NT Chapter 1 Industrial Location: The Location of the Firm in Theory EEE 1.1 Introduction to Classical and Neoclassical Models of Location ‘The level of output and activity of an area depends on the total quantities of factor inputs employed in the area, and the wealth of an area depends on the total payments received by those factors. Observation suggests that some regions exhibit dense concentrations of factors, with large numbers of people and investment located in the same area, whereas other regions exhibit sparse populations and low levels of investment. At the same time, observation also suggests that people are pald different wages in different areas, while land prices vary significantly between locations. Therefore, in order to understand the ‘economic performance of a region itis necessary to understand why particular quantities of factors are employed in that area, and why the factors there earn the particular rewards that they do, Production factor inputs are usually defined in terms of three broad types, namely capital, labour, and land, and the factor payments earned by these factors in the produc- tion process are profits, wages, and rents, respectively. In some analyses ofthe production process, additional factor inputs are also identified such as entrepreneurship and tech- nology. However, in our initial discussion of the causes and reasons for particular types of industrial location behaviour, we will not initially distinguish these additional factors from the broad factor groups. We include entrepreneurship in our description of labour, and technology in our description of capital. Later in our discussion of the causes and reasons for particular types of industrial location behaviour, we will also investigate the ‘additional issues associated with entrepreneurship and technology. In this chapter we will concentrate on the determinants of spatial variations in capital investment, and in later sections of the book we will focus on spatial variations in labour stocks, and variations in land prices ‘We start our analysis by asking the question—what determines the level and type ofINDUSTRIAL LOCATION 7 ‘capital invested in a particular region? When talking about capital, our most basic unit of microeconomic analysis is the capital embodied in the firm. In order to understand the level of capital investment in an area itis necessary to ask why particular firms are located there and why the particular levels and types of investment in the area are as they are, ‘These are the questions addressed by industrial location theory. We begin by discussing three classical and neoclassical models of industrial location behaviour, namely the ‘Weber model, the Moses model, and the Hotelling model, Each of these motels provides us with different insights into the fundamental reasons for, and the consequences of, {industrial location behaviour. After analysing each of these models in detail, we will Aiscuss two alternative approaches to analysing industeial location behaviour, namely the behavioural approach and the evolutionary approach. A broad understanding of these various approaches to industrial location behaviour will then allow us to discuss the concept of agglomeration economies. 1.2 The Weber Location-Production Model Our starting point is to adopt the approach to industria locational analysis originally derived from the nineteenth-century German mathematician Laundhart (1883), but which was formalized and publicized beyond Germany by Alfred Weber (1909). For our analysis to proceed we assume that the firm is defined at a point in space; the firm is therefore viewed as a single establishment. We also adopt the standard microeconomic assumption that the firm alms to maximize its profits. Assuming the profit-maximlzing rationale for the firm, the question of where a firm will locate therefore becomes the question of at which location a firm will maximize its profits. In order to answer this question we will begin by using the simplest two-dimensional spatial figure, namely a triangle. This very simple type of two-dimensional approach will subsequently be extended to more general spatial forms, ‘The model described by Figure 1.1 is often described as a Weber location-production ‘triangle, in which case the firm consumes two inputs In ordet to produce single output. Notation for use with Figures 1.1 to 1.12. ‘my, m, weight (tonnes) of material of input goods 1 and 2 consumed by the firm 'm, weight of output good 3 produced by the firm Pu Bs prices per tonne of the input goods 1 and 2 at thelr points of production P price per tonne of the output good 3 at the market location. ‘My Mz_ production locations of input goods 1 and 2 ‘M; market location for the output good 3 tf tyansport rates per tonne-mile (or per ton-kilometre) for hauling input ‘goods I and 2 ” transport rates per tonne-mile (or per tonne-kilometre) for hauling output goods 3 K the location of the firm. )URBAN AND REGIONAL ECONOMICS FIG. 1.1 Weber location-production triangle |We assume that the firm consumes material inputs 1 and 2, which are then combined by the firm in order to produce an output commodity 3, In the Weber location: production model, we assume that the coefficients of production are fixed, in that theres i ixed relationship between the quantities of each input required in order to produce a Single unt ofthe output. Our production function therefore takes the general form: Ky my ks. ay In the very simplest case k; in which case our production function becomes mm). (1.2 “This representsa situation where the quantity ofthe output good 3 produced is equal fo the combined weight of the Inputs 1 and 2. In other words for the purposes of our analysis here, we can rewrite (1.2) as y= m+ as) “The production locations of the input sources of 1 and 2, defined as M, and My are ‘given, as isthe location of the output market M,, at which output good 3 is sold. The prices per ton of the inputs 1 and 2 are given as p, and fat the points of production My tind M,, respectively. The price per tonne of the output good 3 atthe market location Ms, [s givenas py As such, the firm isa price taker, Moreover, we assume thatthe firm isable to sell untimited quantities of output 3 at the given price py as in perfect competition. The transport rates ae given as ff, and f,, and these transport rates represent the costs of transporting 1 tonne of each commodity 1, 2, and 3, respectively, over 1 mile or 1 Kilo- metre. Finally, the distances d,, d,, and d, represent the distances over which each of the goods 1, 2, and 3 are shipped. “We also assume that the input production factors of labour and capital are freely avail- able everywhere at factor prices and qualities that do not change with location, and that land ts homogeneous. In other words, the price and quality of labour is assumed to be cequal everywhere as isthe cost and quality of capital, and the quality and rental pice ofINDUSTRIAL LOCATION 9 land, However, there is no reason to suppose that the prices of labour, capital, and land are equal to each other. We simply assume that all locations exhibit the same attributes in terms of their production factor availability. Space is therefore assumed to be homogeneous. Ifthe firm is able to locate anywhere, then assuming the firm is rational, the firm will locate at whichever location it can earn maximum profits. Given that the prices of all the input and output goods are exogenously set, and the prices of production factors are invariant with respect to space, the only issue which will alter the relative profitability of different locations is the distance of any particular location from the input source and ‘output market points. The reason for this is that different locations will incur different ‘costs of transporting inputs from their production points to the location of the firm, and ‘outputs from the location of the firm to the market point. If the price per unit of output pis fixed, the location that ensures maximum profits are ‘eamed by the firm is the location at which the total input plus output transport costs are minimized, ceteris paribus. This is known as the Weber optimum location. Finding the Weber optimum location involves comparing the relative total input plus output trans- Port costs at each location. The Weber optimum location will be the particular location at ‘which the sum (TC) of these costs is minimized. The cost condition that determines the ‘Weber optimum location can be described as, To=Min mtd, as where the subscript i refers to the particular weights, transport rates, and distances over Which goods are shipped to and from each location point K, With actual values corres- ‘Ponding to each of the spatial and non-spatial parameters, it Is possible to calculate the total production plus transportation costs incurred by the firm associated with being at any arbitrary location K. Given our assumptions that the firm will behave so as to maxi ‘ize its profits, the minimum cost location will be the actual chosen location of the firm. In his original analysis Weber characterized the problem of the optimum location in ‘terms of a mechanical analogy. He described a two-dimensional triangular system of pulleys with weights called a Varignon Frame. In this system, the locations of the pulleys reflect the locations of input source and output market points, and the weights attached ‘to each string passing over each of the pulleys corresponds to the transport costs associ ated with each shipment. The point at which the strings are all knotted together repre- sents the location of the firm. In some cases, the knot will settle at a location inside the triangle, whereas in other cases the knot will settle at one of the corners. This suggests ‘hat the optimum location will sometimes be inside the Weber triangle, whereas in other ‘cases the optimum location will be at one of the comers. Nowadays, rather than using such mechanical devices, the optimum location can be calculated using computers. However, although itis always possible to calculate the optimum location of the firm in ‘each particular case, of interest to us here is to understand how the location of the Weber ‘optimum will itself be affected by the levels of, and changes in, any of the parameters deseribed above. In order to explain this, we adopt a hypothetical example.10 URBAN AND REGIONAL ECONOMICS 1.2.1 The location effect of input transport costs Letusimagine that Fgure 1.1 represents firm that produces automobiles from inputs of eel and plastic. The output good 3 is defined as automobiles and these are sold atthe narket point M,. We can assume that input 1 stee and input 2 is plastic, and these are produced at locatons M, and M, respectively. I the firm produces a car weighing > tonnes from 1 tonne of steel and 1 tonne of plastic, and the fixed transport rate for steel {s haf that for paste , (given that plastic is much Tess dense than steel, and transport rates are notmmally charged with respect to product bulk), the firm will locate relatively ‘lose to the source of the plastic production. In other words, the firm will ocate close 10 ‘My, The eason is that the firm will wish to reduce the higher total transport costs associ ated with shipping plastic Inputs relative to steel Inputs, ceteris paribus. The frm can do this by reducing the value of dy relative to a. On the other hand, ifthe firm had 9 different production function, such that it produces a car weighing 2 tons from 1.5 tonnes of steel and O.S tonnes of plastic, then even with the same values for the fixed transport rate f, and tain the previous case the fm willnow be incurring higher ta) transport costs associated with stee! shipments, ters paribus. The reason fortis Is that although plastic i twice as expensive to ship per Kilometre as tee, the total quantity of Steel being shipped is three times that of plastic, The result is thatthe firm can reduce it total input transport costs by reducing the value of dy relative tod, The optimum location ff the firm will now tend towards the location of production for the steel Input M, ‘Within this Weber framework, we can compare the effects of different production function relationships on the location behaviour of the firm, For example, we can Imagine thatthe two types of production function relationships described above—one ‘which is relatively plastic intensive, and one which is relatively steel intensive actually efer to the different production functions exhibited by two different competing auto- tmobile producers. Firm A exhibits the plasticintensive production function, and frm 5 ‘exhibits the steel-intensive production function, As we see In Figure 1.2, from the argu tment above we know that firm A will locate relatively close to M,, the source of plastic, IG. 1.2 Relative input transport costs and locationINDUSTRIAL LOCATION 14 while firm B will locate relatively close to M,, the source of steel. This is because, if we were to consider the case where steel and plastic inputs were shipped over identical distances, ie. d,,=dz. for firm A the total transport costs associated with plastic trans- portation would be greater than those associated with steel transportation. It therefore has an incentive to reduce the higher costs associated with plastic shipments by reducing a, and increasing dj. Alternatively, for firm B, for identical input shipment distances, Le. y= dz, the total transpoct costs associated with steel transportation would be greater than those associated with plastic transportation. It therefore has an incentive to reduce the higher transport costs associated with steel by reducing dyy and increasing day 1.22 The location effect of output transport costs Until now we have only considered the transport cost pull of the input sources on the location decision of the firm. However, the market itself will display a pull effect on the location behaviour of the firm. We can imagine the case of a power-generating plant ‘which bums coal and coke, produced at M; and M,, respectively, in order to produce electricity. We can regard the output of the plant as having zero weight or bulk. The ‘output transportation costs of shipping electricity can be regarded as effectively zero, given that the only costs associated with distance will be the negligible costs of booster stations, In this case, the market point of the plant, whether it isa city or a region, will play no role in the decision of where to locate the plant. As such, the optimal location of the plant will be somewhere along the line joining M, and M,, The optimal location problem therefore becomes a one-dimensional location problem. A discussion of this type of problem is given in Appendix 1.1. ‘In most situations, however, the output of the firm is costly to transport due to the weight and bulk of the output product. Different output weight and bulk will affect the ‘optimum location of the firm relative to the location of the market and the inputs. Once again, we can illustrate this point by using our hypothetical example above of two auto- ‘mobile firms, A and B, each consuming inputs of steel and plastic. However, in this case ‘we can imagine a situation where the input production functions of both firms were the same. In other words the relative input combinations for each firm, given as m/ are the same. If both firms pay the same respective transport rates t, and t, for each input shipped, the relative locational pull of each input will be identical for each firm. However, In this situation we also assume that the firms differ in terms of their technical efficiency, in that firm A discards 70 per cent of the inputs during the production process, whereas firm B discards only 40 per cent of the inputs during the production process. Con- sequently, the total output weight m, of firm B is twice as great as that of firm A, for any total weight of inputs consumed. This greater output welght will encourage firm B to ‘move closer towards the market point and further away from the input points than firm A. AS see in Figure 1.3, firm B will therefore be more market-oriented than firm A in its location behaviour. A more common situation in which similar firms exhibit different location behaviour with respect to the market is where the density of the product changes through the production process at different rates for each of the producers. For example, we can imagine our two automobile firms A and B, producing identical weights of output from12 URBAN AND REGIONAL ECONOMICS FIG, 1.3 Relative output transport costs and location identical total weights of inputs. Here, the production functions of both firms are there- fore the same. However, we can also assume that firm A specializes in the production of ‘small vehicles suited to urban traffic, while firm B produces large four-wheel-drive vehicles suitable for rough terrain, As we have already seen, transport rates also depend on the bulk of the product, and products which have a high density will exhibit lower unit transport costs than products with a low density. In this situation firm B produces goods which are very bulky, whereas firm A produces goods which are relatively dense. ‘Therefore, the output of firm B will be more expensive to transport than that of firm A, and this will encourage firm B to move closer to the market than firm A. Once again, as seen in Figure 1.3, fim B will be more market-oriented than firm A. 1.23 The location effect of varying factor prices (Our analysis so far has proceeded on the assumption that labour and land prices are {identical across all locations, although in reality we know that factor prices vary signifi- cantly over space. The Weber approach also allows us to consider how factor price varl- ations across space will afect the location behaviour of the firm. In order to understand this, itis necessary for us to identify the factor price conditions under which a fim will look for alternative locations. ‘We assume that the firm Is still consuming inputs from M, and M, and producing an ‘output for the market at M,, Under these conditions, we know that the Weber optimum K* Is the minimum transport cost location of the firm, and that if all factor prices are ‘equal across space this will be the location of the firm. Our starting point is therefore to ‘consider the factor price variations relative to the Weber optimum K* which will encour- age a firm to move elsewhere, In order to do this, it is first necessary for us to construct a ‘contour map on our Weber triangle, as described by Figure 1.3. These contours are known sdapanes. (Ona standard geographical map each contour links all ofthe locations with the sameINDUSTRIAL LOCATION 13, altitude. On the other hand, each isodapane contour here in a Weber map links all the locations which exhibit the same increase in total input plus output transport costs, pet unit of output m, produced, relative to the Weber optimum location K*. Increasing Is0- dapanes therefore reflect increased total input plus output transport costs per unit of ‘output m, produced, relative to the Weber optimum X*, As the location of the firm moves away from the Weber optimum in any direction, the firm incurs increasing transport costs relative to the Weber optimum. In other words, the locations become less and less efficient, and the firm exhibits successively lower profits, ceteris paribus, We can also say that the firm incurs successively greater opportunity costs as it moves further away from the Weber optimum. If factor prices are equal across space, locations further away from the Weber optimum will become successively less desirable locations for invest- ‘ment. Therefore, we need to ask by how much do local factor prices need to fall relative to the Weber optimum location K* in order for the firm to move there? If we take the case of location R, we can ask by how much do factor prices at R need to, {all relative to the Weber optimum K*, in order for the firm to move from K* to R? As we FIG. 1.4 Isodapane analysis14 URBAN AND REGIONAL ECONOMICS see from Figure 1.4, Ris on the $25 isodapanes. If the costs of the labour and land factor Inputs required to produce one unit of output mat Rare $20 less than at K* it will not be in the interests ofthe firm to move from K* to R. The reason is that the fall in local factor input prices associated with a move from K*to R will not be sufficient to compensate for ‘the increased total transport costs as we move away from the Weber optimum. Ifthe firm ete to move from K* to R in these circumstances, it would experience profits which were $5 unit of output m, les than at K*, On the other hand ifthe local labour and land prices pe unit of output as R were $30 les than at K*, if would be inthe interest ofthe firm to prove. This Is because the reduction in the local input factor costs associated with a move from K*to Rwill now more than compensate for the increase in total transportation costs incurred by the move. Ifthe firm were to move from K* to R in these circumstances, it ‘would experience profits which were $5 per unit of output m, greater than at "This type df analysis can be applied to any alternative locations, such as Q,R, S, and T, in order to determine whether a firm should move and to which location, For example, location Qis on the $10 isodapane, is on the $25 isodapane, $is on the $40 isodapane, and Tis on the $50 isodapane. Let us assume that the costs of the labour and land factor inputs required to produce one unit of output mat Q R, 5, and Tare less than the factor costs at K* by amounts of $12, $20, $35, and $55, respectively. We can determine that the alternative locations Q and Tare superior locations to K*, in that both twill provide greater profits than K*, whereas R and S are inferior locations in that they exhibit reduced profits relative to K*. However, of these superior alternatives, T is the better location because profits here are $5 per unit of output greater than at K* whereas those at Qare only $2 greater. With this particular spatial distribution of local labour and land prices, location Tis the optimum location of the firm. T sa superior location to the Weber optimum location at which total transport costs were minimized, because the lower local factor input prices more than compensate for the increased total transport costs associated with the location of T. ‘This type of approach also allows us to ask and answer a very important question: how will Iocal wages and land prices have to vary over space in order for the fitm’s profits to be the same for all locations? This can be analysed by modifying Figure 1.4, We can con- struct Figure 1.5, by employing Figure 1.4, but then altering it by drawing a line from K* ‘eastwards which passes through Q R, 5, T, a in Figure 1.5. This ine is defined in terms of geographical distance, We can then observe how the isodapanes intercept this line. Prom the above example, we know that location Qis on the $10 isodapane, Ris on the $25 isodapane, Sis on the $40 isodapane, and Tis on the $50 isodapane. The firm's profits ‘will be the same In all locations if the local labour and land factor input prices at each Tocation exactly compensate for the increased total transport costs associated with each ocation, Therefore in Figure 1.5 this allows us to plot the labour and land price gradient ‘with respect to distance which ensures equal profits are made at all locations east of K*, ‘assuming the wage at K* is w*, We can repeat the exercise by drawing a line from K* which passes west through U, V, W, and X, and plotting the local factor prices which will censure the firm makes profits equal to those at K* at all locations west of K*. Combining this information allows us to construct the interregional factor price curve for ou pat- ticular firm which ensures that it makes equal profits at all locations in the east-west direction. This is shown in Figure 1.6INDUSTRIAL LOCATION 415 eo 8 3s fF Di FIG. 1.5 Distance isodapane equilibrium labour prices ‘This slope of the line is the interregional equilibrium factor price gradient for this par- {cular firm along this particular axis. This equilibrium factor price gradient describes the variation in local factor prices, which ensures that the firm will be indifferent between locations. The firm is indifferent between locations along the east-west line, because the profits it can earn are the same everywhere along this line. As such, from the point of view (of this firm, all locations along the east-west line are perfect substitutes for each other. In principle, we can also construct similar factor price gradients for movements in any. other direction away from K’, such as movements passing through locations C, D, or F, in order to generate a two-dimensional equilibrium factor price map of the whole spatial economy. ‘The idea that locations can be perfect substitutes for each other, from the point of view of a firm’s profitability, is important in terms of understanding the spatial patterns of industrial investment. For example, ifa multinational manufacturing firm is looking for a new production site in order to develop its business In a new area, the likelihood of it going to any particular location will depend on the firm's estimate of the profits it can eam at that location. From the isodapane analysis of our Weber location-production ‘model here, we know that the locations of key input sources such as M, and M, and ‘market points such as M,, will automatically mean that some locations are more profit: able than others, with the Weber optimum being the most profitable location, ceteris paribus. Therefore, in order to make other locations attractive for investment, local factor prices have to fall relative to the Weber optimum. The attractiveness of any particular location as.a new investment location for the firm will depend on the extent to which the local factor price falls can compensate for the increased transport (opportunity) costs associated with any suboptimal geographical location. If all local factor prices are inter- regional equilibrium prices, as described by Figure 1.6, the firm will be indifferent16 URBAN AND REGIONAL ECONOMICS Equilibrium interregional wage gradient as x Ww FIG. 1.6 Inter-tegional equilibrium wage gradient between locations, Under these ciscumstances, the firm will be equally ikely to build its new production faclity at any location. In other words, the probability of investment jail be equal for allocations, Over large numbers of firms with similar input requlce- nents and similar output markets to this particular firm, the level of investment in any Tocation should be the same as in all locations. On the other hand, if wages are not in equilibrium over space, certaln areas will automatically appear more attractive as loca- tions for investment, thereby increasing the probability of investment there. ‘Geography confers diferent competitive advantages on different locations, which can only be compensated for by variations in local factor prices. However, in the above ‘example, the equilibrium relationship between local factor prices and distance was only applicable to the particular frm in question here. Tis i because the interregional factor price gradient was calculated with respect to the Weber optimum of ths particular Erm. ‘As we have seen, different firms will exhibit different Weber optimum locations, and this implies that different equilibrium interregional factor price gradients will exist for differ- cent types of firms exhibiting different transport costs, different production functions, ‘and finally different input and output locations. 1.2.4 The locational effect of new input sources and new markets uranalyss hao fardiscused the locational efecto itferent transport cots diferent crrition funtions andthe resting conditions under which a firm wil be wiling to errata atemative Locations. We will now discuss the question of different input and sre yat ceationsand the condition under which aim wil earch fraltematvs. Inthe cranes above, it was possible to use sodapane analysis to Identify the facto price (Pit). Therefore, he hers makes a greater profit from selling automobiles to market M, than to market M,- ssh tum could switch markets completly from M, toM,, Alternatively, it could decide to Supply both markets M, to.M,. Under these conditions, it may be that anew optimum raya of H arses, in which the fm at H buys from two supplies locations M, and Mo, woepealls at two market locations, My and Ms, More complex arrangements are possible Tor example, In order to guarantee sufficient supplies of steel Inputs for the newly txpanied sutomotile market of (y+) the frm may decide to continue to purchase sats fgom both M; and M,, as well as purchasing plastic from M,, Now we have a Weber incation production problem with M;, M, Mi, My and M, as spatial reference points. ‘nue again this will move the Weber optimum away from point H, and wil also alter the inter-regional equilibrium wage gradient “Tis type of geometrical arrangement, In which a firm has multiple put sources and ‘multiple output market location, Is the norm for firms n reality. Although our analysis here has been developed primarily with only two input source locations and one output market location, the Weber location-production arguments and the associated isodapane Snalysis are perfectly applicable to the case of firms with multiple Input and output Tocations. The reason for employing the triangular case of the two input locations and ‘one output market location is that this particular spatial structure is simply the easiest two-dimensional model to explain. The model is designed to help us understand the idvantages which geography confers on particular locations as sites for investment. A first key feature of the Weber model is therefore that it allows us to understand the factor price conditions under which other areas will become more attractive as locations for Tavestment, Secondly, the model allows us to see location as an evolutionary process In which changes n factor prices can engender changes in location behaviour, which them- Selves can change the supply linkages between suppliers, firms, and markets, Industrial location problems are inherently evolutionary in thelr nature as firms respond to new ‘markets and products by changing their locations, and by changing the people they buy from and the people they sell to. All of these are spatial issues. There is one final issue relating to the Weber model which needs to be addressed. In reality, firms are constantly changing their Input suppliers and output markets in response to changes in input and output market prices. From our Weber analysis, these ‘changes will also imply that the optimum location of the firm is continuously changing, ‘and that in order to ensure the profitability of any particular location the equilibrium Inter-teglonal factor price gradient must also be continuously changing. However, obser- vation tells us that fiems in reality do not move very frequently, and this raises the {question of the extent to which the Weber model is @ useful analytical tool to describe industulal location behaviour. ‘The reason why firms are not continuously moving i that the relocation process itself usually incurs very significant costs, such as the dismantling of equipment, the moving of people, and the hiring of new staff. Part of the transactions costs associated with reloca- tion ate also related to information and uncertainty, which are topics we will deal with, later in the chapter. However, within the above framework we can easily IncorporateINDUSTRIAL LOCATION. 19 ‘hese relocation costs, by including the annualized cost of these one-off relocation costs into our isodapane model. The existence of these additional costs simply implies that firms will move only when the factor cost advantages of alternative locations also com- pensate for these additional relocation costs as wel asthe increased transport costs. In other words, the equilibrium inter-regional wage gradient will be even steeper than under the situation where such costs are negligible. The Weber model therefore stil allows us to Identify the optimum location, and consequently the profit-maximizing behaviour of the firm in space, even in situations where relocation costs are significant. The observa- tion that firms do not move frequently does not limit the applicability of the Weber ‘model to real-world phenomena, ‘The one major location issue which the Weber model does not address, Is that of the relationship between input substitution and location behaviour. In order to understand this relationship, we now tum to a discussion of the Moses location-production model —— 1.3 The Moses Location-Production Model —. |! ‘The Weber model assumes that the quantities of each input consumed, m, and m,, are fixed per unit of output m, produced. However, we know from standard microeconomic analysis that substitution is a characteristic feature of firm behaviour, and that efficiency conditions mean that firms will substitute in favour of relatively cheaper inputs, ceteris paribus. Substitution behaviour was first incorporated coherently into the Weber analysis, bby Moses (1958), and in order to see how substitution behaviour affects the location behaviour of the firm, we discuss here the main features and conclusions of the Moses approach, In Figure 1.8, we construct an arc JJ in our triangle M;, M,, My, which is at a constant distance d, from the market point M,. If we constrain our firm to locate along this ar, the distance from the location of the firm K to the market M, will no longer be a variable. ‘Therefore, we can analyse the locational pull on the firm of changes only in the delivered. prices of the inputs produced at M, and M;. For example, if the firm was located at J, the delivered price of input 1, given as (;+td,), will be a minimum, because the distance dj from M, to J will be a minimum, Similarly, the delivered price of input 2, given as (p,+t,4,), will be a maximum, because the distance d; from M; to I will be a maximum. The delivered price ratio, given as (s+ tad.) (P+ td), will therefore be a minimum at location J, On the other hand, ifthe firm now moves to J, the delivered price of input 1 will be a maximum, because the distance d, from M, to Jwill be a maximum. At the same time, the delivered price of input 2will be a minimum, because the distance d, from M, to Iwill be a minimum. Therefore, the delivered price ratio, (p, + td, (p+ ta), will be a maximum at location J In standard microeconomic approaches to firm efficiency, the optimal input combi ation is determined by finding the point at which the highest isoquant attainable is tangent to the budget constraint. In this standard approach, the slope of the budget Constraint is determined by the relative prices of the goods. From the above argument, we20. URBAN AND REGIONAL ECONOMICS My FIG, 1.8 Weber-Moses triangle fudge constraint at Budget constront ot} FIG. 1.9 Budget constraints at the end points and J ‘can draw the budget constraints at locations I and J as shown in Figure 1.9, which repre- sent equal total expenditure on inputs at each Location. The delivered price ratios at locations 1 and / ate given by the ratio of the tangents of the angles a/f; and a/B, respectively. Yet, this arguments also applicable to all locations along the arc. Ifthere are different delivered price ratios for different locations, this implies that for given source prices of the inputs p, and p,, the slope of the budget constraints at each location along J must be different. As we move along the arc IJ from Ito J, the delivered price ratio increases, andINDUSTRIAL LOCATION 24 for every location along the arc I there is a unique delivered price ratio. This means that ‘the usual approach to analysing microeconomic efficiency is not applicable to the firm in space, and must be adapted to incorporate the effects of location on the slope of the ‘budget constraint. In order to do this we must construct the envelope budget constraint, Which just contains all of the budget constraints associated with each of the locations along the arc J. This is done by drawing each of the budget constraints for each of the location points on the arc JJ, as in Figure 1.10, and the outer limits of this set of indi vidual budget constraints will define the envelope budget constraint. ‘The Moses argument Is that we can now apply standard efficiency conditions to this ‘model, by finding the point at which the envelope budget constraint is tangent to the highest isoquant attainable. This is shown in Figure 1.11, where the point of maximum efficiency is at E*. ‘At B+, the optimum input combinations are given as m,* and m,*, However, E* also represents an optimum location K*. The reason is that the optimum input combination is found at a particular point on the envelope budget constraint. Yet, every point on the ‘budget constraint also represents a unique location. Therefore, the optimum input mix and the optimum location of the firm are always jointly determined. One is neve ‘without the other, This is a profound insight. Where input substitution is possible, al location problems become production problems and all production problems become location problems. We can illustrate the argument with an example. In our Weber-Moses triangle, we can imagine that a road-building programme takes place in the area around location M,, the effect of which is generally to reduce the value of f, forall shipments of goods from this location, relative to all other locations. Ifall the other parameters remain constant, this will imply that the delivered price ratio (P, +f) (s+ fxd), at all locations along If will Envelope budget constraint f FIG. 1.10 The envelope budget constraint22 URBAN AND REGIONAL ECONOMICS Envelope budget constraint ‘soquonts FIG. 1.11 Location-production optimum fall. In other words, the slope of each budget constraint becomes steeper, ceteris paribus, and the envelope budget constraint also becomes steeper and shifts upwards to the left, Strictly Speaking, in accordance with the income effect, the envelope will also shift out- wards to the right, because the price of the input has fallen. However, in this discussion ‘we focus only on the substitution effect of the change in slope of the envelope. For a given set of production isoquants, the optimum production combination will change from that represented by E*. "As we see in Figure 1.12, at the new optimum E’, the optimum input mix is now m/ and my. The reason is that the firm substitutes in favour of input 1, which is now rela- tively cheaper than before, and away from input 2, which {s now relatively more expen sive than before. In doing so, the firm increases the relative quantities of input 1 it consumes and reduces the relative quantities of input 2 it consumes. However, this also implies that at the original location K*, the firm now incurs increasing total transport ‘costs (ty, for input I relative to the total transport costs (mfa,) for input 2. Therefore, the firm will move towards M,, the source of input 1, in order to reduce these costs. The new optimum location of the firm K's closer to M, than E*, and so the firm moves towards M;. Te area around M, benefits in two different ways. Firs, the relative quantity of goods produced by the area around M, which are bought by the firm increases. This increases regional output for the area, Secondly, the firm itself locates in the vicinity of M,, thereby increasing the levels of industrial investment in the area. Exactly the same result would have arisen in the case where, instead of a road-building programme, there was a fall in the local wages at M;, which reduced the source price p;, relative to all other locations. Once again, the fallin the delivered price ratio at all loca- tions leads to substitution in favour of the cheaper good and also relocation towards M,. ‘We can contrast this Moses result with that of the Weber model. In the simple WeberINDUSTRIAL LOCATION 23, Envelope budget constraint efter the price change Envelope budget constraint before the price change FIG. 1.12 A change in the location-production optimum ‘model, if the transport rate f, falls, ceteris paribus, the effect on the location of the firm is to move the locational optimum away from M,. The reason is that input 2 now becomes relatively more expensive to transport, and because the coefficients of production are fixed, such that the relative quantities of my, and m, consumed remain the same, the firm will move towards the source of input 2 in order to reduce the total transport costs. The difference between the locatlon-production results ofthe two models is that in the Weber ‘model the Gxed coefficients mean that no input substitution is possible, whereas in the Moses model of variable coefficients, input substitution Is possible. In the latter case, the input substitution behaviour alters the relative total transport costs and consequently the optimum location behaviour of the firm. In reality, there is a continuum of possible location effects, dependent on the technical substitution possibilities. In situations where the elasticities of substitution are zero or very low, the results will tend to mimic those of the Weber model, whereas in situations where the elasticities of substitution are high, the results will tend towards those of the conclusions of the Moses model. A second feature of the Moses model is that it allows us to examine the effect of returns to scale on the location-production behaviour of the firm. In particular, we can ask the question, how will the optimum location of the firm be affected by changes in the level of ‘output of the firm? In order to answer this in Figure 1.13 we construct a series of envelope budget constraints, represented by the dotted lines, which correspond to different levels of total expenditure on inputs. Envelope budget constraints further to the right imply24 URBAN AND REGIONAL ECONOMICS m, FIG. 1.13 Output changes and location production behaviour ‘greater total expenditure levels on inputs. An isoquant map, represented by the solid ‘curves, can now be combined with the envelope map. We can also apply the Moses argument, which states that the optimum point for each level of output and input ‘expenditure is where each particular envelope is tangent to the highest isoquant, to the case of different output levels, By joining all the points of tangency we construct a line ‘ABC, which is an output expansion path. Yet, this output expansion path Is different from the usual form of an expansion path. Each point on the expansion path defines a particular optimum input combination. However, each point on the expansion path also defines an optimum location. If the expansion path is curved downward, such as in the case of ABC in Figure 1.13, it implies that as the output of the firm increases, and the total quantity of inputs con- sumed increases, the optimum input mix changes relatively in favour of input 2. The ‘optimum ratio of m,/m falls and the optimum location of the firm moves towards M,- Alternatively, ifthe expansion path were to curve upwards, this would mean that as the ‘output of the firm increases, the optimum input combination would change in favour of ‘input 1, As the optimum ratio of m,/m, increases, the optimum location of the firm ‘would move towards the market. ‘This argument immediately leads to the conclusion that if the expansion path Is a straight line from the origin, such as FGH in Figure 1.14, both the optimum input mix and also the optimum location of the firm will remain constant as output expands, ceteris ‘paribus. The actual slope of the expansion path is not important, other than it implies a different optimum location, All that is required to ensure that once the firm has found its ‘optimum location it will always remain at this optimum location as output changes, isINDUSTRIAL LOCATION 25 FIG. 1.14 The independent of output optimum location solution that the production function of the firm exhibits a straight line expansion path from the origin, This is the basic Moses result. ‘This basic Moses result holds in the case where the firm is constrained to locate on the arc If ata fixed distance from the market. However, in the more general case where the distance from the market is also part of the location problem, the optimum location of the firm will be Independent of the level of output, as long as both the production function of the firm and the transportation technology of the firm exhibit constant returns to scale, The Weber fixed-coefficients production function will satisfy the Moses requirement, However, there are other more general types of production functions allow- {ng for input substitution, which also satisfy this requirement. These results are detailed In Appendix 1.2. The Moses result can be viewed somewhat as the spatial equivalent of the firm in perfect competition. The firm is a price taker, and once it has determined its optimum production technique and optimum location, the firm will not change its behaviour, ceteris paribus. In other words, unless there are external changes in technology which alter the production function relationships, or changes in transportation technology which alter relative transport costs, or externally determined changes in the location of input {goods sources and output market points, the frm will always remain at the same location employing the same input-output production techniques. It would be wrong, however, to view these spatial results as implying that the spatial economy is essentially static. From ou discussion in section 1.2.4, we saw that the spatial economy exhibits evolution- ary characteristics, with firms searching for new optimum locations in response to factor price changes, and subsequently searching for new input supplier and market outputEE 26 URGAN AND REGIONAL ECONOMICS locations, in response to their relocation behaviour. The key insights, howeves, of the (Weber and Moses models are that production behaviour and location behaviour are completely intertwined issues. Often this point is overlooked in textbook discussions of . eduetrial economics and the theory ofthe frm. This is largely because location adds an fextra dimension to the optimization problems, making the analysis somewhat more complex. . 13.1 The logistics-costs model ‘There are a couple of possible limitations’ to the applicabllity of the Weber-Moses framework to real-world phenomena which need to be considered at this point. The first limitation is that the market price or revenue of the output good plays no role In the determination of the optimum location of the firm in either model. In the Weber model, the optimum location is determined solely by the transportation costs associated with the input and output goods, whereas in the Moses model, the input prices do play a role {in the optimum location. In nelther model does the market price have any effect on the determination of the optimum location. The second limitation of this framework is the emphasis on transport costs asa locational issue: In reality, transport costs tend to be only a very small percentage of total costs for most firms. However, both of these model reaknesses can be largely reconciled within a Weber-Moses framework by employing a broader description of distancesransport costs defined as ‘total logistics costs’, which includes all of the inventory purchasing and carrying costs associated with transporta- tion (McCann 1993, 1997, 1998). Employing this logistics costs approach, it can be dem- ‘onstrated both that the market price and market sales revenue do play a crucial role in determining the optimum location, and also that distance costs are very significant: In particular, as we see in Appendix 1.3, the higher value-adding activities will tend to De nore market-oriented than lower value-adding activities, and will also tend to be less Sensitive to inter-regional labour price changes. As such, market areas will tend to be surrounded by higher-value activities or activities further up the value-chain, whereas Supply sources will tend to be surrounded more by lower value-adding activities ox firms ower down the supply chain. At the same time, total logistics costs can also be shown to be very much more significant than transport costs alone, because each of the inventory ‘purchasing and carrying cost components can be shown to be functions of distance. A final point here is that the total logistics costs approach can also be employed to account for the economies of distance and scale generally observed in transport pricing (McCann 2001) and discussed in Appendix 1.1.INDUSTRIAL LOCATION 27 1.4 Market Area Analysis: Spatial Monopoly Power In our analysis so far we have assumed that the market location is simply a point in space, However, taking geography and space seriously in our models of firm behaviour also requires that we investigate the explicitly spatial nature of market areas. Market areas frequently differ over space, due to differences in spatial population densities, differences in income distributions across space, and differences in consumer demand across space according to regional variations in consumer tastes. However, even if there were no spa- ~ tial variations in population densities, income distributions, and consumer demand pat- tems, space would stl be an important competitive issue. The reason is that geography and space can confer monopoly power on firms, which encourages firms to engage in spatial competition in order to try to acquire monopoly power through location ‘behaviour. In order to sce this we can adopt the approach frst used by Palander (1938). In Figure 1.15, we have two firms A and B located at points A and B along a one- dimensional market area defined by OL. We assume that both firms are producing an identical product. The production cost p,of firm A at location A can be represented by the vertical distance a, and the production costs ps of firm B at location B can be repre- sented by the vertical distance b. As we see, firm A is more efficient than firm B. The transport costs faced by each firm as we move away in any direction from the location of the firm are represented by the slopes of the transport rate functions. As we see here the transport rates forthe two firms in this case are identical, ic. t,= For any location at a distance d, away from firm A the delivered price of the good is given as (p,+ 4), and for any location ata distance dy away from firm B, the delivered price ofthe good is given as (4+ ta Tf we assume that consumers are evenly distributed along the line OL, and vre also assume that consumers, being rational, will buy from the firm which i able to supply at ‘that particular location a the lowest delivered price, the total market area will be divided into two sectors OX and XL. The reason for this is that between O and X, the delivered price of firm A, given as (7, + t,), is always lower than that of firm B, On the other hand, tall locations between X and Lthe delivered price of firm B, given as (p,+ fy), is always lower than that of firm A, Although firm A is more efficent than firm B, and although both firms produce an identical product, firm A does not gain all of the market. The reason Is that location gives each frm some monopoly power over the area around itselt. Firm A cannot capture all of frm B's market, even though it is more efficient than firm B, ‘because the transport costs associated with shipping goods to market locations close to firm B increase the delivered price (p+ td) to an uncompetitive level in market locations close to firm B, In terms of selling to consumers in the vicinity of firm B, firm A is unsuccessful simply because i is too faraway. On the other hand, fr sales in this aea, firm Bis successful simply because itis in the right location, even though tis less efficient in production. This type of analysis can be extended to allow for differences in transport rates between28 URBAN AND RECIONAL ECONOMICS Price/Cost Market offemA Market of frm 8 FIG, 1.15 Spatial market areas: a one-dimensional model with equal transport rates firms as well as differences in production costs. In Figure 1.16a, b, we see that market areas ‘can be divided up in a varlety of ways in situations where the production costs and ‘transport rates vary between the firm, Generally, the size of a firm's market area will be larger the lower are the production costs of the firm and the lower ate the transport rates faced by the firm. However, only in the case where transport rates are zero Is a lower ‘production price sufficient to ensure a firm captures all of the market. The reason is that the existence of transport costs allows less efficient firms such as firm B to survive by ‘providing each firm with some monopoly power over particular market areas. In general, ‘the areas over which firms have some monopoly power are the areas in which the firms are located. For example, Figure 1.16b can be regarded as representing a case such as a local bakery, where firm B maintains a very small local market area in the face of competi- tion from a national bakery, firm A, which produces at much lower unit production costs, and transports in large low-cost shipments. Monopoly power refers to the ability of the firm to increase the production price of the good p, or ps and yet maintain some market share. In general, the greater Is the monop- ‘oly power of the firm, the steeper Is the firm's downward-sloping demand curve. In many textbook descriptions of monopoly or monopolistic power, the slope of the firm's downward-sloping demand curve is viewed as being dependent on brand loyalty, associ- ated with advertising and marketing. However, location is also an important way in ‘which many firms acquire monopoly power. The reason Is that transport costs are a form of transactions costs, and from the theory of the firm, we know that the existence of transactions costs such as tariffs and taxes can provide protection for some inefficientINDUSTRIAL LOCATION 29 PrcefCost Market of ‘) PrcelCost ! MorketofA P'Morket 7 Morket o of8 | ofA FIG. 1.16a, b Spatial market areas: one-dimensional models with varying transport rates and production costs firms, Geography acts in similar manner, because the costs of overcoming space in order ‘to carry out market exchanges incur transport-transactions costs. In the context of Fig ‘ures 1.15 and 1.16, there are two general rules governing the extent to which distance costs provide a firm with spatial monopoly power: i First, the greater are the values of the transport rates t, and fy the lower will be the fall in the market area of the firm, and the greater will be the monopoly power ofthe frm, for any marginal increase in the price of either p, or py, ceteris paribus.30 URBAN AND REGIONAL ECONOMICS Gi Second, the further apart are the firms, the lower will be the fallin the market area of the firm, and the greater will be the monopoly power of the firm, for any marginal increase in the price of elther por Pp, ceteris paribus. ‘Therefore, firms which are located ata great distance from each other, and which face significant transport costs, will consequently exhibit significant local spatial monopoly power. ' 1.41 The Hotelling model of spatial competition ‘The existence of spatial monopoly power provides an incentive for firms to use location asa competitive weapon in order to acquire greater monopoly power. This ts particularly important in industries where firms do not compete primarily in terms of price, but instead engage in non-price competition, such a product quality competition. In com- petitive environments characterized by oligopoly, the interdependence between firms in the determination of output quantities and market share is also a result of locational considerations, as well as interdependence in terms of pricing decisions. The simplest demonstration of this is the Hotelling (1929) model, which describes firms’ spatial interdependence within the context of a locational game. In Figure 1.17 we adapt Figure 1.15 to the case where both the production costs and transport cates of firm A and firm B are identical. In other words, p,= ppand t= 6, and we assume that these prices do not change. As before, we assume that consumers are evenly distributed along OL and we also introduce the assumption that the demand of con- ‘sumers is perfectly inelastic, such that all consumers consume a fixed quantity per time period respective ofthe price. In terms of frm strategy we assume that each firm makes a competitive decision on the basis of the assumption that Its competitor firm will not change its behaviour. In the game theory literature this particular set of rules describing the nature of the competitive environment is known as ‘Cournot conjectures’. Given that the firms are not competing in terms of their production prices, which are assumed to be fixed, each firm can only adjust its location in order to acquire greater market share. Ifthe firms react to each other in sequential time periods, the location result can be predicted easily If we assume that the firms A and B are initially located at one-quarter and the quartets of the way along the market, respectively, firm A will have monopoly power over (OX and firm B will have monopoly power over XL. In this case, both firms will have identical market shares, In time period 1 firm A will therefore move from its original location to a location at C, just to the left of B. In this way firm A will increase its market share from OX to a new maximum value of OC, Firm 8 will still retain market share over BL, although its market share is now at a minimum, Firm B will now assume that firm A will maintain its location at C, and so in time perlod 2, firm B will move just to the left of C In time period 3, firm A will respond by moving to the left of firm B, and this process will continue until both firms are located at X, in the middle of the market, Once both firms are located at X, neither firm has any Incentive to change its location behavious, because any location change will involve a reduction in ‘market shate telative to their location at X. In game theory, any situation in which neither firm has any incentive to change its behaviour is known as a ‘Nash equilibrium’INDUSTRIAL LOCATION 31 PrceCost ran T [> MarketofAintimeperod? = | Morket of jn time period 1 Market of in time period 2 | Market ofintime | period? | FIG. 1.17 The Hotaling location gare ‘The locational result in which both firms are located at the centre of the market is the [Nash equilibrium for this particular locational game. Consequently, once the firms reach this point they no longer continue to move. This is the Hotelling result. The details of this are given in Appendix 1.4 {At the conclusion of the Hotelling game we see that the market share of both firms located at X will be half of the market, exactly the same as at the start of the location game. However, from Figure 1.18 we see that the Hotelling result leads to a fall in ‘consumer welfare relative to the original situation. Given that consumers all consume a fixed quantity per time period of the good produced by firms A and B, there is no substitution effect between the goods produced by firms 4 and B and other consump- tion goods. Therefore, the change in the delivered prices at the each location will accurately reflect the change in welfare of the consumers at each location. The met effect of these welfare gains and losses can be represented by the areas under the delivered price curves, which are arrived at by comparing the delivered prices at the respective locations at the start and the end of the Hotelling location game. The con- sumers who ate located in the centre of the market benefit by generally reduced delivered prices, represented by egih in Figure 1.18, whereas those located at the edges of the market lose by generally higher delivered prices, represented by (def) + (fklm) in. Figure 1.18. The gain in lower prices for the central consumers Is outweighed by the32. URBAN AND REGIONAL ECONOMICS Pricelcost 1e T FIG. 1.18 The welfare implications ofthe Hotellng result increase in prices for the more peripheral consumers, The net effect is therefore a social welfare loss. In one-dimensional space discussed here the Hotelling result holds for two firms. ‘Meanwhile, in the two-dimensional case the Hotelling result olds for the case of three firms. However, beyond these numbers there is no stable equilibrium result as firms keep changing theit location. Moreover, even in the one-dimensional case, the Hotelling result only holds a long as the firms do not compete in terms of prices. If price competi- tion is also a possibility, there is no Hotelling result (d’Aspremont et al. 1979). In Figure 1.19, we can consider the situation where firm A lowers its production price marginally in. time period 1 when both firms are located at X, In time period 2 firm A gains all of the ‘market, From our Cournot conjectures, firm B now assumes that firm A will maintain both its new lower price and its location at X. Therefore in time period 3 firm B also lowers its market price below that of firm A, and now gains all of the market. This process will continue and the long-run Nash equilibrium of this price war is that both firms will end up selling at zero profit while still being located at point X. If the production costs are not zero, but rather are positive, the long-run result ofthis co-location competition will be to drive prices down to the marginal costs of production, which is the typical equilibrium result of a competitive market. However, the Hotelling ‘model is implicitly about monopoly power, with firms able to use location as a means of ‘generating monopoly power over a certain portion of their market. As we see in Chapter 2, the greater is their localized monopoly power, the greater will be the possibilities forINDUSTRIAL LOCATION 23, PrcefCost ‘Market of inte period 1 Market of in time period 1 Market of in tie period 2 with a price cut by fem A FIG. 1.19 The effect of price competition on the Hotelling result the firm to raise additional revenues by employing monopoly practices such as price discrimination. Therefore, in order to generate localized monopoly power, as prices spiral downwards due to the Bertrand problem, each firm has some Incentive to move away from its competitor in order to maintain monopoly power, and consequently positive profits, over some of the market area. However, neither firm has an incentive to move away first, because in doing so, the other firm will then be able to maintain its current prices at the centre of the market and dominate a larger market area than the firm which moved away from the centre. Therefore, unless there is some way in which the firms can ‘mutually agree to move away {rom each other, a price war becomes inevitable with disastrous consequences for both firms. The relationship between the co-location of competing activities and the problem of a price war is known as the ‘Bertrand problem’. ‘Competitor firms will consequently only locate next to each other in situations in ‘which price competition is ruled out either by mutual agreement or by other forms of xnon-ptice competition. Yet, in these types of non-price competitive situations, the spatial clustering of competitor firms is a natural process. Many types of shops and showrooms, for example, such as those for clothing, electronics goods, automobiles, restaurants, and furniture, compete in industries dominated by non-price competition. In these industries prices are used to indicate product quality, and to indicate the types of consumers for whom the good is intended. As such, prices in these industries tend to be fixed. Firms are34 URBAN AND REGIONAL ECONOMICS unwilling to compete by lowering prices because this suggests that the product quality is falling, and this may actually have an adverse effect on sales. The practice of ascribing, prices to products in order to indicate both the product quality and the consumer for ‘whom the product is intended is known as ‘price placing’, and the problem of lower prices implying lower product quality is related to the famous ‘market for lemons’ prob- lem described by Akerlof (1970). At the same time, engaging in non-price competition also implies that the products are not identical, and therefore the Hotelling result would appear not to be relevant. However, in many cases of non-price competition, the differ~ fences between the products are largely superficial, Involving primarily differences in ‘packaging and appearance, The products in essence will essentially till be identical, The fact that firms attempt to make more or less the same products appear very different is known as the ‘Hotelling Paradox’. In these situations, firms will tend to cluster together in space. This is exactly how retail parks and central city shopping areas arise. (On the other hand, where firms produce identical products in which non-price com- petition is extremely difficult, such as the market for gasoline, firms will not cluster together in space. Oil companies which own or franchise out gasoline retail stations will ‘mutually agree not to locate theit outlets too close to their competitors, in order to ‘guarantee some market monopoly power for each station in its immediate vicinity. The only time in which gasoline stations will be located close to each other on the same highway is where they are separated from each other by a centtal reservation, median barrier, or major junction. In these cases, the stations are effectively separated from each other and customers denied the choice between the stations, because drivers are unable easily to switch sides of the road. Therefore, the stations can be considered as not being located together, but rather located away fom each other. ‘The Hotelling result therefore provides us with two important sets of analytical conclu- sions, First, for competitor firms producing the same type of product and which also engage in non-price competition, the spatial competition for markets may encourage such firms to locate next to each other. In other words, spatial industrial clustering can arise naturally where price competition is not paramount. This is particularly important in many examples of retailing. Moreover, in this situation, the market will be split more or less equally between all of the firms in the spatial cluster. This ensures that no firm will be any worse of than its competitor due to an inferior location, a point we will discuss in section 1.5. On the other hand, for firms which produce more or less Identical products for which non-price competition is very difficult to engage in, and in which there are no information problems, spatial competition will encourage such firms to move away fom teach other, The result of this process is industrial dispersion. Secondly, from a welfare point of view, consumers located close to a spatial cluster of firms will tend to experience ‘a welfare gain relative to those located ata great distance away. The reason for this is that the costs of consuming the goods produced by the firms will tend to be much lower for those who are located close to the firms than for those who are located at adistance away. ‘This an important observation concerning agglomeration economies, atopic which we ‘will discuss in the next chapter. ‘When applying these insights of the Hotelling framework to the real world, however, these two observations must be interpreted with caution, because there are some other situations in which price competition and spatial clustering are compatible. This is theINDUSTRIAL LOCATION 35, ‘ase where prices are not predictable and are continually changing, such asin the case of ‘many food markets or gambling activities. In these situations, although price compett- tion is very keen, firms may gain from either short-term first mover advantages, or alter- natively customer inertia in the face of rapid and frequent price changes. The co-location of retailing activites in thts casei justified, as with the case of non-price retail competi- tion, because this may encourage customers to buy more goods in general than they ‘would otherwise if they were not presented with a broad range of consumption alterna- tives and the relevant price information about them. As such, all firms in the cluster are + expected to gain, and co-location ensures that all firms benefit more or less equally. These ‘arguments are related primarily to the questions of information, clustering, and external- ities discussed initially in section 1.5 and at length in Chapter 2. One final point concerning this Palander and Hotelling type of spatial market analysis 1s the criticism that in many real-world cases, individual firms charge the same delivered. price for a given product at all locations. As such, spatial markets are not divided up according to delivered prices which vary with location. On the other hand, where Gelivered prices are invariant with respect to distance within a given market area, this {mplies that the marginal profitability of each delivery will be different according to the location of the customer. This is because the transport costs of outputs must be absorbed by the firm, thereby reducing the net marginal profits from sales as the delivery distance Increases. In other words, the profits associated with deliveries to nearby customers will, be much higher than those for deliveries to distant customers, As such, for any given spatial distribution of markets, the location of the firm will still determine the overall profitability ofthe firm. Moreover, as we see in Appendix 1.3 and Appendix 3.4, even for uniform delivered prices, firms are able to employ changes in the quality of service, such. as changes in delivery frequencies, in order to mimic the spatial price effects of situations in which customers pay the transport charges in addition to the quoted source prices. - 1.5 Behavioural Theories of Firm Location ‘The models discussed so far ely on the assumption that ‘rational’ firms will aim to use thelr Jocation behaviour in order to maximize their profits. We have also assumed that the information available to the firms is suficient for them to do this. However, in reality the information available to firms is often rather limited, Moreover, different firms will ofter-have diferent information available to them. For this reason, some commentators hhave argued that firms cannot and do not make decisions in order to maximize their profits, Rather, they argue that firms make decisions in order to achieve alternative goals, other than simply profit maximization, Therefore, from the perspective of location the- ‘ory, this critique might suggest that the underlying motivation of our models would need to be reconsidered. The critique has three themes, namely bounded rationality, contlict= {ng goals, and relocation costs, The first two themes can be grouped under the general heading of Belavioural Theories, and were not originally directed at location models in particular. The third theme is essentially a spatial question.Le 36 URBAN AND REGIONAL ECONOMICS ‘The arguments conceming ‘bounded rationality’ are most closely associated with ‘Simon (1982, 1959). This critique concerns the fact that firms In the real world face limited information, and this limited information itself limits firms’ ability to be ‘rational’ in the sense assumed in microeconomics textbooks. These arguments are & ‘ote general critique of rationality within microeconomics as a whole. However, they have been argued to be particularly appropriate to the question of industria location behaviour. The reason is that information concerning space and location is very limited, due to the inherent heterogeneity of land, real estate, and local economic environments ‘Therefore, when considering location issues, it would appear that the ability of the firm to be ‘rational’ Is very much ‘bounded’ by the limited information available tot, In these ‘cieumstances, decisions guided by straightforward profit-maximization behaviour ‘appear to be beyond the ability of the firm. Therefore, location models based on this ‘assumption seem to oversimplify the location issue. Location behaviour may be determined primarily by other objectives than simply profit maximization. "Where firms face limited information, Baumol (1959) has argued that firms will focus on sales revenue maximization as the short-run objective of their decision-making. One teason for this is that sales revenue maximization implies the maximum market share for the firm in the short run, Where information is limited, current market share is deemed by many observers to be the best indicator of a firm's long-run performance, Decause It provides a measure of the monopoly power ofthe firm. The logic ofthis approach Is that the greater isthe market share of the firm, the greater is the current monopoly power of the firm, and the greater will be the firm's long-run ability to deter potential competitors ‘through defensive tactics such as limit-pricing and cross-subsidizing. From the perspec- tive of location models, this may imply that the firm will make location decisions primar- fly in order to ensure maximum sales revenues rather than maximum profits. In the Fotelling model above these two objectives coincide. However, ifthe costs of production ot transportation faced by the firm were to vary with location along the line OL, as they do in the Weber and Moses-type models, the two objectives of sales maximization and profit maximization may not coincide at the same location point in the Hotelling model ‘The eventual location resuilt will therefore depend on which particular performance ‘measure the firm adopts and chooses to maximize. “The second critique of profit maximization as the decision-making goal of the firm is that of ‘conflicting goals. This critique is most closely associated with the work of Cyert ‘and March (1963). The argument here is that in a world of imperfect information, the Separation of ownership from decision-making in most major modem firms means that business objectives are frequently pursued which are different from simply profit maxi- ization. Only shareholders have a desire for maximum profits in the short run. On the other hand, in modern multi-activity, multievel, multi-plant, and multinational frm ‘organizations, corporate decisions are the result of the many individual decisions made by a complex hierarchy of people, each with particular business objectives, and many ‘of which are different from profit maximization. The reason is that the performance of different employees within a company is measured in different ways. For example, the directors’ performance may be evaluated primarily by the firm’s market share, whereas the sales manager's performance may be evaluated on sales growth. Similarly, the pro- duction manager's performance may be evaluated primarily by inventory throughputINDUSTRIAL LOCATION. 37 efficiency, whereas the personnel officer may be evaluated according to the number of days lost through industrial disputes. Given that each of these different decision-makers is evaluated on different criteria, the success, promotion, and consequently the wages ‘earned by each of these workers will be evaluated differently. Therefore, the objectives pursued by different employees may be quite different from profit maximization. Under these conditions, the ‘conflicting goals’ critique suggests that firms will aim to ‘satisfice’ In other words, the firm will aim to achieve a satisfactory level of performance across a range of measures. In particular, the firm will initially aim to achieve a level of profit sufficient both to avoid shareholder interference in directors’ activities and also to avoid the threat of a takeover. Once this objective is achieved, the other various goals of the {firm can be satisfied. For example, the firm may aim to achieve market share levels as high as possible without jeopardizing the efficiency cost gains associated with production and logistics operations. Equally, employees’ pay may be increased in order to encourage firm loyalty. The point here is that the overall objective of the firm can be specified in various ways. In Figure 1.20, the firm's total profit function Tx is constructed as the difference between the total revenue function TR and the total cost function TC. The firm may ‘choose to produce at output levels Q,, Q,, and Q,, which represent the minimum cost ‘output, the maximum profit output, and the maximum revenue output levels, respect- ively. All of these levels of output produce a profit level sufficient to maintain a firm's independence z, but only one of these output levels Q. is the profit-maximization level of output. From the point of view of location, if we have a set of spatial total cost and revenue ‘curves, such as those described by Figure 1.20, the firm will make different location Revenuerofit | 3 @ @ Sutporo FIG. 1.20 Profits maximizing, revenue-maximizing, and profit-satisicing138 URBAN AND REGIONAL ECONOMICS decisions, according to whether the firm is aiming to maximize profits in the short run or ‘whether It is aiming to earn satisfactory profits in the short run along with achieving some other goals. For example, in Figure 1.21, if the firm is aiming to maximize profits in the short run it will locate at point P. On the other hand, if is aiming to maximize sales it will locate at S, and if it is aiming to minimize production costs and to maximize production efficiency it wil locate at C. Ifthe firm had perfect information regarding these different spatial cost and revenue curves, we can argue that the firm will always ‘move to polnt P. However, behavioural theories assume that information is imperfect. Given the limited information available and the conflicting goals within the organiza tion, the actual location behaviour of the firm will depend on which is the particular dominant objective of the firm. The third eritique of the classical and neoclassical location models comes from the ‘question of relocation costs. Relocation costs are the costs incurred every time a firm relocates. The models described above all assume that location is a costless exercise. However, relocation costs can be very significant, comprising the costs of the real-estate site search and acquisition, the dismantling, moving, and reconstruction of existing factl- ities, the construction of new facilities, and the hiting and training of the new labour employed. These significant teansactions costs, along with imperfect information and conflicting goals, will mean that firms are unlikely to move in response to small varl- ations in factor prices or market revenues. In Figure 1.21, the areas in which positive profits are made, Le. where TR> TC, are known as ‘spatial margins of profitability’ Rav¥- stron 1958), and are represented by the areas between locations a and b, cand d, andeand The relationship between marginal location change and the profitability of the firm in. these areas is given by 4(TR~TC)/03, and this is represented by the differences in the slopes of the spatial revenue and spatial cost functions as location changes. Inthe spatial ‘margins of profitability in which the slopes of the spatial revenue and spatial cost functions are very shallow, the marginal benefit to the firm of relocation will be very oP >< 3 dec f FIG, 1.21 Spatial cost and revenue curves