Filling the Data Gap: Unlocking Economic Nowcasting

In compiling and releasing official statistics, the Census and Statistics Department (C&SD) is committed to balancing “timeliness” with “data quality”. To ensure the public to grasp the economic situation of Hong Kong as early as possible, C&SD releases advance estimates on Gross Domestic Product (GDP) approximately one month after the end of the reference quarter. Compilation of GDP requires vast data sources including external trade statistics, various survey results and administrative data. However, some data sources especially survey results are not yet available at this early stage.

This is where nowcasting¹ plays a vital role. For some expenditure items that contribute significantly to GDP (e.g. restaurant receipts), C&SD tries to leverage some open-source statistical modules based on non-traditional time series models to nowcast such items. Unlike traditional methods, these modern modules adopt robust algorithms that offer flexibility in handling trends, seasonality and holiday effects, while also incorporating timely market indicators (e.g. visitor arrivals and the diffusion index on business receipts for SMEs) as external regressors, to scientifically fill the critical data gaps.

Nowcasting can provide initial yet reliable estimates for compiling advance estimates on GDP before official survey results are available, helping reduce the chance of significant revisions when survey data are incorporated later.

Looking ahead, C&SD will actively explore more open-source advanced modeling techniques and effective market indicators, striving to enhance the accuracy of nowcasting.

¹ “Nowcasting” refers to the prediction of the very recent past, the present, and the very near future state of an economic indicator.