The Politician, the Liar, and the Obedient Worker: Emerging Behavior of LLM Agents in Hierarchical Games

From Luck to Pluck: The Politics of Australian Reform in an Uncertain Age—The 2026 Freebairn Lecture

In the 2026 Freebairn lecture, I argue that Australia's prosperity can no longer be taken for granted. Despite strong foundations—stable institutions, natural resources and a diverse population — the country faces weakening productivity, unaffordable housing, stagnant wages and a deteriorating global order. I contend that reform failure is not primarily a failure of individual politicians, but of systemic incentives—short election cycles, hyperbolic discounting, information asymmetry and media dynamics that reward outrage over nuance. Drawing on my own tax reform work, I identify four lessons for driving change: (1) start with ‘why’; (2) build unusual coalitions; (3) meet people where they are; and (4) prioritise effectiveness over elegance. However, I also argue better strategies alone are insufficient. I am calling for three institutional reforms: (1) clear outcome goals with rigorous ex‐post evaluation of spending and policy; (2) transparency by default across government processes; and (3) regulatory reform—including deregulation KPIs, sunset clauses and independent regulatory cost assessments. Australia must choose its future deliberately rather than rely on luck. Becoming The Plucky Country requires institutional discipline, political courage and genuine optimism.

Polistemics: Evaluating LLMs as Information Mediators in Politics & Elections

As LLMs increasingly shape the political information citizens rely on, no standard exists to assess whether they do so responsibly. We introduce Polistemics, a theory-grounded diagnostic benchmark for evaluating LLMs as mediators of political information in elections. Prior work has treated this task as reproduction rather than mediation, leaving its epistemic dimensions and interaction with imperfect information unaddressed. We ground the evaluation in Epistemic Modesty, a normative standard derived from citizens'epistemic agency, and test it across controlled settings that vary the clarity, noise, and consistency of the available evidence. Applying the benchmark to three state-of-the-art LLMs across the 2025 German and Dutch elections, we find that high aggregate scores mask systematic failures. Models mediate reliably under clear evidence but break down when it is absent, vague, or contradictory, while flattening the intensity of political language throughout. These failures point to party priors, shifting with party labels and output language. Reliable mediation appears achievable, but no model delivers it consistently.

Majority Lost: Global Evidence and Field Experiments on Electoral Miscoordination

Voters in multiparty systems often fail to coordinate on majority-preferred parties. Using constituency-level election data linked to party-ideology measures across 96 countries, we show that in 20–30% of multiparty elections since 1940 the Condorcet winner does not win, and in roughly 10% the winner is the Condorcet loser. Failures are most common under plurality rule and in fragmented, volatile party systems, and systematically disadvantage opposition and culturally liberal parties while benefiting organizationally strong ones. Can information fix this? In four pre-registered experiments around Indian elections (N > 4,000), constituency-level polls shift voting intentions by at most 2 percentage points, a precisely estimated null, despite modest belief updating, though undecided voters become 7 points more likely to support a viable party. A model with expressive motives rationalizes both results: polls refine beliefs but lack the leverage to overcome entrenched attachments, identifying boundary conditions on polls as coordination devices.

Contested Governance and Civilian Prosocial Behavior: Evidence from Lab-in-the-Field Experiments in India

How does contested governance by states and competing armed actors affect the behavior of the governed? During civil conflicts, rebel organizations frequently provide public goods, judicial arbitration, and security, but these forms of governance often operate alongside local state institutions. Beyond avoiding violence, therefore, noncombatants must navigate the daily uncertainty of local governance by states and armed groups vying for territorial control. This, we argue, shapes their quotidian interactions with one another. Using a series of lab-in-the-field experiments and surveys from Naxal-affected Bihar, India, we find that individuals living under such contested governance exhibit greater interpersonal trust and trustworthiness and less opportunistic behavior. We propose two mechanisms to explain these patterns—fearing intra-community retaliation or substituting for the state with interpersonal reliance. Most evidence is only consistent with the latter. By connecting governance, rebellion, and social capital, this work makes important contributions toward understanding the micro-politics of local order during longstanding armed conflicts.

The Power of Woke and Other Forms of Disproportionate Punishment

Tariff Threats, Macroeconomic Expectations, and Policy Communication Strategies: Experiments Based on a Multi-Agent System

Tariff threats can move household beliefs before policy is enacted, yet their rapidly changing language is difficult to study with conventional surveys. We build a multi-agent system that turns 300 households from the Michigan Surveys of Consumers into persistent large-language-model agents exposed to social-media information over several simulated months. Calibrated agents reproduce some distributional and demographic patterns in human survey data collected after the announcement of Liberation Day tariffs. Simulated experiments indicate that immediacy, rate salience, semantic progression, message complexity, narrative, and sender identity jointly shape inflation and unemployment expectations and their dispersion. Open-ended responses trace these effects to attention, ambiguity, credibility, and causal narratives. A second experiment finds that central-bank explanations can coordinate beliefs, although their effects on average expectations depend on message content. The framework supports disciplined exploration of policy communication, subject to human validation rather than as a substitute for it.