Working Paper No Commercial Revitalization in Low- Income Urban Communities: General Tax Incentives vs. Direct Incentives to Developers

Transcription

1 Working Paper No Commercial Revitalization in Low- Income Urban Communities: General Tax Incentives vs. Direct Incentives to Developers Li Zou University of Alberta January 202 Copyrigt to papers in tis working paper series rests wit te autors and teir assignees. Papers may be downloaded for personal use. Downloading of papers for any oter activity may not be done witout te written consent of te autors. Sort excerpts of tese working papers may be quoted witout explicit permission provided tat full credit is given to te source. Te Department of Economics, Te Institute for Public Economics, and te University of Alberta accept no responsibility for te accuracy or point of view represented in tis work in progress.

2 Commercial Revitalization in Low-Income Urban Communities: General Tax Incentives vs. Direct Incentives to Developers Li Zou University of Alberta January 20, 202 Abstract Tis paper proposes a commercial development model, based on Fujita s (988) monopolistic competition model of spatial agglomeration, to examine stores decisions to enter urban communities. Te model focuses on commercial developers and large stores, and identi es a potential oldup problem in te commercial development market arising because developers incur costs before negotiating wit ancor tenants over pro t saring; te oldup problem is more likely to occur in low-income communities were te pro tability of commercial projects is small. Te model predicts tat direct incentives to developers are preferred to general tax incentives for addressing tis market failure. JEL classi cation: R58, H50, H76 Key Words: urban redevelopment programs, economic agglomeration, oldup problem Introduction Commercial development is a popular urban revitalization strategy tat as been implemented in many U.S. cities including New York, Cicago, and Boston. Its popularity is founded on te belief tat low-income urban communities are under-served by stores. Residents in tese communities spend a greater sare of teir income sopping outside of teir own neigboroods. For example, in 996 Cicago residents living in te neigboroods of Little Village and Sout Sore made about 62% and 70% of teir purcases outside of teir neigboroods, wile residents of te average Cicago neigborood made 37% of purcases outside (Weissbourd and Berry (999)). Lack of locally available commercial goods and services as a negative impact on te welfare of inner-city residents, usually poor people wo rely on public transportation (Glaeser et al. (2008)). Department of Economics, University of Alberta. li.zou@ualberta.ca. Pone: (780) I tank James Rauc, Julie Cullen, Gordon Hanson, Brad Humpreys, Craig MacIntos, Jos Gra Zivin and seminar participants at te University of California, San Diego, and te University of Alberta for comments and suggestions.

3 According to estimates in New Markets: Te Untapped Retail Buying Power In America s Inner Cities, a report by te Department of Housing and Urban Development (HUD) (Cuomo (999)), U.S. inner-city neigboroods ad an unmet retail demand of $8.7 billion. Despite tis documented under-provision in low-income urban communities, we know very little about te nature of tis problem. Wy don t stores enter tese communities to meet te excess demand? Te answer to tis question as important policy implications. If ig crime rates in inner-city neigboroods is wat deters te entry of stores, ten te government sould put more resources to reduce crime. Consistent wit recent commercial revitalization successes in urban low-income communities, tis paper provides a model of commercial development to sow tat low-income communities may be under-served because of lack of economic agglomeration and policies sould aim at generating necessary commercial agglomeration. A well-known example is Harlem, New York, one of te six Round I federal urban Empowerment Zones. Before Harlem was designated as an Empowerment Zone in 994, it ad not ad any large scale commercial development since te 960s, and 70% of te sopping by Harlem residents was done outside of te neigborood, wic ad not ad any large. Wit public subsidy to developers, te community saw te opening of Harlem USA, a 275,000 square-foot retail and entertainment complex, in 2000 and te opening of East River Plaza, a 485,000 square-foot retail sopping center tat is ome to Costco, Target, Best Buy, Marsalls, Old Navy, PetSmart, Bob s Discount Furniture, and many oter tenants, in In 200, tree developers announced on te same day total planned investment of about $6 million in commercial development projects in Harlem. Te commercial development model in Section 2 builds on existing monopolistic competition models of spatial agglomeration (Fujita (988)), in wic consumers ave a taste for di erent varieties of goods and services (Spence (976), and Dixit and Stiglitz (977)) and ave to incur travel costs to te location were goods and services are provided. For a given travel cost, consumers are more likely to sop at a place wit more varieties of goods and services; is consumption pattern creates positive demand externalities among commercial goods and services tat agglomerate at te same location: more coices attract more consumers and lead to iger revenues (Arakawa (2006), Henkel et al. (2000)). Te rst innovation of tis model is to demonstrate te importance of demand externalities in stores decision to enter a community, an aspect tat, to my knowledge, as not been investigated in te literature. Wit xed costs in commercial provision (e.g., rent paid to a landlord and wages paid to some minimum required number of workers), stores at a location wit a very limited number of varieties may incur losses because of low consumer tra c and revenues. Witout a proper coordination mecanism, te market may fail to acieve necessary commercial agglomeration and end up at an ine cient equilibrium wit under-provision of commercial opportunities despite te existence of a more e cient and pro table alternative equilibrium. Te second innovation is to model te coordination mecanism in te commercial market and to investigate te nature of te under-provision problem in low-income communities. Current commercial develop- Janus Partners LLC and Monadnock Construction Inc. and Artimus 2

4 ments often involve two types of agents commercial developers and large stores. A developer acts as te planner of a commercial development project and large stores serve as ancor tenants. Tis arrangement creates a synergy and is bene cial for bot parties. Witout a developer, a large store will ave no mecanism to carge nearby specialized stores for te positive demand externalities tey generate. Most likely, owners of te properties tat are occupied by tese specialized stores will capture te bene ts troug eiter an increase in rent or property value appreciation. 2 A developer, troug lease contracts wit all stores in a commercial center, will be able to internalize te demand externalities around a commercial center. However, a developer cannot start a project witout ancor tenants, te large stores. Te importance of ancor tenants to a commercial development project is a result of te need for economic agglomeration and te existence of signi cant coordination costs. Recruiting small specialized stores to acieve necessary agglomeration may be a bad option because of ig coordination costs. According to te Urban Land Institution s (ULI) Retail Development Handbook (Kramer (2008), page 9), "as a rule, a sopping center will not be built until te developer as secured commitments from key or ancor tenants." Ancor tenants, aware of teir importance in commercial development, will be able to negotiate wit a developer for a sare of pro ts. Evidence sows tat ancor tenants usually pay muc lower rents tan small tenants. According to te estimate in Pasigan and Gould (998), ancor stores receive a per foot rent subsidy of no less tan 72 percent of te rent tat non-ancor stores pay. Gould et al. (2005) point out tat "te most striking feature of ancor contracts is tat most ancors eiter do not pay any rent or pay only a trivial amount". A successful commercial development requires te cooperation of two powerful parties, but one party, te developer, as to incur signi cant costs before negotiating over pro t saring. Before searcing for suitable ancor tenants, a developer as to perform a market analysis, a site suitability analysis, and a regulatory review, wic can easily cost alf a million dollars (Kramer (2008)). More importantly, te searc process can be costly in terms of time and money. Stores tat can serve as ancors in a commercial center usually ave teir own expansion plans and market targets. For example, in 998, Barnes & Noble decided not to sign a lease wit Harlem USA, te rst large-scale commercial development in Harlem, New York for alf a century, wose developer ad been trying to persuade te bookstore to open a sop in te center since 992. Eventually, Harlem USA signed a bookseller, Hue Man Books in All costs incurred before te negotiation wit ancor tenants are sunk to te developer. At te point of negotiation, ancor tenants ave no incentive to sare tese costs. A developer must anticipate saring te payo s from a commercial development wit ancor tenants, but bear te pre-negotiation costs alone. Tis represents a oldup problem tat arises wen one party must pay te cost wile oters sare in te payo. 3 A commercial project wic generates positive pro ts may not be pro table for a developer and terefore no action will be taken. Tis situation is more likely to appen in low-income communities were te pro tability of a commercial project 2 Tis appens if tere are more potential small commercial business owners tan te spaces tat are close enoug to bene t from te large store s consumer base. 3 Acemoglu and Simer (999) study tis kind of oldup problem in te labor market. Proposed market solutions usually require te agents to take actions (e.g., forming joint ventures) before incurring costs. However, in te case of retail development, te costs must be sunk before te developer meets te ancor tenants. 3

5 is slim to start wit. Te commercial market failure in low-income communities justi es government intervention. Section 3 analyzes ow government policies elp to solve te under-provision problem. Te key is to induce actions by developers. Incentives provided directly to developers for projects in under-served communities, in te form of block grants, low-interest loans, or accelerated deductions, sould work. General tax incentives to capital or labor may be less e ective. In some communities, general tax incentives may not be big enoug to trigger developers actions. If, in some communities, tey do induce developers to invest, tey will also lead developers to over-supply sopping spaces and generate unnecessary revenue loss to ancor stores. Te policy analysis contributes to te growing literature on place based policies. 4 As Gottlieb and Glaeser (2008) and Kline (200) point out, justi cation of place based policies requires better understanding of te structure of agglomeration economies. Tis paper sows tat commercial revitalization incentives as place based policies are welfare enancing. Tere will be more private investment, pro ts, permanent jobs, and sustainable economic activities in communities tat successfully implement tese policies. Analysis of te welfare redistribution e ect of commercial revitalization projects suggests tat ome owners and large stores tat serve as ancor tenants of newly-developed commercial centers are sure winners. Existing community residents wo rent teir ouses may not bene t if te increase in teir consumer surplus from easier access to commercial goods and services drives up rents in te neigborood. 2 A model of neigborood commercial development Tis section develops a model of commercial development to analyze te entry of stores in a community. Following Fujita (988), I assume tat eac individual occupies a residence in te community and derives utility from te consumption of a variety of commercial goods and services. In order to purcase te goods and services, consumers ave to incur a travel cost to get to te location were te goods and services are provided. Te production of eac variety of commercial goods and services is symmetric and requires a xed cost. Te commercial market is composed of commercial developers, multi-variety stores (large stores), and single-variety stores (specialized stores). 2. Consumers Consider a circular community. Housing is omogeneous and uniformly distributed along te circumference of a unit circle. Residents of te community earn identical income y and consume a variety of goods and 4 See Gottlieb and Glaeser (2008) for reviews. 4

6 services x(j) along wit ousing services. Te utility function is 2 u(; x(j)) = 4 Z n 0 x(j) ( )= dj 3 5 = were n is te number of varieties of goods and services available, and > is te elasticity of substitution among varieties of goods and services. Te larger te, te weaker te consumer preference for variety. For simplicity, I normalize te consumption of ousing services to. To purcase any of tese goods and services, consumers incur costly travel to te pysical place were te varieties are o ered. Suppose tere is a sopping center in te community (I will refer to it as "te community sopping center" (CSC)) tat provides n varieties of goods and services. For community residents, te travel cost t to te CSC increases wit te distance z between teir residence and te CSC. Outside of te community tere is a sopping center tat provides n o varieties of goods and services (I will refer to tis sopping center as "te outside sopping center" (OSC)). Te travel cost to te OSC is t o for all community residents. Assume tat, for all community residents, it is always less costly to travel to te CSC tan to te OSC. In oter words, t(z) < t o for z 2 [0; =2]; 2) te OSC o ers weakly more varieties of goods and services tan te CSC, i.e., n o > n; and 3) te price of te same variety is te same in te OSC and te CSC. 5 Te consumption decision of residents consists of two steps: rst tey decide were to purcase goods and services, and second tey decide ow muc of eac variety to purcase. Because residents can buy more varieties at te OSC at te same price, residents wo travel to te OSC ave no incentive to make anoter trip to te CSC. Terefore, no consumer will travel to bot te OSC and te CSC. A consumer eiter travels to te CSC at a lower travel cost but fewer varieties or travels to te OSC for more varieties but a iger travel cost. Te consumers problem is solved by backward induction. Consider a resident wo lives distance z away from te CSC (I will call tis consumer "consumer z"). Suppose se purcases goods and services at te CSC, te er budget constraint will be p + Z n 0 x(j)p(j)dj + t(z) = y were p is te price of ousing services and p(j) is te price of variety j. Formally, er utility maximization problem is Max x(j) 2 4 Z n 0 x(j) ( )= dj 3 5 = s:t: p + Z n 0 x(j)p(j)dj + t(z) y. From te rst order conditions of consumer z s utility maximization problem, we get er demand for con- 5 We can see from te next subsection, wen production of te varieties are symmetric in te OSC and te CSC, te price of te varieties will be te same. 5

7 sumption of variety j 2 were P (n) = 4 Z n 0 p(j) 3 dj 5 x(j; z) = y p t(z) p(j) P (n), is te aggregate price index for goods and services available at te CSC. Consumer z s indirect utility from purcasing goods and services at te CSC will be U(z) = y p t(z). P (n) Te indirect utility tat residents get from patronizing te OSC is 2 3 Zn o were P (n o ) = 4 p(j) dj 5 0 U o = y p t o, P (n o ) is te aggregate price index for goods and services available at te OSC. Consumer z will coose to purcase goods and services at te CSC if U(z) > U o. He will coose te OSC if U(z) < U o. Let ez be te distance tat makes a resident indi erent between te CSC and te OSC, i.e., U(ez) = U o. Witout loss of generality, assume tat travel costs t(z) = tz are a linear, increasing (t > 0) function in z, te distance between a consumer s residence and te CSC. We can solve U(ez) = U o to get ez = y p P (n) P (n (y o) p t o ). t Te resident wo lives ez distance away from te CSC is te marginal customer at te CSC. Residents wo live closer to te CSC tan te marginal customer, z ez, will patronize te CSC. Residents wo live farter to te CSC tan te marginal customer, z > ez will patronize te OSC. 2.2 Production of commercial goods and services Te production of variety j requires one unit of land and a xed labor input (j). In addition to tese xed inputs, eac unit of variety j requires (j) units of labor to produce. Assume te commercial market is small relative to te size of te local economy, so labor is supplied competitively at price w and land is supplied competitively at price r. Te pro t function for producers of variety j at te CSC is (j) = q(j)[p(j) (j)w] (j)w r: were q(j) is te total demand for variety j at te CSC, Te varieties are assumed be produced under monopolistic competition: te supplier of variety j takes P (n) and P (n o ) as given and cooses p(j) to 6

8 maximize its pro t. Te rst order condition for tis te pro t-maximization problem yields an expression for te price of variety j p(j) = (j)w. Assuming tat production of varieties is symmetric at bot te OSC and te CSC, eac variety will ave te same price, p(j) = p = purcased by consumer z is w. Terefore, P (n) = n p, P (no ) = n o p. Te quantity of variety x(j; z) = x(z) () = y p tz p (n p) = y p tz np = y p tz nw. An increase in n reduces x(z), te quantity purcased by eac consumer wo sops at te CSC. Wen tere are more varieties at te CSC, consumers wo sop at te CSC buy more varieties but less of eac variety. Te marginal customer ez at te CSC is ez = y p P (n) P (n o) (y p t o ) t = y p ( n n ) (y p t o ) o : (2) t Equation (2) implies tat te weaker consumer preference for variety (te bigger te ), te smaller te number of varieties te CSC needs to attract te same number of consumers, and tat te more competitive te OSC (te bigger te n o or te smaller t o ), te larger te number of varieties te CSC needs to attract te same number of consumers. Note tat ez = 0 wen n < y p t o no. Tis means tat a CSC needs to o er a minimum y p number of varieties y p t o no in order to attract positive consumer tra c. Also, ez = =2 wen y p n > y p t o y p 2 t no. Tis means tat increasing te number of varieties will not attract more consumers wen all community residents sop at te CSC, wic means tat consumers wo live outside low-income communities will not sop in tese communities, a reasonable assumption for inner city communities. For n 2 [ y p t o no ; y p t o y p y p no ], an increase in n increases te mass of consumers wo sop at te 2 t = ( n n o ) ( ) (y p t o ) n o t > 0. 7

9 Te pro t for producers of eac variety at te CSC is = w w q w r = [2 Z ez 0 x(z)dz] w r (3) Te number of varieties at te CSC a ects te pro ts generated by eac variety at te CSC troug its impact on te quantity purcased by eac consumer and te number of consumers. 2.3 Commercial agglomeration and coordination Commercial agglomeration of commercial goods and services means tat te pro t of eac variety increases wit te number of varieties in a commercial center. an increasing function of te total demand for te variety. agglomeration is demand externalities. Equation (3) sows tat te pro t of a variety is Tis means tat te source of commercial An increase in n increases te consumer > 0) and reduces quantity demanded by eac < 0). Wen te number of varieties is small, te positive impact on consumer mass dominates. Commercial goods and services producers all enjoy te positive demand externality tey generate. Wen te number of varieties is large, te negative impact on per-consumer consumption dominates. For a speci c community, tere exists a number of varieties tat maximizes te pro t generated for producers of eac variety. Substitute equation () and (2) into equation (3), to get te pro t of eac variety provided at te CSC = f w r; n < y p t o (y p ( n no (y p t o tn w r; [(y p ) 4 t] n w r n > y p t o y p no no n y p y p t o y p no 2 t y p t o y p 2 t no : I sow, in Appendix A.3, tat te pro t of a single variety is maximized at n = + 2 y p t o y p no. 6 Te agglomeration economies of commercial goods and services is illustrated in Proposition, wic sows tat before a CSC reaces n, te pro t of eac variety weakly increases wit te number of varieties provided at tis CSC: n = 0 for n < y p t o no y p n > 0 for n 2 [ y p t o no ; n ]. y p Proof. See Appendix A. To understand te importance of economic agglomeration e ects in stores decision to enter a community, imagine a commercial market wit only symmetric single-variety stores ("specialized stores" from now on). We can tink of tese stores as small businesses like candle stores, dry cleaners, bakeries, and oter stores 6 n is te optimal commercial agglomeration for eac variety and is di erent from te optimal commercial agglomeration for a commercial center. 8

10 tat sell one type of product. I will focus on communities wit (n ) > 0. Communities wit (n ) < 0 are not an interesting case because entry of stores is not teir only problem. Tese communities will not be able to sustain a sopping center witout furter government subsidy, even if te government built a CSC wit n varieties of goods and services. We are more interested in understanding wy stores do not enter communities wit (n ) > 0, were tey can earn positive operating pro ts. Proposition 2 sows tat an ine cient outcome can arise in tese communities wen te commercial market is composed of symmetric specialized stores: Proposition 2 If a commercial market is composed of symmetric specialized stores, for a community wit (n ) > 0, no goods and services provided in te community, n = 0, is a Nas equilibrium outcome. Proof. See Appendix A.2. Te intuition is tat a specialized store will not enter a community wen tere is no commercial agglomeration to generate positive pro ts. A market wit only specialized stores will face a coordination problem, wic leads to te ine cient equilibrium outcome of no CSC in communities wit (n ) > 0. Successful commercial agglomeration of specialized stores wit no explicit market coordination mecanism usually rely on te presence of natural amenities suc as beaces or istorical attractions. For communities witout suc natural amenities, tis kind of commercial ourising may not appen. 2.4 A commercial market wit developers, multi-variety stores, and specialized stores Te existence of a pro table alternative equilibrium creates an opportunity for agents wo ave coordination ability. 7 In te commercial market, tese agents are commercial developers and multi-variety stores (large stores). Commercial development processes in current market conditions usually requires te cooperation of a developer and one or more multi-variety stores. Tis arrangement is mutually bene cial because developers and multi-variety stores ave di erent coordination abilities. A multi-variety store contains witin-store agglomeration of varieties and will ave more consumer visits. Commercial developers specialize in identifying a suitable locations, creating te rigt store mix, and managing lease contracts for a commercial project. It may be too costly for a multi-variety store to acquire te planning ability of a commercial developer 8 and for a developer to coordinate small business to te necessary commercial agglomeration. Te commercial development model is based commonly observed commercial development processes wic involve developers, large stores, and specialized stores. A developer is te central planner of a commercial center, large stores are ancor tenants, and specialized stores ll sopping spaces tat are not occupied by ancor tenants. A commercial center is assumed to last for in nite time periods once developed. 7 Rauc (993) demonstrates tat developers of industrial parks can eliminate te ine cient city-industry location equilibrium wen te pro ts are large enoug. 8 One exception is Wal-Mart, wic as a division, Walmart Realty, specializing in commercial development located in or around Wal-Mart Stores. 9

11 All relevant decisions are made in period 0 in tree stages: Stage : a developer makes te decision to enter a community or not. If te developer decides to enter, ten e incurs a cost to identify a suitable site, coose te size of te commercial center, and searces for potential ancor tenants. Stage 2: te developer meets wit large stores tat are potential ancor tenants and negotiates te ancor tenant contracts. Stage 3: te developer purcases land, constructs te sopping center, and leases te rest of te retail spaces to specialized stores. From period on, te commercial center operates and te parties execute te contracts signed in period 0. 9 Te period 0 commercial development game is solved using backward induction. In te tird stage, te developer interacts wit specialized stores. Assume tere are a large number of entrepreneurs wo can manage a specialized store in a CSC. Tese potential small business owners compete for te limited retail spaces available in te sopping center. Te competition allows te developer to o er leases tat extract all te expected pro ts from specialized stores. According to te Kramer (2008), "for a retail center wit major tenants, smaller tenants represent te largest income potential for te project. Altoug major tenants may be te primary generator of customer tra c and te nancial foundation for attracting initial capital commitments, in-line sops pay iger rents and generate te greatest pro ts for te developer." Let r s be te rent tat a specialized store pays for eac period. Since no specialized store makes any pro t, we solve (n) + r r s = 0 and get r s = (n) + r, wic is te maximum rent for a space. Given te size of a sopping center, te developer would like to ave more spaces leased to specialized stores. However, te developer as a constraint on te spaces leased to specialized stores. Suppose at te tird stage, te developer let entrepreneurs bid for slots in te CSC tat are not occupied by ancor stores. If te number of varieties committed to te sopping center is so small tat no store makes positive "beforerent pro t", (n) + r < 0, ten no entrepreneur will be interested in bidding for one slot, assuming tat tere is no negative rent bid, i.e., r s > 0. To start te bidding, a developer as to commit at least n 0 varieties at te second stage suc tat (n 0 ) + r = 0. For n 2 (n 0 ; n ], we know (n) + r > 0. In te second stage, te developer and ancor stores negotiate over te total payo s from te CSC, denoted by V. 0 For a CSC wit n varieties, te total pro t eac period equals n(n). Let i > 0 be te interest rate. Te discounted present value of total pro ts V over te life of te CSC equals +i i n(n). Assume te division of V is caracterized by a Nas bargaining solution. Let 2 [0; ] be te sare tat 9 Renegotiation is not allowed in te model for simplicity. If we allow a positive probability of renegotiation after a developer incurs more site-speci c investment, a developer will be more reluctant to invest in low-income communities because renegotiation allows ancor tenants to get larger payo s. 0 Bargaining over rents ancor tenants pay and over te total payo are teoretically te same. Tere are cases tat ancor tenants do not pay rents and also pay lower operation costs. 0

12 goes to ancor stores. A bigger means more bargaining power for te ancor tenants. Te payo to te developer is ( )V. In te rst stage, te developer incurs a cost c. Te cost c is sunk wen te developer negotiates wit ancor tenants in te second stage, so te developer would not be able to convince te ancor tenants to sare te cost. Te pro t of te developer is V d = ( )V c. Te developer cooses n to maximize pro t: V d = + i ( )n(n) c. i From te rst-order condition of te developer s pro t maximization problem, we get Te developer s maximum pro t is n 2(y p t o = [ t( )(w + r) ] 2 + n + o. V d (n ) = + i i ( )n [ (y p ( n n o (y p t o tn w r] c. Te derivation is in Appendix A.3. If V d (n ) > 0, te developer will incur te cost c to initiate te development of a CSC. If V d (n ) < 0, te developer will not initiate te project. Proposition 3 sows tat a developer s maximum pro t from a CSC increases wit te average level of income in te community wen consumers preference for variety is not too weak. Proposition 3 dv d(n ) dy = ( ) dv (n ) dy > 0 wen 2(y p ) t o. Proof. See Appendix A.4. In reality, income per square mile is an important determinant of project revenue and is frequently found in feasibility reports for commercial development projects (Kramer (2008)). Oter tings equal, a community wit iger income per square mile will ave larger purcasing power and a iger demand. It is not surprising tat commercial developments in ig-income communities are more pro table. Tis means tat, if make simultaneous developments in all communities are impossible due to exogenous constraints, ricer communities will ave a iger priority and get commercial development earlier. Tis result alone does not imply tat low-income communities will persistently be under-served by commercial goods and services suppliers. Persistent under-provision may appen in low-income communities due to te oldup problem tat arises wen payo s are sared wile costs are not. In te commercial development game, te total payo s from a CSC are sared by te developer and ancor stores but pre-negotiation costs are borne entirely by te developer. Proposition 4 sows tat, in some communities, a CSC may be pro table but will not get built because te developer s sare of te payo will not cover te pre-negotiation costs:

13 Proposition 4 For y 2 [y ; y 2 ) suc tat V (y ) c = 0 and V d (y 2 ) = ( )V (y 2 ) c = 0, V (y) c > 0 and V d (y) < 0. Proof. See Appendix A.5. Communities wit income level between y and y 2 will be under-served by stores because potentially pro table commercial developments are eld up by ancor tenants, wo will grab a sare of te payo s troug ancor contracts and will not sare te costs tat are sunk to a developer before te negotiation. Wile tis beavior will not deter a developer s decision to invest in communities wit y > y 2, because a developer will still make positive pro t (V d (y) > 0 for y > y 2 ) on tese projects, it will deter a developer s decision to invest in communities wit y 2 [y ; y 2 ) even toug a CSC will be sustainable and pro table in tese communities. Tis market failure can be addressed by government policies. 3 Urban development policies 3. Incentives to developers Te market failure in communities wit income level between y and y 2 justi es government intervention. Based on te model, te key to address tis market failure is to increase te developer s returns to make te developer at least break even on te project. First, consider a lump-sum grant g to a developer. Te impact of a lump-sum grant is summarized in Proposition 5: Proposition 5 For a community wit y 2 [y ; y 2 ), a lump-sum grant g > te developer to build a CSC in tis community and generate V d (y) to a developer will induce. private investment from te developer +i i rn + c; 2. permanent jobs n Q(n ) + n ; 3. sales revenues +i i 4. net pro ts V (n ) c > 0; n Q(n )w; 5. an increase in consumer surplus eac period 2 R z (n ) 0 [U(n ; z) U o ]dz > 0. Proof. See Appendix 5. Oter incentives, for example low-interest loans and accelerated deductions, to a developer wit a value equivalent to g can also induce te developer to act in under-served communities, like a lump-sum grant. Once a CSC is establised in a community wit y 2 [y ; y 2 ), it will operate pro tably witout furter government subsidy, and generate permanent jobs and sustainable economic activity in tis community. Te increase in consumer surplus does not guarantee a welfare increase for existing community residents. 2

14 Proposition 5 suggests tat providing incentives directly to a developer is welfare enancing due to te positive net pro ts generated and te increase in consumer surplus. Te developer s investment in land and te jobs created at te CSC are not considered net gains for a community. In reality, attracting private investment, creating jobs, and stimulating sustainable economic activities in distressed urban communities are important policy goals and included in empirical evaluation standards for local development programs, because low-income urban communities often su er from ig unemployment rates, deterioration of real property, and low levels of economic activity. Commercial revitalization troug direct subsidy to developers as been used successfully in a number of US federal urban Empowerment Zones. As discussed in te introduction, commercial development spurred troug grants from an Empowerment Zone as revived Harlem. Te success of commercial revitalization as a place-based policy, wic empasizes positive canges in a targeted location, does not guarantee its success as a people-based policy, wic empasizes welfare improvement for people living in a targeted location. A closer look at te welfare redistribution of te net pro ts generated by a CSC sows tat owners of ancor stores of a CSC are sure winners because tey get a positive pro t of V (n ), and tat te developer is likely to be a winner if te size of te block grant g exceeds te loss c ( )V (n ). Owners of specialized stores make zero pro t due to competition for te limited retail spaces in te CSC. Generally, owners of specialized stores are more likely to be residents of te community tan owners of ancor stores and developers, wic means te pro t from commercial revitalization is more likely to go to people wo are not residents of te community. Te distribution of te net gain from te increased consumer surplus depends on te ownersip of local residential property and te local residential rental market. Residents wo own teir ome keep te consumer surplus wile tey live in teir ouses and capitalize te consumer surplus into te price wen tey sell teir omes. Residents wo rent will be able to keep te increase in consumer surplus from sopping at te CSC, U(n ; z) U o, if tere are enoug vacant residential rental properties in te community. Oterwise, tey may not bene t from te CSC because te increase in consumer surplus is location dependent, and can be extracted by property owners troug an increase in rents. Te welfare of existing residents wo rent can decrease if commercial revitalization induces gentri cation and forces tem to move out of te community. 2 In 2008, te black population in central Harlem was about 77,000, te lowest since te 920s, wile total population increased from about 09,000 in 2000 to 26,000 in Te relationsip between te decline of te black population and te commercial revitalization success in tis community is an issue tat warrants furter investigation. 2 Te communities are assumed to be closed to immigration, so I cannot analyze gentri cation in tis model. I make te assumption of closed communities to investigate te impact of income-segregation on te provision of commercial goods and services. An alternative model includes a distribution of communities sorted by ousing prices and an income distribution for residents. A CSC in a low income community can induce some residents of ig-income ig-ousing-price communities to move to low income communities because of te lower ousing prices, wic will drive up te ousing price in te low income community and force existing residents of te low income community in te lower tail of te community income distribution to move. 3

15 3.2 General tax incentives Incentives provided directly to commercial developers ave been used by spatially targeted economic development programs, suc as te federal Community Renewal Initiative tat was began in 994. In addition to Employment Tax Credits, te six Round I federal urban Empowerment Zones tat participated in te Community Renewal Initiative eac received $00 million in te form of Title XX Social Services Block Grant (SSBG) Funds, some of wic were used as incentives to commercial developers. In December 2000, te Community Renewal Tax Relief Act autorized 28 urban (and 2 rural) Renewal Communities. Eac designated community may autorize up to $2 million in Commercial Revitalization Deduction (not more tan $0 million per project) eac year from December 3, 200 to January, 200. General tax incentives are more commonly used incentives in spatially targeted economic development programs. Before te federal Community Renewal Initiative program began in 994, 37 states and te District of Columbia ad enacted Enterprise Zone programs. Eac of tese programs ad elements unique to te states, but te most common elements included in tese programs were general tax incentives. Evaluation of te Round I federal urban Empowerment Zones (Hebert et al. (200), Busso et al. (200)) found more encouraging outcomes tan assessments of state enterprise zones, wic generally failed to nd signi cant improvements. Busso et al. (200) report evidence tat some of te positive improvements in te Round I federal urban Empowerment Zones may be te result of block grants rater tan tax credits, and empasize tat disentangling te e ectiveness of block grants and employment tax credits is necessary to reconcile te assessment of te Round I federal Empowerment Zones and te state enterprise zones. Te following analysis o ers an explanation for te di erence in te context of urban commercial revitalization projects. Assume tat employment tax abatements take te form of tax rebates, i.e., tey e ectively lower labor costs regardless of te pro tability of a rm. 3 Let s be an e ective wage subsidy tat is equivalent to te employment tax rebates. Under employment tax rebates, te e ective labor costs of te stores are w( Te developer will coose n s 2(y p t o = [ t( )[w( s) + r] ] and will enter tis community if V d (n s ; w; r; y) > 0. + n 2 + o > n ; In tis context, employment tax abatements are inferior to incentives to developers for two reasons. First, conditioned on inducing commercial developments, te minimum revenue costs to government using employment tax incentives will be muc larger tan te equivalent minimum incentives to a developer, because, under employment tax incentives, te government will also lose tax revenue to ancor tenants and developers will over-supply retail spaces (n s > n ) in order to get more tax bene ts. 3 Employment tax credits in federal urban Empowerment Zones can only be claimed against taxable pro ts of a rm tat employs workers. In tis model, specialty stores do not make any pro t and te developer does not employ any workers (once te commercial center is developed). None of tem would be able to access te employment credit under te current market structure. s). 4

16 Second, and more importantly, if s is not large enoug, V d (n s ; w; r; y) will not be su ciently positive and will not induce developers to invest in under-served communities. We can ardly expect suc policies to work in under-served urban communities, because te size of te employment tax credits is determined at te federal level and is uniform across all empowerment zones. Tis one-size- ts-all federal policy is unlikely to be te rigt size for many under-served communities. If employment tax incentives induce developers to invest in some communities, tese communities will likely ave an income level close to y 2. Tis prediction is consistent wit empirical evidence from te literature on te e ectiveness of general tax incentives, wic suggest tat general tax incentives are most likely to work in areas wit lower unemployment rates and iger income to begin wit (Goss and Pillips (999) and Goss and Pillips (200)). Altoug general tax incentives can induce commercial development in under-served communities, te model developed ere predicts tat policies based on incentives to developers may be more e ective. Tis prediction explains observed outcomes in under-served urban communities based on te assessment of state and federal urban revitalization policies. 4 Conclusion Te commercial development model developed ere takes into account te importance and costs of economic agglomeration in te entry decision of stores, and explains te documented under-provision of commercial goods and services in low-income urban communities in te U.S.. It as a number of important implications for commercial revitalization policy in urban communities. Clearly, if te problem facing low-income communities is tat a developer s sare of te payo s generated by a commercial project cannot cover te pre-negotiation costs, an e ective government policy is to provide incentives to a developer for projects in under-served communities. General tax incentives alone are less likely to trigger commercial developers investment in low-income communities, and may induce a developer to over-supply retail sopping spaces if te tax incentives are large enoug to induce te entry of a developer. Wile te policy discussion ere is based on commercial development, it also as implications for spatially targeted economic development programs. If te problem in distressed urban communities stems from di culty acieving economic agglomeration, te existence of wic as been generally accepted in te economic geograpy literature, te general tax incentives appear to be unlikely to spur sustainable economic development because of te non-linear nature of te agglomeration e ect. Note tat te failure of general tax incentives does not invalidate place based policies as a wole. Policy makers sould adopt policies tat trigger a non-linear increase in economic activities in order to generate te agglomeration economy needed to gt urban decline (GREENSTONE et al. (200)). Te literature on urban poverty (e.g., Kain (968) and Wilson (987)) as focused on te spatial mismatc between employment opportunities and ousing options for urban residents, wic leads to iger unemployment and lower income in inner city communities. Tis model sows tat te welfare of inner 5

17 city residents may be furter reduced due to te under-provision of commercial goods and services in teir communities. Te impact of te spatial mismatc between consumption opportunities and ousing options faced by low-income urban residents may be signi cant 4 and is an area for furter investigation. 4 A growing public ealt literature (e.g., Larsen and Gilliland (2008) and Larson et al. (2009)) suggests tat low-income urban residents ave reduced access to supermarkets and tend to ave less ealtier diets and iger levels of obesity. 6

18 References Acemoglu, D. and Simer, R. (999). Holdups and e ciency wit searc frictions. International Economic Review, 40(4): Arakawa, K. (2006). A model of sopping centers*. Journal of Regional Science, 46(5): Busso, M., Gregory, J., and Kline, P. (200). Assessing te incidence and e ciency of a prominent place based policy. Tecnical report, National Bureau of Economic Researc. Cuomo, A. (999). New markets: Te untapped retail buying power in americašs inner cities. Wasington, DC, Government Printing O ce, 495. Dixit, A. and Stiglitz, J. (977). Monopolistic competition and optimum product diversity. Te American Economic Review, 67(3): Fujita, M. (988). A monopolistic competition model of spatial agglomeration* :: Di erentiated product approac. Regional Science and Urban Economics, 8(): Glaeser, E., Kan, M., and Rappaport, J. (2008). Wy do te poor live in cities te role of public transportation. Journal of Urban Economics, 63(): 24. Goss, E. and Pillips, J. (999). Do business tax incentives contribute to a divergence in economic growt? Economic Development Quarterly, 3(3):27. Goss, E. and Pillips, J. (200). Te impact of tax incentives: do initial economic conditions matter? Growt and Cange, 32(2): Gottlieb, J. and Glaeser, E. (2008). Te economics of place-making policies. Brookings Papers on Economic Activity, 2008(): Gould, E., Pasigian, B., and Prendergast, C. (2005). Contracts, externalities, and incentives in sopping malls. Review of Economics and Statistics, 87(3): GREENSTONE, M., HORNBECK, R., and MORETTI, E. (200). Identifying agglomeration spillovers: Evidence from winners and losers of large plant openings. Journal of political economy, 8(3): Hebert, S., of Housing, U. S. D., and Development, U. (200). Interim assessment of te empowerment zones and enterprise communities (EZ/EC) program: a progress report. US Dept. of Housing and Urban Development, O ce of Policy Development and Researc. Henkel, J., Stal, K., and Walz, U. (2000). Coalition building in a spatial economy*. Journal of Urban Economics, 47():

19 Kain, J. (968). Housing segregation, negro employment, and metropolitan decentralization. Te Quarterly Journal of Economics, 82(2):75. Kline, P. (200). Place based policies, eterogeneity, and agglomeration. American Economic Review, 00(2): Kramer, A. (2008). Retail Development Handbook. Urban Land Institute. Larsen, K. and Gilliland, J. (2008). Mapping te evolution of food deserts in a canadian city: Supermarket accessibility in london, ontario, International Journal of Healt Geograpics, 7():6. Larson, N., Story, M., and Nelson, M. (2009). Neigborood environments:: Disparities in access to ealty foods in te us. American journal of preventive medicine, 36():74 8. Pasigan, B. and Gould, E. (998). Internalizing externalities: te pricing of space in sopping malls. JL & Econ., 4:5. Rauc, J. (993). Does istory matter only wen it matters little? te case of city-industry location. Te Quarterly Journal of Economics, 08(3):843. Spence, M. (976). Product selection, xed costs, and monopolistic competition. Te Review of Economic Studies, 43(2): Weissbourd, R. and Berry, C. (999). Te market potential of inner-city neigboroods: Filling te information gap. Tecnical report, Wasington, DC: Brookings Institution. Wilson, W. (987). Te truly disadvantaged: Te inner city, te underclass, and public policy. Cicago, London. 8

20 A Appendix A. Proof of Proposition Wen n < y p t o y = 2 = 0. Wen n = n o ( n n o (y p t o tn = + ( n n o (y p t o (y p tn 2 Te = 0 yields n = 2 + y p t o y p no, (y p ( n n o (y p t o tn 2 y p t o y p no 2 (n ) = 2(y p tn 3 ( ) < 0. A.2 Proof of Proposition 2 Te decision facing a specialized store is to enter a CSC or not. A store will coose to enter if it makes a positive pro t. For communities wit (n ) > 0, tere exists an en 2 ( y p t o y p no ; n ] suc tat (en) = 0 and (n) = f < 0, for n < en > 0, for en n n In order for te suppliers of variety to make positive pro ts, a CSC as to provide at least en varieties. If n < en, a specialized store will coose not to enter a CSC because (n) < 0. Terefore, n = 0 is a Nas equilibrium. A.3 Te developer s pro t maximization problem Te developer s pro t is V d = +i i ( )n(n) c. Take te rst derivative of V d wit respect to n, we get 9

21 (n) = [ (y p ( n n o (y p t o w r] tn + +n ( n n o (y p t o (y p tn 2 = (y p ( n n o (y p t o w r + tn 2 = ( n n o (y p t o w r. tn Solving te rst-order condition, (n) = 0, we get n 2(y p t o = [ t( )(w + r) ] + ( n n o (y p t o (y p + n 2 + o Te maximum pro t of te developer from a CSC in tis community is V d = + i i = + i i A.4 Proof of Proposition 3 ( )n [ (y p ( n n o (y p t o tn ( ) (y p ( n n o (y p t o t I use te envelope teorem to sow tat dv d dy > 0 wen 2(y p ) t o. We know tat n dv d dy d(n = + i i tn w r] c + i ( )n (w + r) c i 2( )[(y p ) ( n n o (y p t o )] t > n because te developer cooses n to maximize te total pro ts of n varieties supplied, not te pro t of one variety, terefore dv d dy > 0 wen 2(y p ) t o. (y p ) ( n n o (y p > (y p ) ( n n o (y p = (y p ) = (y p )[ t o ) t o ) (y p (y p t o + (y p t o ) y y p t o + ] p > 0 if 2(y p ) t o 20

22 A.5 Proof of Proposition 4 V (y) c > 0 for y > y because dv dy > 0 and V (y ) c = 0: V d (y) < 0 for y y 2 because dv d dy = ( ) dv dy > 0 and V d(y 2 ) = ( )V (y 2 ) c. y = V (c) < y 2 = V ( c dv ) because dy > 0 and c < c. A.6 Proof of Proposition 5 If y 2 [y ; y 2 ), wit a subsidy g > V d (y), te developer s pro t will be V d (y) + g > 0. Terefore, te developer will purcase land wit a value of +i i rn and incur cost c to build a CSC tat provides n variety of goods and services. Every period, te producer of eac variety employs q(n ) + workers and generates revenue q(n )p = q(n )w. Since specialized stores do not make pro t, we nd te total pro t by adding up te pro t of te developer and te ancor tenants, wic equals V d (n ) + g + V (n ) = V (n ) + g c > g. Consumers z 2 [0; z (n ] gain utility from patronizing te CSC, U(n ; z) aggregate consumer surplus due to te CSC is 2 R z (n ) 0 [U(n ; z) U o ]dz > 0. U o > 0. Every period, te 2

23 Department of Economics, University of Alberta Working Paper Series ttp:// : Native Students and te Gains from Exporting Higer Education: Evidence from Australia - Zou : Te Overpricing Problem: Moral Hazard and Francises Eckert, H, Hannweber, van Egteren 202-0: Institutional Factors, Sport Policy, and Individual Sport Participation: An International Comparison Humpreys, Maresova, Ruseski 20-23: Te Supply and Demand Factors Beind te Relative Earnings Increases in Urban Cina at te Turn of te 2 st Century Gao, Marcand, Song 20-22: Tariff Pass-Troug and te Distributional Effects of Trade Liberalization Ural Marcand 20-2: Te Effect of Parental Labor Supply on Cild Scooling: Evidence from Trade Liberalization in India Ural Marcand, Rees, Riezman 20-20: Estimating te Value of Medal Success at te 200 Winter Olympic Games Humpreys, Jonson, Mason, Witeead 20-9: Riding te Yield Curve: A Spanning Analysis Galvani, Landon 20-8: Te Effect of Gambling on Healt: Evidence from Canada Humpreys, Nyman, Ruseski 20-7: Lottery Participants and Revenues: An International Survey of Economic Researc on Lotteries Perez, Humpreys 20-6: Te Belief in te Hot Hand in te NFL: Evidence from Betting Volume Data Paul, Weinbac, Humpreys 20-5: From Housing Bust to Credit Crunc: Evidence from Small Business Loans Huang, Stepens 20-4: CEO Turnover: More Evidence on te Role of Performance Expectations Humpreys, Paul, Weinbac 20-3: External Balance Adjustment: An Intra-National and International Comparison - Smit 20-2: Prize Structure and Performance: Evidence from NASCAR Frick, Humpreys 20-: Spatial Efficiency of Genetically Modified and Organic Crops Ambec, Langinier, Marcoul 20-0: Unit Root Testing wit Stationary Covariates and a Structural Break in te Trend Function - Fossati 20-09: CDS as Insurance: Leaky Lifeboats in Stormy Seas Stepens, Tompson 20-08: Game Attendance and Competitive Balance in te National Hockey League Coates, Humpreys 20-07: Rationalization in te Canadian Retail Gasoline Industry: Te Role of Environmental Regulations Eckert, H; Eckert, A 20-06: Pysical Activity and Healt Outcome: Evidence from Canada Humpreys, McLeod, Ruseski 20-05: Dating U.S. Business Cycles wit Macro Factors Fossati 20-04: Covariate Unit Root Tests wit Good Size Power Fossati 20-03: New measures of te costs of unemployment: Evidence from te subjective wellbeing of 2.3 million Americans Helliwell, Huang Please see above working papers link for earlier papers