Book review
Masahisa Fujita,Jacques-Franc ¸ois This,Economics of Agglomeration:Cities,Industrial Location and Regional Growth,Cambridge University Press,Cam-bridge,2002.
This book explores the economic reasons for and conquence of firms and houholds cluster in regions,cities,and commercial districts.It provides compre-hensive characterizations of the patterns and scope of the clusters when generated under alternative market structures and evaluates the efficiency and equity implications of the resulting allocations.Adopting Papageorgiou’s [10]view that it would be futile to look for the model explaining different types of economic agglomerations,the book us large variety of sophisticated models to explore agglomerations in different b geographical scales Q and b degrees of ctoral details Q .A substantial part of the analysis is bad on earlier works by the authors and other relevant literature.In particular,the book is cloly related to that of Fujita et al.[2].However,the main concerns of the two books and their framework of analysis differ considerably.Fujita et al.[2]endeavors to integrate urban economics,regional science,and international trade in a unified framework bad on the emergence of agglomeration under monopolistic competition due to a preference for varieties and scale economies on the firm’s level.The book under review is mainly concerned with the phenomenon of agglomeration per with
a special focus on cities.Furthermore,it adopts a much wider perspective that allows it to explore the reasons for the emergence of agglomeration under alternative market structures.In its quest to provide comprehensive treatment of the issue,the authors rely heavily on the previous studies on agglomeration in the urban economics and industrial organization literature than Fujita et al.[2].Not less importantly,this book is concerned with the normative aspects of agglomeration,focusing on market failure and the distributive implications of the agglomeration process and its association with economic growth.The issues were completely ignored by Fujita et al.[2],who relegated them to subquent studies.
民间恐怖小故事The book is organized into four parts.The first part,which is subdivided into five chapters,deals with the fundamentals of geographical economics.Chapter 2mainly concerns the failure of the (perfect)competitive paradigm to explain the emergence of specialization,trade and,conquently,agglomeration when space is homogenous (Starrett’s b spatial impossibility theorem Q ).It also discuss the two welfare theorems when equilibrium is viable (e.g.,when the landscape is heterogeneous).
doi:10.sciurbeco.2004.07.001
Regional Science and Urban Economics 35(2005)584–
592
Book review585 Chapter3deals with von Th q nen’s agricultural model and its application to the monocentric urban form paradigm.The imposition of a city which rves as a market or a CBD violates the premi of the spatial impossibility theorem that the landscape is homogeneous and,thus,allows the emergence of an efficient competitive market allocation.
Chapter4concentrates on the fundamental trade-off between scale economies and transportation in generating agglomeration and determining its scope.When communities can capitalize the land rent into their payoff,they can offt the loss associated with scale economies and marginal cost pricing(the Henry George Rule).Then competition among profit-maximizing entrepreneurs yields an optimal allocation.Two alternative micro-economic foundations are ud to explain the reduced form reprentation of scale economies in the city’s aggregate production function.The first is the advantage of large varieties of intermediate goods which are accessible in large cities.The cond is the advantage of large cities in allowing a better match of heterogenous workers and firms’jobs which leads to higher wages.Higher wages compensate the workers for the higher cost of living in large cities,caud by the higher locational costs(commuting and rent). This chapter also discuss spatial competition and how the introduction of land market may help to eliminate the markup and th
e excessive number of firms.1 The first part of Chapter5differs from Chapter4in that the scale economies inherent in the provision of a pure local public good replaces the scale economies in the aggregate production function.The chapter also elaborates on the determination of number and size of public facilities under politics.City residents vote on the number of facilities and their location(in two steps)and the facilities are financed by a proportional income tax.Without land market,the voting outcome maximizes the Rawlsian but not the Benthamite social welfare function.With land market and free mobility,individual’s utility is equalized across locations and the equalized utility is maximized.
The cond part of this volume,which is subdivided into two chapters,is concerned with the structure of a metropolitan areas.Using the findings of a ries of earlier studies, which are bad on Solow and Vickrey’s[11]modelling of communication interactions, the chapter explains the emergence of symmetrical unimodal rent and density functions in urban areas,both in residential cities,when the interactions involve individuals and in the CBD when the interactions involve firms.The chapter continues by examining the effect of two types of interactions:communication interactions among firms and interactions between individuals and firms(employment).The fundamental dilemma is the attraction of firms to each other in order to reduce the impediment of distance on the marginal produ
ctivity of communication,on the one hand,and the attraction of employees’housing to firm’s location to save on commuting costs,on the other.In determining its location, each firm maximizes profit given the locational variation of land rent,wage,and the aggregate net advantage to productivity of communication with other firms.In determining residential location,the individual maximizes utility given the locational variation of wage and land rent.Depending on the functional specifications and 1The introduction of land market,however,does not allow full decentralization through price-taking competition;an appropriate institution(a planner)is still required to allow each entrant to fully capitalize the increment in aggregate land rent given the predetermined expenditure of the consumers.
parameters,the model yields alternative land u patterns,including monocentric,incompletely integrated land u,completely integrated land u,and veral variations of polycentric cities.Independently of the specification,however,the model implies one common result:given the parameter reprenting the impediment of distance on the marginal productivity of communication interactions,a monocentric configuration emerges when commuting costs are sufficiently low and full integration emerges when they are sufficiently high.On a normative level,due to the external effect of communication (the agent ignores the advantage of its location decision on the benefit of communic
ation interactions accruing to other agents),the spatial equilibrium tends to be insufficiently agglomerated.
Chapter 7discuss clustering of firms within cities under monopolistic and oligopolistic competitions,illustrating that imperfectly competitive markets are a major reasons for the existence of agglomeration.The first model describes how alternative residential and commercial land u patterns are generated under monopolistic competition when the reprentative consumer’s utility (entropy-type specification)exhibits preference for varieties.In maximizing its objective function,each type of agent is spatially attracted to each agent of other type,whereas each agent competes on land with other agents,whether belonging to his own type or not.2There are two basic equilibrium land u patterns,mixed land u surrounded by either exclusive residential or exclusive business land u.The one that prevails depends on the average number of customers per store.In each ca,the equilibrium and the first-best land u patterns are the same.3
The cond model is an extension of the Hotelling [5]spatial competition problem.Firms compete on price and location,where the utility of a reprentative consumer exhibits preference for varieties (entropy-type specification).A sufficient product differentiation and low transport costs yield clustering of all the firms in the market center under Nash equilibrium.The reason being that product differentiat
ion relaxes price competition in the cluster and low transport costs implies that only a small gmented market can be cured by moving away from the cluster.This concentration is optimal and,with free entry of firms,their equilibrium number is also clo to optimum.4Location concentration under Nash equilibrium may,however,be higher than first-best optimum,if transport costs are sufficiently high and or product differentiation is sufficiently low.
The third main issue,studied in Section 7.4,is clustering of shops when heterogeneous customers do not know the variety each store offers and,therefore have to travel for arching among stores before choosing where and what to buy.Formally,consumers are distinguished by two parameters:location in a linear geographical space and idiosyncratic ideal variety in a circular characteristics space.The consumers are exogenously distributed on the envelope of a cylinder that portrays the geographical-characteristics space.The issue is to characterize the firms distribution in the geographical space under Nash equilibrium when only the firms’locations is known and the range of varieties offered in 2
充实的英文This description deviates somewhat from verbal description in the book.
3This does not necessarily imply that the Nash equilibrium land u pattern is efficient becau,given the markup pricing,efficiency requires a cond best land u pattern,not a first-best.情感测试
4Once again,it is not necessarily efficient becau,given the markup,a cond-best allocation is required.Book review
586
Book review587 the market,but not the variety offered by each firm.Lack of information about the address where each variety is offered and rational arch strategy is shown to yield clusters of stores which need not be at geographical center.
The formation of employment centers in a land u model is the subject of Section7.5. There is a continuum of identical exporting firms,each lling its output on the world market at a fixed price.Each firm us a fixed amount of labor and land and variable quantities of differentiated intermediate goods.The contribution of each intermediate good to the output of an exporting firm is reprented by an entropy-type function of its quantity.It follows that the exporting firm’s output increas with the mass of the intermediate goods.This model yields similar results to that of the residential–business competition model in Chapter6when we substitute exporting firms and specialized firms for housing and firms,respectively.Thus,integrating the two models allows higher resolution of urban mixed land u and a richer gallery of land u patterns than suggested by each model parately.
The third part,which is subdivided into two chapters,discuss inter-regional industrial location.Chapter8discuss three models that portray industrial agglomeration under Marshallian externalities.According to the first,the centripetal force is generated by the positive scale effect of mobile skilled workers and the centrifugal force is created by the negative crowding effect of the aggregate local labor force,which includes both the mobile skilled and immobile unskilled workers.Each firm’s production function is a product of a CRS function of skilled and unskilled workers and a positive external effect of the aggregate skilled workers in the region.Each firm takes the external effect as given and, therefore,considers its production function to exhibit a CRS production function.Hence, the allocation is determined competitively.Skilled workers are attracted to that region which offers them higher utility.Stability implies that if the external effect of skilled workers is sufficiently strong,the skilled workers are(partially)concentrated in one region,and unskilled workers in that region earn a higher wages than unskilled workers in the other region.Otherwi,only a symmetrical allocation is a stable equilibrium.The cond and the third models discuss cas where immobile population with quadratic utility is distributed equally between two regions.The analysis in both models focus on the effect of localization economies on the firm’s location decision.In the cond model, it is assumed that the advantage of localization economies varies between the regions and two firms compete strategically in price and location.The first-best allocatio
n implies that if transport costs are sufficiently low,the two firms should be located in the region where the localization economies are larger,whereas strategic competition leads to such an agglomeration only if the localization economies in that region are sufficiently large.The third model examines the effect of localization economies with a continuum of firms each of which produces differentiated goods.The model implies that when transportation costs decline,the allocation is transformed from perfect symmetry to agglomeration.For sufficiently low transport costs,all the firms agglomerate in one region.The first-best allocation is never less agglomerated than the market equilibrium.
沙尔曼Chapter9is devoted to the b new economic geography Q where agglomeration in the form of core–periphery structure is generated under monopolistic competition by preference for varieties and scale economies on a firm level.Beginning with the original version of the core–periphery model where the subutility of differentiated products is a
CES function and the transport costs are reprented by melting iceberg,it is shown that agglomeration,reprented by a core–periphery structure,requires low transport costs.This obrvation is also corroborated by a modified model in which the subutility of the differentiated products is quadratic and the transport cost of a unit shipped between the regions is fixed in terms of
the numeraire.In contrast to the original version,however,the modified version of the core–periphery model lend itlf to welfare analysis.It is shown that agglomeration helps the mobile skilled workers who concentrate in the core and the immobile unskilled workers there,but hurts the immobile unskilled workers in the periphery.The loss of the unskilled workers in the periphery may outweigh the gain of the unskilled workers in the core.The cond version of the core–periphery model is extended to include a mobile intermediate ctor and perfect foresight.
The fourth part of the book,which is subdivided into two chapters,is concerned with urban system and regional growth.Chapter 10discuss the stability of von Th q nen’s monocentric structure of and the evolution of the urban system when the population grows.The stability of the monocentric structure depends on low transport costs of the agricultural product relative to the cost of moving manufactured goods.When the population increas,the demand for agricultural products increas and the agricultural hinterland expands.The local prices of differentiated goods increa in the periphery,which attract manufacturing firms and,hence,new cities are established.By adding an intermediate goods ctor to the model,a rich variety of results can be derived such as the emergence of a regular pattern of cities in accordance with some characteristics of the central place theory.Yet,the model does not yield a perfect hierarchy where each city trades with a limited t of cit
ies and its agricultural hinterland.Rather,each city,which,according to the prent model,specializes in its unique t of differentiated goods,trades with each of the other cities.
The volume culminates in Chapter 11,which synthesizes the core–periphery model and growth theories and evaluates the prevailing policy aimed at restricting agglomeration.The basic idea is that knowledge is accumulated at a rate equal to the aggregate production of patents by skilled workers.It is assumed that the productivity of patents production is (indirectly)proportional to the stock of knowledge,where the coefficient of this proportion in a given region increas with its share in the aggregate skilled workers (reflecting the decay of personal knowledge spillover with distance).This assumption underlies the crucial association between the rate of increa of knowledge (aggregate output of patents)and agglomeration.Each new patent allows the establishment of a new firm which produces a new brand of differentiated goods.Hence,the rate of aggregate increa of varieties,which is a measure of the rate of growth,is incread with agglomeration.Solving the steady state growth path,it is shown that,with perfectly mobile patents,the R&D ctor as well as a substantial part of manufacturing firms agglomerate in one region and when the transport cost is sufficiently low,both the R&D ctor and the entire manufacturing ctor agglomerate in one region.Under a core–periphery steady-state growth path,the welfare of unskilled
workers in the core exceeds that of unskilled workers in the periphery.The welfare of the latter may,but need not,be hurt relative to what they enjoy under a symmetric growth path.Yet,a dispersion policy is likely to reduce global economic growth and may even hurt the unskilled workers in the periphery.This policy,however,may be justified when the criterion of horizontal equity is adopted and there is no way to compensate the workers in the periphery.
Book review
588
Book review589 This volume provides the most comprehensive study of the fundamental reasons for agglomeration and the market incentives that drive agents to agglomerate.It is distinguished from earlier studies by its special focus on intra-urban agglomeration, integration of agglomeration analysis in urban and regional economics as well as industrial organization literature,novel exploration into normative issues associated with agglom-eration and growth,and last but not least,integration of agglomeration with growth models.No student of agglomeration can do without this study.Yet,I think that a coherent epilogue(Chapter12)is missing in this book.Indeed,at the book’s outt(page2),the reader is b warned Q not to expect to find the theory of agglomeration.Rather,citin
花青蛙g Papageorgiou’s[10]conjecture that each type of agglomeration requires a different model, the authors offer a rich gallery of models,each adapted to different scale and ctor. Accordingly,the quest to generalize results is confined to the concluding comments of the relevant chapters with only occasional cross reference to other similar models or models that yields opposing results.I do think that even in the ca portrayed by Papageorgiou [10],a coherent epilogue is required for prenting the most conci t of explanations for the emergence of agglomeration,categorizing them,and surveying the main results derived in the book,while spelling out what is common and what is not and why.I will elaborate on this in the quel.
First,consider the following list of cited statements regarding the reasons for agglom-eration(it is not an exhaustive list):
1.b According to Marshall(1890,1920,Chap.X),externalities are crucial in the formation
of Q,(p.7).
2.b The hypothesis that production ts are convex implies that production exhibits no
increasing returns to scale—whatever If the distribution of natural resources is uniform,the
economy is such that each person produces for her own consumption, we therefore have backyard capitalism.Q(p.27).
he advantages that firms producing similar goods may exploit by collocating.
For example,different agents own different bits of information and their gathering yields higher level 2.A allows for a large of intermediate commodities can be ud as inputs to enhance the productivity of the 3.a large city allows for a better average match between heterogeneous workers and firms T job requirements.Q(p.98).
教师考勤制度4.b But why do houholds and firms ek proximity?Fundamentally,this occurs
becau economic agents need to interact and distance is an impediment can view the as an interplay between an interaction field among agents and competition on he need to interact acts as a centripetal force,whereas competition for land has the nature of centrifugal force.Q(p.169).
rnalities are at the root of economic agglomerations.Q(p.210).
孟姜女哭长城
6.b Such an externality effect may explain why economic agents are prepared to pay
high rent to live clo to the centers of large cities where this effect is more inten"
眼镜架品牌(p.211).
he prence of imperfectly competitive markets is another major reason for the
agglomerations.Q(p.258).
