An early risk warning system for Outward Foreign Direct Investment in Mineral Resource-based enterprises using multi-classifiers fusion

Abstract

Outward foreign direct investment in mineral resource-based enterprises (OFDI-MREs) is usually a substantial long-term investment. However, as it is affected by many uncertain factors, the investment process is full of risks. In order to reduce or lessen the investment risk of enterprises and improve the scientific approach to decision-making, it is of great significance to construct an efficient early risk warning system. In this paper, a novel method which combines the coefficient of variation method, system clustering and multi-classifier fusion to early-warn the risk of OFDI-MREs is proposed. The validity of the model is verified by using 173 sample data from 42 MREs in China. The main results are as follows: First, a hierarchically-structured risk warning indicator system with 20 indicators in three dimensions is obtained with indicator reduction; Second, the risks facing OFDI-MREs is classified into four levels based on the rate of return on equity, earnings per share, and capital accumulation rate, and most of the OFDI-MREs are at high risk; Third, the proposed multi-class fusion technology based on self-organizing data mining had higher accuracy and stability than the four widely used single-classifier models (logit regression, support vector machine, neural network, Decision Tree) and the six commonly used multi-classifier fusion methods (such as majority voting, the Bayesian method, and genetic algorithm). Accordingly, some targeted policy implications are put forward in terms of institutional distance, enterprise resource and competency foundation, which may help MREs to reduce the OFDI risks and enhance their risk prevention capabilities.