Data-rich economic forecasting for actuarial applications

With the advent of Big Data, machine learning, and artificial intelligence (AI) technologies, actuaries can now develop advanced models in a data-rich environment to achieve better forecasting performance and provide added value in many applications. Traditionally, economic forecasting for actuarial applications is developed using econometric models based on small datasets including only the target variables (usually around 4-6) and their lagged variables. This paper explores the value of economic forecasting using deep learning with a big dataset, Federal Reserve Bank of St Louis (FRED) database, consisting of 121 economic variables and their lagged variables covering periods before, during, and after the global financial crisis (GFC), and during COVID (2019-2021). Four target variables considered in this paper include inflation rate, interest rate, wage rate, and unemployment rate, which are common variables for social security funds forecasting. The proposed model “PCA-Net” combines dimension reduction via principal component analysis (PCA) and Neural Networks (including convolutional neural network (CNN), Long Short-Term Memory (LSTM), and fully-connected layers). PCA-Net generally outperforms the benchmark models based on vector autoregression (VAR) and Wilkie-like models, although the magnitude of its advantage varies across economic variables and forecast horizons. Using conformal prediction, this paper provides prediction intervals to quantify the prediction uncertainty. The model performance is demonstrated using a social security fund forecasting application.