Procurement Auction Formats English, Dutch, Anglo-Dutch and Sealed-Bid

Auction designs fall into two broad families.

• Dynamic auctions play out over time. Price moves in known steps and suppliers react round by round, learning from the fact that the auction is still running.
• Sealed-bid auctions happen in a single round. Each supplier submits one best-and-final offer, the bids are opened, and the lowest bid wins.

One point trips up newcomers: procurement auctions run in reverse of the sales-room versions. In a procurement English auction the price falls as suppliers compete to undercut each other, while in a procurement Dutch auction the price rises until a supplier accepts. The same toolkit works whether you are running a buy-side tender or a sell-side divestiture — only the direction of the price changes.

Reverse English (descending price). The auctioneer opens with a high starting (reserve) price set deliberately above expected cost, then lowers it in known decrements. Suppliers signal at each step that they want to stay in and drop out once the price falls below their cost. The price keeps falling until the second-to-last supplier exits; the last one standing wins, at the price where that second-to-last supplier dropped. Because the runner-up effectively sets the award price, this format carries a "second-price" character — and it only works with genuine competition, since a falling price itself signals that at least two firms are still bidding.

Reverse Dutch (ascending price). The clock starts at a price so low that no supplier will accept, then rises in known increments until the first supplier accepts the contract. That supplier wins and is paid the price it accepted — a first-price logic. The Dutch design is vulnerable to ties if the opening price is set too high and several suppliers accept at once. Resolving ties by "first to click" is poor practice: it adds time pressure and can let a higher-cost supplier win. Better to pre-rank bidders in an earlier phase, request best-and-final offers below the tied price, or fall back to a random draw. The practical lesson is to start low.

The Hong Kong variant. This hybrid follows the English price path (descending) but ends only when the very last bidder drops out, not the second-to-last. The winner can keep undercutting itself even after its rivals are gone. The price movement looks English, but the strategic logic is Dutch: because bidders cannot tell when they are alone, they cannot tell when they have already won. It only works if the auctioneer hides how many competitors remain.

The information rule. In practice the auctioneer knows more than the bidders and usually keeps it that way. Research generally finds that disclosing less yields better outcomes, with one exception: when supplier costs are correlated, revealing some competitive information can give bidders the confidence to bid more aggressively. Standard practice is therefore to withhold the number of competitors, the indicative offers (which can seed collusion and reveal reserve prices) and prior bid history. When the FCC's spectrum auctions once disclosed who was bidding on what, the result was tacit collusion; from Auction 66 onward it released only the number of new bids per round. The same caution applies in procurement: full transparency undermines competition.

Choosing a dynamic design comes down to three levers — whether price rises or falls, the information rule, and the starting price — and all three hinge on expected competition. With many suppliers expected, a descending English auction works well and the second-lowest-cost supplier disciplines the price. With thin competition, an ascending Dutch auction or a Hong Kong auction is safer, because there the winner's margin is shaped by expected rather than actual competition.

Sealed-bid auctions are simpler and faster: one round, best-and-final offers, lowest bid wins. They come in two variants that differ only in what the winner is paid.

First-price. The lowest bidder wins and is paid its own bid. Simple to run, but it forces strategic reasoning: a lower bid raises the chance of winning while shrinking the margin, so each supplier must balance the two. This is the dominant form of sealed-bid procurement.

Second-price (Vickrey). The lowest bidder still wins, but is paid the second-lowest bid. Because raising your own bid cannot change what you are paid if you win, the dominant strategy is simply to bid your true cost. The format is "truth-inducing" and reaches the same price outcome as the English auction without the live bidding.

Despite its elegance, the second-price auction is rarely used in procurement, for two reasons. The first is optics: if the winning bid is far below the second-lowest, the process can look as though it failed to extract a competitive price. When New Zealand sold UHF licences this way in 1990, one broadcaster's winning bids totalled NZ$6.9m but it paid only NZ$1.2m — efficient in theory, politically toxic in practice. The second is the incentive to cheat: a private buyer can simply understate the second-lowest bid. If the winner bid $200,000 and the runner-up $290,000, claiming the runner-up was $215,000 pockets the buyer $75,000 and the winner still takes the deal. First-price auctions avoid both problems — the buyer has no reason to manipulate, and paying the winner its own stated bid verifies itself.

In his 1961 paper, Vickrey showed that under a tight set of assumptions the first-price, second-price, English and Dutch auctions all produce the same expected cost. Translated to procurement, this is the Cost Equivalence Theorem: if the assumptions hold, the format is neutral and only efficiency matters. But the assumptions are demanding — a single contract, risk-neutral parties, independently drawn supplier costs, no capacity constraints, and a fixed, costless field of bidders.

In the real world those conditions rarely all hold, and that is precisely why the format matters. When suppliers are capacity-constrained, risk-averse, or simply smaller firms without deep credit lines, they value the price certainty of the first-price and Dutch formats, where they lock in their margin before the auction resolves. The second-price and English formats, by contrast, let the losing bid set the winner's price, which could leave almost no margin. Risk-averse buyers working to a fixed budget tend to agree. The English auction earns its place in the opposite case: a large, complex project with only a few capable firms whose costs are correlated, where live bidding reduces uncertainty and draws out aggressive bids.

A vivid illustration comes from the adapted Mona Lisa case in the paper. Acting as a proxy with a $1.5bn ceiling against two rivals valuing the painting at $1.0bn and $0.9bn, an English (ascending, in the sales direction) auction settles at $1.1bn — the runner-up's exit point. A Dutch auction ticking down from a high start, however, pushes the proxy to accept at around $1.4bn rather than risk losing the painting altogether: roughly 27% higher. The lesson carries directly into procurement. When bidders are few and their costs differ widely, the Dutch format prices off expected rather than realised competition and leans on suppliers' risk aversion. In a sale that means a higher price for the seller; in a procurement tender it means suppliers accept earlier, at a lower price than open bidding would have forced — a better result for the buyer.

Few real tenders use a textbook format unaltered. Two hybrids deserve a permanent place in the procurement toolkit.

The Anglo-Dutch auction (Klemperer) bolts the two dynamic formats together. An English phase first lets information flow, lets bidders gauge their competitive position, and produces a ranking; a Dutch phase then leans on risk aversion to squeeze out competitive pricing. It is demanding — two live phases — so a lighter alternative replaces the English phase with a first-price sealed-bid round that both ranks bidders and tests their prices under competition, giving the Dutch phase a credible starting point.

Negotiauctions (Subramanian) blend rule-based bidding with classic negotiation. They earn their keep when a pure auction would deter participation — strong suppliers who can walk away, or weak ones who see no path to winning — and when there is room for win-win moves like value engineering that lower the buyer's price without crushing the supplier's margin. A common pattern is the take-it-or-best-offer: suppliers are ranked using the comparison price approach, the buyer makes the top-ranked supplier a firm offer, and if it declines it submits its best counter before the buyer moves down the ranking. As with any auction, this only works if the buyer has secured internal commitment to honour the offer. The same design discipline extends to how the work is split into lots, which shapes how many suppliers can compete in the first place.

Which auction format is best for procurement? There is no single answer — the right format depends on your supplier pool. For a field that mixes large and small firms, a first-price sealed-bid or Dutch auction is usually the robust choice, because risk-averse and capacity-constrained suppliers value the margin certainty. A descending English auction shines when only a few capable firms compete on a large, complex project. The second-price auction is a niche tool, sensible mainly when many similar suppliers compete for several contracts at once.

Why is the second-price (Vickrey) auction so rarely used? Two reasons. It can look uncompetitive when the winning bid sits far below the second-lowest, and it hands a private buyer a clear incentive to understate the runner-up's bid. First-price designs avoid both, which is why they dominate sealed-bid procurement.

What is a negotiauction, and when should we use it? It is a controlled blend of negotiation and auction. Reach for it when a strict auction would scare off the suppliers you need, or when joint cost-reduction work can expand the pie rather than just split it.

Competitio designs and runs these mechanisms in live tenders — more than €35bn of negotiated volume across 16 industries, supported by a scientific advisory board that includes Prof. Dr. Christian Rieck. To pressure-test which format fits your next tender, talk to our team.