Prediction of poor working capital development. Can poor working capital development be predicted using accounting based financial information?

Abstract: We investigate if accounting based financial information on an operating entity level can be used to predict poor working capital development (WCD) on a quarterly basis. The development is considered poor if the change in net working capital is not driven by an underlying change in customer orders, and hence poor in the sense that the development is unjustified. The study is performed using data from a Swedish multinational industrial company covering the period 2009-2013, resulting in 2,420 entity-quarter observations. The operating entities included are either customer centers (CCs) or production centers (PCs), where the main activity is selling for the former and manufacturing for the latter. The statistical method logit analysis is used to create three models. The model using all data results in a prediction power of 79.21%. Separating CCs and PCs, results in prediction powers of 81.10% and 76.90% respectively. Thus, prediction of poor WCD is possible using accounting based financial information in the form of key ratios. The key ratios found to have explanatory power to predict poor WCD are asset turnover, inventory intensiveness, asset structure, asset liquidity, an entity's historical frequency of poor WCD quarters, growth in equity and growth in debt. If the models are evaluated based on cost savings, the separate models are superior compared to the model estimated based on both CCs and PCs. The study performed is the first in its field in terms of predicting poor WCD and further research on the subject is encouraged.