Papers

During clinical trials evaluating a drug's effect on a survival endpoint, intermediate events often occur in addition to the primary event. The treatment can exert its effect on the primary endpoint along multiple pathways through intermediate events. Assumptions for identifying mediation effects, such as sequential ignorability in natural effects or the dismissible components condition in separable effects, fail because intermediate events act as treatment-induced confounding. To understand the effect along each pathway, we consider hypothetical interventions in transitions between event statuses to mimic the treatment mechanism. The hypothetical interventions adjust for effects through intermediate events and marginalize over unobserved treatment-induced confounding, if any. Based on the derived efficient influence functions for the counterfactual cumulative incidences under hypothetical interventions, we construct multiply robust and semiparametrically efficient estimators for pathway treatment effects. Our proposed framework enables the examination of treatment effects through each transition, on each event, and along each path. By analyzing data from the LEADER Trial, we find that liraglutide significantly reduces the risk of cardiovascular and microvascular events. The reduction in all-cause mortality is primarily mediated by its effects on expanded major adverse cardiovascular events.

This paper studies the nonparametric identification and estimation of additively separable triangular models with continuous endogenous and instrumental variables, allowing for a nonseparable first-stage equation. Under the independence of instrumental variables and unobservables, we show that the outcome function possesses a closed-form expression as a functional of conditional cumulative distribution functions. The resulting plug-in estimators require no regularization and converge at the rate $n^{-m/(2m+1)}$, where $m$ is the order of smoothness the model imposes. Also, no estimator of the outcome function converges faster. We use the empirical bootstrap to construct a uniform confidence band that covers the outcome function at every point of a compact set simultaneously.

We study a dynamic spatial panel model with observed regressors, interactive effects, and contemporaneous and lagged dependence in a large-$N$, fixed-$T$ framework. The spatial model constitutes an $N$-dimensional simultaneous-equations system. In this $N$-equation view, the interactive effects introduce $N$ unit-specific loading vectors. Estimating them individually when $T$ is fixed creates the type of incidental-parameters problem underlying Nickell bias. We use the $N$-equation view to account for spatial simultaneity through the spatial Jacobian. Crucially, however, we view the same model as a $T$-equation system with $N$ observations. Together with a spatially enriched random-loadings (SERL) specification, this $T$-equation view yields a quasi-likelihood without unit-specific incidental parameters. We propose a computationally tractable block-coordinate algorithm. Simulations show small estimation errors and generally near-nominal coverage probabilities. Applying the method to U.S. county female labor-force participation, we find substantial dynamic and spatial dependence and an important role for education in explaining the rise in female labor-force participation.

This paper studies quantile treatment and spillover effects in network experiments. Average spillover effects reveal how treating a unit's neighbors affects its outcome on average, but mask the heterogeneity of these effects across the outcome distribution. We define structural quantile effects that compare outcome quantiles between exposure states, characterizing how own treatment and exposure to treated neighbors affect different parts of the outcome distribution. Building on \citet{leung2020treatment}, we first establish the weak convergence of the estimated quantile-effect process under conditions requiring the stabilization of the degree distribution and the network-dependent covariance structure. Our main contribution is to propose uniform confidence bands (UCBs) based on Gaussian approximations conditional on the realized network, avoiding these stabilization requirements. The proposed method is evaluated through extensive simulation studies and an empirical application to a randomized savings-account experiment in Nepal \citep{prina2015banking}.

Staggered distributional difference-in-differences produces cohort-specific potential-outcome distributions, but applied work typically wants one overall quantile treatment effect. Two natural summaries—averaging cohort quantile treatment effects (QTTs) and mixing cohort distributions before inversion—use the same policy weights yet answer different target-population questions and can disagree even in sign. We derive the exact sharp interval for their gap conditional on cohort quantiles and weights, a globally sharp range-only envelope, and a locally sharp density-tilt representation. We develop joint smooth and mass-point-safe inference and show that neither estimator is uniformly more precise, even when the estimands coincide. In a same-object reconstruction of a public staggered-QTT application, holding data, identification, distributions, and weights fixed while changing only aggregation order reverses reported signs at several quantiles. Aggregation order is therefore part of the estimand and must be chosen before inversion.

Asymptotic normality approximations often fail to hold for extremum estimators when the true value of the parameter is at or close to the boundary of a parameter space. I analyze and develop tests using a quasi-unconstrained estimator, which is asymptotically normal even when the true parameter vector is near or at the boundary. These results generalize previous work with this estimator by allowing for more types of constraints and showing how the method can naturally be modified when a nuisance parameter is also high-dimensional. I show that variations of Wald, Likelihood Ratio, and Lagrange Multiplier tests can control size in a uniform sense, provided the initial constrained estimator is sufficiently accurate. Lastly, I apply the method to an application involving network estimation with panel data.

Global forecasting models pool many series and learn one shared function. Gradient-boosted trees are their most common form. We measure a failure of this design that has not, to our knowledge, been documented. Train a global tree on the individual series of a hierarchy, then ask it for the hierarchical aggregate. The aggregate sits far outside the model's training range, and the forecast collapses. The model under-predicts the total by 30-50x in our production deployment, and by up to 496x in a public M5 reconstruction. The mechanism is known: beyond its training range, a tree predicts a constant. It surfaces at the aggregate because the total dwarfs every training series. The cure is not new. Per-series scaling, the preprocessing step that Montero-Manso and Hyndman (2021) recommend, prevents the collapse. So do a weighted aggregate-level training row and seasonal differencing. Our contribution is the characterization. The collapse reproduces on five panels: a production business-to-business marketplace, a synthetic hierarchy, M5, Australian Tourism, and a public business-buyer panel. It holds on three tree libraries, is invariant across training seeds, and is statistically significant. Its onset is immediate and tracks a simple support bound: a scale gap of only 1.15x already costs a third of the total. No standard configuration change prevents it: pooling every hierarchy level into training fails at scale, and the one knob that fits linear models in the leaves softens it without curing it. Rolling the forecasts forward recursively separates the cures: the aggregate-row cure re-collapses, per-series scaling degrades but stays low, and only seasonal differencing keeps its one-step accuracy unchanged. We close with a three-step procedure for diagnosing and preventing the failure in deployed systems.

Rejection sampling requires a proposal that dominates the target by a known constant, generally unavailable for non-Gaussian state space models. We construct such a proposal for the latent state path, yielding independent exact smoothing draws and an unbiased likelihood estimator whose relative variance is at most $1/p-1$ per draw at acceptance probability $p$. The method covers scalar states with affine Gaussian dynamics and log-concave observation densities, including multivariate observations. Transition twisting makes the log target-to-proposal ratio separable, and tangent-line twists make each term nonpositive, producing an attained, sharp dominating constant. With a companding node placement, the accumulated envelope error is $O(T/G^2)$ for a sample of length $T$ with $G$ nodes per date, so $G\propto\sqrt{T}$ keeps acceptance bounded away from zero; for stochastic volatility, the required conditions hold almost surely. A simpler mode-centered grid shows the same scaling empirically. At $T=2{,}000$, acceptance is $75\%$, versus roughly $10^{-16}$ for the Gaussian envelope.

Short observational pricing panels often contain many data points but very few actual price changes. This paper shows that this sparsity creates a hidden source of error that standard statistical methods miss. When estimating price effects, most of the uncertainty does not come from sample size within a panel, but from the specific history of price movements observed. Standard confidence intervals fail because they only measure variation within the panel, ignoring this broader design-level error. Using simulations, we find that this cross-design variation accounts for most of the estimation error, causing standard methods to significantly understate uncertainty. First, we show that cross-design error decreases predictably as the total volume of price variation increases. Second, adding more data from regions that share the same price trends does not fix the issue; true precision improves only when combining data across units with independent price trajectories. Third, applying a simple variance-component adjustment across independently priced units restores accurate statistical coverage. We confirm these findings in real-world store scanner data, showing that products and pricing zones behave as if they have far fewer independent price movements than their raw counts suggest. Ultimately, reliable inference in passive pricing data requires genuine, independent variation, which can be achieved through controlled regional price testing.

We study causal moderation when treatment assignment is randomized but the moderator is not. We combine the parallel estimation framework with front-door adjustment to identify an average mediated treatment moderation effect. We apply this approach to 165 municipal assembly elections in Tokyo (1987-2023), where pamphlet positions are assigned by lottery and total pamphlet pages are mechanically determined by candidate set size and fixed municipal rules. One additional candidate increases the front-page effect on winning probability by 0.322 percentage points through the page-count channel, 8.0% of its baseline magnitude, while the back-page effect does not increase significantly.

Firms perform online experiments with multi-armed bandits to personalize what consumers are shown while balancing exploration and exploitation. However, third-parties can infer consumers' underlying segments from observing which banners, ads, or recommendations consumers receive. To control this inference, we propose a privacy risk budget that firms can set ex ante to bound such third party belief updating using differential privacy. To spend this privacy risk budget, we propose two strategies: a constant privacy risk strategy and a dynamic privacy risk strategy that spend privacy risk differently across visitor. We study how privacy risk budgets affect experimentation performance in two applications–website design and a recommendation system–under these strategies. For both strategies, we analytically find privacy risk budgets that optimally balance exploration and exploitation. We then extend the idea of an experiment-level privacy risk budget to a firm-wide privacy risk budget. We apply this firm-wide privacy risk budget in an empirical setting with 78 experiments. We find that the dynamic strategy is particularly valuable in longer and more complex experiments, and that optimizing the allocation of a firm-wide privacy risk budget across experiments substantially improves learning performance.

Moment restrictions provide a flexible basis for quasi-Bayesian inference when a full likelihood is unavailable, but the weighting matrix in a quadratic moment criterion determines both the relative importance of the moments and the information scale of posterior updating. We propose curvature-calibrated quasi-Bayesian updating, which uses the inverse of the covariance (or long-run covariance) of the moment conditions evaluated at a self-consistent quasi-posterior center. The resulting fixed-point procedure alternates between covariance estimation and simulation from a fixed-weight quasi-posterior, thereby avoiding parameter-dependent weighting during each simulation run. Under a Bernstein-von Mises condition for the fixed-weight quasi-posterior at the efficient population weight, we show that the calibration map is locally contractive, that its fixed point is consistent at the standard parametric rate, and that the Gaussian approximation continues to hold under the calibrated data-dependent weight, with covariance given by the inverse Godambe information matrix. Under a uniform fourth-moment condition, the scaled quasi-posterior covariance converges to the same matrix, so quasi-posterior and repeated-sampling uncertainty agree to first order. Simulations show improved covariance calibration and interval coverage after a few updates. An application to longitudinal binary-response data illustrates the method with within-subject dependence and overidentified residual moments.

Pairwise randomization can yield substantial efficiency gains in experiments. Yet methodological guidance cautions against pairwise randomization, especially in settings with attrition, partly because common practices for estimation (i.e., pair fixed effects) imply discarding data from incomplete pairs thus exacerbating data loss from attrition. This practice of dropping incomplete pairs reduces statistical power of tests as well as precision of estimates, in paired experiments, compared to designs with less finely stratified treatment assignment. We argue that this concern is misplaced if attrition is independent of treatment status and potential outcomes, and that these issues follow from an inefficient use of the data that remains post-attrition. First, we show how, by using a specific permutation test, it is possible to use all observed units for inference (complete pairs and incomplete pairs where one unit attrits) while still exploiting the pairwise randomization design structure. The test procedure we suggest provides exact size control under the sharp null. Second, we study an optimally weighted estimator that efficiently combines within-pair and across-pair comparisons. Finally, we show that combining these two insights yields a test procedure that dominates the two commonly used inference methods (a paired $t$-test and the two-sample $t$-test) in power, for any level of attrition. Usefully for applied researchers, we show that the efficient procedure can be implemented via a weighted fixed effects regression, straightforward in standard software. In sum, our results provide researchers with practical tools for conducting experiments with pairwise randomization without sacrificing observations or statistical power when facing independent attrition.

This paper develops asymptotic theory and feasible inference for unbounded-kernel order-k U-statistics under clustered sampling and weakly dependent time-series. The analysis first builds the complete order-2 pipeline, moving from clustered data to exact $m$-dependence and then to near-epoch dependence. The same logic is subsequently extended to general order k greater or equal to 2. Under clustered sampling, the theory allows arbitrary within-cluster dependence and growing, unbalanced cluster sizes. Under weak dependence, an i.i.d.-based approximating sequence carries the exact-m theory to near-epoch-dependent processes. The common combinatorial device partitions the sample into columns, separating sampling-generic tuples, where the first-order Hoeffding projection is analysed, from collision terms and higher-order Hoeffding projection remainders, which are controlled explicitly. Cluster-robust and HAC estimators of the covariance of the first-order projection, needed for feasible inference, are shown to be consistent. Empirical applications and data-calibrated simulations for inequality, L-moment, and rank-dependence statistics illustrate the finite-sample performance of the proposed procedures.

We study difference-in-differences (DiD) designs in which a binary treatment changes an endogenous time-varying (continuous, discrete, or mixed) mediator that in turn affects an outcome. We allow for the inclusion of lagged outcomes in the model. Under our model assumptions, we show that the usual DiD estimand mixes the average direct effect on the treated, the average indirect effect, and a trend bias term. A two-way fixed effects (TWFE) regression that controls for the mediator does not recover the average direct treatment effect on the treated. We show that a DiD estimand conditional on the observed mediator path identifies the conditional average direct effect for treated units at that path, and that averaging over the treated path distribution identifies the average direct effect even when unconditional parallel trends fails. A stable average mediator effect assumption helps recover the average mediator and indirect effects. The framework extends to multivariate mediators, nonlinear DiD, and multiple-treatment-period settings. Existing doubly robust estimators can be used to conduct inference. Revisiting the effects of railroad access on agricultural land values, the specification yields a positive direct component not mediated by measured market access, while the corresponding indirect component is small and imprecise.

We study how fast experimental designs can approach the semiparametric efficiency bound in finite samples, as measured by the excess variance of unadjusted treatment effect estimation. We prove an impossibility theorem: under weak conditions, no design can approach the variance bound uniformly over smooth outcome models unless covariate dimension $d \ll \log n$. Even in experiments with thousands of units, this permits only a handful of covariates. Motivated by this, we propose new designs based on discrepancy minimization that instead attempt to control imbalances over restricted-complexity nonparametric function classes. Such designs achieve fast rates to their corresponding restricted efficiency targets, permitting $d \ll n$ covariates in an additive nonparametric specification. They can also be combined with matching to protect against unmodeled outcome variation. In simulations calibrated to 12 published experiments, our designs reduce variance relative to matched pairs randomization in every empirical setting.

When parallel trends fails for some treated cohorts but not others, the average treatment effect on the treated (ATT), an average over all of them, is exactly the target that becomes hard to recover. We propose changing the estimand rather than defending it. The credible-subpopulation local ATT (LATT) is the effect for the subpopulation of cohorts whose parallel trends is credible, and it is point-identified under parallel trends for the selected cohorts alone, a weaker requirement that can hold when the ATT's fails. It is estimated by reweighting standard group-time effects toward those cohorts, and paired with honest sensitivity bounds on the residual violation that a pre-trend screen cannot rule out. The method's advantage grows with how informative pre-trends are about post-treatment violations, as simulations confirm. In an application, a significantly positive pooled estimate of the shale boom's effect on local house prices proves to rest on cohorts already trending before onset, and the credible subpopulation reveals no effect.

This paper analyzes when choice probabilities reveal rankings of deterministic utility indices in semiparametric discrete choice models. It begins with binary choice, where quantile thresholds guarantee ranking recovery, and shows that such thresholds can arise either from behavioral departures from utility maximization (e.g., limited attention) under exchangeable unobservables, or from non-exchangeable unobservables under standard utility maximization. These behavioral and distributional routes are then extended to multinomial choice. Under limited attention, balance restrictions on attention probabilities yield global linear ranking partitions which are robust to the distribution of unobservables and, given sufficiently rich joint variation in the differences of utility indices, are also necessary. Absent the required attention restrictions, opposite rankings can produce overlapping probability images. Under non-exchangeable unobservables, a comparable distribution-uniform partition generally need not exist. Holding the distribution fixed, however, ranking recovery remains possible via an injective nonlinear map from normalized utility differences to choice probabilities under both behavioral and distributional extensions. Together, the results distinguish distribution-robust global ranking partitions from ranking recovery with a fixed distribution of unobservables and clarify the limits of extending binary quantile restrictions to multinomial choice.

Empirical studies of peer effects often exploit conditional random assignment to peer groups within urns. We develop a GMM framework for estimation and inference in this setting. The framework separately identifies endogenous and contextual peer effects and nests tests of random peer-group assignment as a special case. It permits unknown heteroskedasticity and corrects finite-urn bias in variance estimation. Its asymptotic theory allows the number of peer groups to grow through more urns, more groups within urns, or both. We establish the asymptotic validity of the procedures and evaluate their finite-sample performance through Monte Carlo simulations. We apply the method to study peer effects on personality among university students. For traits with positive reduced-form peer effects, the estimates indicate that positive contextual effects are partly offset by negative endogenous effects.

We ask whether COVID-19 lockdown stringency altered national Olympic performance between Rio 2016 and Tokyo 2020, using the Oxford Stringency Index and the 99 countries that won a medal in either edition. As in \citet{liu2024}, mean performance is unaffected: stringency is insignificant in every OLS and ANOVA specification. The distribution is not. Among the 84 non-traditionally dominant nations, medal changes are three to six times more dispersed in high-stringency countries; the difference is absent among dominant nations and concentrated in men's events. A common shock left the competitive hierarchy intact while sharply raising outcome uncertainty for smaller Olympic teams.

We study the existence of a Regional Differential in rugby sevens: whether, in tournaments where no competing team enjoys formal home status, some national sides systematically over- or under-perform depending on where the event is staged. Using the universe of 2672 men's and women's matches from international rugby sevens tournaments played between 2016 and 2025, principally the World Rugby Sevens Series, but also the World Cup Sevens and the Olympic Games, we replace the binary “locality” indicator of the classic home advantage literature with continuous measures of geographic, temporal, and cultural proximity between each team and the host country. We show that aggregate associations between proximity and performance (which appear large and highly significant in naive specifications) are almost entirely an artifact of team composition and scheduling: once team and opponent quality are absorbed through fixed effects in a symmetric team-match panel, no general regional advantage survives. However, this null aggregate masks substantial heterogeneity. For a well-defined subset of teams (most notably France, but also New Zealand, the United States, Canada, Spain, Ireland, Kenya, Samoa, and Brazil), performance declines significantly and monotonically with the distance separating the host city from home. Some of the established southern-hemisphere powers (Fiji, South Africa, Australia, Argentina) are essentially distance-neutral. We further show that for some teams the gradient operates through east-west jet lag rather than pure displacement. The Regional Differential in sevens is therefore real but team-specific rather than universal, a distinction obscured by pooled estimation.

We study how many observations are needed to determine the causal direction between two linearly related variables. Classical LiNGAM theory shows that independent non-Gaussian disturbances identify the direction, but does not quantify the difficulty when the causal effect is weak or the disturbances are nearly Gaussian. Let $\beta$ bound the absolute structural coefficient from below, let $\nu$ measure each standardized disturbance's distance from Gaussianity, and let the disturbance scales lie in $[\underline\sigma,\overline\sigma]$. We prove the sharp local minimax law \[ N_2^\star(\beta,\nu,\delta) \asymp \frac{\log(1/\delta)} {d_\beta^2+\beta^2\nu^2}, \qquad d_\beta= \left[\beta^2- \left(1-\frac{\underline\sigma^2}{\overline\sigma^2}\right)\right]_+. \] Previous theory established population identifiability or assumed a fixed separation between the two directions. By contrast, we establish the sharp sample complexity as a joint function of edge strength, distance from Gaussianity, and scale uncertainty, and characterize when identification comes from non-Gaussian dependence or from covariance alone. The proof was independently generated with GPT-5.6 Sol in Codex's Ultra mode during a two-hour session. The human author supplied the prompt and was responsible only forchecking the proof and revising and polishing the manuscript.

Route and activity choice are distinct transportation problems that both require models of feasible decisions unfolding over networks and time. This critical integrative review connects transportation choice modeling with inverse reinforcement learning (IRL) and imitation learning (IL), while distinguishing evidence from transportation applications, transferable methods from other fields, and emerging proposals. We develop a four-layer sequential mobility choice framework comprising the environment, behavioral objective, stochastic choice mechanism, and observation process. Under stated assumptions, recursive logit, logit dynamic discrete choice, and maximum-entropy IRL use the same soft Bellman recursion linking future opportunities to current choice probabilities. Expected state-action visitation also satisfies conservation equations analogous to network flows. These mathematical connections do not make utility, reward, policy, occupancy, constraints, and observation error behaviorally interchangeable. Transportation evidence is strongest for network-scale planning, context-dependent reward learning, inference from incomplete trajectories, and activity-schedule generation, but remains limited for actual interventions and transfer across networks. We therefore propose a behaviorally disciplined hybrid architecture that keeps feasible actions, interpretable trade-offs, observation processes, and system feedback explicit while using machine learning for scalable computation, contextual representation, heterogeneity, and data integration.

We propose a family of control variate estimators for variance reduction in design-based survey sampling and causal inference, with and without interference. In these settings, inverse probability weighting (IPW) estimators are widely used, but may have large variance when sampling, treatment, or exposure probabilities are small. Building on the observation that several common estimators, including the Hajek estimator, the normalized estimator, the augmented inverse probability weighting (AIPW) estimator, and the targeted maximum likelihood estimator (TMLE), all correct the Horvitz-Thompson estimator by canceling part of its randomness, we provide a unified interpretation of these estimators as special cases of a general control variate estimator. We then construct optimal control variates that can reduce the finite sample variance compared to these common estimators. We parameterize the proposed control variates by their bases and characterize the optimal bases through a stochastic optimization formulation. In survey sampling and causal inference without interference, the optimal bases are characterized by leading eigenvectors of matrices that depend on both the design-based sampling structure and the model-based outcome uncertainty. In causal inference under network interference, the optimal bases solve a nonconvex quadratic optimization problem; we provide a $\frac{1}{2}$-approximate solution and an alternating local search heuristic. We apply the control variate estimators to the Swiss Environmental Panel survey data and the Chinese social network data, and conduct extensive simulations to show that the proposed control variate estimators can achieve substantial variance reduction.

This study investigates the methodological and theoretical properties of session handover in applications that use large language models. A task may continue in a new session when the context reaches the model's input limit, when the application restarts, or when another agent is asked to finish the task. The application must then decide which information from the earlier session to pass on. We formulate handover as the transfer of a task-relative in-context learning (ICL) state and distinguish exact recovery of earlier material from preservation of the target distribution. Under an exogeneity condition, predictive equivalence characterizes the coarsest deterministic sufficient handover and gives a fixed-length bit requirement. The analysis isolates the effects of the memory constraint, the writer, and the continuation procedure, and quantifies the cost of writing before the realized downstream query is known. We propose a three-part record that stores decisions and constraints exactly, uses task-justified statistics for repeated evidence, and retains original observations whose effect is not preserved by those statistics. Gaussian linear regression gives an exact finite-dimensional handover and finite-bit perturbation bounds, while nonparametric regression gives upper and lower bounds that relate memory to squared prediction error. These results provide a theory and method for deciding what a handover must retain and how its memory requirement depends on the continuation task.

This paper employs a Threshold Bayesian Vector Autoregression (TBVAR) to estimate the regime-dependent macroeconomic effects of capital regulation in Hungary. Using the Factor-based Index of Systemic Stress (FISS) as the threshold variable, the model identifies normal and stress regimes consistent with the occasionally binding constraints literature. The TBVAR offers a practical multivariate alternative to Growth-at-Risk for data-constrained economies. Generalised impulse responses reveal a pronounced asymmetry: releasing regulatory capital during stress raises GDP growth at the peak, with effects persisting for roughly twenty months, while the cost of accumulating capital in the normal regime is economically negligible. These findings are robust to alternative Cholesky orderings, sample periods, and credit variable definitions, providing direct empirical support for the countercyclical operation of the capital buffer.

We develop a method for estimating and testing a single block of a macroeconomic model with heterogeneous agents, without placing assumptions on the structure of the rest of the economy. In a large class of models, individual agents' decisions depend on the macroeconomy only through their expectations of the evolution of a finite-dimensional vector of "sufficient statistics" (e.g., asset returns or aggregate earnings). Our estimator selects the structural parameters that provide the best model-consistent fit between empirical impulse responses with respect to identified macro shocks of (a) cross-sectional moments of agent choices (e.g., moments of consumption) and (b) the vector of sufficient statistics. In a simulation illustration, we estimate a two-asset heterogeneous household model block without restricting production, firm investment, financial intermediation, monetary policy, trade, etc.

Limited dependent variable models are central to empirical economics, but likelihood-based inference is infeasible when likelihoods involve high-dimensional integration over latent variables. This paper proposes Stochastically Estimated Gradient Ascent (SEGA), a scalable estimation approach for limited dependent variable models. Using Fisher's identity, SEGA replaces the intractable likelihood score with an unbiased augmented data score evaluated at a single conditional draw of the latent variables, and embeds this score in a stochastic gradient ascent algorithm. With sufficiently many iterations, we show that SEGA is asymptotically equivalent to the infeasible maximum likelihood estimator. A variance estimator based on Fisher's and Louis' identities is proposed that allows inference to proceed in the usual manner. Applications to brand choice and household demand demonstrate the usefulness of SEGA for conducting inference in large-scale discrete choice and censored demand models.

We study minimax-optimal designs and estimators for estimating the sample average treatment effect in finite population randomized experiments, where both design and estimator are unrestricted. For binary potential outcomes, we show this minimax risk is equivalent to the minimax risk $\rho_n^*$ of an estimation problem with $2$ unknown parameters. We leverage this reduction to establish a second-order risk expansion $\rho_n^* = n^{-1} - Cn^{-4/3} + o_n(n^{-4/3})$ for an explicit constant $C$ related to the Airy function. The minimax risk is attained by Bernoulli randomization with a nonlinear shrinkage estimator. Our results show that standard procedures such as complete randomization with difference in means are only minimax optimal up to first order in $n.$ We derive further results on admissibility of these procedures and discuss the practical implications of our results.

When comparison units may also respond to treatment, panel comparisons reflect both the treatment effect and spillovers. If the interference pattern is unknown, observed outcomes alone do not separate the two. I characterize what can nevertheless be learned from panel outcomes under general restrictions, without requiring an exposure mapping or prior classification of affected donors. The framework scales validity bounds for every convex donor weight by its fit before treatment and combines these bounds with prespecified restrictions tailored to the application. The validity bounds constrain the treatment effect relative to spillovers, while the additional restrictions determine its possible values. Together these restrictions yield a sharp identified set. When the additional restrictions have a finite linear representation, checking whether a proposed treatment effect is compatible with the model reduces exactly to asking whether a finite linear system has a solution. Bootstrap calibration tests this condition. Inverting these tests uniformly controls, in large samples, the probability of falsely excluding each compatible value. In an application to the Legal Arizona Workers Act, the resulting 95 percent inversion sets contain effects of both signs across all reported specifications, leaving the sign of the treatment effect unresolved.