Pub Date : 2026-01-28DOI: 10.1016/j.jeconom.2026.106186
Anders Bredahl Kock , David Preinerstorfer
A decision maker typically (i) incorporates training data to learn about the relative effectiveness of treatments, and (ii) chooses an implementation mechanism that implies an “optimal” predicted outcome distribution according to some target functional. Nevertheless, a fairness-aware decision maker may not be satisfied achieving said optimality at the cost of being “unfair” against a subgroup of the population, in the sense that the outcome distribution in that subgroup deviates too strongly from the overall optimal outcome distribution. We study a framework that allows the decision maker to regularize such deviations, while allowing for a wide range of target functionals and fairness measures to be employed. We establish regret and consistency guarantees for empirical success policies with (possibly) data-driven preference parameters, and provide numerical results. Furthermore, we briefly illustrate the methods in two empirical settings.
{"title":"Regularizing fairness in optimal policy learning with distributional targets","authors":"Anders Bredahl Kock , David Preinerstorfer","doi":"10.1016/j.jeconom.2026.106186","DOIUrl":"10.1016/j.jeconom.2026.106186","url":null,"abstract":"<div><div>A decision maker typically (i) incorporates training data to learn about the relative effectiveness of treatments, and (ii) chooses an implementation mechanism that implies an “optimal” predicted outcome distribution according to some target functional. Nevertheless, a fairness-aware decision maker may not be satisfied achieving said optimality at the cost of being “unfair” against a subgroup of the population, in the sense that the outcome distribution in that subgroup deviates too strongly from the overall optimal outcome distribution. We study a framework that allows the decision maker to regularize such deviations, while allowing for a wide range of target functionals and fairness measures to be employed. We establish regret and consistency guarantees for empirical success policies with (possibly) data-driven preference parameters, and provide numerical results. Furthermore, we briefly illustrate the methods in two empirical settings.</div></div>","PeriodicalId":15629,"journal":{"name":"Journal of Econometrics","volume":"254 ","pages":"Article 106186"},"PeriodicalIF":4.0,"publicationDate":"2026-01-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146076155","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"经济学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-23DOI: 10.1016/j.jeconom.2026.106187
Bingxin Zhao , Yuhong Yang
This paper studies minimax rates of convergence for nonparametric location-scale models, which include mean, quantile, expectile and momentile regression settings. Under Hellinger differentiability on the error distribution and other mild conditions, we show that the minimax rate of convergence for estimating the regression function under the squared L2 loss is determined by the metric entropy of the nonparametric function class. Different error distributions, including asymmetric Laplace distribution, asymmetric connected double truncated gamma distribution, connected normal-Laplace distribution, Cauchy distribution and asymmetric normal distribution are studied as examples. Applications on low order interaction models and multiple index models are also given.
{"title":"Minimax rates of convergence for nonparametric location-Scale models","authors":"Bingxin Zhao , Yuhong Yang","doi":"10.1016/j.jeconom.2026.106187","DOIUrl":"10.1016/j.jeconom.2026.106187","url":null,"abstract":"<div><div>This paper studies minimax rates of convergence for nonparametric location-scale models, which include mean, quantile, expectile and momentile regression settings. Under Hellinger differentiability on the error distribution and other mild conditions, we show that the minimax rate of convergence for estimating the regression function under the squared <em>L</em><sub>2</sub> loss is determined by the metric entropy of the nonparametric function class. Different error distributions, including asymmetric Laplace distribution, asymmetric connected double truncated gamma distribution, connected normal-Laplace distribution, Cauchy distribution and asymmetric normal distribution are studied as examples. Applications on low order interaction models and multiple index models are also given.</div></div>","PeriodicalId":15629,"journal":{"name":"Journal of Econometrics","volume":"254 ","pages":"Article 106187"},"PeriodicalIF":4.0,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146015887","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"经济学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-22DOI: 10.1016/j.jeconom.2026.106181
Ye Yang , Wim P.M. Vijverberg
This paper offers hypothesis and specification tests for the best generalized methods of moment estimator (BGMME) of the matrix exponential spatial specification (MESS) developed by Debarsy et al. (2015). First, as the BGMME is a two-step estimator, we formulate corrected standard errors using a modified version of the finite sample correction method in Windmeijer (2005) that accounts for the fact that the BGMME makes more extensive use of the first-stage estimator than the GMM model analyzed by Windmeijer. Second, since the BGMME uses different moment conditions under normal, non-normal, and heteroskedastic disturbances, we propose a pretest strategy to determine which set of moment conditions is most suitable for the data at hand. Third, we consider and examine the performance of test statistics that help choose between MESS(1,1), MESS(1,0) and MESS(0,1) models. The performance of these tools is examined with Monte Carlo experiments, which also allow for varying degrees of spatial correlation in the explanatory variables. The correction in the standard errors is especially useful when the sample size is small, such as in a study with state-level, provincial or country-level data: the corrected standard errors improve statistical inference, yielding better size properties. The pretest strategy is effective when the heteroskedasticity, if present, is correlated with explanatory variables in the model. Spatial lags in the outcome variable are more easily detected than those in the disturbance. An empirical study of housing prices illustrates the new tools.
{"title":"GMM inference in the matrix exponential spatial specification","authors":"Ye Yang , Wim P.M. Vijverberg","doi":"10.1016/j.jeconom.2026.106181","DOIUrl":"10.1016/j.jeconom.2026.106181","url":null,"abstract":"<div><div>This paper offers hypothesis and specification tests for the best generalized methods of moment estimator (BGMME) of the matrix exponential spatial specification (MESS) developed by Debarsy et al. (2015). First, as the BGMME is a two-step estimator, we formulate corrected standard errors using a modified version of the finite sample correction method in Windmeijer (2005) that accounts for the fact that the BGMME makes more extensive use of the first-stage estimator than the GMM model analyzed by Windmeijer. Second, since the BGMME uses different moment conditions under normal, non-normal, and heteroskedastic disturbances, we propose a pretest strategy to determine which set of moment conditions is most suitable for the data at hand. Third, we consider and examine the performance of test statistics that help choose between MESS(1,1), MESS(1,0) and MESS(0,1) models. The performance of these tools is examined with Monte Carlo experiments, which also allow for varying degrees of spatial correlation in the explanatory variables. The correction in the standard errors is especially useful when the sample size is small, such as in a study with state-level, provincial or country-level data: the corrected standard errors improve statistical inference, yielding better size properties. The pretest strategy is effective when the heteroskedasticity, if present, is correlated with explanatory variables in the model. Spatial lags in the outcome variable are more easily detected than those in the disturbance. An empirical study of housing prices illustrates the new tools.</div></div>","PeriodicalId":15629,"journal":{"name":"Journal of Econometrics","volume":"254 ","pages":"Article 106181"},"PeriodicalIF":4.0,"publicationDate":"2026-01-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146015886","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"经济学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.jeconom.2026.106184
Yao Luo , Peijun Sang
We propose a class of sieve-based efficient estimators for structural models (SEES), which approximate the solution using a linear combination of basis functions and impose equilibrium conditions as a penalty to determine the best-fitting coefficients. Our estimators circumvent repeated solution of the structural model, apply to a broad class of models, and are consistent, asymptotically normal, and asymptotically efficient. Moreover, they solve unconstrained optimization problems with fewer unknowns and offer convenient standard error calculations. As an illustration, we apply our method to an entry game between Walmart and Kmart.
{"title":"Efficient estimation of structural models via sieves","authors":"Yao Luo , Peijun Sang","doi":"10.1016/j.jeconom.2026.106184","DOIUrl":"10.1016/j.jeconom.2026.106184","url":null,"abstract":"<div><div>We propose a class of sieve-based efficient estimators for structural models (SEES), which approximate the solution using a linear combination of basis functions and impose equilibrium conditions as a penalty to determine the best-fitting coefficients. Our estimators circumvent repeated solution of the structural model, apply to a broad class of models, and are consistent, asymptotically normal, and asymptotically efficient. Moreover, they solve <em>unconstrained</em> optimization problems with fewer unknowns and offer convenient standard error calculations. As an illustration, we apply our method to an entry game between Walmart and Kmart.</div></div>","PeriodicalId":15629,"journal":{"name":"Journal of Econometrics","volume":"253 ","pages":"Article 106184"},"PeriodicalIF":4.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145976962","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"经济学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.jeconom.2025.106172
Jiafeng Chen
This paper studies nonparametric identification and estimation of causal effects in centralized school assignment. In many centralized assignment algorithms, students face both lottery-driven variation and regression discontinuity- (RD) driven variation. We characterize the full set of identified atomic treatment effects (aTEs), defined as the conditional average treatment effect between a pair of schools given student characteristics. Atomic treatment effects are the building blocks of more aggregated treatment contrasts, and common approaches to estimating aTE aggregations can mask important heterogeneity. In particular, many aggregations of aTEs put zero weight on aTEs driven by RD variation, and estimators of such aggregations put asymptotically vanishing weight on the RD-driven aTEs. We provide a diagnostic and recommend new aggregation schemes. Lastly, we provide estimators and asymptotic results for inference on these aggregations.
{"title":"Nonparametric treatment effect identification in school choice","authors":"Jiafeng Chen","doi":"10.1016/j.jeconom.2025.106172","DOIUrl":"10.1016/j.jeconom.2025.106172","url":null,"abstract":"<div><div>This paper studies nonparametric identification and estimation of causal effects in centralized school assignment. In many centralized assignment algorithms, students face both lottery-driven variation and regression discontinuity- (RD) driven variation. We characterize the full set of identified <em>atomic treatment effects</em> (aTEs), defined as the conditional average treatment effect between a pair of schools given student characteristics. Atomic treatment effects are the building blocks of more aggregated treatment contrasts, and common approaches to estimating aTE aggregations can mask important heterogeneity. In particular, many aggregations of aTEs put zero weight on aTEs driven by RD variation, and estimators of such aggregations put asymptotically vanishing weight on the RD-driven aTEs. We provide a diagnostic and recommend new aggregation schemes. Lastly, we provide estimators and asymptotic results for inference on these aggregations.</div></div>","PeriodicalId":15629,"journal":{"name":"Journal of Econometrics","volume":"253 ","pages":"Article 106172"},"PeriodicalIF":4.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938447","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"经济学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.jeconom.2025.106171
Z. Merrick Li , Xiye Yang
We test for the presence of market frictions that induce transitory deviations of observed asset prices from the underlying efficient prices. Our test is based on the joint inference of return covariances across multiple horizons. We demonstrate that a small set of horizons suffices to identify a broad spectrum of frictions, both theoretically and practically. Our method works for high- and low-frequency data under different asymptotic regimes. Extensive simulations show our method outperforms widely used state-of-the-art tests. Our empirical studies indicate that intraday transaction prices from recent years can be considered effectively friction-free at significantly higher frequencies.
{"title":"Multi-horizon test for market frictions","authors":"Z. Merrick Li , Xiye Yang","doi":"10.1016/j.jeconom.2025.106171","DOIUrl":"10.1016/j.jeconom.2025.106171","url":null,"abstract":"<div><div>We test for the presence of market frictions that induce transitory deviations of observed asset prices from the underlying efficient prices. Our test is based on the joint inference of return covariances across multiple horizons. We demonstrate that a small set of horizons suffices to identify a broad spectrum of frictions, both theoretically and practically. Our method works for high- and low-frequency data under different asymptotic regimes. Extensive simulations show our method outperforms widely used state-of-the-art tests. Our empirical studies indicate that intraday transaction prices from recent years can be considered effectively friction-free at significantly higher frequencies.</div></div>","PeriodicalId":15629,"journal":{"name":"Journal of Econometrics","volume":"253 ","pages":"Article 106171"},"PeriodicalIF":4.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145880565","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"经济学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.jeconom.2026.106188
Shiwei Huang , Yu Chen , Jie Hu , Weiping Zhang
This paper introduces a dynamic panel data quantile regression model with network-linked fixed effects, named DQR-NFE, in which unobserved individual heterogeneity is structured through an underlying network. The corresponding estimator is derived by incorporating a quantile network cohesion (QNC) penalty into the dynamic panel quantile regression framework. This penalty encourages connected units within the network to exhibit similar conditional quantiles, with a particularly increased capacity to capture tail network dependence. Relative to conventional fixed-effects specifications, the proposed framework improves the estimation of unobserved heterogeneity and enables more accurate prediction in cold-start settings where training data are unavailable. We establish the consistency and asymptotic normality of the DQR-NFE estimators within a general nonlinear structural framework. These theoretical guarantees hold under both correctly specified and misspecified network structures, with an explicit characterization of their dependence on the network topology. Simulation studies and empirical applications reveal that the proposed estimator outperforms competing approaches in terms of both estimation accuracy and out-of-sample forecasting.
{"title":"Dynamic panel data quantile regression with network-linked fixed effects","authors":"Shiwei Huang , Yu Chen , Jie Hu , Weiping Zhang","doi":"10.1016/j.jeconom.2026.106188","DOIUrl":"10.1016/j.jeconom.2026.106188","url":null,"abstract":"<div><div>This paper introduces a <strong>d</strong>ynamic panel data <strong>q</strong>uantile <strong>r</strong>egression model with <strong>n</strong>etwork-linked <strong>f</strong>ixed <strong>e</strong>ffects, named DQR-NFE, in which unobserved individual heterogeneity is structured through an underlying network. The corresponding estimator is derived by incorporating a <strong>q</strong>uantile <strong>n</strong>etwork <strong>c</strong>ohesion (QNC) penalty into the dynamic panel quantile regression framework. This penalty encourages connected units within the network to exhibit similar conditional quantiles, with a particularly increased capacity to capture tail network dependence. Relative to conventional fixed-effects specifications, the proposed framework improves the estimation of unobserved heterogeneity and enables more accurate prediction in cold-start settings where training data are unavailable. We establish the consistency and asymptotic normality of the DQR-NFE estimators within a general nonlinear structural framework. These theoretical guarantees hold under both correctly specified and misspecified network structures, with an explicit characterization of their dependence on the network topology. Simulation studies and empirical applications reveal that the proposed estimator outperforms competing approaches in terms of both estimation accuracy and out-of-sample forecasting.</div></div>","PeriodicalId":15629,"journal":{"name":"Journal of Econometrics","volume":"253 ","pages":"Article 106188"},"PeriodicalIF":4.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146034683","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"经济学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.jeconom.2025.106173
Serafin Grundl , Yu Zhu
This paper proposes a new approach to the identification of first-price auctions that is robust to overbidding, but at the same time remains contiguous with the canonical point-identification approach of Guerre et al. (2000) (GPV) and its simple estimators. We show that a weak identifying restriction allows us to reinterpret the GPV estimates as a bound. We demonstrate that the identifying restriction holds in a set of commonly used auction models that can generate overbidding and is satisfied in the bid data from a laboratory experiment. We illustrate the approach in applications to laboratory data and field data. We recommend that practitioners continue to follow the GPV approach, but interpret the estimates as a bound in applications where they are concerned about overbidding.
本文提出了一种新的识别首价拍卖的方法，该方法对超标价具有鲁棒性，但同时与Guerre et al. (2000) （GPV）及其简单估计器的标准点识别方法保持一致。我们表明，一个弱识别限制允许我们将GPV估计重新解释为一个界。我们证明了识别限制在一组常用的拍卖模型中成立，这些模型可以产生过高的出价，并且在实验室实验的出价数据中得到满足。我们在实验室数据和现场数据的应用中说明了这种方法。我们建议从业者继续遵循GPV方法，但在他们担心过高出价的应用程序中，将估计解释为一个界限。
{"title":"A simple, robust identification approach for first-price auctions","authors":"Serafin Grundl , Yu Zhu","doi":"10.1016/j.jeconom.2025.106173","DOIUrl":"10.1016/j.jeconom.2025.106173","url":null,"abstract":"<div><div>This paper proposes a new approach to the identification of first-price auctions that is robust to overbidding, but at the same time remains contiguous with the canonical point-identification approach of Guerre et al. (2000) (GPV) and its simple estimators. We show that a weak identifying restriction allows us to reinterpret the GPV estimates as a bound. We demonstrate that the identifying restriction holds in a set of commonly used auction models that can generate overbidding and is satisfied in the bid data from a laboratory experiment. We illustrate the approach in applications to laboratory data and field data. We recommend that practitioners continue to follow the GPV approach, but interpret the estimates as a bound in applications where they are concerned about overbidding.</div></div>","PeriodicalId":15629,"journal":{"name":"Journal of Econometrics","volume":"253 ","pages":"Article 106173"},"PeriodicalIF":4.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146034686","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"经济学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.jeconom.2025.106178
Nan Liu , Yanbo Liu , Yuya Sasaki
We propose methods for estimation and uniform inference for a broad class of causal functions, such as conditional average treatment effects and continuous treatment effects, under multi-way clustering. The causal function is identified as the conditional expectation of a Neyman-orthogonal signal that depends on high-dimensional nuisance parameters. We introduce a two-step procedure: the first step uses machine learning to estimate the nuisance parameters, and the second step projects the estimated Neyman-orthogonal signal onto a dictionary of basis functions whose dimension grows with the sample size. We consider both full-sample and multi-way cross-fitting approaches to this procedure and derive a functional limit theory for the resulting estimators. For uniform inference, we develop a novel resampling method, the multi-way cluster-robust sieve score bootstrap, which extends the sieve score bootstrap of Chen and Christensen (2018) to settings with multi-way clustering. Extensive simulations demonstrate that the proposed methods exhibit favorable finite-sample performance. We apply our approach to study the causal relationship between mistrust levels in Africa and historical exposure to the slave trade. Accounting for the two-way clustering by ethnicity and region, our inference method rejects the null hypothesis of uniformly zero effects and uncover heterogeneous treatment effects, with particularly strong impacts in regions with high historical trade intensity.
{"title":"Estimation and inference for causal functions with multi-way clustered data","authors":"Nan Liu , Yanbo Liu , Yuya Sasaki","doi":"10.1016/j.jeconom.2025.106178","DOIUrl":"10.1016/j.jeconom.2025.106178","url":null,"abstract":"<div><div>We propose methods for estimation and uniform inference for a broad class of causal functions, such as conditional average treatment effects and continuous treatment effects, under multi-way clustering. The causal function is identified as the conditional expectation of a Neyman-orthogonal signal that depends on high-dimensional nuisance parameters. We introduce a two-step procedure: the first step uses machine learning to estimate the nuisance parameters, and the second step projects the estimated Neyman-orthogonal signal onto a dictionary of basis functions whose dimension grows with the sample size. We consider both full-sample and multi-way cross-fitting approaches to this procedure and derive a functional limit theory for the resulting estimators. For uniform inference, we develop a novel resampling method, <em>the multi-way cluster-robust sieve score bootstrap</em>, which extends the sieve score bootstrap of Chen and Christensen (2018) to settings with multi-way clustering. Extensive simulations demonstrate that the proposed methods exhibit favorable finite-sample performance. We apply our approach to study the causal relationship between mistrust levels in Africa and historical exposure to the slave trade. Accounting for the two-way clustering by ethnicity and region, our inference method rejects the null hypothesis of uniformly zero effects and uncover heterogeneous treatment effects, with particularly strong impacts in regions with high historical trade intensity.</div></div>","PeriodicalId":15629,"journal":{"name":"Journal of Econometrics","volume":"253 ","pages":"Article 106178"},"PeriodicalIF":4.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938446","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"经济学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-01DOI: 10.1016/j.jeconom.2026.106182
Ignace De Vos , Gerdie Everaert
Local projections (LPs) are widely used for estimating impulse responses (IRs) as they are considered more robust to model misspecification than forward-iterated IRs from dynamic models such as VARs. However, this robustness comes at the cost of higher variance, particularly at longer horizons. To mitigate this trade-off, several GLS transformations of LPs have been proposed. This paper analyzes two broad strands of GLS-type LP estimators: those that condition on residuals from an auxiliary VAR, and those that condition on residuals from previous-horizon LPs. We show that the former impose a VAR structure, which leads them to align with VAR IRs, while the latter preserve the unrestricted nature of LPs but end up replicating LP OLS estimates. Consequently, the intended efficiency gains are either not achieved or come at the expense of the very robustness that motivates the use of LPs.
{"title":"GLS estimation of local projections: Trading robustness for efficiency","authors":"Ignace De Vos , Gerdie Everaert","doi":"10.1016/j.jeconom.2026.106182","DOIUrl":"10.1016/j.jeconom.2026.106182","url":null,"abstract":"<div><div>Local projections (LPs) are widely used for estimating impulse responses (IRs) as they are considered more robust to model misspecification than forward-iterated IRs from dynamic models such as VARs. However, this robustness comes at the cost of higher variance, particularly at longer horizons. To mitigate this trade-off, several GLS transformations of LPs have been proposed. This paper analyzes two broad strands of GLS-type LP estimators: those that condition on residuals from an auxiliary VAR, and those that condition on residuals from previous-horizon LPs. We show that the former impose a VAR structure, which leads them to align with VAR IRs, while the latter preserve the unrestricted nature of LPs but end up replicating LP OLS estimates. Consequently, the intended efficiency gains are either not achieved or come at the expense of the very robustness that motivates the use of LPs.</div></div>","PeriodicalId":15629,"journal":{"name":"Journal of Econometrics","volume":"253 ","pages":"Article 106182"},"PeriodicalIF":4.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146034688","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"经济学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号