Non-uniform Bid-scaling and Equilibria for Different Auctions: An Empirical Study

Yuan Deng,Jieming Mao,Vahab Mirrokni,Yifeng Teng,Song Zuo

2023-11-18

Abstract:In recent years, the growing adoption of autobidding has motivated the study of auction design with value-maximizing auto-bidders. It is known that under mild assumptions, uniform bid-scaling is an optimal bidding strategy in truthful auctions, e.g., Vickrey-Clarke-Groves auction (VCG), and the price of anarchy for VCG is $2$. However, for other auction formats like First-Price Auction (FPA) and Generalized Second-Price auction (GSP), uniform bid-scaling may not be an optimal bidding strategy, and bidders have incentives to deviate to adopt strategies with non-uniform bid-scaling. Moreover, FPA can achieve optimal welfare if restricted to uniform bid-scaling, while its price of anarchy becomes $2$ when non-uniform bid-scaling strategies are allowed. All these price of anarchy results have been focused on welfare approximation in the worst-case scenarios. To complement theoretical understandings, we empirically study how different auction formats (FPA, GSP, VCG) with different levels of non-uniform bid-scaling perform in an autobidding world with a synthetic dataset for auctions. Our empirical findings include: * For both uniform bid-scaling and non-uniform bid-scaling, FPA is better than GSP and GSP is better than VCG in terms of both welfare and profit; * A higher level of non-uniform bid-scaling leads to lower welfare performance in both FPA and GSP, while different levels of non-uniform bid-scaling have no effect in VCG. Our methodology of synthetic data generation may be of independent interest.

Computer Science and Game Theory

What problem does this paper attempt to address?

### What problems does this paper attempt to solve?

This paper aims to explore and analyze the performance of different auction formats (such as First - Price Auction (FPA), Generalized Second - Price Auction (GSP) and Vickrey - Clarke - Groves Auction (VCG)) under non - uniform bid - scaling conditions in the autobidding environment. Specifically, the paper attempts to solve the following problems:

1. **The impact of non - uniform bid - scaling**:
- Research the impact of using non - uniform bid - scaling strategies on social welfare and profit in different auction formats.
- Explore whether non - uniform bid - scaling can improve or worsen the efficiency and revenue of auctions.

2. **Differences between theoretical and practical performance**:
- Supplement existing theoretical research and verify the robustness of these theoretical results in practical applications through empirical analysis.
- Compare the differences between the theoretical performance in the worst - case scenario of different auction formats and their performance in actual data generation and simulation experiments.

3. **Exploration of optimal bidding strategies**:
- For auction formats that do not truthfully report values (such as FPA and GSP), study whether non - uniform bid - scaling strategies can be a better bidding strategy.
- Propose an efficient method to optimize non - uniform bid - scaling factors and evaluate their performance under different levels of degrees of freedom.

4. **The impact of introducing user costs**:
- Analyze how the performance of different auction formats changes when user costs are introduced and the impact of these changes on social welfare and profit.

### Formula display

To ensure the correctness and readability of the formulas, the following are some key formulas involved in the paper:

- **Bidding formula**:
\[
\text{bid}=\kappa\cdot\text{pCVR}\cdot\text{tCPA}
\]
where \(\kappa\) is the bid - scaling factor, \(\text{pCVR}\) is the predicted conversion rate, and \(\text{tCPA}\) is the target cost - per - conversion.

- **Definitions of social welfare and profit**:
\[
\text{welfare}=\sum_{i\in N}\text{value}_i - \text{cost}_i
\]
\[
\text{profit}=\sum_{i\in N}\text{spend}_i - \text{cost}_i
\]

- **Uniform bid - scaling strategy**:
\[
b_{i,j}=\kappa_i\cdot\frac{v_{i,j}}{\tau_i}
\]

- **Non - uniform bid - scaling strategy**:
\[
b_{i,j}=\kappa_{i,d}\cdot\frac{v_{i,j}}{\tau_i}
\]
where \(\kappa_{i,d}\) is the bid - scaling factor for the \(d\)-th - level partition.

Through the research of these problems, the paper hopes to provide new insights for understanding auction design in the autobidding environment and provide guidance for future mechanism design.