Robert Inklaar

Some recent new discussions about total factor productivity (TFP) numbers in @PennWorldTable, with the notably reductive post by @Noahpinion as a low point: x.com/Noahpinion/sta… (1/)

@ajpirzada @JonSteinsson We switched to the official data for China, as we explain here: dataverse.nl/api/access/dat…. There is a bigger debate about statistics quality, but not one we address if we only adjust Chinese data.

I have spent the last couple of days looking at convergence in East Asian economies (Japan, South Korea, China, etc.) toward the U.S. The figure below uses the recently released Penn World Table (PWT) release 11.0: rug.nl/ggdc/productiv… In orange, it plots TFP growth in China and in a representative East Asian economy (an average across these economies), always relative to the U.S. A value of 2% in 2016 in the left panel means that China’s TFP grew 2 percentage points faster than U.S. TFP in that year. In blue, I plot average relative TFP growth over 15-year windows. For China, the series starts in 1979, with the beginning of economic reforms. For the representative East Asian economy, each country enters when its per capita income reaches Japan’s 1950 level, the first observation available in the dataset, and roughly the threshold the World Bank uses to define middle-income status. Thus, an x-axis value of 25 in the right panel means relative TFP growth 25 years after reaching Japan’s 1950 income per capita. Start with the right panel. The representative East Asian economy follows a standard TFP catch-up process: adoption of foreign technology and managerial practices, and the reallocation of labor from agriculture to manufacturing and services. This process lasts about 45 years. After year 45, catch-up to the U.S. stops. If anything, there is mild divergence. This pattern closely mirrors Western Europe between 1945 and 1990: rapid convergence that lost momentum after roughly 45 years, followed by small divergence. In both cases, convergence stalls at about 70–80% of U.S. TFP. There appears to be a remaining 20–30% gap that most economies do not manage to close. Now look at the left panel. China seems to be following a similar path. China’s income per capita reached Japan’s 1950 level around 1995 (in 1979, it was extremely low). Although China’s relative TFP growth has been somewhat faster than that of the representative East Asian economy, the slowdown is clear. By the end of the sample, China’s TFP growth is only about 0.5% higher than that of the U.S., and its absolute TFP level remains well below that of other East Asian economies. This reading is consistent with other data coming out of China and, more anecdotally, with my own impressions from a long trip there this summer. Combined with China’s demographics, this suggests grounds for being less than sanguine about the state of the Chinese economy 10 to 20 years from now. Final disclaimer: I am not a big fan of the PWT. I use it here to speak to the profession at large, which would find it odd if I relied on other datasets. P.d. If you see me at ASSA this weekend, please say hi! I will be giving a talk on Saturday on deep learning and networks, and on Sunday on demographics. But we warned, as every time the ASSA meetings are in Philly, it is quite cold🥲

@JesusFerna7026 We conversed about PPP conversions in PWT some years ago)which was not satisfactory for anyone concerned). I still argue that one can use PWT series that match NA growth rates closely (as you seem to do here) or which (like Eurostat) are helpful for subsequent cross sections.