First of all, Happy Year of the Rooster to all Chinese friends, colleagues and others.
According to this article, 乱七八糟 (luàn qī bā zāo), meaning “a total mess”, ranks #2 at list of “the common and useful Chinese idioms”. Just as China comes back from a long holiday, it reminds us the distortion in economic data, which is caused by this break. This one is among many difficulties of dealing with Chinese economic figures (*), but for today, we will only talk about the New Year effect.
While, we are happy to share positive thoughts and best wishes for the Chinese New Year, any economist covering China probably can not help feeling a little unease nowadays. This is because the Lunar New Year holiday does not go very well with economic data.
Seasonal or working-day adjustments are always easier, when some holiday always falls into the same month of the calendar. However things get a little tricky, when these holidays, especially long ones, shift around the calendar. Eid al-Fitr and Eid al-Adha are two most common examples, which are celebrated by Islamic countries. Still, even these holidays can be adjusted in terms of economic data, since their movement around the calendar is almost mechanical. The hardest of all, should be a week-long holiday, which alternates between two months each year, namely the Chinese New Year.
Take this: this year the Chinese New Year was between 27 January and 2 February. Last year, it was 7-13 February. In 2012, it was 22-27 January, etc. Please add the fact that some factories or workers extend the holiday for one more week, on either end (lesser number of others do it on week-long National Day holidays, in early October). We hope by now, you can imagine how big is the distortion in economic data during the first two months of the year. If you haven’t, lets take a look at yoy % change in China’s foreign trade.
At the first two months of each year, the data makes such a blip that it looks more like a heart graph, rather than an economic figure. This can be seen in any data, especially if they are expressed in nominal terms.
The distortion caused by the Lunar New Year on Chinese economic data is so big that even the National Bureau of Statistics opts for not releasing seperate data for key activity indicators, such as year on year change of industrial production, fixed asset investment or retail sales. That’s why, we won’t be seeing any January figures for these and some other data in the coming days, as it has been every year.
Maybe, merging two months’ data for a single indicator solves the problem of seeing a highly volatile data for January and February, every year. However, it creates a new one, which is the mismatch between the number of observations among different series. For example, we have 12 monthly data for yoy % change on trade, but industrial production data includes only 11 observations. Even for the easiest task of putting these two in the same graph, will require an extra observation for industrial production. Here we employ two different solutions depending on which one gives us the smoothest line: (1) duplicating the January-February figure for both months or (2) we take the average of December and January-Febraury figure and put it on January, (or January-February and March average figure and put it on February).
We wonder what other China economists are doing to fix it?
*: We are not, in any way, referring to the doubts the authenticity of official numbers. Although we loosely buy the idea that some economic data could be ‘smoothened’, we think that Chinese economic data are still the best we have at hand.