Skip to content

Invest2 publishers2 min readPublished

Two reports of one NBS release put China's August investment 7.7 points apart

Two publishers describe the same August release from China's statistics bureau, and their fixed-asset investment figures sit 7.7 points apart. Only one of the two prints landed on the consensus it was measured against.

The Investor · Invest desk

Illustration accompanying Two reports of one NBS release put China's August investment 7.7 points apart

What happened

  • Cryptobriefing, citing a National Bureau of Statistics release on September 15, reported August retail sales up 3.4 per cent year on year, slowing from 3.7 per cent in July.
  • The same report put industrial output up 5.2 per cent in August against 5.7 per cent the month before, short of what analysts expected.
  • It also put fixed-asset investment up 0.5 per cent year to date through August, decelerating from 1.6 per cent through July.
  • SCMP, reporting the same statistics bureau release, said retail sales grew 0.4 per cent year on year last month.
  • SCMP also reported overall fixed-asset investment down 7.2 per cent in the January-August period, in line with a Wind forecast of a 7.19 per cent decline.

Compiled by The InvestorSomething wrong?How this is made

Why it matters

  • contradiction Anyone modelling Chinese consumption gets 3.4 per cent from one publisher and 0.4 from the other, a 3-point gap on the single line item that rate cuts and targeted lending were supposed to move.
  • decision Commodity desks have to pick a series before picking a view: cryptobriefing's own reading of investment near zero is subdued construction demand, and SCMP's figure is a contraction in it.
  • constraint Beijing has to reach an approximately 5 per cent 2026 growth target with property investment still falling at a double-digit rate, after a year of support measures that have not lifted consumption.
  • exposure Export strength is what keeps the August picture mixed, and cryptobriefing argues that a slowdown in key export markets would leave the domestic weakness on its own.

The forecasts settle more than the prints do. Cryptobriefing says the Reuters poll ahead of the release looked for fixed-asset investment growth of 1.4 per cent year to date [4]. SCMP says the Wind consensus looked for a 7.19 per cent contraction [8]. That is 8.59 points between two panels of analysts forecasting the same cumulative series in the same month [3], and the two publishers describe that series compatibly: SCMP as covering infrastructure, manufacturing and property construction [15], cryptobriefing as a gauge of infrastructure and business spending [16]. Analysts do not miss each other by 8.59 points on a monthly national statistic. One of the two accounts is describing a different period.

SCMP's number behaves like a print. Its actual sits 0.01 of a point from the forecast it cites [4], while the cryptobriefing figure misses its cited consensus by 0.9 [5]. That is not proof. The cryptobriefing set also hangs together, with real estate investment down 12.9 per cent over the first eight months [9] and urban unemployment at 5.3 per cent in August against 5.2 in July, attributed partly to recent graduates entering the workforce [10], so it is probably a genuine release of some period. Neither report reproduces the bureau's table.

Which one you take changes what you believe about everything outside property. Total fixed-asset investment is a weighted average of its components, and property fell 12.9 per cent in the eight months to August [9]. For the total to come in at plus 0.5 per cent, non-property investment has to be growing faster than 0.5 per cent; for the total to come in at minus 7.2, non-property investment only has to be above minus 7.2, which permits it to be contracting too [9].

The other lines in cryptobriefing's account move less. Retail sales decelerated 0.3 of a point from July and undershot the Reuters poll by 0.5 [6]. Industrial output slowed 0.5 of a point month on month [7]. Fixed-asset investment growth, on that account, fell by 1.1 points in a single month [8], which is the largest move in the set and the one the two publishers cannot agree on.

What would settle it is the bureau's own January-August line. If it shows plus 0.5 per cent, the consensus-matching argument fails and cryptobriefing has the release, in which case the story is a stall in construction spending and not a contraction. On the evidence as filed, I would size China industrial demand off the SCMP figure, and I would treat cryptobriefing's 25 per cent export surge [11] as the part of the print that needs the least revision either way.

What to watch

  • The bureau's next monthly release, and whether cumulative fixed-asset investment prints near zero or near minus seven per cent.
  • Whether other outlets covering the same NBS table match SCMP's 0.4 per cent retail figure or cryptobriefing's 3.4 per cent.
  • Whether copper and iron ore buyers position for construction spending that is flat or for construction spending that is shrinking.
Loading claim ledger
Loading source directory links
Loading share composer
Loading topic controls
Loading related stories