Why Have Housing Pices Gone Up? By EDWARD L. GLAESER, JOSEPH GYOURKO, AND RAVEN E. SAKS* Ove the past 30 yeas, housing pices have isen egulaly, and the dispesion in housing pices acoss Ameican makets has inceased even moe substantially. Since 1970, U.S. Census data show that the standad deviation of pices acoss metopolitan aeas inceased by 247 pecent, compaed with 72-pecent appeciation in aveage pices. This gowing dispesion has occued mainly in the uppe end of the distibution. In many pats of the county, new housing units still ae abundant, and housing pices emain elatively low. In a small, but inceasing numbe of metopolitan aeas (pimaily, but not exclusively, on the coasts), housing pices have soaed, and new constuction has plummeted. Because ising pices have been accompanied by lage eductions in esidential development in these places, the natual explanation fo these changes in metopolitanaea housing makets is that housing supply is limited. These constaints do not appea to be caused by a declining availability of land, but athe they ae the esult of a changing egulatoy egime that makes lage-scale development inceasingly difficult in expensive egions of the county (Glaese and Gyouko, 2003; Glaese et al., 2005a). Changes in housing-supply egulations may be the most impotant tansfomation that has happened in the Ameican housing maket since the development of the automobile, but this change is both unde-studied and undedebated. The positive eseach agenda going fowad should be to undestand why these changes have occued and how they elate to othe majo tends in Ameican society. The nomative policy agenda should be to bette undestand the costs and benefits of limits on new constuction. The costs appea to include highe pices and a misallocation of labo, while * Glaese: Depatment of Economics, Havad Univesity, Cambidge, MA 02138, and NBER (e-mail: eglaese@havad.edu); Gyouko: Whaton School, Univesity of Pennsylvania (gyouko@whaton.upenn.edu); Saks: Havad Univesity (saks@fas.havad.edu). 329 the benefits include intenalization of constuction-elated extenalities. Given the magnitude of this egulatoy shift, the economics pofession could make a majo contibution by analyzing the welfae effects of egulation on the ise in housing pices. I. The Economics of Zoning and Pemitting Ou model is one of a local zoning authoity that decides whethe to appove o eject esidential development. Thee ae two locations: the zoning authoity s town and a esevation locale. Thee ae N total consumes, of which D live in the town. The emainde of the population lives in the esevation locale, and thee ae no constaints peventing people fom moving thee. Total utility in this outlying aea is a deceasing function of the numbe of people living thee, U (N D). In the town, the flow of utility equals U(D) a (housing costs), whee U(D) is deceasing in the amount of development in the city, and a is an individual-specific desie to live in the town. The distibution of a is descibed by a cumulative density F(a) and density f(a). The cost of constuction in the town equals K, which captues both physical costs of constuction and the oppotunity cost of land taken away fom agicultual uses. We nomalize the cost of constuction in the esevation locale to be equal to zeo, so that K eflects the additional cost of building a housing unit in town. Denoting the inteest ate as, the annual cost of housing constuction is K. As in any spatial equilibium, thee will be a maginal consume with a taste fo the town equal to â who is indiffeent between living in the town o the esevation locale. Evey consume with a value of a geate than â will live in the town, and the emaining consumes will live in the esevation locale. The maginal consume must satisfy D N[1 F(â)]; we use the notation â(d) F 1 (1 D/N). The initial population of the town is split into homeownes and entes. We assume that a
330 AEA PAPERS AND PROCEEDINGS MAY 2005 faction h of these units ae allocated to homeownes, and the emainde ae allocated to entes. All individuals ae assumed to live in the community fo exactly L time peiods. Afte that time, individuals ae eplaced by identical consumes so that the total size of the population emains unchanged. Individuals maximize V t 0 e t u(t) dt e V Asset V whee u(t) is the flow of utility at time t, and Asset V is the value of any asset as of time V. Rentes pay the maket-cleaing ent, which is equal to the same annual cost as the inteest payments on a house. Because new constuction inceases the supply of housing units, both ents and housing values will decline with new development. If the town stats with D housing units, then houses in town will be woth U D U N D â D Given these assumptions, thee is a unique amount of development that will maximize the aveage discounted lifetime utility of all cuent esidents of the town, but thee ae two easons why the level of development that maximizes the welfae of cuent esidents will not be socially optimal. Fist, highe population density has a negative impact on the utility of futue esidents of the town and of esidents of the esevation locale that is not intenalized by cuent town esidents. Second, cuent homeownes have an incentive to incease the value of thei homes and do not intenalize the impact that highe housing pices have on nonhomeownes who would like to live in the town. We now conside the decision faced by the town s zoning authoity who decides whethe a new development poject of size will poceed. We simplify the analysis by assuming that utility in the esevation locale is fixed at U. Futhemoe, we will ignoe the incentive of entes to lobby fo moe housing to be built. Essentially, this assumption implies that entes ae not oganized enough to suppot the constuction of new housing. The zoning authoity will eceive net benefits of g C (C R C D ) g T (T R T D ) fom ejecting the poject. The paamete captues the innate distaste of the authoity fo development. C D and C R eflect the cash spent by. developes and town esidents to influence the authoity s decision, and g C is a concave function eflecting the influence that cash will have on the decision-making of the authoity. Similaly, T D and T R eflect the time spent by the develope and esidents, espectively, on influencing the authoity, with a concave function g T eflecting the influence of time on the authoity. We assume that both g functions ae symmetic aound zeo. Finally, is a unifomly distibuted, mean-zeo idiosyncatic tem with density 1. We will assume that the paamete values ae such that thee is always some positive pobability that the poject will be accepted and likewise some positive pobability that it will be ejected. Unde these assumptions, the pobability that the authoity will authoize the poject equals 0.5 g C (C D C R ) g T (T D T R ). The cost of time to the develope equals W D, and the cost to esidents equals W R. The develope theefoe chooses the amount of time and cash spent to influence the zoning boad to maximize (1) 0.5 g C C D C R g T T D T R U D U â D W D T D C D. K Fom the pespective of each cuent homeowne, the development poject will ceate a net loss equal to U D U D e L â D â D. This expession eflects both the negative extenality associated with highe population density and the decline in housing values. Individuals face a cost of influencing the authoity equal to W R T R C R. Because a continuous distibution of esidents implies that each individual peson has a negligible impact on the zoning decision, we assume the existence of a community goup that oganizes town esidents. This oganization includes a popotion of homeownes and maximizes the aggegate utility of its membes. We assume that W D W R so that the oppotunity cost of time is highe to the develope than
VOL. 95 NO. 2 REGULATION AND THE HIGH COST OF HOUSING 331 to the homeownes. Although this assumption seems plausible, it implies that landlods cannot employ entes to lobby the zoning boad at the same time cost faced by homeownes. 1 Taken togethe, this model implies two cental popositions (poofs in Glaese et al., 2005b): PROPOSITION 1: If both the landlod and the homeownes association undetake some lobbying effot, then the landlod will use only cash and the homeownes will use only time. Such specialization of effot seems consistent with much anecdotal evidence on local battles between developes and community goups (Kee Wane and Havey Molotch, 2000). PROPOSITION 2: If both actos engage in a positive amount of lobbying then: (i) the pobability the poject will be appoved will decline with, (ii) the pobability that the poject will be appoved is deceasing with h and, (iii) if g C (x) C g C(x) and g T (x) T g T(x), then the pobability the poject will be appoved declines with T and ises with C, and (iv) if U(D ) U(D ) u, then the pobability the poject will be appoved falls with u. Poposition 2 sets foth a numbe of compaative statics that can potentially explain the change in the zoning envionment within the United States. The fist petains to a change in the pefeence of judges and othe political decision-makes egading development. The second compaative static suggests that the explanation could lie in the ise of homeowneship and the success of community oganization. The homeowneship ate has inceased by about 10 pecentage points ove the past 40 yeas, and political paticipation by homeowne 1 Histoically, it has been ae to see entes fight zoning estictions. Pehaps this absence is due to an agency poblem that pevents developes fom oganizing entes. Stonge homeowne paticipation might also be because homeownes simply enjoy the social activity of potesting new developments. goups has been ising (Eic C. Feund, 1974; Robet H. Nelson, 2004). The thid pat of Poposition 2 points to the changes in the elative effectiveness of using cash vesus time to influence political decision-makes. The final compaative static concens the taste fo density. We should expect to see less development if ising incomes have caused people to place a highe value on living in a low-density community. Othe factos, such as cime and impovements in tanspotation, also may have inceased the desiability of low-density living. II. Evaluating the Explanations fo a Moe Restictive Zoning Envionment A. Judicial Tastes Robet Ellickson (1977 p. 338) noted that sububan govenments ae becoming eve moe adventuesome in thei effots to contol housing development. Ellickson does not discuss easons fo this change but points to judicial decisions such as Nectow v. City of Cambidge that have inceased the difficulty fo landownes to stop municipalities fom esticting new constuction on thei land. William Fischel (2004 pp. 332 33) points to the ideology of judges: Couts, whose judges shae the same envionmental attitudes as middle class homeownes (just as 1920s judges shaed the ideology of heath and home), wee moe sympathetic to claims that the local decision had failed to account fo envionmental impacts than they had been to seemingly selfish claims that neighbos home values wee at isk. Othe cases have made it clea that the couts will allow gowth contols as long as othe conditions ae met: once a community has satisfied its fai shae obligation [a faction of the egion s low-income housing], the Mount Lauel Doctine will not estict othe measues, including lage-lot and open aea zoning, that would maintain its beauty and communal chaacte (Mount Lauel II, 456 A.2d at 421; cited in Fischel [2004 p. 331]). Thee can be little doubt that cout decisions have become fiendlie to anti-development sentiment. While couts clealy ae impotant, ultimately it is unsatisfying to attibute the change in the zoning envionment to changing attitudes of judicial decision-makes. These
332 AEA PAPERS AND PROCEEDINGS MAY 2005 attitudes ae not exogenous, but eflect othe tends in Ameican society. If changes in the tastes of judges and policymakes eflect societal tends like the envionmental movement, then these changes should be viewed as an impoved effectiveness of cetain goups in shaping policy. In the language of the model, this should be viewed as an incease in T R o T, not an exogenous change in. Empiically, we cannot eject the hypothesis that judicial tastes changed, but on theoetical gounds this explanation is so unsatisfying that we will tun elsewhee. B. The Impact of Residents Goups While the influence of developes may o may not have declined, many obseves have noted a sizable incease in the oganization and political influence of local esidents. Alan Altshule and David Lubeoff (2002) examine the histoy of lage-scale govenment pojects ( Mega-pojects ) and suggest that a change began in the 1960s, when citizens became bette able to challenge lage-scale pojects that would impact thei neighbohood. One ealy and stiking example was Jane Jacobs s leadeship of the Geenwich Village movement that stopped Robet Moses s West Side highway poject in New Yok. Though inceasingly sophisticated use of the media, local goups leaned how to tun mega-pojects into public-elations disastes. Thee is abundant evidence of the impact of homeownes and neighbohood goups, but thee is less undestanding of whee this impact comes fom. One hypothesis is that homeownes have become bette oganized (an incease in ). Some analysts have suggested that the oganizational skills of envionmental goups wee leaned fom the oganizational successes of the civil-ights movement and the anti-wa potests. Eithe though imitation of these ealie goups o because of ising education and media savvy, local esidents have simply become bette at using the media and the couts. The typical esidential activist of 2004 seems much moe skilled than he countepat fom 1955. C. Relative Efficacy of Cash and Time in Influencing Local Decision-Makes Anothe possible hypothesis is that developes ability to use cash to influence local decisionmakes has deceased ove time. This influence histoically has come both fom legal payments, in the fom of campaign donations o legal cash tansfes (i.e., a develope employing a politician fo legal wok), o illegal cash payments o bibes. Zoning envionments may have become moe estictive if developes in the 1960s wee moe easily able to bibe local politicians than they can today. In othe wods, the uban gowth machine descibed by Molotch (1976) has weakened as it has become hade fo developes to tansfe cash to politicians. Thee is some evidence suggesting a decline in couption ove time within the United States. Glaese and Claudia Goldin (2004) use newspape ecods to show a decline in the shae of aticles alleging couption between the late 19th centuy and the mid-20th centuy. Howeve, thei coveage does not show a significant change between the 1960s and the 1990s, which is when esidential pemitting and constuction intensity slowed down. Anecdotes about couption in development abound, and it may be tue that such anecdotes wee moe common in the 1960s than today. While this hypothesis emains plausible, thee is pecious little evidence eithe suppoting o efuting a changing ability to bibe local decision-makes. 2 D. The Value of Amenities Anothe natual explanation fo the ise in estictions on new constuction is that ising income levels have inceased the willingness to pay fo high-amenity neighbohoods and, in paticula, fo low-density neighbohoods (assuming low density is a nomal good, of couse). This hypothesis coesponds to an incease in the paamete u, which the model 2 Even if it wee possible to document such a change, it would be desiable to go futhe and undestand why this change occued. One plausible explanation could be that impovements in the news media have caused moe attention to be paid to coupt deals. A second explanation is that the political influence of local paty machines has declined. These machines may have facilitated the flow of funds fom developes (o anyone else) and ensued that legal epecussions fom local justice would be modest. The decline of local machines might also have played a ole in educing the influence that developes wee able to have on local govenments.
VOL. 95 NO. 2 REGULATION AND THE HIGH COST OF HOUSING 333 pedicts should lead to a decease in pemitting as the incentive of homeownes to spend time to block new constuction ises. The empiical analysis in Glaese et al. (2005b) suggests that iche communities ae less likely to build new housing units, but the magnitude of this effect is not nealy lage enough to explain the decline in pemitting we see in the data. Examining the zoning envionment of vey ich places in 1960 also suggests that ising incomes can only explain a small pat of the change in the pemitting envionment. If the income hypothesis is coect, then pemitting in these places should have been as estictive in 1960 as the entie metopolitan aeas of Boston o New Yok in moe ecent yeas. Howeve, places like New Rochelle, NY, San Mateo, CA, and West Oange, NJ, allowed at least 10 times as much development in the 1950s as metopolitan aeas with compaable incomes today. E. Changes in the Housing Maket A final hypothesis is that the impact of new constuction on housing pices has changed ove time. In the 1950s, housing costs wee low, lowe incomes made people less concened about envionmental amenities, and an absence of constuction in pevious decades may have meant that the quality of new housing was significantly highe than olde units. Fo these easons, new constuction may not have led to majo eductions in housing pices fo existing units, and as such, homeownes had much weake incentives to fight new constuction. In 2004, howeve, homeownes appea to believe that new constuction will significantly educe housing pices. Cetainly, empiical evidence fom ecent decades linking ising housing pices to eductions in constuction suggests that they ae ight. As in the case of the pevious theoies, we have little evidence on the elevance of this theoy and we look to futhe eseach to examine this hypothesis and the othes moe thooughly. REFERENCES Altshule, Alan and Lubeoff, David. Megapojects. Washington, DC: Bookings Institution, 2002. Ellickson, Robet C. Sububan Gowth Contols: An Economic and Legal Analysis. Yale Law Jounal, 1977, 86(3), pp. 385 511. Fischel, William. An Economic Histoy of Zoning and a Cue fo Its Exclusionay Effects. Uban Studies, 2004, 41(2), pp. 317 40. Feund, Eic C. Land Development Management: Revolution and Evolution, in Municipal yea book. Washington, DC: Intenational City Management Association, 1974, pp. 277 88. Glaese, Edwad and Goldin, Claudia. Couption and Refom: An Intoduction. National Bueau of Economic Reseach (Cambidge, MA) Woking Pape No. 10775, Septembe 2004. Glaese, Edwad and Gyouko, Joseph. The Impact of Zoning on Housing Affodability. Economic Policy Review, 2003, (2), pp. 21 39. Glaese, Edwad; Gyouko, Joseph and Saks, Raven. Why Is Manhattan So Expensive? Regulation and the Rise of Housing Pices. Jounal of Law and Economics, 2005a, (fothcoming). Glaese, Edwad; Gyouko, Joseph and Saks, Raven. Why Have Housing Pices Gone Up? National Bueau of Economic Reseach (Cambidge, MA) Woking Pape No. 11129, Febuay 2005b. Molotch, Havey. The City as a Gowth Machine. Ameican Jounal of Sociology, 1976, 82(2), pp. 309 30. Nelson, Robet H. The Rise of Pivate Neighbohood Associations: A Constitutional Revolution in Local Govenment. Unpublished manuscipt pesented at a confeence on The Popety Tax, Land Use and Land-Use Regulation, Lincoln Institute of Land Policy, Washington, DC, 13 15 Januay 2002. Wane, Kee and Molotch, Havey. Building ules: How local contols shape community envionments and economies. Boulde, CO: Westview, 2000.