According to the article “China Trade Tops Forecasts in Holiday-Distored Month” from Bloomberg.com, China has seen a great economic recovery from a year ago. Exports have risen 25% and Imports have increased 28.8% from last year. All signs point to a growing and improving economy from these figures, however they aren’t telling the complete story. These numbers are “distorted by the weeklong Lunar New Year Holiday that was in January in 2012 and starts tomorrow for 2013”. This difference in the holiday accounted for five extra work days in 2012 than 2011. Clearly this could skew the data somewhat, but is it telling the whole story? I don’t think five extra workdays could account for the 25-28% increase that China has seen in imports and exports. In my opinion this is a sign of an improving global economy as the rest of the world has increased its demand for China’s goods and that China has increased its demand for the rest of the worlds goods. Is there a model that we could use to gauge the impact of these extra five days? I would suggest taking a weekly average and subtracting the final week of 2012 and then comparing it to 2011. Either way, China’s economy seems to be improving as does the global economy.
Source (photo came from Bloomberg.com)
Do the arithmetic! If you’ve 4.5 weeks then with a 6-day work week you have 26 work days, so 5/26 is 20%. Use a 5-day work week and you get to 25%. I assume that January 1st is a holiday, and if there are other national holidays that cuts the denominator further. Other random factors make monthly trade data volatile, there are big seasonal effects even without doing a work-day correction. But note that sales data in the US are routinely corrected for the number of holidays (esp when both Saturday and Sunday matter), which shift from year to year.
Your suggestion to do year-over-year is good in general, though you still need to do the holiday correction. However, you then have to check whether January 2011 was unusual, which it was due to the lingering global economic slowdown. And I think you simply can’t get weekly data, though if you could your suggestion of using averages would help.