## SMOOTHING THE PATH TO A NEW NORMAL: CHINA’S PROPERTY SECTOR TRANSITION

## Source details

**Canonical URL:** [SMOOTHING THE PATH TO A NEW NORMAL: CHINA’S PROPERTY SECTOR TRANSITION](https://www.imf.org/-/media/files/publications/cr/2024/english/1chnea2024002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2024/english/1chnea2024002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2024/english/1chnea2024002.pdf.json)

---

### A. Introduction and macro-financial context
- Property-related activities accounted for an estimated "20 percent of GDP" through China’s decades of rapid growth.
- Average sales prices rose almost "350 percent" in the 15 years through 2021.
- Policy and market history:
  - "Three Red Lines" (August 2020) limitations: liability-to-asset ratios above "70 percent"; net debt higher than "100 percent" of equity; cash less than "100 percent" of short-term debt.
  - Default of the second largest developer by sales in late 2021 precipitated widespread liquidity stress as presales slowed.
  - Authorities eased demand-side measures (downpayment, mortgage restrictions, existing mortgage rates) but avoided large-scale bailouts.
- Central near-term outlook:
  - Real property investment likely to fall by "30 to 60 percent from end-2022 levels" and rebound only modestly thereafter.
- Policy priorities emphasized:
  - Expedite restructuring of nonviable developers.
  - Support and de-risk surviving developers.
  - Contain buildup of market risks.
  - Reform the presales system.

### B. Drivers of the medium-term decline in new housing investment
- Demographics and household formation:
  - Urban population growth set to decelerate as net migration slows and total population declines.
  - Declining average household size could partially offset population drag on housing demand.
- Publicly funded redevelopment:
  - Share of population living in housing built before 2000 fell from "66 percent in 2010 to 35 percent in 2020" (implying net destruction of "66 million" pre-2000 housing units).
  - Share of housing units without modern amenities fell from "28 to 12 percent (2010 to 2020)".
  - Targets for urban redevelopment scaled down almost "75 percent" for the 14th Five Year Plan (2021-25) versus the 13th (2016-2020).
- Supply-side imbalances and market distortions:
  - Rapid pre-2021 building and sharp sales declines since mid-2021 left large inventories of finished and unfinished housing.
  - Interventions to limit price declines reduced market-clearing flexibility and appear to have exacerbated sales declines.
  - Price evolution uncertain: mortgage credit and household income support prices, while completion risk, distressed inventory liquidation, reduced speculative demand, and weaker income growth create downside risks.

### C. Projection framework and scenarios for fundamental demand
- Methodology:
  - Projections follow Chivakul et al (2015): developer supply responds to expected fundamental demand (additional units for household growth and replacement of net lost units), modified by supply-side distortions.
  - Prices discussed qualitatively (not quantified in projections).
- Two household-size scenarios (UN medium fertility population projections; World Bank urbanization slowdowns):
  - Lower average household size scenario:
    - Average persons per household falls from "2.77 in 2021 to 2.4 in 2034."
    - Replacement demand extends 2010–2020 decline in pre-2000 housing starting from roughly "2.8 million units per year."
    - Average new-build apartment size grows from "114 square meters in 2020 to 121 square meters."
  - Higher average household size scenario:
    - Average household size falls to "2.6."
    - Replacement demand at two-thirds the pace of the upside scenario, generating demand for "2.1 million units of housing per year over a decade."
    - Average apartment size unchanged.

### D. Fundamental demand projections and scenario ranges
- Central quantitative outcomes and ranges:
  - Fundamental demand projected to decline "roughly in the range of around 35-55 percent relative to the past decade, to an average of about 950 million square meters per year."
  - Replacement demand provides the largest overall compression due to the rapid pace of demolition in 2013-2018.
  - Fundamental demand estimates broadly align with recent private estimates cited (examples preserved verbatim in source).
- Key reference figures:
  - Fundamental demand range: "around 35-55 percent" decline; average "about 950 million square meters per year."
  - Central scenario new starts: "715 million square meters" average 2024-2033.
  - 2019-2021 average starts (reference peak): "1.6 billion square meters."

### E. Housing supply assumptions and quantified supply-side drags
- Developers’ supply timing assumption:
  - Supply (starts, floor space terms) equals expected fundamental demand "1.5 years in the future," reflecting a three-year production cycle with presales allowed one year into production.
- Supply-side modules layered onto baseline:
  - Excess inventories: cumulative lagged "1.5 year starts less sales"; inventories clear as starts reduce.
  - New supply from investor sales: conservative estimate of "5 percent of cumulative 2017-2021 sales" → "377 million square meters."
  - Presale completion backlogs: distressed developers assumed to trade off completing "one quarter the estimated presale completion backlog attributed to liquidity-constrained developers, or ten percent of the total" → reduction in new starts of "535 million square meters."
- Combined drag and timing:
  - Combined supply-side drag equivalent to "about 17 percent of projected fundamental demand in the 2024-2033 period."
  - Excess inventories as of end-2023: "just over 1 billion square meters in floor space terms."
  - Supply-side drags assumed concentrated in the first three years, dissipating over the next four years.

### F. Projected starts, real estate GFCF, and macro linkages
- New starts trajectory:
  - New starts average about "715 million square meters in the central scenario" for 2024-2033, about "45 percent of the 2019-2021 average level (1.6 billion square meters)."
  - As supply factors fade and fundamental decline eases, average new starts in the early 2030s rebound to "about 52 percent of the 2019-2021 peak pace" before declining thereafter.
- Real estate GFCF model and projections:
  - Growth model: Growth of real estate GFCFt = β1 × Growth of Floor Space Startst + β2 × Growth of Real Estate GFCFt-1, with coefficients constrained to sum to one.
  - Regression results: β1 = "0.3383" (Standard Error .0528), β2 = "0.6616" (Standard Error .0528).
  - Real estate GFCF projected to decline "roughly 30-60 percent from the end-2022 level by the middle of the 2020s."
  - In the central scenario, real estate GFCF "is expected to fall to about 50 percent of its 2021 peak level in the mid-2020s" and rebounds to "60 percent of its 2021 peak level in the mid-2030s."
- Macro linkages and amplification:
  - Final demand from real estate-related activity accounted for "roughly 20 percent of the economy’s total value-added as of 2020."
  - About two thirds comes from construction/upstream linkages; another third from real estate services tied to existing stock.
  - Additional amplification: falling demand for housing-related goods (furniture, appliances) and wealth effects.

### G. Province-level heterogeneity and adjustments
- Method and fit:
  - Urban population and household-size changes scaled by province contribution in 2017-2021; replacement demand based on pre-2000 housing shares and 2010-2020 pace.
  - Sum of province-level projections differs negligibly from national: province vs national differ by "1.3 and 0.2 percent" for the two fundamental demand components.
- Province outcomes (median and distribution):
  - Median province: fundamental demand for 2024-2033 (urban household formation and replacement only) falls by "46" (percent implied) from 2012-2021, with most provinces seeing declines of "30-55 percent."
  - Factoring supply-side drag, median province projected starts fall "52 percent from their 2019-2021 average pace."
  - Median province faces an average supply-side overhang effect equivalent to "roughly 10 percent of the average pace of starts in the 2019-2021 period."
  - One quarter of provinces face overhangs equivalent to "35 percent or more" of recent pace, bringing projected supply to "30 percent or less of the 2019-2021 pace"; these provinces account for "only 13 percent of nationwide starts in the 2019-2021 period."

### H. Risks, transmission channels, and potential amplifiers
- Key risk channels:
  - Completion and counterparty risk from presales reduces marketable value of unfinished properties.
  - Liquidation of distressed inventories at discounts could exert downward pressure on broader prices.
  - Structural reduction in speculative/investment demand increases effective supply.
  - Worsening developer distress could strain local government finances, potentially tightening regional financial conditions and prompting fiscal consolidation.
- Downside scenario:
  - If actual sales fall below projected fundamental demand, adjustment could be "deeper and more protracted": inventories run down more slowly or rise, developers cut starts further, causing negative spillovers to construction and upstream industries.

### I. Policy implications and recommendations (property market)
- Priority objective: accelerate reduction of excess inventories and restructuring of the developer sector to avoid prolonged procyclical feedback loops.
- Accelerating exit of nonviable developers:
  - Phase out forbearance policies for property lending.
  - Supervisors to guide lenders to assess viability conservatively, recognize losses, and initiate insolvency where necessary.
  - Use corporate restructuring and insolvency regimes more, with strengthened liquidation and out-of-court settlements.
- Supporting housing completion:
  - For commercially nonviable projects with unfinished presold housing, use a central-government-backed support scheme for homebuyers awaiting delivery (complete projects or provide partial compensation, whichever is less costly).
- Reforming the presale model and de-risking viable developers:
  - Prohibit use of presale funding as general liquidity.
  - Backstop presale funding with stronger financial/legal protections (stricter escrow rules, third-party completion insurance).
  - Consider government-backed completion guarantees with conservative underwriting.
  - Guide surviving firms to use mergers, asset disposals, equity raising, and debt restructuring to rebuild capital and liquidity.
- Public housing and demand-side measures:
  - Step up public and rental housing construction to meet lower-income demand and repurpose private inventories, recognizing local fiscal constraints.
- Price and land market adjustments:
  - Allow market-based adjustments in house and land prices; avoid using macroprudential and other policies to impede necessary price corrections.
- Measures to shift household incentives away from property:
  - Introduce a nationwide property tax.
  - Introduce alternative saving options (third pillar pensions, voluntary supplementary medical insurance).
- Fiscal reforms to reduce LG reliance on land sales:
  - Tax reforms to generate new local revenues; increase central-to-local revenue shares where needed; make transfers automatically responsive to local conditions; phase out provincial growth targets that encourage excessive investment financing.

### J. LGFVs, local government fiscal risks, and recommended framework changes
- LGFV sector size and exposures:
  - LGFVs: interest-bearing debt about "50 percent of GDP."
  - Total intersectoral liabilities and equity of LGFVs: "111 percent of GDP."
  - Direct claims on physical assets estimated "56 percent of total LGFV assets."
  - Central Government net financial assets: "47 percent of GDP"; Local Governments net financial assets: "42 percent of GDP"; Households net financial assets: "187 percent of GDP."
- Shocks and simulation findings:
  - Stylized land shock (prices/activity "20 percent lower during three years"):
    - Decline in LG land sale revenue lowers LG financial net worth by "3.3 percent of GDP."
    - Decline in market value of LGFV real estate assets lowers LGFV debt and equity liabilities by "0.4 and 5.1 percent of GDP," respectively.
    - After equity spillovers, ultimate loss absorption concentrates nearly all losses on LGs absent policy reallocation.
- Fiscal reaction and constraints:
  - Empirical result: each additional percentage point of official debt-to-GDP of an LG is associated with an increase by "0.3 percentage points" in the general budget primary balance the following period.
  - Panel regression highlights: L.Offical Debt coefficient "0.30**"; D. Health Exp. coefficient "-2.78***"; L.LGFV spread (AA) coefficient "-1.61***".
- Policy options for LGFV distress:
  - Restructuring via insolvency and write-downs can reduce economy-wide leverage and force creditors (including banks and investors) to absorb losses, improving market discipline.
  - Alternatives (LG bailout, CG bailout, use of SOEs/lenders to absorb debt) shift burdens differently and carry risks (debt sustainability, moral hazard).
  - Staff-simulated loss allocation from face-value reductions: distributions cited as "30 28 42" labeled "Local Governments Central Government Other Sectors" (percent shares).
- Complementary reforms and safeguards:
  - Multiyear fiscal consolidation plans for affected LGs with CG monitoring; one-off targeted CG transfers to sustain public services; support for vulnerable households.
  - Strengthen bank restructuring/resolution frameworks; consider temporary financial resolution fund; coordinate sectoral restructurings.
  - Upgrade LG fiscal reporting and transparency (GFSM and IPSAS alignment); create public registry of public-sector company debt issuance; require timely audited LG financial statements.
  - Include LGFVs in official budgets once their debt has been restructured.

### K. Tax and revenue mobilization priorities linked to property and LG fiscal reform
- SIP revenue mobilization strategy highlights:
  - Package estimated to raise tax-to-GDP ratio by "5-6 percentage points over the next 5 years," with about half accruing to local governments.
- Priority measures with estimated yields (percent of GDP by year 5) and local government (LG) share:
  - Personal Income Tax — lower thresholds, increase rates: "2.7"; LG share "40 percent."
  - Social Security Contributions — reduce contributions: "-1.3"; LG share "100 percent."
  - Capital income tax — raise rate to "25 percent", reduce mortgage deductibility: "0.3"; LG share "40 percent."
  - Wealth Taxation — property, estate, gift taxes: "0.9"; LG share "100 percent."
  - Withdraw Covid measures: "1.5"; LG share "50 percent."
  - VAT — reduce number of rates and exemptions: "0.4"; LG share "50 percent."
  - Carbon pricing — expand ETS pricing: "0.6"; LG share "TBD."
  - Corporate Tax reforms and excise tax increases: each "0.4"; LG share varies.
- Property tax reform specifics:
  - No comprehensive nationwide recurrent property tax currently; recurrent property taxes typically yield around "0.7 percent of GDP in the Asia Pacific region."
  - Introducing a nationwide real property tax combined with reduced income-based housing concessions recommended to improve efficiency and provide stable local revenue.
  - Property tax reform is politically sensitive and requires strong political will, administrative planning (valuation and record keeping), and gradual implementation to restrain one-off wealth losses to current owners.
- Fiscal transfers and intergovernmental design:
  - Shifting revenue from SSCs to PIT implies a shift from local to central revenues; offsets include reallocating PIT shares or introducing a nationwide real property tax accruing to local governments.

*Source: IMF staff chapter on the People’s Republic of China property market (content unit: 1chnea2024002).*

### References _____________________________________________________________________________ 18

### SMOOTHING THE PATH TO A NEW NORMAL: CHINA’S PROPERTY SECTOR TRANSITION

### A. Introduction and key macro-financial context
- Property-related activities accounted for an estimated 20 percent of GDP through China’s decades of rapid growth.
- Average sales prices rose almost 350 percent in the 15 years through 2021.
- The authorities introduced the “Three Red Lines” in August 2020, which imposed regulatory limitations on developers with: liability-to-asset ratios above 70 percent; net debt higher than 100 percent of equity; and cash less than 100 percent of short-term debt. The rules were set to take effect in mid-2023.
- After the default of the second largest developer by sales in late 2021, liquidity stress spread through large segments of the developer sector as new presales—relied on as a key form of working capital—slowed sharply.
- Authorities eased demand-side policies (relaxing downpayment requirements, easing mortgage restrictions, lowering rates for existing mortgages) but avoided large-scale bailouts to defaulting developers.
- Real property investment is likely to fall by 30 to 60 percent from end-2022 levels and rebound only modestly thereafter.
- Policy priorities emphasized in the text:
  - Expedite the resolution of underlying supply-side imbalances, most importantly by restructuring nonviable developers.
  - Support and de-risk surviving developers.
  - Take steps to contain the buildup of risks in the property market.
  - Reform the pre-sales system.

### B. Drivers of the medium-term decline in new housing investment
- Demographics and household formation
  - Growth in China’s urban population is set to decelerate as net migration to urban areas slows and total population declines.
  - Declining average household size could partially offset population drag on housing demand.
- Publicly funded redevelopment
  - Shantytown redevelopment accelerated demolition of older housing, particularly in 2013-2018.
  - Share of population living in housing built before 2000 fell from 66 percent in 2010 to 35 percent in 2020 (implying net destruction of 66 million pre-2000 housing units).
  - Share of housing units without modern amenities fell from 28 to 12 percent (2010 to 2020).
  - Targets for urban redevelopment programs were scaled down almost 75 percent for the 14th Five Year Plan (2021-25) from the 13th Five Year Plan (2016-2020).
  - Near-term scaling up of redevelopment is constrained by fiscal costs for local governments and a smaller remaining stock of pre-modern housing.
- Supply-side imbalances
  - Rapid pre-2021 building and sharp decline in sales since mid-2021 left developers with large inventories of finished and unfinished housing.
  - Large share of unoccupied investment properties may enter the secondary market or be reallocated within families, reducing new construction needs.
  - Developer liquidity stress will weigh on new construction while authorities prioritize completion of large backlogs of unfinished presale homes.
- Prices and market distortions
  - Interventions to limit house price declines have reduced market-clearing flexibility and appear to have exacerbated declines in sales.
  - Price evolution is uncertain: mortgage credit and household income trends support prices, while completion risk, liquidation of distressed inventories, structural reduction in speculative demand, and weaker income growth create downside risk.

### C. Projection framework and scenarios for fundamental demand
- Methodology
  - Projections follow Chivakul et al (2015): developer supply responds to expected fundamental demand, then is modified by supply-side distortions. Prices are discussed qualitatively (not factored quantitatively).
  - Fundamental demand equals additional units needed to accommodate projected household growth and replace net lost units from the housing stock.
- Two scenarios for average household size (both use UN medium fertility population projections and World Bank urbanization slowdowns):
  - Lower average household size scenario
    - Average persons per household falls from 2.77 in 2021 to 2.4 in 2034.
    - Replacement demand derived by extending 2010–2020 decline in pre-2000 housing stock but starting from roughly 2.8 million units per year.
    - Average new-build apartment size grows from 114 square meters in 2020 to 121 square meters.
    - Rationale: household size approaches levels similar to Japan or Korea; stronger redevelopment policy and income-driven demand for larger housing.
  - Higher average household size scenario
    - Average household size falls to only 2.6.
    - Replacement demand occurs at two-thirds the annual pace of the upside scenario, generating demand for only 2.1 million units of housing per year over a decade.
    - Average apartment size remains unchanged.
    - Rationale: greater drag on net household formation from persistent youth unemployment, weak consumer confidence, and continued cohabitation.

### D. Risks, transmission channels, and potential amplifiers
- Completion risk and counterparty risk introduced by widespread pre-sales reduces marketable value of unfinished properties.
- Liquidation of distressed developer inventories at a discount could exert downward pressure on broader house prices.
- Structural reduction in speculative or investment demand (including reduction of vacancy rates via sales of investor-held vacant units) increases effective supply relative to income.
- Worsening developer distress could strain local government finances, raising the likelihood of tighter regional financial conditions and future fiscal consolidation.

### E. Policy implications emphasized in the chapter
- Expedite restructuring of nonviable developers to resolve supply-side imbalances and reduce uncertainty stemming from unfinished presales.
- Support and de-risk surviving developers with conservative business models and better governance to enable orderly completion of projects.
- Reform the presales system to reduce completion/counterparty risk embedded in homebuying transactions.
- Step up publicly supported construction of public and rental housing (recently signaled as intensified support), while recognizing fiscal constraints on local governments.
- Balance trade-offs of price-stabilizing interventions: measure against growth costs of a sharp price adjustment and wider spillovers to household and financial sector balance sheets.

*Prepared by Henry Hoyle (APD).*

### 15.      These estimates project fundamental demand in the coming years will decline roughly

### 1chnea2024002 - 15.      These estimates project fundamental demand in the coming years will decline roughly

### Fundamental demand projections
- Projected decline in fundamental demand in the coming years is "roughly in the range of around 35-55 percent relative to the past decade, to an average of about 950 million square meters per year."
- Replacement demand provides the largest overall compression, "given the rapid pace of demolition in the 2013-2018."
- The decline in average household size accounts for the largest difference between the upside and downside scenario, reflecting its importance for net household formation and uncertainty in its future path.
- These fundamental demand estimates "are broadly in line with recent estimates by private analysts in China." (Examples cited: Wu and Xu (2021) 43 percent decline; Xia and Xu (2022) 996 million square meters through 2035; Song and Zhang (2022) 960 million square meters in 2026-2030.)

### Housing supply assumptions and supply-side drags
- Developers’ supply function:
  - Supply (starts, floor space terms) is assumed to equal expected fundamental demand 1.5 years in the future, reflecting a three-year production cycle with pre-sales allowed one year into production and developers aiming to complete pre-sales before end of year two.
- Supply-side modules layered onto baseline supply:
  - Excess housing inventories: "Excess inventories are calculated as the cumulative sum of lagged 1.5 year starts less sales." Developers reduce new starts when realized sales fall below projected demand; inventories are cleared over time by assumed reduction in starts.
  - New supply from investor sales: A conservative estimate of "5 percent of cumulative 2017-2021 sales" is used to reflect increased secondary market supply from investors reducing holdings of unoccupied investment properties.
  - Presale completion backlogs: Developers facing liquidity stress reduce new starts to complete overdue presold housing. The developer sector is assumed to face the trade-off for "one quarter the estimated presale completion backlog attributed to liquidity-constrained developers, or ten percent of the total."
- Combined drag from supply-side factors: "equivalent to about 17 percent of projected fundamental demand in the 2024-2033 period."
  - Excess inventories as of end-2023: "just over 1 billion square meters in floor space terms."
  - New supply from investor sales: "377 million square meters."
  - Reduction in new starts related to distressed developer backlogs: "535 million square meters."
- Timing assumption for drag: supply-side drags are "assumed to be concentrated in the first three years, and then gradually dissipate over the following four years."

### Projected starts, real estate GFCF, and macro linkages
- New starts:
  - New starts in the ten years starting in 2024 average about "715 million square meters in the central scenario," about "45 percent of the 2019-2021 average level (1.6 billion square meters)."
  - The drag from supply-side factors generates about one-quarter of this decline; declines in fundamental demand and other factors contribute the remainder.
  - As supply factors fade and fundamental decline falls further in the first five years of the 2030s, average new starts would rebound to "only about 52 percent of the 2019-2021 peak pace before starting to decline."
- Cross-country context:
  - Among thirteen advanced economies with relevant data, five experienced declines in housing starts of over 50 percent in three years, "roughly matching the 60 percent three-year decline in China from 2021-2023."
  - In those five countries, housing starts averaged "26 percent of peak levels over the subsequent five-year window."
- Real estate gross fixed capital formation (GFCF):
  - Growth model: Growth of real estate GFCFt = β1 × Growth of Floor Space Startst + β2 × Growth of Real Estate GFCFt-1, with coefficients constrained to sum to one.
  - Regression results (Text Table 1): β1 = "0.3383" (Standard Error .0528), β2 = "0.6616" (Standard Error .0528).
  - Projected path: Real estate GFCF projected to decline "roughly 30-60 percent from the end-2022 level by the middle of the 2020s."
  - In the central scenario, real estate GFCF "is expected to fall to about 50 percent of its 2021 peak level in the mid-2020s" and begins to rebound in 2028, reaching "60 percent of its 2021 peak level in the mid-2030s."
  - Comparable experience: In countries with a three-year decline in starts exceeding 50 percent, real estate GFCF averaged "57 percent of peak levels in the ten years after peak," compared to "58 percent estimated over the same period in China" (average of two household size scenarios).
- Macro linkages:
  - Final demand from real estate-related activity accounted for "roughly 20 percent of the economy’s total value-added as of 2020."
  - About two thirds of that comes from "real estate’s imputed portion of construction activity" and upstream linkages; another third comes from "real estate services" connected to the existing housing stock.
  - Additional amplification channels: declining demand for housing-related goods (furniture, appliances) and wealth effects.

### Province-level adjustments and heterogeneity
- Method:
  - Urban population growth for each province is scaled to the national projection by province contribution to urbanization growth in 2017-2021.
  - Change in average household size is scaled similarly by province 2017-2021 change.
  - Replacement demand calculated per province based on share of pre-2000 housing in 2020 stock and pace of reduction in 2010-2020.
  - The sum of province-level projections differs negligibly from national projections (province vs national differ by "1.3 and 0.2 percent" for urban household formation- and replacement-driven fundamental demand).
- Province-level outcomes:
  - Median province: fundamental demand for 2024-2033 (urban household formation and replacement only) falls by "46" from the 2012-2021 period, with most provinces seeing declines of "30-55 percent."
  - Factoring supply-side drag, average annual projected housing starts in 2024-2033 are likely to fall "52 percent from their 2019-2021 average pace in the median province."
  - Distribution of supply-side overhang: median province faces an average supply-side overhang effect equivalent to "roughly 10 percent of the average pace of starts in the 2019-2021 period."
  - For one quarter of provinces, the effect is "equivalent to 35 percent or more of the recent pace of building activity," bringing projected supply to "30 percent or less of the 2019-2021 pace." This segment accounts for "only 13 percent of nationwide starts in the 2019-2021 period" and tends to be among more fiscally vulnerable provinces based on official LG debt to GDP ratio.

### Policy implications and recommendations
- Tailor housing market policies to localized supply and demand projections.
  - For provinces with relatively limited net fundamental demand in 2024-2033, "large-scale stimulus to support new housing construction should be avoided in favor of policies that make better use of the existing housing stock."
  - Local supply and demand considerations should inform resolution of unfinished presold projects of financially distressed developers.
  - For projects with significant unsold housing in weaker-demand markets, "partial monetary compensation to homebuyers may be more effective than completing the project."

### Risks and scenarios
- Downside risk: If actual sales fall below projected fundamental demand, the adjustment could be "deeper and more protracted."
  - In that case, developers’ excess inventories or investor secondary market supply "will be run down more slowly than projected or could even rise."
  - Developers would respond by cutting starts further, generating negative spillovers to construction and upstream industries.
- Scenario ranges and key reference figures:
  - Fundamental demand range: "around 35-55 percent" decline vs past decade; average "about 950 million square meters per year."
  - Central scenario new starts: "715 million square meters" average 2024-2033.
  - 2019-2021 average starts (reference peak): "1.6 billion square meters."
  - Combined supply-side drag: "about 17 percent of projected fundamental demand in the 2024-2033 period."
  - Excess inventories end-2023: "just over 1 billion square meters."
  - Investor-sales-induced new supply: "377 million square meters."
  - Distressed developer backlog reduction in starts: "535 million square meters."
  - Real estate GFCF projected falls: "roughly 30-60 percent from the end-2022 level by the middle of the 2020s."
  - Real estate GFCF central scenario: fall to "about 50 percent of its 2021 peak level in the mid-2020s"; rebound to "60 percent of its 2021 peak level in the mid-2030s."
  - Real estate-related final demand share: "roughly 20 percent of the economy’s total value-added as of 2020."
  - Province-level median fundamental demand decline: "46" (percent implied in context) with most provinces "30-55 percent" declines; median province projected starts fall "52 percent" vs 2019-2021.

*PEOPLE’S REPUBLIC OF CHINA — INTERNATIONAL MONETARY FUND (excerpt).*

### 30.      The controlled and slow pace of house price adjustment is one factor likely to suppress

### 30. The controlled and slow pace of house price adjustment is one factor likely to suppress sales relative to fundamental demand, prolonging the property market contraction

### Slow house price adjustment and market clearing
- Controlled and slow pace of house price adjustment likely suppresses sales relative to fundamental demand and prolongs the property market contraction.
- Limited price adjustment appears to partially reflect efforts by LGs to limit the pace of price declines.
- Price declines generally play an important role in clearing the housing market and restoring equilibrium between supply and demand.
- Drawn-out price adjustments could:
  - Reinforce supply pressures and entrench expectations for persistent house price declines.
  - Weaken actual sales relative to fundamental demand if homebuyers expect gradual further price declines.
  - Undermine restructuring of developers’ balance sheets by prolonging the imbalance between supply and demand.

### Developer distress and its effects on sales
- The share of defaulted or distressed private developers continued to grow through mid-2023, surpassing 40 percent of the sector in 2020 market share terms.
- Restructuring has largely been delayed amid widespread use of forbearance to encourage completion of large backlogs of unfinished presold housing, resulting in continuing deterioration of distressed developers’ financial positions.
- Selected balance sheet items of distressed developers (Billions of renminbi):
  - Presale Liabilities: -12.7%
  - Net Debt: 2.9%
  - Equity: -57.9%
  - Cash: -42.7%
- Notes: Includes 24 distressed developers with available financial statements for 2022. Distressed developers have either defaulted within last two years or have average bond prices below 40.
- The decline in presale liabilities in 2022 (text chart) proxies the slow pace of delivery of presold housing and reinforces homebuyer caution for unfinished homes except from the strongest developers (for instance those with strong central government backstops).
- Uncertainty over resolution of troubled developers’ large stock of unfinished properties, residential land, and arrears to suppliers may further depress sales relative to fundamental levels.

### Interlinkages with local government finances
- Local governments have long relied on land sales to property developers and real estate-related taxes to cover significant structural fiscal deficits.
- Widespread liquidity distress in the developer sector has resulted in a sharp decline in land revenues and likely in land valuations, exacerbating on- and off-budget financial difficulties for many local governments and contributing to reports of rising local government and LGFV arrears.
- Cross-province evidence suggests land sales declines in 2022 were most acute in provinces with pre-existing fiscal weaknesses proxied by provincial government debt-to-GDP ratios at end-2021, suggesting a potential negative feedback loop between declining property activity and fiscal vulnerabilities.
- Province-level regression reported in the text:
  - y = -0.3143x + 20.693
  - R² = 0.2499

### Public housing production: benefits and trade-offs
- The authorities’ plan to increase production of public housing would:
  - Be broadly beneficial to construction activity and help offset decline in construction sector activity.
  - Provide more affordable alternatives to the private market for relatively lower-income household segments.
  - Absorb fundamental demand for new housing units, shrinking the equilibrium supply growth needed in the private market and potentially delaying resolution of oversupply in some regions.
- The impact could be partially mitigated if developers received new revenue streams from overseeing the production of public housing.

### Policies for smoothing the transition (priority: expedite resolution of supply-side imbalances)
- Policy objective: Accelerated reduction of excess inventories and restructuring of the property developer sector to avoid prolonged procyclical feedback loops involving house prices, developer balance sheets, and local government fiscal vulnerabilities.

- Accelerating exit of nonviable property developers:
  - Phase out forbearance policies for property lending.
  - Supervisors should guide lenders to adopt a conservative approach in assessing developer viability and collateral values and require banks to recognize losses and initiate insolvency proceedings as necessary.
  - Use the existing corporate restructuring and insolvency regime more, with strengthened liquidation proceedings and out-of-court settlements.

- Supporting housing completion:
  - For commercially nonviable projects with unfinished presold housing units, projects should be taken over by a central-government-backed support scheme for homebuyers awaiting delivery (as described in Box 1, IMF 2022).
  - The scheme would either complete taken-over projects or provide partial compensation to affected homebuyers, whichever is less costly.

- Allowing market-based adjustments in house and land prices:
  - Macroprudential and other housing and development policies should not be used to impede market-based price adjustments, which are needed to restore stability and confidence in regions with excess supply.
  - Cross-country evidence shows faster home price depreciation episodes are associated with stronger growth in GDP and productivity compared to modest but prolonged home price declines (Jinjarak et al, 2016).

### Policies to right-size and de-risk viable developers
- Reforming the presale model:
  - Strictly prohibit developers’ use of presale funding as a general source of liquidity.
  - Backstop presale funding with stronger financial and legal protections for presale homebuyers, for instance by introducing stricter escrow rules and third-party completion insurance.
  - Consider government-backstopped completion guarantees backed by conservative underwriting to help restore homebuyer and creditor confidence.

- Expanding housing access:
  - Boost policy support for public and rental housing construction to meet fundamental demand for lower-income segments and reduce the need for household savings to purchase private housing at elevated valuations.
  - Design public housing programs to re-purpose private inventories and provide alternative revenue streams for developers.

- Assisting surviving developers to repair balance sheets:
  - Guide surviving firms to speed up use of mergers, asset disposals, equity raising, and other tools (including debt restructuring) to boost capital and liquidity buffers.

### Policies to address fundamental incentives and fiscal reliance on property
- Guide households towards alternative investment options:
  - Introduce a new nationwide property tax.
  - Introduce alternative saving options such as “third pillar” pensions and a voluntary supplementary medical insurance plan to reduce investment-motivated demand for housing.

- Fiscal reforms to reduce local governments’ reliance on land sales and property activity:
  - Implement tax reforms laid out in the accompanying SIP on revenue measures to generate new revenue for local governments.
  - Increase the share of revenue allocated from the central to the local government, where needed, to be commensurate with local governments’ spending mandates and reduce vertical fiscal imbalances.
  - Make transfers to local governments automatically responsive to local economic conditions to attenuate the need for off-budget financing.
  - Phase out provinces’ growth targets that encourage excessive investment financing and growing indebtedness at the local level.

*Source: IMF staff chapter on the People’s Republic of China property market (content unit: 1chnea2024002).*

### 4.      This SIP proposes a revenue mobilization strategy to achieve fiscal consolidation while

### 1chnea2024002 - 4.      This SIP proposes a revenue mobilization strategy to achieve fiscal consolidation while

### Overview
- The SIP proposes a revenue mobilization strategy to achieve fiscal consolidation while contributing to a more sustainable and inclusive growth path.
- The paper presents a menu of tax policy reforms China could progressively implement over the coming years, with estimates of revenue potential (by level of government) and assessments of accompanying benefits, drawing on international experience and previous IMF technical assistance reports on China.
- The suggested reforms focus on: increasing reliance on personal income taxation; reducing tax-favored status of owner-occupied housing; leveraging international tax reform; and improving the efficiency of indirect taxation.

### Key reform areas and high-level expected impact
- Short-term priorities:
  - Raise capital taxation.
  - Reverse untargeted pandemic relief measures.
  - Leverage ongoing international tax reform.
- Medium-term and longer-term measures:
  - Move to specific and higher excise tax rates.
  - Adjust the PIT schedule to raise average rates while reducing SSCs.
  - Introduce a nationwide property tax.
  - Reduce the number of VAT rates and exemptions.
  - Increase revenue from carbon emission permits.
- Aggregate impact:
  - The package of reforms is estimated to raise the tax-to-GDP ratio by 5-6 percentage points over the next 5 years, with about half that amount accruing to local governments.
- Ancillary benefits:
  - Increase economic efficiency.
  - Reduce inequality.
  - Improve cyclical stabilization properties of the tax system.
  - Support private consumption and have positive spillovers to health and climate outcomes.

### Estimated yields by measure (summary of Text Table 1)
- Personal Income Tax — lower PIT thresholds, increase PIT rates: 2.7 (% GDP by year 5); LG share 40 (%)
- Social Security Contributions — reduce social security contributions: -1.3 (% GDP by year 5); LG share 100 (%)
- Capital income tax — raise tax rate to 25%, reduce mortgage deductibility: 0.3 (% GDP by year 5); LG share 40 (%)
- Wealth Taxation — introduce property tax, estate and gift taxes: 0.9 (% GDP by year 5); LG share 100 (%)
- Withdraw Covid measures: 1.5 (% GDP by year 5); LG share 50 (%)
- VAT — reduce number of rates and exemptions: 0.4 (% GDP by year 5); LG share 50 (%)
- Carbon pricing — expand ETS pricing: 0.6 (% GDP by year 5); LG share TBD
- Corporate Tax — streamline incentive regime, introduce minimum tax in line with Pillar 2, introduce excess profit tax: 0.4 (% GDP by year 5); LG share 40 (%)
- Excise Taxes — increase specific rates on tobacco, alcohol, and fuel: 0.4 (% GDP by year 5); LG share 100 (%)
- Notes: Revenue estimates are approximate and based on international comparisons. Sequencing rank is lower for measures that can be implemented sooner.

### Labor income taxation: features, problems, and reform scenarios
Findings on current system
- China’s PIT is dual structured: progressive schedules on labor income and a lower flat rate on capital.
- PIT revenue accounts for only 8.4 percent of total tax revenue in China; the share is 15 percent for China’s synthetic control, and 37.5 percent among other comparators.
- The large standard deduction of RMB 60,000 a year effectively removes PIT liability for about 70 percent of the population.
- Aggregate effective PIT tax rate is less than 5.5 percent, the lowest among comparators.
- China’s redistributive capacity (difference in pre- and post-tax Gini) is weaker than most comparators except Russia.
Social security and wedge statistics
- Top combined employer and employee contribution amounts to around 38.5 percent of labor cost.
- G20 contribution rates are around 30 percent.
- Weighted average of Indian and Russian contribution rates (the synthetic control) is just 27 percent.
- Example (Shanghai): for an employee earning RMB 65,000, the combined marginal government take of an additional yuan is almost 40 percent; for an employee earning RMB 450,000 the marginal government take is reduced to 22 percent (due to SSC caps at roughly 300 percent of a region’s average wage).

Reform options and scenarios (Text figure 4 summaries)
- Current system (as presented)
  - First threshold (rate): 60,000 (3 percent)
  - Top threshold (rate): 1,020,000 (45 percent)
  - Increase cap for SSC by: -
  - SSC rate in percent: 37
  - Redistributive capacity: 2.2
  - PIT revenue in percent of GDP: 1.2
  - Combined revenue (PIT/SSC) in percent of GDP: 9.5
- Revenue neutral reform (Column 2)
  - Key changes: reduce SSCs by 3 percentage points, increase their cap by 25 percent, halve the standard deduction, increase the marginal tax rate in the first bracket to 5 percent, and reduce thresholds including top threshold to 750,000.
  - First threshold (rate): 30,000 (5 percent)
  - Top threshold (rate): 750,000 (45 percent)
  - Increase cap for SSC by: 25 percent
  - SSC rate in percent: 34
  - Redistributive capacity: 3.2
  - PIT revenue in percent of GDP: 2
  - Combined revenue (PIT/SSC) in percent of GDP: 9.5
  - Effect: Increase net income for the bottom 90 percent of the income distribution.
- Revenue raising reform (Column 3)
  - Key changes: cut SSCs to 35 percent, increase SSC cap by 30 percent, reduce tax rate thresholds more aggressively.
  - First threshold (rate): 15,000 (10 percent)
  - Top threshold (rate): 500,000 (45 percent)
  - Increase cap for SSC by: 30 percent
  - SSC rate in percent: 35
  - Redistributive capacity: 5
  - PIT revenue in percent of GDP: 3.9
  - Combined revenue (PIT/SSC) in percent of GDP: 10.9
  - Effect: Increase revenue by 1.4 percent of GDP by increasing the tax burden on the top 40 percent of the income distribution.

Sequencing and intergovernmental effects
- SSCs fully accrue to local governments; only 40 percent of PIT collections accrue to local governments.
- Shifting from SSCs to PIT implies a shift from local to central government revenues.
- Possible offsets: change PIT revenue allocation, introduce a nationwide real property tax that accrues to local governments, or allow provinces to impose additional payroll taxes within centrally set bands.
Reform of deductions
- Current deductions (child education and care, major medical expenses, elderly care, mortgage interest and rent) reduce liabilities in proportion to marginal tax rates and thus benefit higher-income taxpayers more.
- Recommendation: reverse recent expansion of deductions; consider scaling back or abolishing them; convert high-social-value deductions into refundable tax credits to benefit all income groups.

### Capital income taxation: findings and reform options
Findings
- Capital income taxation is low by international comparison and contrasts with China’s high top marginal labor income tax rate.
- Capital income is taxed at up to 20 percent in China.
- Several exemptions reduce effective tax on capital income and create distortions, e.g. exemptions of interest from deposit and savings accounts with Chinese banks and from government and corporate bonds for natural persons; dividends and stock gains from listed companies are exempt under certain conditions.
- Text Table 2: Top tax rates applying on labor income and selected types of capital income (maximum tax rates applicable to residents)
  - Capital income columns: Interest Income, Dividends, Short-term Capital Gains, Rental Income; Labor Income column shows top labor rate.
  - China: 20 20 20 20 45
  - Brazil: 22.5 0 22.5 27.5 27.5
  - India: 30 30 30 30 30
  - Indonesia: 20 10 35 10 35
  - Russia: 15 9 15 15 15
  - USA: 37 37 37 37 37

Reform options
- Quantify costs and benefits of existing tax concessions for specific capital income types (e.g., publish Tax Expenditure Reports).
- Increase tax on capital income to curb tax planning, increase effective taxation of high-income individuals, reduce wedge between labor and capital returns, and facilitate taxation of small businesses.

### Property and wealth taxation: findings and reform options
Findings
- No comprehensive nationwide recurrent property tax; instead, transaction-based taxes vary by location, property type, and transaction value, and a progressive Land Appreciation Tax applies to property transfers.
- Preferential tax treatment of owner-occupied housing contributed to the property sector boom via:
  - Mortgage interest deductibility and imputed rental income not taxable.
  - Tax-exempt contributions to the housing provident fund: mandatory contributions deductible from PIT, fund investment income exempt from capital income taxes, and no tax on distributions; marginal subsidy ranges up to 45 percent (the top marginal income tax rate).
  - Limited taxation of capital gains: capital gains on owner-occupied homes are exempt from PIT if the property has been the only dwelling for the family for five years.
- The lower tax burden on owner-occupied housing reduces efficiency of capital allocation and financial market development and inflates aggregate demand for real estate relative to a neutral tax system.

Reform options
- Introduce a nationwide real property tax in combination with reduced income-based tax incentives.
  - Rationale: recurrent property taxes are efficient (land and buildings are relatively immobile) and, with a carefully chosen exempt threshold, progressive and stable.
  - Evidence: recurrent property taxes yield around 0.7 percent of GDP in the Asia Pacific region.
- Increased property revenue could allow reduction in reliance on distortive transfer taxes and, combined with reforms of income-based concessions, increase efficiency of capital allocation by reducing differences in effective tax rates across assets.

### Indirect taxation, corporate tax, excises, and carbon pricing (summary)
- VAT: remains the most important source of tax revenue; new draft VAT law (released end-2022) includes improvements but further reforms are necessary to reduce distortions and administrative costs; scope to reduce number of VAT rates and exemptions.
- Excise taxation: scope to move to specific and higher excise tax rates on items such as tobacco, alcohol, and fuel.
- Carbon pricing: increase revenue from carbon emission permits by expanding ETS pricing; estimated yield 0.6 (% GDP by year 5).
- Corporate income tax: relatively efficient by international standards but tax incentives for certain companies/activities can distort competition and complicate administration; the Inclusive Framework’s international tax reform is an opportunity to review incentives and implement a domestic minimum tax.

*Source: IMF staff estimates and analysis in the SIP chapter provided.*

### 22.      Property tax reform is politically sensitive, requiring gradualism and careful planning.

### 22.      Property tax reform is politically sensitive, requiring gradualism and careful planning.

### Property tax reform: challenges and implementation requirements
- Property taxes are highly visible and the link with benefits may be unclear to citizens.
- If a property tax is capitalized in property prices, present owners will suffer a one-off loss in wealth through a change in the price level.
- Comprehensive property tax reform often confronts opposition.
- Successful implementation requires:
  - strong political will;
  - detailed planning of administrative reform (including on valuation and record keeping);
  - gradual implementation to restrain losses of current owners.

### Estate (inheritance) taxation: potential and constraints
- Estate taxation can curb excessive wealth inequality.
- Real property is a major store of value, but the wealth of top-income individuals is typically held in corporate stocks, financial securities, or unincorporated business equity.
- Inheritance (or estate) taxation can help reduce income inequality even if it typically yields no more than 0.2 percent of GDP.
- Tax-free thresholds that exempt small inheritances can reduce the compliance burden and are equitable.
- Inheritance taxation requires even more planning than property taxation, leaving this project for the medium term.
- Inheritance taxation would need to be coupled with a gift tax to restrict avoidance.
- Implementation considerations:
  - administrative planning, public consultations, and comprehensive information campaigns are required to increase knowledge of and reduce resistance to the tax.
  - particular care is needed if taxes are levied on land and if commercial property is not included in the base.

### Property tax revenue context (regional comparison)
- Text Table 3. China: Property Taxes as a Percentage of GDP in Different Regions (In 2020)
  - OECD: 1.9
  - Africa: 0.3
  - Asia-Pacific: 0.7
  - Latin America: 0.8

### Leveraging international corporate income tax reform (context and implications)
- China’s CIT revenue has declined and is now below the averages of comparator groups despite a relatively efficient CIT system.
- The Inclusive Framework’s (IF) 2-Pillar framework will support revenue collection and change longstanding norms.
- Estimated direct effects on China’s CIT revenue (quantifications from Klemm and others (2021)):
  - Pillar 1 could increase China’s tax collections by around 0.4 percent of current CIT.
    - Several Chinese-owned multinationals in scope account for 8 percent of total global residual profit.
    - Overall revenue impact of Pillar 1 is limited.
  - Pillar 2 could yield up to 0.5 percent of current CIT revenues.
    - As the headquarters of some large multinationals, China may raise additional revenue from profits of these firms that are currently lightly taxed, depending on responses of companies and other countries.
- Given the statutory tax rate of 25 percent, a 15 percent global minimum rate would still preserve a positive tax differential in China relative to the rest of world, maintaining incentives for outward profit shifting.
- Policy implications and recommendations:
  - Use the reform moment to review the incentive regime and consider implementing a domestic minimum tax in line with the IF approach.
  - Evaluate options including redesign of tax incentives and implementation of a domestic minimum tax.
  - Equalizing the statutory taxation of all firms at 25 percent and reducing the generosity of selective tax base narrowing measures would increase competitiveness, support potential growth, and yield revenue.
  - An excess profit tax: a permanent well-designed excess profit tax would tax economic rents without identifying sectors; Hebous and others (2022) estimate the revenue potential of a non-distortive surcharge of 12.5 percent on China’s corporate rents at more than 0.5 percent of GDP.

### Improving indirect taxation: VAT, excises, and carbon pricing
- VAT
  - In 2022, VAT contributed close to 30 percent of China’s total tax revenue—or about 4 percent of GDP—while the average contribution among comparators is only 25 percent.
  - A slightly declining share of consumption in total GDP reduced VAT revenue by 2.5 percent of GDP since 2017.
  - The standard VAT rate of 13 percent is low in international comparison, supporting the transition to consumption-led growth.
  - Draft updated VAT law (released by end-2022) includes improvements:
    - enhance adoption of OECD international VAT guidelines by adopting the place of consumption approach;
    - allow refunds of excess input VAT credits;
    - convert VAT pilot programs and previous VAT/business tax rules into a single substantive law.
  - Remaining reform priorities that could raise revenue:
    - Remove several pandemic relief measures in the short term, which could raise about 1 percent of GDP.
    - Review treatment of the financial sector and residential property.
    - Two major design margins could improve VAT efficiency and revenue productivity, yielding around 0.4 percent of GDP:
      - Multiple rates: reduced from 6 to 4 but still too many by international standards; multiple rates complicate administration, raise compliance issues, and can lead to revenue losses through fraudulent VAT refund claims.
      - Exemptions: fragmented exemptions (agriculture, health, culture, education) reduce efficiency; agriculture should be fully taxed while equity considerations could be addressed through a limited number of exemptions for selected basic food items; fewer exemptions reduce tax avoidance and tax cascading.
- Excise taxes
  - Excise taxes in China account for about 1.5 percent of GDP.
  - Excises on goods with low price elasticities (e.g., alcohol, tobacco) can generate substantial revenue and improve health outcomes.
  - Increasing the specific tax on cigarettes until the retail price doubles would yield around 0.2 percent of GDP and improve health outcomes; currently excise taxes on the most-smoked brand account for just 36 percent of the retail sales price (above 40 percent among close comparators and above 52 percent among other comparators).
- Carbon pricing
  - Wider use of carbon pricing would raise revenue and contribute to climate change mitigation.
  - The existing Emissions Trading System (ETS) grants for the most part free permits to emitting firms.
  - Introducing partial auctioning of ETS permits and transitioning towards a full cap and trade system with wider sectoral coverage would improve efficiency, curb emissions, and raise fiscal revenue.
  - China’s potential for carbon revenue is about 0.6 percent of GDP annually by 2028 (IMF 2023b).

*Source: 1chnea2024002 - 22.      Property tax reform is politically sensitive, requiring gradualism and careful planning.*

### 2.      LGFVs have become significant sources of macro-financial risk, despite longstanding

### LGFVs have become significant sources of macro-financial risk, despite longstanding central government (CG) efforts to rein in their activities.

### Overview: structure, size, and governance weaknesses
- LGFVs are legally state-owned corporations, typically majority-controlled by local governments (LGs), but are not included in the budget and lack formal reporting or governance requirements.
- Close cooperation with LG-controlled banks has allowed LGFVs to accumulate heavy debt despite weak cash flows.
- Estimated sector size and exposures:
  - LGFVs: interest-bearing debt of about 50 percent of GDP.
  - Total intersectoral liabilities and equity of LGFVs: 111 percent of GDP.
  - LGFVs’ interlinkages on the asset side: totaling 50 percent of GDP.
  - Direct claims on physical assets (land, real estate, capital equipment) estimated to be 56 percent of total LGFV assets.
  - Machinery accounts for 4.5 percent of LGFV assets; analysis assumes roughly half of LGFVs’ remaining physical assets are land or assets deriving value from land.
  - Households: net financial assets of 187 percent of GDP.
  - Central Government (CG): net financial assets of 47 percent of GDP.
  - Local Governments (LGs): net financial assets of 42 percent of GDP.

### Shocks since 2020: pandemic and real estate slump
- Key developments and magnitudes:
  - Official LG debt: grew to 22 percent of GDP by end-2019.
  - LGFV debt increased by 22 percentage points of GDP between 2015 and 2019.
  - In response to the pandemic and real estate slump, official and LGFV debt issuance added another 18 percentage points of GDP in total debt in the four years through 2022.
  - Health expenditures (general budget of the general government) grew by 0.3 percentage points of GDP in 2022.
  - Mobility restrictions and tax relief contributed to revenue declines of 2.5 percentage points of GDP between 2019 and 2022.
  - Gross land transfer revenues and real estate related taxes declined by 23 percent in 2022; land-related revenue had accounted for nearly 40 percent of total local government revenues in 2020.
  - Land and real estate represent a significant portion of LGFV balance sheets, estimated conservatively at around 25 percent of total assets.
- Consequences:
  - Declines in land sale revenue and falling collateral values have widened LG financing needs and reduced LGFVs’ ability to obtain market-based credit.
  - LGFV spreads and credit market spreads for smaller LGFVs and those with weaker parent LGs widened significantly in 2023.
  - LGFV defaults on commercial bills and other payables remained relatively rare but increased significantly in 2023.
  - Cash shortages have led to reports of local government arrears and disruptions to public services.

### Intersectoral balance sheet approach and simulations
- Methodology and purpose:
  - Constructed estimated matrices of cross-sector debt and equity claims across government, households, nonfinancial corporations, financial institutions, LGs, LGFVs, and the rest of the world using public financial statements, official statistics, and other data.
  - Rows show claims of a given sector on every other sector; columns show each sector’s claim on a given sector.
  - Used to simulate impact of shock and policy scenarios and trace spillovers and loss absorption across sectors.
- Stylized land shock simulation:
  - Scenario: real estate prices and activity 20 percent lower during three years.
  - Direct effects:
    - Decline in LG land sale revenue lowers LG financial net worth by 3.3 percent of GDP.
    - Decline in market value of LGFV real estate assets lowers the value of LGFV debt and equity liabilities by 0.4 and 5.1 percent of GDP, respectively.
  - Transmission:
    - Immediate losses affect LGs, lenders, firms (via account payable exposures), and households (first round).
    - Losses translate into multiple rounds of equity spillovers (equity absorption).
    - Ultimate loss absorption concentrates nearly all losses on LGs after rounds of spillovers, unless reallocated by policy actions.

### Key findings from matrices and indicators
- LGFVs’ financial position and liquidity:
  - LGFVs show consistently negative aggregate operating cash flows adjusted for capitalized interest costs, implying limited capacity to deleverage or stabilize debt levels.
  - Aggregate sector cash fell for the first time in several years in 2022; LGFVs’ cash relative to total short-term liabilities fell further in 2022.
- Interconnectedness and spillover potential:
  - LGFVs hold significant claims on LGs and nonfinancial firms in the form of arrears and receivable exposures, and own significant equity in local firms and banks.
  - LGFVs have sizeable non-interest-bearing liabilities to local firms (largely account payables) and bond liabilities to households (exposed via asset management products).
  - LGFVs provide extensive guarantees for local firms, creating additional contingent exposures.
- Fiscal dynamics and risks:
  - Continued large deficits and rising debt servicing needs could crowd out primary spending and raise the risk of intensifying macro-financial spillovers and a disruptive adjustment.
  - A disorderly scenario could trigger fiscal tightening in weaker provinces and intensify negative macro-fiscal feedback loops.

### Policy options and recommendations (high-level synthesis from analysis)
- Restructuring options for LGFV debt:
  - Bailouts by LGs risk jeopardizing debt sustainability in weaker provinces and could induce unintended fiscal tightening.
  - CG bailouts could generate moral hazard.
  - Restructuring LGFVs using insolvency mechanisms is an option; debt write-downs would lead to losses for LGs and the CG but could shift a fraction of the burden to other sectors and lower overall cost if asset sales offset losses.
  - Debt write-downs should occur in parallel with strengthened measures to ensure fiscal sustainability and financial stability.
- Complementary measures needed alongside restructurings:
  - Temporary CG support as needed to limit systemic disruption.
  - Stronger insolvency frameworks and a comprehensive strategy for restructuring weak banks.
  - Broader fiscal reforms to limit future debt-generating flows, including:
    - Tax reforms (see accompanying SIP “A Revenue Mobilization Strategy”).
    - Upgraded transfer arrangements.
    - Strengthened transparency and monitoring.
- Authorities’ recent actions and intent:
  - Central authorities have required banks to avoid cutting credit to weak LGFVs to prevent near-term financial stress.
  - In October 2023, the Central Finance Work Conference signaled intent to comprehensively address local government debt risks via a “long-term local debt risk resolution mechanism” and called for optimizing the balance of central and local government debt.

*Prepared by IMF staff; source content from the specified chapter.*

### 16.      Indeed, higher LG government debt is estimated to create fiscal drag at the province

### Indeed, higher LG government debt is estimated to create fiscal drag at the province level

### Fiscal drag and LG fiscal reaction functions (2019-22)
- Panel data across provinces for 2019-22 is used to estimate fiscal reaction functions at the LG level.
- Key empirical result: each additional percentage point of official debt to GDP of an LG is associated with an increase by 0.3 percentage points in the general budget primary balance (before CG transfers) in the following period (Table 2), indicating fiscal tightening when debt is higher.
- Table 2 (regression highlights):
  - Dependent variable: general budget primary balance.
  - L.Offical Debt coefficient: 0.30** (reported as "L.Offical Debt0.30**").
  - D. Health Exp. coefficient: -2.78*** (reported as "D.Healh Exp.-2.78***").
  - L.LGFV spread (AA) coefficient: -1.61*** (reported as "L.LGFV spread (AA)-1.61***").
  - Other variables: D. Land Sales coefficient: -0.17; L. LGFV Debt coefficient: 0.02; L.Offical Int. Rate coefficient: 3.51; RGDP growth coefficient: 0.08.
  - Sample size and fit: N74; R_20.37.
- Interpretation:
  - LGs choose or are forced to tighten fiscal policy when official debt is higher.
  - Higher official interest rates are associated with fiscal tightening.
  - Lower land sale revenue and higher real GDP growth are associated with fiscal tightening (though the latter two are not statistically significant).
  - Rises in health expenditure (associated with Covid in the sample) lead to a more than proportional increase in the deficit.
  - Higher LGFV spreads are associated with fiscal loosening (through higher primary expenditure), possibly because financially constrained LGFVs are less able to undertake public projects.

### Credit market access, fiscal drag, and financial instability trade-offs
- Provinces with higher official debt have historically seen:
  - Slower LGFV debt growth.
  - Higher LGFV funding costs.
- Once weakening on-budget finances limit LGFV credit access, LGFVs may sharply reduce large ongoing investment flows, exacerbating off-budget fiscal tightening.
- Reported consequences in certain highly indebted provinces:
  - Sharper curtailment of credit access forcing LGs to re-direct fiscal or SOE resources to LGFV debt service payments, potentially worsening liquidity stress.

### Authorities’ policy response (measures and practices)
- Authorities focus on containing LGFV debt and strictly controlling new issuance.
- Key reported measures and actions:
  - Guidelines to strictly limit growth in borrowing for overindebted regions and specific troubled LGFVs, and stronger controls on investment project approvals.
  - In October 2023, LGFV liabilities of roughly RMB 1.1 trillion were reported to be swapped for official LG debt in heavily indebted provinces (a practice used selectively since 2015).
  - Selective restructurings requiring banks (including LG- and CG-owned entities) to provide maturity extensions and lower borrowing rates.
  - Central bank indicated it may provide liquidity support for some LG debts using its own balance sheet.
  - Policy banks or other CG-owned financial institutions have consolidated and refinanced LGFV debts in new asset management vehicles.
  - Intent to accelerate clearing of LG arrears to LGFVs and step up asset sales.

### Addressing unsustainable LG debts: intersectoral balance sheet perspective
- Recommendation: assess sustainability of LGFV and LG debts on an entity-by-entity basis with conservative future cash flow projections.
- Unsustainable debts should be restructured as part of a CG-coordinated program of fiscal framework reforms and balance sheet restructuring.
- Policy approaches differ in how losses are redistributed across sectors and their potential for financial contagion.

### Scenarios for LGFV debt reduction and implications (as simulated)
- LG refinancing (LG bailout):
  - Simulated as one-time transfer of LG liquid financial assets to LGFV sector, funded by additional LG official debt.
  - Cost borne entirely by LG sector; reduces immediate default risk but undermines LG debt sustainability and erodes LG fiscal space.
  - Complicates eventual debt restructuring because LGFVs can be restructured under corporate insolvency while LGs lack an insolvency framework.
- Use of LG-owned SOEs and lenders to absorb LGFV debt:
  - Economically similar to LG bailout in burden-sharing; initially spares LG balance sheet but weakens SOE sector and transfers ultimate losses back to LGs via equity claims.
- Use of CG balance sheet (CG bailout):
  - CG official debt is only about one-third of official LG debt, so a transfer would greatly ease LG liquidity problems.
  - Creates significant moral hazard and could lead to renewed LG debt accumulation if not tightly constrained.
- Reductions in face value via insolvency (write-down):
  - Market value of LGFV debt is reduced proportionately for all claim holders; subsequent rounds reduce equity owners’ value.
  - Text description: the official LG and CG sectors each absorb about 30 percent of the total losses, with the remaining 40 percent absorbed by other sectors.
  - Staff-simulated distribution chart shows values as 30, 28, 42 (reported as "30 28 42" and labeled "Local Governments Central Government Other Sectors").
- Financial institutions’ reprofiling of LGFV debt:
  - Similar to a write-down economically but does less to improve LG finances and increases banking system risks by reducing banks’ economic capital and loss-absorption buffers.
  - Unlikely to restore LGFV long-term viability given weak income streams; contingent liability risks for LGs remain.

### Benefits of insolvency-driven write-downs and complementary measures
- Insolvency-driven write-downs:
  - Reduce economy-wide leverage and general government leverage.
  - Force bank shareholders and investors to absorb losses and add capital, reducing moral hazard and improving market discipline.
  - Better distribute losses across sectors: text cites a 28 percent share for LGs and a 30 percent share for CG (see sequencing below).
  - The wealthiest ten percent of households owns an estimated 60 percent of non-deposit financial assets; household losses would be concentrated among wealthier households.
- Asset sales and recoveries:
  - Sales of sizeable public assets, including privatization of profitable SOEs, could help finance debt restructuring.
  - Corporate insolvency provides a direct legal mechanism for maximizing creditor recovery, potentially reducing ultimate cost of write-downs.
- Complementary legal mechanisms:
  - Corporate insolvency may need to be complemented with a specialized CG-administered mechanism to restructure quasi-fiscal LGFVs.
  - CG receivership and an ex ante LG insolvency mechanism would help manage asset sales and avoid overburdening the legal system.

### Fiscal and social safeguards during restructuring
- Debt write-downs should be accompanied by:
  - Multiyear fiscal consolidation and debt servicing plans for affected LGs, subject to close CG monitoring as the basis for CG assistance.
  - One-off targeted CG transfers to ensure continuity of public services and avoid unintended sharp fiscal consolidation in the weakest LGs.
  - Support for vulnerable households during restructuring.
  - Broader fiscal reforms to raise LG revenues and maintain LG debt sustainability going forward.

### Financial sector implications and required safeguards
- Potential scale of restructuring needs could be large given LGFV debt is reported as 48 percent of GDP and persistent lack of operating cashflows across the sector.
- Significant bank credit losses may require CG-coordinated capital injections and resolution of weak banks.
- Possible spillovers: curtailed credit to other sectors and sharp declines in asset prices if market-based asset sales reveal overvaluation.
- Recommended actions:
  - Strengthen bank restructuring and resolution frameworks to facilitate orderly exit of weak banks without endangering financial stability, including establishment of a temporary financial resolution fund.
  - Coordinate restructuring across sectors and preposition financial and administrative resources.
  - Enhance crisis preparedness and operational capacity of the financial safety net.
  - Continue asset management and credit market reforms to ensure retail investors are transparently informed of risks in exposures to LGFV credit in bank wealth management and other investment products.

### Strengthening LG fiscal frameworks and preventing future debt accumulation
- Upgrade fiscal reporting and monitoring following GFSM standards; classify and report public-sector-controlled non-market producers as part of general government (this could include a substantial share of LGFVs).
- Increase transparency consistent with IPSAS for LGs, SOEs, and state-owned banks.
- Concrete transparency steps:
  - Create a public registry of all public sector company debt issuance and transactions with LGs.
  - Require timely reporting of audited LG financial statements, including cash flows.
- Upgrade LG fiscal risk monitoring systems using quantitative indicators of debt sustainability encompassing the broader GFSM perimeter and interlinkages with the financial system.
- Inclusion of LGFVs in official budgets should occur once the debt of such LGFVs has been restructured.

*Source: IMF staff analysis and simulations as presented in the chapter text.*

### 31.      LG fiscal frameworks should be upgraded to reduce future debt-generating flows. Tax

### 31.      LG fiscal frameworks should be upgraded to reduce future debt-generating flows. Tax

### Major fiscal policy recommendations for local governments (LGs)
- Tax revenue reform to close LG fiscal gaps; about half of additional revenue would accrue to LGs.
- Consider allocating a larger share of tax revenue to LGs, commensurate with their expenditure mandates, to reduce vertical imbalances (Wingender 2018).
- Reform the central government (CG) transfer system to increase risk sharing among provinces and the CG (IMF 2020) to:
  - remove pressure for LGs to resort to non-standard financing channels; and
  - attenuate feedback loops in economically weaker provinces.
- Public financial management (PFM) reforms to:
  - clarify LG expenditure mandates;
  - integrate financial management and reporting systems; and
  - phase out provinces’ growth targets, which encourage excessive investment financing and growing indebtedness at the local level.
- Require that all future borrowing for projects of a fiscal nature be included in LG budgets to enhance transparency.
- Control options for LG borrowing (Saxena 2022) include:
  - direct control by the CG (examples cited: Peru, Spain, and Thailand);
  - subnational fiscal rules (examples cited: Brazil, India, and Philippines); or
  - market discipline (examples cited: Canada, South Africa, and the United States).
- Consider transitioning from the current mechanism of LG debt quotas to rules-based constraints to strengthen market discipline in the long term.

### Financial sector and supervisory measures to limit LG-related risks
- Align financial policies to support reduction of debt-generating flows.
- Use bank supervision and regulatory policies to guide banks to limit exposure to unsustainable LG-related debts.
- Supervisory tools that could be applied include:
  - identifying credit exposures to LG-owned entities outside of the general government perimeter; and
  - requiring lenders to use higher risk weights, additional capital charges, or concentration limits for exposures that lack income-based debt repayment capacity (e.g., based on a defined debt to earnings threshold).

### Loss allocation and shock methodology (Annex I summary)
- Distribution of losses across sectors follows Garcia-Macia (2021) using matrices of cross-sectoral asset holdings.
- First-round impact d on sectors i holding asset k issued by sector j:
  - d_{i k} = −∆p_{ik} a_{i k} if i ≠ j (notation preserved as in source).
- Distinguishes sectors that are entirely owned by other sectors (LGFV, DI and NFC) from ultimate shock absorbers (LG, CG, HH, and RoW).
- LGFV, DI and NFC pass on their financial losses to ultimate shock absorbers via equity cross holdings (column “equity absorption”).
- Total financial loss for ultimate shock absorbers equals the sum of first-round impacts and knock-on impacts from equity holdings.

### Land sale shock and assumptions
- Loss in LG net worth assumes land sale expenditures have a 50 percent elasticity with respect to land sale revenue.
- Elasticity of LGFV spreads to land sale revenues from province-level panel data: -21 percent.
- Assumed average outstanding duration of LGFV debt: 3 years.
- Loss in the value of equity is the residual given the total loss in asset value.

### Data sources, matrices construction, and sector allocations
- Asset management product (AMP) sector holdings apportioned pro rata: FIs = 31 percent; LGFVs = 10 percent; HHs = 45 percent; NFCs = 10 percent.
- Debt claims and equity matrices use multiple sources and bottom-up estimations (CEIC, Capital IQ, PBC, WIND, China Central Depository & Clearing Co Ltd., Shanghai Clearing House, BIS).
- Key sectoral estimation rules (verbatim terms and shares preserved):
  - Local governments (LG):
    - Liabilities: official debt liabilities distributed across sectors based on sectoral bond ownership data produced by China Central Depository & Clearing Co Ltd. data, accessed via CEIC.
    - Assets: LGs’ credit claims on DIs are estimated as half of total financial institutions' fiscal deposits, as reported by the PBC.
  - Local government financing vehicle (LGFVs):
    - Liabilities: LGFV debt liabilities estimated from aggregated balance sheet data of nearly 2200 LGFVs, obtained via S&P Capital IQ.
    - Senior bonds and notes attributed 97 percent to domestic AMPs and 3 percent to non-residents.
    - Other interest-bearing debt liabilities apportioned to AMPs (75 percent) and FIs (25 percent).
    - Other payables apportioned to NFCs (75 percent) and NBFIs (25 percent).
    - Assets: debt claims on NFCs calculated as 20 of total receivables and half of Other Long-Term Assets; debt claims on LGs estimated as 80 percent of total receivables.
  - Financial Institutions:
    - Liabilities: Banks’ non-bond debt liabilities calculated from the PBC’s Depository Institution Balance Sheet data series.
    - Banks’ bond liabilities apportioned using sectoral bond ownership data from China Depository & Clearing Co. Ltd and Shanghai Clearing House.
    - Assets: Banks’ non-bond credit assets from the PBC’s Depository Institution Balance Sheet data series.
  - Households:
    - Liabilities: household loans as reported in the PBC’s Depository Institution Sources and Uses of Funds data.
    - Assets: household deposits as reported in the PBC’s Depository Institution Sources and Uses of Funds data and HH’s pro-rata share of the financial claims held by the AMP sector.
  - Nonfinancial Corporates:
    - Liabilities: includes estimated receivables owed to LGFVs; debt liabilities to banking system from PBC data less LGFV loans.
    - Assets: nonfinancial enterprise deposits as reported in PBC data.
  - Central Government (CG):
    - Liabilities: official debt liabilities distributed across sectors based on sectoral bond ownership data produced by China Central Depository & Clearing Co Ltd. data, accessed via CEIC.
    - Assets: CG’s credit claims on FIs estimated as half of total financial institutions' fiscal deposits, as reported by the PBC.
  - Rest of World:
    - Liabilities and assets based on PBC and BIS locational/banking statistics as detailed.

- Equity matrices notes:
  - LGFV reported minority interest equity assumed held by the NFC sector; remaining LGFV equity attributed to the LG sector.
  - For financial institutions, LGFV claims assumed to be 30 percent of LGFV’s long-term investment sourced from Capital IQ.
  - Claims on AMPs distributed pro-rata to sectors using AMP ownership shares noted above.

### Estimation of LG fiscal reaction function (Annex II)
- Regression specification:
  - ∆pb_{it} = β d_{it−1} + γ X_{it−1} + δ_i + ε_{it} (formula preserved as in source).
  - i indexes provinces and t years; pb denotes the general budget primary balance as a share of GDP; d official debt-to-GDP; X is a vector of fiscal variables and other controls; δ are province fixed effects; ε the error term.
- Data:
  - Annual data on fiscal variables (both on-budget and LGFV) at the provincial level for 2019-2022.
  - Most variables from CEIC; LGFV debt data from Capital IQ aggregated with bottom-up methodology; LGFV spreads from WIND.
  - Sample period constrained by availability of LGFV spread data at the province level, which starts in May 2019. Values for spreads in 2019 are imputed from the average of the available months.
- Summary statistics (Annex II. Text Table 1, preserved exactly):
  - Primary Bal. units: percent of GDP; Mean: -14.6; Std. dev.: 19.3; p25: -18.5; p75: -4.9; N: 108
  - Healh Exp. units: percent of GDP; Mean: 2.3; Std. dev.: 1.2; p25: 1.5; p75: 2.7; N: 111
  - Land Sales units: percent of GDP; Mean: 5.7; Std. dev.: 2.8; p25: 3.8; p75: 7.1; N: 112
  - Offical Debt units: percent of GDP; Mean: 32.4; Std. dev.: 15.1; p25: 21.6; p75: 41.8; N: 116
  - LGFV Debt units: percent of GDP; Mean: 44.3; Std. dev.: 26.4; p25: 23.4; p75: 59.2; N: 116
  - Offical Int. Rate units: percentage points; Mean: 1.9; Std. dev.: 0.5; p25: 1.6; p75: 2.3; N: 108
  - LGFV spread (AA) units: percentage points; Mean: 3.0; Std. dev.: 1.5; p25: 2.1; p75: 3.8; N: 106
  - RGDP growth units: percent; Mean: -0.9; Std. dev.: 4.3; p25: -3.6; p75: 0.8; N: 116

*People’s Republic of China: chapter content as provided in the source PDF.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2024/english/1chnea2024002.pdf_