location for such an activity, the first step in the selection process might well be to winnow down the alternatives to a limited places. Thus one would not ordinarily expect to find patent lawyers, opera houses, investment bankers, small cities. atial pattern of advantage is not obviously systematic at all-that is, it random. Tax rates this category. Some general statements can be r merit further ity or local market, and those involving labor cost reasons s, for which transportation costs are in ansfer costs on oment of a systematic le for emphasis on costs are functionally inated by differential oth. Similarly, we can model, it will be newed as a service; thus, one n of transfer r se. It also al distances we are to nging inputs tota ut The total costs. profits must take ng to maximize profits (revenue less cost) ul to start al points, but that the unit is too small to have and it lelivering its output, so se as distance from th tion on the input side as we have just done on the output side. I s, but at each source the supp ght. Consequently, on to the already mentioned Our third tion or scale of operations. ns, which will be addressed cation to the much simpler prob such factors as processing-co ing prices by the business unit under ce Fina m per ton mile, regardless of d postpones(until the next chapter) a recognition of the various lly appear in transte costs in If the unit in question uses only one kind of transferable input (say, wood)and produces one kind of transferable output (say, baseball bats), then the choice of the most profitable location is easy to describe. The first question to be settled is that of input orientation versus output orientation. Will it be preferable to make the bats at a wood source, or at the market, or at some9 location for such an activity, the first step in the selection process might well be to winnow down the alternatives to a limited set of sufficiently large places. Thus one would not ordinarily expect to find patent lawyers, opera houses, investment bankers, or major league baseball teams in towns or small cities. Finally, there are location factors for which the spatial pattern of advantage is not obviously systematic at all—that is, it cannot be described or predicted in any reasonably simple terms, although it is not necessarily accidental or random. Tax rates, local water supply, labor supply, and quality of public services seem to fall into this category. Some general statements can be made to explain the broad outlines of the pattern (such a statement is attempted for labor costs in Chapter 10); but for making comparisons for actual selection of locations there is no way of avoiding the necessity of collecting information about every individual location that we wish to consider. Among the kinds of patterns of differential advantage that location factors may assume, three in particular merit further discussion: those determined by transfer costs, those determined by size of city or local market, and those involving labor cost. We turn here to the transfer cost case, reserving the other two for consideration in later chapters. 2.5 TRANSFER ORIENTATION Until fairly recently, location theory laid exaggerated emphasis on the role of transportation costs, for a number of reasons. Interest was particularly focused on interregional location of manufacturing industries, for which transportation costs are in fact relatively more important and obvious than for most other kinds of activities. Moreover, the effect of transfer costs on location is more amenable to quantitative analysis than are the effects of other factors, so that the development of a systematic body of location theory naturally tended to use transfer factors as a starting point and core. A basic rationale for emphasis on transfer advantages is given by Walter Isard: "Only the transport factor and other transfer factors whose costs are functionally related to distance impart regularity to the spatial setting of activities."7 We can speak of a particular activity as transfer-oriented8 if its location preferences are dominated by differential advantages of sites with respect to supply of transferable inputs, demand for transferable outputs, or both. Similarly, we can call an activity labor-oriented where the locational decisions are usually based on differentials in labor cost. Let us look first at a simple model of transfer orientation. In order to facilitate the development of this model, it will be helpful to consider the concept of production. In traditional nonspatial economic theory, production is viewed as a transformation process. One uses factors of production in some combination in order to produce a good or service; thus, one "transforms" inputs into outputs. Later in this chapter, we shall find that the nature of that transformation process may itself influence the location decision. However, for our immediate purposes, it is important to recall from the discussion of transf er factors earlier in this chapter that the activity of a locational unit involves much more than transformation per se. It also involves the acquisition of inputs and the distribution of output, both of which may require transfer over substantial distances. The same might be said about the activity of a household or other nonprofit establishment. Space plays an essential role in economic activity. Given this, it is easy to recognize that the costs incurred by the firm also have a spatial component. If we are to understand the behavior of business establishments, we must be concerned with the costs associated with bringing inputs together and distributing outputs, just as we are concerned with the costs of transforming inputs into output. The total costs, therefore, include these three components, and a locational unit that is seeking to minimize costs or maximize profits must take them all into consideration. Let us focus now on the behavior of a single-establishment business firm aiming to maximize profits (revenue less cost) and seeking the best location for that purpose. We shall see that the problem can be quite complex, so it will be helpful to start off with some simplifying assumptions that can later be relaxed. First, we shall assume that there are markets for this unit's output at several points, but that the unit is too small to have any effect on the selling price in any of those markets. In other words, demand for the unit's output is perfectly elastic, and it must take the prevailing prices as given, regardless of its volume of sales. The firm has to pay for the costs of delivering its output, so there is some incentive to locate at or near a market. Costs associated with distribution of output rise as distance from the market increases. We simplify the case further by making exactly the same kind of assumption on the input side as we have just done on the output side. In other words, the kinds of transferable inputs our unit uses are available at different sources, but at each source the supply is perfectly elastic, so the price can be taken as given regardless of how much of the input is bought. Consequently, there will be a cost incentive for the unit to locate at or near a source of transferable inputs, in addition to the already mentioned incentive to locate at or near a market. Our third assumption is that the unit's processing costs (using local inputs) will not vary with either location or scale of operations. These three simplifying assumptions bypass some highly important factors bearing on the choice of locations, which will be addressed in later chapters. What we have done for the present is to reduce the problem of a maximum-profit location to the much simpler problem of minimizing transfer costs per unit of output, by postponing consideration of such factors as processing-cost differentials, economies or diseconomies of scale, and control over buying or selling prices by the business unit under consideration. Finally, we can simplify the problem of minimizing transfer costs by letting transfer costs be uniform per ton mile, regardless of distance or direction. This assumption of what is called a uniform transfer surface postpones (until the next chapter) a recognition of the various differentials that typically appear in transfer costs in the real world. If the unit in question uses only one kind of transferable input (say, wood) and produces one kind of transferable output (say, baseball bats), then the choice of the most profitable location is easy to describe. The first question to be settled is that of input orientation versus output orientation. Will it be preferable to make the bats at a wood source, or at the market, or at some