China PMI 49.2, CPI 0.5%, Chip Exports +117% (July 2026): The Two-Speed Economy in One Month of Data — and What It Means for the Aussie and Kiwi
China's July data set is complete: PMI 49.2, CPI 0.5%, PPI 3.5%, exports +23.9%, chip exports +117%. Two economies in one month — and what reaches AUD and NZD.
China PMI 49.2, CPI 0.5%, Chip Exports +117% (July 2026): The Two-Speed Economy in One Month of Data — and What It Means for the Aussie and Kiwi
China's July data set is now complete, and it describes two economies. Both purchasing managers' surveys contracted — manufacturing 49.2, non-manufacturing 49.0, the composite at a post-2022 low of 49.3. Consumer inflation then slowed to 0.5% on 9 August, the weakest since January, with food prices down 1.5%. Yet exports grew 23.9% in the same month and chip exports alone rose 117%. A country whose factories are shipping record volumes of semiconductors while its households will not pay more for food is not one economy having a bad month. It is two economies, and only one of them buys iron ore.
The preview below was written on 27 July, when consensus expected stability at 50.3. The scenario that landed was the downside one, and it landed harder than the scenario table allowed for: not a headline flattered by production while forward demand rolled over, but a broad decline in which production, new orders and export orders all fell together. The trade and inflation data that followed in the first week of August confirmed the split the survey had already flagged. What follows is the outcome, the mechanism by which it reaches the commodity currencies, and why the exchange rate has not simply followed.
- Official manufacturing PMI: 49.2 in July, down from 50.3 in June, versus a consensus near 50.1. First sub-50 reading since February. Non-manufacturing fell to 49.0 and the composite to 49.3, the lowest since December 2022.
- The internals were worse than the headline: new orders 48.5 (from 51.2, the weakest since 2023), production 49.9 (from 51.4), new export orders 49.6 (from 50.1). New orders is the component that reaches raw materials.
- July CPI, released 9 August, rose 0.5% year on year against 0.8% expected and 1.0% in June — the slowest since January. Core CPI 0.9%, food prices −1.5%, and month-on-month CPI −0.1% against an expected +0.2%.
- Factory-gate prices rose 3.5%, down from 4.1% and below the 3.8% expected — a three-month low. Producer prices are still positive only because mining and raw materials are pulling them up, an energy shock imported from outside China.
- Exports rose 23.9% in July and chip exports 117%, with a $112.5bn surplus. The export engine and the domestic consumer are running at completely different speeds, and the commodity currencies care about the second one more.
- AUD/USD rose to US$0.7063 and NZD/USD to US$0.5877 by 7 August, higher than before any of this data, because US payrolls fell 23,000 and the dollar leg moved further than the China leg. See how the commodities, risk and rates factors are scoring right now on the live meter.
Update, 9 August: the two-speed economy, quantified
The survey came first, and the hard data has now caught up with it in both directions at once.
On Sunday 9 August the National Bureau of Statistics reported that consumer prices rose 0.5% year on year in July, down from 1.0% in June and below the 0.8% economists expected — the slowest pace since January. Core CPI, excluding food and energy, rose 0.9%. Food prices fell 1.5%. On a monthly basis prices edged down 0.1%, against an expected 0.2% rise and following a 0.3% fall in June. Factory-gate prices rose 3.5% year on year, easing from 4.1% to a three-month low and undershooting the 3.8% expected in a Reuters poll (South China Morning Post; Reuters).
Two days earlier, customs data had gone the other way. Exports grew 23.9% year on year in US dollar terms, beating the 22.2% expected, though easing from June's 27% — the fastest pace since October 2021. Imports rose 27.5%, marginally under the 27.9% forecast. The trade surplus was $112.5 billion against roughly $107 billion expected, narrower than June's $125.6 billion. Chip exports surged 117% from a year earlier in July alone, and mechanical and electrical products made up more than 60% of total shipments across the first seven months (CNBC).
| July 2026 release | Expected | Prior | What it describes | |
|---|---|---|---|---|
| Manufacturing PMI (31 Jul) | 49.2 | ~50.1 | 50.3 | Forward demand for inputs |
| Composite output (31 Jul) | 49.3 | — | 50.6 | Whole-economy activity |
| Exports, y/y (7 Aug) | +23.9% | +22.2% | +27.0% | External demand, AI-led |
| Imports, y/y (7 Aug) | +27.5% | +27.9% | +36.0% | Domestic absorption |
| CPI, y/y (9 Aug) | +0.5% | +0.8% | +1.0% | Household demand |
| PPI, y/y (9 Aug) | +3.5% | +3.8% | +4.1% | Imported energy and raw materials |
Note what the two columns of misses have in common. Every domestic-demand indicator undershot; the one external-demand indicator beat. Zhiwei Zhang, chief economist at Pinpoint Asset Management, tied the inflation data directly back to the survey this post was written about, saying the weakening in July was "consistent with other activity data such as the PMI index, which also dropped more than expected" — while separately expecting China's export engine to remain strong through the third quarter.
Zhaopeng Xing, ANZ's senior China strategist, put the mechanism plainly: "Lower oil prices, combined with weakening demand, caused both (consumer and producer price inflation) in July to come in below expectations." On the policy response he added that "the effect of faster fiscal spending in the second half of the year is likely to be felt with a lag of about one quarter" — which is the timing constraint that matters for anyone waiting on the Politburo's pledges to show up in commodity demand. ANZ forecasts full-year PPI of 2.5% and CPI of 1%.
The export composition deserves one further note, because it is the same divide the PMI's sub-indices showed. A 117% surge in chip exports and a 60%-plus share for mechanical and electrical goods is an export boom concentrated in exactly the sectors that consume the least iron ore per dollar of output. The global build-out of AI infrastructure is doing the work, and it reaches Australia's terms of trade very weakly — a point the demand side of the memory market makes from the other direction in our piece on Chinese DRAM inside HP, Asus and Acer laptops. A record month for Chinese semiconductor shipments and a soft month for Chinese steel are perfectly compatible, and only the second one shows up in how the commodities factor scores the Australian dollar.
One honest caveat on timing: the inflation data landed on a Sunday, before currency markets reopened. There is no exchange-rate reaction to report yet, and the levels quoted in this post are Friday 7 August closes.
What actually happened on 31 July
The National Bureau of Statistics released all three surveys together at 09:30 Beijing time. Manufacturing printed 49.2, down 1.1 points from June's 50.3 and below the roughly 50.1 consensus (South China Morning Post). Non-manufacturing, which bundles services with construction, fell to 49.0 from 50.2. The composite output index fell to 49.3 from 50.6.
A diffusion index at 49.2 does not mean output fell 0.8%. It means more surveyed purchasing managers reported deterioration than improvement. The magnitude is small; the sign change is what matters, because a series that had spent four months grinding along just above the line has now crossed it — and crossed it on both the industrial and the services side at once.
The sub-indices carry the real signal. Production fell to 49.9 from 51.4. New orders fell to 48.5 from 51.2, the weakest reading since 2023. New export orders fell to 49.6 from 50.1. That combination rules out the more comfortable interpretation: this was not a month in which firms worked through a backlog while demand held, nor one in which a single external shock hit exports. Domestic and foreign demand softened together while output fell with them.
Huo Lihui, the NBS chief statistician, attributed the cooling to a high base of comparison after recent rapid growth and to the traditional off-season in some manufacturing sectors, noting that both production and market demand eased. Economists added detail the official line leaves out: Capital Economics pointed to weak domestic goods demand and soft building activity, and to several typhoons in July that disrupted work. Lynn Song, chief economist for Greater China at ING, called it "an unpromising start to the first wave of economic data for the second half of the year" (Reuters, via The Globe and Mail).
The narrow base got narrower
The pattern this survey has shown all year is an aggregate held above water by a small number of advanced-manufacturing sub-sectors. July did not break that pattern; it removed the water. Specialised equipment and other hi-tech sub-sectors still reported production and new-order readings above 53, while ferrous metal smelting and the automotive industry showed weak supply and demand.
That split is the crux for anyone trading the Aussie rather than an equity index. Semiconductors, servers and precision equipment are not iron-ore-intensive. Steel, construction and heavy industry are. A survey in which the strong sub-sectors are 53-and-above and the weak ones are ferrous metals is close to the worst available composition for bulk commodity demand, even at a given headline level. Our earlier China Q2 GDP piece traced the same divergence through the national accounts, where Q2 growth of 4.3% undershot the 4.5%–5% target range Beijing set in March (CNN; the target itself was the lowest on record).
The scenario that landed, factor by factor
The preview mapped three outcomes. The downside case arrived. Here is what each fundamental channel actually did, rather than what the headline implies.
| Channel | What the July print did | What it means for AUD / NZD |
|---|---|---|
| Commodities | New orders 48.5, the weakest since 2023, with ferrous metals explicitly weak | Clearly negative. This is the component that translates into future iron ore, coal and copper purchases, and it fell 2.7 points in a month |
| Risk sentiment | Composite at a post-2022 low, both surveys contracting together | Negative, but partly absorbed: global equity risk appetite that week was set by US mega-cap earnings, not by Chinese diffusion indices |
| Growth | Production 49.9 and export orders 49.6 — softness is domestic and external at once | Negative for the China-linked growth read; the tech export channel that carried H1 is no longer offsetting |
| Rates | Unchanged by this release. The PBoC held its loan prime rates on 20 July for a fourteenth month | Neutral on the day. The Politburo's language raises the probability of easing later in the year, which is a China story, not an AUD rate story |
| The other side of the pair | The Fed held rates steady on 29 July and the dollar softened afterwards | Positive for AUD/USD and NZD/USD — and large enough to cancel the China signal on the day |
That last row is why AUD/USD sat near US$0.70 after a print that damaged two of its factors, and why NZD/USD held near US$0.578. Both pairs are two-sided. A currency can weaken on its own fundamentals and still trade flat or higher against a dollar that is weakening faster. This is precisely the case for scoring each of the eight majors on its own fundamentals rather than reading a single pair as a verdict — the approach we set out here.
Australia's domestic calendar reinforced the point. Q2 CPI landed on 29 July at 3.8% year on year, down from 4.0%, with the trimmed mean at 3.6% and the quarterly print below forecasts (Australian Bureau of Statistics). We mapped those scenarios in the Australia Q2 CPI post. Softer inflation ahead of the RBA's 11 August decision is a mild negative for the Aussie's rate factor — arriving in the same week the commodity factor deteriorated, and still not enough to push the pair down, because the dollar was doing more work than either.
The policy answer arrived a day before the data
The Politburo held its economic meeting on Thursday 30 July, and the readout landed before the survey that would have justified it. The leadership acknowledged the difficulties in the economy and pledged to intensify counter-cyclical adjustment and roll out incremental measures in the second half, with faster fiscal spending and use of bond proceeds, support for employment and for stabilising the property market, and a specific emphasis on expanding services consumption (Caixin). Local-government special-purpose bond issuance had reached 2.1 trillion yuan in the first half, about 47% of the annual quota, marginally behind the equivalent 2025 pace — which is the mechanical basis for "faster delivery" meaning something.
The preview expected firmer easing language and accelerated fiscal delivery rather than a large new package, and that is what the readout delivered. The consequence for commodity currencies is the one worth holding onto: stimulus aimed at consumption, employment and services supports Chinese growth and global risk appetite, but it is far less iron-ore-intensive than the property-and-infrastructure cycles that trained a generation of traders to read "China stimulus" as "buy the Aussie". The same package can be a positive risk-sentiment impulse and a neutral commodities impulse at the same time. On the monetary side, the People's Bank of China held its loan prime rates unchanged on 20 July for a fourteenth consecutive month, at 3.0% and 3.5% (People's Daily).
The raw-material market had already moved
The most useful confirmation that this print reaches the commodity currencies is that the underlying market had begun repricing before the survey printed. Dalian iron ore futures were at 714.5 yuan a tonne on 31 July, down about 2.5% over the month. Chinese steel mill economics explain why: average losses at Tangshan mills had widened beyond 100 yuan per tonne, and average daily hot metal output had fallen for a third consecutive week to 2.38 million tonnes as of 23 July, the lowest since 3 April (market data).
That sequence — mill margins, then hot metal, then ore — is the transmission chain the PMI's new-orders component is a leading indicator for. It also revises the more constructive line in the original preview, which noted that iron ore had "at least stopped falling" after recovering to around US$102.73 a tonne CFR in mid-July. It resumed. We traced the earlier leg of that episode in why the Australian dollar was falling with a rate above the Fed's, and the mechanism has not changed: Australia's terms of trade answer to Chinese heavy industry, and Chinese heavy industry is the part of this survey that is contracting.
What to watch from here
Three things carry the story forward, and none of them is the next headline number in isolation.
First, whether new orders stabilises. A single month below 50 in a diffusion index is noise; two months with the forward-looking component in the 48s is a demand trend that steel output has to answer. The immediate cross-check has since arrived and pointed the same way: the private-sector survey compiled by S&P Global, published under the RatingDog name and skewed toward smaller exporters, slowed to a four-month low, with weakening new orders in both data sets. The August surveys, due at the start of September, now decide whether this is a trend.
Second, whether the non-manufacturing weakness persists. The composite at a post-2022 low was the genuinely new information in the July release, and it matters more for New Zealand than for Australia — the Kiwi's China exposure runs through dairy and the consumer, not through blast furnaces. The 9 August CPI, with food prices down 1.5% and core at 0.9%, is the first hard confirmation that the consumer side is where the softness sits.
Third, what the Politburo's "incremental policies" turn out to be in practice. Announced intent and delivered spending are different variables, and the commodity currencies only respond to the second. ANZ's estimate of a one-quarter lag between faster fiscal spending and its effect on prices sets the earliest realistic window for any of it to show up in the data. The composition will matter as much as the size: bond issuance routed to property completion reaches iron ore, while consumption vouchers and services support largely do not.
Fourth, the other side of the pair, which has decided every move so far. The Reserve Bank of Australia meets on 11 August with Q2 CPI at 3.8% behind it, and the US data flow after a July payrolls print of −23,000 is what set the dollar leg in the first week of August — the mechanism we traced in the July jobs report post.
The honest summary of July is that China's activity data got worse, its consumer prices confirmed it, its export machine had its second-best month in five years anyway, and the Aussie rose through all of it — because the US dollar was moving more. Those are four separate statements, and collapsing them into one is how a reader ends up believing that a bad number means a lower price.
Educational macro context only — not investment advice.