Editor’s note: This is general educational information about how auction pricing works in a public offering. It is not investment advice, and it does not describe any current offering. Everything below is drawn from the official filings and rules listed at the end.
Most initial public offerings are priced in a negotiation that the buyers never see. A syndicate of banks collects soft indications of interest from institutions, forms a view of demand, and then sets one price and decides who gets stock. An auction offering replaces that negotiation with a rule. Investors submit binding bids stating a quantity and a maximum price, the bids are ranked, and the price is read off the point where the quantity demanded matches the quantity for sale. The best documented example of the structure in a large equity offering remains the prospectus Google Inc. filed with the Securities and Exchange Commission on August 18, 2004, which describes the auction it ran in unusual mechanical detail.
How the bids turn into a single price
The starting point is a sealed bid rather than an indication of interest. In the Google offering the minimum bid was five shares, and every bidder had to obtain a bidder identifier before submitting a quantity and a price per share through a participating underwriter. Bids could be withdrawn until they were accepted, and once an underwriter sent an electronic notice of acceptance the bid could not be cancelled or rejected.
When bidding closes, the bids are sorted from the highest price down and the quantities are accumulated. The prospectus defines the resulting number precisely: the auction clearing price is the highest price at which all of the shares offered, including shares subject to the underwriters’ over-allotment option, may be sold to potential investors. Bids below that level receive nothing. Bids at or above it are successful, and they all pay the same amount, whatever price each of them actually wrote down. A bidder who was willing to pay well above the clearing price does not pay a penny more than the bidder who scraped in at the line.
Allocation is then a separate step, and it is deliberately blind to how aggressive a bid was. Google’s prospectus states that the allocation process will not give any preference to successful bids based on bid price, and describes two mechanical methods. Pro rata allocation divides the shares offered by the shares represented by successful bids and applies that percentage to every successful bidder: in the filing’s own worked example, 20,000 shares offered against 21,200 shares of successful bids gives an allocation percentage of 94.34 percent, applied to each bidder in turn. Maximum share allocation instead fills small bids completely and caps large ones at whatever ceiling exhausts the offering. The company said its objective was an offering price at which successful bidders receive at least 80 percent of the shares they bid for, and that it did not intend to disclose which of the two methods it used.
Where discretion survives
An auction narrows discretion; it does not remove it. Google reserved the right to set the offering price below the auction clearing price, explaining that it might do so to achieve a broader distribution of the stock or to reduce downward volatility in the days after listing, while warning that neither result was assured. The final price was $85.00 a share on 19,605,052 shares, of which the company sold 14,142,135 and selling stockholders 5,462,917, with underwriting discounts of $2.3839 a share against $82.6161 of proceeds.
Discretion also survives in who is allowed to bid and in what happens to bids the issuer treats as manipulative or disruptive under the standard it set for itself. The prospectus reserved the right to treat a bid as manipulative or disruptive if the company believed it did not reflect the number of shares the bidder actually intended to buy, to reject every bid that bidder had submitted, and not to inform the bidder that the rejection had occurred. Bidder suitability remained with each underwriter, so eligibility to participate still ran through an intermediary relationship. The pricing rule was mechanical; entry to the mechanism was not.
The government auction the equity version borrows from
The single price auction was a government bond technique long before it was an equity technique, and the rule that defines it is written into United States regulation. Part 356 of Title 31 of the Code of Federal Regulations defines a single-price auction as one in which all successful bidders pay the same price regardless of the yields, discount rates, or discount margins they each bid. The Treasury describes the same sequence its rules require: it accepts all compliant non-competitive bids first, then works through the competitive bids from the lowest rate or yield upward until the offering is fully awarded, and every successful bidder receives the rate, yield, or discount margin of the highest accepted bid.
Two features of the Treasury version have no counterpart in the equity auctions. Non-competitive bidders, who accept whatever the auction produces, are limited to $10 million and are barred from bidding competitively in the same auction. And the rules impose a net long position reporting threshold, typically 35 percent of the offering amount, so that a bidder who already holds a large when-issued or futures exposure to the security has to disclose it. Both rules exist because the same buyers return to the same auctions on a published calendar, which is not the position an issuer selling stock once ever occupies.
Analysis: what the mechanism decides, and what it leaves open
The auction settles two questions that book building answers privately: what price clears, and who gets stock at it. Because allocation follows a published formula, an underwriter cannot reward a favoured account with a larger fill, and because every winner pays the clearing price, a high bid buys priority rather than a worse fill price. That is the whole of the structural change. It does not make the price more accurate, and Google’s own risk factors said so in plain terms, warning that the offering price might have little or no relationship to traditional indicators of value such as sales, earnings, comparable company multiples, or research analyst views, and describing the winner’s curse that successful bidders face when they infer there is little incremental demand above the price they paid.
The comparison with the Treasury auctions is where the limits show. A single price auction works smoothly for Treasury securities because the instrument is standardised, the calendar is known months ahead, and a professional bidding base returns week after week, which is why the rules can afford to be purely mechanical and still police concentration through position reporting. An equity offering is a single event with a bidder base assembled once, no history of comparable auctions in the same security, and no obvious way to distinguish a serious institutional bid from a speculative one except through the discretionary anti-manipulation power the issuer reserved.
For a reader in Tel Aviv the relevant point is comparative rather than local. The Tel Aviv Stock Exchange is Israel’s sole stock exchange, and its TA-35 and TA-125 indices rose 51.6 percent and 52.0 percent in 2025, so pricing new issues into a market that has repriced that far in a year is a live question for any issuer. What the documents cited here establish is narrow and useful: an auction is a rule for converting bids into one price and a fixed allocation, applied to whoever the underwriters let bid. What they do not establish is that the rule produces a better price, and the filings that describe the mechanism most carefully are also the ones that say so.
