Papers

Generative models can reproduce an observational distribution while encoding an incorrect causal structure. We study a sequential game in which a structural causal generator proposes observational and interventional distributions, while an adversarial experimentalist selects interventions intended to maximally falsify the generator. The discriminator is therefore not merely a real-versus-synthetic classifier: it is indexed by an intervention and tests whether the generator reproduces the corresponding post-intervention law. We introduce Adversarial Causal Intervention Falsification (ACIF), formulate oracle and implementable versions of the game, and distinguish three objects that are often conflated: observational fit, interventional equivalence over an admissible query class, and point identification of a structural causal model. For finite model and intervention classes, we prove: (i) an exact reduction of the adversarial objective to a worst-intervention integral probability metric; (ii) identification up to interventional equivalence, with point identification under a separating intervention family; (iii) existence of mixed-strategy equilibria; (iv) finite-sample uniform convergence and margin-based model-selection guarantees; and (v) a logarithmic elimination guarantee for a disagreement-driven sequential design under a balanced-separation condition. We also give a complete linear-Gaussian example in which two observationally indistinguishable causal directions are separated by a single well-chosen intervention. The framework clarifies what an adversarial causal discriminator can and cannot certify, and provides a principled bridge between causal generative modeling, active causal discovery, and experimental design.

We study a binary choice problem in which an agent chooses between two actions whose payoff depends on a continuous state. The agent chooses how much effort to invest in learning about the state. Equivalently, we can think of the state as the strength of a stimulus, with the agent exerting costly effort to be more responsive to it. Taking as given the Fisher information cost introduced by Hebert and Woodford (2021), we analyze the optimal state-dependent choice rule using a variational approach. The main result is that agents' optimal response is an S-shaped function of the state under mild conditions. This prediction is aligned with the widely documented psychometric curve response profile observed in the experimental literature in psychology and economics.

Which social norms are self-correcting under rational learning? We show that conduct sustained by false beliefs cannot persist if a single departure from prevailing behavior generates evidence against those beliefs. We study the overlapping-generations learning model of Fudenberg and Levine (1993), in which finitely lived Bayesian agents are repeatedly and randomly matched with agents in other player roles, observe only their own matches, and learn from experience. In simple extensive-form games with nodewise-independent, nondegenerate priors, as agents live increasingly long lives and become sufficiently patient, every limiting game outcome is path-equivalent to a subgame-confirmed equilibrium. This establishes the converse of Fudenberg and Levine (2006). The mechanism is endogenous experimentation: uncertainty about the consequences of a potentially profitable departure gives patient agents an incentive to test it, generating the observations that correct beliefs and discipline continuation play.

This paper provides an axiomatic foundation for catastrophic thinking, a form of pessimism in which an agent evaluates uncertain alternatives by attending only to a subset of adverse outcomes. We introduce Catastrophic Attention Preferences (CAP), under which an act is evaluated by its subjective expected utility conditional on the worst outcomes, up to a subjectively determined probability threshold $q$. The resulting functional is a subjective counterpart of Expected Shortfall: both the agent's belief $\mu$ and her threshold $q$ are derived from preferences rather than assumed, without a probability distribution given as a primitive. Our main result is a complete behavioral characterization: six axioms, one of which, Catastrophic Complementarity, carries the behavioral content of catastrophic thinking, together with two standard richness conditions, are equivalent to the existence of a CAP representation, and the parameters $(\mu, q)$ are unique. The parameters are fully identified from probability equivalents of events, simple binary bets that can be elicited experimentally. We characterize comparative ambiguity aversion within the class: with common beliefs, ambiguity aversion is completely ordered by $q$; with different beliefs, we provide a necessary and sufficient condition on the two belief-threshold pairs. The model admits an equivalent multiple priors representation with a closed-form set of priors, nests subjective expected utility at $q = 1$, and converges to maxmin expected utility as $q \rightarrow 0$.

A foundational question in General Equilibrium theory—when is an equilibrium stable?—has remained largely unresolved for over 150 years despite important contributions. This paper proposes a resolution.

Equality before the law is a human right, and access to high-quality legal aid for indigent defendants is essential to enforce it. In a context where all defendants have access to a lawyer, I study the impact of denying legal aid on court outcomes. I combine double machine learning and a new administrative dataset linking aid to court outcomes in New South Wales, Australia, to learn the assignment function, whose inputs are known. I find that applicants who fail the means test and hire private lawyers are 10 percentage points less likely to be incarcerated than if they passed and relied on legal aid. Given an average incarceration length of nearly four years, this gap is significant. However, I find evidence suggesting that they spend more time in jail if they are incarcerated. A government preference for broad access to aid over allocated time per case could explain this pattern. Keywords: Indigent Defense, Crime, Criminal Justice. JEL: I30, K14, H44.

Choice alternatives are often multidimensional and risky. We introduce and axiomatize the structured multidimensional expected utility representation, a unified framework that generalizes existing approaches to evaluating such alternatives. The representation uses a rooted clustered tree to organize the joint, separate, and conditional evaluation of risk across dimensions within a common structure. We analyze the uniqueness of the representation and characterize useful special cases. We apply the representation to inequality across individuals, groups, and generations and to multisource income, characterizing the implications of bracketing for stochastic dominance and the avoidance of multidimensional risk.

Every SVAR result is conditional on two choices: the restrictions that identify the shock and the variables on which they operate. The literature disciplines the first; the second is chosen by hand. We develop a Bayesian methodology that constructs information sets, uses an out-of-sample criterion, and retains the largest system it admits. Under recursive identification, output rises with housing production rather than household credit alone. For monetary policy, an anchor-free joint Bayesian proxy SVAR with multiple instruments strengthens the credit spread channel. A core system augmented with the selected corporate spread identifies expected default risk as a potent transmission margin.

Representation theorems in decision theory establish that behavior satisfies certain axioms if and only if it can be rationalized by a well-defined objective. I argue that this “if and only if” structure provides a potentially useful foundation for label-free evaluation and regularization of LLMs and other AI systems. Axiom compliance can be checked from the model's own responses to synthetic choice problems, with no external labels or human feedback, and the penalties are readily computable. Because the axioms are necessary and sufficient, the resulting checks exhaust the implications of the relevant rationality standard for the elicited data: a model that passes cannot be rejected on rationality grounds by any further test of the same data. I discuss three instantiations: probabilistic coherence via a theorem of de Finetti, preference rationality via Afriat's theorem, and subjective expected utility via a theorem of Echenique and Saito (2015), each yielding a continuous penalty that is zero whenever behavior can be rationalized. Since coherence does not restrict which objective rationalizes behavior, these penalties complement rather than replace other evaluation and training signals.

I develop an estimation and inference framework for distribution regression in dyadic network settings with two-way fixed effects that vary across thresholds of the outcome. I show that identification of the structural parameters is achieved through binarization of the outcome at each threshold, and estimate the model by conditional maximum likelihood, which "differences out" the fixed effects and circumvents the incidental parameter problem. The estimator remains asymptotically unbiased under sparsity, whether from the network structure or binarization at extreme thresholds. The second novelty is to establish the joint asymptotic distribution of the estimators across multiple thresholds with different convergence rates, and to develop simultaneous confidence bands and tests for equality of coefficients across thresholds. Monte Carlo simulations confirm small bias, valid inference, and correct simultaneous coverage under sparsity. An application to bilateral trade finds that coefficients vary substantially across the distribution, with equality rejected for key trade barriers.

The United States is a currency union where multiple risk-sharing mechanisms— migration, fiscal transfers, income diversification and credit markets—buffer consumption from local income fluctuations. We show that risk sharing not only directly smooths consumption but also indirectly stabilizes income by dampening the local multiplier. Combining causal estimates from regional military buildups with a multi-region quantitative model, we find that current levels of risk-sharing cut state-level consumption volatility by a factor of four. Crucially, the indirect stabilization of income accounts for nearly half of this effect, implying substantially larger benefits from integration than conventional measures suggest.

Bergemann and Morris (2016) show that one information structure is more informative than another exactly when it induces a smaller set of Bayes correlated equilibrium outcomes in every game. We build the quantum analogue. An information structure becomes a family of density operators indexed by the payoff state, which the mediator observes. We show that obedience is equivalent to a Loewner domination between operators on one player's subsystem. The equilibrium set is then a nonempty compact spectrahedron computable by semidefinite programming, classical structures embed exactly, and under quantum individual sufficiency more information shrinks the equilibrium set in every game.

This paper explores how financial institutions pass interest rate risk through to product markets using the life insurance industry as a setting. We show theoretically that it is optimal for insurers to distort product issuance across maturities to offset duration gaps. We examine insurers exogenously exposed to interest rate risk through their variable annuity liabilities after the 2008 financial crisis. Consistent with our mechanism, exposed insurers developed negative duration gaps, increased markups on long-duration products, and rebalanced product issuance toward shorter-duration products to hedge. This response reduced long-duration life insurance coverage by 12.1% of GDP between 2005 and 2023.

In fixed-effect regressions with many groups, fixed effects can absorb most identifying variation, leaving a handful of observations to carry what remains. When variation is this concentrated, conventional $t$-tests can reject a true null more than half the time, and any fixed critical value is either invalid or so conservative it has essentially no power. This paper builds an exact test from the design alone. A nuisance-annihilating contrast is a linear combination of the treatment and fixed-effect dummies that eliminates the fixed effects without touching the outcome; sign-flipping these contrasts is then an exact symmetry of the null distribution at every sample size, under arbitrary heteroskedasticity. In two-way designs — worker-firm, firm-time — these contrasts are exactly the cycles of the bipartite mobility graph, so the movement that identifies the treatment effect is what makes exact inference possible. Exactness costs power: relative to an oracle test, a chosen set of cycles has an observable capture ratio $\kap\in[0,1]$ and standard-error premium $\kap^{-1/2}$, and a packing algorithm resolves the capture-granularity trade-off. In the Grunfeld investment regression (single-observation score concentration $73.9\%$), 32 cycle contrasts capture $\kap=0.627$ of the identifying variation, giving an exact $95\%$ confidence interval of $[0.150,\,0.450]$.

A common monetary standard can expand credit while weakening the alternative practices needed to recover from its failure and identify its cause. I study a growing credit economy in which safeguarded real settlements supply recovery capacity, matched observations for a reform decision, and apprenticeships in an alternative practice. The efficient reserve grows in absolute size but becomes a vanishing share of the economy. Protection based on a fixed population share therefore eventually fails, even along a path with expanding capacity and evidence. Voluntary cooperation provides only bounded readiness, while cheaper drills can preserve recovery without producing evidence. Apprenticeship and expected political survival reinforce one another, allowing exposed and protected continuations. I characterize protection indexed to the required service and the least-cost funding of replacement if the original institution is cancelled. The resulting arrangement combines verified procurement, training support when needed, and safe real collateral held by an outside custodian.

We develop a hierarchical Bayesian panel quantile regression model in which unit-specific coefficient paths are smoothed across quantiles by Gaussian processes, while a common time effect absorbs aggregate shocks. Componentwise-monotone Bernstein polynomials, perturbed by unit-specific deviations, deliver soft noncrossing, and we provide identification conditions together with a bound on the crossing probability. Applying the model to 33 countries over 1979–2023, we find that global temperature shocks generate a systemic, non-diversifiable downside risk to output growth. This risk is concentrated in the lower tail and disproportionately affects emerging markets. Finally, we apply our framework to risk analysis and show that the model reduces out-of-sample tail-risk forecast loss by roughly one-third relative to country-specific quantile regressions.

Estimating the dynamic effects of economic shocks in short and very short samples is impeded by a lack of degrees of freedom. We offer a solution based on a Bayesian hierarchical framework for estimating local projection (LP) impulse response functions across a panel of related time series. The framework explicitly accommodates unbalanced panels in which some series are substantially shorter than others, allowing the short series to borrow information from longer ones at horizons where the short series carry little or no own data. Since series might exhibit heterogeneous dynamics, we develop a sparse finite mixture pool that clusters units by similarity of their impulse response profiles. We show in simulations that our approach substantially improves LP estimation accuracy relative to the standard approach if the time series are short while producing similar LPs for longer time series. Using a US price dataset, augmented with survey responses, we find that supply-chain and oil shocks trigger heterogeneous reactions of different price measures, with headline price indices responding more sharply than their core counterparts and goods prices changing more than services prices.

Formula One outcomes reflect the joint contributions of drivers and constructors, but these contributions are unobserved and vary over time. We propose a Bayesian state-space model that disentangles dynamic driver and constructor abilities using two observed outcomes: fastest qualifying lap times and race rankings. Both outcomes depend jointly on latent driver and constructor states that evolve at the Grand Prix level, while the race equation additionally accounts for starting-grid position. The decomposition is supported by constraints that center the driver and constructor abilities at zero, together with variation in driver-constructor assignments over time. Bayesian inference is performed using the No-U-Turn sampler under weakly informative priors that treat driver and constructor abilities symmetrically. Applying the model to the Formula One hybrid era from 2014 to 2021, we find substantial heterogeneity in both driver and constructor abilities. Driver abilities are generally more stable over time, whereas constructor abilities exhibit greater variation and, for many driver–constructor combinations, contribute more strongly to observed performance.

Credit stress testing requires impulse responses of portfolio default probabilities, not only macro-financial drivers. We derive closed-form generalized impulse responses for the mean, quantiles (PD-at-Risk), and expected shortfall in a modular framework combining a Bayesian VAR, a Gaussian satellite, and the Merton-Vasicek model underlying Basel IRB regulation. Results extend to any probit-Gaussian mapping of a latent factor. Nonlinearity makes responses depend on conditional means and variances; plug-in evaluations understate projected default probability levels by 6-8% and miss tail quantiles. For U.S. geopolitical risk shocks, 99%-quantile responses exceed mean responses by 50%, and peak responses vary 4.6-fold across the credit cycle.

On two-sided content platforms, symmetric two-sided isolation (assigning matched fractions of creators and viewers to isolated treatment and control submarkets) is widely used for creator-side and cold-start experiments because it removes cross-arm marketplace interference. Isolation, however, thins each viewer's candidate catalog, and intuition suggests the resulting engagement cost should fade as the platform grows: a small fraction of a vast catalog is still vast. We show that, in an order-statistics model of engagement, whether this intuition holds depends on the upper tail of match quality. Extreme-value theory yields tail-class loss laws with a sharp dichotomy: for light or bounded tails the loss vanishes as the candidate pool grows, whereas under heavy tails it converges to a size-independent constant, so expanding the candidate pool, even by orders of magnitude, does not asymptotically eliminate the cost. Evidence from two production experiments on a platform with millions of active creators is consistent with this picture: a pure A/A traffic sweep reveals a measurable, depth-graded engagement cost; a one-sided catalog ablation independently shows that per-viewer thinning contributes to the loss; and a tail index calibrated on the small exploration pool predicts an effect consistent with the one observed in the far larger full-catalog ablation. Isolation thus carries a price that experimenters should budget for, like any other cost. We give practitioners a preflight procedure that estimates it before launch, sizes traffic accordingly, and recommends a fallback design when the predicted cost exceeds a chosen tolerance.

In this paper, I study to what extent countries differ in their preferences for quality and their technologies for improving quality. The paper also quantifies the contribution of those differences to the differences in gains from trade across countries. I adopt ANTONIADES, which allows endogenous quality choice of firms. This paper extends that framework into multi-country and multi-sector setting. The new feature is that bilateral trade liberalization yields spillover effects on the competitiveness of another country. This paper structurally estimates the model under multi-country and multi-sector framework. It is found that richer countries generally have stronger valuation for quality. The quantification demonstrates that variations in the strength of quality preferences and in the technology of improving quality across countries add to their heterogeneities in market competitiveness. Subsequently, I simulate a 5% increase in the trade barriers. If the quality channel is shut down, countries with stronger preferences for quality have larger degrees of underestimations in their losses from the trade barrier. Finally, gains from a universal rise in quality preference are unequal among countries, with larger economies generally gaining more than smaller economies.

We study auctions of k identical objects to n bidders, each of whom wants at most one. The objects have a common but unknown value and the bidders receive private signals about this value. The discriminatory price auction and the uniform-price auction are compared in terms of how informative the resulting auction prices (not bids) are in conveying the true value to an outside observer/investor. Since both auctions have symmetric, monotone equilibria, the problem reduces to comparing the informativeness of the highest order statistic of a sample to the (k+1)st highest. We find sufficient conditions under which the highest order statistic is superior—in the sense of Lehmann—in this regard. The sufficient conditions involve the informativeness of high versus low signals and the ratio k/n of objects to bidders. These conditions are also qualitatively necessary.

We comment on the optimally-transported generalized method of moments (OTGMM) estimator proposed by Schennach & Starck (2026a) and give counterexamples to Theorems 2-6 under their stated assumptions. First, the assumptions used in the small-error analysis are insufficient for consistency in Theorem 2 and asymptotic normality in Theorem 3. Next, we consider the large-error analysis, in which Theorem 4 states that the OTGMM estimator is equivalent to a GMM estimator with modified moments. We show that in a scalar model, Theorem 4 selects a value that differs from the unique OTGMM minimizer and violates the OTGMM sample moment restriction. In an overidentified model satisfying the assumptions used in Theorems 5 and 6, the first component of the Lagrange multiplier has different probability limits under the OTGMM estimator and the GMM estimator with modified moments. Under misspecification, the population value selected by OTGMM depends on the transport metric and on which variables may be adjusted. We give a sufficient condition under which solutions of the modified moment equations also solve the original constrained problem at a given parameter value, and separate conditions for consistency and asymptotic normality of the OTGMM estimator. We also show that Assumption 16 does not imply the matrix bound used in the supplemental proofs and replace Assumption 16 with a matrix condition that yields the bound.

Providers deploying imperfect AI must decide how much to keep humans engaged: engagement costs output but builds the fallback skill needed when it fails. A liability pledge whose expected cost falls in that skill restores single crossing on a type that is neither drawn nor embodied once in a good, but built, decaying and poachable. One engagement decision thus produces both the client-facing signal and the trajectory on which mobile workers sort. That conjunction is what is new. It yields a two-sided band of served work, a stakes threshold above which building beats free-riding, and a signal that never dies.

Objectives: While causal analysis of travel behavior is an emerging field, estimating heterogeneity in mode choice through causal modeling remains unexplored. This study demonstrates the application of a novel causal method, causal forest, to quantify the heterogeneity in travel mode choice shifts caused by the COVID-19 pandemic. Methods: We applied causal forests, a non-parametric causal machine learning method, to 802,935 trip records from the 2017 and 2022 waves of the National Household Travel Survey. The 2017 wave serves as the pre-pandemic control group, while the 2022 wave represents the treatment condition. Within the potential outcomes framework, we estimate average treatment effects (ATE), heterogeneous treatment effects (HTE), and conditional average treatment effects (CATE) across diverse socio-demographic groups and trip characteristics. Findings: Our results reveal an estimated ATE of a 1.86 percentage point (pp) increase in car-mode share, contrasted with decreases of 0.38 pp and 1.57 pp in public transit and walking, respectively. The largest increases in car use appeared for short-distance trips (one mile or less), households with annual incomes exceeding USD 200,000, and female travelers. Novelty: This is one of the first applications of causal forests to travel mode choice, and the first to use causal machine learning to estimate the pandemic's causal effect on mode choice analysis. Practical Applications: This study discusses methodological advantages, inherent assumptions, and limitations of causal forests within the context of transportation planning. This methodology is applied to COVID-19 travel data to illustrate how causal heterogeneity analysis can offer a deeper understanding of changes in mode choice. These insights are valuable for planners and policymakers in making policies related to mode shifts under an intervention.

Does generative artificial intelligence (AI) widen or narrow productivity gaps across workers? We study this in a randomized online experiment with 1,174 adults aged 25-45 who completed a workplace-style problem-solving task with or without a generative AI assistant, followed by an unassisted module. AI improves performance for all participants, but gains are larger among those with less education. Without AI, higher-education participants outperform lower-education participants by 0.548 standard deviations; with AI, the gap falls to 0.139, closing about three-quarters of the initial difference. Chat logs show that lower-education participants obtain substantial assistance, while higher-education participants use AI more effectively. Gains are not purely due to delegation: treated participants do not perform worse once AI is removed, and lower-education participants retain part of their improvement, although a sizable gap re-emerges. Intensive AI use raises assisted performance regardless of participants' own effort, but follow-up performance improves only when intensive use is combined with sustained effort. Generative AI narrows effective productivity differences in task execution, while human-capital differences continue to shape unassisted performance and tool use.

The set of $n\times n$ correlation matrices, known as the elliptope, has volume decaying at the super-exponential rate $\exp\{-\tfrac14 n^2\log n\}$. We characterize where this vanishing volume concentrates. A uniform draw is entrywise close to the identity yet globally far from it and nearly singular. Its maximum absolute correlation is of order $\sqrt{\log n/n}$, its Frobenius distance is asymptotic to $\sqrt n$, its empirical spectral distribution converges to the Marchenko-Pastur law with ratio one, and its smallest eigenvalue has the exact $\operatorname{Beta}(1,d)$ distribution, where $d=n(n-1)/2$, and is therefore of order $n^{-2}$. Distinct off-diagonal entries are exactly pairwise independent under every $\operatorname{LKJ}(\eta)$ law, which yields a Chen-Stein proof of the extreme-correlation point-process limit for the whole LKJ family and an $O(n^{-1})$ total-variation bound for finite-dimensional exceedance counts relative to Poisson laws with their exact finite-$n$ means. Finally, for a bounded, centered i.i.d. off-diagonal perturbation of any correlation matrix, the nearest-correlation projection removes a fraction of the squared perturbation that tends to one and recovers the original matrix in average squared error per entry.

This paper considers difference-in-differences identification strategies when the parallel trends assumption holds after conditioning on covariates that may themselves be affected by the treatment (often referred to as "bad controls"). We show that common approaches such as simply dropping bad controls are often ill-advised and develop two alternative approaches that allow bad controls to function as genuine controls despite being affected by treatment. First, we derive explicit conditions that rationalize conditioning only on pre-treatment values of the bad control, leading naturally to the Callaway and Sant'Anna (2021) estimator with pre-treatment values as covariates. Second, under a covariate unconfoundedness condition, we develop imputation and double/debiased machine learning estimators that recover the average treatment effect on the treated. We extend these results to staggered treatment adoption, provide pre-tests for the identifying assumptions, and apply the methods to study the effects of job displacement on earnings.

Empirical work often removes fixed effects, latent factors, or high-dimensional controls before estimating structural relationships. These transformations reduce confounding but may also remove identifying variation. We study linear panel IV after one equation-compatible nuisance projection under two-way dependence. The projected Jacobian determines which structural directions remain visible; the projected-score law determines their precision; and, on Gaussian fixed-rank strata, they combine in a Projected Information Matrix. We derive weak-identification limits with dimension-specific information accumulation, feasible factor-transfer conditions, identification-robust tests, bootstrap procedures for non-Gaussian interaction limits, and inference for the projected spectrum, rank, subspaces, and information matrix. Simulations show that a raw first-stage statistic above 500 can support the wrong sign while projected diagnostics reveal weak valid information. In an international monetary application, common projection substantially attenuates apparent foreign-output persistence, while Gaussian-reference Anderson–Rubin sets remain unbounded. Identification should therefore be assessed after nuisance removal.

In the standard Downsian model, two office-seeking parties converge to the median voter. However alienated voters may abstain and turn out only for a party within their tolerance radius. For single-peaked voter distributions, convergence survives but relocates to a central voter, the median of the electorate that participates at the convergent platform. However single-peakedness of the voter distribution is an empirically contested assumption. So we first characterize pure-strategy equilibrium for any continuous voter distribution. For general distributions, pure-strategy equilibrium can fail to exist or be non-unique, and existence of equilibrium need not persist as the tolerance radius of the voters increases. In order to resolve these issues, we propose a fundamental object: centripetal structure for which there is a single anchor platform toward which competition always pulls. We show this structure produces convergence at equilibrium under alienation based abstention. Our main result concerns the emergence and persistence of this new structure as the tolerance radius increases. Even though equilibria for office-seeking parties themselves can vanish and reappear as the radius grows, once centripetal structure emerges, it persists as long as the midpoint voter is not alienated. Moreover, the centripetal structure always emerges, and this structure classifies equilibrium completely when parties are policy motivated.