Event count estimation. (7th February 2022)

• Record Type: Journal Article
• Authors: Balazsi, Laszlo; Chan, Felix; Matyas, Laszlo
• Is Part Of: Econometric reviews. Volume 41:Number 2(2022)
• Page Start: 147
• Page End: 176
• Publication Date: 2022-02-07
• Subjects: Big data -- outliers -- robust estimation -- event count estimation -- censoring; excess zeros; C13 -- C24 -- C55; Econometrics -- Periodicals; 330.015195
• DOI: 10.1080/07474938.2020.1862505
• Languages: English
• ISSNs: 0747-4938

Abstract

Abstract: This paper proposes a new estimation procedure called Event Count Estimator (ECE). The estimator is straightforward to implement and is robust against outliers, censoring and 'excess zeros' in the data. The paper establishes asymptotic properties of the new estimator and the theoretical results are supported by several Monte Carlo experiments. Monte Carlo experiments also show that the estimator has reasonable properties in moderate to large samples. As such, the cost of trading efficiency for robustness here is negligible from an applied viewpoint. The practical usefulness of the new estimator is demonstrated via an empirical application of the Gravity Model of trade.