RDP 2015-13: Seasonal Adjustment of Chinese Economic Statistics 4. Concluding Remarks

The strong growth of the Chinese economy over recent decades has led to an increasing degree of attention being paid to monthly and quarterly releases of Chinese economic data. For China's trading partners, such as Australia, high-frequency movements in Chinese data may have implications for exchanges rates, stock markets and the real economy (for example, through the effect of Chinese domestic growth on its demand for imported commodities).

This paper argues that seasonal adjustment procedures (such as X-12-ARIMA and SEATS) can be helpful in interpreting Chinese data in real time. Unlike simpler methods often used in studies of the Chinese economy, such as regressing the time series on fixed seasonal dummies, these methods allow for the possibility that seasonality is time-varying. Changing seasonality may be important in transition economies such as China, where the evolution of macroeconomic aggregates is subject to rapid structural change.

The paper also proposes strategies to control for moving holidays such as Chinese New Year, the Dragon Boat festival and the Mid-Autumn festival. It generalises the approach of Lin and Liu (2003) by suggesting a simple procedure to optimise the selection of moving holiday regressors, and extending the method to moving holidays other than Chinese New Year. This procedure uses an information criterion to select an ‘optimal’ choice of moving holiday regressors from a large number of possible alternatives. The paper considers two variants of the procedure that utilise information regarding historical public holiday dates in China differently. Seasonal adjustment using these approaches yields results that compare favourably with rule-of-thumb techniques such as January–February averaging or the computation of year-on-year growth rates. In particular, it is found that the potential of Chinese New Year effects to spill over into the month of March reduces the reliability of simpler approaches.