Guan Yu Zheng

Statistics New Zealand

Viability of reducing survey frequency using administrative data

Administrative data is data collected to fulfill administrative purposes, instead of statistical purposes. Statistics New Zealand would like to expand the use of administrative data in various ways. One way is to investigate the possibility of reducing survey frequency by using administrative data in time series. Using administrative data to reduce survey frequency requires that administrative data possess highly similar statistical properties to survey data. In this paper, we explore whether Electronic Card Transactions (ECT) as an administrative data source is a good leading indicator of the Retail Trade Survey (RTS), as a first step towards assessing the viability of reducing survey frequency. For Statistics New Zealand, a good leading indicator should provide high correlations and similar seasonal adjustment estimates to the actual ones. We found ECT and RTS are highly cointegrated with constant seasonal variation. In one step ahead forecasting, forecasts are unbiased and the differences in seasonal adjustments are relatively small for subgroups of RTS.