CEIC China GDP Nowcast

The GDP Nowcast methodology is based on a set of selected high-frequency indicators, employed in a dynamic factor model. Dynamic factor models outperform standard univariate models such as random-walk or autoregressive models in estimating GDP. By using this approach, we update our GDP Nowcast, the estimate of GDP growth in the current and the next quarter, every time when new data for any of the high-frequency indicators is available. The model uses a dynamic factor of 1.

For China, a total of 23 indicators have been selected, spanning the whole range of the economy such as external trade, domestic demand, industry, financial sector, prices, and property market. All these indicators have had a strong correlation with the GDP growth historically. Most of them are published monthly, except rainfall, which is a daily indicator aggregated to a monthly frequency to fit the model and be consistent with the other indicators. Since timeliness is an important aspect of the real-time monitoring of the economy, the maximum data release lag for each indicator is 30 days.

Current quarter means the quarter for which the GDP data is expected to be released by China’s National Bureau of Statistics (NBS), the official source of China GDP data, while next quarter refers to the following quarter. NBS usually releases the quarterly GDP data around two weeks after the end of the reference period.

Yes, all components used to calculate China GDP Nowcast have equal weights as it has been proven empirically that defining different weights or factors does not improve the model predictive power.

The GDP Nowcast for the current and the next quarters are revised every time when a new data for any of the underlying indicators is released. The new data improves the GDP Nowcast accuracy as it increases the volume of information used by the model to estimate the current and the next quarter GDP. The variance of the estimate also declines with the new data releases.

While the movement of individual indicators is suggestive of the direction of the economy, our model interprets the values of these indicators differently, since they are treated as part of the set. Hence, the values could decline, like in the case of air passenger traffic post April 2020, and yet the impact on the GDP be positive, and vice versa.