A Neural Network Approach to the Measure of Management Accounting Practices

Management accounting is an important component of managerial decision making to help guide the overall business strategy in organizations. A management accounting index (MAI) can provide a quantitative measure of management accounting practices (MAPs) and help stakeholders determine an organizations’ maturity of MAPs. In this study, based on previously defined benchmark MAPs, we propose MAIs at the firm level by adopting word2vec, a machine learning based neural network language model, to measure six established value creation functions of management accounting. Based on a longitudinal analysis (from 2003 to 2017), we construct and validate our MAIs. We find firms that are either lacking resources or facing severe market competition tend to have high MAI scores. In addition, MAPs vary across Chinese firms with different levels of organization centralization and proportions of institutional shareholders. Lastly, we also discover significant and positive relations between environmental uncertainty and MAIs.

Keywords: Management Accounting Index, Management Accounting Practices, Neural Network Language Model

JEL Classification: M41;G14

Suggested Citation: Suggested Citation

Qiu, Fangjuan and Hu, Nan and Hu, Nan and Liang, Peng and Dow, Kevin, A Neural Network Approach to the Measure of Management Accounting Practices (September 9, 2020). Available at SSRN: https://ssrn.com/abstract=3710331 or http://dx.doi.org/10.2139/ssrn.3710331