When algorithms meet ECM - Google’s decision to opt for a rarely used modified or “dirty” Dutch auction to complete its US$1.67bn IPO in 2004 was far from an unreserved success but the idea retains a place, if a rarely accessed one, in the ECM toolkit.
Google’s decision to opt for a rarely used modified or “dirty” Dutch auction to complete its US$1.67bn IPO in 2004 was far from an unreserved success but the idea retains a place, if a rarely accessed one, in the ECM toolkit.
From the perspective of IPO investors, time has proven Google’s IPO to be a spectacular investment, delivering a 900% gain in the nine years since it went public as the online search firm has prospered. Yet the mechanisms of the auction, algorithmic in execution and rooted in academia, resulted in a less-than-stellar outcome on valuation and investor participation.
“[Google founders] Sergey Brin and Larry Page felt that if there was going to be a financing event, the Google users responsible for that growth should be able to participate in the transaction,” said Lise Buyer, founder of capital markets consultancy Class V Group and a former investment banker.
“Too many companies go public at [as an example] US$20, open at US$50, and that profit goes into the hands of people that had nothing to do with the development of the company,” said Buyer, who was hired by Google in late 2003 as director of business optimisation and worked closely on the IPO.
Setting initial price talk of US$108–$135 on what was originally a 24.6m-share offer, an intentionally high dollar share price, was designed to force retail investors to validate their interest. Investors, both retail and institutional, were required to register for bidder IDs with one of 28 banks in an underwriting syndicate led by Morgan Stanley and Credit Suisse First Boston.
Operating under the mantra “Don’t be evil”, Google management elected not to provide formal earnings guidance. Part of the rationale, recalled Buyer, was to avoid tripping securities laws that prohibited selective disclosure, given the high level of retail investors that would not be privy to the guidance. Another was that Google was growing so rapidly at the time that management lacked visibility to properly guide analysts of the underwriting firms.
The break from tradition did alienate some institutions. Retail demand was also muted by the decision not to offer a selling concession, a move that meant retail brokers of the underwriting firms had no incentive to push the deal to their clients.
The lack of subscriptions from both institutions and retail was evident in a reduction of the price talk to US$85–$95, and the subsequent decision by VC-backers and insiders to pare a secondary component from 10.5m shares to 5.5m shares, reducing the overall deal size to 19.6m shares – investors that had not submitted for bidder IDs from the outset were not allowed to do so after price talk was revised.
The net result was pricing at US$85.
Pricing was set on the discretion of the company and its underwriters, rather than at the marginal price needed to clear investor demand, with all investor orders scaled back on a pro rata basis. So, for example, an investor that submitted bids for 1m shares at US$100, 500,000 shares at US$95, and 500,000 shares at US$90 was scaled back to 1.6m shares if, for example, overall demand totalled 23.5m – 1.2 times coverage.
“The allocation process in a true auction is typically just a mathematical exercise, so the real benefit of an auction is the ability to create a perfect demand curve, often down to penny increments as it relates to price,” said Cully Davis, a Credit Suisse managing director specialising in tech ECM.
One result of such a scientific approach, combined with the reluctance to provide forward guidance, was that it took several quarters to on-board institutions into the company’s registry, though Google returned with follow-on sales of stock in September 2005 and March 2006 at US$295 and US$389.75, respectively.
Despite what many viewed as a flawed outcome on Google, and limited applicability subsequently, auction IPOs do have a future role. Credit Suisse rolled out its auction platform for NetSuite’s IPO in 2007, though the exercise was more designed to maximise pricing, and was mandated by Liquidnet, alongside Goldman Sachs, in 2008 on what was to have been a hybrid auction, though the deal never came to fruition.
Hybrid auctions, where there is a dedicated retail tranche but pricing is dictated by the larger, institutional pot, is gaining traction as a way to democratise participation but with a more sophisticated approach toward price discovery. DePaul University finance professor Ann Sherman, an adviser to Google on its IPO, has suggested that Twitter adopt a hybrid auction for its forthcoming IPO.
The reality is that Google offered a rare crossover of high-growth and brand awareness that suggests auction IPOs are episodic events. Lise Buyer argues that auctions may be appropriate on high-profile, wide brand-name recognition deals where there is unlimited demand from unsophisticated investors.
“One of the biggest misconceptions about an IPO is that everyone is trying to accomplish the same thing,” Buyer says. “For some companies, but not most companies, there is an alternative to the traditional bookbuild IPO.”