Streetwise Professor

August 20, 2018

Goodhart’s Law on Steroids, PCP, and Crack: Chinese GDP

Filed under: China,Commodities,Economics,Politics — cpirrong @ 6:46 pm

Goodhart’s Law states that if a measure becomes a target, it ceases being an informative measure.  If you want to see an illustration of Gooodhart’s Law in action on a humungous scale, just look at China.

Michael Pettis has a piece in Bloomberg which, in brief, says that China has a GDP target.   If it appears that the country will fall short of the target, local governments get the high sign to invest in infrastructure, construction, and the like.  Local governments control credit creation (by guaranteeing bank debts) so banks are willing to lend to finance this investment: further, frequently the government will jawbone banks, or will twiddle the knobs in the banking system (e.g., lowering reserve requirements) to get banks to supply the necessary funds.

The investments are guaranteed (though what revenue stream or assets back the guarantees Pettis doesn’t say, and there are reasons to doubt the value of these guarantees in a crunch).  Hence, banks never have to write down the debt even if the investments turn out to be junk, with a value far less than the cost incurred to create the underlying assets.

So basically, the Chinese government can produce any GDP number it wants.  Voila, apropos Goodhart, the GDP number is useless.

You’d like GDP to measure the value of goods and services (including investment goods) created.  Instead, in China on the fixed asset side in particular, it measures cost, which may bear little relationship to value when economic decisions are made according to the process that Pettis describes.  In market economies where banks and borrowers have hard budget constraints, investments that don’t pan out are written down, and the losses are deducted from income.  That doesn’t happen in China.

So what is national income in China?  I’d start with consumption, though even there due to issues with price indices/inflation measurement that may be overstated.  Then I’d add a constant X times reported fixed investment, where X<1.  Probably a lot less than 1, to take into account the fact that much investment has a cost that exceeds value.  Further, I’d deduct some fraction of accumulated past investment to reflect writedowns that should be made, but aren’t.

The focus of this analysis should be on determining X.  X should be a function of something related to estimated shortfall of GDP from target absent stimulus: the bigger the shortfall, the smaller X (because more bad investment is likely when the shortfall is big, as it’s then that the government encourages investment to make up the shortfall).  It could be a function of the increase in fixed asset investment, or construction investment, with a smaller X when investment in those categories shoots up.

A few other remarks.

First, it is stories like Pettis’ that convince me that modern China represents the most colossal misallocation of capital in history.

Second, it also makes me skeptical about Scott Sumner’s use of state-owned-enterprise (SOE) share of employment as a measure of centralized control of the economy. Most of the capital, and related employment, that results from the GDP targeting channel that Pettis analyzes flows through private firms.  The government controls/affects resource allocation via incentives given to local governments, which in turn incentivize banks and private firms to achieve the government objective.

Spitballing here, but I think a better measure would be something along the lines of the ratio of the volatility in fixed investment to the volatility in GDP.  Or maybe the ratio of the volatility in credit creation to the volatility of GDP.  Chinese GDP volatility, especially post-crisis, is laughably low.  The channel that Pettis identifies stabilizes GDP (reducing its volatility) by changing investment/credit creation in response to changing economic conditions (thereby increasing its volatility).  The only problem with this measure is that there is a real risk it will become infinite.

In (relatively) market-oriented economies, investment is the most volatile component of GDP, so the ratio I propose would be positive in those economies.  But that could serve as a market economy benchmark against which to compare China.  I’m guessing that China’s ratio would be substantially larger.

Third, when looking at the demand for commodities, the potential for shortfalls of economic performance from government target should be decisive.  These shortfalls induce the turning of the credit spigot which juices the demand for commodities.

In sum, what matters in China is not whether or not GDP hits the target–it will! The question is what the government has to do to hit it.

Print Friendly, PDF & Email


  1. Very informative, Professor!
    You have peeled the scales from my economically-illiterate eyes with this post.
    I do want to check if you are using the Latin “pace” correctly, in the phrase
    “Voila, pace Goodhart the GDP number is useless.”
    I believe the “pace” usually means “contrary to” which I believe is the opposite of what
    I think you intended.

    Respectfully Mudakych

    Comment by Mudak Mudakych — August 20, 2018 @ 7:53 pm

  2. @Mudak–Thanks, for the compliment, and the correction. I’ve changed accordingly.

    Comment by cpirrong — August 21, 2018 @ 8:01 am

  3. Hi professor, great analysis as always.

    I’d be willing to ask you about the latest regional problem in the Americas, Venezuela. What do you think about the situation at large? How long do you think it will for the country to recover after Maduro is gone?

    Thanks as always.

    Comment by WZ — August 21, 2018 @ 1:36 pm

  4. So, Chinese GDP is made from pot metal too.

    Comment by dearieme — August 22, 2018 @ 3:51 am

  5. So, didn’t the Japanese government jawboning and twiddling the banks knobs cause the bubble economy that collapsed in the 1990’s? Are the Chinese doing anything differently or are they willing to trade a stagnant economy for a few decades of rapid growth? You could argue that Japan is ok, except they work themselves to death.

    Comment by Daniel Rust — August 22, 2018 @ 2:48 pm

  6. Why i hated national income accounting vol. XX. After nearly 40 years it still gives me nightmares.

    Comment by Sotos — August 26, 2018 @ 1:31 pm

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress