TL;DR
This paper argues that traditional auctions like English or simultaneous auctions create significant inefficiencies when buyers need to invest in costly information gathering before making a purchase. The Dutch auction, however, avoids these inefficiencies by guaranteeing a purchase price at the time of information acquisition, leading to better search coordination and potentially larger gains from trade.
What Problem Does This Paper Address?
When buying something complex or uncertain, like a startup company or a timber tract, buyers often need to do extensive "due diligence" – a costly and time-consuming information-gathering process. Traditional auction formats, such as English (ascending price) or simultaneous auctions, add another layer of uncertainty: buyers don't know the final price they'll have to pay until the auction concludes. This uncertainty can lead to two major problems:
- Wasteful Search: Buyers might spend time and money investigating items that are already out of reach or have already been claimed by someone else.
- Discouraged Search: The risk of wasted effort might deter buyers from conducting necessary investigations altogether, potentially preventing mutually beneficial trades from happening.
While there are theoretically efficient auction mechanisms (like dynamic Vickrey-Clarke-Groves), they are often too complex and impractical for real-world use. The paper asks: can a simpler, commonly used auction format like the Dutch auction (descending price) offer a practical solution to these search coordination problems?
Methods
The core of the paper's argument revolves around the concept of a "strike price" and "covered call value." For any bidder, there's a strike price at which it becomes just worthwhile to investigate the item. By selling a "call option" at this strike price, a bidder can finance their information gathering without bearing the risk of wasted costs. The paper then analyzes how different auction formats handle this "real option" aspect of information acquisition.
The authors formally model this using a "real option" framework, drawing on work by Dixit and Pindyck. They then demonstrate that the Dutch auction, by its nature of offering a descending price, allows bidders to know the maximum price they'll pay at any given moment. This guarantee enables rational planning for information acquisition and effectively allows bidders to "exercise in the money" – meaning they only proceed with costly investigation if they are confident they can secure the item at a price that justifies the investment.
The paper uses theoretical analysis, including proofs and lemmas, to establish the equivalence between Dutch auctions with information costs and equivalent auctions without information costs but with modified bidder values (covered call values). They also employ numerical calibrations for specific markets (start-up acquisitions and timber auctions) to quantify the welfare losses and revenue implications of different auction formats.
Datasets
Not discussed in the paper.
Key Insights and Ideas
- Dutch Auction as a "Real Option" Manager: The Dutch auction's descending price mechanism inherently manages the "real option" of information acquisition. Bidders know the maximum price they'll pay, allowing them to rationally decide whether to invest in costly search.
- "Exercising in the Money" is Key: The Dutch auction's guarantee allows bidders to "exercise in the money," meaning they only incur search costs if they are likely to secure the item at a price that justifies the investigation. This is crucial for efficient search.
- Standard Auctions Fail at Search Coordination: Ascending and simultaneous auctions fail to coordinate search effectively because bidders don't have price certainty when they need to decide whether to investigate. This can lead to significant welfare losses.
- Covered Call Value Equivalence: The paper shows that the welfare properties of a Dutch auction with information costs are equivalent to an auction without information costs but where bidders' values are replaced by their "covered call values." This simplifies analysis and highlights the Dutch auction's strengths.
- Negligible Asymmetry Losses: While bidder asymmetries can be a concern in standard auctions, the paper finds that these losses are negligible in the Dutch auction context, especially compared to the gains from improved search coordination.
Key Results
- Dutch Auction Efficiency: The Dutch auction is shown to be highly efficient in coordinating search when information acquisition is costly, performing as well as an auction without information costs.
- Standard Auction Inefficiencies: Simultaneous and English auctions can lead to unboundedly small welfare fractions of the optimal outcome in certain search-intensive scenarios.
- Calibrated Losses: In start-up acquisition and timber auctions, standard formats incur welfare losses of a few percent due to poor search coordination, while the Dutch auction's losses are typically less than 1%.
- Revenue Gains: The Dutch auction often achieves significantly higher revenue than second-price auctions, especially in scenarios with high inspection costs.
Limitations and Future Work
- Instantaneous Inspection: The baseline model assumes information acquisition is instantaneous. The authors acknowledge that in reality, this process takes time and has implications for sequencing and market design, suggesting this as a direction for future research.
- Exogenous Information: The model assumes information arrives only endogenously through costly search. Future work could explore markets where information also arrives exogenously.
- Market Design Extensions: The paper suggests that the insights about descending prices and search coordination could extend to matching markets and other allocation problems where transfers are not available.
- Common Values: While the paper briefly touches on common values, a more thorough analysis could be beneficial.
- Posted Pricing: The paper notes that posted pricing mechanisms can also avoid the "exposure problem" but sacrifice price discovery benefits, suggesting further exploration of hybrid mechanisms.
Summary updated