Incorporating Google Trends Data in Predicting Consumer Confidence in Sri Lanka

The paper employs a spike-and-slab technique to select Google Trends categories into an augmented autoregressive model to nowcast Sri Lanka’s consumer confidence. The paper then examines the benefits of incorporating Google Trends data in predicting consumer confidence in Sri Lanka: i. Increasing prediction accuracy: The Central Bank of Sri Lanka is able to forecast consumer confidence with greater certainty. ii. Reducing time lag: The Central Bank is able to make real-time prediction of consumer confidence. The first benefit results from the ability of using Google search queries to reflect consumer confidence. Google Inc. classifies search queries into categories via Google Trends service. Incorporating Google Trends data increases the predictive power of the forecast model, compared to AR(1) baseline. The second benefit arises because Google Trends data are available in real time. Economic time series, like consumer confidence, are reported infrequently, often monthly or quarterly. Using Google Trends data reducing the time lag of consumer confidence reporting