2006-06-13

Indicators for exchange rate: is RMB under-valued?

Is RMB undervalued? If you are a poor student frequenting the McDonald's, yes. But if you are a bourgeois who prefer to pay more for the couch in Starbuck, then no. The Economist commented:
  • "Many readers will find burgernomics and lattenomics hard to swallow. Both are flawed as measures of PPP, because they are distorted by differences in the cost of non-tradables such as rents. Yet they are surely a more fun way to understand exchange rates than textbooks...
  • Burgers and coffee are therefore likelier to give some clues about currencies."
Well, they give more clue than the abstract PPP which doesn't account enough for 'quality' of goods or the nominal rates which missed the non-tradables. The real number should be somewhere in between BigMac and Tall Latte.

---

A more thorough treatise is by Chinn/et al. As we know, currencies of rich countries tend to be "overvalued" by the market vs the PP value. Chinn's work attempted to see whether RMB is above or under that trend line. Results: it is slightly under (not statistically significant for an action), but it already was during the 1997-98 Asian financial crisis, when everybody said RMB was over-valued.
  • "As can be seen,the Renminbi (Rmb) appears to be substantially undervalued, even after taking into account the fact that absolute PPP doesn't hold. The regression line is for a pooled sample encompassing over a 174 countries over the 1975-2003 period, with plus/minus 1, and plus/minus 2 standard error bands (adjusted for serial correlation). The point estimate indicates that each one percent increase in relative (to US) per capita income induces a 0.25 percent real appreciation (these real exchange rates are the Penn World Tables "price level").

  • One key result is the confirmation that according to this criterion, the Rmb is indeed undervalued, and undervalued by a very large amount (in log terms, over 70%; in level terms about 50%). A second, equally important result, is that the 2003 estimate is within one standard error of the conditional mean. Hence, the corresponding p-value for the hypothesis that the Rmb is undervalued is larger than conventional significance levels. In other words, by conventional levels of statistical significance, the Rmb is not undervalued.

  • Speaking for myself, I do not take this as proof positive that the Rmb is not undervalued. After all, failing to reject the null hypothesis is not the same as accepting the null hypothesis. Rather, I would say it tells us something about the difficulties of defining exchange rate misalignment. One notices that the degree of measured undervaluation was not that much smaller in 1997-98, when the Rmb was widely acknowledged as overvalued, or not misaligned."



Categories:

No comments: