial negative externalities for properties located on bigger lots. We cannotreject sizeable negative effects, particularly for higher quality homes.5 ConclusionWe explore the impact of the right to build a laneway home, and the impact of exercisingthat right on one’s own property and one’s immediate neighbors. Exercising the rightin a newly built home generally means reducing the size of the garage. The announcedright to build a laneway home appears to have had a briefly positive impact on pricesaround the announcement date in 2008. The extension of that right in 2013, howeverdid not have a positive impact. The small impact of the right to build a laneway homematches two features of the data documented here. First, there has not been a rush tobuild laneway homes – fewer than 5% of homeowners granted this option have exercised.Second, building a laneway home appears to add roughly the same value to propertiesas it costs to build.We find only a small and insignficant effect of building a laneway on neighboring prop-erty values. For neighbors with larger yards, perhaps more impacted by the unpleasant-ness of more people nearby, there is a significantly negative impact on price.Given that the magnitude of our estimated effects is small, the results may be inter-preted as casting doubt on the common “NIMBY” concern that neighbors will reduce5 We predict home value based on characteristics and split the sample on predictedvalue.19property values. On the other hand, laneways only introduce a small amount of density.Multiplying our point estimate of 1.3% by 10, we would find an economically meaningfulimpact of building a multiunit property adjacent to a given home.206 ReferencesAhlfeldt, Gabriel M., Redding, Stephen J., Strum, Daniel M., and Nikolaus Wolf, 2015.The Economics Of Density: Evidence From The Berlin Wall. Econometrica. 83 (6),21272189.Brinkman, Jeffery C., 2016. Congestion, Agglomeration, and the Structure of Cities.Journal of Urban Economics. 94, 1331.Clapp, J.M., Salavi, K., 2010. Hedonic pricing with redevelopment options: a new ap-proach to estimating depreciation effects. Journal of Urban Economics 67, 362377.Clapp, J.M., Bardos, K.S., Wong, S.K., 2012. Empirical estimation of the option pre-mium for residential redevelopment. Regional Science and Urban Economics 42, 240256.Crecine, Paul, Davis, A. and J. E. Jackson, 1967. Urban property markets: Some empir-ical results and their implication for municipal zoning. Journal of Law and Economics.10, 79-99.Diamond, R. and T. McQuade. 2016. Who Wants Affordable Housing in Their Back-yard? An Equilibrium Analysis of Low- Income Property Development. Working Paper.Dye, Richard F. and Daniel P. McMillen, 2007. Teardowns and Land Values in theChicago Metropolitan Area. Journal of Urban Economics. 61 (1), 45-64.Fischel, William A., 2001. The Homevoter Hypothesis: How Home Values InfluenceLocal Government (Cambridge, MA: Harvard University Press.Glaeser, Edward, and Joseph Gyourko, 2017. Economic Implications of Housing Supply.Wharton Zell/Lurie Working Paper #802.Glaeser, Edward, Joseph Gyourko, and Raven Saks. 2005. Why is Manhattan so ex-pensive?: Regulation and the rise in house prices. Journal of Law and Economics,48(2):331370.21Grether, D. M. and P. Mieszkowski, 1980.. The Effects of Nonresidential Land Uses onthe Prices of Adjacent Housing: Some Estimates of Proximity Effects. Journal of UrbanEconomics. 8, 1-15.Hamilton, Bruce W. 1975. Zoning and property taxation in a system of local govern-ments. Urban Studies, 12(2):205211.Lucas, Robert E. and Enrico Rossi-Hansberg, 2002. On the internal structure of cities.Econometrica, 70 (4), 14451476.Mark, Jonathan H. and Michael A. Goldberg, 1986. A Study of the Impacts of Zoningon Housing Values Over Time. Journal of Urban Economics. 20, 257-273.Maser, S.M., Riker, W. H. and R. N. Rosett, 1977. The Effects of Zoning and Externali-ties on the Price of Land: An Empirical Analysis of Monroe County, New York. Journalof Law and Economics. 20, 111-132.McMillen, Daniel, P. and Arthur OSullivan, 2013. Option Value and the Price of Tear-down Properties. Journal of Urban Economics. 74, 107-129.Moretti, Enricoand and Chang-Tai Hsei. 2017. Housing constraints and spatial misallo-cation chang-tai hsieh. Working paper, University of Chicago and NBER.Rueter, F. H., 1973. Externalities in Urban Property Markets: An Empirical Test of theZoning Ordinance of Pittsburgh. Journal of Law and Economics. 16, 315-350.Sa, F. (2015). Immigration and House Prices in the UK. The Economic Journal, 125(September), 1393-1424.Sagalyn, Lynn B. and G. Stemlieb, 1972. Zoning and Housing Costs: The Impact of LandUse Controls on Housing Price. Center for Urban Policy Research, New Brunswick, NJSaiz, Albert, 2010. The Geographic Determinants of Housing Supply. Quarterly Journalof Economics. 1253-1296.22Saiz, A. and S. Wachter. (2011). Immigration and the neighborhood. American Eco-nomic Journal: Economic Policy, 3(2), 169-88.Strange, William C., 1992. Overlapping neighborhoods and Housing Externalities. Jour-nal of Urban Economics. 32, 17-39.Stull, W.J., 1975. Community Environment, Zoning, and the Market Value of SingleFamily Homes. Journal of Law and Economics. 18, 535-557.Turner, Matthew A., Andrew Haughwout, Wilbert van der Klaauw, 2014. Land UseRegulation and Welfare. Econometrica. 82 (4) 1341- 1403.Wheaton, W.C., 1982. Urban residential growth under perfect foresight. Journal ofUrban Economics 12, 121.237 Figures and Tables24Fig. 1. Location of newly constructed houses.25Fig. 2. Location of transactions that include a laneway.26Fig. 3. Location of transactions that are adjecent to a new house.27(1) (2) (3) (4) (5)VARIABLES N mean sd min maxlnP 39,318.000 13.929 0.686 11.513 16.118Dummy, =1 if parcel re-zoned for laneway July 28 2008 39,318.000 0.945 0.229 0.000 1.000Dummy, =1 if parcel newly re-zoned for laneway July 2013 39,318.000 0.055 0.229 0.000 1.000Property has a Laneway unit 39,318.000 0.030 0.171 0.000 1.000Lot size 000sf 39,318.000 5.220 2.248 1.920 36.339Lot size squared 39,318.000 32.302 44.773 3.686 1,320.523Finished area 000sf 39,318.000 1.738 0.775 0.588 6.539Finished area squared 39,318.000 3.620 3.546 0.346 42.759Number of bedrooms 39,305.000 4.544 1.373 0.000 12.000Number of bathrooms (full+half) 39,305.000 3.458 1.635 0.000 12.000Dummy, =1 if land use is single family w/ suite 39,318.000 0.504 0.500 0.000 1.000Number of stories 39,318.000 1.496 0.492 1.000 3.000Dummy, =1 if has full basement 39,318.000 0.664 0.472 0.000 1.000Less than 5 years old 39,318.000 0.148 0.355 0.000 1.000Age - renovation adjusted 39,318.000 27.372 19.325 0.000 106.000Age Squared 39,318.000 1,122.676 1,280.742 0.000 11,236.000Had major renovation 39,318.000 0.466 0.499 0.000 1.000Years since major renovation 39,037.000 28.340 19.762 0.000 106.000Years since major renovation - Squared 39,037.000 1,193.693 1,333.053 0.000 11,236.000Dummy, =1 if has single car garage 39,318.000 0.266 0.442 0.000 1.000Dummy, =1 if has multi-car garage 39,318.000 0.553 0.497 0.000 1.000Table 1The table reports summary statistics of the full sample, including all controls used in all regressions.28(1) (2) (3) (4) (5) (6)2007 falsification 2007 falsification 2008 rezoning 2008 rezoning 2013 rezoning 2013 rezoningVARIABLES +/- 6 mos +/- 12 mos +/- 6 mos +/- 12 mos +/- 6 mos +/- 12 mos1.postJuly20071.laneAllow1 1 -0.007 -0.009(-0.30) (-0.48)1.postJuly20081.laneAllow1 1 0.058* 0.015(1.80) (0.80)1.postJuly20131.laneAllow2 1 0.025 0.028(0.56) (0.70)Observations 4,436 6,797 3,247 7,583 3,793 7,458R-squared 0.614 0.609 0.592 0.623 0.717 0.719Neighborhood/time effects + controls Yes Yes Yes Yes Yes YesRobust t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1Table 2The table reports the estimation of Equation (1) for the 2008 and 2013 announcements based on zone classificationas well as a falsification test for July, 2007. The coefficient on the on the interaction term post-July 2008X laneAllow1 and post-July 2013 X laneAllow2 report the announcement effect on properties with zoning thatallowed laneway houses as of July, 2008 and July, 2013, respectivelly. The interaction term is large and marginallysignificant at +/- 6 months of the 2008 announcement, but becomes smaller and not significant for longerestimation windows or for the 2013 annoucement.29(1) (2) (3) (4) (5) (6) (7) (8) (9)Full Sample Restricted Sample Full Sample Restricted Sample Full Sample Restricted SampleFull Sample Restricted Sample New New New New New NewVARIABLES New House New House Old House Big House Big House High Price High Price Big Lot Big Lot1.haslaneway 0.110*** 0.144*** -0.030 0.137** 0.112 0.031 -0.018 0.094 -0.055(2.92) (2.74) (-0.97) (2.28) (1.53) (0.34) (-0.19) (1.04) (-0.46)Observations 4,549 2,746 26,398 3,583 1,989 2,241 933 2,231 688R-squared 0.822 0.797 0.714 0.804 0.797 0.635 0.624 0.804 0.849Neighborhood/time effects + controls Yes Yes Yes Yes Yes Yes Yes Yes YesRobust t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1Table 3The table reports the impact of having a laneway unit using Equation (2) We report the estimates for the full sample and a sample restricted to properties that meetthe laneway requirements and have lot width between 25 and 48 feet and length less than 148 feet. We further break down the sample by age and by house size, lotsize, and transaction price (above and below median). The presence of a laneway unit is positive and significant for all new houses, and for big new houses. However,it is NOT significant for expensive houses or houses on big lots.30(1) (2) (3) (4) (5)VARIABLES N mean sd min maxLot size 000sf 339.000 4.747 1.657 2.750 27.080Finished area 000sf 339.000 1.554 0.608 0.648 4.041Number of bedrooms 338.000 4.701 1.559 2.000 10.000Number of stories 339.000 1.462 0.473 1.000 2.000Dummy, =1 if has full basement 339.000 0.743 0.437 0.000 1.000Dummy, =1 if has multi-car garage 339.000 0.454 0.499 0.000 1.000Dummy, =1 if has single car garage 339.000 0.327 0.470 0.000 1.000Dummy, =1 if land use is single family w/ suite 339.000 0.596 0.491 0.000 1.000Property has a Laneway unit 339.000 0.112 0.316 0.000 1.000Dummy, =1 if parcel re-zoned for laneway July 28 2008 339.000 0.979 0.142 0.000 1.000Dummy, =1 if parcel newly re-zoned for laneway July 2013 339.000 0.021 0.142 0.000 1.000lnP 339.000 14.087 0.541 12.182 15.409Number of bathrooms (full+half) 338.000 3.444 1.720 1.000 12.000Age - renovation adjusted 339.000 31.112 22.399 0.000 77.000Years since major renovation 334.000 33.404 22.928 0.000 84.000Age Squared 339.000 1,468.192 1,535.708 0.000 5,929.000Less than 5 years old 339.000 0.192 0.394 0.000 1.000Lot size squared 339.000 25.271 40.617 7.562 733.326Finished area squared 339.000 2.782 2.250 0.420 16.330Had major renovation 339.000 0.383 0.487 0.000 1.000Years since major renovation - Squared 334.000 1,639.973 1,643.883 0.000 7,056.000Table 4The table reports the summary statistics for transactions with a newly built neighbor with a laneway.31(1) (2) (3) (4) (5)VARIABLES N mean sd min maxLot size 000sf 1,157.000 5.361 2.284 2.011 36.339Finished area 000sf 1,157.000 1.902 0.747 0.663 6.198Number of bedrooms 1,156.000 4.614 1.416 1.000 10.000Number of stories 1,157.000 1.670 0.464 1.000 3.000Dummy, =1 if has full basement 1,157.000 0.659 0.474 0.000 1.000Dummy, =1 if has multi-car garage 1,157.000 0.603 0.489 0.000 1.000Dummy, =1 if has single car garage 1,157.000 0.241 0.428 0.000 1.000Dummy, =1 if land use is single family w/ suite 1,157.000 0.456 0.498 0.000 1.000Property has a Laneway unit 1,157.000 0.041 0.199 0.000 1.000Dummy, =1 if parcel re-zoned for laneway July 28 2008 1,157.000 0.941 0.235 0.000 1.000Dummy, =1 if parcel newly re-zoned for laneway July 2013 1,157.000 0.059 0.235 0.000 1.000lnP 1,157.000 14.378 0.646 12.206 15.999Number of bathrooms (full+half) 1,156.000 3.983 1.815 1.000 9.000Age - renovation adjusted 1,157.000 24.529 21.699 0.000 100.000Years since major renovation 1,150.000 26.346 22.668 0.000 100.000Age Squared 1,157.000 1,072.126 1,407.415 0.000 10,000.000Less than 5 years old 1,157.000 0.296 0.457 0.000 1.000Lot size squared 1,157.000 33.953 54.043 4.044 1,320.523Finished area squared 1,157.000 4.175 3.446 0.440 38.415Had major renovation 1,157.000 0.449 0.498 0.000 1.000Years since major renovation - Squared 1,150.000 1,207.503 1,521.489 0.000 10,000.000Table 5The table reports the summary statistics for transactions with a newly built neighbor without a laneway.32(1) (2)VARIABLES LABELS Has Laneway Has Lanewaylotsize Lot size 000sf 0.219(1.29)lotsize2 Lot size squared -0.013(-1.23)floorarea Finished area 000sf -0.532(-0.85)floorarea2 Finished area squared 0.059(0.44)bedrooms Number of bedrooms 0.135***(3.89)baths Number of bathrooms (full+half) 0.090*(1.88)suite Dummy, =1 if land use is single family w/ suite 0.462***(3.12)stories Number of stories 0.956***(4.87)base full Dummy, =1 if has full basement 0.145(1.16)lnPhat Predicted log-Price 0.061(0.27)Observations 2,912 2,912Neighborhood/time effects Yes YesRobust z-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1Table 6The table reports the probability of having a laneway as a function of the predicted transaction price as estimatedby a probit regression. The sample is restricted to new houses (less than 4 years old) that are eligible for a lanewayhouse. The predicted transaction price is based on a model that includes all controls listed in the summarystatistics tables except the age-related and garage-related variables. The age-related variables are not includedbecause the sample is restricted to new houses, and the garage-related variables are not included because theyare direclty correlated with a laneway house. The predicted transaction price is not a significant predictor of theprobability of having a laneway in any of the samples we report.33(1) (2) (3) (4) (5) (6) (7) (8)VARIABLES Tract Tract 5-digit code 5-digit code High Price Big House Big Lot New HouseShare of new w laneway out of all new 0.055 0.050(0.83) (0.98)Share of new w laneway out of all new 0.035 0.022 -0.018 0.043 0.030 0.004(1.15) (1.16) (-0.57) (1.29) (1.06) (0.07)Observations 8,585 8,548 8,450 8,414 919 3,562 4,484 1,383R-squared 0.495 0.613 0.496 0.618 0.538 0.537 0.627 0.708Neighborhood/time effects Yes YesNeighborhood/time effects + controls Yes Yes Yes Yes Yes YesRobust t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1Table 7This table verifires the random assignemnt of laneway houses among all newly built properties. We first computethe share of new houses with laneways out of all new houses built in the 2011 - 2016 period by census tractand by 5-digit postal code. We then include these shares in hedonic regressions for all single family transactionsprior to 2010. If laneways are systematically built on superior or inferior properties, the share of laneway houseswould have a significnat impact on pre-2010 transaciton prices. Instead, the share of new houses with lanewaysout of all post 2010 new houses has no explanatory power in any of the model specifications we considered. Weinterpret this result as evidence that laneway houses are randomly assigned among all new houses.34(1) (2) (3) (4) (5) (6)VARIABLES Full sample High predicted price Very High predicted price Big house Big lot New houseNew neighbor has laneway -0.018 -0.065 -0.143** -0.009 -0.112** -0.002(-0.70) (-1.53) (-2.46) (-0.21) (-2.28) (-0.04)Observations 1,483 743 487 753 730 401R-squared 0.788 0.621 0.598 0.755 0.780 0.878Neighborhood/time effects + controls Yes Yes Yes Yes Yes YesRobust t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1Table 8The table reports the estimation of Equation (3) only for transactions with a newly built neighbor. We breakdown the sample by predicted price, house size, lot size, and age of house. The results in the full sample indicateno singificnat negative ipact of a neighboring laneway house. The coefficient is negative but small and relativelyprecisely estimted, allowing us to rule out a large negative effect. The coefficient for higher-priced propertiesis much more negative and not precisely estimated, indicating that we cannot rule out large negative impacton higher-valued properties. The coefficient for big lots is highly negative and significant, indicating that largeproperties suffer a significnat negative impact form a neighboring laneway.35


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items