National Statistical Offices are being asked to produce more granular information in a more timely manner but to have the same high quality as the less detailed and timely data that have been produced for many years. Many feel that the solution to this conundrum is as easy as using alternative data. However, it is becoming clear that alternative data is not necessarily always the answer due to challenges such as conceptual differences, quality of the data, access to the data, etc.. This does not mean that alternative data should not be consider as they have a lot of upside in terms of being used as auxiliary data for modelling purposes. This session considers several ways that alternative data can be combined with more traditional data through models to produce official statistics with higher quality than the alternative or the traditional data themselves. It includes presentations on producing small area estimates, nowcasting methods and imputation methods to improve sample designs for the estimation of rare characteristics by three well known researchers from the UK, the USA and Brazil, as well as a well known discussant from Canada.

Organiser: Dr Wesley Yung 

Chair: Wesley Yung 

Speaker: Paul Smith 

Speaker: Denise Silva 

Speaker: Katherine Thompson 

Discussant: Mr Jean-Francois Beaumont