On the Quantity and Quality of Chinese GDP

Balding's World

A lingering question about the Chinese economy is the reliability of GDP data but a less studied question the quality of Chinese GDP.  Let’s begin with questions about the accuracy of Chinese GDP data.  Last week the Peterson Institute of International Economics and Nicholas Lardy wrote a blog post criticizing anyone casting a critical eye at Chinese GDP data.  Lardy make three main points.  First, why did China wait so long before devaluing the RMB? Second, China is transitioning its economic growth. Third, a lack of high frequency data.

While I think highly of Nicholas Lardy and PIIE, these are straw man arguments that do nothing to build up the accuracy of Chinese data and only weakly make the case against critics. To briefly rebut the Lardy points. First, China has been unveiling a wide range of economic and financial support measures the entire year. The RMB policy is merely one intended to support growth. Second, transitioning economies do not throw basic mathematics aside. If the data does not support the headline number, the data does not support the headline number. Third, there is large amounts of high frequency data whether it is from official or quasi official agencies through to better recording of unofficial data.  This is why people work so hard to compare a mountain of data to official Chinese GDP and are so concerned when they simply do not reconcile.

The bigger problem with the Lardy argument is that there is no evidence that he has clearly considered or even attempted to rebut the clear and convincing data compiled by a variety of people demonstrating the problems with NBSC data.  I will draw from my own research on the topic as I know it best, though there are a variety of others from consulting firms, investment banks, and academics that have raised a variety of very specific and technical concerns about data manipulation in Chinese GDP that are ignored by Lardy.

In compiling inflation data, the National Bureau of Statistics China (NBSC) has clearly manipulated the data.  According to official inflation data compiled by the NBSC, urban housing CPI rose from 2000-2011 by 6%.  Let me emphasize, that is 6% TOTAL over 12 years, not 6% annually.  Even extending that through the latest available data year brings the total to only around 10%.  That number is so patently absurd that no one believes it.  Rather than using third party and independent data to convince you how absurd that number is, which I could do, I am rather going to show you how the NBSC manipulated the data so badly that cannot even reconcile its own data.

To arrive at a low national housing CPI number from 2000-2011, the NBSC utilized an 80/20 urban/rural housing CPI weighting.  The NBSC was assuming in 2000 that China was already an 80% urban country.  It did this because according to their data, rural housing prices were rising much faster than urban housing prices.  Only by using an 80/20 urban/rural weighting can the NBSC arrive such amazing final numbers.  Now elsewhere, the NBSC is reporting that that China was only approximately 35% urban in 2000.  In other words, the NBSC inflation data did not match their own nationwide population data.

I could write at length about how obviously manipulated Chinese economic data is but I feel this ground is well covered and want to move on to other issues but I will leave this topic with a few brief points.  First, if you want to criticize people that are skeptics or critics of Chinese GDP data, then at least examine the very high quality work that has been done on the topic and be prepared to rebut it.  Second, read the numerous and varied work on Chinese statistics.  This isn’t a case of sour grapes or uneducated speculation.  Third, given the enormity and obviousness in one major line item, it is extremely unlikely that this is not happening throughout a variety of other areas of national statistics. Fourth, the size of the discrepancies here are not minor or rounding errors.  These are nothing less than clear and blatant examples of statistical manipulation.  Fifth, this specific line item matters because if plausible adjustments are made to inflation factoring in reasonable housing inflation, this would reduce total real Chinese GDP by a little more than 10% or one trillion USD.  Sixth, whoever wants to defend Chinese GDP statistics needs to begin by agreeing with 6% urban housing CPI in China in 12 years and an 80/20 urban/rural weighting beginning in 2000. After that I will gladly discuss or debate anything the author wishes including but not limited to unicorns, a Beatles reunion tour, and whether Donald Trump is too withdrawn and wonkish to be considered a viable presidential candidate.

I think a factor that received decidedly less attention is the quality of GDP.  The old simple example of what I mean by quality of GDP is this: two guys make up a country decide that they are going to drive up GDP.  One guy decides to pay the other guy to dig him a hole.  The other guy decides to pay the first gentleman to fill up the hole for him.  This process repeats itself infinitely and GDP rises rapidly.  Now very little has been accomplished but GDP goes up.  Now before proceeding, it must be emphasized that this is a very simple example.

While not fitting this exact pattern, there is significant evidence that the quality of Chinese GDP has been quite low.  There are multiple ways to consider the quality of GDP.  First, there is capacity and pricing data.  The Chinese economy remains heavily dependent on the steering of the Chinese government.  Though many companies are by lax accounting standards “private”, an issue covered in , large amounts of economic activity and investment direction remain steered by the government.  We see this quality of GDP showing up in lower pricing, surplus capacity, and failing government debt.  By official accounts, approximately 30% of the 2008-2009 debt fueled fiscal stimulus was wasted and local governments just received a 3.6 trillion debt restructuring that kept them out of default.  Especially in commodities like steel and other metals which had been targeted for expansion by Beijing, surplus capacity is enormous causing deflationary pressure on global markets.  Most airports in China are losing money and there are examples of vast sums of money being spent on airports that see few travelers.  Even in real estate, there is nearly three years of unsold inventory waiting for buyers.  While this binge may push up GDP in the short term, this is difficult to maintain over the long run and is low quality GDP.

Second, while there is no specific data on this, there is evidence that a Chinese variation on hole digging and refilling is occurring.  As an anecdote, the Shenzhen airport where I live is was recently unveiled and it is beautiful.  The problem is this.  The Shenzhen airport that was closed was only approximately 20 years old and having travelled through it frequently was a fine airport, definitely nicer than Los Angeles International. Though straining at capacity, the decision was made to shutter the old airport rather than upgrade it even though it was in fine working order.  This is nearly the airport version of redigging the hole even though the other hole just needed to be bigger.  Having lived in China for six years, one witnesses this type of replication work constantly.  While new capital is being created, there are also enormous capital losses regularly.  People have paid great attention to the new infrastructure, but there is little attention paid to whether it is being used and how quickly it is being replaced.  The capital losses associated with both are simple staggering.  Though I know of no explicit study on the issue, I would wager significant money that depreciation and capital losses are much higher as a percentage of GDP than other countries, whether measured against developer or emerging market.

The reason this matters is that most talk of economic health are both dependent on maintaining high growth rates and high quality growth rates.  I have serious doubts about both in China and I believe that all non-official data supports this position.