Taking the normal approach as it is, we can calculate the “estimation error” as the initial estimate minus the actual outcome, looking at differences over time. We’ve looked at experience errors for gilts alone and gilts compared to other measures. Our tentative conclusions are here.