July Is Not One Number

A monthly data release invites a scoreboard. July resists one. Industrial output and the manufacturing survey describe production from different angles. Profits describe covered firms over seven months. Retail sales describe nominal goods and catering transactions. Fixed investment and property track cumulative spending, floor area, sales, and financing. They overlap, but they are not the same object.

The first rule is therefore grammatical: keep every subject attached to its verb. Industrial value added grew. Industrial profits grew in aggregate. Retail sales edged higher. Fixed investment fell. Property measures contracted. Purchasing managers reported a deterioration from June. Removing the subjects would turn six bounded observations into one unearned claim.

This is a source-audited data interpretation, not verified Mainland Dispatch reporting. NBS is the primary authority for what it measured, published, and defined. The World Bank, OECD, IMF, and BIS add institutional analysis, forecasts, and empirical mechanisms; they do not independently reproduce the July tables.

The cleanest positive observation is real industrial value added: 4.5 percent above July 2025 for enterprises above the designated size and 0.11 percent above June on the seasonally adjusted series. Manufacturing rose 5.5 percent, while mining fell 4.2 percent. High-technology manufacturing grew 16.9 percent. Those are output measures, not financial results.

The breadth was narrower than the headline. NBS reported year-over-year growth in 25 of 41 major industries and 279 of 626 listed products. Computers, communications, and other electronic equipment rose 19.1 percent in value-added terms, while some materials categories fell. Industrial resilience is defensible when it names this population and these differences.

The PMI supplies a different comparison. Its 49.2 reading asks purchasing managers whether conditions improved or deteriorated from the previous month and weights the responses into a diffusion index. The 50 threshold divides reported expansion from contraction; it does not translate into percentage output growth. July can therefore have a below-50 PMI and positive year-over-year industrial output without contradiction.

The two clocks matter. PMI moved from 50.3 in June to 49.2 in July, while industrial value added compared July 2026 with July 2025. One suggests month-to-month loss of breadth or momentum; the other says the covered output level remained above a year earlier.

July retail sales reached 3.9022 trillion yuan and rose 0.6 percent year over year. Excluding automobiles, they rose 2.5 percent. The difference matters because motor-vehicle retail sales at units above the designated size fell 17.0 percent. Food, telecommunications equipment, cosmetics, furniture, and building materials also moved at different rates.

The January-July total was 28.7744 trillion yuan, up 1.2 percent. Online sales of goods and services rose 4.8 percent over that period, but the release says its platform coverage was expanded and the measure is not comparable with the earlier online-retail series. A faster subseries cannot be treated as the whole consumer market.

Retail sales are not total household consumption. They cover physical goods sold to individuals and social groups plus catering revenue. Many services enter other statistical series, and the published retail growth rates are not adjusted for prices. Calling 0.6 percent nominal retail growth a real household-consumption rate would change both the population and the unit.

The World Bank and OECD interpret cautious consumption partly through saving behavior, social protection, and property adjustment. That is an analytical frame around the release, not a causal finding produced by the July retail table.

Fixed-asset investment excluding rural households totaled 26.0328 trillion yuan from January through July, down 6.7 percent year over year. The release covers projects with planned investment of at least 5 million yuan plus all real-estate development. It is cumulative, not price adjusted, and calculated against revised prior-year data on a comparable basis.

The components were uneven. Infrastructure investment fell 3.6 percent, manufacturing investment fell 1.7 percent, and property development investment fell 19.2 percent. Non-governmental investment fell 9.4 percent. Those declines explain why selected expanding categories did not lift the total.

Intellectual-property-product investment rose 9.1 percent. Information-transmission investment rose 26.0 percent, aerospace-equipment investment 18.7 percent, and investment in computers, communications, and other electronic equipment 7.8 percent. These observations support a rotation or divergence reading only when their distinct category weights remain unknown.

The NBS summary adds a high-technology-industry investment measure of 5.0 percent growth. It is useful official context, but it should not be silently merged with the intellectual-property or named industry series. Similar labels do not guarantee identical classifications or weights.

Property is the broadest cluster of negative observations in the packet. Development investment fell 19.2 percent, new starts by floor area fell 24.0 percent, the floor area of newly built commercial buildings sold fell 11.8 percent, and sales value fell 13.1 percent. Funds available to development enterprises fell 20.3 percent.

The units prevent false arithmetic. Investment, sales value, and funds are monetary flows. New starts and sales area are square meters. The figures cover January through July and are compared with the same period of 2025. They cannot be added, averaged, or described as one property index.

The IMF's 2025 Article IV staff work links property downturns to weaker consumption and disinflation through investment, local-government finance, and household wealth. BIS researchers using Alipay transactions from 2017 through early 2023 find positive housing wealth effects in 33 larger cities but different results in smaller cities and among younger people. These sources make the transmission channel plausible and heterogeneous; they do not measure its July 2026 size.

The alternative reading is not that property is irrelevant. It is that a property contraction can coexist with growth in industrial output and selected investment categories, and that coexistence says more than either a national collapse story or a national acceleration story.

The six signals line up most clearly as a composition problem. Supply-side industrial measures remained positive. The household-facing retail measure was positive but soft. The manufacturing survey moved below its threshold. Investment fell overall while selected technology-related categories expanded. Property contraction cut across activity, sales, and financing.

Profit growth adds evidence of industrial earnings, not a universal demand verdict. The 17.6 percent aggregate included a sharp electronics rise and a sharp automotive fall. Receivables rose 8.5 percent and finished-goods inventories 10.8 percent at the end of July. Those facts are consistent with several business conditions, and the release does not identify one cause.

The official NBS synthesis calls the economy generally stable and says the imbalance between strong supply and weak demand remained acute. That wording is a government characterization. The World Bank, OECD, and IMF use different models and assumptions but also distinguish industrial or export resilience from property and consumption weakness. Their overlap is interpretively useful without becoming independent verification of every official number.

A bounded conclusion can carry all of this at once: July's measured activity was uneven by sector, demand channel, and statistical clock. The next release may change the direction or breadth. It will not make these July definitions interchangeable after the fact.