Historical economic data are often uncertain due to sampling and non-sampling errors. But data uncertainty is rarely communicated quantitatively. An exception are the “fan charts” for historical GDP growth published at the Bank of England. We propose a generic loss function based approach to extract from these ex ante density forecasts a quantitative measure of unforecastable data uncertainty. We find GDP data uncertainty in the UK rose sharply at the onset of the 2008/9 recession; and that data uncertainty is positively correlated with popular estimates of macroeconomic uncertainty.