Lies, Damned Lies And Chinese Statistics

Posted January 9, 2009

Believe it or not, China has not yet set an official growth target for 2009. At its recent Economic Work Meeting, the government did declare an intention to maintain "relatively fast growth rate." Everyone seems to assume that that target will be 8%, the number set in years past, and as far as economic myth goes, the number below which chaos descends upon China. Many believe that whatever happens, the official numbers will tell a story in which they more or less meet the mythical 8%.

For instance, in a recent chat with a local People’s Bank of China official, I asked what the local growth rate was for first half 2008. "The city reported 10%," he replied before shaking his head, "but there’s no way it was really that. But they would not dare report anything less."

Which raises the question: How will we know how fast China is really growing in 2009, if even government officials have lost confidence in the GDP statistics? We have been here before, of course. Back in the late 1990s, some economists wondered if Beijing was "adding water" (as the Chinese euphemism has it) to its headline GDP numbers. But then some serious economists thought that China actually got through the Asia Financial Crisis with the relatively fast growth they posted. The debate is super relevant for understanding what may happen over the next two years—but the devil is in the details. So it’s worth delving into that messy detail, adjudicating where we now stand, and outlining how to track activity over 2009-10.

Thomas Rawski at the University of Pittsburgh argues that China’s official 1998-99 GDP data is "totally divorced from reality." He believes that maximum growth in each of these years was 2% and that cumulative growth over the period 1998 to 2001 was somewhere between 0.4% to 11.4%, compared to the official (pre-2004 Economic Census revision) number of 34.5%. To support his view, Mr. Rawski cites reports of stagnating rural incomes, mass layoffs, inventory accumulation, lower steel production, etc. He shows that household-expenditure growth is much weaker than official retail-sales data. He argues local governments came under pressure to report numbers that hit the center’s target.

Mr. Rawski also uses a number of alternative measures of economic activity, what we will call "growth proxies," to show the official figures smell. First, there was energy production, which grew at a moribund rate of only 2.5% on average in 1998, compared to 4.5% year on year in 1997. Air travel grew only 3.4% in 1998. Nonprocessing imports fell 5% year on year in 1997 and had basically zero growth in 1998, before picking up rapidly in 1999 to 31% year on year.

Harry Wu at Hong Kong Polytechnic University has recently added to these doubts. He has spent considerable time digging around in the 2004 National Economic Census data, the numbers which caused the National Bureau of Statistics to revise up real GDP growth over the 1994 to 2004 period by some 0.5 percentage points a year on average, and as a result to revise up the size of the economy in 2004 by some 17% to 16.1 trillion yuan ($2 trillion) from 13.7 trillion yuan. Mr. Wu, however, shows that while the NBS did not collect data on prices, its method for revising the real GDP numbers should have been reliant upon price data. Mr. Wu believes that the NBS derived a GDP trend it thought looked reasonable and then revised the deflators (overall changes in prices) to match.

The NBS also had the opportunity to revise down the 1998 number after the census, but did not do so. In fact, it did not touch 1998 growth (7.8%), while it did revise up all the other years’ numbers (including 1999). Mr. Wu supposes that it might have been an admission of error (or worse) to revise down the 1998 number given all the attention it had drawn. But the very mild slowdown in 1998 still looks strange. Over the period 1978 to 2002, Mr. Wu estimates an average growth rate of 16.2% year on year, compared to the official average of 20.4% year on year. For 1998, he reckons the industrial economy contracted by close to 7%, after basically not growing at all over 1996-97.

In Defense of the NBS

A number of economists have taken issue with the skepticism though. They include Nicholas Lardy at the Peterson Institute in Washington, D.C., who believes that while the official numbers have some problems, the problem is not that bad. Mr. Lardy first looked at tax-revenue growth. These grew 22.5% year-on-year on average in 1997, 10.7% in 1998, and 19.4% in 1999. Given that taxes are paid on the back of value-added activities, they should follow growth, and these numbers suggest that there was a quick V-shaped recovery. Mr. Lardy also looked at imports. They began to pick up in 1999, which leads him to reject the idea that the whole of the period 1998 to 2001 was really slow.

Mr. Rawski responds that official tax revenues may not be a reliable indicator, particularly in the bad times. This is because local government officials are strictly judged on their ability to raise taxes. They thus have an incentive to "fake" taxes, moving money in and out of the accounts to exaggerate revenues.

Moreover, over the past few years, local governments and central ministries have been transferring monies which had previously gone into their informal, unseen accounts into the main Ministry of Finance account. This process may well have accelerated in hard times in order to ensure that officials met their official revenue quotas. As far as imports go, Mr. Rawski thinks antismuggling efforts during the late 1990s may have forced more trade to go through official ports, making this proxy is unreliable also.

Carsten Holz at Hong Kong University of Science and Technology then joins the fray. In a 2003 paper, he makes a strong defense of the data. He shows that the NBS has improved its collection methods by getting large firms to input data directly via the Web and doing its own surveys—both ways of avoiding local officials’ numbers. More precisely, he shows that many of the proxies have their own problems.

Take energy, for instance. Mr. Holz shows that the official survey’s coverage is terrible. Others have shown that closures of small coal mines, expansion of small power plants that fall outside the official statistics, and increased use of power generators by firms themselves all weaken the data as a growth proxy.

Freight numbers have their issues too, Mr. Holz argues. On the plus side, the Transport Ministry and NBS work to collect data directly from railways, ports, and large transport firms, and they also do surveys. This means much private-sector activity should be captured. On the downside, though, in contrast to rail, much river and road traffic is likely going unreported, as it is handled by small, private operators.

Indeed, a study of petrol consumption by haulage firms found that it was growing much faster than official road freight numbers, implying that the latter numbers were underestimated. All of that said, though, when Mr. Holz crunched the numbers, he found that freight turnover was indeed correlated with industrial output in the 1990s (though not in the 1980s). This is bad news for the conspiracists, as it suggests that overall activity did not decline as much as they claim.

And there the story might have ended. However, Mr. Holz then had a long look at the 2004 Economic Census revisions. He found numerous inconsistencies and evidence suggestive of manipulation. He concludes: "… One may begin to wonder about the possibility and likelihood of professional statistical work in China. For the time being, the 2006 benchmark revision implies that Chinese statistics have to be taken with a rock of salt."

How does one conclude, then? The key problem is the 1998 GDP number. Whatever their problems, we cannot ignore the fact that all of the growth proxies indicate a sharper slowdown than the official data allows. However, the 2% growth estimate appears exaggerated when the fiscal and import numbers are taken into account. An estimate of 5% growth for 1998 seems reasonable. The evidence, especially the census work, also suggests that headline GDP numbers can still be problematic—particularly at times when the central government is assuring everyone that a certain level will be reached.

A Taxing Task

So, what do those proxies tell us about today? They show an economy which has slowed considerably in the last few months, but they are not all suggesting a severe slowdown.

Let’s begin with a look at fiscal revenues. These have crashed—falling 9% year on year in November, after falling 0.3% year on year in October 2008. Has the economy simply stopped growing then? We are reticent to draw such a conclusion, for three reasons. First, 2007 was a year of huge land-sale revenues, for instance. They reportedly rose to 50% of revenues in some areas. In theory, these revenues should have been included in local government budgets in 2007, but much likely are still not included. To the extent that they are, this will have pushed up fiscal receipts in 2007, and will be dragging them down now. Second, companies may be delaying tax payments for cash flow reasons—and local tax bureaus may be going easy on them.

Third, funny things can go on at local government finance bureaus. Each year, officials at each level of government are given a revenue quota to meet. In theory, once it is met, their job is done. In practice, officials normally make sure they exceed their target by a healthy margin. That is the game. This year, however, the incentives in the game have changed. With the economy slowing, many local officials may have met their revenue quotas in the third quarter. But now they will want to save as much revenue as they can for next year, when times will be tougher.

Moreover, they will want to show Beijing that they really need that large fiscal transfer to finance their new bridge/metro system/high-tech development zone/enterprise bailout fund. This means they may be tempted to report low numbers now and "back-pocket" the rest, parking the funds in nonofficial accounts, all to be transferred into next year’s budget.

I suspect this is happening now, and it may be artificially depressing the revenue growth rate. That said, there are also reports of tax officials (who have not met their quota) demanding firms to pay up right now—and even urgently demanding payment of back taxes. But, overall, I believe that this proxy is probably exaggerating the scale of the current slowdown.

Now turning to electricity production. After so many years of plus 10% year-on-year growth, the sudden drop since the summer is even more, ahem, shocking than the decline in fiscal revenues. What do we make of it? Here again, we need to exercise some caution. Two-thirds of electricity is used by industry, so these numbers only tell us about one part of the economy. And within industry, heavy industry is the key, since most electricity is used by makers of aluminum, steel, copper, etc. It is these sectors that have borne the brunt of the construction slowdown and of the sudden correction in commodity prices in third quarter, which has created large inventories and crushed production incentives. As a result, the electricity numbers are also likely exaggerating the severity of the current slowdown.

There has also been a clear slowdown in moving stuff around. Rivers are a cheap and reliable form of transport and have been particularly useful for coal and steel products. With demand for electricity down and stockpiles of coal and iron ore up, perhaps that explains some of the waterway decline.

Looking at rail traffic, the recent direction is down, but the trend is unclear. Supply constraints matter. Look at first half 2004, when we know there was an explosion of economic activity. Both waterway and road use expanded rapidly, but the railways did not. That was because of a supply constraint—there were not enough rail carriages or track. There still aren’t. As a result, as the economy slows, since rail costs are lower than road and waterway costs and more capacity is currently being added, rail-freight growth will likely remain healthier for longer.

And then we have road freight, which is still pretty strong. One potential problem here is that the numbers may not be capturing the slowdown in activity by private firms with one or two trucks, or manufacturers using their own trucks (Mr. Holz shows that only dedicated freight firms are included in the survey). But the resilience is still interesting.

When it comes to passenger air traffic, airlines are having a terrible time. Passenger kilometer growth is already well into negative territory and revenue per passenger is falling as well as the airlines cut prices. Local media has made much of 10 yuan ($1.30) tickets. Air travel has been an early casualty of business cut-backs. The Ministry of Finance’s postearthquake call to cut down on government travel has probably had an effect too.

Folk are still using the railways to travel, though. But if this is simply to return home to Yunnan or Anhui province from a closed factory in Dongguan then that is obviously not a positive for growth. According to the Ministry of Labour and Social Security’s surveys, some 10 million migrant workers have already returned home, that’s around 7% of the migrant workforce.

Finally, we end with an indicator that should have some utility as a leading indicator—bank-credit growth. We show nominal and real bank loan growth in the chart on page 17. A bank loan is only as important as its purchasing power and inflation has played havoc with corporates’ buying power in recent months. So real growth is key. We have data to November—real loan growth has now picked up. How sustainable is this recovery? The credit quota for 2009 is 4.7 trillion yuan ($690 billion), which is aggressive. Given the private sector’s lack of appetite, all depends on the fiscal stimulus package.

Eggs in One Basket

We put several of the proxies we like into a single index. We used industrial goods production, freight, bank credit and major imports, shown on this page.

The past few months have seen a further deterioration on a weakening trend that began in late 2007. This is a much more serious downturn than in 1998. There is very weak growth momentum going into 2009.

Predicting the bottom of the current cycle is not going to be easy. We think we will see continued deterioration into January/February 2009 driven by the inability of the fiscal package to get any purchase before the second quarter, banks’ unwillingness to lend to the private sector, continued property price declines, high levels of risk aversion in the private sector before fiscal spending kicks in, more bad news from the United States, Europe, and Asia generally, and a cash crunch that could hit in January before Chinese New Year.

All eyes will then be on the various proxies to indicate when the economy ends its slide. I expect the economy to stabilize in second quarter 2009 as bank loans loosen corporate-liquidity problems, extra fiscal spending kicks in to create private investment demand, looser monetary policy supports private-sector investment, and housing transactions increase after a further price correction. If this happens, China would be set for a restrained recovery in the second half of 2009. But at the core, private investment in the economy will be weak, in everything from real estate to heavy and light industry, and that is why GDP growth is likely to come in at just 6.8% for 2009 before recovering to 8% for 2010.