Wednesday, February 26, 2014

Tonight We're Going to Party Like It's 1999

(originally published on March 27, 2011 on the Launching Tech Ventures course blog)

Are we in a new bubble, as Steve Blank recently wrote, or do current high valuations for early- and late-stage consumer Internet companies reflect sound fundamentals, as argued by Ben Horowitz?

From an academic's perspective, this is a difficult question, and I won't tackle it here. Instead, I'll share some data on the performance of Internet startups launched during in the late-1990s boom.

The table below compares the market value at the end of 2001—the trough of the valuation cycle that began in the mid-1990s—to total capital raised since inception (private and public) for all 2,121 U.S-based Internet companies that had ever got funding from VCs or public markets (see appendix below for my definitions and methods). The firms had an aggregate market value of $99 billion at the end of 2001, and they had raised $85 billion of capital. This doesn't imply an attractive return for someone who invested pro rata in all rounds, especially if you consider the time value of money and the fact that investors had to share part of the $99 billion with company founders. But it's not a wipe-out, either.

Of course, focusing on sector-level, aggregate results may mask serious problems with overinvestment in individual markets, for example, online pet supply retailing—where four rivals burned through over $500 million and all failed.  Also, it's important to note that returns from individual companies were very skewed. As the table below indicates, only 139 (7%) of the 2,121 Internet companies completed an IPO. Among these public companies, only 37 had a market value at the end of 2001 that exceed the total capital they had raised historically. A full 42% of the sector's $99 billion in yearend 2001 market value was accounted for by just five companies: Amazon, eBay, E*Trade, Yahoo!, and WebMD.

In summary, if the Internet sector had been subject to grossly excessive rates of entry and investment during the last bubble, as conventional wisdom holds, we would have expected a significant sector-level capital loss, rather than a shoulder shrug, "got-my-money-back" aggregate return. Conventional wisdom exaggerates the economic damage wrought by the late-1990s bubble. In considering the impact of valuation bubbles, it's important to separate stock market gains and losses—transfers of wealth between traders—from poor long-term returns on the capital invested directly in companies. It is also important not to conflate the high business failure rates that result from excessive rates of entry with the high failure rates we would normally expect to observe in new markets with “winners-take-most” potential and low entry barriers—like those targeted by many Internet companies. Given lottery-style payoffs, high rates of entry and failure are consistent with rational economic behavior.

As a new bubble emerges around social media startups, we should expect to see similar performance patterns.

APPENDIX


Definitions. The table above includes all U.S.-based Internet companies that raised capital from professional investors prior to 2002. Internet companies were defined as firms that relied on the Web as their principal channel for delivering products or services to consumers or businesses. This definition excludes: 1) companies that earned most of their revenue by providing professional services, software, or hardware to Internet companies; 2) firms that provided Internet access or hosting services; and 3) the online units of established, brick-and-mortar corporations. Professional investors were defined as venture capital firms (including corporate venture funds) and institutions that purchased public securities. Firms funded exclusively by angels or by strategic investments from non-Internet companies were excluded, although strategic investments were included in estimates of total capital raised by VC-backed firms.

Valuations. Enterprise values were estimated as of December 31, 2001. Valuations for private Internet companies reflect only the market value of equity; few such firms had any debt. Enterprise values for public firms reflect the market value of their equity as of yearend 2001, plus the book value of any debt.

For active, private Internet companies that raised funds sometime during 2001, yearend 2001 equity market value was assumed to equal 98% of total capital raised historically, based on average post-round valuations for funding transactions completed during the second half of 2001. For active private firms that did not raise funds during 2001, yearend 2001 equity market value was assumed to equal 49% of total capital raised historically, based on analysis of performance for VC funds launched during 1999.

For merged firms, valuations reflect their contribution to the yearend 2001 market values of the companies that acquired them, rather than proceeds realized by the merged firms’ shareholders. 
For mergers involving two Internet companies, the acquired company’s value is included in the acquiring company’s yearend 2001 market capitalization. Capital raised by the acquired company is added to the capital raised by the acquiring firm. 

For acquisitions of Internet companies by non-Internet companies completed during 2001, yearend 2001 market value was assumed to equal announced merger transaction proceeds, which averaged 76% of total equity capital raised historically by the acquired companies. For such firms acquired prior to 2001, yearend 2001 market value was conservatively assumed to equal the total capital they raised historically.  

Source: Eisenmann, "Valuation bubbles and broadband deployment," Ch. 4 in The Broadband Explosion, edited by Austin & Bradley, HBS Press, 2005