Turnover Flash Estimation by Purposive Sampling and Debit Card Transactions

Improving timeliness is ever more urgent for official statistics, not least due to the potentials of various non-survey big data that in principle can be made more quickly available than traditional sample surveys. Taking Retail Turnover Index as the case-in-point, we develop new approaches of model learning aimed to achieve flash estimation of acceptable accuracy, as well as the associated methods of uncertainty assessment, when one does not have the target observations that would have been required for unbiased inference by established statistical theories. Applications to the Norwegian data will be used to demonstrate the efficacy of our proposals.