On Tuesday we had official PMIs from China:
Today we get the privately surveyed manufacturing PMI from Rating Dog/S&P Global. Rating Dog is the best name for an economic indicator I have ever heard of. Anyway, back to it.
The two sets of PMIs out of China each month routinely tell different stories. Knowing why saves you from reading a divergence as a data error.
The official one: NBS PMI
Compiled by China's National Bureau of Statistics, a government agency. It's the official read, and it leans toward the parts of the economy government policy actually touches.
The private one: now RatingDog, formerly Caixin
This is the one that changed names. S&P Global has compiled the survey throughout and continues to do so; what changed is the sponsorship. Caixin Media held the naming rights for years, but in mid-2025 a Shenzhen credit-research firm called RatingDog bought the exclusive naming rights from S&P Global, and the index was rebranded the RatingDog China PMI from the August 2025 release onward. Same methodology, same survey, new name on the masthead. You'll likely still see it referred to informally as "the old Caixin" for a while as the market adjusts.
Why they diverge
NBS skews toward state-owned, policy-supported industry. RatingDog skews toward private, market-facing, often export-exposed firms. When the two move in opposite directions, look at what's driving each survey's sample rather than treating one as "wrong." A weak NBS print with a firm RatingDog print, for instance, often points to private and export demand holding up while state-linked and domestic-policy sectors lag (or vice versa).
Reading them together
Used side by side, they're a decent cross-check on how much of any given month's China story is policy support versus organic private demand.
---
Also on the data calendar is the Bank of Japan Tankan. There are tow of these also!
Japan's two Tankans: a guide
Same idea as the China PMIs: two surveys, same rough purpose, different scope and cadence. Reuters Tankan and the Bank of Japan's own Tankan get bundled together in the data calendar, but they're not interchangeable.
The official one: BoJ Tankan
Compiled and published by the Bank of Japan itself. This is the headline number markets actually move on.
The interim one: Reuters Tankan
A private monthly survey, not affiliated with the BoJ, designed explicitly to track and anticipate the BoJ's quarterly number. It was originally called the Telerate Tankan before Reuters (now Thomson Reuters) acquired Telerate and put its own name on it; unlike the Caixin to RatingDog switch, there's no new rebrand here, it's been Reuters Tankan for years.
Why they diverge
The core difference is sample composition and timing rather than methodology. Reuters Tankan is large-company-only and monthly, so it can move ahead of the BoJ print, picking up shifts in sentiment among major exporters and manufacturers before the broader survey catches them. The BoJ Tankan adds in the small and mid-sized firm layer and is the one the BoJ, government and market actually treat as the authoritative quarterly read, but its survey window means it can be stale on fast-moving shocks by the time it's published.
Reading them together
In practice, Reuters Tankan is best used as a leading indicator into the BoJ print, and the BoJ Tankan as the print itself, the one that actually moves yen positioning and BoJ policy expectations.
I am not sure if anyone cares much about China PMIs and BoJ Tankan given later on Wednesday we get new Fed Chair Warsh speaking. Like Powell, we'll all be waiting! Although Warsh says he doesn't do forward guidance so I don't know what he'll be talking about. I'll get a preview posted separately.
This article was written by Eamonn Sheridan at investinglive.com.
Source: Investing Live - News