CDS Seminar Series

2026 Edition

We are pleased to continue our in person research seminar series hosted by the Centre for Decision Sciences (CDS) - formerly Centre for Mathematical Social Sciences (CMSS) - at the University of Auckland, New Zealand, which we call "There will be donuts."

The series has been running periodically since August 2022. On this website, you can find upcoming talks, as well as an archive of all presentations held since the series was launched.

If you want to talk about theory (your own, or an interesting paper from the literature) or even give a mini-course on some area of theory you want to tell us about, then please sign up! We encourage informality, brainstorming of potential research ideas, and general enthusiasm for all things theoretical. Book your slot by emailing one of the organisers. If you want to learn some theory, then please come along. Students are particularly welcome (in the audience and on stage). And of course, donuts will be provided! Hope to see you at the seminars!

Best,

The Organisers: Simona Fabrizi (s.fabrizi@auckland.ac.nz), Steffen Lippert (s.lippert@auckland.ac.nz) and Matthew Ryan (mryan@aut.ac.nz)

Time and place: Thursday, 13 August 2026, 11am-noon, Room 260-005 (CaseRoom 1, level 0 of Sir Owen G Glenn Building)

Speaker: Yuanyuan (Lydia) Li [PhD Candidate in Economics, UoA]

Title: Algorithmic Noise, Epistemic Opacity, and the Limits of Platform Control (joint work with Simona Fabrizi and Steffen Lippert [UOA])

Abstract: We study optimal network interventions when strategic interactions are governed by epistemic complementarities rather than classical action spillovers. In our framework, agents strategically signal private beliefs, but platform interventions mechanically inject algorithmic noise that endogenously corrupts the social learning channel; we show that the resulting epistemic state—which we formalize as the filtering barrier—is globally unique. We identify a Tragedy of Interpretability: a platform failing to internalize this friction systematically over-intervenes relative to the social optimum, driving the network toward an absorbing state of complete epistemic opacity. Crucially, we demonstrate Hub Fragility: contrary to classical network heuristics, highly connected hubs are structurally the most vulnerable to interpretability loss and inflict the maximum collateral epistemic damage when targeted. Unconstrained pursuit of topological leverage creates an Illusion of Control that systematically distorts the network hierarchy. Consequently, optimal governance requires an interpretability-adjusted rule that endogenously penalizes high-centrality nodes to preserve the credibility of the social learning channel. The canonical centrality-based targeting heuristic emerges merely as a frictionless special case; under realistic algorithmic noise, safeguarding network interpretability is fundamental to the integrity of digital public discourse.

Time and place: Thursday, 30 July 2026, 11am-noon, Room 260-005 (CaseRoom 1, level 0 of Sir Owen G Glenn Building)

Guest Speaker: André De Palma [Emeritus Professor at CY Cergy Paris University and an instructor at EPFL - Swiss Federal Institute of Technology Lausanne]

Title: Targeted Persuasive Advertising (joint work with Simon Anderson [University of Virginia, USA])

Abstract: This paper develops a theory of persuasive advertising in asymmetric oligopoly, extending the literature beyond the traditional monopoly framework. While most theoretical work on advertising focuses either on monopoly or informative advertising (e.g., Grossman and Shapiro, 1984), the paper analyzes how persuasive advertising operates in competitive markets with heterogeneous firms. Firms differ in intrinsic quality and compete by investing in persuasive advertising that shifts consumer preferences within a random utility (logit) framework (Anderson, de Palma and Thisse, 1992). The analysis begins with a covered-market duopoly. In this benchmark case, persuasive advertising is a zero-sum activity: firms choose identical advertising levels, advertising merely redistributes market shares, and equilibrium positions remain unchanged. The incentive to advertise is strongest when firms are closely matched, confirming the traditional view that advertising wars are most intense in tight oligopolies. The next contribution concerns markets with more than two firms or with an outside option. In this more general setting, the symmetry of the duopoly disappears. Higher-quality firms advertise more, obtain larger market shares, and further reinforce their initial advantages. Persuasive advertising therefore leverages existing competitive advantages, increasing market concentration rather than simply reshuffling demand. This provides a theoretical explanation for why dominant firms may benefit disproportionately from targeted advertising. The paper establishes the existence and uniqueness of equilibrium using the theory of aggregate games (e.g., Corchón, 1994; Anderson, Erkal and Piccinin, 2020), yielding transparent comparative statics and equilibrium rankings. The framework is then extended to endogenous pricing, showing that higher-quality firms simultaneously advertise more, charge higher prices, earn larger markups, and obtain higher profits. Finally, the welfare implications depend on the nature of advertising. If advertising merely reallocates consumers, it generates excessive expenditures with little social benefit, echoing the classic critique of Dixit and Norman (1978). If, however, advertising enhances consumers' valuation of products—as in the complementary advertising framework of Becker and Murphy (1993)- equality advertising may instead be socially insufficient. Overall, the paper shows that persuasive advertising is neutral only in the special case of a covered duopoly. In general, oligopoly amplifies firm heterogeneity, increases industry concentration, and strengthens market dominance, offering a new theoretical perspective on the role of targeted advertising in digital markets. We develop a new welfare analysis (for the MLM and the nested logit model) when the information underlying individual choices is only partially reliable. We show how the standard formula should be simply amended in this case.

Selected References:

Discrete Choice Theory of Product Differentiation — Anderson, de Palma and Thisse (1992).
Grossman, G. M., & Shapiro, C. (1984). Informative advertising with differentiated products. The Review of Economic Studies, 51(1), 63-81.
Dixit, A., & Norman, V. (1978). Advertising and welfare. The Bell Journal of Economics, 1-17.
Becker, G. S., & Murphy, K. M. (1993). A simple theory of advertising as a good or bad. The Quarterly Journal of Economics, 108(4), 941-964.
Corchón, L. C. (1994). Comparative statics for aggregative games the strong concavity case. Mathematical Social Sciences, 28(3), 151-165.
Anderson, S. P., Erkal, N., & Piccinin, D. (2020). Aggregative games and oligopoly theory: Short‐run and long‐run analysis. The RAND Journal of Economics, 51(2), 470-495.Chiappori, P.-A. (1988). “Rational Household Labor Supply”. Econometrica 56.1, pp. 63–90.

Short bio: André de Palma (PhD in Physics and Economics) is an Emeritus Professor at CY Cergy Paris University and an instructor at EPFL - Swiss Federal Institute of Technology Lausanne. In collaboration with Simon Anderson and Jacques-François Thisse, he is one of the fathers of new industrial organization and a contributor to discrete choice models. He also works in Transportation Economics, particularly in developing dynamic models with Moshe Ben-Akiva, Richard Arnott, and Robin Lindsey. He has created several software programs, including the dynamic equilibrium simulator METROPOLIS, which is available online and in use worldwide (https://metropolis2.org/). He is also the co-founder, with Nathalie Picard, of a spin-off, RiskDesign, which provides decision-support tools to private investors and has developed risk-attitude analysis tools currently used by more than 2 million people. de Palma has published 10 books and 320 articles in internationally refereed journals in Industrial Organization, Urban Economics, Transportation Economics, Decision Making, and Risk-Taking. He has managed grants in Canada, Belgium, Switzerland, the USA, and France.

Time and place: Thursday, 30 July 2026, 3pm-4pm, Room 260-005 (CaseRoom 1, level 0 of Sir Owen G Glenn Building)

Guest Speaker: Nathalie Picard [University of Strasbourg]

Title: Negotiation of joint discrete decisions in dual-earner couples (joint work with Youssef El Yaakoubi [BETA, University of Strasbourg, France], André de Palma [THEMA, CY Cergy Paris Université, France], Michel Bierlaire [TRANSP-OR, EFPL, Switzerland])

Abstract: We develop a collective discrete-choice framework for joint decisions in dual-earner couples. Following Chiappori (1988, 1992), we assume that the decisions collectively made by couples are Pareto-efficient, and can thus be represented by a weighted sum of individual utilities, in which the Pareto weight captures intra-household bargaining power. Chiappori's collective framework, initially developed for continuous decisions, is here extended to discrete decisions. We further assume egotistic preferences, and consider an assignable-good structure for individual utilities. We then prove the identification of individual utilities (up to a money-metric normalization) on the one hand, and of Pareto weights on the other hand. Our framework requires no observations on prices, income, or budget shares, which makes it applicable to a wide range of empirical settings beyond traditional consumption or labour supply. In the empirical application, we embed the collective structure in a mixed logit model with shared error components, and we estimate if using Bayesian methods. We illustrate our methodology on joint commuting and vehicle ownership choices jointly made by dual-earner couples in Paris region, using census data.

Selected References:

Chiappori, P.-A. (1988). “Rational Household Labor Supply”. Econometrica 56.1, pp. 63–90.
Chiappori, P.-A. (1992). “Collective Labor Supply and Welfare”. Journal of Political Economy 100.3, pp. 437–67.
de Palma, A., Picard, N., and Inoa, I. (2014). “Discrete Choice Decision-Making with Multiple Decision Makers within the Household”. Handbook of Choice Modelling. Ed. by Hess, S. and Daly, A. Edward Elgar Publishing.
Picard, N., Dantan, S., and de Palma, A. (2018). “Mobility decisions within couples”. Theory and Decision 84.2, pp. 149–180.
Train, K. (2009). Discrete choice methods with simulation. Cambridge university press.
Walker, J. L., Ben-Akiva, M., and Bolduc, D. (2007). “Identification of Parameters in Normal Error Component Logit-Mixture (NECLM) Models”. Journal of Applied Econometrics 22.6, pp. 1095–1125.

Short bio: Nathalie Picard (PhD in Economics, engineer in Statistics and Econometrics, higher degree Mathematics at Ecole Normale Supérieure) is an Economics Professor at the University of Strasbourg. She is an econometrician working on family decisions and public policy evaluation in various domains such as transports, urban economics, real estate and labour markets, and development economics. She developed the LUTI (Land Use and Transport Interaction) software EUrbanSim and applied it to evaluate large transport projects in Paris region and other metropolitan areas. She is the co-founder, with André de Palma, of a spin-off, RiskDesign, which provides decision-support tools to private investors and has developed risk-attitude analysis tools currently used by more than 2 million people. She is a founding member, with, among others, André de Palma and Moshe Ben-Akiva (MIT), of the research initiative AICC (Action versus Inaction facing climate Change) and of the associated youtube channel https://www.youtube.com/@AICC-academia. They organize each year the AICC conference (see https://sites-aicc.sciencesconf.org/) and AICC symposium. Next one will be in Hamburg, in June 2027. Her research areas include Behavioral finance and Behavioral economics, Experimental economics, Decision Theory, Risk analysis, Collective and other Within-family decision models, Urban economics, Transportation economics, Development economics, Economic demography, Sustainable development and Climate Change.

Time and place: Tuesday, 16 June 2026, 3pm-4pm, room 260-210G (Board Room)

Speaker: Yunjie Shi [PhD Candidate in Economics, UoA]

Title: Learning Private Values in Public Good Contributions (joint work with Simona Fabrizi and Steffen Lippert [UoA])

Abstracts: We study voluntary contributions to a threshold public good by two players who learn about their private valuations through private signals. Only bad signals can arrive. The absence of signals makes players increasingly optimistic about their own valuations, whereas a prolonged history of no contribution makes them increasingly pessimistic about the other player’s willingness to contribute. We characterize the unique symmetric perfect Bayesian equilibrium. We show that equilibrium behavior evolves from waiting to partial contribution and ultimately to full contribution by a single player. Private-value learning can eventually eliminate free-riding in the private provision of public goods.

Time and place: Tuesday, 2 June 2026, 3:15pm-4:15pm, room 260-6115

Speaker: Matthew Ryan [AUT]

Title: Luce Models and Conditional Probability Space (joint work with José Rodrigues-Neto and James Taylor [ANU])

Abstracts: This paper connects some recent literature on stochastic choice with an older literature on conditional probability structures. In the stochastic choice context, Cerreia-Vioglio et al. (2021, Theorem 2) proved that Luce's (1957, 1959) famous choice axiom characterises a two-stage model of behaviour in which the “acceptable” alternatives are selected according to a rational choice function (Arrow, 1959), then “indifference” resolved randomly using a Luce model. In the conditional probability context, the choice axiom characterises a conditional probability system (CPS): Rényi (1955) and Császár (1955). The CPS structure was introduced into game theory by Myerson (1986) to model off-equilibrium beliefs. Two alternative structures that also permit conditioning on zero-probability events were subsequently proposed by McLennan (1989a,b) and Blume, Brandenburger and Dekel (1991a,b). Hammond (1994) proved that all three conditional probability structures are isomorphic, at least for a finite state space. We show that part of Hammond's isomorphism is mathematically identical to the special case of Cerreia-Vioglio et al. (2021, Theorem 2) in which the universal set of alternatives is finite. In the other direction, we show that Cerreia-Vioglio et al. (2021) can be used to extend a different part of Hammond's isomorphism to infinite domains. Finally, we observe that Rodrigues-Neto, Ryan and Taylor (2025b) permits a further generalisation of both results to arbitrary families of (finite) conditioning events.

Slides

Recording

Time and place: Tuesday, 10 March 2026, 3:15pm-4:15pm, room 260-6115

Speaker: Binyamin Oz [ISOM, Faculty of Business and Economics, University of Auckland]

Title: The Social Benefit of Priority Service

Abstracts: In most cases, society is indifferent to the specific order in which homogeneous customers are served in a queue. Hence, it is only the server who may benefit from selling priority service, and the revenue from such activity comes at the expense of reduced consumer surplus. In this work, we show that when strategic customers are faced with an additional decision, on top of whether to pay for priority service, selling priority service by revenue-maximizing servers may not only improve social welfare but even maximize it in some cases. We exemplify this principle by studying two models. In the first, a monopolistic server charges for priority while customers decide whether to join the queue, and if they do, whether to pay for priority. In the second, multiple revenue-maximizing servers compete on the priority service price, while customers decide which queue to join and whether to pay for priority at the queue they join.

Page updated

Google Sites

Report abuse