On the (Un)foreseeable path of the UK economy through uncertain times

Abstract

Against the backdrop of uncertain times we assess the out-of-sample forecasting performance of UK GDP growth, CPI inflation, unemployment rate and the policy interest rate of the Bank of England (BoE). A large time-varying Bayesian model with domestic variables (such as economic policy uncertainty, Divisia M4 growth, and financial stress) and international variables (such as geopolitical risk, the US interest rate, and US financial stress) is ranked first for CPI inflation, followed by the BoE's forecasts in second place. The large Bayesian model is ranked first for GDP growth at the four-step-ahead forecast horizon and remains inferior only to the BoE's forecasts at shorter forecast horizons. Smaller versions of the Bayesian model provide superior unemployment rate forecasts and predict, together with market expectations of interest rates, the policy interest rate better than competing models. An Augmented AutoRegressive (AR) model of global supply chain pressures, pandemic effects, and trade uncertainty offers significant forecasting power for UK inflation and very short-term GDP growth. The Augmented AR model makes no assumptions about the future path of the policy interest rate. This might be an attractive and alternative option to the BoE's forecasting analysis which currently relies on market expectations of interest rates.