We propose a flexible framework for modeling the predictive distributions of nonlinear, possibly multivariate time series. Our approach expresses a general predictive distribution in an appropriate generative representation that is based on a folklore result from measure theoretic probability. This representation provides a direct simulation-based approximation to the predictive distribution, enabling straightforward computation of forecasts for the conditional mean and variance, fan charts, value at risk, expected shortfall, joint tail risks, and other quantities of interest. We estimate this generative representation using a version of conditional generative adversarial networks and provide a formal statistical analysis of estimation under weak temporal dependence. Specifically, estimation is expressed as a particular minimax problem and we establish consistency of its approximate solutions in Hausdorff distance. The empirical relevance of the approach is illustrated using applications to equity returns, realized variance, and realized covariances. The proposed method is also computationally manageable, with estimation in our applications taking approximately one minute on a standard laptop.

This paper defines the deep projections of the indifference and the offer hypersurfaces. These projections are used to measure how much the trade hyperplane must be curved to reach these other two canonical manifolds. In particular, it is shown that the factor $λ_{n}(p)\partial e(p,v_{n}(p))/\partial u>0$ measures how much more bent the offer hypersurface is relative to the indifference hypersurface, where $λ_{n}(\cdot)$ is the Lagrange multiplier and $v_{n}(\cdot)$ is the indirect utility associated with the normalized Walrasian demand $x_{n}(\cdot)$. These definitions and results are then applied to a consumption-loan overlapping generations economy to provide general statements for the sufficiency and necessity of the Cass criterion based on $\sum^{\infty}_{t=1}1/(\Vert p_{t}\Vert\sum_{h\in G_{t}}\Vert c^{h}_{t}\Vert)=\infty$ (thus allowing unbounded dynamics for both the demography and per capita endowments) under assumptions that reveal its local nature.

Central banks monitor macroeconomic risk through two traditions: scenario analysis, regularly used since the mid-1990s, and distributional forecasting, practiced since the late 1960s. The two are complementary but separate: scenarios provide narratives without probabilities, while predictive distributions provide probabilities with limited economic interpretation. Treating baseline forecasts and scenarios as conditional predictive densities, and distributional forecasts as reference predictive distributions, places both within a common framework and clarifies their roles. The Scenario Synthesis assigns weights to scenarios consistent with the reference distribution, offering a practical and reproducible tool for risk assessment and policy deliberation under deep uncertainty.

Energy policy is often guided by a small set of least-cost pathways to net-zero emissions, despite wide uncertainty in technology performance, fuel prices, demand and weather. To avoid overstating confidence in any single pathway, we quantify the likelihood of alternative technology pathways and identify the assumptions driving divergence, including the conditions under which technologies reach critical tipping points in competitiveness. We couple a sector-linked national optimisation model with Monte Carlo sampling (10,000 runs) across two European power systems (Germany and Great Britain) to generate probability distributions of capacity expansion and robust cost thresholds for key technologies. Results reveal substantial ambiguity in the future roles of wind versus solar, gas with carbon capture, and negative-emissions options. Tipping points vary widely with system conditions, while cross-country differences highlight the role of institutional constraints and resource endowments. Britain exhibits an either-or decision around nuclear power, investing if costs in 2035 fall below EUR 4700/kW, otherwise favouring offshore wind. Germany's uncertainty centres on dispatchable low-carbon options: gas with carbon capture (below EUR 2100/kW), biomass with carbon capture (below EUR 4200/kW), or hydrogen if electrolysis is below EUR 560/kW. We reframe scenario analysis as risk management by linking uncertainty to cost targets and minimum deployment requirements for robust net-zero strategies.

Risky consumption generates information when uncertainty is resolved. This paper axiomatically characterizes the consumption-information trade-off even when the analyst does not observe an agent's future problems. A subjective future menu underpins the agent's willingness to sacrifice current consumption for future information. By carefully separating objective risk from subjective risk, we decompose the certainty equivalent of an act into a standard risk premium and a novel information premium. To facilitate applications, we introduce an Arrow-Debreu-Pratt parameterization that yields a tractable model, capturing risk aversion and information incentives with a single coefficient for each. Finally, we show that heterogeneity in risk-taking may arise from differing opportunities to capitalize on information, rather than being solely attributable to differences in risk aversion.

This paper develops identification and estimation methods for dynamic partially linear logit models when social networks are endogenous and evolve over time. The outcome equation includes a lagged dependent variable and an unknown function of time-varying unobserved social characteristics that also govern link formation. Standard panel logit approaches, including those augmented with network controls, produce biased estimates when these latent traits are present. I show that combining conditional likelihood arguments with network-type matching across agents eliminates both individual heterogeneity and the unknown social influence function, achieving point identification of the slope parameters and the state dependence coefficient without imposing parametric restrictions on the network formation process. I propose a feasible kernel-weighted conditional maximum likelihood estimator that matches agents using codegree similarity and local smoothing over time-adjacent covariates. Consistency and asymptotic normality are established under weak regularity conditions. Monte Carlo simulations demonstrate that the estimator substantially reduces bias relative to standard dynamic logit specifications across a range of network formation mechanisms and sample sizes. An empirical application to adolescent smoking behavior using longitudinal friendship network data illustrates the method and suggests that standard approaches overestimate state dependence by confounding it with endogenous network sorting.

Over the past decade, U.S. policies have increasingly aimed to preserve artificial intelligence (AI) leadership by promoting domestic free-market policies while controlling global technological chokepoints, particularly advanced semiconductors and computational infrastructure. These measures raised the cost of Chinese AI development, but they also increased the strategic value of open and locally adaptable AI systems. Before raising export controls on high-performance chips, both the U.S. and China promoted policies that included support for open-source AI. During the period following major U.S. export-control shocks, China increasingly embedded open-source AI into national technology strategy through proposed ecosystem building, standards coordination, and resilience-oriented deployment. Moreover, Chinese developers increased engagement with open-source large language model repositories substantially more than U.S. developers did, consistent with a shift toward open infrastructure under geopolitical constraints. Subsequently, Chinese-origin open models diffused widely through open-source communities and scientific research. Even though such models remained largely absent from U.S. patent disclosures, American commercial entities use them in open-access research, suggesting their undermeasured importance within the foundation of U.S. commercial activity. These findings suggest that technological containment policies may unintentionally accelerate open innovation ecosystems as a competitive response, with implications for global leadership in both academic and commercial artificial intelligence.

Production is a sequence of steps that can be executed (1) manually, (2) augmented with AI, or (3) fully automated within contiguous AI-executed steps called ''chains.'' Firms optimally bundle steps into tasks and then jobs, trading off specialization gains against coordination costs. We characterize the optimal assignment of humans and AI to steps and the firm's resulting job structure, showing that comparative advantage logic can fail with AI chaining. The model implies non-linear productivity gains from AI quality improvements and admits a CES representation at the macro level. Empirical evidence supports the model's key predictions that (1) AI-executed steps co-occur in chains, (2) dispersion of AI-exposed steps lowers AI execution at the job level, and (3) adjacency to AI-executed steps increases the likelihood that a step is AI-executed.

A firm that discloses data about one customer moves a downstream seller's belief about every correlated customer, pricing third parties it never transacts with. This privacy externality equals the change in downstream deadweight loss, is signed by which side of the pricing threshold a customer is on, and falls hardest on those just carried across. Disclosure is privately optimal on an open set of imperfect correlations; incentive compatibility prices it through a distorted allocation and rations the discount at the top under a continuum of types. Selling tips disclosure past a liquidity threshold; consent dominates both data minimization and laissez-faire.

A game of information concerns two players transmitting messages that are obscured by noise. A receiver digests the combination of the two information sources and makes an assessment rationally. The aim of the players is to generate opposing assessments for the receiver by choosing signal-to-noise ratios of their information. It is shown that this problem can be reduced into an elementary infinite game on the square, thus admitting a complete equilibrium solution. Three generalisations of the game are proposed.

First, we present a new theory of "modular natives." A modular native is a basic building block that is born modular, e.g., a solar cell. The theory predicts that using modular natives in building things reduces complexity and improves predictability, resulting in better outcomes and faster scale-up. Second, we test the theory on the largest dataset of its kind. We find, at a high level of statistical significance, that modular natives operate under a fundamentally different risk regime than other project types, with finite and predictable risk, in contrast to non-natives that have infinite and unpredictable risk. The findings help explain why modularity is key to successful building while bespokeness often leads to failure. Third, we relate our findings to economic and geopolitical development, arguing that China understands modular natives and scale-up better than any other geography and that this is key to China's swiftly growing dominance in renewables, batteries, EVs, robots, etc. We argue that China's mastery of modularity and scale-up is a major innovation in its own right, among the greatest and most impactful in human history, falsifying the common notion that China cannot innovate. Business and government outside China ignore these findings at their peril. Finally, we spell out policy and practice implications and identify areas for further research.

This paper develops an approach for estimating demand for a new product. Taking willingness to pay (WTP) as primitive, it establishes a general and yet analytically simple demand function, and proposes an estimation procedure that consistently recovers the underlying demand function from the WTP data. Monte Carlo simulations find the estimation procedure works well in identifying the demand function. This approach complements existing methods of demand estimation, and can be applied both within and outside academia, for example in teaching economics, for a business to launch new products, and for policymakers to conduct non-market valuation.

The classical expected utility (EU) axioms are not sufficient for finite datasets. There are a number of anomalies (violations of EU) where axioms are satisfied. This paper studies axioms that are immune to this problem and definitively delineate between EU and non-EU. We discuss implications for experimental design and explore the automatic generation of anomalies.

Although discrete choice (choice-based conjoint) analysis has become a widely used technique for the elicitation of consumer preferences and hence a foundation for product design, to the best of our knowledge, there exists neither free and open-source nor commercial software that covers the game-theoretic simulation of competitive reactions among firms based on discrete choice models to improve decision making beyond traditional product (line) optimization. The R package cash (conjoint + Nash) does not only provide functions to fill this gap but comprises an entire simulation pipeline including the upstream processes of discrete choice analysis itself. cash ranges from preference generation, choice design, error and response simulation, through Bayesian model estimation and evaluation, to Nash equilibrium computation. Doing so, it partly draws from established R packages concerned with discrete choice analysis. While the structure of cash generally aims towards end-to-end simulation as well as simulation of competitive dynamics based on real data, all its key elements mentioned above may be of use independently of each other.

Organizations increasingly pursue multiple objectives, yet we know little about how boundedly rational firms search for better strategies when performance is multidimensional. We use corporate social responsibility (CSR) -- pursuing social and financial outcomes simultaneously -- as a motivating context. We distinguish multi-objective decision making (choosing among known alternatives) from multi-objective search (discovering alternatives through path-dependent local moves), and argue that preferences matter through the heuristics that implement them during search. In a dual-landscape NK simulation, we compare five prominent heuristics: Maximize (improve financial performance only), Combine (raise the sum of financial and social performance), Alternate (pursue one goal until stuck, then switch), Penalize (maximize financial performance while deducting shortfalls below a social threshold), and Satisfice (prioritize financial performance only after meeting a social threshold). Across varying complexity, goal correlation, and thresholds, these heuristics produce systematically different trajectories and joint outcomes. Importantly, by escaping local financial optima, Alternate often discovers "oblique" strategies that match or exceed the financial performance achieved by profit-only local search while also improving social performance. We conclude that implementing the same multi-goal preferences through different search heuristics steers firms toward different regions of the social-financial frontier, shaping both the compromises they reach and the opportunities they discover.

Inferring the strength of conditional dependence and testing conditional independence are fundamental problems in statistics. A recent breakthrough by Azadkia and Chatterjee introduced, for the first time, a conditional dependence measure that equals $0$ if and only if the variables under study are conditionally independent, and equals $1$ if and only if they are conditionally perfectly dependent. They further proposed a computationally efficient and strongly consistent estimator, $T_n$, based on an ingenious use of ranks and nearest neighbors. Despite these attractive features, the asymptotic theory of $T_n$ has remained largely undeveloped. This paper closes that gap. We prove that, under general dependence, $T_n$ is asymptotically normal and its limiting variance admits a closed form. We also construct consistent variance estimators that are computationally efficient and implementable in $O(n\log n)$ time. Taken together with existing bias-correction methods, these results provide a complete inferential theory for $T_n$.

We study how artificial intelligence (AI) interacts with social communication networks to shape the stability of collective knowledge. Agents exchange information through a network while receiving AI-generated content, and AI systems retrain on the aggregate social information they influence. This interaction generates two feedback forces: an AI contagion channel, through which distortions diffuse across the network, and an AI social distortion multiplier, through which retraining amplifies past errors. Despite the high dimensionality of the environment, we show that the long-run behavior of the system admits a two-dimensional representation whose spectral radius determines whether AI-mediated information systems are dynamically stable or unstable. We characterize a sharp regulatory frontier identifying the minimum filtering required for stability and show how network topology shapes systemic informational risk.

Large language models are increasingly used as binary classifiers when the true label is latent. We study partial identification of the prevalence $θ= P(X^* = 1)$ from panels of LLM reports whose errors may be arbitrarily dependent given the truth. The design of replication determines the observable, and hence the identifying content: repeated prompts to one model yield a count, several named models a response vector, and both a response matrix. Cast as a two-component finite mixture, the problem makes the identification failure transparent: absent restrictions that separate the latent components, the prevalence $θ$ is completely unidentified, and weak stochastic-ordering restrictions (first-order dominance, monotone likelihood ratio, mean ordering) leave the identified set at $[0,1]$. Identifying power comes instead from externally calibrated scores and events, which discipline the mixture in the spirit of the misclassification and corrupted-data literature. We characterize the resulting bounds, establishing validity and sharpness, and give an exact account of the identifying information in the full score distribution beyond its mean. When named models are asked repeated versions of the same question, what identifies $θ$ is not the number of positive answers but which models agree across prompts -- a feature a vote count discards. An extension derives implied bounds on regression coefficients when $X^*$ is a regressor of interest that is not directly observed.

In the archetype discovery problem a researcher wants to summarize N heterogeneous policy effects of interest that vary over a discrete set of covariates. The goal is to partition the set of covariates into K<N groups -- the archetype sets -- and to provide a summary of the policy effects for each group. We use decision theory to show that, under a weighted mean-squared-error criterion, a procedure analogous to the Sorted Group Average Treatment Effects (GATES) solves the archetype discovery problem. The key difference is that, in the optimal procedure, archetype sets are obtained by weighted K-means clustering of the N heterogeneous policy effects, instead of relying on K equally-spaced quantiles. We show that the procedure that minimizes average risk for a given prior can be obtained by clustering the different values of the posterior mean estimate of the policy effects of interest. Similarly, an approximately minimax procedure in large samples can be obtained by clustering a consistent estimator of the policy effects. In both of these cases, an exact solution to the weighted K-means clustering problem can be found using a simple and well-known dynamic programming algorithm.

This paper provides identification results to characterize a fairness-accuracy (FA) frontier, and statistical inference tools to test hypotheses and build a confidence set for the FA-frontier, when outcomes are observed only for selected individuals. When the selection process is unrestricted but loss is measured in specific ways, we provide a characterization of the sharp identification region of the FA-frontier. Under an assumption of unconfoundedness conditional on observables (and unrestricted loss functions), we obtain point identification and propose a debiased machine learning estimator, derive its asymptotic distribution, and show how this can be used to carry out inference for the FA-frontier. In work in progress, we extend the partial identification results to a broader class of loss functions.

Models whose predictions depend on only a handful of well-constrained parameter combinations, termed sloppy models, are ubiquitous in nonlinear stochastic systems. The information-geometric approach to sloppiness advocates using the symmetrized Kullback--Leibler divergence and its associated Hessian, the Fisher Information Matrix (FIM), as the natural loss function. However, prior applications have relied on analytically known or parametrically fitted distributions. In practice, for general agent-based or stochastic models the distribution must be estimated from simulation data. I demonstrate, using Kirman's ant recruitment model as a worked example, that a standard kernel density estimate (KDE) converges to the analytical FIM eigenvectors and eigenvalues with simulation budgets accessible in practice. I derive the analytical Hessian in closed form, show numerical convergence of the KDE-based estimate as a function of simulation data, and demonstrate how the stiff direction enables efficient phase exploration across the model's unimodal and bimodal regimes.

When a parent company acquires a horizontal competitor on the same side of a multi-sided market, it must decide whether to fully integrate the acquired platform or keep it as a separate brand. We study this in the context of Uber's acquisition of Postmates, using novel consumer receipt data that tracks food delivery spending. Employing an Age-Period-Cohort (APC) decomposition, we isolate the merger's effect on consumer spending while controlling for lifecycle and cohort effects. We find that Postmates users sharply reduced their spending on the platform after the merger, but spending shifted not only to UberEats, but also to competitors like DoorDash and Grubhub. Consumers who used multiple platforms and had low pre-merger activity on Postmates were more 'sticky', showing little change. Comparing our APC results with a standard Difference-in-Differences (DiD) design, we find the DiD underestimates the merger's total impact by missing market-wide effects. Our findings suggest that in multi-sided markets, keeping acquired platforms separate can be beneficial; dissolving them may push demand to competitors, and some sticky multihoming users may not shift spending at all.

Agentic AI systems are increasingly being deployed as productive resources in organizational workflows, yet existing evaluation methods primarily measure isolated technical performance rather than economic contribution. This paper introduces \emph{Agentomics}, a workflow-based framework for valuing, attributing, and pricing human and artificial agents. The framework models a workflow as a configuration of heterogeneous agents whose collective performance determines gross value, deployment cost, reliability, and expected failure loss. Workflow value is treated as a team-level quantity that may include complementarities, substitution effects, bottlenecks, and nonlinear production; additive stage-level value is only a special case. Building on this workflow model, the paper formulates AI deployment as a coalition-formation problem and defines coalition value as the incremental net surplus generated relative to a benchmark human workflow. The Shapley value is then used to attribute economic surplus among participating AI agents, yielding a principled connection among valuation, accountability, and market pricing. The resulting Shapley pricing equilibrium provides a normative benchmark for assessing whether agent prices reflect expected marginal contribution. A security-operations case study illustrates how the framework accounts for productivity gains, deployment costs, reliability losses, and coalition-level complementarities in hybrid human--AI workflows.

Economic decisions often depend on many noisy estimates of quantities such as neighborhood effects, school quality, and hospital performance. Shrinkage estimation can improve decisions by pooling information across related units, but geography, adjacency, and shared characteristics each define a different notion of relatedness, and each implies a different way of pooling. We treat the choice of relatedness as part of the estimation problem, using Stein's Unbiased Risk Estimate (SURE) to form a weighted average over a library of flexible shrinkage estimators. This comparison among the candidate estimators treats no prior or latent covariance structure as a correctly specified model for the parameters being estimated. Each candidate is judged by its SURE value. Under smoothness conditions on the estimators, the SURE-weighted average performs nearly as well as the best fixed weighted average of trained candidates, including nonlinear rules whose reported values use the full vector of noisy estimates. In an application to Opportunity Atlas economic mobility data from 20 commuting zones, the best individual spatial specification varies across zones, yet the SURE-weighted average tracks the best in each zone and reduces estimated mean squared error by about 27% relative to the best-performing non-spatial empirical Bayes baseline in our library of estimators.

Teams often compete on multiple fronts: parties contest districts for majority control, contractors field specialized units for procurement awards, and squads play match by match for titles. The prize accrues collectively, but incentives depend on its internal division. We study a majoritarian team contest in which two rival managers simultaneously split their team-prize budgets among heterogeneous members, each battle resolved by a homogeneous-of-degree-zero technology. Under a log-odds curvature condition, a unique pure-strategy equilibrium exists: both managers choose identical relative allocations -- whatever the heterogeneity in costs or values -- with each battle's share proportional to its discriminatory power, closeness, and pivotality.

Advances in quantum computing challenge the hardness assumptions underlying widely deployed public-key cryptography in blockchain systems. Although post-quantum cryptography (PQC) standards are emerging, understanding quantum risk remains fragmented across research, engineering, governance, and investment communities. This demo presents Quantum Futures Interactive, a live interdisciplinary demonstration combining educational visualization, participatory interaction, and demonstrative post-quantum artifact generation using a toy LWE-based construction. Participants engage in a structured seven-stage interaction flow covering quantum threat education, sentiment capture, technology prioritization, infrastructure tradeoff exploration across simulators and QPUs, and artifact generation. The system integrates distributed trust concepts and sustainability-aware infrastructure considerations within an interactive decision framework.

I study how organisations choose selection procedures in a competitive environment. Two firms compete to hire candidates of unknown productivity from a common pool. Firms simultaneously post a selection procedure which consists of a test and an acceptance probability for each test outcome. After observing the firms' selection procedures, each candidate can apply to one of them. Firms can vary both the accuracy and difficulty of their test. The firms face two key considerations when choosing their selection procedure: the statistical properties of their test and the selection into the procedure by the candidates. I show that there is a unique symmetric equilibrium where the test is maximally accurate but minimally difficult. Intuitively, competition leads to maximal but misguided learning: firms end up having precise knowledge that is not payoff-relevant. In contrast, when firms face capacity constraints or have the possibility of making a wage offer, they use more difficult tests in equilibrium. I also consider asymmetric equilibria where one firm is more selective than another.

This paper studies sensitivity analysis in stochastic frontier models by developing relaxations of the baseline assumptions imposed on the latent inefficiency and noise components, and characterize bounds for a benchmark technical-efficiency object under such relaxations. We then derive the associated breakdown frontier for conclusions about conditional technical efficiency and illustrate the procedure using a well-known dataset. We show the estimation and inference of the breakdown frontier. We also extend the analysis for widely used alternative specifications of the stochastic frontier error structure, that is, the Normal-Truncated Normal, Normal-Exponential, and Normal-Half Normal cases. Finally, we suggest avenues for extending this analysis under heteroskedasticity, the relaxation of input exogeneity, and applications using panel data. Code for empirical implementation is also provided.

Stochastic frontier models have attracted considerable attention due to the incorporation of an inefficiency term in addition to the conventional error term. In this paper, we propose a general estimation framework for panel stochastic frontier models that accommodates potential heterogeneity through latent group structures. The framework is tailored to the distinctive features of stochastic frontier models and is paired with a practical hybrid estimation procedure that combines individual-level and joint panel estimation. We illustrate the estimation framework using a panel stochastic frontier model that treats the inefficiency term as a random effect, and show that it can be readily extended to a range of fixed effects specifications common in the literature. Simulation studies indicate strong finite-sample performance, and we further demonstrate the practicality of the approach in an empirical application to the cost efficiency of the U.S. commercial banking sector.

We revisit size controllability results in Pötscher and Preinerstorfer (2025) concerning heteroskedasticity robust test statistics in regression models. For the special, but important, case of testing a single restriction (e.g., a zero restriction on a single coefficient), we povide a necessary and sufficient condition for size controllability, whereas the condition in Pötscher and Preinerstorfer (2025) is, in general, only sufficient (even in the case of testing a single restriction).