Measuring the level of global economic activity is of key interest. But the measures we have on variables like industrial production don’t cover all countries and are only available with a significant lag. Michigan Professor Lutz Kilian suggested in an influential paper published in 2009 that we could get a useful timely indicator by looking at average shipping costs. I recently had a chance to look into the details of how that series is constructed and have some suggested improvements.
Kilian’s basic idea is that changes in world economic activity are the primary driver of the demand for shipping and that this higher demand shows up in the short run as an increase in the real cost of shipping. Kilian calculated a nominal index of shipping costs (which I’ll denote by x) by starting with a value for 1968:M1 of x = 1 and adding an average of the monthly change in the log of different shipping costs each month to get a new value for xt for each month t up to 2007:M12. Since 2008, Kilian has simply added the monthly change in the log of the Baltic Dry Index (BDI) of shipping costs to the previous month’s value of x. That procedure implies that for data since 2008,
The value of c0 turns out to depend on the value of x and of the BDI in 2008: M1. Kilian has never reported the values for x itself or the value of c0, but it is possible to uncover both exactly from data he has made public, as I describe in this note. The value of c0 turns out to be -5.236.
Kilian’s next step was to take the log of x, subtract off the log of the CPI, and regress the difference on a time trend:
The residuals from this second regression are Kilian’s index of real economic activity, which he regularly updates on his website and which have been used in dozens of economic studies.
