China: Balance Sheet Recession and the Case for Longer-Term AUD Bearishness
Date: 19 August 2026.
1. What the data says
Chinese activity data for July disappointed across the board:
| July 2026 | expected | prior | |
|---|---|---|---|
| Industrial production (y/y) | 4.5% | 5.0% | 5.3% |
| Retail sales (y/y) | 0.6% | 1.5% | 1.0% |
| Fixed-asset investment (Jan–Jul, y/y) | −6.7% | −6.2% | — |
| Surveyed unemployment | 5.2% | — | 3-month high |
| M2 (y/y) | 7.7% | 7.9% | 8.0% |
Fixed-asset investment is contracting outright. Rising investment in manufacturing can no longer offset the fall in real estate investment; the absolute level of manufacturing investment remains very high but is now shrinking rather than growing. That leaves exports as the only remaining source of growth, hence the rising trade surplus, which some analysts note has been partly obscured in the headline data by a surge in gold imports.
Chinese auto exports are growing at over 3m cars a year — Japan at its peak exported 6m in total — yet total auto production is flat, because domestic demand is falling. Record export success is producing zero net output growth.
2. Credit is now contracting, not merely slowing
July's credit data is the important release, not the activity data:
- Net new yuan loans −CNY 340bn (−USD 50.4bn) against a Reuters consensus of +CNY 45bn — a record monthly contraction and the second this year, after April.
- Household loans −CNY 460bn; corporate loans −CNY 130bn.
- Repayments to the real economy CNY 590bn, the most since 2002.
- Total social financing still +CNY 1.4tn, of which CNY 1.32tn was government bond issuance — roughly 94% of the total.
- Jan–Jul new loans CNY 10.38tn versus CNY 12.87tn a year earlier, −19%.
Households are deleveraging harder than corporates. This is a property-led balance sheet recession, not the corporate-led version Koo described in 1990s Japan. The policy implications differ - household balance sheets repair more slowly and are less responsive to corporate tax or credit measures.
3. Why Chinese yields fall while the world's rise
With bond yields rising almost everywhere — US 10y ~4.65–4.70% near a 19-month high, Bund 3.21% at its highest since 2011, 20y JGB at levels not seen since 1999 — with the exception of China (10y ~1.68%, lowest in over a year), we think it's because China is in a balance sheet recession. With the property bubble deflated, households and firms are minimising debt rather than maximising profit, so credit demand fails at any policy rate: the LPR has sat at 3.0% (1y) and 3.5% (5y) for ten months and the PBOC calls its stance "appropriately loose", yet July still delivered a record contraction in lending. Banks that cannot lend buy government paper instead, so yields fall even as the state issues at a record pace. That is the mirror image of the rest of the world, where the private sector still borrows and fiscal expansion competes with it for savings. The 10y at 1.68% is not evidence that easing is working — it is the price of credit demand failure, and no policy rate makes a household with negative equity want a second mortgage.
The debt picture follows. Private debt is being repaid while public debt expands to fill the gap: the stock keeps growing, the composition shifts from private to sovereign. Some analysts put the scale at among the fastest increases in a debt-to-GDP ratio in history over the past fifteen years. Reining that in requires investment growth to slow sharply, perhaps to go negative. Consumption growth lags the GDP target, so investment can only slow without missing the target if the trade surplus absorbs the difference. The surplus grows as the residual of that attempt, and it holds only as long as the rest of the world keeps absorbing it. The other option we have seen being circulated was that it would be possible to spend more on infrastructure investment, which would mean increased demand for steel, of which iron ore is the key component, and which could positively impact the Australian dollar.
4. Decoupling raises the probability that it won't be absorbed
We read the US administration's direction as decoupling: tariffs, restrictions on AI hardware access, and successive trade measures, most recently involving Mexico. Disruption in the Strait of Hormuz falls disproportionately on China as the largest buyer of Gulf and Iranian crude; we treat this as an effect that compounds the pressure.
If external demand is constrained precisely when China depends on it to offset private deleveraging, the adjustment has to come through domestic demand or the currency. Neither is favourable for commodity exporters.
5. Implication
We are longer-term bearish AUD: China takes roughly three-quarters of seaborne iron ore, and iron ore is Australia's largest export. New supply from Simandou and Vale compounds the demand problem. But short-term, we are neutral and watching for Chinese infrastructure spending, which could mean we lean long AUD short-term.
