Cash flow prediction using a grey-box model

This paper tackles the problem of financial forecasting by extending methods developed in automation, engineering and computing science. Current methods existing in the literature for firm-level cash flows are first analysed. Then a grey-box modelling method is developed to elevate the performance of cash-flow prediction. Linear panel data modelling is used as a benchmark model. Experiments with out-of-sample tests are used to validate the grey-box approach. Encouragingly, nonlinear grey-box modelling outperforms linear panel data modelling in both one-period-ahead and multi-period-ahead predictions.