# German Export in August and the Toll Index

German Exports in the month of August were announced today by Destatis. The data is still preliminary but it does suggest an increase of 7.2% with respect to August 2016. According to the German Press (for example finanzen.net) these results “surprised German Economists”.

On September 21 I published the Toll index for the month of August. How much less surprised would you have been if you used the Toll Index to forecast exports?

If $E$ are monthly exports (in billions of Euros, $T$ is the Toll Index (border crossing trucks), $\Delta$ is the monthly difference operator and $m_i: i= 1,\ldots,12$ monthly dummies  and you estimated a model like this:

$\Delta(E) = \alpha \Delta(T)+\displaystyle{\sum_{i=1}^{12}} \beta_{i}m_i,$

you would have come up with +5.7% which would imply you would be much less surprised. There is still some difference to 7.2% but first this is a preliminary estimate, second there is some noise in the world and third the model is build and run in under 5 minutes. Of course you would be almost just as informed if you just saw this graph:

This entry was posted in Big Data, data, economics, nowcasting, technology and tagged , , . Bookmark the permalink.