The estimation of corporate liquidity management using artificial neural networks

Abstract

In this paper computational intelligence techniques are applied based on Artificial Neural Networks (ANNs) in order to investigate the liquidity performance of Greek listed firms. The effectiveness in managing their liquid assets is estimated with Multi-Layer Perceptrons (MLPs) and Probabilistic Neural Networks (PNNs) based on a set of financial ratios. Various MPL architectures are examined after modifying the number of nodes in the hidden layer, the transfer functions and the learning algorithms. The PNNs use spread values from 0.1 to 50, and 3 or 2 neurons in output layer, according to the coding of corporate liquidity management outcome. A PNN of 10-55-2 architecture is implemented to estimate the effectiveness of liquid assets management of Greek listed firms. We conclude that PNNs outperform MLPs, proved to be an appropriate computational intelligence technique for liquidity management estimation since the proposed PNN evaluate firm’s liquidity classification with 98.46% accuracy over the testing cases.