Forecasting corporate earnings: integrating financial ratios and earnings management predictors

Purpose This study aims to develop the traditional financial ratio-based earnings prediction model and the earnings management predictors-based prediction model to access each firm’s earnings status to minimize agency costs. Design/methodology/approach This study utilizes data from 322 companies from 2009 to 2022, which accounted for a total of 3,220 firms-year observations. The forty-five (45) traditional financial ratios and eight (8) earnings management predictors are chosen from previous literature. Stepwise logistic regression is used to develop the earnings prediction models, while the maximum likelihood ratio is used to determine the selection of the final model. The first step begins with the development of the traditional financial ratio-based earnings prediction model, followed by the earnings management predictors-based earnings prediction model and finally the earnings prediction model. Findings The findings of the last step identified the five discriminatory traditional financial ratios, where four (4) variables belonged to the statement of financial position and one (1) variable belonged to the statement of profit and loss, with a predictive accuracy of 80% of the overall model. Research limitations/implications The external validity of the developed earnings prediction model can be tested on large data sets of different financial markets. Practical implications The traditional financial ratio-based earnings prediction model developed will reduce agency conflict, which will benefit the firm. Using the predictive model, stakeholders can access a firm’s earnings status and management’s strategy for short-term and long-term firm decisions. Originality/value The developed earnings prediction model’s discriminatory variables are the proxies of working capital policy, working capital investment policy, leverage, short-term liquidity risk and firm size.