Financial statement complexity contributes to a reduction in the reliability of accounting data used in firm valuation. The present paper examines whether the complexity of financial statements can be accurately forecasted. Assessing forecast accuracy is important because adjustments to a firm’s value are appropriate only when complexity is expected to persist in future years. In the absence of sufficient information to assess reliability, this study used the number of items in the balance sheet and income statement as a proxy for financial statement complexity. Ratios from a sample of 3037 Italian Micro and Small Enterprises (MSEs) over a 10-year period were used to forecast the compound annual growth rate of financial statement items, employing logistic regression and neural network models. The results show that the percentage of correct out-of-sample predictions ranges from 67.4 to 73.9%, depending on the forecasting model. Confusion matrices provide a more detailed view of model performance, while logistic regression highlights the variables with the greatest statistical significance.
Pierluigi Santosuosso (Wed,) studied this question.