依旧强大：互联网的隐藏秩序

CREDIT SUISSE FIRST BOSTON CORPORATION

Americas

Michael J. Mauboussin 1 212 325 3108 [email protected]

Alexander Schay 1 212 325 4466 [email protected] Equity Research

U.S. Investment Strategy July 7, 2000

Still Powerful

The Internet’s Hidden Order

• Despite a correction in the Internet sector over the past six months, the relationship between rank and market value of online firms continues to exhibit power law behavior.

• This winner-take-all, market-capitalization distribution supports the notion that competition on the Web has some unique characteristics. For example, large firms can grow as quickly as small firms and the positive reinforcing nature of the medium provides leaders with a source of competitive advantage.

• We would continue to focus on the companies with customers,

capital, and viable long-term business models.

Executive Summary

Despite the Internet’s apparent complexity, hidden order exists. Studies demonstrate that Web pages are distributed among sites according to a universal power law. Power laws are distribution functions for measurable quantities—such as the number of earthquakes that occur at a particular intensity, or the number of cities that exist within a given range of population. The number of Web pages adheres to a remarkably stable regularity. Roughly described, power laws hold 1 that many sites have few pages and that few sites have many pages.

The distribution of Web pages is not the only online phenomenon that adheres to a power law. The distribution of visitors per site follows a power law as well: many sites have very few users, and a few sites have very many users. Although intuitive, the precise nature of these two regularities carries profound implications. Online firms that depend on the ability to attract and monetize users face significant challenges. The observed power laws suggest a low probability that 2 newly established sites will attract a significant number of users. In fact, from January 1999 to April 2000, 20 of the top 25 companies in the MediaMetrix “Top 3 Properties” remained in the top 25.

4 We first explored power laws and the Internet in late 1999. Despite a sharp correction in Internet stock prices over the last six months, the relationship between rank and the market value of online firms is still best expressed with a power law. This winner-take-all distribution continues to support the notion that competition on the Web has unique characteristics. Not only can large firms grow as quickly as small firms, but the positive-reinforcing nature of the medium also translates into a source of sustainable competitive advantage. This report explores some of the factors that may contribute to the observed distribution.

Powerful Web

Let there be no mistake—activity on the Web is highly concentrated. To show the breakdown of Web site popularity, we ranked the number of unique site users on the horizontal axis and plotted page views on the vertical axis, using MediaMetrix data for the top 500 domains in April 2000. The result is a straight line when using a double logarithmic scale. The data show that the top 5%, or 25 sites, account 5 for 31% of the total volume of traffic. This is consistent with broader studies of the Web showing that the top sites account for a very large percentage of total traffic volume. That is, one-tenth of one percent of the total, or the top 119 sites, 6 account for 32% of the total volume of site traffic. Very few Web sites get large, and very few Web sites get a lot of user traffic.

Figure 1 Power Laws

T o p 5 0 0 , Ap r il 2 0 0 0 10 5

Hits (000)

10 4

10 3

10 2 10 0 10 1 10 2 R an k

Source: MediaMetrix and CSFB analysis.

A logarithmic plot of rank versus frequency that is a straight line with slope near unity (as seen in Figure 1) illustrates Zipf’s law. This regularity has been observed in diverse realms, including the magnitude and frequency of earthquakes, rank and city sizes, and the distribution of words in a language. Extremities are apparent when Zipf distributions are plotted on a linear scale. A few elements score very high, a medium number of elements score in the middle of the road, and an enormous number of elements score very low. (The long tail of the distribution hugs the x-axis.)

Power laws are unusual—they represent special situations. Gaussian, or normal, distributions are much more common in nature than power law distributions. For example, the velocity of gas molecules in a room, the length of new-born babies, and the distribution of market capitalizations for the broader stock market all follow Gaussian distributions. One common trait of Gaussian systems is independence. To use gas molecules as an illustration, the velocity of any given molecule will not explicitly depend on the velocity of the other molecules. Even when the molecules collide, there is only a small degree of interaction.

In contrast, power laws occur when there is a high degree of interaction. The canonical example is a forest fire. The probability that a given tree in the forest

will burn is highly dependent on whether other trees in the forest are burning and the location of the tree relative to its neighbors. The underlying “network” position of the trees determines whether or not the blaze dies out, is confined to a small area, or becomes a conflagration. This complexity gives rise to power law distributions in the sizes of forest fires. The concentration of hits on a Web site also translates into a concentration of market value, so the most frequently used sites generally have the best opportunity to create viable business models. It’s interesting that the valuation of Internet companies still conforms to a power law, despite a pronounced downturn in the segment over the last six months. This can be seen in Figure 2.

Figure 2 Internet Market Capitalization Power Law

Internet 10 6

Market Value (000,000)

Source: CSFB analysis.

This analysis provides compelling evidence that the market continues to value Internet companies correctly on a relative basis. It is important to note that these data remain mute on the issue of absolute valuations. Contrary to popular perception, there is some discernible order to Internet valuations. So the question becomes, Is this order unique to Internet stocks? Do other industries exhibit the same characteristics, where a disproportionate share of the total market capitalization resides in a handful of firms? We analyzed 46 sectors from 18 different industry classifications in an attempt to answer this question. This broad survey revealed that 9 sectors exhibited strong power law characteristics. They are displayed in the Appendix.

The Strong Get Stronger

While no pure Internet brand is reinforced by the benefit of a physical presence, many particular characteristics of the online medium allow those companies that are already ahead to stay ahead. One example noted in “Absolute Power” is that new sites tend to link to already-popular sites in a bid for increased traffic. Such a reinforcing link structure is reconfirmed in a number of subsequent studies of the Web.

The most comprehensive study of the Web to date reveals some interesting insights into its macroscopic structure. Researchers at IBM’s Almaden Research Center in San Jose, California, examined over 200 million pages and 1.5 billion links (roughly one-fifth of the estimated current Web). They found that the Web’s connectivity resembles a bow tie. On the left side of the tie is what they dub the “newbies.” This group of sites consists of 43 million pages (or 21% of the total) that are relatively new. These sites link to the center, or what is called the “Strongly Connected Component” (SCC). However, these established players in the center—with over 56 million pages—do not link back to the newbies. Sites in the SCC consist of portals and many of the top corporate sites in the MediaMetrix data. On the right side of the bow are the “introverts.” This group includes a lot of lesser-known e-commerce sites. Although the SCC is linked to these introverts, 7 the introverts do not provide links back to the strongly connected center.

Most search engines rely on so-called “crawlers.” Crawlers work by indexing a Web page, jumping to other pages linked to it, then indexing those, and so on.

The Almaden study contradicts earlier suggestions that any two Web pages are connected by a relatively small number of hyperlinks. In order to give wide breadth of coverage in the future, search engines will have to crawl from a 8 greater diversity of starting points than they do today. Until this occurs, they will continue to channel traffic to the most popular and entrenched Web sites— without any reciprocity for the newbies. With only a fraction of the growing Web appearing in most search queries, traffic patterns to the existing strongly connected sites are positively reinforced.

Advertising Spiral So how does this translate into economic value? One example of the value concentration is advertising. Advertisers aren’t taking any chances with the “new” medium: about 95% of Web advertising is spent on 1% of all ad-supported Web sites. And while the absolute number of Web sites receiving online ad revenues continues to grow, the percentage share of ad dollars being spent at the top 50 sites continues to rise.

Table 1 Publishers’ Share of Total Web Ad Revenues

Source: eMarketer estimates.

Although this profile may change as advertisers shift their dollars away from portals and toward more highly targeted vertical Web sites, advertisers continue to gravitate toward high-traffic sites.

Web Competition Models that try to explain the characteristics of electronic markets—where a few sites thrive while most are driven to the verge of extinction—conclude that sudden transitions occur. As the competition between Web sites heats up, a profound shift can take place from an environment where many sites simultaneously thrive to a winner-take-most market where a handful of sites attract most users. This is attributable to a nonlinear interaction among sites, which effectively reduces the growth rate of some, owing to extreme competitive 9 pressure from others.

With firms spending enormous sums to sustain growth, it is easy to see that it is possible to reach a critical spending point where funding dries up (in tandem with the business fundamentals). This phenomenon can occur even with only two competing sites, and is comparable to the “principle of mutual exclusion” in ecology. Two predators of the same prey cannot coexist in equilibrium when 10 competitive predation is very strong.

Once the fixed development cost of operating a Web site is spent, it is relatively inexpensive to scale capacity to meet increased demand. So as Web sites proliferate, focus on aggregate demand becomes the order of the day.

Unfortunately, many Web company “predators” are competing for the attention and resources of the same consumer “prey.”

Conclusion

Figure 3 Trailing Venture Capital Money Flows

Dollars (billions)

Source: VentureOne Corp. and CSFB analysis.

Investment by venture capital firms has been unprecedented in the last two years. In the fourth quarter of 1999 alone, $14.9 billion was spent on new business opportunities. This quarterly total was higher than any annual amount in the history of U.S. venture capital. Picking up where 1999 left off, venture-backed firms raised $17.8 billion in the first quarter of the new millennium. About 84% went to Internet-related ventures. We register our skepticism today, as we did six months ago, about the ability of most online firms to survive the unforgiving rule of the power law.

Since the beginning of the second quarter there has been a marked change in perception of Internet businesses. Investors are now focusing on a new acronym:

P2P—the path to profitability. With some high-profile firms at 52-week lows and others closing their doors altogether, investors must remember the inherent advantage of today’s market leaders. They have the customers. They have the capital. And they benefit directly from the self-reinforcing nature of the Web.

N.B.CREDIT SUISSE FIRST BOSTON CORPORATION may have, within the last three years, served as a manager or co-manager of a public offering of securities for or makes a primary market in issues of any or all of the companies mentioned.

1 This was initially reported in “Growth Dynamics of the World-Wide Web.”

Huberman, B.A., Adamic, L. Nature Vol. 401, September 9, 1999. Since then the results have been reconfirmed in “Graph Structure in the Web.” Broder, Andrei,

et. al. IBM Almaden Research Center, San Jose, CA., June 2000. 2 ibid. Huberman. It’s important to note that the observed power law behavior was derived from usage logs for all web sites, not usage logs for commercial web sites alone. Corporate strategy can influence the ability to crack the line-up. 3 Six properties were absorbed by competitors that remained in the top 25 in April. The balance of the firms remained in the top 100.

4 Michael Mauboussin, Alexander Schay, Stephen Kawaja, “Absolute Power,”

Credit Suisse First Boston Equity Research, December 21, 1999.

5 Although we could find no direct statistical correlation between site traffic and market capitalization, 14 of the top 25 companies in site volume were also in the top 25 in terms of market value.

6 “Growth Dynamics of the World-Wide Web.” Huberman, B.A., Adamic, L. Nature Vol. 401, September 9, 1999.

7 “Graph Structure in the Web.” Broder, Andrei, et. al. IBM Almaden Research Center, San Jose, CA., June 2000.

8 “Souped Up Search Engines.” Nature Vol. 405, May 11, 2000 www.nature.com 9 “Competitive Dynamics of Web Sites.” Maurer, Sebastian M., Huberman, Bernardo A. Xerox Palo Alto Research, March 17, 2000.

10 Ibid.

Appendix

Complements We analyzed 46 sectors from 18 different industries for possible power law trends. Data are from FactSet industry classifications. This broad survey shows that 9 of these industries show strong power law characteristics. (See Table 2.) In every industry studied we found that the largest companies best follow a power

law. We show the relative size of this subset of companies in the first column. The second column gives a measure of the extent to which each industry is dominated by its largest companies, and the third column quantifies how well the companies fit a power law. (Smaller numbers closer to zero correspond to better fits.) Here we employ the standard statistical practice of taking the square root of the mean squared residual as a goodness of fit test.

Source: CSFB analysis.

To further illustrate the procedure used to establish the results in Table 2, we present the Internet industry results in greater detail.

Figure 4 Internet Services Industry

In te r n e t In d u s tr y 10 8

Market Value (000)

10 4

10 0

1 0 -4 10 0 10 1 10 2 R ank

Source: CSFB analysis.

Figure 4 displays the market value (in thousands) of each of the 297 companies in the Internet industry. As a function of rank, the top 140 companies (47%) closely follow the fit line. Figure 4 demonstrates how the fit line was chosen. Each point is the result of a power law fit to a portion of the total industry. When more than 140 companies are included in the fit, the RMS values begin to systematically increase, which indicates divergence from power law behavior.

Therefore the fit line in Figure 5 is chosen to best fit the top 140 companies.

Figure 5 Internet Services Industry

Best Fit at (140, 0.152)

0.4

Source: CSFB analysis.

– 10 –

Figure 6 Power Law Segments

Internet 0.4 10 8

Market Value (000,000)

Rank Points in Test

Biological Technology 6 1.0 10

Market Value (000,000)

Rank Points in Test

Savings and Loans 10 5 0.25

Market Value (000,000)

Rank Points in Test

Source: CSFB analysis.

– 11 –

Source: CSFB analysis.

0 RMS

Source: CSFB analysis.

Copyright Credit Suisse First Boston, and its subsidiaries and affiliates, 2000. All rights reserved.

This report is not directed to, or intended for distribution to or use by, any person or entity who is a citizen or resident of or located in any locality, state, country or other jurisdiction where such distribution, publication, availability or use would be contrary to law or regulation or which would subject Credit Suisse First Boston or its subsidiaries or affiliates (collectively "CSFB") to any registration or licensing requirement within such jurisdiction. All material presented in this report, unless specifically indicated otherwise, is under copyright to CSFB. None of the material, nor its content, nor any copy of it, may be altered in any way, transmitted to, or distributed to any other party, without the prior express written permission of CSFB. All trademarks, service marks and logos used in this report are trademarks or service marks or registered trademarks or service marks of CSFB.

The information, tools and material presented in this report are provided to you for information purposes only and are not to be used or considered as an offer or the solicitation of an offer to sell or to buy or subscribe for securities or other financial instruments. CSFB has not taken any steps to ensure that the securities referred to in this report are suitable for any particular investor.

Information and opinions presented in this report have been obtained or derived from sources believed by CSFB to be reliable, but CSFB makes no representation as to their accuracy or completeness and CSFB accepts no liability for loss arising from the use of the material presented in this report where permitted by law and/or regulation. This report is not to be relied upon in substitution for the exercise of independent judgment. CSFB may have issued other reports that are inconsistent with, and reach different conclusions from, the information presented in this report. Those reports reflect the different assumptions, views and analytical methods of the analysts who prepared them.

CSFB may, to the extent permitted by law, participate or invest in financing transactions with the issuer(s) of the securities referred to in this report, perform services for or solicit business from such issuers, and/or have a position or effect transactions in the securities or options thereon. In addition, it may make markets in the securities mentioned in the material presented in this report. CSFB may, to the extent permitted by law, act upon or use the information or opinions presented herein, or the research or analysis on which they are based, before the material is published. CSFB may have, within the last three years, served as manager or co-manager of a public offering of securities for, or currently may make a primary market in issues of, any or all of the companies mentioned in this report. Additional information is available on request.

Past performance should not be taken as an indication or guarantee of future performance, and no representation or warranty, express or implied, is made regarding future performance. Information, opinions and estimates contained in this report reflect a judgment at its original date of publication by CSFB and are subject to change. The value and income of any of the securities or financial instruments mentioned in this report can fall as well as rise, and is subject to exchange rate fluctuation that may have a positive or adverse effect on the price or income of such securities or financial instruments. Investors in securities such as ADRs, the values of which are influenced by currency fluctuation, effectively assume this risk.

Structured securities are complex instruments, typically involve a high degree of risk and are intended for sale only to sophisticated investors who are capable of understanding and assuming the risks involved. The market value of any structured security may be affected by changes in economic, financial and political factors (including, but not limited to, spot and forward interest and exchange rates), time to maturity, market conditions and volatility, and the credit quality of any issuer or reference issuer. Any investor interested in purchasing a structured product should conduct its own investigation and analysis of the product and consult with its own professional advisers as to the risks involved in making such a purchase.

This report is distributed in Europe by Credit Suisse First Boston (Europe) Limited, which is regulated in the United Kingdom by The Securities and Futures Authority (“SFA”). It is not for distribution to private customers as defined by the rules of The SFA. This report is distributed in the United States by Credit Suisse First Boston Corporation; in Canada by Credit Suisse First Boston Securities Canada, Inc.; in Brazil by Banco de Investimentos Credit Suisse First Boston Garantia S.A.; in Japan by Credit Suisse First Boston Securities (Japan) Limited; elsewhere in Asia/Pacific by Credit Suisse First Boston (Hong Kong) Limited;. Credit Suisse First Boston Australia Equities Limited; Credit Suisse First Boston NZ Securities Limited, Credit Suisse First Boston Singapore Branch and elsewhere in the world by an authorized affiliate.

In jurisdictions where CSFB is not already registered or licensed to trade in securities, transactions will only be effected in accordance with applicable securities legislation, which will vary from jurisdiction to jurisdiction and may require that the trade be made in accordance with applicable exemptions from registration or licensing requirements. Non-U.S. customers wishing to effect a transaction should contact a CSFB entity in their local jurisdiction unless governing law permits otherwise. U.S. customers wishing to effect a transaction should do so only by contacting a representative at Credit Suisse First Boston Corporation in the U.S.

NI2571.doc

1 This was initially reported in “Growth Dynamics of the World-Wide Web.” Huberman, B.A., Adamic, L.

Nature Vol. 401, September 9, 1999. Since then the results have been reconfirmed in “Graph Structure in the Web.” Broder, Andrei et. al. IBM Almaden Research Center, San Jose, CA., June 2000.

2 ibid. Huberman. It’s important to note that the observed power law behavior was derived from usage logs for all Web sites, not usage logs for commercial Web sites alone. Corporate strategy can influence the ability to crack the line-up.

3 Six properties were absorbed by competitors that remained in the top 25 in April. The balance of the firms remained in the top 100.

4 Michael Mauboussin, Alexander Schay, Stephen Kawaja, “Absolute Power,” Credit Suisse First Boston Equity Research, December 21, 1999.

5 Although we could find no direct statistical correlation between site traffic and market capitalization, 14 of the top 25 companies in terms of site volume were also in the top 25 in terms of market value.

6 “Growth Dynamics of the World-Wide Web.” Huberman, B.A., Adamic, L. Nature Vol. 401, September 9, 1999 7 “Graph Structure in the Web.” Broder, Andrei et. al. IBM Almaden Research Center, San Jose, CA., June 2000 8 “Souped Up Search Engines.” Nature Vol. 405, May 11, 2000 www.nature.com 9 “Competitive Dynamics of Web Sites.” Maurer, Sebastian M., Huberman, Bernardo A. Xerox Palo Alto Research, March 17, 2000.

10 Ibid