…through the Lens of Multiple Regression
I hesitate to add to the welter of commentary on President Hu Jintao’s state visit (e.g., ), but Professor Mark Wu’s recent NYT Op-Ed China’s Currency Is Not Our Problem inspired me to think again about how econometrics can inform policy.
Bivariate vs. Multivariate
In the article, Professor Wu states:
Many Americans believe that the Chinese jobs being preserved by an artificially low currency come at the expense of American jobs. There are three common explanations behind this theory.
First, a stronger currency would increase the purchasing power of Chinese consumers and decrease the relative cost of American goods in China, spurring more Chinese to buy more American products. Second, a stronger currency increases the relative cost of Chinese goods in third markets, like Europe or Latin America. So if the renminbi appreciates, consumers in other countries will shy away from Chinese products in favor of American products. Third, a stronger currency would increase labor costs in China, making it less attractive for American companies to move jobs to China and thus keeping more people employed at home.
These claims, however, are more wishful thinking than actual truths. Consider the first idea, that a strengthened Chinese currency would increase the growth rate of American exports to China. From 2005 to 2008, the renminbi appreciated nearly 20 percent against the dollar. Yet, American exports to China over those three years grew at a slightly slower pace than in the previous three-year period when the renminbi did not appreciate at all (71 percent versus 89 percent).
The description of the data is hard to deny. And, in addition, the Chinese global trade balance also rose as the renminbi appreciated on a trade weighted basis. Figure 1 illustrates the time series.
Figure 1: Log CPI deflated trade weighted value of Chinese yuan (blue line, left axis), Chinese trade balance, monthly in billions of USD, customs basis (red line, right axis), and 12 month trailing moving average (purple line, right axis). NBER defined recession dates shaded gray. Source: BIS for exchange rate, IMF, International Financial Statistics for trade data, updated with news reports, NBER, and author’s calculations.
The correlation between CNY value and the trade balance is positive, thereby seemingly buttressing Professor Wu’s skepticism. But this is why multiple regression was developed; that is, we typically think of macroeconomic relationships as involving more than just two variables.
Briefly put, Chinese exports could at a minimum be thought of as depending on rest-of-world economic activity, productive capacity in China, and the relative price of Chinese exports, a function itself of the exchange rate and the price level. Imports could be thought of as depending on Chinese economic activity, the relative price of Chinese imports, a function itself of the exchange rate and the price level. Hence, the Chinese trade balance at a minimum should depend on Chinese GDP, rest-of-world GDP, Chinese manufacturing capacity, and the real exchange rate. If one thought that there were other determinants, such as trade barriers, that would make the list of relevant determinants even bigger. Making this realization means at a minimum, bivariate correlations are highly suspect. (Another variable of interest would be productivity, which we know has grown rapidly in China; this suggests the proper relative price would be the unit labor cost deflated real exchange rate [primer], as suggested on page 12 of the last IMF Article IV on China).
Professor Wu continues, by arguing that even if there is an effect, it is likely to be minimal, since US exporters and Chinese exporters don’t compete head to head in third markets, and further US exports are capital goods (aerospace and power generation are the examples cited). These arguments make sense, but for me, I’d want to refer to what the statistical evidence indicates about quantitative magnitudes of responses, rather than a priori judgments.
New Econometric Results
Three years ago, I would’ve agreed substantively with Professor Wu’s assessment that the degree of substitutability between Chinese and non-Chinese goods was low, and hence price elasticities low. But since then, variation in trade flows have provided evidence of greater sensitivity to relative prices. For instance, Ahmed (2009) obtains a quasi-long run export price elasticity of 1.8, estimated over a sample ending in 2009 (Ahmed’s figure is for growth rates on growth rates, since he uses a first differences specification). Further, as I noted in a post from September 2010, I also obtain a pretty high elasticity of exports with respect of the real exchange rate.
To obtain this conclusion, I used an error correction specification, where exp is log real total Chinese goods exports, and y* is rest-of-world log GDP, and q is the log trade weighted exchange rate for the China (CPI deflated, from the IMF).
Δ exp t = β 0 + φ exp t-1 + β 1 y t-1* + β 2 q t-1 + β 3 z t-1 + γ 0 Δ exp t-1 + γ 2 Δ y *t-1 + γ 3 Δ y USt-2 + γ 4 Δ q t-1 + γ 5 Δ q t-2 + u t
Estimating this over specification over the 2000Q1-2009Q4 period yields an adjusted-R2 of 0.94, SER=0.032, and passes a LM test for serial correlation of order 2, at the 10% msl. The implied long run price elasticity is 0.75 (and is statistically significant). The rate of reversion is 0.46 per quarter, implying a half-life of a deviation is a little over 1 quarter.
This 0.75 long run elasticity is quite a bit higher than the Cheung et al. (2010) estimate of between 0.34 to 0.64, estimated over the 1993-2006 period. (For caveats about the empirics, see the original post).
Where I agree with Professor Wu is that it is unclear that rest-of-world exports to China (global Chinese imports) would increase with an appreciation of the Chinese yuan. That’s because, like Marquez and Schindler and Cheung et al. (2010), I was unable to obtain a reasonable estimate of a price elasticity for Chinese imports using an analogous ECM specification incorporating Chinese GDP and the real exchange rate. This result might be in part due to inappropriate aggregation of ordinary and processing imports, but I think that is only part of the story, since in Cheung et al. (2010), we are unable to obtain sensible price elasticities even after disaggregation. One hypothesis is the period includes substantial import substitution.
One could assume that an appreciated currency would induce greater imports, as Cline (2010) does. In the absence of reasonable estimates, I prefer to remain agnostic. (By the way, this explains why Cline obtains a substantially larger $170 billion impact on the trade balance for the 10% appreciation.)
However, one can still make some conclusions regarding exports. The results indicate that, holding all else constant, a 10% appreciation of the trade weighted real value of the yuan will induce a 7.5% reduction in Chinese exports. To get a feeling for the quantities involved, Chinese nominal exports in the 2009Q3-10Q4 period were $1.385 trillion. A 7.5% reduction of this amount is $104 billion; but a change in the value of the yuan should induce a partially offsetting $32 billion increase in total nominal value of exports (assuming 25% pass through, and 10% appreciation). The net dollar impact would then be only about $72 billion. This is not an unsubstantial impact, given 2009Q3-10Q2 trade balance was $155.6 billion (or the $200 billion forecast by China Economic Quarterly, or $212 billion in the IMF’s Article IV review of China).
Yin-Wong Cheung and I are now starting work to undertake an updated, and more comprehensive, look at Chinese elasticities.
China, and China Plus East Asia
The assertion the America’s trade balance would be largely unaffected by Chinese appreciation is in part predicated by the view that only Chinese exports would be affected. But as I stressed in this October IMF panel on China, trend Chinese yuan appreciation would likely induce trend appreciation of other East Asian currencies. Figure 2 (from this post) highlights the 2005-07 experience.
Figure 2: Chinese yuan (black), and other East Asian currencies against the USD, all normalized to 2005M06=1. “Down” is defined as appreciation. Source: IMF, International Financial Statistics, St. Louis Fed FRED II, and author’s calculations.
Professor Wu concludes:
…it is unlikely that a stronger renminbi would bring many jobs back home. Instead, companies would most likely shift labor-intensive production to Vietnam, Indonesia and other low-wage countries. And in any case many high-skilled jobs will continue to flow overseas, as long as cheaper talent can be found in India and elsewhere. Only in a few industries, like biomedical devices, would a stronger Chinese currency combined with quality issues tempt American companies to keep more manufacturing at home.
I agree that if it were only China shifting its exchange rate, Professor Wu’s assessment would be make sense. However, it has never been “just China”. From Thorbecke and Smith (2010) ([wp version]):
China’s global current account surplus equaled 9% of Chinese GDP in 2006 and 11% of GDP in 2007. Many argue that a renminbi appreciation would help to rebalance China’s trade. Using a panel dataset including China’s exports to 33 countries we find that a 10% renminbi (RMB) appreciation would reduce ordinary exports by 12% and processed exports by less than 4%. A 10% appreciation of all other East Asian currencies would reduce processed exports by 6%.A 10% appreciation throughout the region would reduce processed exports by 10%. Since ordinary exports tend to be simple, labor-intensive goods while processed exports are sophisticated, capital-intensive goods, a generalized appreciation in East Asia would generate more expenditure-switching towards US and European goods and contribute more to resolving global imbalances than an appreciation of the RMB or of other Asian currencies alone.
See also this post.
Will renminbi appreciation solve the problems of the US. Here I agree with Professor Wu; on sustaining long term growth and reducing the current account deficit over the long term, increasing investment in human and physical capital and reining in the long term structural budget deficit are important. But to get to the long term, we need to survive the short term, and here expenditure switching via exchange rate changes is one important tool that can be effected much more quickly than in restructuring the US educational system.
Statistical Evidence on Foreign Official Flows and Current Account Balances
Finally, Professor Wu’s implicit argument that China’s current account surplus would be largely unaffected by reduced foreign exchange intervention flies in the face of statistical correlations. Using a cross-country panel regression methodology similar to that implemented by Chinn and Ito  , Joseph Gagnon writes:
External financial policies are the most important drivers of current account imbalances in
developing economies, but they are not significant drivers in advanced economies. These policies include,
for example, preventing exchange rate appreciation after positive shocks to net exports by sterilized
intervention in foreign exchange markets. Such policies enable current accounts to remain positive
indefinitely because they shut off the normal adjustment channel of real exchange rate appreciation. These
policies are more common in developing economies than in advanced economies, in part because greater
capital mobility in advanced economies makes sterilized intervention less effective. The current account
tends to increase by 40 to 50 percent of any increase in net official financial outflows (including reserve
accumulation), and the effect may be even larger in some developing economies. This result supports calls
for increased exchange rate flexibility (i.e., reduced accumulation of foreign exchange reserves and other
official assets) in developing economies with current account surpluses.
This suggests to me that US, the world — as well as China   — would benefit from more rapid appreciation of the renminbi, and the associated decrease in Chinese foreign exchange reserve accumulation.