Papers

Artificial intelligence capabilities are often measured as resources accumulated within firms, yet they also circulate across organizational boundaries through worker mobility. Drawing on a database of approximately 535 million employment records across 58 countries, we reconstruct the inter-firm network of AI talent flows over 2010–2022. We find that AI talent inflows are becoming increasingly concentrated among a small set of leading firms, with stronger preferential attachment than in general labor mobility. Yet the network core remains contestable: central positions in AI networks change hands far more often than in non-AI networks. Network position also carries information that goes beyond workforce size. This network position is economically consequential: AI network centrality is associated with higher enterprise value, after accounting for workforce scale, assets, and fixed effects. Event-study estimates further indicate that sharp improvements in network position are followed by higher enterprise value relative to matched firms with similar pre-event trajectories. Together, these findings suggest that corporate AI advantage depends not only on how much talent firms accumulate, but also on where they sit in its circulation.

North-star metrics such as customer lifetime value are often too slow and noisy to decide a short A/B test. Teams therefore rely on a proxy metric, commonly chosen by how closely its effects tracked the north star's across past experiments. Validating that choice, or any method for making it, is hard: the only benchmark is the noisy north star, and the number of available past experiments is limited. In addition, proxy and north-star effects are estimated on the same customers, so their sampling errors are correlated. Recent work at major experimentation platforms removes this shared error as contamination, improving estimates of the true-effect covariance. Choosing a proxy, however, is a ranking problem, and a better estimate need not give a better ranking. We measure agreement free of shared error by estimating the two effects on disjoint random halves of each experiment's customers. In an archive of 262 experiments and 69 candidate proxies, the shared error ranks the candidates in a similar order to this agreement (Spearman correlation 0.65): it carries information about proxy quality. The more of it a correction removes, the worse the ranking because removal discards part of the signal but leaves the main sources of ranking noise, the noisy north star and the limited number of experiments, untouched. Archive-calibrated simulations, in which the correct ranking is known, confirm this even when every correction receives the true sampling covariance. Held-out real experiments, evaluated on disjoint customer halves so that shared error cannot bias the comparison, closely reproduce the predicted ordering (Spearman correlation 0.93). Correction can still pay off with more experiments, but the number needed rises steeply with the north star's noise. We map this crossover and give platform teams three inexpensive checks for deciding from their own archive whether and how strongly to correct.

The growing importance of intraday trading in Europe, driven by the increasing penetration of renewable energy sources, has led to higher volatility and periods of market stress. Understanding how order flow behaves under varying liquidity conditions requires models that adapt to the state of the market. This paper proposes a multivariate state-dependent Hawkes process to model the flow of orders across different liquidity regimes in the intraday continuous electricity market. The observed market states are identified using a proposed Liquidity Stress Index (LSI), which quantifies stress using the bid-ask spread in the limit order book, order book volume, and mid-price volatility. Transitions between states induce changes in the intensity parameters, resulting in piecewise dynamics that reflect regime-specific order-flow behavior. The model is calibrated on January-March 2024 Dutch intraday continuous cross-border intraday (XBID) hourly order data from EPEX SPOT. The estimated dynamics show that order flow is strongly self-exciting and near-critical, that executed trades trigger new same-side orders while the reverse effect is negligible, and that liquidity stress mainly strengthens transaction self-excitation rather than reshaping the cross-side structure. Robustness checks over the LSI composition and the state partition show that the results are driven by the regime construction itself: alternative index compositions fit within a narrow band, three states fit better than two, and removing the states clearly degrades the fit relative to a single-regime Hawkes benchmark. The results highlight both the value of conditioning order-flow models on liquidity stress and the limits of a single-exponential specification when event frequencies are highly imbalanced.

This study examines whether universities shape the regional innovation effects of public R&D expenditure and whether this role varies across regional development trajectories. We analyse a panel of 1166 EU NUTS 3 regions over the period 2007-2020, combining data on European public R&D funding, patenting performance, university presence and regional development traps. Toa ddress the endogeneity of public R&D expenditure, we employ a control function aproach. The results show that public r&D expendirue positively affects patenting performance in both university and non-university, but the difference between them is not statistically significant. University presence therefore does not sistematically amplify the returns to public R&D. However, its moderating role becomes more pronounced in regions exposed to development trap risk and in regions experiencing persistent stagnation, while evidence is weaker for trap intensity. These fundings suggest that universities act as contingent rather than unviersal catalyst of public R&D. Their contribution appears particularly relevant in vulnerable and persistently stagnant regional trajectories, but less able to offset sever structural constraints. The study contributes to debates on the geography of public R&D and regional development traps by showing that the effectiveness of research funding depends not only on institutional presence, bu also on the development rajectory in which that institutional capacity is embedded.

We give a unified analytic account of a finite bid–ask spread in a two-field reaction–diffusion order book. The model retains separate bid and ask densities in operational time, with diffusion, cancellation, reaction and external order creation. On the symmetric equal-coefficient branch, the imbalance field determines the reaction price, while the total standing density carries the spread geometry. A market-making source is introduced by withholding order placement over a finite interval around the reaction boundary, representing round-trip replenishment displaced from the quote. Under a separated, quasistationary and weak-overlap approximation, transport penetrates inward from the placement edges over the cancellation length, giving the quoted spread as the sub-threshold core of the placement-source-free interval. Market orders remove standing density at the observed quote, whereas replenishment is delivered outside it and must diffuse back across the penetration depth. We derive the resulting delayed and attenuated quote-level response in operational time. On that branch, finite spread and price impact arise from distinct response sectors, while their calendar-time appearance requires a separate observation clock.

The middle-income trap is usually identified by comparing a country's income to a threshold. This reveals that convergence has stalled but not where the blockage lies, and it leaves unexplained why the trap is still debated in countries that have already passed the threshold. This study treats the trap as a configuration problem, measuring development along three dimensions - future, outside, and inside potential - with indices built from standardised indicators. The indices are calculated for 34 OECD members in 2000, 2010, and 2020, so that changes in configuration become visible, and the resulting country clusters are interpreted through factor analysis. Two traps and two transition strategies emerge. The first strategy attracts foreign investment and technology, but can itself become a trap: domestic value added stays limited and rising wages erode export competitiveness. The second adds a developmental state focused on the quality of education, building the human capital that raises productivity and innovation. All six CEE members of the OECD had adopted the first strategy by 2010; by 2020 only Estonia had completed the second. Configuration levels and changes track relative income convergence.

This paper proposes a framework for modeling probability of default via functional data analysis. By representing a series of credit variables as functions, we investigate whether intra-monthly information improves default predictability in linear models. We further show how a range of widely used variables, among them available funds, utilization rate, and overdraft, can all be derived from three quantities observed over time. Namely, 1) the type of each account, 2) the balance of the account, and 3) the size of the credit limit on the account. Retaining these quantities as continuous-time processes, rather than reducing them to monthly aggregated values, yields a continuous faithful representation of the borrower. When analyzing these three processes, we discovered that recurring events associated with the ordinal position among banking days created strong cyclical patterns. We therefore develop a relative time framework that aligns the recurring events across borrowers, ensuring that borrowers possess the same cyclical pattern, regardless of real time. We assess the framework using functional logistic regression. This approach accommodates the continuous representation while remaining closely related to a logistic regression model commonly used in credit risk practice, due to strict regulatory constraints. We show that, when equipped with an effective functional representation of transactional trajectories, the proposed model attains predictive performance on par with XGBoost while consistently outperforming logistic regression. Importantly, the model balances predictive performance of a machine learning model with the interpretability of linear default models, potentially enabling financial institutions to use the model, even under strict regulation.

In actuarial science and quantitative finance, convex order provides a natural way to compare risks with the same mean. In dimension one, convex order is well understood through several characterisations. In higher dimensions, a natural approach is to compare all one-dimensional projections, but, although necessary, the resulting condition is in general not sufficient for multivariate convex order. A simple counterexample due to Pinelis exploits Popoviciu's inequality. We revisit this example using gauge functions, and show that gauge functions themselves do not suffice to characterise multivariate convex order for symmetric distributions, even in dimension two. We then show that the radial case makes the source of the failure transparent and construct radial counterexamples in every dimension at least two. Nevertheless, projection tests are sufficient for some important families of distributions. In particular, this is the case for linear transformations of a common radially distributed random vector. This framework includes elliptical distributions generated from a common radial distribution and, in particular, centred Gaussian distributions. Beyond comparisons between Gaussian distributions, Jourdain and Pagès recently studied the comparison of a Gaussian distribution with a Gaussian mixture, obtaining necessary conditions and sufficient conditions whose equivalence was left open in general. In this paper, we settle the comparison of a centred Gaussian distribution with a mixture of two centred Gaussian distributions: in every finite dimension, the Gaussian is dominated by the mixture in multivariate convex order if and only if all one-dimensional projections are ordered in convex order.

Mitigation pathways require fossil power plants to retire well before the end of their technical lives, yet retirement decisions are typically derived from cost-minimizing system models rather than from the decision facing the plant's owner. This study investigates the carbon price threshold at which closure becomes more advantageous for an operator, as well as the ongoing costs associated with continued emissions. A finite-horizon Markov decision process is solved plant by plant for the European fossil fleet of 850 units, with a carbon price that moves stochastically and plants that age deterministically. The price process is constructed so that widening its range leaves the expected price unchanged, isolating the effect of uncertainty from the effect of price level. Closure thresholds fall below recent EU allowance prices for all fuels considered, and the sensitive range lies well beneath them, so further price increases are unlikely to be the binding instrument for retiring this fleet. Because fuel costs rather than emissions determine how much margin a carbon price must erode, the least carbon-intensive capacity retires first. Uncertainty raises the thresholds by about a tenth at an unchanged expected price: a credibly stable carbon price retires the same capacity as a volatile price roughly a tenth higher. Policy credibility therefore substitutes for price level.

Deep hedging learns trading policies from historical or simulated market trajectories, yet under nonstationarity these training paths may not represent future market conditions. We propose WRAP (Wasserstein-Reweighting Adversarial Perturbation), a drift-aware adversarial training framework derived from a two-budget distributionally robust optimization (DRO) formulation. The formulation is anchored to a weighted empirical reference distribution whose fixed baseline weights are chosen to balance sampling uncertainty against temporal drift. Around this reference distribution, the ambiguity set addresses two complementary forms of distributional misspecification by allowing an adversary to reweight the observed trajectories subject to a $\phi$-divergence constraint and perturb their paths subject to an optimal-transport (OT) constraint. We derive a joint first-order expansion in which the leading-order increase over the nominal expected loss decomposes into a reweighting contribution determined by the dispersion of hedging losses across trajectories and a transport contribution determined by the sensitivity of the loss to path perturbations. This expansion yields an explicit finite-dimensional adversarial attack that replaces the distributional inner supremum with a tractable first-order approximation. Across stationary and nonstationary Heston dynamics and a generalized affine diffusion (GAD), the experiments show complementary benefits from reweighting and transport, with joint adversarial training providing the largest gains under nonstationarity.

Experimental studies of generative AI in education mostly report average effects, yet field evidence shows that AI can narrow attainment gaps or widen them. We argue that the direction depends on the judgement burden, the expertise a learner must supply to screen AI output before learning from it, and that expert verification before release moves this burden from students to an accountable tutor. We test the argument in a two-cohort difference-in-differences design in which one half of a compulsory firstyear university economics course received AI-generated podcasts, FAQs and quiz-based study guides, produced with a source-grounded model and checked by a named graduate teaching assistant (170 students; 340 examination marks). Access was associated with a 2.34-mark advantage on a 50-mark component. The share of marks below the upper-second classification boundary fell by 24.7 percentage points relative to the counterfactual, effects were significant at every threshold from 23 to 31 marks and at none above, and roughly three-quarters of the average originated in the bottom quintile. The threshold estimate is robust to removing the lowest-scoring students from the pre-intervention cohort; the average effect is not. Interviews and feedback from 36 students indicate that the verification label gave students a reason to engage with AI-generated material without ending their scrutiny of it. Evaluations of AI learning resources that report only mean effects cannot detect whether the students the resources are meant to help are the ones who gain.

Compared to self-employment or performance-based pay, fixed wages reduce earnings uncertainty but are prone to market distortions through moral hazard and adverse selection. Using a model of wage contracts under asymmetric information, I show that these distortions map to marginal treatment response functions in a marginal treatment effects (MTE) framework. I apply this framework to a field experiment in which data-entry workers choose between a randomized fixed wage and a piece rate. I find evidence of both moral hazard and adverse selection: Fixed wages reduce worker productivity by an estimated 8.74 percent relative to the mean. Meanwhile, a 10 percent increase in the wage offer attracts a marginal worker whose productivity is higher by 1.90 percent of the mean. Using semi-parametric MTE estimation, I calculate the welfare loss associated with asymmetric information and the marginal values of public funds (MVPFs) for policies aimed at recovering this loss. I find that a tax on piece rates can mitigate adverse selection into fixed wages, generating tax revenue at social costs as low as $0.88 per dollar.

Unions are important economic and political actors at the local level, but the primary source on US union membership, the Current Population Survey (CPS), is unreliable below the state level. Researchers have therefore turned to union-reported membership in federal "LM" filings, geolocating unions and aggregating their reported membership to estimate union density for counties, commuting zones, and congressional districts. We ask whether this practice is sound. We develop and defend multilevel regression with poststratification (MrP) on the CPS as a benchmark, then validate LM-based estimates against it from the state level down. LM, CPS, and MrP are highly correlated at the state level, but this correlation is roughly halved at the congressional-district, commuting-zone, and county levels. The breakdown is driven in part by reporting "lumpiness," which produces improbable values, including union densities exceeding 100%. Missing public sector union members explains only a small part of this divergence. These issues are substantively consequential: in a standard commuting-zone "China shock" regression, swapping density measures changes the import-exposure coefficient enough to alter its magnitude, significance, and sign. We conclude that LM records remain a rich resource for studying unions as organizations, but aggregating LM-reported membership is not a reliable way to measure union density at fine geographic resolution.

Population prompts are widely used to generate synthetic survey responses, but they combine information supplied at inference with associations already encoded during pretraining. We introduce an alternative construction that maps declared aggregate preference anchors into group-indexed choice policies. For each population, the signs of six Global Preferences Survey (GPS) coordinates deterministically label a shared bank of paired synthetic responses, and Direct Preference Optimization fits a parameter-efficient adapter to those comparisons. We evaluate the adapters on candidate World Values Survey (WVS) items using prompts that omit country names and distinguish four questions: recovery of the imposed labels, transfer of the anchor signal to new text, coherence between the GPS anchors and human WVS responses, and agreement between adapter and human scores. The adapters recover the imposed pairwise labels. On a purposively selected sixteen-country development panel, adapter trust scores completely separate the two GPS-sign groups and have a rank correlation of (0.74) with continuous GPS trust scores. Human-GPS and adapter-human associations remain unresolved on the same panel, and results for the other preference dimensions are heterogeneous. These findings show that an anchored policy can retain a declared aggregate signal without thereby reproducing human response patterns. The contribution is therefore both an inspectable construction and an evaluation framework that separates anchor transfer from human criterion agreement.

This article studies a retirement planning problem from a new perspective in which a retiree delegates an initial lump sum to a professional fund manager. The fund is managed dynamically to deliver lifelong benefits while satisfying a guarantee that, at all times, wealth remains above a prescribed solvency level and the benefit rate remains above a minimum level. We formulate this problem as a continuous-time stochastic control model, in which the controls are the benefit rate and the portfolio allocation, and the objective incorporates general utility functions over benefits, management fees, and terminal bequest. Conventional approaches such as standard viscosity solution and martingale methods are not immediately applicable here due to the simultaneous influence of (i) the control variable nature of the benefit rate; (ii) the guarantee constraints; and (iii) the running wealth-dependent utility. We develop a new approach that deals directly with the associated fully nonlinear Hamilton-Jacobi-Bellman equation and establish the unique existence of a classical solution. One particular trick for resolving this equation is to apply an intricate transform that diverts the former to a semilinear auxiliary equation. A major contribution is to establish the existence of the solution to the auxiliary equation and characterize its precise growth, derivative, and boundary behavior. These results are crucial for establishing the verification theorems and deriving the optimal strategy. Besides, numerical experiments indicate that guarantee constraints induce more conservative policies, particularly at low wealth levels, while providing effective downside protection for both wealth and benefits. The associated well-being loss relative to the unconstrained framework remains small, suggesting that incorporating guarantees into retirement products is practically appealing.

We develop a posteriori primal-dual bounds for numerical approximations of European option prices in the Uncertain Volatility Model. Given a smooth candidate approximation of the value function, a feedback control induced by its Hessian yields a primal lower bound. On the dual side, we derive a representation based on a matrix-valued Gamma field, which provides an upper bound through a nonnegative Hamiltonian penalty. We further show that, when the dual field is generated by the candidate itself, this upper bound admits an equivalent representation in terms of the residual of the associated Black-Scholes-Barenblatt equation. We then study discrete-time approximations of these primal and dual quantities and quantify the corresponding discretization errors. Finally, we investigate their numerical evaluation for candidates obtained by stochastic policy-gradient and physics-informed neural-network methods, and compare the resulting estimates with a martingale dual approach from the stochastic-control literature. The numerical experiments highlight the importance of derivative accuracy, and in particular of second-order information, for obtaining tight a posteriori bounds.

Trade-sign errors can change measured trading costs even when classification accuracy is high. We validate the side of Polymarket's public trade prints against the taker leg of each print's on-chain settlement. On twelve selected days between April and August 2026, spanning both exchange generations, 92.1% to 100.0% of prints match a settled taker leg, with exact side and token agreement on all 24.5 million matched pairs. Mint-and-merge settlement makes that taker leg essential: pooling maker and taker legs changes the measured buy share. Signing the full cached tape before settlement selection yields 16.6 million prints signed by every rule; equal-day balanced accuracy is 0.942 for Lee-Ready, 0.768 for the tick test and 0.670 for retrospective bulk volume classification. On 16.4 million identical eligible fills, Lee-Ready raises effective spread by 0.429 cents per share under equal-fill weights; its realised-spread difference is -0.209 cents. Under share-volume weights, taker five-minute midpoint impact is 0.689 cents, while tick and bulk classifications give -0.374 and -0.119 cents. That aggregate sign reversal disappears when tied print rows are excluded. Distortion depends on error-weighted signed outcomes, sample selection and weighting. Receipt-time ordering and unobserved future-quote age limit these delivered-quote accounting quantities; they do not identify causal impact or private information. Supporting venue and collector analyses provide descriptive diagnostics.

Administrative boundary reforms in China alter the assignment of local authority and growth incentives without changing physical geography. We study two reforms with opposite institutional content—county-to-district (centralization: a county absorbed into a municipality as an urban district) and county-to-city (empowerment: a county upgraded to a county-level city)—using a firm-year pollution panel (1998–2013) and a border-pair research design that compares firms on opposite sides of the same county border. Centralizing reform lowers local NOx intensity by 14.8% through cleaner production, without shrinking output or employment; empowering reform raises COD, dust, and wastewater levels by 87–95% through dirty growth. The local abatement is not net abatement: NOx rises in same-prefecture neighbor counties (+0.77, p<0.01), reallocation follows administrative boundaries rather than geographic distance, is confirmed in satellite ambient data, and is not driven by wind or water transport. The reallocation operates through incumbent neighbor firms becoming dirtier, not through firm relocation or new entry. A back-of-the-envelope welfare calculation yields a statistically insignificant welfare gain from the only significant abatement (NOx, \textyen 47 million/yr, CI crossing zero) and a damage of -\textyen 2.3 billion/yr from county-to-city reform. These findings show that boundary reforms reshape the spatial configuration of pollution across administrative boundaries rather than reducing it.

Entropic risk optimisation is a general framework for pricing and hedging financial derivatives in incomplete markets that can be used to decompose P&L into market and model risk contributions. When the prices are quadratic Gaussian, the coupled equations for price and hedge ratios are solved in closed form. This enables comprehensive analysis of trading P&L, with a decomposition of the model value-at-risk into convexity, dimension and funding contributions that are attributed in the explanation of realised P&L. The equations of the quadratic Gaussian model are directly applicable when the underlying prices follow Gaussian processes, such as fractional Ornstein-Uhlenbeck processes. The model also provides simple parametric expressions for hedge ratios that can be used for regressions in deep hedging.

Personality is associated with economic outcomes, but whether these relationships are causal remains unclear. We exploit within-family variation in genetic propensity toward the Big Five generated by random genetic transmission from parents to offspring. Combining polygenic indexes from the Lifelines cohort study with 19 years of administrative tax records in the Netherlands, we find that greater genetic propensity toward emotional stability increases earnings and household wealth, while genetic propensity toward conscientiousness increases household wealth. The other three traits show no consistent pattern across outcomes and specifications. The effect of genetic propensity toward emotional stability on earnings remains positive as we progressively account for genetic propensity toward educational attainment, cognitive performance, and depression. These findings strengthen the evidence that personality causally affects earnings and wealth.

Two coefficients enter nearly every quantitative argument about monetary policy: the Phillips-curve slope and the policy rule's response to inflation. For seventy years, each generation that estimated them identified the coefficient conditional on assumptions it maintained, and its successor relaxed one of them, naming the bias this exposed. This survey reconstructs the two programs as a single identification problem. It distinguishes five senses in which a coefficient has been called identified, reads each generation of the curve and of the rule in that key, and shows that the programs fail together because each borrows its identifying condition from the other. The same sequence ran on the latent inflation target and on the optimal rule. The survey closes on what the credible designs license when they leave a range rather than a point: bounds, signs and orderings, and decisions designed for a range.

The use of run-off triangles (ROTs) is a common industry practice in estimating the loss given default (LGD) risk parameter when predicting credit losses in banking. We benchmark this industry practice using credit card data against a more sophisticated (though classical) regression-based approach, which is able to leverage various types of input variables in producing loan-level LGD-estimates. This regression-based approach can demonstrably recover the typical characteristics of the 'U-shaped' empirical LGD-distribution, which the ROT-based approach cannot do. First, we critically review the ROT-based approach and identify multiple demerits using data-driven diagnostics. We then estimate a two-stage regression-based LGD-model and favourably assess the model performance of each component (or 'stage'). Finally, we aggregate the LGD-estimates produced by each approach over time, and compare each time series to the mean empirical loss rate over time. The ROT-based aggregates diverge substantially from the empirical rate over most time periods, whilst the regression-based aggregates follow the empirical trends much closer. These results underscore the greater prediction accuracy of the regression-based LGD-model, relative to the ROT-based one. By implication, the former approach is probably better than the latter ROT-based approach when estimating the LGD under the IFRS 9 accounting framework, which prioritises accuracy.

On-chain transfer volume is often used to describe activity in tokenized real-world assets (RWAs), although recorded transfers may reflect very different economic functions, including token creation, destruction, issuer-related operations, and movements between other addresses. Ownership measures face a similar problem because a large address balance may belong to a pool or intermediary that represents multiple investors. We address these two measurement problems by decomposing transfer activity and making ownership coverage explicit. Transfer turnover is calculated before and after observable event categories are separated, using the same asset, observation window, and supply denominator. For ownership, we compare concentration measured against total supply with concentration among the addresses that remain after identified intermediaries are excluded. Using Ethereum-based tokenized funds from December 2025 to August 2026, we find that transfer classification can materially change measured turnover. We also show that removing an intermediary can lower concentration when measured against total supply while concentration among the remaining addresses increases. Transfer activity and address concentration need to be interpreted with their underlying transaction functions and ownership coverage in mind. These measures provide evidence about liquidity-related activity, but they do not directly measure market liquidity. The code and data used in the analysis are available at https://github.com/atytmr07/rwa-liquidity.

Personal AI assistants are beginning to transact for consumers, and early adopters capture real savings. Whether those savings survive, and what happens to consumers who have no assistant, depends on how sellers respond – a question single-user evidence cannot answer. We build an agent-based rental market in which language models play consumers, assistants, and six adaptive sellers guided by an algorithmic pricing tool. Half the population receives an assistant under an advisory or an executing mandate; the contract pairs a fee only the renter's physical action avoids with a pre-selected add-on an authorised assistant can cancel online. An analytical benchmark and a behaviourally calibrated rule market supply ex-ante predictions, and paired branches with frozen versus adaptive sellers separate adoption effects from market feedback. Across thirty simulated markets, executing assistants cut adopters' spending by 13.7 USD per renter-day when sellers are frozen; adaptation claws back about a third, leaving 8.7, with the gains arriving both as lower bills and as rentals completed at all. Sellers raise headline rates while cutting fees, and the calibrated forecast of the burden on unassisted consumers (+3.6) does not transfer: their mean spending change is +0.4, confidence interval -0.6 to +1.3. Seller-model swaps and a within-market transfer of fee-setting to the pricing tool show that fee conduct, and with it the division of the gains, is decided on the seller side. Assistants, we conclude, should be evaluated at market level – completion, total spending, and non-users included – and the comparison layers locate exactly where a calibrated behavioural forecast fails in a language-model market.

Credibility theory combines individual experience with portfolio information for insurance pricing, but classical formulations focus primarily on conditional means and expected premiums. We propose a Dirichlet mixed-membership model (DMMM) for multivariate distributional credibility. Policyholder risk is represented by a stable composition over latent risk classes: baseline characteristics determine its a priori assessment, while repeated multivariate experience progressively updates its a posteriori assessment. Support-specific expert distributions accommodate heterogeneous outcomes, with dependence induced through the shared latent structure. The resulting posterior predictive distribution can be written exactly as a convex combination of portfolio and experience components, extending the familiar credibility-factor structure beyond the conditional mean. We study the framework through a simulation experiment and a vehicle telematics application. The simulation shows that relatively simple experts can capture complex multivariate insurance distributions and learn policyholder-specific risk as experience accumulates. In the telematics application, recent claims and driving behaviour provide complementary information for future claim-frequency prediction, allowing similar policyholders to receive different experience-rated assessments. The DMMM therefore provides a flexible and interpretable framework for combining heterogeneous insurance experience while retaining the structure of classical credibility.

Market-regime models typically assume a finite set of discrete latent states. We examine whether high frequency limit-order-book dynamics exhibit distinct regime separation or apparent regimes result from discretising an underlying continuum, analysing deep limit-order book data for EURO STOXX 50 index futures across 987 clean trading days from 2022 to 2025 using 30, 45 and 60-second aggregation windows. Market states are represented by symmetric positive definite covariance matrices and analysed under Log-Euclidean and affine-invariant geometries; the resulting state cloud has an effective dimension of approximately 1.4. A causal anomaly layer filters statistical, calendar and rollover contamination, while label-free economic validation evaluates market-state separation independently of volatility-based proxy labels. A 17-method zoo spanning conventional and Riemannian representations, reinforced by full-cloud geometric certificates, consistently favours a continuum over a discrete regime structure. A dominant latent coordinate captures between 83.66% and 85.27% of variation in the covariance-state geometry and tracks VSTOXX without regime labels, remaining stable across years although its average level shifts with market conditions, indicating that conventional regimes are coarse quantisations. A divisive hierarchy spanning macro, macro-subregime, micro and micro-subregime tiers is evaluated using expanding-year walk-forward tests, with out-of-sample performance strongest at the macro tier while finer tiers fail to generalise; continuous fine-scale information nonetheless remains predictive, and its discretisation causes the performance loss. Forecasting performance peaks near ten minutes, reaching a maximum pooled R2_OOS of 0.7483. Overall, the latent stress coordinate defines a stable, low-dimensional geometric continuum that captures market stress and its predictive dynamics.

Feller random measures generalize the Feller diffusion (the CIR process) by giving it memory. They arise as the scaling limits of nearly unstable Hawkes processes, and include the rough CIR process and its hyper-rough and discontinuous relatives. We show that every Feller random measure is the occupation measure of a Dawson–Watanabe superprocess whose spatial motion is a killed Lévy subordinator, integrated over the branching time. This is a continuum analogue of the Hawkes–Oakes cluster representation. It yields existence, stability in the parameters, and stochastic equations driven by an explicit noise. The structure of this noise depends on whether the kernel has an atom at the origin. Without an atom, the noise is a Brownian motion time-changed by the distribution function of the measure. This leads to a martingale characterization and to sharp results on densities and their Hölder regularity. With an atom, the noise is a compensated inverse-Gaussian process time-changed by the compensator of that distribution function, and the measure is purely atomic.

Learning effective representations helps characterize the structure and dynamics of equity markets from financial data with a low signal-to-noise ratio. Black-box deep models can capture complex patterns but may overfit sample noise and lack explicit economic structure. Meanwhile, classic linear financial models provide interpretable references, but their oversimplified assumptions leave non-linear signals uncaptured. To combine the strengths of these two directions, we propose Stock-JEPA, a joint-embedding predictive framework that learns predictable incremental revisions relative to a point-in-time financial prior. First, we leverage a low-complexity financial model to produce fixed statistics summarizing multi-horizon return and risk. A prior projector then maps these statistics into the target encoder's latent space as an anchor. Second, we design a context-conditioned revision predictor to estimate the future representation's predictable displacement from the anchor. Separate losses update the two branches: the anchor learns from prior statistics, while the revision captures additional predictable information from historical context. Third, we freeze all representation modules and train a downstream readout, evaluating its forecasts through cross-sectional ranking and portfolio performance. Theoretically, we prove that optimal revision reduces the prior anchor's expected squared error for the same future representation by exactly $\mathbb{E}[\|\boldsymbol{\Delta}\|_2^2]$. This non-negative gain is the expected squared magnitude of the additional signal predictable from historical context. Experimentally, Stock-JEPA outperforms 13 strong baselines across large-scale China and U.S. equity universes on 5 key evaluation metrics. Ablation studies and representation analysis further demonstrate the value of the learned revisions for representation learning in equity markets.

Firms increasingly evaluate deployed AI agents using repeated human ratings, yet observed score changes may reflect the measurement process as much as changes in the agent itself. Firms also rarely know how long an evaluation remains informative as deployed systems evolve. We study reliability as a latent state using 2,611 rubric-scored production interviews from a deployed voice-and-video interviewing agent, overlapping human reviewers, the agent's change log, and support tickets. Reviewer severity varied substantially: two reviewers assessing the same batches differed by 0.79 standard deviations on the composite index, and a shift in reviewer composition flattened the raw trend. After adjusting for reviewer effects and smoothing batch-level estimates, evaluated reliability increased by 0.53 standard deviations from March to August 2026. Reliability also shifted materially across logged deploys. At the estimated rate of unaccounted movement, forecast uncertainty reaches the magnitude of the largest observed deploy contrast after about five weeks, although this horizon is imprecisely estimated and based on a small number of step-like changes. An independent operational outcome moved in the same direction: interview-related support tickets fell by about 10% per month relative to technical-issue tickets. These data do not identify a causal effect of evaluation on performance. Instead, they show that longitudinal AI evaluation requires three checks: comparability across raters, evidence currency under drift, and corroboration with consequential operational outcomes.

Operational research evaluates tournament formats by their efficacy, the probability that the design identifies the stronger competitor, and computes it almost exclusively by simulation under an assumption that is rarely tested: that the trials a format aggregates are independent. We show that this assumption systematically overstates what shortening a contest costs, and we measure the correction. When trials are serially correlated, a format carries N/[1+(N-1)rho] effective trials rather than N, so a twenty per cent cut costs 10.6 per cent of discriminating power under independence but only 5.2 per cent at the correlation we estimate, and only 2.1 per cent in a sixty-trial format. The ingredients are estimated from an implemented change in contest length: in September 2023 professional Counter-Strike cut regulation rounds per map from thirty to twenty-four. Across 72,576 maps we measure discrimination as the ability slope over the residual standard deviation on the first half of a map, a contest fixed by rule at fifteen or twelve rounds with no stopping rule. Against the calibrated benchmark of 0.948, discrimination falls by thirty-five per cent in a ten-month window around the switch (0.648, 95 per cent interval 0.52-0.81) and by ten per cent over the full panel (0.901, 0.82-0.99). The correction needs one number, recoverable from ordinary results data as a design effect.