Recent GDP data for Italy have rekindled concerns about how well some countries are moving out of the global financial crisis. Professor Justin Wolfers plotted a comparison between real GDP in Italy and the United States that shows the dismal Italian “recovery” and hints at the possibility of a triple-dip recession. (FRED lets you plot this graph pretty quickly.) Several Italian commentators have also made comparisons between Italy and Japan. But these FRED graphs show that the path of Japan’s GDP is more similar to that of U.S. GDP. And, as Professor Wolfers points out, U.S. GDP hasn’t been all that bad in an international context.
Italy’s GDP appears even more dismal if you consider real GDP per capita, which smooths out differences in population growth:
In terms of real GDP per worker (a ratio also used as a measure of labor productivity), Japan’s trend has diverged from the U.S. trend only since the global financial crisis. Because there is a tighter relationship between employment and GDP in the United States than in Japan, real GDP per worker in the United States hardly reveals a recession at all: As GDP was falling in 2008-09, the number of employed workers was also dropping. In Japan, however, workers were not being laid off in such large numbers, so the ratio declined more. Chalk that up to stark differences in the labor markets of these two countries.
Yet, the divergence of Japan from the United States is dwarfed by that of Italian real GDP per worker, showing a dismal protracted reduction since the global financial crisis.
How these graphs were created: The first and second graphs simply use data on real GDP and real GDP per capita, rebasing them to 100 in 2001 using the options under the “EDIT DATA SERIES” tab: Select “Index (Scale value to 100 for chosen period)” and choose the 2001 option. Note that this is a default option for rebasing the series, but one can also choose different dates. Construct the third graph as follows: Create the ratio of the original series (real GDP = a and civilian employees = b; a/b) and then apply the transformation “Index (Scale value to 100 for chosen period)” and again choose 2001. Finally, remove the legend axis on this last graph, which reduces the clutter.
Suggested by Silvio Contessi