Why do firms contrive shortages? The economics of intentional mispricing.

Haddock, David D. ; McChesney, Fred S.

Economists typically inveigh against government-imposed price controls, finding them and the shortages they create to be Kaldor-Hicks inefficient.(1) But not all observed shortages result from government price controls. When firms experience sudden changes in costs or demand, they sometimes voluntarily but temporarily refuse to clear the market rather than alter prices.(2) Depending on whether the temporary price is above or below market-clearing levels, a firm's intentional refusal to alter price forces potential buyers or sellers to queue, so that rationing is required and both the firm and its clients incur additional costs [Barzel 1974]. When the abnormal events pass, mispricing and rationing disappear as well. But price alteration was always open to firms in those situations, and no widely accepted price-theoretic explanation exists for their insistence on mispricing.(3) That is the phenomenon investigated in this article.

Economists' inability to derive a convincing model of episodic mispricing, and in particular to test various models using conventional empirical methods, has begun to spawn new theories at variance with received economic doctrine (e.g., Basu [1987]; Kahneman, Knetsch and Thaler [1986]). One recent study by Blinder [1991] has addressed the problem through interviews, in the belief that price stickiness can be explained by asking business executives themselves for the reasons. Though preliminary, that study has already attracted substantial attention.(4)

Those few, more orthodox price-theoretic inquiries into voluntary mispricing fall into two distinct categories. In the first category, researchers model inventories and order backlogs in which all participants are ignorant of present or impending market supply and demand, due largely to stochastic features of markets.(5) Inventories or backlogs are expectable, because it is not economic (or possible) for market participants to predict market supply and demand with precision. Indeed, in some of those situations the apparent queues are not even real. For example, De Vany and Frey [1982] show that where production processes are especially time consuming, apparent queues are actually a listing of forward orders by buyers who would not even want delivery at present.

By contrast, in the second category, no participant need be ignorant of impending market conditions.(6) Rather, queues arise either because buyers prefer to consume some products (restaurant meals, rock concerts, athletic events) as part of a crowd [e.g., Becker 1991]; because under-pricing of some goods is preferable when the alternative is other buyers just taking the goods [Cheung 1977]; because the cost of waiting in a queue is low relative to the cost of continuously adjusting price (e.g., ski lifts [Barro and Romer 1987] or peak-time telephone service(7)); or even because--jumping outside price theory--ill-defined ethical considerations interdict market-clearing price adjustments (e.g., following catastrophes [Kahneman et al. 1986]).

A third logical category exists--one in which participants on one side of a market are substantially better informed about market conditions than are participants on the other. That third category is the one studied here. In retail markets, sellers are able to ascertain the nature and likely duration of market alterations more often than are buyers. In upstream markets, some buyers survey a larger part of the market than do sellers, and so are in a better position to interpret market conditions.

Logically, the analysis here would apply to both situations, but the examples that we consider are of the former sort. Thus, this article considers situations in which perfectly informed sellers tolerate shortages that cost them money in the short run in order to protect long-term, profitable relationships with consumers who are not as well informed.(8) The analysis is based upon a conventional price-theoretic model and specifies empirical tests showing why and how profit maximizers nevertheless run shortages intentionally.

In the spirit of Occam's Razor, our approach is simply to adapt the existing theory of specific capital [Klein, Crawford and Alchian 1978; Williamson 1979] to consumers' product-specific information. The theory of specific capital begins by noting that contracting parties often benefit from investments that are specific to their mutual ongoing relationship. In other words, it is advantageous to the group if some resources are devoted to producing capital (in our case intangible informational capital) that is of less or no use if the relationship breaks down. To make such an investment, a party must expect an enhanced flow of benefits from the relationship, which is the return on the relation-specific investment.

But, in the absence of special precautions, a completed relation-specific investment is highly vulnerable. If the relationship breaks down, the return that was the motivation for the initial investment disappears. What would have seemed a good investment on the assumption that the relationship would continue becomes a poor investment. Thus, parties are less inclined to make valuable relation-specific investments the less secure the relationship. That means that the parties to a relationship have an incentive, individually and collectively, to enhance expectations that the returns to relation-specific investments will be forthcoming.

In the case at hand, we focus on potential buyers making investments in ascertaining the price/quality ratios of competing products. Once a buyer has identified the product with the best ratio for his purposes, he is in possession of intangible, relation-specific informational capital that is valuable not only to himself but to the producer of the product so identified. Such an investment is costly, and buyers will not renew a search until and unless they have reason to believe that their informational capital has become obsolete. When a seller, who ordinarily surveys more of the market than does a buyer, believes that the buyers' informational capital is sound for the long run, the seller will benefit by preventing buyers of his product from renewing the costly search process merely because transitory events might make it appear that the capital is obsolete. Buyers also benefit, by being dissuaded from costly new search that (the seller knows) is unnecessarily.

As background, section I notes several instances of intentional underpricing. Section II then applies economic theory to explain such privately contrived shortages in the face of increased costs or demand. The analysis hinges on the value of firm-or product-specific investments in information that customers make, and investments to provide information that firms make specific to their customers. Under some conditions, price changes will erode so much information-capital value that they will be avoided. The result does not require that buyers prefer consumption as part of a crowd, as in Becker [1991]. Sellers need not make estimation errors, either; even a fully informed seller would, under the hypothesized conditions, choose to contrive shortages for a time rather than alter prices when demand or costs increase. Nor does the result require an exogenously assumed ethical revulsion against clearing market by price, as in Kahneman et al. [1986].

The model has refutable empirical implications, as section III shows. Counter-intuitively, firms sometimes have an incentive to increase advertising and other promotional devices when shortages develop. Perhaps most interesting are the implications for queuing theory. If firms intentionally contrive shortages, they have an incentive to structure queues in predictable ways, sometimes including discouragement of the formation of any queue at all. The model also yields predictions about firms' policies concerning resale of items that are in short supply. Section IV presents qualitative corroborating evidence, drawn from the examples introduced in the first section. Our model will not end lunch-table discussions about excess demand; one still needs several independent models to explain the entire set of observed private shortages. Ideally, a single model would explain the entire set, and such a model may still be in the offing. But for the moment, the model here contributes to a superior understanding of shortages by encompassing several seemingly distinct phenomena within a more general price-theoretic structure.

I. THE SHORTAGE PUZZLE

Price changes are only one equilibrating mechanism firms use to control imbalances arising between quantities demanded and produced. Movement along the firm's short-run marginal cost curve dampens price changes. The firm's long-run analogue is a capacity adjustment. Both alterations substitute quantity changes for at least some price adjustments. In most industries also, firms use input and output inventory fluctuations to respond to transient supply or demand shifts. As marginalists, economists are not surprised that firms ordinarily use several arrangements simultaneously.

Conventional theory can explain a firm's dispensing entirely with a particular adjustment mechanism. A firm that fails to vary output with price must have inelastic short-run marginal costs. One that fails to adjust capacity must already be operating at the minimum point of its long-run average cost curve, so that adjustments in industry capacity will come from entry or exit. If the firm carries no inventories of outputs, as in service-producing industries, the costs of maintaining them is assumed to be prohibitive. And so on.

But when, occasionally, a firm resorts to shortages rather than change prices, economic puzzlement intensifies. Implicitly, most microeconomists assume price changes are nearly costless, and so prices should be varied promptly and continuously. Yet, sellers in several settings persistently and intentionally hold the line on prices, leading to excess demand and queues.

To take one example, Olmstead and Rhode [1985, 1044] note that "[i]n the spring and summer of 1920 a serious gasoline famine crippled the entire West Coast, shutting down businesses and threatening vital services.... [W]estern oil marketers voluntarily suppressed price advances and, instead, created and administered a complex allocation scheme." The authors note similar episodes, like the "coal famine" in the winter following a 1902 miners' strike.

More contemporary examples of shortages without external price controls are ubiquitous. Airlines intentionally create regular shortages in two ways: by offering fewer seats at certain fares than the quantity demanded, and by purposely over-booking flights. When natural disasters like Hurricane Andrew occur, many merchants choose shortages and queues over price increases. Foreign and domestic auto companies have sometimes maintained prices below market-clearing levels, rationing their product among dealers and discouraging them from increasing price, an episode reminiscent of Henry Ford's underpricing of the original Model T. L. L. Bean once responded to an upsurge in demand by refusing to send catalogues to those who were not already on its mailing list. Newspapers typically do not vary the number of papers printed or the price charged, even on days when a particularly newsworthy event makes it likely that the issue will sell out [McChesney 1987]. Parisian restaurants maintain prices below market-clearing levels during summers when demand, particularly by tourists, increases abnormally [Ricklefs and Kamm 1985]. Intel Corporation persistently ran shortages of its 386 chips for personal computers.

In all those cases, profit-maximizing producers have foregone price increases, although no external price controls existed. Likewise, in few if any of the cases were the sellers surprised that the queues persisted, once in place: in some industries, for example, occurrence of seasonal shortages can be predicted with virtual certainty. The issue analyzed here is why some sellers on occasion prefer queues and rationing systems to price changes that promptly would clear markets.

II. THE MODEL OF PRIVATE PRICE CEILINGS AND RATIONING

Information about product attributes is a common element in all the shortage examples above. Various marketing practices demonstrate that purchasers invest in firm- and product-specific information. For example, "free" samples are given away because they provide free information, so some consumers become repeat purchasers. The logo or trademark by which customers can identify a firm or product is valuable only because customers use it to save costs of locating or identifying what they want time after time without rounds of costly procurement of information.

Once buyers judge a product to have acceptable price-quality attributes, making further search uneconomic, they continue to purchase that product until something indicates that search is again worthwhile. "Consumer loyalty" implies that the seller has amassed a stock of intangible capital, often called goodwill. That capital spares consumers costly acquisition of information. Goodwill associated with a product or firm assures superior purchaser satisfaction. Customer goodwill is demonstrably valuable to sellers. Many firms are bought and sold at prices well above the replacement cost of plant and personnel.

Firms' investments to create customer goodwill also indicate its value.(9) To consumers, those investments are a substitute for their own search; the value of goodwill will be greater, the higher the costs of buyers' search [Telser 1973, Klein and Leffler 1981]. Advertising, economists recognize, is properly treated as an investment rather than a current expense when it increases the likelihood of repeat purchases [Ayanian 1975]. Because a reputation for honesty increases the likelihood of repeat dealings, sellers forego short-lived profits they could earn if they cheated customers [Telser 1980].

A price increase may upset the no-search equilibrium, however, and thereby devalue a firm's hard-won goodwill capital. The price increase conveys an ambiguous signal to rationally ignorant buyers. It may indicate an upward shift in demand or in costs for all similar products, which therefore will experience a similar increase in price. If so, buyers' ideal strategy ordinarily would not be to switch brands, but merely to reduce the quantity purchased.

But a buyer discovers only ex post, by search, whether the price increase is general. His original loyalty arose because he searched only until the expected marginal benefit of search equaled its marginal cost, not until the marginal benefit was zero. Consequently, in learning whether the recent price increase argues for a shift to a new product, the buyer may chance upon a brand that has always been a better bargain than the present choice. If that happens, the buyer's loyalties switch.(10) (Whether a shortage also proves destructive of consumer loyalty depends endogenously on the firm's allocation of the product and its management of queues, as discussed in the next section.)

Alternatively, suppose the buyer discovers that Brand A's price has increased relative to alternatives; demand or costs for only that brand have increased. Then the buyer will more likely switch his loyalty to a competitor, Brand B. Two possibilities then arise. Suppose, first, that Brand A's price rose because demand for it has increased. On net, Brand A would still seem to have benefited. It has lost the loyalty of some old buyers, but new customers paying higher prices have replaced them.

That conclusion sometimes is unwarranted, however. A demand increase may be temporary, reflecting a fad among transient buyers. A price rise would diminish the firm's stock of goodwill among loyal customers. Regaining the loyalty of the old clientele will be costly. A priori, the cost of regaining clientele is not necessarily less than the opportunity cost of foregoing sales to transitory buyers. The firm must estimate which course is apt to be more profitable. So a firm believing that a demand increase is transitory might quite rationally restrain prices and serve only loyal buyers, thus creating excess demand and potential queues of transitory buyers.

Even if a demand increase for Brand A is permanent, the optimal policy may be to favor customers with large specific investments in information. An unanticipated, permanent demand increase will leave the firm temporarily with insufficient capacity to clear the market without transitory price increases. If capacity expansion appears warranted, it will take time, because the cost of expanding capacity varies inversely with the time taken to expand [Alchian 1959]. Temporary price increases in the interim will depreciate loyal buyers' information about anticipated future price/quality relationships, while causing potential new buyers to accumulate a stock of inaccurate information ("Brand A is and will continue to be high priced"). Correcting inaccurate information is costly. In some instances, the long-run costs of lost sales may exceed the revenues foregone by a failure to increase price in the short run. Indeed, a supplier might even reduce price during a shortage, if the long-run price following capacity expansion is expected to be lower than the current price (because of scale economies, for example).(11)

Similar issues would arise when Brand A's price rises because of cost (rather than demand) increases that only producer A faces. Production cost may increase temporarily (e.g., a labor strike, transport disruption, or the like, affecting only one firm). But passing along price increases costs the firm customer goodwill. If customers forsaking Brand A make product-specific investments in Brand B, when normal cost conditions return those customers cannot be induced costlessly to renew their investments in and purchases of Brand A. Even if the cost increase is expected to be permanent, the firm may expect to mitigate its effects by longer-term adjustments (e.g., by altering capital-labor ratios in response to higher labor costs). If so, the firm will find full price adjustments along the steeper short-run marginal cost curve unattractive, ceteris paribus, because they will convey inaccurate information to current purchasers about the firm's long-run prices [Brennan, Buchanan and Lee 1983].

This simple model of shortages merely extends the now-familiar Klein-Crawford-Alchian [1978] model of specific capital to include an intangible asset, customer loyalty. Earning loyalty, or anything else of value, is costly. Foregoing an ability to increase price is simply the opportunity cost some firms voluntarily accept in return for more valuable longer-term customer loyalty. In that sense, running a shortage is the analogue of investments in reputation in the Klein-Crawford-Alchian model: costly in the short run but more than paying for themselves in the long run. Obviously, the higher the cost or the lower the value of consumer loyalty, the less likely firms will make the investment; thus, there will be a distribution of firms, some endeavoring to maintain consumer loyalty and others not.

III. EMPIRICAL IMPLICATIONS

A principal failing of almost all models of non-market-clearing prices is the inability to distinguish among them empirically. Because "all the theories share exactly the same prediction: that prices are sticky...," Blinder [1991, 90] asks, "how are we to discriminate among them?... A natural idea is to use each theory to derive other, auxiliary predictions, and then to test these. Unfortunately, often there are no such predictions...." Blinder's point, however, does not apply here. The contrived shortages model has an array of testable implications beyond that of (tautological) price rigidity.

Unexpected Demand and Cost Increases

First, firms will more likely adopt a policy of intentional shortages, refusing to increase prices in the face of rising demand, when certain predictable conditions are met. Intentional shortages will more likely emerge when (a) customer demands or input costs are rising unexpectedly but the seller can predict they will move back toward long-run equilibrium levels, or (b) unexpected demand increases are believed to be permanent but will later be matched by increased long-term production.(12) For example, new products--goods that, if especially successful, will maintain a long-run demand growth sparking increased capacity--will be ones whose firms adopt a policy of transitory shortages rather than episodic price increases.

Unexpected increases in demands or costs are necessary but not sufficient for contrived shortages, however. The costs that customers bear to learn firms' price-quality ratio must also be high, relative to their valuation of that information. The stock of specific information capital that consumers hold is a negative function of the cost of acquiring it and the rate at which it depreciates. Suppose that, with a demand increase for its Brand A, Firm A could increase price but customers could know instantaneously and costlessly when A's price returned to its long-term level (after the fad abates, the season passes, or Firm A expands capacity). In those instances, many current consumers of A who switch to Brand B would likely return to A when its price returned to the prior level, and Firm A could be charging a higher price to those customers who purchase in the interim.(13) The incentive to run shortages in such an environment is therefore slight. Holding constant the initial stock of consumer knowledge and its depreciation rate, higher consumer search costs increase a firm's incentive to allow shortages in the face of demand or cost increases.

Consider a product like gasoline. Consumers obtain information about gas prices almost costlessly, as a by-product of just driving past gas stations, rather than searching out prices. Under the model here, there would be no reason for gas stations to "hold the line" on price as demand or cost changes occurred. And in fact, gasoline prices routinely change with changes in demand (rising during the summer, for example) and cost (increasing during Mideast wars). Similarly, if exogenously some buyers in a market search more than others, sellers have less incentive to maintain price, as demand or cost increases, for the more intensive searchers. Commercial buyers, who enter the market more regularly and purchase larger lots (e.g., fleet buyers of cars), will search more frequently and more often revise their price-quality perceptions. Therefore, a seller is more likely to adjust prices for commercial buyers rather than impose shortages.

Conversely, holding constant the initial knowledge stock and the cost of assembling new information, an increase in that information's depreciation rate decreases the firm's incentive to allow shortages. For example, a high rate of inflation erodes consumers' stocks of useful information about the relative prices and other attributes of Brand A and competing brands. There is less reason for firms to forego the immediate benefits of increased prices, given that the value of buyers' information is diminishing exogenously [van Hoomissen 1988]. In times of high inflation, therefore, firms would contrive shortages less frequently, all other things equal. Similarly, if for a given stock of consumer knowledge the cost to the firm of maintaining its consumers' knowledge increases, the likelihood that the firm will choose to contrive shortages will fall. For example, during the oil disruptions due to the recent Persian Gulf war, a refiner attempting to maintain prices to customers in the face of substantial crude price increases would quickly have been driven into bankruptcy.

Customer Loyalty and Firm Goodwill

To make voluntary underpricing worthwhile, a firm's customer loyalty must be substantial. When customers are fickle, running a shortage means foregoing increased short-run net revenues with no compensating long-run gain. Repeat purchases by customers and advertising by sellers are two phenomena that are positively correlated with customer goodwill.

Indeed, the model has a counter-intuitive implication concerning advertising. If a shortage indicated merely that the firm had chosen the "wrong" price, one would expect the firm to reduce advertising until the shortage was alleviated by finding the "right" price. Why advertise for customers when current quantities demanded cannot be supplied? But if firms voluntarily incur shortages to preserve customer goodwill, one would expect them to continue advertising, or any other policy designed to maintain existing customer loyalty. Rather than waiting for the shortage to abate and then undertaking a massive advertising investment, the firm will invest to maintain at least a part of its consumer loyalty, despite its being unable to capitalize on it completely in the immediate future.(14)

In fact, firms rationally might increase advertising as a way of avoiding loss of goodwill. This is apparently the strategy that Perrier adopted for its mineral water when impurities forced it to withdraw the product from the market for several months. During that time, the company stepped up advertising. Similarly, if the firm expects to satisfy a permanent increase in demand by increasing capacity, it might also increase its advertising in anticipation, even though the firm was unable to satisfy existing demands at the time.

Queue Management

To date, there has been little economic analysis of the composition of queues when firms choose to create shortages.(15) Any departure from market-clearing prices is treated as unfortunate, but who gets the item demanded typically is seen as unimportant to sellers. The model here predicts the opposite: firms contriving shortages will care whose demands are satisfied, and will incur costs to manage the length and types of queues that may develop.

When shortages result from demand increases, firms have an incentive to channel the available quantities to loyal customers. If a firm allows new, transitory customers to get the available supply, established customers must go elsewhere. The firm has foregone the short-run profit available from raising prices with no compensating long-run benefits. Firms are not indifferent, that is, about who gets the product and who is left empty-handed. They will prefer that loyal customers have favored places in line and that relative newcomers to the product go to the rear. If non-pecuniary competition is allowed to determine product allocation, it is not certain that the customers the producer seeks to favor will actually obtain the product.

Without some rationing mechanism, that is, the firm cannot be certain that loyal customers will be favored. Designing and managing a rationing scheme will not be costless, and sometimes the cost will be prohibitive--but not always. The firm will seek to control such costs, since queuing does not benefit the firm and is detrimental to its customers.

In some cases, it may be virtually certain that existing customers will take all the available supply. If so, it will be advantageous to both the firm and would-be new customers to have newcomers excluded altogether. Queuing imposes dead-weight losses not just on those who wait; since queues ordinarily must be managed, they impose costs on firms themselves. So predictably, firms sometimes will simply refuse certain orders. That is, some firms will refuse to organize a queue in the first place.

When shortages develop because costs have risen and quantities produced fallen, the firm has a slightly different task. The firm will not want to produce enough to satisfy every old customer. Efficient management of the resulting queue in the increased-cost case thus will require discrimination among loyal customers, with some being rewarded with better places in line.(16)

A measure of existing customers' losses from excess demand due to cost increases can be derived from individual demand elasticities. Ceteris paribus, consumer losses from quantity restrictions are greater as demand becomes more inelastic.(17) So, when sellers are able to ration available supply among current users with different demand elasticities, they predictably will favor users with lower demand elasticities. Indeed, if the elasticities differ substantially between market segments, some segments may be entirely excluded for a period. Some may be provided with enough output to clear the market at the maintained price.

Figure 1 illustrates the importance of demand elasticity for seller rationing to minimize the burden on loyal buyers. Suppose the market consists of two purchasers, A and B. Both currently consume [Q.sub.1] at price P (i.e., total quantity sold is 2[Q.sub.1]), given their respective demand curves ([D.sub.a] and [D.sub.b]). Assume that costs rise so that the optimal quantity sold must decline by one half. The firm could treat A and B identically, allowing each to purchase [Q.sub.2] = 1/2[Q.sub.1]. The loss to B would be measured by the area SUV, the loss to A by SUT. But the firm would reduce the total burden by selling nothing to B and letting A purchase as much as desired, imposing an additional burden of U[P.sub.1]RV on B but saving the greater area SUT for A.

Prevention of Unwanted Resales

A fourth set of implications concerns resales when increased demand leads to shortages. By opting for shortages coupled with a policy of favoring long-time customers, firms create an incentive for favored customers to purchase "wholesale" for resale to newcomers. If the firm channels the product to A, the higher-valuing user in the long run, A has an incentive to resell to B, a higher-valuing user at the moment. For the same reasons that the firm will try to channel the product to A$in the first place, resale to B may not be in the firm's interest. At least some loyal customers who resell to newcomers then may resume search for the best substitute brand or product.(18)

Predictably, then, voluntary shortages will be encountered more frequently when undesired resales either can be discouraged by the firm at relatively low cost, or are unlikely to occur for other reasons. For example, the transaction costs of reselling may be so high relative to expected transitory trading profits that resale is not a problem for the firm opting for shortages over price changes. That firm will pay no attention to resales.

Resales are a concern because higher-valuing, long-term users then take their patronage elsewhere. So, by definition, firms predictably will not impede resales that can be managed without long-term patronage being lost. In fact, as noted below, there are situations where resales solve problems of queue management without costing the firm any long-run loyalty. In those cases, as the model here would indicate, sellers not only allow but facilitate resales.

IV. EMPIRICAL EVIDENCE

The implications of the model just discussed appear to be verified in all the episodes of private shortages presented in section I.(19) In each case, consumer loyalty, reflected by seller advertising and repeat customer purchases, was important. Sellers did not simply let queues develop haphazardly, but evinced considerable concern about which customers would get the available product. And in every instance, either the firm expended resources to prevent resales, the transaction costs of reselling rendered that problem unimportant, or reselling was managed so as to negate any long-term lost loyalty.(20) Most of the shortages to be discussed were due to demand surges, but the final episode presented was due to sudden cost increases.

Shortages from Demand Surges

As discussed above, demand-induced shortages may arise when the firm knows either (a) the demand increase is a short-term transitory phenomenon ("fad"), such that the long-term price will be unchanged, or (b) the demand increase is permanent, but long-term price will still be unchanged because of capacity increases. In either situation, though, the firm faces problems of minimizing queuing costs and guarding against resales not in the firm's interest.

Short-Term Demand Increases. There are numerous examples of shortages due to unanticipated but transitory demand increases.

Food and Drink. Restaurants in France face a problem of transitory, unpredictable demand from tourists, due in part to vagaries of exchange rates, terrorism and other factors. There is a variable influx of tourists every summer, and few of them return to France (not to mention the same restaurants) often enough for customer loyalty to be relevant. That makes it difficult to establish a stable summer/winter price differential to clear the market consistently. The loyal customers are the local ones. Predictably, then, sudden shortages caused by unusual surges of tourism would be managed to favor locals at the expense of the tourists.

In fact, French restaurants do use a mechanism to avoid disappointing natives during summers of heavy tourist demand without also conveying price information now that proves fallacious in the future. Tourists are simply turned away when they request reservations, while French speakers are added to the list [Ricklefs and Kamm 1985]. The restaurants' behavior is not to accumulate back-orders, or to manage lengthy queues, but completely to preclude some would-be customers from any meaningful chance at being served. That discourages formation of queues in the first place.(21) Under the circumstances, reselling of meals by Frenchmen to tourists is unlikely to be a problem.

Faddishness in demands for food and drink is not infrequent. It occurred some years ago with Jack Daniel's whiskey, when "Frank Sinatra caused all the Jack Daniel's that was ready to be drunk to be drunk."

It happened back in the '50s when Ol' Blue Eyes, at the peak of his popularity, started carrying Jack Daniel's on stage and talking it up. Soon other Hollywood Rat Packers were sipping Tennessee whiskey, and just like that, it became the "in" drink. So in, in fact, that they "ran out of whiskey," says Jack Daniel's historian. [Cawthon 1991, p. M4]

Realizing that the demand surge was just a fad, Jack Daniel's was concerned about ensuring the availability of its whiskey to loyal long-term customers. But the way that the product was retailed made channeling it to those buyers almost impossible; there was no cost-effective way to keep newcomers from taking bottles of Jack Daniel's off the shelf before long-time buyers did. Unwilling simply to raise the price, the distillery undertook a campaign to compensate long-term aficionados of the product for their tribulations. "While more sour mash was aging in Lynchburg, the distillery got loyal drinkers to stay loyal by designating them Tennessee Squires.... To belong, you must be recommended by another member. Squires receive an invitation to an annual coon hunt..." [Cawthon 1991, p. M4].

Airlines. Airlines routinely run two-sorts of queues. Initially, tickets are priced by categories, with only a certain number of seats per category. Shortages thus may develop for the lower-priced seats. Airlines do not run queues at all in these cases; customers are offered seats on other flights. Practically, those getting tickets cannot resell to disappointed demanders, since the two groups cannot locate one another at a reasonable cost.

Queues also develop at the gate, due to the airlines' policy of protecting themselves from no-shows by selling more tickets than there are seats on the plane. In this situation, queues could be managed so as to allow higher-valuing users to fly by assigning places in the queue as customers arrive and then, because all flyers are present, allowing resale among them. But the airlines do not permit this. Allowing long-term customers to sell to short-term higher-valuing customers creates a risk that the long-term flyer will take his business elsewhere, which often is easy in large, high-volume airports. Optimal allocation without loss of customer loyalty is instead achieved by the airlines themselves auctioning off places on the overbooked flight. Those willing to wait are ticketed for the next flight--on the same airline--and given a free stand-by ticket on any flight within a certain period--but only on that airline.(22) So structured, resales are not objectionable to the seller.

Natural Disasters. Perhaps the best known example of the "fairness" explanation of shortages [Kahneman et al. 1986, 729] concerns the seller of snow shovels who does not raise his prices the morning after a blizzard, because his customers supposedly would think it "unethical" to do so. The same "unfairness" constraint was said to limit sellers of items suddenly in great demand in South Florida following Hurricane Andrew [Lohr 1992]. But the "fairness" explanation has nothing to say about the importance of queue management and preventing unwanted resales, the principal implications of the model here.

In fact, the events following Hurricane Andrew are those predicted by our model. Some sellers did increase prices of items suddenly in greater demand, such as plywood; other sellers did not. The difference depended on who was in the market for the long term. "The big companies performed far differently than [sic] the price-gougers selling ice, water and lumber from the back of pickup trucks at wildly inflated prices.... But unlike the carpetbagging vendors, who drove away at sunset, the big companies have a long-term stake in the South Florida market. For them, the good will of local customers...is a valuable asset" [Lohr 1992, C1-2].

One aspect of temporary demand increase following natural disasters like Hurricane Andrew is unique. For the long-term sellers, queue management did not require separating the "faddish" demander from long-term buyers. All customers were local homeowners representing a long-term clientele; the "fad" is temporarily higher demand from existing customers. Since nearly all were affected by Andrew, there would be no important demand-elasticity difference among most customers. Thus, sellers would predictably try to satisfy the most urgent demands of all customers. In Figure 1, let all buyers have the same demand, [D.sub.a]. Assume that prior to the disaster price [P.sub.1] and quantity [Q.sub.2] cleared the market. Temporarily, though, [Q.sub.1] is demanded by each customer at the same price. Since [Q.sub.1] = 2[Q.sub.2], a shortage must result at pre-disaster prices. Maintaining those prices, a seller could give half his customers [Q.sub.1] and turn the other half away. But the consumer surplus generated for favored customers (TUS) in moving from [Q.sub.2] to [Q.sub.1] would be less than that lost by disfavored customers (WTU[P.sub.1]). A rational seller would instead give each of his customers half of what each wanted.

In fact, partial satisfaction of all demands, rather than first-come-first-served full satisfaction of just some buyers' demands, was the policy followed after Hurricane Andrew by sellers relying on queues rather than price increases. The South Florida Home Depot stores initiated a policy of quantity limits on plywood, roofing shingles, roofing felt and polyethylene sheeting. For Home Depot, the quantity limits solved a second problem: they prevented "operators from buying large quantities from Home Depot and reselling the items at inflated prices" [Lohr 1992, p. C2].

Newspapers. Formerly, more newspapers were sold in the streets than by home delivery. Publishers would increase print runs when a major event increased the demand for news, even publishing "extra" editions. Today, newsworthy events do not cause an increase in numbers of papers printed; publishers prefer to leave demand unsatisfied at constant prices. It seems puzzling that newspapers do not increase their prices on days when newsworthy events (increased demand) make it possible to do so.

The answer lies in the investments that newspapers make in learning about their customers. On net, newspapers lose money on circulation itself. They are profitable ventures on the whole, however, because they "sell" readers to advertisers. Advertisers care about how many people read the paper (hence advertising rates are set in part according to circulation figures audited by an independent agency), and in particular about who the readers are. Readers' demographic profiles (incomes, locations and other attributes) are essential information to advertisers. Such data are a major determinant of advertising rates, and newspapers generally provide the information themselves. The demographic data usually refer only to home-delivery customers, in part because it is relatively costly to obtain similar information for newsstand purchasers, but principally because today home delivery makes up the bulk of sales.

The duration of contracts between advertisers and newspapers is typically several months. Increased circulation runs that are not expected to persist beyond a day or two therefore are not of value to newspapers. Circulation itself loses money; one- or two-day circulation increases will not increase long-term advertising rates, which are based on the number and demographics of loyal customers, not temporary demanders [McChesney 1987]. Thus, publishers allow temporary shortages to arise when demand for the paper temporarily increases, even though the shortage is foreseeable before the papers "hit the street," and even though prices could easily be increased.

However, shortages develop only at newsstands; home delivery customers are never short-changed. Newspapers know, by subscribers' revealed preferences, that home purchasers are the highest long-run valuers of the paper. The fact that most circulation is done via home delivery also means that queuing is not a problem. Reselling is not a problem either, since it would occur only after the first recipient--the highest-valuing user--has had an opportunity to read the paper. Reselling therefore would not be costly to loyal customers, nor to the publisher or advertisers, who would already have gotten the demographics they contracted for.

Long-Term Demand Increases. In several instances, demand for a product is higher than present capacity to satisfy demand at current prices. Under the shortages model presented here, manufacturers have a disincentive to increase prices if demand is expected to grow in the long term but capacity adjustments will soon restore price to its current level. This pattern is evident from several recent shortage incidents.

Automobiles. Car manufacturers often run shortages and queues rather than raise prices: the Datsun Z-car, the Mazda RX-7 and the Miata, and the General Motors Saturn are some recent examples [DeGraba 1993, Mitchell 1992]. But the phenomenon is not new. When the demand for Japanese cars rose in the mid-1980s, wholesale prices were maintained at below market-clearing levels, and dealers were also encouraged to keep prices at levels that would cause queues to form. Customer loyalty toward products is stronger than loyalty toward any one dealer of that product. Thus, when some dealers tried to hide price increases by requiring buyers to purchase additional options in order to move quickly through the queues, the manufacturers responded by making nearly all the manufacturer-produced "options" standard equipment. It is interesting also that the United States government had already imposed quantity restrictions on the import of Japanese vehicles; politically, therefore, price increases would have been welcome. How, then, to explain the Japanese decision to keep prices low?

Two complementary factors were apparently at work. An unanticipated demand increase for small cars followed the oil embargo by OPEC. But a long-run supply response could be anticipated from the CAFE-induced increase in the production of small autos by the "Big Three" domestic companies.(23) A purchaser buying a Japanese auto at a temporarily inflated price might well conclude after several years of use that the car had been overpriced. The inevitable initial engineering flaws of domestically produced alternatives would be solved, and domestic output would increase. On the other hand, most new car purchasers are in possession of an older car at the moment an order is placed. Consequently, being unable to obtain a new model for a month or two does not impose serious costs on most buyers. For those who would experience a serious loss, a heart-to-heart talk with the dealer would likely succeed, as dealers managed the queue. In the case of the recent Saturn shortage, for example, buyers were allowed to take demonstrator cars home while they waited for their own Saturn to arrive, and even were given free rental cars when delays persisted [Mitchell 1992].

But obtaining the new car immediately, at a temporarily elevated price (but one not known by buyers to be transitory), creates expectations about the price-quality relationship that will prove erroneous in the longer run. Erroneous price expectations are particularly damaging in an industry like automobiles, where manufacturers invest so much in learning about customer desires and customers invest so much in learning about cars, much of it through "learning-by-doing" use of the vehicle.(24) A firm's advertising stresses the fact that Customer X has never bought anything but a Belchfire; four generations of Belchfire owners talk about their brand-name loyalty. When a new price niche is established, existing model names may occasionally be upgraded, but they are never intentionally down-graded. New names are created for new models (Falcon, Corvair, Pinto, Protege) when a market niche appears at a lower quality level than exists at the time, so loyal customers are not misled into purchasing a model that is inappropriate for their requirements. That policy is consistent with the analysis developed here.

One interesting implication of the present model is vindicated by the car shortage episodes. Manufacturers typically continue advertising their autos even as shortages develop in this country. Saturn ads proliferated during the 1992 shortage, for example [Mitchell 1992]. There are two other testable implications for which there is no evidence at present. First, manufacturers (and their dealers, if dealer incentives were appropriately structured) should be attempting to favor loyal customers at the expense of newcomers to the market. For example, offering a Mazda trade-in should reduce the delay in getting a new Mazda. Second, an important difference between automobiles and some other commodities discussed here is the incentive for resale. The difference between the market-clearing price and the actual price charged by manufacturers could be so large that in many instances resale by loyal customers could be a problem. Therefore, the company predictably would attempt to discourage resales, perhaps through non-transferable warranties; if the value of the warranty exceeds the resale profit, no resales would occur.(25) L. L. Bean. L. L. Bean for a time simply excluded new customers from the queue by refusing to mail catalogues to those who were not already Bean shoppers. This means of managing queues assured that the inconvenience of the shortages generated by the price restrictions would not fall on the preexisting clientele. Eventually, that policy was reversed, and now an enlarged L. L. Bean actively solicits new customers through magazine advertisements.

Beans's initial catalogue policy is consistent with two hypotheses. Bean may have believed the demand shift to be transitory, and did not want to drive off loyal old customers for the benefit of temporary interests. In fact, it was widely believed by many observers that the upsurge in L. L. Bean sales represented merely a "preppy fad." If so, that belief was a mistake, ultimately corrected by capacity expansion, not price increases.

Computer Chips. Until recently, Intel Corporation has been the sole manufacturer of new-generation microprocessor chips for IBM-compatible personal computers, such as the 386 chip.(26) The chips, now including the 486 and the Pentium, have been extremely popular since their introduction. From 1989 to 1990, unit shipments of 386 chips doubled, but even so there were widespread, persistent shortages. By early 1990, the company was announcing that it was unable to meet existing demand and began accumulating back orders. The shortages persisted throughout 1990 and the first half of 1991 [e.g., Brennan 1991].

This episode corroborates several implications of the shortages model developed here. Throughout the 1990 shortages, Intel continually reduced prices of its 386 chips [Fisher 1990]. But Intel also was expanding capacity enormously; in January 1991 it announced plans to spend up to $1 billion on new plant and equipment. In 1992, Intel announced another $1 billion dollar expansion of an existing factory, plus plans "for yet another factory of a similar huge scale" [Ybarra 1992, p. B6].

In the meantime, Intel managed the queue for 386 chips carefully. Large, established purchasers (such as Compaq, IBM and NEC) likely to continue to dominate the computer market were favored, while small newcomers with more uncertain futures were snubbed. (Disfavored purchasers complained of this policy, causing the FTC to investigate Intel [Yoder 1991].) There was no worry about losing customers to other companies through resale, because search for substitute chips was futile: at that time Intel was the sole maker. Moreover, Intel continued advertising during the shortage period, even inaugurating a new campaign to persuade users of the 286 chip to switch to 386's. The campaign, based on the slogan, "Now You Can't Afford to Wait," ran at exactly the time Intel was announcing that you would have to wait--it could not meet current orders [Fisher 1990].

Cost-Induced Shortages

As Olmstead and Rhode discuss, the gasoline famine of 1920 and the coal famine of 1902 arose from supply shifts, but producers had good reason to expect the situations to be temporary. The coal shortage arose from a miners' strike and the gasoline shortage from a railroad strike.

The information stock of particular relevance in this case is specific to the time, explaining why refiners' reactions to the recent Gulf War were so different. During the 1920s, farmers were deciding if and when to replace draft animals with gasoline motors. Some had decided to wait, and some to convert to machines.(27) Gasoline producers, with better predictions of the future of the new industry, wanted to forestall consumers' perception that the cost of gasoline-powered machinery was more variable than it would shortly become if the market grew.

The model here predicts that a firm adopting a policy of temporary shortages will attempt to distinguish instances of high-cost product unavailability from instances that impose lower costs. As discussed, demand elasticities are closely connected to those relative costs; lower elasticities imply greater losses from shortages. Ceteris paribus, demands will be less elastic as there are fewer substitutes. In fact, Olmstead and Rhode [1985, 1045] note that the major oil companies attempted "to maintain supplies at historic levels to 'essential users' and force 'nonessential' consumers to bear the entire shortfall." Consumers were divided into those (mostly rural) who could cope with shortages only at high cost and those (mostly urban) who could cope at lower cost, there being no convenient way for buyers to switch groups. Urban owners of private automobiles could more easily utilize public transportation, walking, car pooling, even telephones to substitute for much local travel. Even those users, however, would have occasional uses with inelastic demands (e.g., emergency hospital trips, out-of-town travel). But urban markets were provided with some gasoline, so each family was able to obtain and allocate a restricted supply to its less elastic uses. The fact that favored buyers were geographically separated from disfavored demanders also made the transaction costs of resale high, relative to the benefits.

V. CONCLUSION

Nothing in the model and implications here concerns shortages resulting from exogenous imposition of government price controls (or fear thereof). Price controls, with their inevitable shortages and queues, are deemed undesirable for three reasons: producers have inadequate incentives to produce; buyers have incentives to dissipate resources over places in the queue; and there is no assurance that resources go to highest-valuing users. But privately contrived shortages need entail none of those welfare costs.

Understanding the first issue, producer incentives, requires distinction between the short and long run: to find a shortage attractive, a rational producer must be endeavoring at least to maintain the same long-run production level (in the case of fluctuating costs or temporary increases in demand) or to increase long-run production (in the case of permanent demand increases). The same is true of the second issue, allocation to highest-valuing users: firms contrive shortages precisely to ensure that their outputs go to those who attach the highest discounted present value to the stream of repeated consumption. And any problem of resource dissipation via queuing ignores the ability of private firms to control or even obviate the costs of queuing.

The model here is to be distinguished from prior shortage models in two ways. Substantively, in this model shortages arise from neither seller error and ignorance, nor from buyers' preference for consumption in crowds. They arise as an intended short-run palliative to protect valuable firm goodwill in the face of exogenous demand or cost shocks whose effects will be mitigated in the long run. Methodologically, the model of privately contrived shortages allows derivation and testing of hypotheses not implied by other models. Several examples discussed here indicate that in many actual cases of private shortages, firms do behave as the model predicts.(28)

With respect to predictive power, we close with an illustration. In commenting on an earlier draft of this article, Gary Becker noted that the model here differs from his in yielding the implication that "regular customers are treated better during periods of upward pressure on price." Gordon Tullock went one step further. At a conference where this paper was presented, he said that a company of his sometimes ran shortages, but that he had always assumed the firm filled orders in the order they came through the door. To test the model in this paper, he then asked several of his managers about this. In a subsequent letter to us he wrote, "I asked whether they simply filled the orders as they came in and in each case they denied it--in fact, looked as if I was a helpless moron to even think they might. They gave priority to favored customers."

1. For example, Alchian [1969], Frech and Lee [1987], Kornai [1980], Lindsay and Feigenbaum [1984], and Weitzman [1991]. See Friedman [1990, 482-93] for a concise summary of most of the issues raised by price controls. An exception to the inefficiency conclusion concerns certain controls placed on monopolists (or monopsonists). By altering the marginal revenue (or marginal value product) curve, properly designed price controls could induce such firms to select more efficient outputs [McCloskey 1985, 360-64].

2. Olmstead and Rhode [1985] report a number of such paradoxes without attempting to explain them.

3. Widespread unintentional mispricing has long been a staple of macroeconomic theorizing. For example, Alchian [1969], Clower [1970], Keynes [(1936) 1957], Leijonhufvud [1968], Means [1935], Okun [1981], and Roberts [1992].

4. The impact of the Blinder team's work on those in attendance at the 1990 American Economic Association meetings is reported by Wessell [1991].

5. E.g., Carlton [1991], De Vany [1976], Hay [1970], Kenny and Klein [1983], Lazear [1986], Pindyck [1982], Stigler [1963], and Zarnowitz [1962].

6. Barro and Romer [1987], Basu [1987], Becker [1991], and Kahneman, Knetsch and Thaler [1986].

7. Telephone companies do not revise the rate schedule when all circuits are busy, even though the times when overloading will occur (e.g., Christmas, Mother's Day) are usually predictable. For discussion of how the phone companies handle the overflows, see "'All Circuits Are Busy' Preparing for Mother's Day," Atlanta Journal-Constitution, 7 May 1993, p. G1.

8. McNicol [1975] also investigated situations in which long-term relationships would be depreciated by short-run market-clearing pricing. But his model depends upon some firms in the market being vertically integrated with buyers, a condition not satisfied in the cases examined here. Our implications differ from McNicols' as well. In his model firms would be interested in continuing to misprice, but eventually cannot because of competitors' decisions. In our model, by contrast, firms are relieved when they can eliminate mispricing voluntarily.

9. The model here takes firms' specific capital as given. Of course, firms within an industry differ in the amounts of capital they create. Similarly, specific capital is greater in some industries than in others.

10. As Hirshleifer and Riley explain [1979, 1394], when a buyer with a given prior probability distribution of beliefs about price receives new price information, "that in general will lead to a revision of probability beliefs. And thus in turn, to a possible revised choice of action." Brennan, Buchanan and Lee [1983, 544-45] explain how this phenomenon will constrain the long-run pricing strategy of a monopolist, whose customers otherwise would have a greater incentive to invest in finding substitutes.

11. For other reasons why firms' prices at any moment are not a function of just momentary costs, but are set with a view toward long-run costs, see Spence [1981] and Dick [1991].

12. Contrast Becker [1991], Basu [1987] and Barro and Romer [1987], who model price rigidity and apparent excess demand as long-term equilibrium phenomena, not arising as temporary expedients.

13. Low search costs also mean that if some Brand B was already a better buy than Brand A at A's old price, most customers would already have known about it.

14. Investments in knowledge are costlier when made more rapidly [Alchian 1959].

15. But see De Vany and Frey [1982] and Barro and Romer [1987]. On the composition of the queues at government enterprises, see Lindsay and Feigenbaum [1984].

16. Thus an elderly buyer, one apt to move, become insolvent, or otherwise cease buying in the near future, may be disfavored in times of shortage. From the seller's point of view, because that buyer's firm-specific intangible capital is becoming obsolete, the firm will expend fewer resources to retain it.

17. Frech and Lee [1987] demonstrate this point in a second-best world in which shortages arise from government price controls. But it is equally applicable to cases in which firms adopt shortages as a first-best solution to conserve consumers' stocks of information.

18. Loyal customers may not search elsewhere, however. For example, A may resell then choose to go without the product during a shortage, rather than search for a substitute. When the shortage disappears, he would then resume his former consumption pattern of the same product. Or the product may be a durable that A can use first and completely exhaust his demand for the product (e.g., a book or a computer program), then resell to B. Neither of these practices will harm the firm. Indeed, the firm is benefited in both instances. If A prefers to resell and do without, the long-term value of the product to him is greater. This is also true if the nature of the product permits A to use it first, then resell to B. Hence, A's long-term reservation price will be higher if on such occasions he is able to resell.

19. The empirical implications of the model here are reconcilable with Blinder's [1991] findings, but are both more precise and more numerous. The executives he surveyed were asked to evaluate a number of distinct reasons for price rigidity. In the context of the present model, two reasons given frequently, reliance on delivery lags and inventories, should be treated as the same phenomena with different signs. (Nearly two-thirds of the executives surveyed by Blinder listed inventories/delivery lags as an important reason for resisting price changes, the highest score earned by any of Blinder's suggested reasons.) When the sign of inventory holdings is negative, one observes delivery lags or, in our nomenclature, privately contrived shortages. Whether any particular firm relies on inventories, delivery lags, or both, depends largely on the attributes of the industry. For instance, a producer of services or highly perishable goods ordinarily cannot hold substantial inventories, and so would have listed that category as totally unimportant in Blinder's survey. Such a firm must rely entirely on negative inventories--"delivery lags" to Blinder, "shortages" in the model here. Similarly, a retailer of rattlesnake serum cannot make use of delivery lags.

The results regarding another of Blinder's categories, "implicit contracts" between buyers and sellers that prices will not vary as demands or costs change, is also informative. A third of the sellers in Blinder's sample deny that such contracts even exist in their industries. In the context of our model, such a result seems reasonable. Where consumer loyalty is low, or the cost of managing queues appropriately is high, firms will not voluntarily contrive shortages; there will be no implicit contracts. But for industries in which consumer loyalty is high and the cost of queue management low, executives will be sensitive to the costs that their price changes impose on buyers. In other words, they will recognize a "duty" toward, or unstated contract with, their loyal buyers. Interestingly, among that subset of Blinder's respondents who perceived an implicit contract between their customers and themselves, the "respondents generally think such contracts are an important source of price stickiness: the mean response within this group is ... stunning" [1991, 95].

Blinder seems to seek the one reason that producers generally hesitate to adjust prices. Our model, instead, implies that some firms will not hesitate, and those who do will hesitate for different reasons. Some firms can hold inventories, which will dampen price movements. Others cannot, but consumer loyalty will make producers reluctant to adjust prices anyway. Contrived shortages should be the exception, not the rule: as this paper demonstrates, several conditions are necessary for producers voluntarily to choose to run shortages.

20. Also, contrary to the models presented in Basu [1987], Barro and Romer [1987], Becker [1991], De Vany and Frey [1982] and Lindsay and Feigenbaum [1984], each instance of shortage discussed below was a temporary response to changing conditions, not a permanent feature of the particular market affected.

21. More profitable alternatives could perhaps be imagined. Predictably, a tourist slipping a tip to the maitre d' would discover that a segregated pricing mechanism was, in fact, already in use.

22. Since the free tickets are stand-by, there is no out-of-pocket cost to the airline, and no cost at all unless the customer would have purchased a ticket for the same flight anyway. (In many instances, however, the free ticket results in the purchase of an additional ticket for the free flyer's companion.) Because the free ticket typically entails no cost for an airline, it would be more valuable to customers at no short-run cost for airlines if the free ticket were valid for any airline. But of course, that would entail the risk of lost loyalty that leads the airline to organize the resale market in the first place.

23. The Corporate Average Fuel Economy (CAFE) standard is a government requirement that the aggregated pool of automobiles produced by a given manufacturer and sold in this country achieve a specified average mileage per gallon of fuel. Because the standard increases over time, domestic small car production would obviously be anticipated to increase as well, as it in fact has.

24. It is noteworthy that automobile manufacturers sell cars to rental firms at reduced prices, and then advertise jointly with them. The motivation is apparently to increase the number of likely car buyers who learn about the manufacturer's product by renting during a business or vacation trip.

25. It is interesting, finally, that the Japanese strategy of contriving shortages rather than raising price in the face of increasing demand is precisely the policy followed by Henry Ford, as recounted in the famous Dodge v. Ford Motor Co. case (204 Mich. 349, 170 N.W. 668 (1919)). The case pitted Ford against the Dodge brothers, minority shareholders in the Ford Motor Company. At then--current production levels, Ford could sell his entire output at $440 per car, but instead charged $360, costing the firm a $48 million flow in annual net revenues. The court questioned what it saw as a failure to maximize the value of the company. But as shown both by Ford's trial testimony and by subsequent events, particularly the construction of Ford's fabulous River Rouge plant, foregoing the revenue-flow was only a short-run expedient in anticipation of tremendous expansion of long-run capacity. So determined was Ford of the correctness of his business strategy that, rather than alter his policy after losing in court, he bought the Dodge brothers' shares at a substantial premium. By 1921, Ford was manufacturing 56 percent of the new cars sold in America, a market share never attained by any automobile company before or since. Ford reached that share despite competition from a determined and experienced new competitor, the Dodge brothers.

26. One industry newsletter refers to Intel's position in the chip market as "one of the few real monopolies left outside the Middle East." Pollack, "Business Technology: Intel's Chip Monopoly Challenged," New York Times, Sept. 5, 1990, p. D1, col. 3. That position has now eroded somewhat.

27. The U.S. Army was in the midst of making similar decisions about its continued use of transport animals versus shifting to mechanized vehicles. See Anderson and Hill [1975, 175-76].

28. To take still another example, while this paper was being reviewed, it was reported after the 1994 earthquake in Southern California that many businesses were running shortages rather than raise prices, and in particular tending to their regular customers. [Holden 1994] A carpet-cleaning firm, for example, "received more than 65 calls from potential new customers immediately after the quake.... But [the owner] turned away practically all of them, because he is booked through mid-February."

REFERENCES

Alchian, Armen A. "Costs and Outputs," in The Allocation of Economic Resources, by M. Abramovitz, et al. Palo Alto: Stanford University Press, 1959, 23-40.

-----. "Information Costs, Pricing and Resource Unemployment." Western Economic Journal, June 1969, 109-28.

Anderson, Terry L., and P. J. Hill. "The Evolution of Property Rights: A Study of the American West." Journal of Law and Economics, April 1975, 163-79.

Atlanta Journal-Constitution. "'All Circuits Are Busy' Preparing for Mother's Day." 7 May 1993, Sec. G, p. 1.

Ayanian, Robert. "Advertising and Rate of Return." Journal of Law and Economics, October 1975, 479-506.

Barro, Robert J., and Paul M. Romer. "Ski-Lift Pricing, with Applications to Labor and Other Markets." American Economic Review, December 1987, 875-90.

Barzel, Yoram. "A Theory of Rationing by Waiting." Journal of Law and Economics, April 1974, 73-95.

Basu, Kaushik. "Monopoly, Quality Uncertainty and 'Status' Goods." International Journal of Industrial Organization, December 1987, 435-46.

Becker, Gary S. "A Note on Restaurant Pricing and Other Examples of Social Influences on Price." Journal of Political Economy, October 1991, 1109-16.

Blinder, Alan S. "Why are Prices Sticky? Preliminary Results from an Interview Study." American Economic Review: Papers and Proceedings, May 1991, 89-96.

Brennan, Geoffrey, James Buchanan, and Dwight Lee. "On Monopoly Price." Kyklos, 36(4), 1983, 531-47.

Brennan, Laura. "Persian Gulf War Adds Wrinkle to 386 Chip Shortage for Intel." PC Week, 28 January 1991, p. 129.

Carlton, Dennis W. "The Theory of Allocation and Its Implications for Marketing and Industrial Structure: Why Rationing Is Efficient." Journal of Law and Economics, October 1991, 231-62.

Cawthon, Raad. "How Jack Went Hollywood." Atlanta Journal-Constitution, 7 April 1991, Sec. M, p. 4.

Cheung, Steven N. S. "Why Are Better Seats 'Underpriced'?" Economic Inquiry, October 1977, 513-22.

Clower, Robert W. "The Keynesian Counter-Revolution: A Theoretical Reappraisal," in Monetary Theory: Selected Readings, edited by R. Clower. Baltimore: Penguin Books, 1970, 270-97.

DeGraba, Patrick. "The Economics of Buying Frenzies and Persistent Excess Demand." Photocopy, Cornell University, 1993.

De Vany, Arthur S. "Uncertainty, Waiting Time, and Capacity Utilization: A Stochastic Theory of Product Quality." Journal of Political Economy, June 1976, 523-41.

De Vany, Arthur S., and Gail Frey. "Backlogs and the Value of Excess Capacity in the Steel Industry." American Economic Review, June 1982, 441-51.

Dick, Andrew R. "Learning by Doing and Dumping in the Semiconductor Industry." Journal of Law and Economics, April 1991, 133-59.

Fisher, Lawrence M. "Profits Rise by 48% at Intel on Demand for Its PC Chips." New York Times, 13 April 1990, Sec. D, p. 1.

Frech, H. E., III, and William C. Lee. "The Welfare Cost of Rationing-by-Queuing across Markets: Theory and Estimates from the U.S. Gasoline Crises." Quarterly Journal of Economics, February 1987, 97-108.

Friedman, David D. Price Theory: An Intermediate Text, 2nd. ed. Cincinnati: South-Western Publishing Company, 1990.

Hay, George A. "Production, Price, and Inventory Theory." American Economic Review, September 1970, 531-45.

Hirshleifer, Jack, and John G. Riley. "The Analytics of Uncertainty and Information--An Expository Survey." Journal of Economic Literature, December 1979, 1375-421.

Holden, Benjamin A. "Quake Price-Gouging Is Tempting but Shortsighted." Wall Street Journal, 27 January 1994, Sec. B, p. 2.

Kahneman, Daniel, Jack L. Knetsch, and Richard Thaler. "Fairness as a Constraint on Profit Seeking: Entitlements in the Market." American Economic Review, September 1986, 728-41.

Kenney, Roy W., and Benjamin Klein. "The Economics of Block Booking." Journal of Law and Economics, October 1983, 497-540.

Keynes, John M. 1936. Reprint. The General Theory of Employment, Interest, and Money. New York: Harcourt Brace, 1957.

Klein, Benjamin, Robert G. Crawford, and Armen A. Alchian. "Vertical Integration, Appropriable Rents, and the Competitive Contracting Process." Journal of Law and Economics, October 1978, 297-326.

Klein, Benjamin, and Keith B. Leffler. "The Role of Market Forces in Assuring Contractual Performance." Journal of Political Economy, August 1981, 615-41.

Kornai, Janos. Economics of Shortage (2 vols.). Amsterdam: North Holland,