Making investment decisions using XBRL filing data

Abstract

Purpose

The Securities Exchange Commission mandated eXtensible Business Reporting Language (XBRL) filing data provide immediate availability and easy accessibility for both academics and practitioners. To be useful, this data should provide information for decisions, specifically, investment decisions. The purpose of this study is to examine whether the XBRL database can be used with models, developed in previous studies, predicting the directional movement of earnings. The study does not attempt to examine the validity of these models, but only the ability to use the data in the analysis of financial statements based on these models.

Design/methodology/approach

The study analyzes New York Stock Exchange companies’ XBRL data using a two-step logistic regression model. The model is then used to arrive at the directional movement of earnings between current and subsequent quarters. Additional models are created by dividing the sample into industry membership.

Findings

The results classified companies as realizing an increase or a decrease in earnings. The final model indicated a significant ability to predict earnings changes, on average about 65 per cent of the time, for the entire model, and 71 per cent, for the industry-based models (higher than those of previous studies based on COMPUSTAT). The investment strategy created average quarterly return between 2.8 and 10.7 per cent.

Originality/value

The originality of this study is in the way it examines the quality of XBRL data, by examining whether findings from prior research which relied on traditional databases (such as COMPUSTAT) still hold using XBRL data. The use of XBRL allows not only easier and less-costly access to the data but also the ability to adjust the models almost immediately as current information is posted, thus providing a much more relevant tool for investors, especially small investors.

Citation

Palas, R. and Baranes, A. (2019), "Making investment decisions using XBRL filing data", Accounting Research Journal, Vol. 32 No. 4, pp. 587-609. https://doi.org/10.1108/ARJ-01-2018-0002