Posterior Variance Test: Ex ante Evaluation of Grey Forecasting model

Abstract

Scholars frequently use ex post evaluation metrics such as the Mean Absolute Percentage Error (MAPE) to estimate the forecast accuracy. However, ex ante metrics are essential to know whether a given forecasting model is suitable for a given variable irrespective of the outcome of ex post evaluations. The ex ante measures help us ensure that the forecast is accurate, not by chance. The current study presents the Posterior Variance Test (PVT), which can serve as an ex ante measure of grey forecast accuracy. The study forecasted the methane emissions from Australia and India using a grey forecasting model and found that even though the MAPE generated "accurate forecasts" for both cases, the PVT invalidated the model's suitability for one of the two cases. The data visualization also corroborated the outcome of the PVT.