Mahsa Derakhshan,Emily Ryu,S. Matthew Weinberg,Eric Xue
Abstract:The competition complexity of an auction setting is the number of additional bidders needed such that the simple mechanism of selling items separately (with additional bidders) achieves greater revenue than the optimal but complex (randomized, prior-dependent, Bayesian-truthful) optimal mechanism without the additional bidders. Our main result settles the competition complexity of $n$ bidders with additive values over $m < n$ independent items at $\Theta(\sqrt{nm})$. The $O(\sqrt{nm})$ upper bound is due to [BW19], and our main result improves the prior lower bound of $\Omega(\ln n)$ to $\Omega(\sqrt{nm})$.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to determine the competition complexity in the auction setting. Specifically, the author studies in auctions how many additional bidders are required in order for simple auction mechanisms (such as selling items separately) to generate more revenue than complex optimal auction mechanisms (such as randomized, Bayesian - truthful mechanisms that depend on prior distributions). The main result of this paper addresses the competition complexity problem when bidders' valuations are additive and items are independent.
### Research Background and Motivation
In the field of multi - dimensional mechanism design, especially in computer science, Myerson's classic work has solved the optimal auction problem in the one - dimensional setting. However, in the multi - dimensional setting, even in the restricted two - item instance, the optimal mechanism can become extremely difficult to handle. Therefore, many studies have turned to simple mechanisms. Although these mechanisms are usually not optimal, they can achieve near - optimal revenues in many cases.
However, constant - factor approximations do not fully explain the reasons for the widespread use of simple auctions in practice. The resource - augmentation paradigm offers a different perspective: running a complex auction mechanism has costs. Is it possible to more effectively increase revenues by recruiting additional bidders to participate in a simple auction mechanism? This is the question that this paper attempts to answer.
### Main Contributions
The main contribution of this paper lies in solving the competition complexity problem in the "large \(n\)" case, that is, when the number of bidders \(n\) is greater than the number of items \(m\), the competition complexity is \(\Theta(\sqrt{nm})\). Specifically:
1. **Upper Bound**: Previous work has already proven the upper bound of \(O(\sqrt{nm})\).
2. **Lower Bound**: This paper improves the previous lower bound of \(\Omega(\ln n)\) to \(\Omega(\sqrt{nm})\), thus completely solving the competition complexity problem in the "large \(n\)" case.
### Technical Highlights
To prove this result, the author designs a Bayesian Incentive - Compatible (BIC) auction mechanism, which is applicable to bidders' valuations drawn from the truncated Equal Revenue curve. Here are several technical highlights:
1. **Separate Sale with Near - Optimal Welfare**: When \(n\gg m\), selling each item separately is already very close to the optimal welfare, leaving very limited room for further improvement.
2. **Non - Explicit DSIC Auction Design**: The author designs a BIC but non - DSIC auction mechanism and proves its feasibility and incentive compatibility.
3. **Price Adjustment**: By adjusting the prices, the author ensures the feasibility of the auction mechanism and makes it close to BIC.
### Independent Results
During the research process, the author also obtains two independent results:
1. **Worst - Case Competition Complexity**: For any \(n\), the worst - case competition complexity is witnessed by independently and identically distributed truncated Equal Revenue curves.
2. **Performance of Un - Truncated Equal Revenue Curves**: For some absolute constant \(c\), \(VCG_{n + cm\ln(n)}(ER^m)\geq Rev_n(ER^m)\), which means that \(O(m\ln(n))\) additional bidders are sufficient to make the revenue from separate sales exceed the optimal revenue.
These results not only help in understanding the main conclusion of this paper but also provide deeper insights into the field of multi - dimensional mechanism design.
### Conclusion
Through rigorous mathematical analysis and innovative auction mechanism design, this paper solves a key problem in multi - dimensional auctions, that is, it determines the competition complexity in the case of additive bidders and independent items. This result not only fills a theoretical gap but also provides theoretical support for simple auction mechanisms in practical applications.