Application of Artificial Neural Network to Corporate Liquidity Determination

Artificial neural network (ANN) is used to predict corporate liquidity (cash holdings) based on the sample for 45 countries during the period from 1994 to 2004. The benchmark determinants of corporate liquidity are used to conduct the outside tests. We find that the average R-value is above 0.8, suggesting that ANN outperforms the conventional regression models in predicting corporate liquidity. In addition, ANN helps identify five major determinants of corporate liquidity, confirming the results of previous studies. Further, training based on the combination of these five factors performs as well as that based on all the benchmark factors, suggesting that future corporate liquidity models should focus on these major factors rather than including too many variables.