Ronald Coase's theory of the organization (1937), which attributes the boundary between the market and the organization to transaction costs, has been accused of lacking empirical content. In this note, I operationalize three concepts that are both critical to Coase's theory and ambiguous in meaning: organization, cost, and transaction. I also provide a commonplace example that demonstrates the validity of Coase's fundamental insight. This example was inspired by several of Coases recent observations on the nature of transaction costs (1988a). Coase explains that, although the "gains which accrue from the existence of the organization come from a reduction in transaction costs , the main transaction costs that are saved are those which would otherwise have been incurred in market transactions between the factors now cooperating within the organization" and the organizers of the organization. Coase (1988b) further explains that these factors include both people and things that agree to obey the directions of the organizations organizer for remuneration. He also notes with respect to things, "the times of delivery, the quantities to be dispatched, and the places to which they are to be delivered are not, for purchases of most commodities, matters of minor importance" (41-42). He concludes that the relationships between the costs of organizing and the costs of transacting "are extremely complex, involving pricing practices, contractual arrangements and organizational forms (47).
This example also demonstrates that neither opportunism nor asset specificity is needed to explain the existence of the organization or its size, which is also Coases position (1988a). Coase notes that many economists "seem to believe that vertical integration comes about mainly when there is asset specificity, because of the incentive for opportunistic behavior to which this gives rise," but he doubts that "there is such a systematic relationship." Jointness, indivisibilities, transport costs, and perhaps ultimately ignorance will suffice to explain vertical integration. Nevertheless, the following case also shows that organizations will be larger in the presence of opportunism than in its absence. Finally, it indicates why, despite the inherent power and utility of the concept, transaction costs are difficult to operationalize and discussions of asset specificity are frequently inconclusive.
Coase implies that an organization is simply a bundle of things or assets -- the bigger the bundle, the larger the organization. This bundle can and usually will include both fixed capital and working capital. By fixed capital I mean both tangible assets (e.g., real property, plant and equipment, etc.) and intangible assets (human capital, technologies, products, brands, distribution channels, etc.). Working capital is the difference between short-term or liquid assets -- such as cash, inventories, accounts receivable, and prepaid expenses -- and short-term liabilities.
Presumably, an organization's value will be greater than the sum of its parts: bundling should increase the value of an organization's "cooperating" assets over and above the value of those same assets unbundled. Otherwise the organization's existence (as a value-creating, profit-seeking entity) is not justified. This means that a rational manager should willingly replace part of a organization's current bundle of assets (cash can be obtained by liquidating an existing asset or by acquiring an additional liability) with an alternative asset if and only if the expected residual value of the bundle would thereby be increased. Conversely, if the liquidation value of an asset exceeds its contribution to the bundle, it should be transformed to cash -- thereby reducing fixed capital and increasing working capital.
The organization is a stock concept -- a precisely dimensioned reality at an identifiable point in time. Accountants usually define costs as flows; costs reflect changes in stocks over a fixed temporal interval. In this instance, however, I can avoid combining apples and oranges by adopting Armen Alchian's definition of cost as:
the change in equity caused by the performance of some specified operation, where for simplicity of exposition, the attendant change in income is not included in the computation of equity. Because of logical difficulties in converting this present value concept into a satisfactory rate (per unit of time) concept I measure costs in units of present value or equity. Hereafter, the unmodified expression "costs" will always mean the present worth, capital value concept of cost.
Finally, the accounting definition of a transaction is: an identifiable operation carried out by or through an organization that transforms or converts an asset. All transactions conform in theory to the dual aspect principle: every transaction affects at least two accounts and obeys the fundamental accounting equations that assets equal liabilities plus owners' equity, and debits equal credits. Of course, in reality many of the transactions carried out by and through an organization are not recorded in its financial accounts.
The best measure of a debt-free organization's size is its share value. According to the capital asset pricing model, the share value of a debt-free organization is equal to the risk-adjusted discounted net present value of its future cash flows (or, given long-term liabilities, both explicit and implicit, expected future cash distributions to shareholders). If an organization's balance sheet were complete and accurate, it too would accurately measure organization size. Balance sheets, however, are neither complete nor accurate.
Their incompleteness and inaccuracy has a direct relationship to the organizations purpose as a profit seeking entity and is due in no small measure to asset specificity. An asset is said to be "specific" if it makes a necessary contribution to the provision of a good or service and has a significantly lower value in alternative uses. If it were costless to redeploy an asset to alternative uses, its liquidation value would by definition be equal to its contribution to the organization. And if an asset's liquidation value were its market price, its liquidation value would also be equal to its replacement cost. Hence, if all of a organization's assets were valued at replacement cost (rather than historical cost), their sum -- as depicted in the organization's balance sheet -- would equal the organization's market value (although there would be no reason for the organization and therefore presumably no organization).
The discrepancy between an asset's historical cost and its contribution to organization value, between replacement cost and liquidation value, or between opportunity cost ex ante and measured cost ex post is likely to be greatest where intangible fixed capital -- human and intellectual capital -- is concerned and least where working capital is concerned. At the limit, an organization's cash balances are presumed to be absolutely liquid and perfectly redeployable. Tangible fixed capital in the form of plant or equipment is usually an intermediate case.
One possible reason for the inconclusiveness of the transaction cost literature is that most transaction cost analysts have focused on difficult, complex problems involving intangible capital, fixed assets, or even capital structure. Fortunately, it is possible to make several points of fundamental importance about the nature of transaction costs and their relevance to the boundary between the market and the organization by focusing on simple problems involving working capital.
The balance sheets of many organizations, including all those that specialize in marketing consumer goods, are dominated by working capital in the form of raw material and parts stocks and work-in-progress and finished-goods inventories. Where this is the case, it may be stipulated that organization size is a function of inventory levels. The implication here is that if inventory were a retail organizations only asset and its inventory fell to zero, the organization would have zero size, i.e., the boundary between the market and the organization would disappear. It should perhaps also be noted that in principle a retail organization could contract with a second party, say a warehouse, to handle reordering, shipping, and storage services for it. Hence, the decision of a retail organizations organizers to hold inventory in house is directly analogous to the vertical integration decision that tends to dominate the transaction cost literature. It is a settled question in economics that transaction costs determine optimal inventory levels. Hence, it follows that insofar as inventories are concerned, transaction costs directly influence the location of the boundary between the organization and the market.
To show that this is the case, I will summarize the basic inventory problem and its solution. The inventory problem is similar to all other asset and acquisition (or divestment) decisions, including the make or buy decision and the maintenance level problem, in that its proper formulation is as a capital budgeting problem.
In this case a retailer sells precisely Q* units of commodity n each year; i.e., Q* is given and known to the retailer. The price of the commodity n is predetermined, and demand for it is spread evenly throughout the year. The retailer is a price taker in both retail and wholesale markets. The inventory problem can be stated as: how many units of n should the retailer hold on average (n*) to minimize costs? Moreover, this problem can be restated in terms of the optimal number of orders (o) of quantity D of commodity n, or the optimal order quantity (D). Thus where the stock of n varies between a maximum of D and a minimum of zero:
o = Q*/D (1)
n* = [D+0]/2 = D/2 (2)
In this problem there are two kinds of cost: (1) carrying costs, which include the cost to the retailer of capital invested in inventory, storage costs, etc., and (2) transaction costs, which include the cost of ordering, shipping and receiving, processing, and otherwise handling deliveries. One could of course think of carrying costs as transaction costs, especially where buffer stocks are held as a hedge against uncertainty. Moreover, the relationship of carrying costs to what we have called transaction costs is that of dual to primal. Nevertheless, the distinction between carrying costs and transaction costs has practical merit. If every relevant cost is a transaction cost, transaction costs cannot possibly explain anything in particular. In this instance, total carrying cost per annum are assumed to be equal to the annual carrying cost per unit (k*) times the average number of units in inventory (D/2). Total transaction costs are assumed to be equal to the shipping and handling cost per item (b*) plus the cost per order (a*) for such activities as accounting for and processing the order. Transaction costs per order are thus equal to a* + b*D, and total transaction costs are o(a* + b*D) or a*Q*/D + b*Q*. Hence, the retailer's total inventory cost (C) is:
C = k*D/2 + a*Q*/D + b*Q* (3)
By differentiating C with respect to D, I obtain the optimal value of D, i.e., that which minimizes total inventory cost.
dC /dD = k*/2 - a*Q*/D2 = 0 (4)
k*/2 = a*Q*/D2 (5)
Multiplying both sides of (5) by 2D2/k* and rearranging the equation gives:
D2 = 2a*Q*/ k* (6)
D = [2a*Q*/ k*]1/2 (7)
Solving equation (7) gives the retailer's optimal order quantity (D) and her optimal average inventory (D/2) for any given level of sales volume (Q*), per-unit carrying cost (k*), and fixed reorder cost (a*). Inspection of this equation clearly demonstrates that the optimal average inventory level (D/2) increases when transaction costs in the form of reorder costs increase.
Note that in this most elementary formulation of the inventory problem, demand was certain, opportunism was absent, and inventories were implicitly assumed to be convertible to cash at any time. Hence, this example demonstrates that, in the limiting case, organization size (and therefore the boundary between the market and the organization) is uniquely determined by the cost of executing transactions.
One implication of this result is that opportunism and asset specificity are not necessary to an explanation of the location of the boundary between the market and the organization. This does not mean that uncertainty, opportunism, and asset specificity are unrelated to the question of organization size. The inventory problem can easily be expanded to show that uncertainty about prices and demand increases the level of the retailers optimal inventory (although increased undiversifiable risk would also increase the organizations cost of capital, which would tend to have an offsetting effect on inventory). Furthermore, insofar as variables such as opportunism increase the cost of conducting transactions, they also have the effect of increasing inventory levels and, in the instant case, organization size, as well as decreasing technical efficiency in precisely the same way that friction reduces thermodynamic efficiency. For example, the possibility of error may be sufficient to explain accounting for the receipt of shipments and the like, but cannot account for the panoply of administrative controls one observes in practice. Clearly the threat of employee pilferage, kickbacks in purchasing, or embezzlement leads to an increase in the scope and severity of a organizations accounting controls, which increases the cost of ordering, shipping and receiving, processing, and otherwise handling deliveries and therefore the size of its inventories. Of course these predictions could easily be tested and corroborated or refuted.
The inventory problem provides the simplest, most elementary case for the relevance of transaction costs to the determination of the boundary between the market and the organization. The reasons I can effortlessly apply transaction cost analysis to this case also show why it is so hard to do so in others. As James Buchanan explained, Coases theory of the organization is a theory of choice. In it transaction costs are subjective -- they exist in the mind of the decision maker and nowhere else, they are necessarily forward looking or ex ante, and they "cannot be measured by someone other than the decision-maker because there is no way that subjective experience can be directly observed."