Publication of KISDI basic research (21-09) ‘A study of auctioning methods that take into account sunk cost effects’

“Sunk costs produce different effects in an auction situation, depending on the impact factors.”
.................................................................................................................
Auction testing in a situation involving sunk costs produced the following results: The sunk cost amplification effect based on responsibility was found to be insignificant, with ‘observability’ amplifying the sunk cost effect

An empirical analysis of the impact of sunk cost related factors (responsibility, observability, budgets) on auctions was carried out.

▲ Responsibility brings a positive effect to the sunk cost, but the effect is not statistically significant.
▲ Both ‘observability’ and ‘budget’ had the effect of amplifying the sunk cost effect.
▲ It will be necessary to take observability and budget into account in the future in auctions involving sunk costs.

KISDI (President Kwon Ho-yeol) recently published KISDI basic research (21-09): ‘A study on auctioning methods that take into account sunk cost effects’.

As a means of efficiently distributing limited resources, auctions are widely used in such situations as frequency allocation, power distribution, and the sale of treasury bonds. Using experimental methodologies, KISDI conducted a study on how sunk costs affect auctions.

The term “sunk cost” refers to costs incurred in the past that are difficult to recover, while “sunk cost effect” refers to the phenomenon whereby the larger the upfront investment is for a particular project or the purchase cost is for a particular product, the greater the follow-up investment required. A great example is the Franco-British joint project to develop Concorde. During the supersonic aircraft’s development, the estimated development cost was estimated to eventually exceed the expected revenues from the sale of the plane. Despite this, both countries plowed ahead with the development, securing additional funds to complete the project. There are also other instances of how sunk costs could impact a project in a competition situation, as in the case of mobile communication bandwidth auctions involving mobile service providers who already own the right to use the same or a similar bandwidth, or real estate auctions involving operators who own land in a specific area.

This report aims to determine the role of sunk costs in a competitive situation like an auction. It contains the results of a study which investigated the kinds of effects produced when variables (responsibility, observability) that are highly related to sunk-cost effects are at work together in a competition situation like an auction. For this purpose, a two-stage auction scenario was designed in which sunk costs are produced and an experiment was conducted, wherein 120 ordinary citizens were invited to participate in a virtual auction that simulated a bandwidth auction environment, and the bidders’ behavior when under the influence of the factors and the sunk cost effect was analyzed. The key findings of the study are summarized below.

First, it was discovered that the sunk cost effect influenced the behavior of the bidders to a statistically significant extent. In other words, bidders who invested heavily to win the first stage of the auction also tended to submit a higher bid at the second stage for the goods they had won in the first stage of the auction.

Second, it was observed that the assignment of responsibility had a positive effect on the occurrence of the sunk cost effect, although this effect was not statistically significant. Here, “responsibility” refers to whether the process leading up to winning the good in the first stage of the auction is based on the bidder’s own decision or an arbitrarily determined result. The assumption was made that if, during the first stage of the auction, the payment for the good was the winning bid price that one had directly decided personally, then in the second stage, the payment would be seen as a sunk cost for which one is responsible. However, that would not be case if one did not decide the bid price and the willingness to take responsibility becomes diminished. According to this analysis, when responsibility is assigned, the sunk cost effects appear higher than when responsibility is not assigned; but the effect is not statistically significant.

Third, it was discovered that, based on whether observability is granted, the bidding behavior of the bidders could be affected in a statistically significant way. Observability refers to the ability to see at what price another bidder wins the right to purchase a particular good during the auction process. If observability is granted, competition between the bidders heats up and the bid price is shown to go up as a result. Furthermore, granting observability also leads to statistically significant differences in the sunk cost effects. This implies that observability is a factor that amplifies the sunk cost effect even more than responsibility. In addition, the difference in budgets was also determined to be a variable that could amplify the sunk cost effect of the bidder.

In conclusion, the higher the sunk cost injected, the higher the amount of additional investment made by the bidder for the same asset. The amount of budget held by the bidder, and observability among the bidders were found to be factors that could amplify the sunk cost effect. On the other hand, responsibility granted to the bidder did not produce any statistically significant effect if measured separately from the observability variable.

This study confirmed through experimental studies how the sunk cost effect will affect when two or more complex variables act in a competitive situation such as an auction. This study verifies through experiment how the sunk cost effects are impacted when more than two comprehensive variables are at work in a competition situation like an auction. This study differs from existing studies in that it verifies through experiment on ordinary citizens how the sunk cost effect is impacted when the responsibility variable and the observability variable act independently of each other.

Research fellow Kim Hee-chun remarked, “Through this study, we have empirically analyzed the factors (and their effects) that impact the decision-making behavior of bidders in an environment where sunk costs have been incurred. By empirically observing the factors that affect the behavior of bidders and the size of the effects in an environment that replicates the uniqueness of auctions, it will be possible to improve the efficiency of transactions in the future because, based on our observational findings, auctions could be designed better.”