Pith. sign in

REVIEW 3 major objections 6 minor 29 references

Rethinking Pricing in Energy Markets: Pay-as-Bid vs Pay-as-Clear

T0 review · 3 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper proves that Pay-as-Bid weakly dominates Pay-as-Clear, meaning in every modeled energy market the worst equilibrium price under PB is at most that under PC.

desk verdict New and worthwhile worst-case comparison of PB vs PC, but the proof of the key PC lower bound has a gap that needs fixing before the main theorem is established. read the letter →

arxiv 2507.06035 v1 pith:ZOICUWNF submitted 2025-07-08 cs.GT

Ioannis Caragiannis , Zhile Jiang , Stratis Skoulakis This is my paper · ORCID
classification cs.GT MSC 91B2691A1091A80
keywords pay-as-bidpay-as-clearmarket-clearingpriceNashequilibriummechanismdesignelectricitymarketsno-regretlearningworst-caseanalysis
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper asks whether energy markets should pay producers their own bid (pay-as-bid, PB) or a uniform clearing price (pay-as-clear, PC), and answers in favor of PB on worst-case grounds. In every instance of its market model, the highest unit price that can arise at a mixed Nash equilibrium under PB is no larger than the highest equilibrium unit price under PC, and in non-degenerate instances it is strictly lower. The paper also proves that no mechanism can be strictly better than PB, PC, or VCG in every market instance, so the PB advantage is a worst-case robustness property rather than a claim of universal superiority. Simulations where producers update bids with a no-regret Hedge algorithm show lower average unit prices under PB, matching the theoretical comparison.

What carries the argument

The argument is carried by two instance-dependent thresholds. For each producer $i$, $b_i^H$ is the highest bid that can appear as a best response for $i$ when every other producer bids either its true cost or its true cost plus one; $b_i^L$ is the smallest price at which $i$, selling its full capacity, would earn at least the utility of its best response to truthful opponents. The pivotal agent $\tau(c)$ is the first producer in bid order whose capacity completes total demand, and the market-clearing price is that agent's bid. Theorem 5 bounds every PB equilibrium inside $[\max_{i \preceq_c \tau(c)} b_i^L - 1, \max_{i \preceq_c \tau(c)} b_i^H]$, while Theorem 4 pushes PC's worst pure NE price up to at least $\max_{i \preceq_c \tau(c)} b_i^H$. The thresholds turn the strategic comparison into a sandwich: PB's worst case sits at or below the same level that PC's worst case must reach.

What would settle it

Inspect the pure Nash equilibrium constructed in the proof of Theorem 7 (Appendix B). If one can exhibit a market instance—say, with a small number of producers, integer costs, and bids on a grid—where some agent $j$ with $c_j < b_i^H$ increases its utility by bidding just below $b_i^H$ and thereby becoming the pivotal, price-setting agent, then the unit price of that profile falls below $\max_{i \preceq_c \tau(c)} b_i^H$, and the lower bound used to prove Theorem 2 collapses for that instance.

Watch

Extended reading notes

Core claim

The paper's main result is Theorem 2: Pay-as-Bid weakly dominates Pay-as-Clear, while the reverse dominance fails. It is obtained by proving two facts: Pay-as-Clear always has a pure Nash equilibrium whose unit price is at least $\max_{i \preceq_c \tau(c)} b_i^H$ (Theorem 4), while every mixed Nash equilibrium of Pay-as-Bid has support inside $[\max_{i \preceq_c \tau(c)} b_i^L - 1, \max_{i \preceq_c \tau(c)} b_i^H]$ (Theorem 5). Thus PB's worst equilibrium price is bounded above by a value that PC's worst equilibrium price must at least reach. Along the way the paper shows PC is not truthful, PB need not have a pure Nash equilibrium, VCG does not weakly dominate PC (there is an instance where VCG's unit price is $\Theta(\log n)$ times PC's worst-case price), and no mechanism strictly dominates PB, PC, or VCG.

Load-bearing premise

The lower bound for Pay-as-Clear rests on a claim in the construction behind Theorem 4 (Appendix B): an agent whose true cost is below the proposed price cannot profitably deviate to a lower bid because the clearing price would stay fixed; that monotonicity step is asserted rather than proved, and it could fail if a lower bid makes that agent the marginal, price-setting supplier and pulls the clearing price down.

Editorial extensions

If this is right

  • A regulator who cares about worst-case outcomes under strategic bidding gets a formal reason to treat pay-as-bid as the more manipulation-robust of the two mechanisms.
  • The two-producer finding that PB yields lower prices than PC under strategic behavior extends to arbitrary many producers, for worst-case equilibrium prices.
  • Pay-as-Clear's expensive outcome is not a mixed-strategy artifact: it is attained at a pure Nash equilibrium, so it is stable and reachable by best-response dynamics.
  • Truthfulness is not a safeguard: VCG can be $\Theta(\log n)$ times more expensive than Pay-as-Clear's worst equilibrium in some instances.
  • In the large family of instances covered by Theorem 6, PB's worst-case price is strictly below the level that PC must reach, so the gap is real, not just non-strict.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the unproved monotonicity in the pure-NE construction behind Theorem 4 fails, PC's worst-case bound would need repair and PB's dominance might rest on weaker ground; a direct check of the construction on small random instances is a cheap way to test it.
  • The cyclical bidding pattern observed in the PB simulations—prices decaying toward the lower threshold, then jumping back—suggests the relevant worst-case object for dynamic markets may be the time-average price under no-regret learning rather than the mixed-NE support; a formal price-of-learning bound would connect the simulations to the theory.
  • The model restricts bids and costs to a shared integer grid $[M]$; whether the PB-over-PC worst-case ordering survives continuous bid spaces and richer cost curves is an immediate open question.
  • The same threshold sandwich ($b_i^L, b_i^H$) could be reused to compare PB with other uniform-price variants, such as paying the pivotal agent's true cost rather than its bid, or with supply-function equilibria.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper studies a procurement auction for unit energy demand with n producers having capacities s_i and integer marginal costs c_i in {0,...,M}. A mechanism selects a cost-minimizing allocation and a payment rule; the authors compare Pay-as-Bid (PB), Pay-as-Clear (PC), and VCG. They define strong and weak dominance relations in terms of mixed Nash equilibrium unit prices and state three main theorems: (1) no mechanism strictly dominates PC, PB, or VCG; (2) PB weakly dominates PC in worst-case mixed Nash equilibrium price, while the converse fails; and (3) VCG does not weakly dominate PC and can be a factor Theta(log n) worse. The proofs use instance-dependent parameters bL_i and bH_i, a pure-Nash-equilibrium construction for PC (Theorem 7), and support bounds for PB (Theorems 9 and 10). The paper also reports Hedge-learning simulations that, on the tested instances, yield lower average prices under PB than under PC.

Significance. If the results are correct, Theorem 2 is a notable worst-case separation between two practically important pricing rules in a clean, stylized model, and Theorem 1's non-dominance statement is an interesting contribution to mechanism design. A strength of the paper is that the quantities bH_i and bL_i are defined directly from model primitives and are not fitted to the simulations; the theoretical claims are self-contained. The narrative is clear and the paper engages seriously with the energy-market literature. However, the proofs as written contain gaps in load-bearing places: the pure-Nash-equilibrium construction for PC is incomplete, and the perturbation argument in Theorem 1 leaves the stated integer-cost model. These gaps must be repaired before the central claims can be considered established.

major comments (3)
  1. [Appendix A, Theorem 1] The perturbed-cost argument uses instances with marginal costs epsilon_k = 2^{-k}, but the model in Section 2 restricts costs to integers in {0,...,M} and bids to [M]. For non-integer costs, the game and the best-response sets are not defined under the paper's own definitions, and the limiting mixed equilibrium sigma* is taken for games outside the model. The theorem may be salvageable by scaling or by explicitly extending the model to real costs, but as written the proof does not establish Theorem 1 within the stated domain.
  2. [Appendix C, Lemma 1] Lemma 1 is false as stated. For n=2, s1=s2=1, c1=5, c2=0, and b=(4,4) with M >= 5, agent 1 sells 1 unit at price 4 in profile b, giving U1(b) = -1, whereas against the truthful bid b2=0 agent 1 sells zero and obtains utility 0. Thus U1(b) >= U1(b1,c_{-1}) fails. The monotonicity argument in the proof of Lemma 1 is valid only when the utilities involved are nonnegative, for example when all bids in the support are at least costs and selected agents receive at least their costs. Since Lemma 1 is used in the proof of Theorem 8, the statement and proof need to be corrected or restricted to equilibrium bid profiles.
  3. [Appendix F, Theorem 9] The proof of Theorem 9 asserts that 'since agent i sells a positive amount once bidding bH_i and since bL_i <= bH_i, we are ensured that P_{c_j < bL_i} s_j < 1.' This implication is not valid: the positive sale by i at bid bH_i only implies that the total supply of agents j != i with bids below bH_i is strictly less than 1; the set {j : c_j < bL_i} can include i itself and agents whose bids are not below bH_i. Consequently the subsequent claim that every j != i with c_j < bL_i sells its full supply by bidding bL_i - 1 is not established. The lower-bound half of Theorem 5 therefore needs a corrected proof.
minor comments (6)
  1. [Section 3, Remark 1] Remark 1 states that the worst-case NE unit price of PC is 'actually strictly smaller' than that of PB, which is the reverse of the paper's main claim and of Theorem 2; it should presumably say that the PC worst case is strictly larger (or PB strictly smaller). The sentence immediately before Remark 1 also conflates 'the unit price of any mixed NE of PB' with the worst-case NE of PB.
  2. [Section 5, Theorem 4] There is a typo in the statement of Theorem 4: 'Pas-as-Clear' should be 'Pay-as-Clear'.
  3. [Definition 4 and Theorem 1] The paper uses both 'strongly dominates' and 'strictly dominating' without aligning the terminology; Theorem 1 and the introduction should use the term defined in Definition 4.
  4. [Appendix K, Figure 8 caption] The caption reports s5 = 26, which is infeasible because supplies are in (0,1]; this should be a typo for 0.26 or similar.
  5. [Section 7] The experimental section does not report the Hedge learning rate, the number of independent runs, or error bars; including these details would make the simulations reproducible.
  6. [Section 1] There is a typo in the introduction: 'game-theretic study' should be 'game-theoretic study'.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the derivation is self-contained and uses only model primitives; the flagged Appendix B concern is a proof-gap/correctness issue, not a circular reduction.

full rationale

The paper's central claims are game-theoretic theorems, and the proof chain does not fit parameters or rename predictions. bH_i and bL_i are defined directly from costs, supplies, and best responses to truthful or cost-plus-one profiles (Definition 5), and they are then used in Theorems 4, 5, 7, 9, and 10 to bound equilibrium prices. The weak-dominance conclusion in Theorem 2 follows by sandwiching the worst mixed-NE unit price of PB from above by max bH_i (Theorem 5) and the worst-case PC price from below by the same quantity (Theorem 4 via Theorem 7). This is a real mathematical argument: the lower bound for PC requires the nontrivial construction of a pure Nash equilibrium in Appendix B, and the upper bound for PB requires support arguments based on best-response guarantees in Appendix G. No quantity is fitted to simulation output and then reported as a prediction; the simulations in Section 7 are illustrative and use the already-defined theoretical thresholds. There are no load-bearing self-citations: the references are external literature, and no 'uniqueness theorem' by the same authors is invoked. The most relevant flagged issue, noted by the skeptical reader, is in Appendix B's proof of Theorem 7: the claim that for an agent j with c_j < bH_i, bidding below bH_i keeps the clearing price at bH_i (or keeps the allocation monotone) is asserted rather than proven, and a lower bid could in principle change the pivotal agent and lower the clearing price. That is a correctness or proof-completeness concern, not circularity: it does not make the theorem's conclusion equivalent to its inputs by definition, and it involves no fitted parameter, no self-citation, and no redefinition of a known result. Accordingly, the circularity score is 0.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The theory uses only the model primitives and standard Nash existence; no parameters are fitted to data. The proof of Theorem 1 introduces perturbed costs outside the integer-cost domain (an ad hoc device). The analysis adopts complete information as a solution-concept assumption despite the private-cost framing. The only hand-chosen parameter in the paper is the unreported learning rate of the Hedge algorithm in the simulations.

free parameters (1)
  • Hedge learning rate
    The simulations use the Hedge algorithm but do not report the learning rate, number of iterations per run, or number of runs. This parameter is chosen by the experimenters and affects the reported price trajectories, but it is not fitted to any target result.
assumptions (5)
  • standard math Finite games admit mixed Nash equilibria (Nash's theorem)
    Used in Section 2.1 to assert existence for every mechanism; not proven in the paper.
  • domain assumption Costs and bids are integers in [0,M] and total demand is normalized to 1
    Model setup in Section 2; defines the instance space.
  • domain assumption The market uses a cost-minimizing allocation with lexicographic tie-breaking
    Definition 1 in Section 2; the basis of the pivotal-agent definition.
  • ad hoc to paper Perturbed costs ε_k = 2^{-k} are allowed in the proof of Theorem 1 even though the model restricts costs to integers
    Used to construct a limit argument; extends the model beyond stated domain.
  • domain assumption Producers have complete information about costs when optimizing; Nash equilibrium is the relevant solution concept
    Definition 2 in Section 2.1; the paper motivates NE via repeated interaction but does not model the learning process.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Rethinking Pricing in Energy Markets: Pay-as-Bid vs Pay-as-Clear." pith.science (2026). https://pith.science/paper/ZOICUWNF

@misc{pith2026250706035,
  author       = {Pith},
  title        = {Pith review of: Rethinking Pricing in Energy Markets: Pay-as-Bid vs Pay-as-Clear},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZOICUWNF}},
  note         = {Machine review of arXiv:2507.06035}
}
abstract

The design of energy markets is a subject of ongoing debate, particularly concerning the choice between the widely adopted Pay-as-Clear (PC) pricing mechanism and the alternative Pay-as-Bid (PB). These mechanisms determine how energy producers are compensated: under PC, all selected producers are paid the market-clearing price (i.e., the highest accepted bid), while under PB, each selected producer is paid their own submitted bid. The overarching objective is to meet the total demand for energy at minimal cost in the presence of strategic behavior. We present two key theoretical results. First, no mechanism can uniformly dominate PC or PB. This means that for any mechanism $\mathcal{M}$, there exists a market configuration and a mixed-strategy Nash equilibrium of PC (respectively for PB) that yields strictly lower total energy costs than under $\mathcal{M}$. Second, in terms of worst-case equilibrium outcomes, PB consistently outperforms PC: across all market instances, the highest possible equilibrium price under PB is strictly lower than that under PC. This suggests a structural robustness of PB to strategic manipulation. These theoretical insights are further supported by extensive simulations based on no-regret learning dynamics, which consistently yield lower average market prices in several energy market settings.

Figures

Figures reproduced from arXiv: 2507.06035 by the authors.

Figure 4
Figure 4. The normalized unit price changing by iteration in an energy market with M = 800 and n = 4 producers with supplies s1 = s2 = s3 = s4 = 0.3 and marginal costs c1 = c2 = c3 = c4 = 0. For every agent i ∈ [n], we have b H i = 800 and b L i = 267. 0 10 000 20 000 30 000 40 000 50 000 0.0 0.2 0.4 0.6 0.8 1.0 Iteration Normalized unit price Pay-as-Bid Pay-as-Clear [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗
Figure 6
Figure 6. The normalized unit price changing by iteration in an energy market with M = 800 and n = 2 producers with supplies s1 = s2 = 0.99 and marginal costs c1 = c2 = 0. For every agent, we have b H i = 800 and b L i = 9. 0 20 000 40 000 60 000 80 000 100 000 120 000 140 000 0.0 0.2 0.4 0.6 0.8 1.0 Iteration Normalized unit price Pay-as-Bid Pay-as-Clear [PITH_FULL_IMAGE:figures/full_fig_p025_6.png] view at source ↗
Figure 8
Figure 8. The normalized unit price changing by it [PITH_FULL_IMAGE:figures/full_fig_p026_8.png] view at source ↗
Figures from the paper (1 more)
Figure 10
Figure 10. Figure 10: The normalized unit price changing by iteration in the energy market with M = 800 and n = 4 producers with supplies s1 = 0.75, s2 = 0.75, s3 = 0.1, s4 = 0.05 and marginal costs c1 = 0, c2 = 100, c3 = 400, c4 = 600. We have that maxi⪯cτ(c) b H i = 601 and maxi⪯cτ(c) b …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

29 extracted references · 28 canonical work pages

  1. [1]

    Electricity market design, European Commission (2025), https://energy.ec.europa.eu/topics/ markets-and-consumers/electricity-market-design_en 1

    work page 2025

  2. [2]

    In: Proceedings of the 24th ACM Conference on Economics and Computation

    Ahunbay, M.c., Bichler, M., Kn¨ orr, J.: Pricing optimal outcomes in coupled and non-convex electricity markets. In: Proceedings of the 24th ACM Conference on Economics and Computation. p. 59 (2023) 1

    work page 2023

  3. [3]

    pay-as-bid pricing

    Akbari-Dibavar, A., Mohammadi-Ivatloo, B., Zare, K.: Electricity market pricing: Uniform pricing vs. pay-as-bid pricing. In: Nojavan, S., Zare, K. (eds.) Electricity Markets: New Players and Pricing Uncertainties, pp. 19–35. Springer International Publishing (2020) 3

    work page 2020

  4. [4]

    Proceedings of the National Power System Conference (NPSC) pp

    Bajpai, P., Singh, S.: Bidding and gaming in electricity market: An overview and key issues. Proceedings of the National Power System Conference (NPSC) pp. 338–346 (2004) 3

    work page 2004

  5. [5]

    In: Proceedings of the 36th Annual Hawaii International Conference on System Sciences (HICSS’03)

    Cramton, P.: Electricity market design: The good, the bad, and the ugly. In: Proceedings of the 36th Annual Hawaii International Conference on System Sciences (HICSS’03). p. 54.2 (2003) 1, 2

    work page 2003

  6. [6]

    The RAND Journal of Economics 37(1), 23–46 (2006) 3

    Fabra, N., von der Fehr, N.H., Harbord, D.: Designing electricity auctions. The RAND Journal of Economics 37(1), 23–46 (2006) 3

    work page 2006

  7. [7]

    The Electricity Journal 15(7), 72–81 (2002) 3

    Fabra, N., von der Fehr, N.H., Harbord, D.: Modeling electricity auctions. The Electricity Journal 15(7), 72–81 (2002) 3

    work page 2002

  8. [8]

    Journal of Regulatory Economics 24, 175–211 (02 2003) 3

    Federico, G., Rahman, D.: Bidding in an electricity pay-as-bid auction. Journal of Regulatory Economics 24, 175–211 (02 2003) 3

    work page 2003

Show all 29 references
  1. [9]

    Electric Power Systems Research 234, 110693 (2024) 1, 2

    Getaneh Abate, A., Majdi, D., Kazempour, J., Kamgarpour, M.: Learning to bid in forward electricity markets using a no-regret algorithm. Electric Power Systems Research 234, 110693 (2024) 1, 2

    2024

  2. [10]

    In: Teglio, A., Alfarano, S., Camacho-Cuena, E., Gin´ es-Vilar, M

    Guerci, E., Rastegar, M.A.: Comparing system-marginal-price versus pay-as-bid auctions in a realistic electricity market scenario. In: Teglio, A., Alfarano, S., Camacho-Cuena, E., Gin´ es-Vilar, M. (eds.) Managing Market Complexity: The Approach of Artificial Economics. Spring...

    2013

  3. [11]

    Applied Mathematics Letters 19(8), 814–819 (2006) 3

    H¨ ast¨ o, P., Holmberg, P.: Some inequalities related to the analysis of electricity auctions. Applied Mathematics Letters 19(8), 814–819 (2006) 3

    2006

  4. [12]

    CoRR abs/1909.05207 (2023) 11

    Hazan, E.: Introduction to online convex optimization. CoRR abs/1909.05207 (2023) 11

    2023 arXiv

  5. [13]

    Kahn, A.E., Cramton, P., Porter, R.H., Tabors, R.D.: Pricing in the California power exchange electricity market: Should california switch from uniform pricing to pay-as-bid pricing? Blue Ribbon Panel Report, California Power Exchange 2 (2001) 3

    2001

  6. [14]

    IEEE Transactions on Energy Conversion 16(4), 352–360 (2001) 3

    Kumar David, A., Wen, F.: Market power in electricity supply. IEEE Transactions on Energy Conversion 16(4), 352–360 (2001) 3

    2001

  7. [15]

    International Journal of Electrical Power and Energy Systems 43(1), 696–702 (2012) 3

    Liu, Z., Yan, J., Shi, Y., Zhu, K., Pu, G.: Multi-agent based experimental analysis on bidding mechanism in electricity auction markets. International Journal of Electrical Power and Energy Systems 43(1), 696–702 (2012) 3

    2012

  8. [16]

    Oxford University Press (1995) 4

    Mas-Colell, A., Whinston, M.D., Green, J.R.: Microeconomic Theory. Oxford University Press (1995) 4

    1995

  9. [17]

    World Bank Publications (2011) 3

    Maurer, L., Barroso, L.A.: Electricity auctions: an overview of efficient practices. World Bank Publications (2011) 3

    2011

  10. [18]

    Energy Economics 49, 308–316 (2015) 3

    Nazemi, A., Mashayekhi, M.: Competitiveness assessment of iran’s restructured electricity market. Energy Economics 49, 308–316 (2015) 3

    2015

  11. [19]

    Energy Economics 73, 194–211 (2018) 3

    Ocker, F., Ehrhart, K.M., Belica, M.: Harmonization of the european balancing power auction: A game-theoretical and empirical investigation. Energy Economics 73, 194–211 (2018) 3

    2018

  12. [20]

    IEEE Transactions on Power Systems 19(4) (2004) 3

    Ren, Y., Galiana, F.: Pay-as-bid versus marginal pricing-part I: Strategic generator offers. IEEE Transactions on Power Systems 19(4) (2004) 3

    2004

  13. [21]

    IEEE Transactions on Power Systems 19(4), 1777–1783 (2004) 3

    Ren, Y., Galiana, F.: Pay-as-bid versus marginal pricing-Part II: market behavior under strategic generator offers. IEEE Transactions on Power Systems 19(4), 1777–1783 (2004) 3

    2004

  14. [22]

    Cambridge University Press (2016) 4

    Roughgarden, T.: Twenty Lectures on Algorithmic Game Theory. Cambridge University Press (2016) 4

    2016

  15. [23]

    INFOR: Information Systems and Operational Research 48(4), 261–266 (2010) 3

    Skoulidas, C.C., Vournas, C.D., Papavassilopoulos, G.P.: An adaptive learning game model for interacting electric power markets. INFOR: Information Systems and Operational Research 48(4), 261–266 (2010) 3

    2010

  16. [24]

    IEEE Transactions on Power Systems 19(4), 1990–1998 (2004) 3

    Son, Y.S., Baldick, R., Lee, K.H., Siddiqi, S.: Short-term electricity market auction game analysis: uniform and pay-as-bid pricing. IEEE Transactions on Power Systems 19(4), 1990–1998 (2004) 3

    2004

  17. [25]

    Energy Economics 93, 105035 (2021) 1, 3 13 A Proof of Theorem 1 Theorem 1

    Viehmann, J., Lorenczik, S., Malischek, R.: Multi-unit multiple bid auctions in balancing markets: An agent-based q-learning approach. Energy Economics 93, 105035 (2021) 1, 3 13 A Proof of Theorem 1 Theorem 1. There is no mechanism strictly dominating Pay-as-Clear, Pay-as-Bid,...

    2021

  18. [26]

    There exists an infinite subsequence S ′ ⊆ Ssuch that for any σk ∈ S′, support(σk A) ̸= {0} and support(σk B) ̸= {0}

  19. [27]

    We will establish the latter for two cases separately

    There exists an infinite subsequence S ′ ⊆ Ssuch that for any σk ∈ S′, support(σk A) = {0} Our proof consists of showing in any case the sequence S ′ converges to a point limk→∞ σk = σ⋆ that is a mixed NE for the original game (marginal costs of the agent are 0) while the paym...

  20. [28]

    We have bi ̸= t, so the agent j with cj = t − 1 does not affect the utility if it bids cj + 1

    Agent i’s utility while fixing a specific bid is minimized when the allocation is minimized; 2. We have bi ̸= t, so the agent j with cj = t − 1 does not affect the utility if it bids cj + 1. Putting Inequality (6) and (7) together, we get U (t, c−i + d⋆ −i) = max bi∈[M ] Ui(bi...

  21. [600]

    In such degen- erate instances, the unit price provided by worst NE in PB and PC is the same

    This instance is a degenerate example where maxi⪯cτ (c) bL i + 1 = maxi⪯cτ (c) bH i . In such degen- erate instances, the unit price provided by worst NE in PB and PC is the same. 26

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.