Michael Pettis Should Read More, Blog Less

Skip the maths, will ya?

I have in the past tied Michael Pettis to the whipping post for uncritically accepting Bernanke's self-serving notion of a "global savings glut" amidst falling levels of savings worldwide and a fantasyland assertion that the US household savings rate would hit double digits. It didn't, and I still don't quite understand why someone who keeps making factually inaccurate statements and predictions remains popular. If you read blogs to keep ahead of the curve of mainstream media or to understand the antecedents of various economic events, well, I would not be so sure about that if you are a regular China Financial Markets reader.

I do not like to keep doing this, but he has done it again (and again and again). In a recent blog, he takes Arvind Subramanian and Martin Kessler to task for their argument that there is a "yuan bloc" emerging in East Asia. The PIIE authors base this argument on movements of several regional currencies covarying more with the Chinese yuan and less with the US dollar. However, Pettis argues that this argument is artifactual:

Well actually you can argue with the math, or at least you can argue with the interpretation of the math. There are alternative – and much simpler, I think – explanations for the increased “co-movement”, and these do a much better job, I think, of explaining what is happening than reserve currency displacement.

Assume for a moment a global scenario in which the largest exporter of manufactured goods in the world has a significantly undervalued currency. Assume further that many of its competitors also have undervalued currencies, and would like to revalue in order better to manage their domestic monetary policies. Assume finally that the world is in crisis, and exporting nations are having trouble maintaining the necessary growth rate of their exports, so they cannot allow their currencies to rise faster than that of their main export competitors.

In this scenario which currency would the currencies of the smaller exporting countries track, the US dollar, or the undervalued currency of the largest and most competitive exporter of manufactured goods in the world? Almost certainly the latter, right? The smaller exporters would want their currencies to rise, but the rise in their currencies would be limited by the rise in the currency of their largest competitor. This would happen not because they are tracking a new reserve currency but only because they are in export competition with that currency.

That is all well and good if Subramanian and Kessler didn't bother to consider this scenario as Pettis clearly suggests, but they did. On page 11, the PIIE boys write:

i. Is the RMB bloc related to trade integration with or competition against China?
A currency could co-move with the RMB because it is integrated with China in terms of common supply chains. A related but distinctly different reason for co-movement could be if policy targets the RMB because countries do not want to lose competitive advantage vis-à-vis Chinese exporters and domestic manufacturers. In other words, the reason for the co-movement could be competition against rather than integration with China.

How can we distinguish the two? One way of measuring competition is to see if a country exports products similar to China’s. Mattoo, Mishra and Subramanian (2012) develop such an index of competition relative to China. Unfortunately, they compute this index for fewer emerging market countries than contained in our sample.

When we introduce this index of competition (which is country-specific), it has consistently the right sign (the more a country competes with China, the more likely its currency to track the RMB). But is not consistently significant in a statistical sense (in Table 6, the index is statistically significant in column 2 but not in column 1). And when we run a horse race between this competition variable and a pure integration variable, the latter consistently trumps the former. So, the evidence, albeit limited, favors an explanation for co-movement that is more related to trade integration than competition, although a role for the latter cannot be ruled out. One reason for that last caveat relates to the findings reported in Table A6. It seems that outside East Asia, more countries track the RMB when it depreciates than when it appreciates. Moreover, the average magnitude of the CMCs outside East Asia more than doubles in such instances. So, we cannot rule out entirely a competitive pressure motivation for currencies to track the RMB.

PIIE boys did their homework, period. Some points:

1. I don't know why Big Name Authors think they can get away with such sloppy writing;
2. My policy of not having a comments section is vindicated by the echo chamber over at Pettis' blog. 59 comments...and not one who points this out either. I guess the Pettis Fan Club takes his word for gospel truth for better or worse--rather worse here;
3. I'm afraid that this is another example of inept blogging. Pettis writes--and he sure does write with posts extending to thousands and thousands of words--but he clearly doesn't do so from a position of knowledge of the material he himself links to;
4. This demonstrates why some quantitative analysis skills are worth learning for those interested in the subject matter. How do you measure covariance? How can alternative variables be included in models to account for alternative hypotheses? Pettis is a poet with limited numbers chops, so this probably explains his lack of awareness about such things. Do the math.

Note to Mike: If you link to something, at least try to understand what you're criticizing. I am not convinced that the yuan is going to take over as the world's dominant currency soon either, but I think it's at least worth trying to understand the arguments of Subramanian and Kessler before criticizing them for faults they didn't commit.