China's Inflation Rates: Signs of Market Imperfections

Ryan Avent points out that if one uses a different measure of inflation in China to convert its nominal exchange rate into a real exchange rate, the Chinese yuan (CNY) has actually appreciated quite a bit in real terms in recent years. Using China's GDP deflator instead of the CPI, in fact, the CNY has appreciated in real terms by close to 50% since 2005. Using the CPI the other day I had calculated the real appreciation to be much smaller.

The larger figure obtained using the GPD deflator provides a reassuring confirmation of our priors, but it raises a very interesting question: why are the two measures of China's inflation rate so different? As seen in the chart to the right, the GDP deflator has been consistently rising much faster than the CPI in China. Why?

While there are many differences in how the two measures of inflation are calculated, the biggest and most relevant distinction in this case is that the CPI only measures the prices of things that consumers buy, while the GDP deflator also measures the prices of things that non-consumers -- i.e. businesses, the government, and the foreign sector -- buy.

Let's make the (reasonable, I hope) assumption that consumers buy more services than goods, while non-consumers buy far more goods than services. Combining this assumption with the distinction noted above tells us that the prices of tradable goods have been rising much faster than the prices of locally-provided services in China. And this has three important implications.

1. Manufacturing wage growth in China outpaces productivity growth.
For simplicity, let's think of China's economy as producing two types of goods: tradables (i.e. manufactured products) and non-tradables (i.e. services). We typically think of wages as being equal to the marginal revenue provided by labor, that is:

(1) w^t = p^t * MPL^t

where 'w' stands for wages, 'p' is the price level, 'MPL' stands for the marginal product of labor, and the 't' superscripts indicate that variables pertain to the 'tradable' sector of the economy. If w and MPL in the tradable sector both rise at the same rate, then p^t should remain constant. The fact that p^t has been rising indicates that wages are rising faster than productivity in this sector of the economy.

This in turn raises its own interesting question: why are wages rising faster than productivity in China's manufacturing sector? This apparently violates classical labor market assumptions. While I don't pretend to be an expert in China's labor market, I suspect that this is a form of catch-up. Prior to China's great economic growth of the past 15-20 years, wages in the manufacturing sector were far below what worker productivity would normally have warranted in the absence of market imperfections. (My guess is that this was due to the communist, command-economy system that dominated in China prior to the 1990s, which kept wages artificially depressed.) And what we've been seeing over the past 10-15 years is the process of that discrepancy being corrected.

This ties back to an argument I have previously made that the incredible movement of manufacturing to China that has happened over the past two decades was a once-in-a-lifetime event, in which multinational firms could essentially take advantage of an arbitrage opportunity in China that existed because labor in China was cheaper than its marginal product. That difference between wages and productivity in China is now rapidly being arbitraged away.

2. There is poor intersectoral labor mobility in China.
The fact that the prices of manufactured goods in China are rising faster than the prices of services also tells us that there must be poor labor mobility between China's manufacturing sector and its services sector. How can we infer that? Take a look at an equation similar to (1) above, but for the non-tradables sector of the economy:

(2) wⁿ = pⁿ * MPLⁿ

We know that pⁿ is rising more slowly than p^t. There's a mountain of empirical evidence that tells us that MPLⁿ almost always rises more slowly than MPL^t, i.e. that labor productivity grows faster in manufacturing than in services. Putting those together, it must be the case that wⁿ is growing more slowly -- much more slowly, in fact -- than w^t.

But if workers could move between the tradable and non-tradable sectors of the economy, they would equalize wages between them. So we can reasonably conclude that workers can NOT move between the two sectors of the economy. This corresponds with casual observation in China (manufacturing jobs are highly prized in China, but not available to everyone), but it is a sign of an important labor market imperfection that we must keep in mind.

3. The real appreciation of the CNY may be near an end.
We've now established that wages are growing faster than productivity in the manufacturing sector in China, and that workers can not easily move between the manufacturing and service sectors of China's economy. This has an important implication for China's real exchange rate.

Developing countries typically experience trend appreciations in their currencies as they become more developed. This is explained by what's widely referred to as the Harrod-Balassa-Samuelson (HBS) effect. The HBS effect can be summed up like this: relatively low productivity gains in the services sector compared to the manufacturing sector means that the relative price of services in the developing economy rises. This raises the overall price level in the developing economy relative to where it used to be (and relative to its trading partners), causing a real appreciation of the currency. The Economist article that Ryan refers to in yesterday's post cites the HBS effect as one reason why inflation in China has been and should be higher than inflation in other countries.

But inferences #1 and #2 above directly contradict the underpinnings of the HBS effect. Instead of experiencing faster inflation in the services sector as the HBS story predicts, China is experiencing faster inflation in the manufacturing sector. And instead of labor mobility equalizing wages between the two sectors of the economy, there are clearly barriers that prevent that in China.

This means that the HBS effect does not apply to China, and can not be expected to drive a continued real appreciation of the CNY. China's currency has indeed undergone a real appreciation in recent years, though as Ryan correctly pointed out, exactly how much depends greatly on which inflation measure you use. (Please see this paper by Menzie Chinn for an excellent primer on which inflation rate to use when calculating real exchange rates. In short: there's no single right answer.)

But the real appreciation of the CNY over the past several years seems to have been driven by the catch-up of manufacturing wages with manufacturing productivity. And that means that once that catch-up process is complete -- i.e. once the difference between labor's productivity and wages has been arbitraged away -- this mechanism for real appreciation will go away. Given the various distortions and imperfections in China's labor markets that this simple analysis can illuminate, I hesitate to have faith that market forces will solve the problems of China's massive imbalances, either internal or external.