Smaller Asymptotic Variance Research Articles

Comparison of GMM and MD estimators for a panel regression model with endogenous regressors and an AR(1) disturbance

ABSTRACT In this paper, we investigate the asymptotic properties of nonlinear GMM and linear MD estimators for a panel regression model with endogenous regressors and an AR(1) disturbance. We demonstrate that the linear MD estimator has a smaller asymptotic variance than the nonlinear GMM estimator when a non-optimal weighting matrix is used. However, when the optimal weighting matrix is used, both the nonlinear GMM and linear MD estimators have the same asymptotic distribution. The simulation results indicate that the linear two-step MD estimator is a practical choice due to its performance and simplicity.

Content Promotion for Online Content Platforms with the Diffusion Effect

Problem definition: Content promotion policies are crucial for online content platforms to improve content consumption and user engagement. However, traditional promotion policies generally neglect the diffusion effect within a crowd of users. In this paper, we study the candidate generation and promotion optimization (CGPO) problem for an online content platform, emphasizing the incorporation of the diffusion effect. Methodology/results: We propose a diffusion model that incorporates platform promotion decisions to characterize the adoption process of online content. Based on this diffusion model, we formulate the CGPO problem as a mixed-integer program with nonconvex and nonlinear constraints, which is proved to be NP-hard. Additionally, we investigate methods for estimating the diffusion model parameters using available online platform data and introduce novel double ordinary least squares (D-OLS) estimators. We prove the submodularity of the objective function for the CGPO problem, which enables us to find an efficient [Formula: see text]-approximation greedy solution. Furthermore, we demonstrate that the D-OLS estimators are consistent and have smaller asymptotic variances than traditional ordinary least squares estimators. By utilizing real data from a large-scale video-sharing platform, we show that our diffusion model effectively characterizes the adoption process of online content. Compared with the policy implemented on the platform, our proposed promotion policy increases total adoptions by 49.90%. Managerial implications: Our research highlights the essential role of diffusion in online content and provides actionable insights for online content platforms to optimize their content promotion policies by leveraging our diffusion model. Funding: R. Zhang is grateful for the financial support from the Hong Kong Research Grants Council General Research Fund [Grants 14502722 and 14504123] and the National Natural Science Foundation of China [Grant 72293560/72293565]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2022.0172 .

Graphical tools for selecting conditional instrumental sets

Summary We consider the efficient estimation of total causal effects in the presence of unmeasured confounding using conditional instrumental sets. Specifically, we consider the two-stage least-squares estimator in the setting of a linear structural equation model with correlated errors that is compatible with a known acyclic directed mixed graph. To set the stage for our results, we characterize the class of linearly valid conditional instrumental sets that yield consistent two-stage least-squares estimators for the target total effect and derive a new asymptotic variance formula for these estimators. Equipped with these results, we provide three graphical tools for selecting more efficient linearly valid conditional instrumental sets: first, a graphical criterion that, for certain pairs of linearly valid conditional instrumental sets, identifies which of the two corresponding estimators has the smaller asymptotic variance second, an algorithm that greedily adds covariates that reduce the asymptotic variance to a given linearly valid conditional instrumental set and, third, a linearly valid conditional instrumental set for which the corresponding estimator has the smallest asymptotic variance that can be ensured with a graphical criterion.

Closed form estimators for a bivariate beta distribution

We propose closed form estimators for the most popular bivariate beta distribution known to date and derive their asymptotic variances. The asymptotic properties of the proposed estimators are compared to asymptotic properties of the maximum likelihood estimators. The proposed estimators can have smaller asymptotic variances and smaller asymptotic covariances compared to maximum likelihood estimators.

New Closed Form Estimators for the Beta Distribution

In this paper, we detail closed form estimators for beta distribution that are simpler than those proposed by Tamae, Irie and Kubokawa. The proposed estimators are shown to have smaller asymptotic variances and smaller asymptotic covariances compared to Tamae estimators and maximum likelihood estimators. The proposed estimators are also shown to perform better in real data applications.

Closed form estimators for a multivariate gamma distribution

Nawa and Nadarajah [Statistics, 2023, doi: 10.1080/02331888.2023.2168004] proposed closed form estimators for a bivariate gamma distribution. But these estimators do not have a direct multivariate extension. In this paper, we propose alternative closed form estimators with direct extensions to the trivariate and multivariate cases. These estimators can have smaller asymptotic variances and smaller asymptotic covariances compared to Zhao et al.'s estimators, method of moments estimators and Nawa and Nadarajah's estimators.

New closed form estimators for a bivariate gamma distribution

Motivated by Zhao, Jang and Kim [Journal of Multivariate Analysis, 191, 2022, article number 105109], we propose new closed form estimators for a bivariate gamma distribution. The new estimators are simpler. In addition, they can have smaller asymptotic variances and smaller asymptotic covariances compared to Zhao et al.'s estimators and method of moments estimators. The new estimators can also perform better in real-data applications.

Volatility of Volatility Estimation: Central Limit Theorems for the Fourier Transform Estimator and Empirical Study of the Daily Time Series Stylized Facts

Abstract We study the asymptotic normality of two feasible estimators of the integrated volatility of volatility based on the Fourier methodology, which does not require the pre-estimation of the spot volatility. We show that the bias-corrected estimator reaches the optimal rate n1/4, while the estimator without bias-correction has a slower convergence rate and a smaller asymptotic variance. Additionally, we provide simulation results that support the theoretical asymptotic distribution of the rate-efficient estimator and show the accuracy of the latter in comparison with a rate-optimal estimator based on the pre-estimation of the spot volatility. Finally, using the rate-optimal Fourier estimator, we reconstruct the series of the daily volatility of volatility of the S&P500 and EUROSTOXX50 indices over long samples and provide novel insight into the existence of stylized facts about the volatility of volatility dynamics.

Improved Inference for Imputation-Based Semisupervised Learning Under Misspecified Setting.

Semisupervised learning (SSL) has been extensively studied in related literature. Despite its success, many existing learning algorithms for semisupervised problems require specific distributional assumptions, such as "cluster assumption" and "low-density assumption," and thus, it is often hard to verify them in practice. We are interested in quantifying the effect of SSL based on kernel methods under a misspecified setting. The misspecified setting means that the target function is not contained in a hypothesis space under which some specific learning algorithm works. Practically, this assumption is mild and standard for various kernel-based approaches. Under this misspecified setting, this article makes an attempt to provide a theoretical justification on when and how the unlabeled data can be exploited to improve inference of a learning task. Our theoretical justification is indicated from the viewpoint of the asymptotic variance of our proposed two-step estimation. It is shown that the proposed pointwise nonparametric estimator has a smaller asymptotic variance than the supervised estimator using the labeled data alone. Several simulated experiments are implemented to support our theoretical results.

A note on efficient minimum cost adjustment sets in causal graphical models

Abstract We study the selection of adjustment sets for estimating the interventional mean under an individualized treatment rule. We assume a non-parametric causal graphical model with, possibly, hidden variables and at least one adjustment set composed of observable variables. Moreover, we assume that observable variables have positive costs associated with them. We define the cost of an observable adjustment set as the sum of the costs of the variables that comprise it. We show that in this setting there exist adjustment sets that are minimum cost optimal, in the sense that they yield non-parametric estimators of the interventional mean with the smallest asymptotic variance among those that control for observable adjustment sets that have minimum cost. Our results are based on the construction of a special flow network associated with the original causal graph. We show that a minimum cost optimal adjustment set can be found by computing a maximum flow on the network, and then finding the set of vertices that are reachable from the source by augmenting paths. The optimaladj Python package implements the algorithms introduced in this article.

Berry–Esseen Bounds of the Quasi Maximum Likelihood Estimators for the Discretely Observed Diffusions

For stationary ergodic diffusions satisfying nonlinear homogeneous Itô stochastic differential equations, this paper obtains the Berry–Esseen bounds on the rates of convergence to normality of the distributions of the quasi maximum likelihood estimators based on stochastic Taylor approximation, under some regularity conditions, when the diffusion is observed at equally spaced dense time points over a long time interval, the high-frequency regime. It shows that the higher-order stochastic Taylor approximation-based estimators perform better than the basic Euler approximation in the sense of having smaller asymptotic variance.

Monotonicity-constrained nonparametric estimation and inference for first-price auctions

In the independent private values framework for first-price auctions, we propose a new nonparametric estimator of the probability density of latent valuations that imposes the monotonicity constraint on the estimated inverse bidding strategy. We show that our estimator has a smaller asymptotic variance than that of Guerre, Perrigne and Vuong’s estimator. In addition to establishing pointwise asymptotic normality of our estimator, we provide a bootstrap-based approach to constructing uniform confidence bands for the density function.

Efficient estimation in heteroscedastic single-index models

In this article, we focus on the efficient estimation in single-index models with heteroscedastic errors. We first develop a nonparametric estimator of the variance function based on a fully nonparametric function or a dimension reduction structure, and the resulting estimator is consistent. Then, we propose a reweighting estimator of the parametric component via taking the estimated variance function into account, and the main results show that it has a smaller asymptotic variance than the naive estimator that neglects the heteroscedasticity. Simulation studies are conducted to evaluate the efficacy of the proposed methodologies, and an analysis of a real data example is provided for illustration.

Efficient adjustment sets in causal graphical models with hidden variables

Summary We study the selection of adjustment sets for estimating the interventional mean under a point exposure dynamic treatment regime, that is, a treatment rule that depends on the subject’s covariates. We assume a nonparametric causal graphical model with, possibly, hidden variables and at least one adjustment set comprised of observable variables. We provide the definition of a valid adjustment set for a point exposure dynamic treatment regime, which generalizes the existing definition for a static intervention. We show that there exists an adjustment set, referred to as optimal minimal, that yields the nonparametric estimator of the interventional mean with the smallest asymptotic variance among those that are based on observable minimal adjustment sets. An observable minimal adjustment set is a valid adjustment set such that all its variables are observable and the removal of any of its variables destroys its validity. We provide similar optimality results for the class of observable minimum adjustment sets, that is, valid observable adjustment sets of minimum cardinality among the observable adjustment sets. Moreover, we show that if either no variables are hidden or if all the observable variables are ancestors of either treatment, outcome or the variables that are used to decide treatment, a globally optimal adjustment set exists. We provide polynomial-time algorithms to compute the globally optimal, optimal minimal and optimal minimum adjustment sets. Because static interventions can be viewed as a special case of dynamic regimes, all our results also apply for static interventions.

Semiparametric estimation for average causal effects using propensity score-based spline

When estimating the average causal effect in observational studies, researchers have to tackle both self-selection of treatment and outcome modeling. This is difficult because the parametric form of the outcome model is often unknown and there exists a large number of covariates. In this work, we present a semiparametric strategy for estimating the average causal effect by regressing on the propensity score. Furthermore, we show that regression error terms usually depend on the propensity score as well, which could cause heteroscedastic variances, and thus construct a refined estimator to improve the estimation efficiency. Both estimators are shown to be consistent and asymptotically normally distributed, with the latter one having a smaller asymptotic variance. The simulation studies indicate that our methods compare favorably with many competing estimators. Our methods are easy to implement and avoid hazardous impact due to extreme weights as often seen in weighting estimators. They can also be extended to handle subgroup effects with known structure. We apply the proposed methods to data from the Ohio Medicaid Assessment Survey 2012, estimating the effect of having health insurance on self-reported health status for a population with subsidized insurance plan choices under the Affordable Care Act.

A Comparison of Maximum Likelihood Estimations for Normal Mean under Right Censoring

For efficiently estimating the normal mean ([Formula: see text]) under right censoring (threshold =[Formula: see text], [Formula: see text] is known), we compare two approaches within the maximum likelihood estimation (MLE) framework. Approach I is a hierarchical MLE for which only the empirical censoring probability is utilized. Approach II is the direct MLE for which expectation-maximization (EM) algorithm is applied to all individual observations. We use discrete approximation to explain that the asymptotic variance of Approach II estimate equals the inverse Fisher information calculated from the full log-likelihood. We prove that Approach II gives a uniformly smaller asymptotic variance than Approach I and the variance ratio is a decreasing function of [Formula: see text]. We further prove some supportive results and graphically demonstrate that EM algorithm monotonically converges to the unique MLE.

First Difference MLE and Dynamic Panel Estimation

First difference maximum likelihood (FDML) seems an attractive estimation methodology in dynamic panel data modeling because differencing eliminates fixed effects and, in the case of a unit root, differencing transforms the data to stationarity, thereby addressing both incidental parameter problems and the possible effects of nonstationarity. This paper draws attention to certain pathologies that arise in the use of FDML that have gone unnoticed in the literature and that affect both finite sample peformance and asymptotics. FDML uses the Gaussian likelihood function for first differenced data and parameter estimation is based on the whole domain over which the log-likelihood is defined. However, extending the domain of the likelihood beyond the stationary region has certain consequences that have a major effect on finite sample and asymptotic performance. First, the extended likelihood is not the true likelihood even in the Gaussian case and it has a finite upper bound of definition. Second, it is often bimodal, and one of its peaks can be so peculiar that numerical maximization of the extended likelihood frequently fails to locate the global maximum. As a result of these pathologies, the FDML estimator is a restricted estimator, numerical implementation is not straightforward and asymptotics are hard to derive in cases where the peculiarity occurs with non-negligible probabilities. We investigate these problems, provide a convenient new expression for the likelihood and a new algorithm to maximize it. The peculiarities in the likelihood are found to be particularly marked in time series with a unit root. In this case, the asymptotic distribution of the FDMLE has bounded support and its density is infinite at the upper bound when the time series sample size T approaching infinity. As the panel width n approaching infinity the pathology is removed and the limit theory is normal. This result applies even for T fixed and we present an expression for the asymptotic distribution which does not depend on the time dimension. When n,T approaching infinity, the FDMLE has smaller asymptotic variance than that of the bias corrected MLE, an outcome that is explained by the restricted nature of the FDMLE.

Quantile treatment effects and bootstrap inference under covariate‐adaptive randomization

In this paper, we study the estimation and inference of the quantile treatment effect under covariate‐adaptive randomization. We propose two estimation methods: (1) the simple quantile regression and (2) the inverse propensity score weighted quantile regression. For the two estimators, we derive their asymptotic distributions uniformly over a compact set of quantile indexes, and show that, when the treatment assignment rule does not achieve strong balance, the inverse propensity score weighted estimator has a smaller asymptotic variance than the simple quantile regression estimator. For the inference of method (1), we show that the Wald test using a weighted bootstrap standard error underrejects. But for method (2), its asymptotic size equals the nominal level. We also show that, for both methods, the asymptotic size of the Wald test using a covariate‐adaptive bootstrap standard error equals the nominal level. We illustrate the finite sample performance of the new estimation and inference methods using both simulated and real datasets.

Estimating overdispersion in sparse multinomial data.

Multinomial data arise in many areas of the life sciences, such as mark-recapture studies and phylogenetics, and will often by overdispersed, with the variance being higher than predicted by a multinomial model. The quasi-likelihood approach to modeling this overdispersion involves the assumption that the variance is proportional to that specified by the multinomial model. As this approach does not require specification of the full distribution of the response variable, it can be more robust than fitting a Dirichlet-multinomial model or adding a random effect to the linear predictor. Estimation of the amount of overdispersion is often based on Pearson's statistic X2 or the deviance D. For many types of study, such as mark-recapture, the data will be sparse. The estimator based on X2 can then be highly variable, and that based on D can have a large negative bias. We derive a new estimator, which has a smaller asymptotic variance than that based on X2 , the difference being most marked for sparse data. We illustrate the numerical difference between the three estimators using a mark-recapture study of swifts and compare their performance via a simulation study. The new estimator has the lowest root mean squared error across a range of scenarios, especially when the data are verysparse.

Semiparametric random censorship models for survival data with long-term survivors

In this article, we study a semiparametric random censorship model for survival data in the presence of long-term survivors. Local likelihood method is employed to estimate the conditional mean regression function of binary random variables. The proposed estimators for the survival function and the cure rate, as well as their asymptotic properties, are investigated based on empirical and U-statistical processes. In particular, the proposed estimator for the cure rate is shown to be superior over the previous estimator considered by Maller and Zhou in the sense of having a smaller asymptotic variance. This semiparametric random censorship model with related estimation methods provide an efficient alternative for survival analysis with long-term survivors.