China’s Deflation Trap

House prices in China have been falling for more than four and a half years. Eurasia Group named China’s structural deflation its seventh top risk of 2026, observing that the wealth destruction involved is comparable in scale to America’s 2008 property crash, except it is still accelerating ¹. Property has been widely estimated at roughly a quarter to a third of Chinese GDP at its peak. Consumer confidence, private investment, and domestic demand have contracted with it.

Beijing bet that high-tech manufacturing would fill the gap. Instead, state-driven investment has created overcapacity in sectors from electric vehicles to steel to solar panels, and weak domestic demand means there are not enough buyers to absorb it. The result is what Chinese economists call “involution” ⁴: too many firms chasing too little demand, slashing prices to survive, compressing wages, weakening demand further.

The numbers are worse than the paper recorded. China’s share of industrial enterprises operating at a loss reached 30% in 2025, and 35% once private companies are excluded (the state-owned sector is the larger drag) ³. Since the 2021 peak, more than $6 trillion has been wiped from Chinese equity markets ². Evergrande was delisted from the Hong Kong stock exchange in August 2025, rendering its shares effectively worthless ⁶. China Vanke, once the country’s largest homebuilder, narrowly avoided a $284 million bond default in December 2025 and still faces hundreds of millions of dollars of debt repayments in 2026 ⁵. Morningstar analysts now expect home prices will not rebound until 2027 ⁷.

The structural dimension runs deeper than the property cycle. China’s fertility rate has fallen to approximately 1.0, its population began declining in 2022 for the first time in six decades, and the demographic drag on domestic demand is now self-reinforcing ¹⁹. This is structural, not a cyclical correction awaiting stimulus.

Japan is the natural reference point, but the analogy has limits. Both economies ran through a property-led credit boom, both absorbed a demographic peak, and both ended up with a long deflationary tail in the real economy. There the similarities thin out. Japan’s corporate cash sat structurally inside keiretsu cross-shareholding networks; China’s cash sits inside a state-owned sector that is the majority source of the industrial loss share and the primary zombie-company backstop. Japan liberalized capital flows through the lost decade; China tightened them. Japan’s banking system was bank-led and effectively on the sovereign’s balance sheet for years; China’s has been less transparently backstopped but no less implicitly guaranteed. And Japan eventually stabilized its fertility; China’s continues to fall. The useful comparison is that technologically advanced economies can absorb long deflationary adjustments without political rupture. It is not a forecast that China’s path will rhyme with Japan’s quarter for quarter.

With the 21st Party Congress coming in 2027, the political calculus favors control and technological supremacy over the consumption stimulus and structural reforms that could break the deflationary cycle. Beijing has the means to prevent a crisis. So far it has chosen gradualism.

The People’s Bank of China held the one-year loan prime rate at a record-low 3.0% in March 2026, the tenth consecutive month without a cut ¹⁰. The five-year rate, the benchmark for mortgages, stayed at 3.5%. Beijing set its 2026 growth target at 4.5 to 5%, the weakest target since 1991.

A nuance worth pricing: China’s producer price index turned positive in March 2026, rising 0.5% year over year after 41 months of decline, the longest deflationary streak in decades ⁹. Consumer inflation eased to 1.0% in March from a near three-year high of 1.3% in February ⁸. On paper, the deflation story is weakening. In practice, the CPI rebound was driven by the March 2026 oil shock from the Iran conflict, and the PPI reversal reflected the same commodity pass-through more than any return of domestic demand. The PBOC’s continued hold is the more honest signal. Beijing is still choosing gradualism.

The World Bank projects Chinese growth at 4.4% in 2026, respectable in headline terms but increasingly propped up by export surpluses rather than domestic demand ¹². Inward foreign direct investment turned negative in the third quarter of 2023, the first such print since Chinese FX data began in 1998, and has not meaningfully recovered ²². Capital is leaving.

The global transmission matters for every allocator, not just those with direct China exposure. For bond investors, China’s deflation exports disinflation through cheap manufactured imports, muddying the inflation signal that central banks rely on. For equity investors, the margin compression in sectors exposed to Chinese competition (electric vehicles, solar, batteries, steel, industrial robotics) is structural rather than cyclical.

The capital markets decoupling adds another layer. Western institutional capital is progressively reducing China exposure, not in a single dramatic exit but in the slow, persistent reallocation that shows up in FDI data, index reweightings, and LP conversations about emerging market allocations. Chinese and Western financial systems are operating on increasingly separate rails.

This is a prolonged, managed divergence, not a conventional crisis. For allocators who built models assuming China was a stable source of growth and an accessible investment market, the adjustment is structural and incomplete.

For the venture investor, China’s deflation trap transmits through three channels.

First, competitive commoditization. DeepSeek’s January 2025 emergence showed that Chinese AI development, operating under export controls, could match frontier model performance at a fraction of Western training costs ¹³. A year on, the pattern has held: DeepSeek’s V4 model now runs on Huawei’s domestic chips, blunting the effect of US export controls ²⁰. The company is raising its first external round of $300 million at a valuation above $10 billion ¹⁵. Alibaba, Tencent, Moonshot, and Zhipu have all shipped competitive open-source models ¹⁴. Chinese models are increasingly the default free tier in the Global South.

Second, hardware margin compression. Companies in robotics, electric vehicles, drones, and industrial AI face direct pressure from Chinese manufacturers exporting overcapacity at subsidized margins. These are precisely the sectors that absorbed the most Western VC capital in the 2021 to 2025 cycle.

Third, LP base contraction. As geopolitical decoupling progresses and outbound capital controls tighten, Chinese sovereign and family office participation in Western VC funds is declining. Three channels, compounding.

The counter-argument deserves a hearing. Chinese AI firms face genuine compute shortfalls. ByteDance’s Seedance 2.0 has struggled to meet user demand, with queue times stretching into hours for short clips, and Zhipu has raised prices on its GLM-5 series after complaints about delays and usage limits ²¹. The export controls are biting, even as Chinese firms work around them.

This refines the thesis rather than contradicting it. Chinese AI can compete on model quality and cost; it cannot yet compete on scale of deployment. The gap is closing slowly, not because China lacks talent or capital but because building leading-edge fab capacity takes a decade. For now, the most likely equilibrium is a bifurcated AI stack: Western hyperscaler infrastructure serving Western applications, Chinese models serving the rest, with neither side fully displacing the other.

For the portfolio, this translates into two exposures worth pricing explicitly. The first is margin compression in Chinese-overlap sectors (EV, solar, batteries, industrial robotics, commodity electronics). The second is disinflation drag on any thesis that assumes secularly rising Western goods prices. China does not need to grow for its deflation to matter. It needs only to continue exporting the excess capacity it is unwilling to rationalize. And that, for now, is what Beijing has chosen to do.