PUBLICATIONS
1."Bootstrap inference for pre-averaged realized volatility based on non-overlapping returns," with Sílvia Gonçalves and Nour Meddahi, 2014, Journal of Financial Econometrics, 12(4), 679-707.
2."Validity of Edgeworth expansions for realized volatility estimators," with Bezirgen Veliyev, 2016, Econometrics Journal, 19(1), 1-32. This paper received the Denis Sargan Econometrics Prize for 2016, the Royal Economic Society.
3."Bootstrapping pre-averaged realized volatility under market microstructure noise," with Sílvia Gonçalves and Nour Meddahi, 2017, Econometric Theory, 33 (4), 791-838.
4."Bootstrapping integrated covariance matrix estimators in noisy jump-diffusion models with non-synchronous trading," 2017, Journal of Econometrics, 197(1), 130-152.
5."A local stable bootstrap for power variations of pure-jump semimartingales and activity index estimation," with Rasmus T. Varneskov, 2017, Journal of Econometrics, 198(1), 10-28.
6."Is the diurnal pattern sufficient to explain the intraday variation in volatility? A nonparametric assessment," with Kim Christensen, and Mark Podolskij, 2018, Journal of Econometrics, 205(2), 336-362.
7."Bootstrapping high-frequency jump tests," with Prosper Dovonon, Sílvia Gonçalves, and Nour Meddahi, 2019, Journal of American Statistical Association, 114, 793-803. Online Appendix.
8."A local Gaussian bootstrap method for realized volatility and realized beta," 2019, Econometric Theory, 35, 360-416.
9."The local fractional bootstrap," with Mikkel Bennedsen, Asger Lunde, and Mikko Pakkanen, 2019, Scandinavian Journal of Statistics, 46, 329-359. PDF.
10."Inference for local distributions at high sampling frequencies: a bootstrap approach," with Rasmus T. Varneskov, 2020, Journal of Econometrics, 215(1), 1-34. (Lead Article).
11."Estimating the variance of a combined forecast: bootstrap-based approach," with Kajal Lahiri, 2023, Journal of Econometrics, 232(2), 445-468.
12."A wild bootstrap for dependent data," 2023, Econometric Theory, 39(2), 264-289 .
13." Are Some Forecasters Really Better Than Others? A Further Note," with Kajal Lahiri, 2023, Journal of Money, Credit and Banking, 55, 577-593.
14."Bootstrapping two-stage quasi-maximum likelihood estimators of time series model", with Silvia Goncalves, Andrew Patton and Kevin Sheppard, PDF. MATLAB Toolbox, 2023, Journal of Business and Economics Statistics, 41(3), 683-694.
15."Bootstrapping Laplace Transforms of Volatility," with Rasmus T. Varneskov, and Zhi Liu., 2023, Quantitative Economics, 14(3), 1059-1103.
16."Heterogeneity in Carbon Intensity Patterns: A Subsampling Approach," with Johnson Kakeu and Li Lu, Energy Economics, Forthcoming.
17."Serial Dependence Robust Bootstrap Test for Cross-Sectional Correlation in Large Panel Models," with Chihwa Kao and Min Seong Kim, accepted at The Econometrics Journal.
18."A modified wild bootstrap procedure for Laplace transform of volatility," with Zhi Liu and Rasmus T. Varneskov, 2025, Economics Letters, 247, 112177
19."Forecasting economic time series in presence of weak factors: multiple supervised learning-based approach," with Zhendong Li, 2026, International Journal of Forecasting, 42(2), 414-433.
20."Wild Bootstrap Inference with Multiway Clustering and Serially Correlated Time Effects," with Jiahao Lin, 2025, accepted at Journal of Business and Economics Statistics.
21."Misspecification-robust bootstrap t-test for irrelevant factor in linear stochastic discount factor models," with Antoine Djogbenou, 2025, accepted at Journal of Econometrics.
22." Can mutual fund "stars" really pick stocks? New evidence from a wild bootstrap analysis," with Jiahao Lin, 2025, accepted at Journal of Empirical Finance.
WORKING PAPERS
1."A nonparametric test for diurnal variation in spot correlation processes," with Kim Christensen and Zhi Liu, 2025, Quantitative Economics, Revise and Resubmit.
Abstract: The association between log-price increments of exchange-traded equities, as measured by their spot correlation estimated from high-frequency data, exhibits a pronounced upward-sloping and almost piecewise linear relationship at the intraday horizon. There is notably lower---on average less positive---correlation in the morning than in the afternoon. We develop a nonparametric testing procedure to detect such variation in a correlation process. The test statistic has a known distribution under the null hypothesis, whereas it diverges under the alternative. We run a Monte Carlo simulation to discover the finite sample properties of the test statistic, which are close to the large sample predictions, even for small sample sizes and realistic levels of diurnal variation. In an application, we implement the test on a high-frequency dataset covering the stock market over an extended period. The test leads to rejection of the null most of the time. This suggests diurnal variation in the correlation process is a nontrival effect in practice. We show how conditioning information about macroeconomic news and corporate earnings announcements affect the intraday correlation curve.
2."Jackknife standard errors for two-way clustering with serially correlated time effects," with Jiahao Lin, 2025. Journal of Econometrics Economics, Revise and Resubmit.
Abstract: Chiang, Hansen, and Sasaki (2024) and Chen and Vogelsang (2024) developed cluster-robust variance estimators (CRVEs) for handling arbitrary serial dependence in linear regressions with two-way clustered panel data. However, conventional CRVEs often perform poorly in finite samples. We propose improved jackknife CRVEs to enhance inference accuracy. Through extensive simulations, we show that the novel jackknife CRVEs deliver remarkably precise inferences. This strong performance holds even in the presence of two-way fixed effects. Notably, one of our new approaches significantly mitigates issues of undefined standard errors when CRVEs are not positive definite, ensuring robust and consistent inference across scenarios.
3."Model Averaging in Semiparametric Estimation of Quantile Treatment Effects," with Sergio Firpo, Antonio Galvao, and Li Lu, 2025.
Abstract: This paper proposes model-averaging methods for estimation and inference of quantile treatment effects (QTE and QTT) under treatment selection based on observables. To address propensity score misspecification, we develop an estimator averaging QTE/QTT across models. To assign optimal model weights, we employ a model-averaging estimator that minimizes the asymptotic mean squared error. We establish the asymptotic distributions of the estimators, enabling the construction of valid confidence intervals, under two different scenarios. First, when the covariate set for each sub-model satisfies the unconfoundedness assumption. Second, when some (but not all) covariate sets violate the unconfoundedness assumption. Moreover, we construct a test for unconfoundedness for a given covariate set, assuming that the union of all covariates satisfies the assumption. We also introduce the ``unconfoundedness signature plot'' as a data-driven tool for assessing covariate relevance. Simulations show strong finite-sample performance of the proposed methods. We also illustrate the approach by estimating the effect of inherited control on firm performance.
4." Projection-Based Wild Bootstrap under General Two-way Cluster Dependence with Serial Dependence," with Jiahao Lin, 2025.
Abstract: This paper introduces a projection-based wild bootstrap (PWB) method for inference in linear regression models with two-way clustered data. We examine all possible scenarios for the asymptotic distribution of the estimator—Gaussian and non-Gaussian—classifying them into five distinct cases. In one scenario, no procedure can achieve uniform consistency under a fully unspecified DGP; to the best of our knowledge, our method is the first to cover the remaining four. We identify and apply two diagnostic factors to distinguish between these scenarios. In addition, our procedure accommodates arbitrary serial dependence. Simulation results demonstrate the accuracy and flexibility of the proposed method, making it a robust tool for empirical work involving complex clustering structures.
5."Robust Two-Sample Mean Inference under Serial Dependence," with Min Seong Kim, 2025.
Abstract: We propose robust two-sample tests for comparing means in time series. The framework accommodates a wide range of applications, including structural breaks, treatment–control comparisons, and group-averaged panel data. We first consider series HAR two-sample t-tests, where standardization employs orthonormal basis projections, ensuring valid inference under heterogeneity and nonparametric dependence structures. We propose a Welch-type t-approximation with adjusted degrees of freedom to account for long-run variance heterogeneity across the series. We further develop a series-based HAR wild bootstrap test, extending traditional wild bootstrap methods to the time-series setting. Our bootstrap avoids resampling blocks of observations and delivers superior finite-sample performance.
6. “Two-way Clustering Robust Variance Estimator in Quantile Regression Models,” with Jiahao Lin, 2026.
Abstract: We study inference for linear quantile regression with two-way clustered data. Using a separately exchangeable array framework and a projection decomposition of the quantile score, we characterize regime-dependent convergence rates and establish a self-normalized Gaussian approximation. We propose a two-way cluster-robust sandwich variance estimator with a kernel-based density ``bread'' and a projection-matched ``meat'', and prove consistency and validity of inference in Gaussian regimes. We also show an impossibility result for uniform inference in a non-Gaussian interaction regime.
7. “Estimation and Inference for the τ-Quantile Individual Heterogeneous Coefficient,” with Antonio Galvao, and Jiahao Lin, 2026.
Abstract: This paper proposes a two-step quantile estimation framework for analyzing heterogeneity in individual coefficients within panel data. Unlike conventional panel quantile regression, which focuses on outcome heterogeneity, our approach targets the cross-sectional distribution of individual-specific slopes. We establish asymptotic theory under both stochastic and deterministic designs, with convergence rates √N and √(N√T), respectively, and validate two corresponding bootstrap procedures. The method requires weaker growth conditions than fixed-effect quantile regression and accommodates large N settings. Simulations and an application to mutual fund performance illustrate its ability to capture diverse patterns of slope heterogeneity.
8. “Smoothed GMM for panel data quantile models,” with Antonio Galvao, and Jiahao Lin, 2026.
Abstract: This paper develops a theory for feasible estimators of finite-dimensional parameters identified by general conditional quantile restrictions for panel data models with fixed effects. This includes instrumental variables panel quantile regression as a special case. More specifically, we consider a set of unconditional moments implied by the conditional quantile restrictions, providing conditions for local identification. Since estimators based on the sample moments are generally impossible to compute numerically in practice, we study feasible estimators based on smoothed sample moments. We propose a generalized method of moments estimator and establish its consistency and asymptotic normality. Inference procedures are discussed. Numerical simulations illustrate the finite-sample properties of the methods.
9. “A Free Probability Theory based LM Test for Cross-Sectional Dependence in Large Panel Data Models with Serial Correlation,” with Chihwa Kao and Min Seong Kim, 2026.
Abstract: This paper develops a robust test for cross-sectional independence in high-dimensional panels with serially correlated observations. Existing tests based on the trace of the sample correlation matrix assume temporal independence, which can lead to severe size distortions in economic and financial panels. We derive analytical corrections to the mean and variance of Tr(R²) , where R denotes the sample correlation matrix, under general Toeplitz temporal dependence. The corrections depend only on spectral moments of the temporal covariance matrix and yield a test statistic that remains asymptotically normal under the joint growth of the cross-sectional and temporal dimensions. The proposed method avoids computationally intensive resampling procedures while maintaining accurate size and competitive power.
10. “Supervised Mixed-Frequency Learning for Macro-Financial Forecasting When Factors are Weak,” with Zhendong Li, 2026.
Abstract: Factor-MIDAS regression models are used to forecast a target variable by extracting common factors from a large panel of high-frequency predictors using principal component analysis (PCA). While PCA mitigates the curse of dimensionality, it relies on the pervasiveness of factors, an assumption often violated in practice when weak factors are present, particularly in macro-financial forecasting. To address this, we propose and theoretically justify the application of the so-called supervised scaled principal component analysis (SsPCA) in the context of mixed-data sampling. The SsPCA integrates supervised data weighting to shrink noise components and selectively exploit relevant predictors, enhancing weak factor identification and thereby improving predictive accuracy. Simulation results indicate that SsPCA outperforms other PCA-based or supervised methods, particularly when weak factors are prevalent. In addition, applying mixed-frequency machine learning techniques such as boosting to the cleaner factors extracted by SsPCA yields further gains in predictive performance. Finally, an extensive empirical application to U.S. macro-financial forecasting provides evidence that SsPCA identifies economically meaningful predictors and improves forecasts of GDP, inflation, unemployment, asset prices, and financial market volatility.
11."Two-Stage Model Averaging for Impulse Responses: Local Projection- and VAR-based Approaches" with Seojin Jung, 2025.
Abstract: This paper proposes a model averaging-based estimators of impulse response functions, aiming to effectively combine estimators from a wide range of possible approaches for the effects of the same economic shock. Our proposal allows estimating the weighted combination not only from different models but also from different estimation methods, i.e., using Local Projection (LP)-based estimators and/or vector autoregression (VAR)-based estimators of impulse response functions. Our proposal also allows computing the averaged impulse response functions estimators by using different statistical paradigms (i.e., Bayesian or frequentist). Our simulation study and empirical applications illustrate a typical bias-variance trade-off: estimators of LP-based approaches exhibit lower bias, whereas estimators of VAR-based approaches exhibit lower variance. Additionally, they demonstrate how the weights assigned to different approaches leverage their benefits and balance the properties of LPs and VARs.