12

Internet Exceptionalism Runs Rampant

I have this vague feeling that, just as every generation thinks it invented sex, every generation thinks it created a new economy.

—Andy Grove, CEO, Intel¹

WebVan impatiently launched in June 1999 after incorporating in 1997, and eventually lost more than $800 million.² It proposed to revolutionize the grocery business by delivering groceries to households who ordered their groceries online. It recruited a well-known executive, George Sheehan, and gained financial backing from several big-name venture capitalists. It laid out an extensive plan to grow in major cities across the United States. It settled on a customer-friendly plan that would deliver groceries within thirty minutes of an order. It aspired to operate out of warehouses in less expensive locations, cutting out expensive retail space, so it could lower its costs enough to make it profitable to charge very low prices for groceries.

There were a few problems with this proposal, but the biggest was most apparent to anyone who had studied the business years earlier, when online grocery ordering and delivering had been tried on a small scale and had not yielded dramatic success. The logistics of delivering perishable products imposed large costs on operations. The prior experiences had not been flawed in any obvious way, and the previous generation of entrepreneurs had executed reasonably well. Try as they might, WebVan’s business was going to be more similar to prior attempts than different from them, despite being shinier, newer, and better publicized. WebVan could improve on the past in only incremental ways at most, by delivering the service at high scale in new facilities.

The economics could be stated in simple terms. WebVan incurred a large sunk cost to entering and incurred high fixed costs during operations. To justify the costs of those facilities and operations, WebVan had to generate large revenue over operating expense. Since the delivered groceries were generally priced at competitive levels and did not generate high margins, and delivery costs were high and did not command high prices, large revenue could be achieved only one way, through a high volume of orders. WebVan’s viability, therefore, came down to a simple economic question in every locale: how many households in a geographic area wanted groceries delivered, and could it become very large?

There were good reasons to be skeptical. The experience of one of the earlier entrants, Peapod, was well known and illustrated the constraints. Peapod started a small-scale version of the grocery delivery business in the Chicago area by partnering with existing grocery stores.³ They had operated this entrepreneurial business since 1989, improving it constantly, and had showed it was viable. Peapod’s experience also illustrated the challenges. No amount of operational cleverness could reduce the delivery costs—gasoline, vehicle maintenance, a driver’s time. It was expensive to send a van from a warehouse to a customer. No amount of clever marketing—online, in magazines, flyers on door posts, and even from the grocery store partner—could convince most users to pay much money for a delivery service, or use it at all.

The problem did not appear to be the software. It went to something for which no technical solution existed, to something innate in human behavior. Stated simply, users changed their shopping habits with reluctance.

That was so of even the savviest online users in the early 1990s. It was quite sensible, therefore, to expect later users to be more reluctant. After all, these were the most inexperienced online users. No amount of website genius could induce a new online user to break with old habits and do their shopping for groceries on a web page. No clever marketing tool could convince many shoppers to trust the selection of bananas to someone else. Many shoppers wanted to examine the day’s daily specials in person. Many wanted a tactile grocery experience, merely as a way to stretch their legs. Many parents with young children wanted an excuse to get out of a claustrophobic house to roam the aisles. All those simple factors prevented many households from making use of the online service.

None of WebVan’s senior management had extensive experience in the grocery business, or in the online grocery business. It is not surprising, therefore, that WebVan’s management convinced itself that it had something that had eluded previous pioneers. They proposed to build brand new warehouses to gain efficiencies from scale and reduce costs as low as they could go. They believed that the newer features of their well-funded web-based grocery service would generate high demand.

The financial vulnerability of the proposal was plain to see: it only could be viable if it achieved a scale of use that no prior pioneer in online groceries had ever come close to achieving. Expensive warehouses and fancy operations would be too expensive if not employed at full capacity. The costs of delivery trucks and drivers would be too high unless demand grew to a point that spread those costs over many customers. The entire cost structure for the business would not be—could not be—low enough unless the business generated a large set of customers to use the capacity at or near its maximum.

In normal times that type of vulnerability would have led investors to call for a cautious expansion plan, reining in the aspirations of the management. Investors might have asked for less cavalier uses of their money, requiring WebVan to, say, first experiment with different mixes of marketing and operational novelties in a friendly location, such as San Francisco, which had a technically sophisticated population spread around a dense urban location. Yet following the prevailing view, WebVan’s management eschewed caution. It avoided cautious exploratory practices. WebVan’s management opened in several cities without first testing the concept in one city for an extended period of time.⁴

What happened? Soon after starting it became obvious that WebVan did not differ that much from any of the pioneers. The online grocery business could generate some interest, but not nearly enough to gain the advantages of scale. WebVan could not make any of its financial goals. Not enough households wanted to use the service to have revenue cover operations expenses—namely, to make that business viable.

Looking back, by the norms of standard business practices, WebVan’s impatient experiment cost its investors dearly.⁵ A more conventional expansion plan—patient, systematic, and incremental—would have led to limited initial trials in a few cities, demonstrated the issues, and limited the downside losses from experimentation, or stretched out the losses longer to allow more time for learning.⁶

WebVan was far from the only example of managers who invested with impatience. This chapter will describe a few others. In retrospect these examples beg the question: What led otherwise sensible managers to act so impatiently and so imprudently? What persuaded otherwise rational investors to put their money behind ventures that behaved this way?

Prior chapters have suggested that some part of the impatience was inevitable. The commercial Internet could not be built without simultaneous investments from many distinct participants—households, businesses, carriers, software developers, ISPs, retailers of all stripes, and more. That juggernaut of investment began to accelerate in the late 1990s, as all participants perceived how pervasive the commercial Internet had become, and all forecast it would become more pervasive in more households and businesses.

Prior chapters have also hinted at what could go wrong with so much uncoordinated investment. Once started, that juggernaut could not instantly or gradually slow. Nobody coordinated it. It was left to each participant to be informed about the activities of the others, and that was impossible to do, so each participant became only partially informed about others. As it continued, it became inevitable that the rates of activities in one group of suppliers would not match one another. One commercial supplier continued to invest at a rapid rate while another slowed down, and one set of users continued to adopt at a rapid pace while another did not. After the fact it appeared as if more capacity was built for more adopters and users of the Internet than actually showed up. In layman’s terms, the building boom overshot its target.

This chapter delves deeply into an additional cause: the near inevitability of overshooting when egged on by poorly conceived strategies. Overbuilding became a certainty because a flawed prevailing view emerged to justify an aggressive approach to building new firms, one that accelerated the amount of investment and reduced the degree of caution. The prevailing view supported a range of investments in the commercial Internet and did so with gusto and optimism. After the fact the buildup became known as the dot-com bubble, its ending as the dot-com bust, and the flawed prevailing view as Internet exceptionalism. This chapter provides details about how Internet exceptionalism distorted the direction of investment in the commercial Internet, wasting many of the fruits of innovation from the edges.

The Prevailing View behind the Dot-Com Boom

A prevailing view is not a scientific hypothesis in which a deep theory links premises to conclusions. It is a social construct, a by-product of countless conversations, each one assessing the planning process at many firms. These conversations occur between investors and analysts, business partners and customers. Over time, a prevailing view always emerges from this collective conversation, and it evolves, becoming refined with myriad adjustments. Every well-run firm in a market is cognizant of the prevailing view, and every analyst and firm uses it as benchmark, assessing where their strategy overlaps with others and where it differs.

A prevailing view changes in reaction to events that arise, grow, and settle in a disorganized way. Chance events refine these beliefs and understandings, twist them occasionally, and obscure the origins of assumptions. They coalesce into a collection of opinion, facts, alleged consequence, promised payoff, and narrative arcs with biographical triumphs and failures. Some logical fallacies can persist for a time, and so can underexamined assumptions and hypotheses.

Prevailing views can linger, and for multiple reasons. No law forces every market participant to test his/her views against hard evidence. Conclusive statements can lack tangibility until hard numbers match revenue with operational costs. There is no requirement or norm that a single coherent document answers the many dimensions for assayers. No entrepreneurial participant must meet a requirement for full transparency, and many private firms can hide their financial results from public scrutiny. No law requires every single participant to explicate all of its positions for pundits, skeptics, and news reporters.

Initially the foundations for the prevailing view behind the dot-com bubble arose for sensible reasons grounded in observable experience. The fast diffusion of e-mail and web browsing in 1996 and 1997 was unusual for the speed with which mainstream users adopted it. The number of browser users reached an uncommonly high level in the United States, and across the globe, and by 1998 showed no signs of slowing. Some very visible entrants with new conceptions of business grew quickly alongside household and business adoption, such as Yahoo, Cisco, AOL, Netscape, and Amazon, so commentators further (correctly) pointed out that the new technology could become viable businesses at an extraordinarily fast rate. Some early entrepreneurial ventures—Vermeer, HoTMaiL, or ICQ—also achieved their commercial goals, merging with other leading firms, reaping returns on investment far in excess of normal outcomes. That too suggested (correctly) that some early movers had earned high returns for their investors.

The dot-com bubble emerged because the early experience supported optimism about 1999, 2000, and beyond. It suggested that viable business could be developed quickly, and, crucially, that a lucrative outcome could arise from an incautious approach to growing a business for users as quickly as possible, often deferring the generation of revenue to later. It presumed all the other elements of the economy would raise the value of adoption, enhancing the value of services, reinforcing further incentives to grow. Eventually the reasoning about the economic value of growing a new business became detached from grounded reasoning about what type of value the business actually created, how much revenue a new business could realistically expect, and what would be needed to cover operational expenses.

What led experienced and sensible entrepreneurs to throw out normal practices and invest with such incautious expansive plans? After the fact, it is challenging to make sense of the psychology of the times. Yet every explanation comes down to the same psychological mantra; these actions made no sense unless everybody believed the Internet made their business exceptional, that it would defy standard economic rules, gain many users in an extraordinarily short period of time, and then turn those users into a regular source of revenue in a profitable business with frictionless ease. No other rationale seems consistent with the facts.

What encouraged this view? It was tempting to fall back on cliché. It appeared as if some investors fell in love with these ventures, and their managers threw caution to the wind, as if hope triumphed over experience, dismissing tropes about the frailty of business success. Skewed social norms also encouraged an almost willful “misattribution bias,” allowing entrepreneurs to believe commercial success largely grew out of their own efforts and had little connection to prior invention and investment in the Internet and the web. There also was a grain of truth in such cliché’s. Building new businesses required the assent and enthusiasm of people who worked at these firms—entrepreneurs, programmers, managers, administrative assistants, and engineers. The pursuit of dreams could and did motivate extraordinary sacrifice, and in this moment of time, the prevailing view for the late 1990s informed many participants that it was time to pursue a once-in-a-lifetime opportunity. The promise gave purpose to work that often lacked it, such as routine programming, and motivated many talented people to change their careers and take risks.

The history of financial bubbles suggests a deeper explanation. Many investors questioned the relevance of traditional valuation processes for the rise of electronic commerce and advertising-supported online media. There were few observable and measureable activities, and most of the value for these firms lay in the future. The distance between the present activity and later value made these markets unlike a stable industry with a long historical record. Analysts of electronic commercial and advertising-supported media did not have sufficient records to construct models that predicted future revenue and costs. If certain factors changed—say, unemployment fell or the price of oil rose—analysts had no model to forecast the effects on demand and the cost of production. They had no finely calibrated model to predict the value of a firm under optimistic and pessimistic scenarios.

As a practical matter many analysts were left grasping for anything they could measure, such as web traffic. For example, small changes in the growth of traffic to a site—the number of visitors, the average length of stays, and any other aspect—would be seized on as informative. These analysts then had to layer heroic assumptions on top of such data, for example, presuming that more traffic at one point led to more advertising revenue at a later point.⁷ These models were, at best, a shot in the dark, and only the most extreme outcome—such as little or no traffic to a website—could reject an optimistic thesis.

How did such shots in the dark take on the status of historical models? The history of financial bubbles also provides guidance. Investors need a prevailing view to explain and rationalize for participants why familiar long-term determinants of value ceased to determine firm investment and why familiar business risks did not apply with the same brutal force.⁸ Internet exceptionalism offered that outlook.

Internet exceptionalism generally took on one of two forms. One form, common among technologists and engineers, stressed the unique technical basis for the Internet and saw it as a central, relegating any economic factor to a secondary effect. A second and related form, very common during the dot-com boom, believed that the Internet followed its own unique economic rules, having little in common with other important historical episodes. Economics mattered, or would matter eventually, but not for some time, and in the short run it certainly would not reflect the recognizable principles practiced only a few years earlier in non-Internet markets.⁹ In either form, Internet exceptionalism forecast the replacement of existing non-Internet-related activities by the new entrepreneurial entities. Without specifying how it would happen precisely, there was a presumption that eventually revenue would appear as activities moved to the new and better Internet economy.

If there was a grain of truth to Internet exceptionalism, it arose in the recent experience with fast household and business adoption of browsers. It offered some basis for an optimistic forecast about the scale of sales in an Internet-based business. The strong rhetoric for the entrepreneur-led new economy built upon this optimism and went further, claiming a general trend of additional success for entrepreneurs in the majority of markets built around the web. Eventually the actual data caught up to this claim, but in 1997, 1998, and 1999 there was little to refute it. At that point, any investor could only speculate about a world characterized by pervasive adoption and use of the Internet, and what the world would resemble when many new firms served all those web surfers.

Cliché alone does not explain the phenomenon in its entirety for another reason, and this leads to a deeper inquiry. Investors did not apply brakes to the incautious actions of new ventures, as normally would have occurred. Sober and hard-nosed financial reasoning failed to limit the number of ill-conceived businesses. Financial markets failed to value an asset on the basis of the likely discounted sum of the revenues over costs, and, accordingly, failed to place brakes on imprudent investment. How did that happen? Answering that question is one of the goals of this chapter.

Timing had to play a role. Some risky investments made at the outset of any commercial gold rush will appear to be incautious later. It is not surprising that financial markets approved of some risky investments in 1996 and even 1997. Some risk leads to some return and some ruin. Some ruin could be excused due to the coincidence of the backbone’s privatization and the growth of the World Wide Web, which generated considerable returns as well.

That explanation only goes so far, however. When discussing many investments in 1998 and 1999, events had gone well beyond the initial gold rush, and virtually all informed participants understood the key building blocks of technology. Financial analysts knew sufficient examples to illustrate how to make and build businesses using Internet technology, and, crucially, how not to. Yet the knowledge about what could go wrong did not stop the wave of investment.

Four explanations for this wave receive prominent attention in this chapter: the wisdom and grounded analysis of many seasoned industry experts became sidelined in favor of analysts that parroted Internet advocates; many financial advocates gave in to the prevailing view and also argued for Internet exceptionalism; enough investors followed this advice for a long enough time to set in motion enough demand to bid up prices for entrepreneurial firms, and far away from their long-term fundamental determinants; a number of institutional practices within financial markets in the United States suppressed information inconsistent with the prevailing optimistic view. In other words, this was a situation ripe to produce too much money chasing too few good deals.¹⁰

Cheerleaders

An examination of cheerleaders and financial institutions of Wall Street can help explain how a prevailing view can sustain itself. The actions of author and pundit George Gilder, a longtime and well-known commentator in conservative political circles, can begin to illustrate how the prevailing view began to take shape.

No law prevented a pundit from trying to shape the prevailing view about the sources of value in the future, and Gilder worked hard at becoming a recognized opinion shaper. That effort, by itself, put him in the same group with scores of others with similar ambitions. He wrote books,¹¹ and during the 1990s he published the “Gilder Technology Report” for subscribers. He differentiated himself by focusing particularly on broadband firms and activities taking advantage of abundant, pervasive, and inexpensive high-speed networking. At his peak Gilder had 75,000 subscribers, which generated $20 million in revenue.¹²

Nobody doubted his sincerity or integrity, as Gilder displayed an enthusiasm for his subject, clearly believed what he wrote, and invested his own money in the firms he liked.¹³ Cynics publicly worried that Gilder merely played a role from which others profited and reached conclusions on the basis of many flawed and incautious observations. As Fred Hickey, editor of a newsletter called the High-Tech Strategist, said:

If there was no George Gilder, the venture capitalists and investment bankers would’ve invented one. They needed some kind of pied piper to put the words on paper to justify the insanity of paying any price for anything that offered any kind of technical promise.¹⁴

Gilder acquired a distinct status somewhat by accident, as an unwitting result of crossing the line between pundit and investor. Ostensibly in order to be transparent, he began listing his own investments on the last page of his newsletter, calling them his favorite firms.¹⁵ Soon thereafter other observers noted that such a designation could move the stock prices of those listed firms, and the movement often occurred whether or not any other fundamental factor had come to light about that firm. Labeled the “Gilder effect,” other pundits and short-term investors began to pay attention to Gilder’s list, whether they agreed with him or not.

As the history of short-term financial bubbles suggests, once a person is designated a prophet of price movements, such a prophet’s pronouncements could become self-fulfilling.¹⁶ It was no surprise, therefore, that self-interested investors learned to understand Gilder’s well-documented preferences and tendencies, and they made efforts to get the firms in which they invested to tailor their appeal to his optimistic forecast about networking’s value. Firms with stakes in technology markets, especially in networking, began to go to great lengths to influence Gilder’s views, paying his fees to give speeches to their clients, as much as $100,000 per speech. Such firms hoped to use the opportunity to place a favored start-up in front of Gilder, hoping for an endorsement of their favored entrepreneurial firm on his list.

After the dot-com bust a mini-industry of critics, perhaps motivated by schadenfreude, began documenting the number of ways in which Gilder had erred in his forecasts and analysis, as well as why he did.¹⁷ Such observations largely missed the point, that Gilder was a product of his time. Gilder, or any cheerleader, could not have had such an inordinate influence without the presence of a prevailing view that fostered a speculative financial bubble.¹⁸

Further illustration of this larger point comes from the experience of Henry Blodget, who also acquired some of the same status, albeit from a very unlikely path. After graduating from Yale, Blodget joined the corporate finance training program at Prudential Securities in 1994 and next moved to Oppenheimer and Company to do equity research. Blodget was young and bold, blessed with an ability to write clear prose, and not colored by association with an established line of analysts. He schooled himself in the new economy and made forecasts that gained attention. In his own words:

In 1997, I recommended that my clients buy stock in a company called Yahoo; the stock finished the year up more than 500 percent. The next year, I put a $400-a-share price target on a controversial “online bookseller” called Amazon, worth about $240 a share at the time; within a month, the stock blasted through $400 en route to $600. You don’t have to make too many calls like these before people start listening to you; I soon had a global audience keenly interested in whatever I said.¹⁹

Blodget was then offered a position at venerable Merrill Lynch, to which he moved. To longtime and established analysts, this represented a shocking departure from the norm—the ascension of a young child who had not paid his dues. For Merrill Lynch, this was merely an attempt to prevent clients from going elsewhere, an attempt to affiliate their brand with a newly minted analyst-turned-celebrity. It also illustrates just how quickly the established firms of Wall Street could and would jettison experienced and tempered analysts and those who used long-term fundamentals, replacing them with analysts who reflected the new prevailing view.

Blodget’s job was ostensibly as an analyst, but after this appointment Blodget’s role became similar to Gilder’s—part pundit, part advisor, and part prophet of prices. From this post, and at the height of the bubble in 1998 and 1999, Blodget influenced the stocks of many firms. Many paid attention to him because they had to, and enormous efforts went into influencing his views. To be fair to him, also like Gilder, Blodget appears to have believed what he wrote. He invested his own money in the sector, losing it in the dot-com bust.²⁰ Indeed, Blodget eventually lost all credibility after the bust. He accepted a buyout offer from Merrill Lynch in 2001.²¹

In a retrospective written many years after his time there, Blodget rationalizes his role by identifying the factors encouraging short-termism in all facets of the financial markets. According to this analysis, Wall Street firms lack institutional memory, because few money managers familiarize themselves with the history of prior bubbles and many have careers too short to recall prior bubbles, which happen only once every few decades. While that might have described Blodget’s own biography, this argument can only go so far, as it did not apply to the senior management at any of the major firms, including those at which Blodget worked. Blodget’s senior managers should have known about the need to limit downside risk and taken action to limit the downside risk to their firms from following a prevailing view with myopic features.

Perhaps this is why Blodget offers his second argument, that short-term pressures to ride a bull market overwhelm all other norms, such as valuing firms according to long-term fundamentals. To illustrate, Blodget tells three tales, each based on a real biography. In the first, an established money manager loses most of his investors for refusing to invest in hightech stocks, which leads him to close the fund (just before the dot-com bust). In the second tale, a money manager gains considerable funds to invest because he does what the craziest clients want, only to lose it all in the bust. In the third tale, a reluctant money manager compromises between his traditional practices and the calls for more technology investments. That leads him to maintain a portfolio that loses considerable value after the bust, at which point he retires.²² From these examples, Blodget argues that a perverse prevailing view will have its day, because no analyst or money manager can resist the pressure. These three stories together compose the “Blodget rationalization” for Wall Street’s behavior.

There is a deeper answer to the Blodget rationalization, but it would require Blodget to explain why the skeptics did not speak when they normally do in financial markets. The answer involves the incentives of those who have information to aid the skeptics of the prevailing view, and it requires Blodget to condemn the behavior of his former bosses and senior colleagues. Said another way, Wall Street did not lack for seasoned experts in financial firms. Many of them, particularly the senior management at many of these firms, knew that the values of dot-coms departed from fundamental economic determinants. Why didn’t they say so and authorize others to say so? Said simply, it would have cost them money. Many executives would have permitted craziness to continue as long as there was big money to be made.

During this time, self-serving rationalizations had captured the prevailing view within Wall Street, run amuck in managerial decisions, and no regulatory intervention corrected for this imbalance. Many aggressive Wall Street firms felt entitled to do whatever profited them, even if it meant undercutting others who sought to serve their clients honorably, and undercutting those who sought to serve society’s long-ferm needs and public obligations. It was this type of behavior—exhibited at many firms and collectively making the whole much worse than the sum of its parts—that lay behind the inability of Wall Street to self-correct.

One contributing factor, of course, was the gullibility of investors and employees, which was fostered by the genuine technical complexity of the situation and the inherent absence of hard data to forecast future events.²³ Although lack of sophistication, by itself, helps explain some investor behavior, it cannot explain the dot-com financial bubble in its entirety. The presence of gullible investors, even in a large scale, is not sufficient to produce a bubble. A sufficiently large set of informed and sophisticated investors normally could have moved financial prices in the appropriate direction, profiting handsomely from outwitting the unsophisticated investor.

Yet sophisticated investors did not disappear in the late 1990s. Why did these participants not play their usual role?

First, there is a bias on Wall Street toward more activity, of any type. Many firms in the financial-services business profited from any activity, whether or not the investor ultimately profited. For example, many firms profited from fees for services supporting mergers, for related services, such as underwriting IPOs, and for investment services, such as arranging to invest in growth funds consisting of entrepreneurial firms.²⁴ Such actors facilitated investment with any prevailing view.

How could that be permissible? Traditional US practices stressed caveat emptor, buyer beware, and placed the emphasis on buyers facing the onus of asking the right skeptical question. If the deals went through, the financial firms had little incentive to correct that view even if it was in error. In practice, many met the letter of their legal obligations by listing risks and potential issues, but few had any reason to meet the spirit of their obligations.

Second, many banking firms were less than fully forthcoming about their incentives to price IPOs properly. During the boom, many Wall Street firms gave access to early stock sales to privileged clients and then—either explicitly or implicitly—asked for quid pro quos, such as favorable public remarks or more banking business. This was known as IPO spinning, and the run-up in stock prices for IPOs from technology firms played a role in that behavior, making it especially easy to reward insiders.²⁵ Similarly, many firms also used laddering. That is, to purchase shares at a given price, investors also must agree to purchase additional shares at a higher price. Executed well, this guarantees higher prices for insiders who buy stock at low prices.²⁶ The profitability of spinning and laddering further muted any incentive for insiders to work with those who asked skeptical questions or make public information that corrected an error in a misguided prevailing view.

Third and finally, just plain self-serving unethical behavior gave financial firms incentives to arrange for IPOs, whether or not they made sense to investors. For example, many analysts worked at firms with conflicts of interest, facing pressure to make favorable remarks to keep some banking businesses at other parts of the firm.²⁷ That too underlay incentives not to correct an error in a prevailing view.

The misplaced incentive of many financial institutions provides an answer to Blodget’s view. Merely because a dollar was on the line, firms in the business of advising clients did not go out of their way to release information inconsistent with an overoptimistic prevailing view. Many skeptics held their tongue, or their questions about presumed future growth received little notice. It went on long enough to shape a majority of investment decisions. In this sense, Wall Street’s reinforcement of the prevailing view contributed to the era, both accelerating the rate of investment and shaping its direction, though not necessarily for the better.

The Bust after the Boom

The dot-com bust, as it became known, can be viewed as either a systemic event or as a parochial one. As a systemic event it did not come suddenly or occur in a single moment in time. There was no singular epiphany or moment of revelation, or a single event that obviously caused later declining events. Rather, the first signs of the decline appeared in 1999, and a few participants recognized the canary in the coal mine. Only with the benefit of hindsight do those signs come to the foreground of a narrative explanation, and this chapter describes these signs in the description below.

As a parochial event many participants experienced the dot-com bust in much the same way a nearsighted surfer on an ocean wave experiences the feelings of ascent and decent as waves grow or decline or crash. The ascent started in the mid-1990s and generated unending adrenaline rushes unlike any employment experience of the past or any investment returns in near memory. Many experienced the descent from peak only after the spring of 2000, as financial markets began to decline and, with zigs down and up, reached a low point in 2002.

At a systemic level this bust was entirely a financial and business phenomenon, not a technical one. The Internet’s engineering never failed. The Internet and the web continued to function properly and reliably, and users continued to make ample use of many valuable services, even after the bankruptcy of the largest backbone provider in the country, WorldCom, or the exit of some of the seeming leaders of the first era, such as Excite, or the entry of some of the leaders of up-and-coming firms, such as Google.

What caused the dot-com bust? It is quite common in popular discussion to frame this question around the timing of the peak, that is, the beginning of the decline. This common framing presumes that a trigger or spark caused the ascent to stop and reverse into a descent. This framing leads to what might be called “the fallacy of the last snowflake.” Focusing on the date of the financial peak is similar to asking why a specific snow-flake generated sufficient pressure to trigger an avalanche. Asking this question creates an unsatisfying approach to analyzing an avalanche. Why should the last snowflake take the blame? All the snow caused the avalanche, as did many other facets of the setting, such as water content of prior snowfalls, the topography of the mountain, and the location of trees. Focusing on that last snowflake takes away from assigning responsibility to all the other relevant factors.

Consider all the responsible factors that contributed to the dot-com bust, not just the last snowflake. The economic setting had been building to it for some time. First, as noted in earlier chapters, simultaneous investments in multiple sectors increased the value of other investments, as the network expanded and attracted more users. Overinvestment was inevitable in this setting, due to the lack of coordination between investments in so many facets of the Internet. The situation was ready for a change in the prevailing view that supported investment in the dot-com sector.

The first signs of overinvestment did not arise in the dot-com sector, as it happened. It arose in the building of the network infrastructure. Financial decline in this sector became known as the “telecom meltdown.”

What happened? Firms had built backbone capacity to support far more traffic than anybody would need for many years.²⁸ Competitive local exchange companies, or CLECs, also had entered with abandon. While they supported local ISPs, CLECs were a child of a new deregulatory era in communications, as well as the implementation of the 1996 Telecom Act, which laid out new competitive rules for interconnecting CLECs with the existing network.²⁹

The key regulatory issues interested longtime telephone attorneys and activists, but these issues left many others confused or ill informed. In brief, the national billing system for telephony assumed that interconnecting firms made as many calls as they received. In the US compensation system, a firm paid for only one of these—the calls made, not those received.

The problems arose when the rules behind billing system met CLECs and CLECs met the Internet. If a CLEC received about as many calls as they sent, then no issue would have arisen. Taking advantage of the intercarrier rules, however, a number of CLECs set up businesses to receive ISP calls from households but send very few, which effectively billed other telephone companies for “reciprocal compensation.”

This strategy went unnoticed by regulators at first, and consequently, regulatory decisions for reciprocal compensation of CLECs encouraged CLEC entry, especially as the Internet grew in 1997 and 1998. The practice received attention as it grew, resulting in hearings at the FCC in 1998. The FCC passed a set of rules to do away with the practice in February 1999,³⁰ which made some CLEC business no longer viable.

In a short time the financial decline in stock prices hit firms in infrastructure markets particularly hard. Investors in CLECs and their business partners’ firms, such as large-scale national ISPs, experienced declines in revenue unless they found other sources of revenue in the dot-com sector.

For many other participants in the Internet, the decline in CLECs was a distant event, and that perception illustrates the general challenges of the time. No single observer could easily grasp all the myriad forces at work. Behind the meltdown lay multiple complex causes, both commercial and regulatory, that enabled the meltdown to escape much public scrutiny. This lack of attention prevented it from serving as a warning for the majority of the Internet investment community.³¹ The dot-com bust followed the telecom meltdown, as it had to, and, yet, it took many by surprise.

The dot-com bust was not a sudden crisis but closer to a creeping realization, and one of diminished possibilities. For many participants the first symptoms of decline became visible in the early spring of 2000. On March 10, 2000, the NASDAQ reached its peak.³² It came down over several weeks, taking a dive on April 4,³³ only to rebound again over the next few months. Figure 16.1 illustrates this on the NASDAQ index, one of the important stock exchanges for young and newly public Internet firms.

At first, proponents of the new economy began to characterize the private market as “hostile,” and it delayed many IPOs. Those young firms and their financial backers expected a rebound. Indeed, signs of the rebound showed up in the summer, bringing false temporary hope. Then the decline returned, starting in the autumn of 2000. Skeptics of the new economy stressed the return to “normal,” and virtually all technology IPOs ceased in the face of no demand to invest in dot-coms. Financial indexes reached their lowest point after the World Trade Center attacks on September 11, 2001, when travel by air shut down for several weeks across the country. Pervasive doubts about the future of economic activity and the war in Iraq and Afghanistan cast a pall of uncertainty over many investments for many months thereafter.³⁴

These events led to an extraordinary loss in paper wealth. Stock market wealth, for example, reached bottom at $9.9 trillion in the third quarter of 2002. Overall, stock market wealth fell $8.3 trillion (or 46 percent) between its peak in the first quarter of 2000 and its trough in the third quarter of 2002.³⁵

Many of the high-flying stocks of the late 1990s experienced declines, including Amazon, Cisco, and other firms in electronic commerce. Other firms in advertising-supported businesses, such as Yahoo, AOL, and many others, also saw their stock value decline.

Many of the most technically adept start-ups of the era—focused on fiber optics and high-speed landline facilities—suddenly found themselves without buyers for their products, or any buyers for their firms. The frontier had been successfully stretched and it went too far. So many of those start-ups succeeded in their technical goal, but did not get to sell their products because there was no demand.

The wake of the bust directly shaped investments in start-ups and fed back into the willingness to start firms. Few VCs could forecast when their existing investments could cash out, and many of them declined to fund new firms. Consequently, financing for Internet start-ups sharply declined after the bust, nearly stopping altogether in 2002.

That brings the discussion back to the most difficult systemic question: What were the systemic causes behind the timing of the general growth and the decline? Again eschewing the fallacy of the last snowflake, and taking a wider view, what fundamental economic factors changed between the fall of 1999 and the fall of 2000?

The first candidate is interest rates. Interest rates had been low throughout 1999, and that ended when the Federal Reserve chose to raise short-term rates at the start of the new year. That factor alone would have led to a decline in borrowing for financial investment, and as long-term rates rose, a decline in the valuation of the future earnings of firms. The change in interest rates did not precisely coincide with the change in the stock market, but there was no reason for any coincidence of timing when Alan Greenspan was chair of the Federal Reserve. He deliberately tried to shroud the policies in the short run, and investors only gradually inferred what those policies were, and as they did, they reacted.

Another contributing factor was the turn of the millennium, which exposed a false forecast about massive failures in installed mainframe computers. Known as the Y2K problem, many old mainframe systems found at longtime users of enterprise IT contained programs that far outlived the designer’s expectations. These programs embedded a myopic error. The impending year was coded as two digits—“00”—instead of four—“2000.” If left uncorrected, the programs with two digit expressions would express “January 1, 2000” as “January 1, 1900.” Many of these issues could be forecast in advance, and many enterprises undertook large infrastructure upgrades to avoid them. IT firms had sold networking and computing services to large enterprise buyers as part of a general package to avoid problems with the expression of 2000. After there were minimal computing problems with the transition to the new millennium, it was almost inevitable that many of these budgets were going to decline. Consulting firms, hardware firms, and a wide variety of integrators for enterprises saw their businesses decline as well.³⁶

The last (and, perhaps, ultimate) contributing factor came from the performance of dot-com firms themselves. As the next chapter explains, online advertising did not become effective right away, and, as a result, did not generate enormous growth in revenues, as so many business plans had assumed. Moreover, despite some success in firms such as eBay, Amazon, and a variety of converts from catalog retailing, many of their cousins in electronic retailing did generate performances on anything close to the scale necessary to justify earlier impatient investments. The experience of WebVan, described earlier, was one such example. There were many others who were equally as impatient in their investment behavior, and some are described below. In this telling, the lack of performance in the Christmas season of 1999 would take special blame, as the poor results were reported in detail over the winter of 2000 for so many firms.³⁷

Irrational Exuberance

A popular phrase emerged to describe the frenzied financial investment that had characterized the moments prior to the crash. That phrase came from the chairman of the Federal Reserve Board, Alan Greenspan. In December 5, 1996, he made remarks that contained the phrase “irrational exuberance.”³⁸

By the time of the bust, the phrase had taken on a new interpretation. Greenspan had focused on the challenges asset values posed for monetary policy, using the experience in Japan and the crisis in 1987 as illustrations. He had not aimed any remarks at the US technology sector.³⁹ After this speech the phrase took on a life of its own, and after the dot-com crash it acquired pejorative tones—as shorthand for business behavior that lacked common sense.

Perhaps the most frequently cited example of unbridled irrational exuberance came from a merger at end of the boom, the merger between Time Warner and AOL, the largest media firm in the United States and the largest Internet access firm, respectively. Justified by many anticipated synergies created by the merger, content firms at Time Warner were supposed to feed content to AOL’s users, which then would give AOL excellent targeted advertising, and thereby gain the advantages of scale. In turn, this was supposed to raise the value of content from Time Warner, allowing them unprecedented opportunities to develop new content.

In hindsight, the merger looked like an obvious mistake, but nevertheless, the integration problems were so immediate and concrete it is a wonder nobody anticipated them.⁴⁰ Although the vision did not run into resistance among investors, its execution left plenty to be desired. In particular, financial incentives could not generate any cooperation out of Time Warner, which contained many established and successful magazines, and processes for generating stories and ads that had long successful histories.

Gains from combining the firms turned out to be a fantasy at worse, or an untested theory at best. Implementation never got close to the theory. Employees at both organizations were not enthusiastic about working with the other’s employees, and many division managers guarded their autonomy, not perceiving any visible or short-run gains from cooperation. Lacking any reason or incentive to cooperate, most did not. Alleged synergies never arose, not even a little bit. Before management had time to develop long-term plans, the dot-com crash reduced advertising from many of AOL’s partners. The decline in the US economy further reduced the value of many of Time Warner’s businesses. That took the air out of further programs to generate new initiatives. Going forward the combined divisions continued their activities as if independent of one another.⁴¹

Why did anyone think the merger of AOL and Time Warner was a good idea? There is really no good explanation for why the obvious challenges and risks were not anticipated. Blaming the prevailing view is the easiest explanation. Impatiently getting large was all that mattered, not carefully vetting the potential for raising revenue. The merger was consistent with the prevailing view at the point when crucial decisions were made, and the small set of executives responsible for it never got the message from any skeptic.

AOL and Time Warner were far from an isolated example of such managerial foolhardiness.⁴² Additional poster children for irrational exuberance touched many high-profile firms, such as PSINet, which was still operated by its founder Bill Schrader more than a dozen years after its founding. Weakened by the telecom meltdown, the dot-com bust accelerated the decline of many infrastructure firms. Dot-com firms could no longer pay for access services and more. Eventually, PSINet faced financial difficulties that led to bankruptcy.

PSINet’s bankruptcy, in June 2001, represented a comeuppance for Schrader, who had become one of the most vocal critics of the reluctance of telephone companies to invest in the Internet. By the end of the decade, the firm had become well known for its bold acquisitions, growing into one of the largest ISPs in the world. It was also well known for unbridled ambition—for example, purchasing the naming rights for the football stadium of the Baltimore Ravens.

The firm became the victim of two factors, a risky merger and leveraged debt. It merged with a large consulting firm, Internet consulting company Metamor Worldwide Inc., just before the dot-com bust. It was supposed to be a diversification, expanding PSINet into a business with a revenue flow distinct from its other lines of business.⁴³ In fact, it did not diversify at all against the bust. Metamor’s revenue base declined rapidly during the dot-com bust, and at the same time as PSINet saw revenue decline in many of its retail ISP businesses.

Second, over many years PSINet had primarily financed expansion through debt, not equity. Rather quickly after the bust, therefore, it faced cash flow issues related to making interest payments, and eventually it defaulted on these.⁴⁴ It is hard to fault PSINet’s management for this choice, since bank loans appeared to be cheaper, and so many of its direct competitors and business partners used the same source of finance.

Nonetheless, to see the risk inherent in that choice, contrast PSINet’s experience with Level(3), another infrastructure firm that also had expanded rapidly. Level(3) had financed itself through equity. While stockholders lost value after the bust, Level(3) neither had to declare bankruptcy nor cease operations because its interest (cash) payments were not as high. Instead, stockholders lost paper value. Stockholders were not happy, of course, but that is not the key point: at least their firm survived.

Other candidates for exuberance included WebVan, Pets.com, and JDS Uniphase, all of whom eventually filed for bankruptcy. These all appeared to be wasteful investments or excessive expansions made by faithful practitioners of the philosophy to “get big fast” who drove their firms off a cliff and wasted investor money. Each company was, in fact, a different variant of the illustration of impatient excess. As recounted earlier in the chapter, WebVan was the victim of impatient exploration, a flawed model unfurled too quickly.

In Pets.com the management committed a different type of mistake as a by-product of its impatience. It spent lavishly on publicity, especially its commercials involving a sock puppet, which became widely ridiculed. Its bankruptcy symbolized that no amount of advertising could make up for a poorly conceived business. As should have been apparent right from the outset, the shipping costs for large volumes of pet food rendered the service too expensive for a typical household, which could buy appropriate amounts from a local retailer. Once again, limited trials would have tested that theory and limited the downside losses.

JDS Uniphase, in contrast, had been successful for a time as one of the largest upstream suppliers of hardware components for a wide range of equipment. Yet after the crash it had few customers for its equipment and became a symbol of incautious expansion.

As with the other examples, it is hard to understand in retrospect how management made itself so vulnerable by expanding so quickly. As it turned out, the primary cause of its downfall was a classic textbook problem that any seasoned manager would have guarded against—namely, a whipsaw effect, which sharply reduced their revenue as demand declined (and would have reduced revenue in any event). A whipsaw effect turns moderate demand downturns into big declines for upstream suppliers and is a risk for an upstream supplier facing a downstream buyer with volatile demand. It can arise when the upstream supplier faces more volatility than the downstream supplier because a downstream supplier may not recognize a decline in demand soon enough and, consequently, adjust to new demand by drawing down on (excessive) inventory holdings rather than placing new orders. As a result, the upstream supplier can see sharp declines in demand from moderate changes downstream.

Guarding against the whipsaw would not have been difficult for JDS Uniphase to do, and, indeed, other firms did guard against such actions. For example, in contrast to JDS Uniphase, Cisco had outsourced much of its upstream activity to others, which left Cisco holding many fewer assets when its revenue declined. After the dot-com bust, Cisco continued to operate with only one large layoff.⁴⁵

Irrational exuberance also became a phrase for improper financial dealings in which the prevailing view enabled a firm to escape scrutiny (for a short while). For example, executives at WorldCom and Enron faced scandal as a result of accounting improprieties that came to light after the dotcom bust. First consider WorldCom, then Enron. Both will illustrate how impatient investors failed to ask skeptical questions, which allowed managerial impropriety to survive far longer than it should have.

WorldCom’s bankruptcy, filed on July 22, 2002, also represented a comeuppance for a newcomer. The firm had expanded through merger to become the largest backbone provider in the United States and the second-largest long-distance telephone company.⁴⁶ It had swallowed up several viable businesses, including Metropolitan Fiber Company, which included UUNET, and the infrastructure division of CompuServe.⁴⁷

WorldCom had become so large that antitrust action had been deployed to limit the firm’s actions. In February 1998, it completed its merger with MCI, the second-largest long-distance telephone company in the United States. The Department of Justice required WorldCom to spin off considerable backbone assets—those that were part of MCI—as a condition for the MCI merger. WorldCom’s attempt to merge with Sprint, announced in October 1999, ran into so much opposition in Europe that the firm had to abandon the proposal, even before the US Department of Justice finished its review. It was abandoned in July 2000.

After the fact, these interventions by government antitrust regulators appeared especially sagacious. As it had turned out, WorldCom’s management had been cooking its books, delaying the day that it had to represent its true financial state to outsiders.

On June 25, 2002, an internal auditor reported a newly discovered accounting fraud. WorldCom’s CFO had reclassified assets as expense, giving WorldCom the appearance of being in a rosier financial condition than was true in practice. The ensuing trial exemplified the excess, as the CEO, Bernie Ebbers, seemed to rely heavily on his CFO for financial expertise. In trial he appeared to have little grasp of the underlying cost structure of the firm he had built.⁴⁸

The comeuppance for Enron came at much the same time, and was especially dramatic. Its demise hurt investors, its employees, and a major national accounting firm, which had been intimidated into certifying financial rectitude even when Enron did not deserve it.⁴⁹

The firm’s experience became symbolic of business untethered from managerial competence, laced with charm and flamboyance that bordered on systematic deception. Although it was primarily an energy company, Enron had expanded into a range of prominent Internet businesses during the 1990s. It had a division that traded on contracts in backbone data exchanges, which had been the darling of scores of new-economy gurus. As came to light later, the trading had been a Potemkin village, nothing more than a showpiece, mostly for reporters and the financial analysts, and no large-scale trading had ever taken place.⁵⁰

It did appear to have real business aspirations behind the show. For example, the Enron Broadband Services group signed a deal with the video-rental chain firm Blockbuster for providing video on demand. That is not possible without real infrastructure, real pipes that deliver data at high volumes between points. Of course, the actual revenue from that deal was far in the future, while many of the costs were experienced early. Even in the face of those factors, however, and as it had done in other business units, Enron’s accounting division could not resist bringing the revenue forward, giving the appearance that the project was profitable, as if the division had made its financial targets.⁵¹

These particular deceptions did not doom Enron as much as its practices for borrowing money. Its financial officers instituted a program of complex financial transactions that let it borrow money against future revenues without paying extra interest for the riskiness of the practice. Once investors uncovered the extent of the practice, it became clear that Enron had become addicted to living on borrowed money in excess of future revenue, and the company’s financial position declined rapidly and irreversibly.⁵²

Would the dot-com expansion have happened without Enron’s deception? Yes, probably, but the absence of this particularly deceptive example might have dampened some of the false perception that the future had arrived so soon. That, in turn, would have reduced enthusiasm of the advocates for the prevailing view in the new economy, and the dot-com bust might have arrived sooner, wasting few resources for as long as it did.

These are but examples among scores that came to light after the dotcom bust. The decline of financial support for investments came as a shock to many with a parochial view of impatient investment. Such impatient investment was not sustainable without returns, and there were many signs that many would not experience such returns. In addition, the tales of deception, imprudence, and incompetence suggested unsound economic foundations, as if many investors had been duped. After so much scandal, the dot-com sector became tainted with suspicion, and the prevailing view supporting actions in the earlier era did as well.

Correcting Prevailing Views

Prevailing views don’t arise, grow, or settle in an organized or orderly way. They involve a collection of opinions, unverified hypotheses, facts, alleged consequences, promised payoffs, narrative arcs with fragile foundations, and logical fallacies. More to the point, prevailing views lack tangibility. No single document declares its many dimensions for assayers. Instead, analysts compete to spot the flaws in general presumptions, which become visible during moments when an event exposes the gap between fact and fantasy. Lessons can become understood only after the facts falsify bogus premises, and analysts pour over traces of volatile economic events revealed within the shards of financial flotsam and jetsam.

In retrospect it is obvious that the prevailing view behind the dot-com bubble was flawed, and by the beginning of 2001 it had become inescapable. The news hit many participants like a hangover, as if economic fate had a sense of humor and favored jokes delivered with a sardonic tenor. It was a cruel joke, at best, to those who had sacrificed so much and worked so hard to build businesses that fundamentally relied on false premises and value propositions that could never become viable. It was a devastating punch in the gut to those who had identified with something they regarded as a greater purpose, and especially to those who had believed the promises to improve the world while building the commercial network.

For skeptics of Internet exceptionalism, in contrast, the bust appeared to be a well-deserved comeuppance. Internet exceptionalism made a challenging situation more economically wasteful than it had to be by rejecting decades of lessons about assessing financial risk and return. Because of the uncoordinated investment in the commercial Internet, some part of the bust, some degree of overshooting, was inevitable. In addition, because innovation from the edges generated many new economy opportunities that required experimentation and entrepreneurial energy to explore, it was inevitable that many investments would fail. Flawed and overly optimistic financial institutions, which were insufficiently supervised, were an additional factor, and it distorted behavior. It oriented toward selfish short-termism at the expense of investors and the economy as a whole.

The dot-com bust did not eliminate innovation from the edges, but it did reorient participants’ and investors’ understanding of it. After the turn of the millennium, the commercial Internet moved forward, with many participants chastened. Many survivors adopted a view grounded in actual user behavior, realistic assessments of the costs of supplier operations, and modest expectations about the costs of building a customer base for a web-based business.

¹ Quoted in Heilemann (2001).

² Delgado (2001).

³ See Greenstein (2004).

⁴ At its peak it served ten cities: the San Francisco area, Dallas, San Diego, Los Angeles, Chicago, Seattle, Portland, Atlanta, Sacramento, and Orange County. It made plans to serve twenty-six but went bankrupt before those could be built.

⁵ WebVan raised nearly $400 million in four venture rounds, and it raised an additional $430 million more in its IPO.

⁶ Since WebVan was financed through equity, even if the concept had been proven unworkable after a couple trial cities it is rather doubtful the management would have returned any of the investor money. That still does not take away from the observation that their expansion plan departed from conventional practices, which are designed to learn lessons over time, and not put them into practice on the next implementation until they are shown to be profitable.

⁷ For more of this view see, e.g., Rajgopal, Kotha, Venkatachalam (2000) and Demers and Lev (2001).

⁸ Reinhart and Rogoff (2009).

⁹ The latter attitude explains why many businesses could read Shapiro and Varian (1999) and its call for a return to traditional economic analysis, and, yet, hold its message at a distance.

¹⁰ See Gompers and Lerner (2003).

¹¹ See Gilder (1989).

¹² See Prince (2006), and Rivlin (2002).

¹³ A reflection of these views can be found in, e.g., Gilder (2000).

¹⁴ Rivlin (2002).

¹⁵ Gilder followed his own advice and eventually lost most of his wealth to the dot-com bust. Rivlin (2002).

¹⁶ This statement is necessarily a simplification of a vast literature. See, e.g., Reinhart and Rogoff (2009).

¹⁷ For popular press pieces with such themes, see, e.g., Prince (2006), and Rivlin (2002).

¹⁸ For an analysis much closer to this observation, see Odlyzko (2010). He shows how Gilder erred in his analysis of networking and stresses that investors could have but did not ask sufficient skeptical questions before investing. See Odlyzko (2010).

¹⁹ Blodget (2008).

²⁰ Eventually he lost a considerable fraction of his wealth during the dot-com crash. Blodget (2008) states:

In the late 1990s, as stocks kept roaring higher, it got easier and easier to believe that something really was different. So, in early 2000, weeks before the bubble burst, I put a lot of money where my mouth was. Two years later, I had lost the equivalent of six high-end college educations.

²¹ Later still Blodget was charged with securities fraud for stating public positions that contradicted private opinions documented in e-mails. He paid a fine and was barred from Wall Street analysis.

²² For review, see, e.g., http://www.pbs.org/now/politics/wallstreet.html, retrieved October 2012.

²³ Andrew Odlyzko (2010), an expert in measuring networking traffic, identifies these factors as particularly salient in the late 1990s. 1. Innumeracy: Many lacked the ability to understand the implications of claims that traffic grew at such a fast rate, doubling every several months. 2. Misuse of basic engineering terms. Many investors measured capacity, not traffic. 3. Unchallenged inconsistencies: Many investors did not ask skeptical questions about different growth projections, with some predicting rates out of scale with each other.

²⁴ For a review of the evidence, see, e.g., Brau and Fawcett (2006), or Liu and Ritter (2010).

²⁵ See Liu and Ritter (2010). For a description of this behavior in the IPO for E-Toys, see Nocera (2013).

²⁶ Hao (2007).

²⁷ Reuter (2006).

²⁸ As noted in Greenstein (2005, 40), “excessive” entry in infrastructure typically means some form of redundancy—two or more facilities serving the same geography. Some degree of overlap was inevitable at a national level because the Internet permitted data to take multiple pathways to accomplish the same tasks. Moreover, in most business ventures some “building ahead of demand is a calculated gamble,” and the backbone business was no exception. Hogendorn (2011) goes on to show, however, that backbone firms faced fewer issues in building their physical infrastructure than in swaps and leases, the components that comprised virtual infrastructure. He also argues that much of the alleged overbuilding was a by-product of double counting of leases.

²⁹ For different interpretations, see Goldstein (2005), Greenstein (2005), or Goldfarb and Xiao (2011).

³⁰ FCC docket No. 99-38, Implementation of the Local Competition Provisions in the Tele-communications Act of 1996, Inter-Carrier Compensation for ISP-Bound Traffic, released February 26, 1999.

³¹ See, e.g., Goldstein (2005).

³² On March 10, 2000, the NASDAQ composite index closed at 5,048 after reaching a peak of 5,132 during the day.

³³ This is the day after the announcement of the Microsoft findings of fact. In circles critical of the court’s findings this coincidence of dates leads to the spurious conclusion that the court’s findings caused the dot-com bust. Such a conclusion is an illustration of the snowflake fallacy. The dot-com bust was going to happen at some point because many factors caused the overvaluation of stocks. The news on a particular day was, at best, a single snowflake, and, at worse, merely a coincidence.

³⁴ In fact, the NASDAQ composite index reached its low almost a year later, closing at 1,108 on October 10, 2002.

³⁵ This comes from the flow of funds accounts, maintained by the Federal Reserve. These are available at www.federalreserve.gov/releases/z1.

³⁶ That explanation is consistent with the general timing of the decline, albeit, it does not explain why it occurred in March instead of February or April.

³⁷ Goldfarb, Kirsh, and Miller (2007), and Goldfarb, Kirsch and Pfarrer (2005), argue that the decline in stock values was attributable to poor reports from electronic commerce retailers in the just-completed Christmas season.

³⁸ It appeared in a speech given at the Annual Dinner and Francis Boyer Lecture of the American Enterprise Institute for Public Policy Research in Washington, DC.

³⁹ The original speech makes that clear:

Clearly, sustained low inflation implies less uncertainty about the future, and lower risk premiums imply higher prices of stocks and other earning assets. We can see that in the inverse relationship exhibited by price/earnings ratios and the rate of inflation in the past. But how do we know when irrational exuberance has unduly escalated asset values, which then become subject to unexpected and prolonged contractions as they have in Japan over the past decade? And how do we factor that assessment into monetary policy?

See http://www.federalreserve.gov/boarddocs/speeches/1996/19961205.htm, accessed November 2012.

⁴⁰ See the critique in chapter 12, which benefits from hindsight: “January 2001: Merger of AOL and Time Warner,” in Bruner (2005), 265–91.

⁴¹ The financial value of the combined entities began to drop not long after the merger. Many years later the entire deal was taken apart at enormous loss to the participants. See the extensive description in chapter 12, “January 2001: Merger of AOL and Time Warner,” in Bruner (2005).

⁴² A little subculture grew up around the most spectacular stories and declines, with websites devoted to collecting stories. See, e.g., Kaplan (2002).

⁴³ Bill Schrader, private communications, July 2008.

⁴⁴ See, e.g., Woolley (2001).

⁴⁵ Cisco survived through one large layoff and, in effect, through transferring much of the risks of its expanding product line to its parts suppliers, typically Asian subcontractors, who were left holding inventories.

⁴⁶ Acquisitions included Advanced Communications Corp. (1992); Metromedia Communication Corp. (1993); Resurgens Communications Group (1993); IDB Communications Group, Inc. (1994); Williams Technology Group, Inc. (1995); and MFS Communications Company (1996), which included UUNET. Remarkably, after the Sprint merger was stopped, WorldCom attempted one more, the acquisition of Digex in June 2001.

⁴⁷ The online presence went to AOL, while the part of AOL, what was left of ANS (originally from IBM), was sent to WorldCom.

⁴⁸ News analysts speculated whether this was a feigned lack of understanding to escape culpability or evidence that he had been over his head, acting as cheerleader for a business he did not understand. See, e.g., Jeter (2003).

⁴⁹ These events reached epic levels for their incompetence, scale, and drama. See, e.g., Eichenwald (2005).

⁵⁰ The scale of the deception is rather remarkable, partly for its audacity. The division did nothing useful, nor was it profitable. Nobody really knew the truth until well after Enron imploded—not the reporters who profiled the firm during its rise in the 1990s, not the auditors, or stock analysts, or consultants. See, e.g., Eichenwald (2005).

⁵¹ See http://phys.org/news68469936.html, accessed February 2014. Also see Eichenwald (2005), for a much more detailed explanation of the accounting, and how the firms systematically realized future revenue in its accounting prior to its actual collection.

⁵² Eichenwald (2005).