Economics focus

An aberrant abacus

Coming to terms with China's untrustworthy economic numbers

AS CHINA'S importance in the global economy increases, investors are paying more attention to its economic numbers. Yet the country's official statistics are notoriously ropy. Some commentators accuse China's government of overstating GDP growth for political reasons, others complain that the official inflation rate is fraudulently low. So which data can you trust?

One reason to be suspicious of GDP figures is that China is always one of the first countries to report them, usually only two weeks after the end of each quarter. Most developed economies take between four and six weeks to produce them.

Amazingly, most economists reckon that China has understated its growth in recent years. The country's National Bureau of Statistics (NBS) has recently revised China's GDP growth up by half a percentage point for both 2006 and 2007, to 11.6% and 11.9% respectively, thanks to stronger growth in services, which government statisticians find harder to count than industry. Yet even these revised numbers may be conservative.

Chinese provinces independently report GDP, and a weighted average of their figures consistently gives higher rates of output and growth than those reported by the central government (see chart). True, local officials have an incentive to inflate growth numbers because promotion depends upon economic performance; however, experience suggests that number crunchers in local government are more accurate than Beijing's. For instance, the figures first published for 2004 showed that the sum of the provincial GDPs was 19% bigger than the reported national figure. Lo and behold, in 2005, after a national economic census picked up more services, the NBS revised its GDP up by 17%; it also lifted the annual growth rate over the previous decade.

Stephen Green, an economist at Standard Chartered, calculates that in 2007 the combined output of the provinces was 10% more than that reported by Beijing. Their average growth rate of 13.1% was also still 1.2 percentage points higher than the revised national growth rate, although the gap has narrowed from almost three points in 2005. Perhaps, suggests Mr Green, central NBS folk have decided that they should trust their local counterparts more. But just as local officials have an incentive to inflate numbers, so Beijing has had reason recently to understate them: it wants to slow the red-hot economy. China's true GDP growth may therefore be higher still—which may appear to add to fears of overheating.

Distrust of GDP has led many China-watchers to track alternative monthly measures of growth. Jonathan Anderson at UBS uses one based on production (eg, industry, electricity and construction) and another based on expenditure (retail sales, fixed investment and net exports). Neither gauge shows the same sharp acceleration since 2004-05 as does GDP. One explanation is that the reported jump in GDP growth may be an attempt to correct previously understated growth figures; if so, this could ease overheating concerns.

The government also smoothes quarterly GDP growth; other less politically sensitive indicators, such as industrial production, are much more volatile. For instance, despite severe snow storms and weaker net exports, first-quarter GDP growth slowed by less than expected and by much less than did industrial production. The government may well have made the figures look stronger to avoid criticism of its tighter credit policy.

What does—and does not—add up

The right-hand chart ranks the reliability of other Chinese statistics, based on an analysis by Goldman Sachs. The closely watched figures for fixed-asset investment are among the least reliable. They include purchases of land, which only reflect changes in ownership, not an increase in capacity or value added. Rising land prices in recent years have therefore led to a big overstatement of the level and the growth of investment. In contrast, consumer spending is almost certainly much higher and growing faster than official figures suggest. Retail sales are often used as a proxy for private consumption, but they exclude services, the fastest-growing slice of households' budgets.

China's true inflation rate is probably higher than the consumer-price index (CPI) reports. One problem is that the CPI appears to be based on the prices of state-provided health, transport and education while ignoring their increasingly important private counterparts. Data for 36 cities collected by the National Development and Reform Commission show that inflation for medical care and education has been running at 5-10% since 2001, well above the 1-2% reported in the CPI. However, even if the official measure understates inflation, the changes in it may still be a fair gauge over time. Goldman Sachs therefore ranks it relatively high in terms of reliability.

Foreign trade is perhaps the most accurate economic indicator. Critics accuse China of fiddling its trade figures, because the value of its exports as measured by the importing country is always much bigger than what the Chinese report. This discrepancy reflects the fact that China's bilateral trade figures exclude goods shipped to Hong Kong before being re-exported. But this should not affect total export figures and detailed Hong Kong data are available to adjust bilateral trade flows.

The prize for the dodgiest figures goes to the labour market. The quarterly urban unemployment rate is meaningless because it excludes workers laid off by state-owned firms as well as large numbers of migrant workers, who normally live in urban areas but are not registered. Wage figures are also lousy. There has recently been much concern about the faster pace of increase in average urban earnings. But this series does not cover private firms, which are where most jobs have been created in recent years.

Now that China is such an engine of global growth, it urgently needs to improve its economic data. Only a madman would drive a juggernaut at full speed with a faulty speedometer, a cracked rear-view mirror and a misty windscreen.