Network fixed effects regressions are widely used to control for unobserved heterogeneity across connected units, but their validity depends on observing the network accurately. This paper studies fixed effects regression when network links are misclassified. Misclassification happens when either some true links are missing or some recorded links are spurious. We show that the standard estimator is generally inconsistent, and that the two errors play very different roles: spurious links drive the bias. Because spurious links behave as if every pair of units were weakly connected, they attenuate broad, smooth variation in the fixed effects — differences at the level of regions, communities, or sectors — far more than localized variation, systematically washing out the large-scale heterogeneity that is typically of interest. We propose a bias-corrected estimator that removes the spurious-link component of the observed network and rescales for missing links. The corrected estimator is consistent when the true network is sufficiently well connected, supports asymptotically normal inference for the smooth functionals, and remains feasible because the misclassification rates can themselves be recovered from repeated measurements of the network.

In this paper, we develop online updating methods for linear panel regression models. Online updating refers to procedures for sequentially updating parameter estimates as new data become available. In practice, the potential size of the dataset or data confidentiality constraints may preclude researchers from storing or accessing the entire dataset. We propose an online updating procedure for widely used linear regression models in panel data, where data expansion can occur through either (1) the arrival of new units or (2) the arrival of additional time periods for existing units. The proposed procedure yields closed-form expressions for updating both the point estimates and associated standard errors in each scenario.

Award: Best Third-Year Paper, Department of Economics, Seoul National University.

Presentations: SNU Econometrics Workshop, SETA 2025 (University of Macau), University of Sydney, KERIC 2025 (SNU), SNU Workshop on Recent Advances in Econometrics, SETA 2026 (University of Tokyo).

Researchers often use fixed-effects and lagged-dependent-variable (LDV) estimates as upper and lower bounds on a treatment effect. This paper studies this comparison in staggered-treatment designs and shows that the conventional ordering of the two estimates is not generally preserved. The direction of the difference depends on cohort- and event-time-specific treatment effects, outcome persistence, and the alignment of treatment timing with unit and time heterogeneity. Thus, either estimator may serve as the lower or upper bound, and the ordering may vary across group-time cells. Monte Carlo simulations further show that the interval formed by the two realized point estimates frequently fails to contain the true group-time average treatment effect in finite samples. The results suggest that fixed-effects and LDV estimates should be viewed as a descriptive comparison rather than as guaranteed bounds.