China Musings: The Deleveraging Trap – A Fragile Balance Sheet

In this report, we examine the evolution of Chinese households balance sheet since the housing downturn.

Key Takeaways

• China's household deleveraging began with housing in 2021, but since mid-2023 has become increasingly confidence-driven.
• Borrowing capacity has improved for three years, yet credit demand has not. Desired debt levels continue to decline as income expectations weaken.
• Balance sheets are repairing, but partly through debt reduction and higher savings, leaving consumption to bear part of the adjustment.
• With households, local govs, and parts of industry tightening, balance sheet repair risks feeding through to weaker nominal income growth + renewed deflation.
• The economy needs an offsetting source of leverage, but elevated aggregate debt and weak investment returns limit the scope for a meaningful offset.

Detailed Musings

Two phases of household deleveraging

Housing-led deleveraging, 2H21 to 1H23. The property downturn that began in mid-2021 fundamentally altered the household leverage cycle. The first phase of deleveraging, from 2021 to mid-2023, was primarily a housing-price shock. As expectations for housing appreciation declined, the incentive to take on additional leverage weakened sharply. Balance-sheet metrics reflected a shock-driven, largely passive adjustment. Debt accumulation slowed and the debt-service burden eventually stabilized as interest rates fell, but the liability-to-asset ratio continued to rise.

Confidence-led deleveraging, from mid-2023. Household credit stabilized briefly in early 2023 as the economy reopened and property transactions recovered, but subsequently resumed its decline despite two years of policy easing, lower mortgage rates, and relaxed purchase restrictions. The weakness also broadened. Short-term consumer lending has remained persistently soft and increasingly contractionary, despite improvements in both the debt-to-income ratio and debt-service ratio. In this context, the primary constraint on household borrowing appears to be demand rather than affordability. Persistent housing weakness has been compounded by deteriorating labor-market expectations and weaker expected income growth, pushing desired debt levels below the existing stock of household debt.

横向滑动查看完整图表

Three leverage ratios, one message: balance sheet repair without spending recovery

Debt-to-disposable income turned first. This stock measure captures how long it would take households to repay debt using current income. The ratio peaked at around 110% in 2022–23 and had fallen to below 100% by June 2026. The decomposition shows income growth continuing to lower the ratio, while the contribution from debt has faded and turned negative in several recent quarters. The improvement is therefore no longer driven by income outpacing an expanding debt stock. Instead, households are actively reducing the stock of debt itself.

The debt-service ratio then fell more visibly. This cash-flow measure captures the share of disposable income absorbed by interest and principal payments. The four-quarter average DSR peaked at around 12.4% in late 2022 and early 2023 before declining to about 10.8% by June 2026. The contribution from debt servicing has progressively diminished as interest rates repriced lower and debt growth slowed, while income growth has continued to reduce the burden.

The liability-to-asset ratio was the last metric to turn. This balance-sheet measure is the closest proxy for household solvency and net worth. It continued to rise throughout the property downturn and only peaked around mid-2025 at nearly 14%, before easing to roughly 13.5% by June 2026. That lag is significant: falling housing wealth and weak asset prices initially offset slower liability growth. More recently, the decomposition points to a combination of asset-side stabilization and outright debt reduction.

How the repair is taking place matters just as much as the repair itself. A benign balance-sheet adjustment is denominator-led: disposable income and asset prices rise, leverage ratios fall, and households can increase spending without taking on additional debt. A contractionary adjustment is numerator-led: households reduce liabilities and increase financial savings, leaving current expenditure to absorb part of the adjustment. China currently sits somewhere between the two. Financial asset valuations are helping to stabilize the liability-to-asset ratio, but outright debt reduction and cautious consumer spending remain central to the adjustment process. In other words, households are repairing their balance sheets, but they are doing so partly at the expense of consumption.

Someone needs to lever up

Micro-level repair may not survive the macro feedback loop. The acceleration in household deleveraging is arriving at a difficult juncture. The property market remains in adjustment, with the new pre-sale rules likely to weigh on construction capex in 2027–28. Local government financing remains constrained. Parts of the industrial sector continue to grapple with overcapacity and are cutting investment. Meanwhile, the fiscal stance is tightening rather than easing, particularly with the recent strengthening of tax collection. If most sectors are deleveraging simultaneously, either a wider central government deficit or a larger current-account surplus must absorb the excess savings. Otherwise, nominal growth must adjust lower. Slower nominal growth, in turn, leads to weaker income growth, reinforcing households' incentive to save and reduce debt. Even as debt levels decline and balance-sheet metrics improve, the desired level of debt may continue to fall faster than the actual stock, raising the risk of a self-reinforcing deleveraging–deflation cycle.

The external surplus is doing much of the heavy lifting, but it is not a sufficient offset. China's current-account surplus widened to 3.8% of GDP in 2025 from just 0.6% in 2019, reversing the structural narrowing trend that had been in place since 2007 (Exhibit 9). However, a large surplus in an economy the size of China inevitably creates greater trade friction. Moreover, export income and profits increasingly accrue to high-tech and capital-intensive industries, limiting the spillover effects to the broader labor market and household income expectations (Exhibit 10).

The US post-GFC experience shows how private-sector deleveraging can be offset by public-sector releveraging. US households repaired their balance sheets through debt repayment, defaults, and reduced borrowing, while parts of the corporate and financial sectors also reduced leverage. The public sector moved in the opposite direction. Federal government debt rose from roughly 64% of GDP in 2Q08 to 93% by 1Q11, and subsequently to around 110% following the pandemic. At the same time, the Federal Reserve adopted quantitative easing and maintained a low-yield environment for an extended period. Together, fiscal and monetary support underpinned nominal income growth, reduced real debt burdens, and ultimately helped stabilize house prices. The household deleveraging cycle ended once asset and income reflation replaced liability contraction as the primary adjustment mechanism. As house prices and employment recovered, stronger household wealth and disposable income supported a re-acceleration in borrowing without worsening debt-to-asset ratios.

Even with active public balance-sheet expansion, US households required five years to deleverage. Despite a substantial policy offset, including a 29 percentage-point increase in federal debt in less than three years and large-scale QE, US household deleveraging persisted for roughly five years before debt growth resumed. We believe China is undergoing a similar balance-sheet adjustment, but without an equivalent fiscal or monetary offset.

China entered the current deleveraging cycle with less fiscal headroom than the US had in 2008. Economy-wide leverage was already high at roughly 275% of GDP in 2021, while broad government debt, including local governments and LGFVs, stood at around 88% of GDP, compared with US federal debt of 64% of GDP in 2Q08. The property downturn has further weakened the fiscal capacity of the broader public sector.

This constrains Beijing's ability to replicate the post-GFC US playbook of shifting private-sector leverage onto the sovereign balance sheet. The central government can still use its relatively stronger balance sheet to refinance local liabilities and provide targeted support to demand. However, the case for large-scale releveraging is less compelling given already elevated public debt, diminishing returns on past debt-funded investment, and the risk that another infrastructure-led cycle merely transfers the balance-sheet problem rather than resolving it.

Rising growth downside risks

Household deleveraging will likely continue, and improving balance-sheet metrics should not be mistaken for an imminent consumption inflection. While household balance sheets have become more resilient, this alone is unlikely to reduce precautionary savings. A more durable improvement would require fiscal and structural measures that directly support household disposable income and service-sector employment, alongside a credible stabilization in housing wealth expectations. However, we expect fiscal policy to remain focused on cushioning downside risks rather than driving a meaningful economic rebalancing, while the broader policy framework remains supply-side oriented. As a result, employment and income expectations may take longer to recover.

Growth downside risks are rising. We expect real GDP growth to remain near the lower end of the 4.5–5.0% target range, supported by capex related to the "Six Networks" initiative and continued export resilience. However, downside risks from weak domestic demand are becoming increasingly apparent. Nominal GDP growth could soften once the current boost from imported inflation fades. We forecast the GDP deflator at 0.8% in 2026 and 0.2% in 2027, with upside risks to the 2026 outlook stemming from oil prices, but increasing downside risks in 2027.

On the RMB, we continue to see only modest appreciation potential rather than a large valuation gap. In the post-housing equilibrium, a weaker RMB is consistent with structurally weak domestic demand, high savings, and a lower equilibrium interest rate. Recent policy actions are also inconsistent with a deliberate strategy of maintaining a materially undervalued currency. In the near term, asymmetric policy tightening abroad, the restructuring of outbound capital flows, and an FX conversion ratio that remains above its recent average should support some appreciation momentum. However, rapid appreciation is not an effective rebalancing tool. It would disproportionately affect lower-value-added but labor-intensive sectors. A stronger RMB should emerge as a consequence of domestic reflation and economic rebalancing.

Potential policy measures to manage the household deleveraging cycle

• Systematic support for service consumption and employment
• Broader mortgage subsidies to help households repair balance sheets
• Faster improvements to the social welfare system following tighter contribution enforcement
• Increased support for fertility, childcare, and child-rearing costs