Going, going, gone

Luther Martin of Voltage Security looks at a parallel between the market for IT and a Dutch auction, which may explain some IT project failures.

The £23.6 billion ($44.3 billion) of transactions in 2005 on eBay shows that auctions have become a popular way of selling goods.

Auctions have, of course, been around for thousands of years. One of the most notable auctions of all time took place in AD 193 when the Praetorian Guard auctioned the entire Roman Empire to the highest bidder.

The winner of this auction, Didius Julianus, offered each soldier 25,000 sesterces, or 10 times their annual salary, and became the next emperor only to be overthrown and slain by Septimus Severus 66 days later.

It seems likely that toward the end of his life that Didius Julianus came to regret his purchase, and felt what is commonly known as buyer's remorse.

The consequences of winning most auctions are not usually as severe as those suffered by Didius Julianus, but economists tell us that we should expect buyer's remorse to be fairly common because we can expect auction winners to pay too much.

Their reason for believing this assumes that bidders in an auction will not know the exact value of what they are bidding on, so that the bidder who overestimates this value the most will end up winning the auction and suffering the winner's curse of having paid too much for their purchase.

The lower-than-expected returns earned by winners of auctions for oil drilling rights or wireless spectrum licenses seem to provide evidence that this does indeed happen.

Economists call the familiar 'going, going, gone' auction that is used by Christie's and Sotheby's an English auction. In this type of auction, bids increase until only one bidder remains.

Another common type of auction is the Dutch auction, which is named after the way in which flowers are sold in The Netherlands. In this type of auction, the price starts high and is progressively lowered until a buyer is found.

Dutch auctions are also used by the Federal Reserve Bank of New York to sell options on overnight repurchase agreements, and were used to sell Google shares in Google's initial public offering.

Although the analogy is not perfect, there is a parallel between the market for information technology (IT) and a Dutch auction: new technology is usually introduced at a high price, but drops over time, just like the price of an item sold in a Dutch auction. At some point the price of the technology may become low enough so that some firms can justify its purchase.

To consumers of IT, this looks much like a Dutch auction, except for the fact that prices may continue to drop after a purchase is made.

So if purchasing information technology is like a Dutch auction we can expect the winner's curse to affect IT purchases and expect that many firms will pay more than they should for IT because they overestimate the return on investment that they use to justify its purchase.

Early adopters of technology seem particularly prone to this problem because they tend to pay higher prices for technology than others who wait until the technology drops in price.

The overestimation of the value of IT purchases may also be caused by problems in the deployment of the technology. The Standish Group, a US consultancy that specializes in tracking the rates of failure in IT projects, estimates that the chance of a trouble-free completion of any IT project is small.

Their CHAOS Report tracks the state of IT implementations, and in 2004 estimated that the chances for a trouble-free deployment ranged from roughly 2 per cent for larger projects to roughly 46 per cent for smaller ones.

This report also found that while only 15 per cent of IT projects resulted in total failure, higher-than-expected costs were common, deployments often took longer than expected and often resulted in fewer capabilities than first planned. Our understanding of auctions may provide an explanation of why so many projects end up troubled.

Firms tend to allocate funding to the projects that have the highest return on investment. So we can think of different projects as bidders in an auction, with funds going to the projects that estimate the highest rate of return on the investment that they require. This means that firms will tend to fund projects for which they have overestimate their value the most.

Similarly, if a firm uses a system integrator for an IT implementation, the contract for the project is usually awarded to the lowest bidder. The lowest bidder tends to be the one who underestimated the true costs of a project the most, which then tends to result in projects with difficulties with costs and schedules once the inaccuracy of the estimates is discovered.

Understanding why some IT projects may result in difficulties can provide insight into ways to address the problem. If you are planning to deploy a relatively new technology, you should carefully consider your business case for adopting the technology. In some cases the benefits will be clear and you should proceed with the deployment of the new technology.

In other cases you may find that unrealistic expectations of the benefits of the new technology have led you to a bad decision and that you are on your way to experiencing the winner's curse as you pay too much for it.

The safe strategy is to assume that you have overestimated the value of the new technology and to revise your projections downward to compensate for this to avoid a possible case of the winner's curse.

Similarly, you should ensure that all of your IT projects will have an appropriate return on investment in the event that its deployment encounters difficulties.

Costs will often be higher than first anticipated, schedules will often slip and deployed technologies will often not offer all of the features that you had anticipated.

Most projects will encounter some of these difficulties, and being prepared for this will let you increase your chances of success and avoid feeling buyer's remorse for your IT investments. Again, the safe strategy is to assume that costs and schedules have been underestimated and to plan accordingly.

November 2006