A Mixed Data Sampling Approach to Accounting Research

Category: Accounting Research Seminar
When: 10 October 2022
, 14:15
 - 15:30
Where: RuW - RuW 4.201 / via Zoom


This paper examines a mixed data sampling (MIDAS) approach to accounting research. MIDASregression models parsimoniously incorporate variation embedded in existing economic datathat are observed at much higher frequencies than accounting data. The additional source ofdata variation creates an opportunity to address new and important questions in accountingresearch. We develop and outline four new MIDAS models within the general frameworkthat are simple to estimate and capture economic properties that are relevant to accounting research.We demonstrate the efficacy of our models with empirical applications to the Januaryeffect and to earnings response coefficients, which together illustrate the potential to expandthe boundaries of accounting research by getting more out of high-frequency data.