Bank of England publishes working paper on structural VAR forecast analysis, links UK inflation forecast revisions to supply, demand and past shock reassessments

The Bank of England has published a staff working paper that sets out a framework for using the structural form of vector autoregressive models to explain forecast errors and forecast revisions, rather than only reporting how far forecasts missed the eventual outturn. The paper argues that structural analysis can give policymakers and researchers a clearer narrative of what drove changes in forecasts in real time. In a stylized UK application focused on the post-pandemic inflation surge, it finds that upward revisions to a four-variable VAR inflation forecast reflected not only contractionary supply-side forces, but also expansionary demand-side shocks and revised estimates of earlier shocks. The framework decomposes forecast errors and revisions into the effects of new shocks hitting the economy, changes in how long past shocks are expected to persist, changes in simulated shocks used in conditional forecasts, updates to the model’s estimated unconditional mean, and other deterministic components. To illustrate the approach, the paper applies a structural VAR for the UK using Bank rate, real gross domestic product, the consumer price index and real oil prices, and identifies demand, supply, energy and monetary policy shocks. In its real-time exercise, the paper concludes that the initial inflation surge and the upward revision to the 2022 second-quarter inflation forecast were driven by a mix of inflationary supply and energy shocks as well as expansionary demand and monetary policy shocks, while part of the forecast errors and revisions in 2022 and 2023 reflected a reassessment of the importance of past shocks.