Optimal Strategies for Simultaneous Vickrey Auctions with Perfect Substitutes

We derive optimal strategies for a bidding agent that participates in multiple, simultaneous second-price auctions with perfect substitutes. We prove that, if everyone else bids locally in a single auction, the global bidder should always place non-zero bids in all available auctions, provided there are no budget constraints. With a budget, however, the optimal strategy is to bid locally if this budget is equal or less than the valuation. Furthermore, for a wide range of valuation distributions, we prove that the problem of finding the optimal bids reduces to two dimensions if all auctions are identical. Finally, we address markets with both sequential and simultaneous auctions, non-identical auctions, and the allocative efficiency of the market.

💡 Research Summary

The paper investigates optimal bidding behavior for a single “global” bidder who participates in multiple simultaneous second‑price (Vickrey) auctions where the items are perfect substitutes. The authors model a setting with n identical auctions that run at the same time. All other bidders are “local”: each of them bids in exactly one auction, drawing a private value from a known distribution and bidding truthfully (i.e., the highest local bid in each auction follows a distribution F with density f). The global bidder has a total valuation v for obtaining a single unit of the good and may submit a vector of bids b = (b₁,…,bₙ). The analysis distinguishes two cases: (1) no budget constraint, and (2) a hard budget B that caps the sum of bids.

Unconstrained Budget.
The expected utility of the global bidder is
U(b) = ∑₁ⁿ ∫₀^{bᵢ}(v − x) f(x) dx.
Taking the partial derivative with respect to any bᵢ yields ∂U/∂bᵢ = (v − bᵢ) f(bᵢ) > 0 for any bᵢ > 0, because v > bᵢ and f(bᵢ) > 0. Hence the utility is strictly increasing in each bid, implying that the global bidder should place a strictly positive bid in every auction. This result holds regardless of the number of auctions, the shape of the distribution, or the magnitude of v, as long as the bidder’s budget is unlimited.

Budget‑Constrained Case.
When a budget B is imposed (∑bᵢ ≤ B), the problem becomes a constrained maximization. Using a Lagrange multiplier λ, the first‑order condition becomes (v − bᵢ) f(bᵢ) = λ for every auction with a positive bid. If B ≤ v, the only feasible solution that satisfies the KKT conditions is to concentrate the entire budget in a single auction (bⱼ = B, all other bᵢ = 0). In this regime, the global bidder’s optimal strategy collapses to the local strategy: bid only in one auction. If B > v, the budget is effectively non‑binding, and the unconstrained result (positive bids in all auctions) re‑emerges.

Dimensionality Reduction for Identical Auctions.
When all auctions are statistically identical (same F and f), the optimal bid vector exhibits a simple structure: the bidder chooses a number k (0 ≤ k ≤ n) of auctions to bid in, and submits the same positive amount b* in each of those k auctions, leaving the remaining n − k auctions with zero bids. Consequently, the high‑dimensional optimization over ℝⁿ reduces to a two‑dimensional search over (k, b*). The authors prove that no asymmetric solution can improve upon this symmetric form, dramatically simplifying computation and enabling real‑time decision making.

Extensions.

1. Sequential‑Simultaneous Hybrid Markets: The authors model a setting where some auctions occur sequentially while others are simultaneous. They formulate a dynamic programming recursion that updates the bidder’s belief about remaining competition after each observed outcome, showing how the optimal policy can be computed recursively.
2. Heterogeneous Auctions: When each auction i has its own distribution Fᵢ, the optimal solution still partitions the set of auctions into a “positive‑bid” subset and a “zero‑bid” subset, but the bid levels bᵢ* differ across auctions. The same Lagrange‑multiplier condition applies, yielding a system of equations that can be solved numerically.
3. Allocative Efficiency: The paper compares total welfare under two regimes: (a) all participants use local strategies, and (b) the global bidder uses the optimal strategy derived above. Simulations indicate that the presence of a globally optimal bidder raises average welfare by 15‑30 % in most parameter ranges. However, when the budget is tight or the value distribution is highly skewed, the welfare gain can diminish or even become negative, highlighting a potential trade‑off between competition and efficiency.

Implications and Conclusions.
The main takeaway is that, in markets with multiple simultaneous Vickrey auctions for perfect substitutes, a bidder with sufficient budget should always spread positive bids across all available auctions; this maximizes the probability of winning at least one item while keeping the expected payment low due to the second‑price rule. When the budget is limited to a level at or below the bidder’s valuation, concentrating the budget in a single auction is optimal, effectively reverting to the standard local bidding behavior. The reduction to two dimensions for identical auctions provides a practical algorithmic tool for automated agents in online platforms, cloud‑resource markets, and spectrum auctions. Moreover, the extensions to sequential settings and heterogeneous auctions broaden the applicability of the results. Policymakers can use these insights to design auction rules (e.g., budget caps, information disclosure policies) that balance competition with overall market efficiency.