Approximating the DGP of China's Quarterly GDP

We demonstrate that the data generating process (DGP) of China’s cumulated quarterly Gross Domestic Product (GDP, current prices), as it is reported by the National Bureau of Statistics of China, can be (very closely) approximated by a simple rule. This rule is that annual growth in any quarter is equal to annual growth in its previous quarter plus an error term that is only nonzero in the first quarter of each year and with small variance. We show that this rule fits the data for 1992Q1 to 2005Q4 well, for total GDP as well for its three sector-specific components. It also gives accurate forecasts for 2006Q1 to 2009Q4. We also study the time series properties of GDP growth in constant prices, and show that these series behave as random walks, with much larger error variance.