Thursday, July 29, 2010

Global Trade Statistics

Following up on the other day's post about the apparent small contraction in global oil supply since February, I started looking around for other timely indicators of the state of the global economy.  It turns out that the World Trade Organization compiles both quarterly and monthly global statistics for world trade.  The data in the graph below are the quarterly data for global imports and exports.  The most recent point is for Q1, 2010 (published 6/2/10).  On a global basis, imports and exports should balance, so the difference basically provides some idea of the measurement error.


The data are a bit tricky to interpret:
  • We only have data from 2005 on, so there isn't much experience with this series.
  • The data are not inflation adjusted (no inflation adjustment is documented, and I confirmed by email with the WTO stats people that this is current US$), so presumably if they were, the rises would appear a little less pronounced, and the falls would be a little more pronounced.
  • There is some sign of modest seasonality in the data - note how there is generally a bit of an abatement of the trend between Q4 of one year, and Q1 of the next (green circles).  However, it's a bit scary to attempt a quantitative seasonal adjustment with so little data and such a huge non-seasonal boom/bust overlaid on it.
Nonetheless, it rather looks as if Q1 2010 trade was down relative to Q4 2009 (pink circle), and more so than was ever seen for the corresponding season in the period of rising trade from 2005-2008.

Turning to the monthly data, we gain the advantage of data through May 2010, but in addition to the problems of the quarterly data above, we gain the following additional problems:
  • There is quite a bit of month-to-month noise
  • There is even less data, as it only starts at the beginning of 2006.
  • There is missing data in recent months for some countries
I overcame the last problem by constructing an index which starts at 100 in January 2006, and then scaled it upward or downward based on the weighted average of whatever countries had available data.  There are isolated data points for June, but nowhere near enough to support a reasonable preliminary estimate.

Anyway, the resulting monthly series are here:


You can see that the drop-off in Q1 is made up of a very strong drop in Jan/Feb, followed by a strong rebound in March.  Then April and May declined from there.

Given the difficulties of interpreting the data, I don't want to draw too strong a conclusion here.  It's certainly consistent with the idea that the global economy is starting to contract again, but the data are too noisy to consider this as strong additional evidence at present.

No comments: