While a central coordinated auction or exchange is of great value, the potential leakages of private information may be costly. The traditional approach is to appoint well-reputed and trusted institutions or individuals to safeguard the private information. Still, individuals are granted full access to the information, an approach that is not only prone for human mistake but also provides a focal point for collusive behavior *. With Partisia software no such focal point exists and the private bids are kept encrypted forever - this is what we call deep confidentiality. Read more about the Technology >> and how it allows us to address Collusion >>.

On this page we discuss some reasons for having deep confidentiality *:

• Keep positions in related negotiations
• Avoid aggregation of private values
• Basic reluctance to share private information

Keeping positions in related negotiations

The private bids in an auction or exchange may contain information that can be used by opponents in related negotiations or by authorities in a regulatory setting.

An example is provided in the Danish Sugar industry, which uses Partisia Contract Exchange to reallocate production contracts among sugar beet growers. The contracts are regularly renegotiated among multiple types of growers and the sugar-processing company Danisco. Since the bids and asks on the exchange provide valuable information about the growers’ willingness to pay * or accept *, insufficient confidentiality may weaken the growers’ positions in these negotiations.

Another example could be procurement auctions. Sellers often have multiple contracts with the buyer, say a municipality, and the terms of these contracts are gradually adjusted. To impact this ongoing adjustment, a seller may therefore manipulate his bids on new contracts.

With deep confidentiality the participant does not weaken his or her position in related situations, no matter how the future is going to turn out.

Avoid aggregation of private values

Aggregation of private bids across time and auctions can generate valuable information about the participants' preferences. Such information may for example be used to exercise more price discrimination at the cost of the participants.

Computerized trading agents have been used for decades not least in the financial sector. Today, many Internet auctions, such as E-bay, provide so-called proxy bidding, where e.g. a buyer provides a price ceiling and allows a computer to overbid others within that ceiling.

It is extremely important to safeguard the private information feed to such trading agents. This was early acknowledged by Professor Hal Varian:

“Hence privacy appears to be a critical problem for ‘computerized purchasing agents’. This consideration usually does not arise with purely human participants, since it is generally thought that they can keep their private values secret. Even if current information can be safeguarded, records of past behavior can be extremely valuable, since historical data can be used to estimate willingness to pay. What should be the technological and social safeguards to deal with this problem?"

Since the date of this quote (1995), the concern has only increased. The use of auctions in general as well as more advanced auctions has increased. Advanced auctions may include multiple commodities or multiple attributes and extract more information from the participants.

Deep confidentiality effectively eliminates these problems since non-winning bids remain secret forever.

Basic reluctance to share private information

Apart from explicit, rational arguments, there might be a general reluctance to share private information.

Private information revelation is a point of no return and it may be hard to foresee future use/misuse of the revealed information. This provides an implicit rationale for the basic reluctance. There may also be more subtle psychological reasons behind a basic reluctance.

With deep confidentiality any reluctance to participate with the required private information is removed and better results are promoted.

References

• Varian, Hal, 1995, Economic mechanism design for computerized agents, In USENIX workshop on Electronic Commerce, p. 13-21