(Originally Issued February 20, 2001)

    The above initiatives share an implicit assumption that a strategy of using limit orders to participate in what are believed to be auctions will benefit the consumer. This Auction Countdown explores why all three strategies are bound to disappoint their users, as Priceline and Optimark already have. The disappointment in the case of SuperMontage and similar electronic book proposals is likely to be especially galling. Not only are they promoted, as were Priceline and Optimark, as the latest in electronic efficiency, but they also carry the implicit imprimatur of official regulatory reform. The truth is, however, that none of these are really auctions. Rather, they are price discrimination machines, and price discrimination, as we will see, is incompatible with what people expect from auctions. To make a bad situation far worse, the collapse of the trading tick – the primary intended effect of decimalization – will greatly magnify the potential for price discrimination.

    The basic problem here is that, even though you name your own price, chances are you won’t be happy with it, because others can get better prices at the same time. These systems share a built-in version of what game theorists call the "winner’s curse." It is a fatal flaw, because it makes you feel foolish twice: once for naming the wrong price, and again for believing the hype about a fair auction. Having been made a fool of, you won’t come back or, if you do, won’t bid aggressively. In academic parlance, you will "shave your bids." This is a problem for these systems because, once the word spreads, the supply of fools will dry up, and there may be no one to trade with or, in the case of Priceline, there will be insufficient bids for services at prices that their vendors will accept. In theory, price discrimination can enhance profits by enabling the capture of "consumer surplus," and enhance "efficiency" by enabling more consumers to trade at prices that satisfy them. But, these benefits will remain theoretical if consumers feel scammed. And they will, as long as there remains a large gap between expectation and reality, such as inevitably follows from touting these systems as auctions. No one has yet figured out how to make sitting ducks remain seated after the shooting starts.

    In the case of SuperMontage and other visible limit order book systems, one effect of lessened aggressiveness is less transparency, as bid (and offer) shaving moves orders out of the range of tradability, cuts down on their size, or moves them out of the book altogether. The loss of transparency is a particularly ironic result in that the most common reason put forward by reformers for both such books and decimals is the need to improve transparency. The theoretical errors behind such seemingly trivial technicalities are not merely academic, nor are these examples unique. Consumers, regulators, business method consultants and entrepreneurs all seem to be falling en masse for the most improbable snake oil. Both the embrace by experts of these auctions that are not really auctions, and the inarticulate consumer revulsion they cause, are signs of a collective analytical breakdown, which I believe may be the result of excessive Government intervention in the formation of commercial structures. To get a handle on where these errors are coming from, let’s start with some background on auctions.

    Behind all their high-tech terms, the "auctions" in question are hiding a crude and actually somewhat old-fashioned merchandising mechanism, a throwback to watch-your-wallet dealing bazaars, or to shopping in 19th century American stores, where bargaining and price discrimination were the norm. But just as the price tag unified prices for most shoppers in the 20th century, true auctions provide a more civilized alternative for many consumers to such primitive multi-price structures. And, more to the point here, they provide a result that is consistent with what is expected from any mechanism that calls itself an auction; consistent, that is, with what is implied, but cannot be delivered, by Priceline, Optimark and SuperMontage. A true auction, as discussed below, is one in which you not only name your own price, but – if competition produces a better one – you get the better price paid by all. Such competition is an expected feature of an auction, as is a single-price result. Because consumers reasonably assume that the act of bidding brings all these features into play, they understandably feel scammed when – without warning – the competition and single-price features are left out.

    Before looking at specific situations, let’s first acknowledge that there is today no common or popularly accepted definition of what an auction is. The standard dictionary definition¹ is at best only a distant and loosely suggestive relative of the systems that today claim auction status. Consequently, the use of "auction" to describe them invites so much play with the concept that, in the end, the term elicits only vague notions of a mechanism that somehow involves other vague but positive-sounding concepts. Among these are: public openness, officially licensed and regulated status, vigorous and unhindered competition, efficiency and fairness. But, while use of the term implies the presence of at least some of these benefits, the lack of definitional certainty means that today’s "auctions" may leave some or all of them out. So egregious can these omissions be, that often even the core auction concept of the sale going "to the highest bidder" is more honored in the breach than the observance. Nonetheless, it is clear that auctions have a positive image, and that any confusion over what is meant by the term usually works in their favor, as it allows both speakers and listeners to assume the best if in doubt. While each individual may have a specific meaning in mind when someone utters the word "auction," it is almost impossible to pin down that meaning, and – for that reason – almost impossible to discredit a sham auction.

    Let’s also acknowledge that some newer auction structures are mind-numbingly complex. This, too, works in favor of their image in the public eye. If the people who hear an auction pitch can’t figure out precisely how it works, but assume that the experts who designed it have, they can easily maintain their positive expectations of it. All of us are used to giving the benefit of the doubt to experts who design the many high-tech miracles we use daily. So we should not be surprised when the complexity that prevents personal comprehension of an auction process may actually move us to be very positive, even messianic, about it.

    Now let’s move away from auctions for a moment to see why certain features of some common merchandising methods are attractive or not. We will then apply what we learn to make some educated guesses at what people really mean when they use the term "auction."

    Economists sometimes refer to giving-the-benefit-of-the-doubt-to-experts as "rational ignorance." In spite of its negative sound, rational ignorance is a very useful tool. It allows us to maneuver successfully through life without being expert at everything, so that we may focus on what we do best. But for rational ignorance to work effectively, we must be able to trust the experts not to take advantage of us. Every honestly represented commercial structure meets this test, whether it is safe enough to use in rational ignorance of its details, or whether it clearly requires us to be wary. Examples of the latter are buying a car from a traditional car dealer or an antique from an antique dealer, both of which are understood to require consumers to be knowledgeable and willing to negotiate. While such requirements make many of us uncomfortable, we are generally aware of the need to be wary, and so will have our rational ignorance mechanism turned off.

    In contrast, one example of a structure in which we can let our rational ignorance operate safely is buying a car under the new one-price selling method (i.e., "haggle free" dealing, which was introduced in this country in the mid-‘Nineties by Saturn’s "no-dicker sticker"). As long as we know that bargaining – by us or anyone else – will not result in a changed deal, we are relieved of the need to engage in it. We can then focus on comparing the haggle free car to those available from other companies, whether or not offered through haggle free selling methods. We know that it is safe to allow rational ignorance to operate because, at least in the haggle free showroom, the salesman is not our adversary. The key feature that makes this situation safe is that the same cars are sold to everyone at the same price. Of course another, and better-known, example of this concept is buying merchandise marked with non-negotiable price tags.

    These methods of one-price dealing do not visibly display the competition that makes them fair in the showroom or store, itself. However, the very act of giving everyone the same price correctly implies the existence of effective competition between the products’ provider and the providers of alternative products the consumer could buy. In effect, the vendor’s reputation as an honest competitor is implied – and is on the line – when setting the price that all consumers can scrutinize and evaluate en masse. And to restate for emphasis what is probably its most attractive feature from the consumer’s perspective, one-price selling relieves the buyer of much or all of the need to engage in purchase strategies, such as soliciting competing offers, bargaining for discounts, or asking other shoppers if they got discounts. Relieved of the need to bargain, the consumer can safely allow rational ignorance to operate.

    Now let’s look at some key differences between how one-price and multiple-price (negotiated) selling methods affect consumers, middlemen and producers, respectively. Understanding these differences can help us gain perspective on what might be called the political dimension of how industries choose one method over another. First, consumers: many, perhaps most of us, consider bargaining an unwanted and often intimidating chore, and are generally willing, therefore, to pay a higher price for an item if it is sold at a single price that everyone pays. Middlemen, in contrast, are the customary beneficiaries of haggling, because they become expert at distinguishing the naïve and/or bargaining-challenged consumers from the ones that need discounts to open their wallets. Consequently, middlemen are generally opposed to one-price selling methods, because their intermediation role diminishes or disappears under such methods.

    Finally, producers, although often dependent on middlemen for sales, may prefer one-price selling methods for two major reasons: 1) the middleman’s slice of sales revenues is cut out, and 2) sales revenues themselves are higher, because the customer will pay a higher price if he doesn’t have to haggle. Although producers might, in theory, gain some "consumer surplus" if they could sell to different people at different prices, such discrimination opportunities generally go to middlemen who become expert at mining them. Producers sell to everyone, rather than to only the few reached by any one dealer in their products. As a result, they usually cannot afford the reputational problems they encounter when trying to discriminate. Thus, because they benefit from higher single-price selling revenues, and because any potential discrimination benefits are usually available only to middlemen anyway, producers often find common cause with consumers in support of switching to single-price methods when circumstances or new technologies, such as the Internet, will allow it.

    To effect such changes, producers must first wean themselves off of their need for dealers as distributors. Second, and always far more difficult, official Government protection of the dealer’s role must be removed. Such protection is pervasive in many industries, including car sales and stock trading.² Although invariably characterized by middlemen and regulators as consumer protection, Government oversight of sales processes invariably turns into protection from competition for middlemen. Such protection is maintained by a combination of the raw political power of dealer associations, and the inherent inertia of long-standing regulatory processes. The argument in support of such arrangements generally boils down to the following sophistry: dealers sometimes rip off customers. Therefore, strong oversight of the dealing process is necessary. To maintain strong oversight, all other sales processes are declared illegal, especially those that don’t use dealers.

    Such "logic" has so far blocked the wide adoption of Internet car sales at single prices, since only producers have the economic incentive to offer haggle free buying. It has also blocked the adoption by Nasdaq of a single-price open. Politically active dealers in support of their positions characterize both industries. Car dealers often demand and get political support for their position that only dealers can be relied upon to honestly sell and service a car to the consumer’s satisfaction. This is clearly absurd. While many dealers have no doubt made progress in reversing their industry’s poor reputation among consumers, it is ridiculous to say that manufacturers could not sell an honest product without the help of dealers. The case of the Nasdaq open is particularly demonstrative of the regulatory inertia problem. Not only has SEC Chairman Arthur Levitt repeatedly recommended a single-price open over the past two years, but so have several prominent and progressive dealers. Nevertheless, Nasdaq itself has so far ignored these recommendations, after conducting a "study" which purportedly demonstrated that a single-price open was both difficult to implement and unnecessary. Instead, they claimed, SuperMontage will address all perceived problems with their market, including the raucous opening, in which a handful of "wholesale" dealers have all the information about the all-important overnight orders to be executed at the open.

    Coming back to our rational ignorance concept, the key difference between a commercial structure in which we can safely allow it to operate and one where we can’t is that, in the former, there is little or no price discrimination while, in the latter, multiple prices are the norm. In spite of the advantages mentioned above for one-price selling, neither structure is necessarily good or bad; there are plenty of examples of each, and each has many consumers that prefer it to the other. Surveys show that nearly two thirds of Americans dislike the dominant dealer-negotiated method for buying cars, which implies that there are also many buyers – more than one third of them – who are quite happy with haggling. Apparently, they consider themselves better than average bargainers who will get better than average deals. Not only is there a constituency for each method, but there is abundant evidence that consumers can safely make the choice on their own, even in the new and often confusing world of e-commerce.

    Consider the violent reaction to Amazon.com’s "dynamic pricing" experiment with its DVD customers. The flap began when some of them noticed that, by turning cookies off, or coming in through another computer or a new account, they got better discounts. They also discovered that they could get any difference between the best price and what they had paid refunded to them – if they would only complain to Amazon. In response to the furor, the company reversed course by announcing a policy of automatically refunding any dynamic pricing difference in the future, even if the customer doesn’t complain. No wonder. Some complained strenuously not only about the prices, but also about the practice of price discrimination itself, and the fact that they were not told by the company about it.³

The episode highlights the importance of knowing your niche and, perhaps, the utility to that end of listening to your customers before your professional "business method" consultants. These DVD fans were way ahead of those economists touting the potential to capture consumer surplus through hidden price discrimination on the anonymous Internet. And, although Amazon stumbled in pushing this particular envelope, the company demonstrated an ability to make quick changes in response to evident consumer preferences. Among the things they learned: 1) Their customers trust them well enough to pay a premium price for the service they expect. 2) They do not want to see others getting better prices by strategizing their purchases (turning cookies off, clearing caches, checking out the prices others are getting, asking Amazon for a better discount, etc.). 3) They don’t want to have to strategize their own purchases. 4) The trust of their customers depends on Amazon not surreptitiously engaging in price discrimination. While Amazon never said it wouldn’t discriminate, the clear sense of its DVD customers was that there was an implicit understanding not to.

    As one of Amazon’s DVD customers quoted in note three implies, a major feature of the landscape upon which price discrimination does or does not occur is antitrust law. Although Amazon’s customers clearly needed no Government help making their choice, the Robinson-Patman Act⁴ is often invoked (if seldom by name) to contest perceived or real discrimination. This is not at all irrational: the very existence of any Government interest in price discrimination implies that you may have a cause of action if you have been discriminated against. And the more imprecise and obscure the law, and this one is both, the more likely you are to imagine you might have a case. In fact, you generally don’t, because most antitrust actions regarding price discrimination are at the wholesale level, and applied in service of such esoteric goals that often the only thing consumers notice is that regulators are preventing them from getting discounts.⁵ Naturally, they are confused – and they aren’t the only ones. Experts often disagree strenuously over whether price discrimination is an appropriate target of antitrust, and, if so, to what end or ends the various overlapping and conflicting statutes should be applied.⁶

    Unlike the consumer protection regulations discussed earlier, which only accidentally affect price discrimination, antitrust deliberately involves itself in the issue, with even more disastrous results. This is not a case of intervention causing more harm than good. Here, the harm is immense, and there are no benefits whatsoever (unless one counts the greater employment of bureaucrats as a benefit). The very fact of Government involvement implies many things, among them that intervention is needed, that consumers can’t effectively make the right decisions on their own, that there is always something wrong, even sleazy, about price discrimination, and that Government is on the case and can stop and punish discriminators. None of these implications is correct. But the detailed involvement is distorting the formation of commercial structures on a grand scale across the economy, because whether or not to accept price discrimination is perhaps the most fundamental choice that consumers need to be free to make if those structures are to evolve to their liking.

    The paternalistic approach not only denies consumers those crucial choices, but probably over time destroys their ability to make them. This, of course, makes them even more dependent on Government to choose for them, however misguided and confused the bureaucrats may be. To the average consumer, his Government’s involvement in the price discrimination issue looks something like this: first you think antitrust’s price discrimination law is there to defend you from discrimination, but then you learn it’s not. Then you find that the discriminators they do go after were actually giving you discounts. And finally you find that the main effect of antitrust’s most important effort, to bust monopolies under the Sherman Act, is anticonsumer with respect to price discrimination in two important ways: while one hand prevents discrimination that benefits you, the other causes discrimination that harms you. Trustbusters actually block or retard pricing standards you may want, such as Saturn’s no-dicker sticker. And they promote discrimination that endlessly confuses and frustrates you by fragmenting every industry into so many competitors and, thereby, so many permutations and combinations of products and vendors, that you can’t possibly compare prices effectively, much less collectively decide on the right price for a standard product.

    Having retarded the natural formation of standard single-price processes, regulators then set about designing and providing their own versions of the standards they have blocked. That’s when things really get ugly.

    An example of this pattern is found in today’s stock market. Having busted the exchanges for fixed commissions, fixed increments and fixed spreads, regulators have now set about making sure all investors always get "best execution."⁷ The result is that investors have been forcibly moved from a situation where they paid relatively known, common and explicit costs for execution (a beneficial side effect of fixing their key components), to one in which the target is moving too fast to trust. They were thereby moved from an environment in which they were able to allow rational ignorance to operate safely, to one in which they must monitor the markets second-by-second to see if someone else gets a better deal. As University of Memphis Professor Robert Wood put it to the Wall Street Journal in an article about his preliminary study of the effects of decimalization:

Decimalization is only the latest initiative in a three-decade long effort to create a "National Market System" (NMS), all of them in service of the goal of providing everyone with best execution. Every one of these initiatives has had the effect of increasing the number of possible trade prices and, thus, the risk and burden of price discrimination. The result is that your "best execution" today often looks bad within seconds compared to someone else’s "best execution." How could the single term "best execution" apply to so many different prices at or nearly at the same time? Good question. But don’t imagine that regulators are asking – much less answering – it. They have adopted a policy of leaving the concept of best execution undefined and vague.

    Now let’s have another look at what is meant by the term "auction." Specifically, what are those academics and regulators talking about when they call the evolving NMS an "electronic order-driven auction"? Is each fleeting price the result of a separate auction? Or is the whole grab-bag of prices an auction? Obviously, you can’t have a sale to the highest bidder when there are multiple sales to multiple bidders at different prices only "sub-seconds" apart. So where did the idea come from that these continuous "flicker markets," as they are sometimes called, are auctions?

    The most common use of the term "auction" in modern times is no doubt the ubiquitous reference to The New York Stock Exchange as a "continuous agency auction." Sometimes the words "continuous" or "agency" are left out, but one never hears the NYSE’s trading process spoken of as anything other than an auction. But is it? The usual purpose of the reference is to distinguish the Big Board’s structure from Nasdaq’s "dealer market." And it is certainly true that the NYSE’s specialists have traditionally enabled many more customer-to-customer transactions than Nasdaq’s dealers did. But why does the fact that the Exchange’s process is not a dealer market automatically mean that it is an auction? Why do we assume that an agency matching service, however fair and cost-effective it may be, is an auction?

    As for the dealer market, Nasdaq never really had a shot at calling itself an auction, if only because of NYSE’s success at defining non-Nasdaqness as, per se, an auction. But if one ignores that argument, it is not at all clear that Nasdaq’s continuous trading is any less auction-like than the Big Board’s. In the first place, the "customer" on the other side of your Big Board trade is often as professional as the specialist or a Nasdaq dealer, so the presumed lack of intermediation is often bogus. Thus, even by its own chief criterion, the Big Board is really just another dealer market with a spread to pay. But that does not mean that either of them – the NYSE or Nasdaq – are bad markets. In terms of their abilities to charge customers a reasonable and standard price for liquidity such that they can deal on those markets safely in rational ignorance of their operating details, both markets have historically been quite effective – or were until the reforms kicked in.

    Prior to Christie-Schultz, the 1994 study that ignited an antitrust investigation, the price of a Nasdaq stock would bounce regularly between the bid and offer of an easily visible spread of a quarter point, or twenty-five cents. While some thought that egregiously wide, and instigated antitrust proceedings to block the "tacit collusion" that produced it, the fact was that Nasdaq customers were more likely to get the same price in the old days than they are today. Not that buyers got the same price as sellers; that seldom happens on any continuous market. But buyers got the same price as other buyers trading at around the same time, and so did sellers relative to each other. Like no-dicker stickers or price tags, the comparatively stable and visible spreads allowed customers to trade comfortably without strategizing, to let the process handle their trades, because they couldn’t do any better with any strategizing techniques that were reasonably available to them.

    Now, post the "order handling" reforms of 1997, and halfway into the radical increase in the number of possible prices known as decimalization, customers are getting very wary. Strictly in terms of whether they can allow rational ignorance to operate safely, they were better off before. Now the tape spews out so many different prices within seconds of their executions that they must worry that their counterparty made better strategic use of the available information, such as CNBC, Level II, etc. And, now that the Big Board has fully decimalized, another worry – that their limit orders are much more likely to be front-run by specialists or other insiders – has emerged. Thus, even though spreads and apparent average costs have shrunk, the burden is now on the investor, just as it was for a while on the Amazon DVD customer, to be on his toes, to strategize. While the average per share trading cost in the old days may have been a quarter, give or take a nickel, the average cost today might be a dime, give or take a dollar. And the hypothetical fifteen-cent improvement for the average investor could be cold comfort if the average investor disappears. That could happen, if professional traders wake up to the opportunity, as they are certain to do. Once that fully happens, average investors will be hard-pressed to prevent the most expensive outcomes possible (i.e., a dollar in our hypothetical), unless they take the time to become professionals themselves. So much for rational ignorance and being able to trust yourself to a safe process.

    Whatever the effect of the reforms on investors, this is not what they think of when they hear the term "auction." The greater number of prices and prints around any potential execution time would probably lead most investors to say that both the Big Board and Nasdaq felt more auction-like in the old days than they do today. And when Nasdaq joins the Big Board in full decimalization in April, the potential for price discrimination will explode again there, too, and with it the implied burden to micromanage your execution strategies, to protect yourself against professional front-running, etc. Furthermore, it must be remembered that reforms are forcing all these markets to become more automated and anonymous, which separates the investor from any previous comfort he may have gotten from the branded execution quality of his broker-dealer. With all executions commoditized by best execution, customers are facing the fact that, "democratized" as they may be, every trade is now suspect. And it is hardly reassuring that one of the fastest growing brokers in the fastest growing category (semi-professionals, once called day-traders) invites its customers to avail themselves of its lightening-quick execution capabilities, as if they were Bruce Lee at a keyboard. What are the rest of us supposed to do?

    The dissatisfied customers of Priceline, after learning that the hidden potential for price discrimination was much greater than they expected, even in a "reverse auction," bid less aggressively, bailed or sued. Optimark never really got off the ground, arguably because word spread as it was trying to get going that the thing had a fearsome winner’s curse built into its multi-price "auction" algorithm. Either system may make a comeback, since the markets within which they operate are so fraught with either natural or regulator-induced price discrimination that, if honestly presented, the systems could fit right in. Honest presentation probably means dropping the word "auction," or at least explaining to their customers how far from what they normally expect of auctions their results could be.

    A larger question than the prospects for those private systems is how consumers of air travel and stock trading services will feel about the pricing of products generally in those industries going forward. Although both have been subject to "deregulation" for some time, one effect of which has been to dismantle or prevent efforts to standardize pricing, I suspect that air travel has a head start. If it does, the example is ominous. That industry is in a constant state of crisis, characterized by poor profits, deteriorating service and the "rage" of its customers. Although it is difficult to quantify the degree to which such rage is exacerbated by discriminatory pricing, it is hard to imagine that the sense of unfairness that inevitably accompanies discrimination helps. In any case, the rapid ratcheting up of price discrimination in the stock market as a result of decimals and such anonymous electronic systems as SuperMontage is potentially explosive. Like the Priceline convoy, each of these reforms has been sold as the best thing since sliced bread; in the case of decimals, the benefits seemed so obvious that even Congress got on the bandwagon. The coming consumer letdown is likely to have far reaching and long lasting effects on their willingness to support capital formation through equity investing.