Prediction of technical efficiency and financial crisis of Taiwan’s information and communication technology industry with decision tree and DEA

This study aims to analyze the business performance and technical efficiency of Taiwan’s ICT industry with the Malmquist productivity index of data envelopment analysis. The regression method is used to verify the influence of ICT industry labor input and research input on yield. The tested units include Taiwan companies among the top 250 companies of the ICT industry in the world, as well as companies with great contributions to the ICT industry in Taiwan, for a total of 16 objects. Using the data mining classification method, the usage variables in the financial crisis model are utilized to predict whether the technology of the tested units is efficient. The research results are as follows. First, during the test period, 3, 3 and 6 companies have, respectively, increasing returns, constant returns, and decreasing returns to scale. This suggests that in Taiwan’s ICT industry, only 3 companies can continuously grow to seek the maximum benefit, while 6 are in a period with stable efficiency, and 6 are in a period with declined operating efficiency and increasing costs. Second, labor input has a significant positive correlation with yield, and the influence of labor input on yield is relative to other inputs. The influence of R&D on yield has a positive but insignificant correlation, which is different from previous research. This is because the effect of research input is not relatively obvious, thus requiring to increase the items to be input. Third, by combining two different variables, this study uses the financial crisis precaution model and data envelopment model to predict technical inefficiency before and after the financial tsunami of 2008. No matter whether before or after the financial tsunami of 2008, the financial crisis precaution model is more accurate than the data envelopment model in terms of prediction of technical inefficiency.