Scott Kelly @ on
Should I use Gravity Method or RAS?
Hello IO hive mind
We are updating the 2007 IRIO tables for China for 2012. The official tables won't be released for some time so we are updating the tables ourselves to do some analysis.
Many people have used the gravity model to update IO tables, but there are many different variants of this method and it has been applied inconsistently - obviously leading to very different end results. There does not appear to be any consensus on the 'right' coefficients or values to use in a gravity model (i.e. distance, travel time etc)
My favoured approach at this steage is to use the 2007 matrix as an initial condition and then use the RAS technique to update the table using total output values for 2012. My assumption here is that the best guess at what a table looks like in 2012 would be the previous known table, which was 2007. The gravity model appears to throw out this information in favour of the theory that regions that are closest trade the most.
What are your thoughts, does my much simpler method using 2007 data with RAS to update for 2012 hold water?