## wp18261

## Source details

**Canonical URL:** [wp18261](https://www.imf.org/-/media/files/publications/wp/2018/wp18261.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp18261.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp18261.pdf.json)

---

### The Long-Run Trend of Residential Investment in China
- Residential housing investment grew from about 4 percent of GDP in 1997 to the peak of 15 percent of GDP in 2014, with residential investment accounting for more than two thirds of total housing investment.
- By cross-country standards, China’s current residential investment as a share of GDP is well above that of countries with similar real household income levels (adjusted for the purchasing power parity and inflation).
- Market distortions increasing susceptibility to price misalignment and overbuilding:
  - Supply-side distortion: local governments’ control over land supply and reliance on land sales to finance spending.
  - Demand-side distortion: housing attractive as an investment instrument due to a history of robust capital gains, high savings, low real deposit interest rates, lack of alternative financial assets, capital account restrictions, and rapid increase in mortgage loans as banks seek to maintain asset expansion amid slower corporate credit growth.
  - Government tendency to boost housing activity during downturns, creating expectation that housing is “too important to fail.”
- Announced “long-term mechanism” key elements:
  - Increase land supply in cities facing stiff upward pressure by converting rural collective land and urban commercial land to residential land.
  - Enhance affordability by developing the rental market and expanding social housing including through the shanty town renovation program.
  - Reduce property market pressure in mega cities by developing urban clusters and satellite cities with improved public services and transportation.
  - Maintain prudential standards on mortgage lending to help contain speculation.
  - Introduce a recurrent property value tax.
- Many underlying drivers behind past strong housing growth—strong household income growth and urbanization—face secular slowdown as China transitions to a “new normal” and experiences demographic changes.
- Focus of the analysis: examine drivers of the residential investment-to-GDP ratio (RIY) using a two-step approach that controls for endogeneity of house prices and structural changes.

### Methodology (Two-step approach and decomposition)
- Two-step approach:
  1. Decompose residential investment to GDP ratios into three terms plus a residual based on an accounting identity (Davis and Heathcote (2007) framework).
  2. Conduct cross-country regression analysis on the residual term to identify drivers and predict long-run equilibrium for China.
- Decomposition identity (notations from the source):
  - RIYt ≜ ItS / Yt = PtS (St − (1 − δS)St−1) / Yt = PtCCt / Yt × PtS / d tS × (1 − ((1 − δS)Pt−1CCt−1 / PtCCt d tS d t−1S)) eεt
  - Definitions:
    - dtS = PtS − (1 − δS) Pt+1S R t is the user cost of housing structures.
    - St is the residential structures stock; change in structures stock equals residential investment in national accounts.
    - Yt is nominal GDP; PtCCt is nominal consumption; PtS is the deflator of residential investment; δS is depreciation rate of residential structure; εt is the residual.
- Economic interpretation of the three main terms:
  - Consumption-to-GDP ratio:
    - Consumption-to-GDP ratio increased from 48 percent in 2010 to 54 percent in 2017 and is projected to rise to 58 percent by 2024 according to IMF staff’s latest projection (IMF 2018).
    - Rebalancing toward consumption is associated with the rise in RIY.
  - Deflator-to-user-cost ratio of residential structures:
    - Increases in expected returns to owning a house raise the price deflator relative to the cost of structures, increasing RIY.
    - Structure prices are mainly determined by productivity differentials between construction and the rest of the economy; decomposition uses the deflator of structures rather than house prices.
  - Growth of consumption relative to growth of the cost of residential structures:
    - If the cost of structures grows faster than consumption, downward pressure on RIY follows.
    - Construction sector is labor intensive; user-cost-of-structures is sensitive to labor costs.
    - Annual nominal wage growth in China has remained above 10 percent since the early 2000s, plausibly contributing to rising cost of residential structures.
- Residual term may reflect:
  - Urbanization (change in urban share in population).
  - Expectations about future housing price and income growth, including speculative demand.

### Empirical strategy for the residual term
- Cross-country panel regression of residual ε_{i,t} uses sample of 34 countries spanning 1980 to 2016.
- Residual regression specification (equation (2)):
  - ε_{i,t} = α_i + μ_t + β_1 (Pop_{i,t}^Urban / Pop_{i,t}) + β_2 (Pop_{i,t}^Urban / Pop_{i,t})^2 + β_3 ΔPop_{i,t}^Urban / Pop_{i,t} + ρ g_{i,t−5→t}^{House price} + φ ω_{i,t} + v_{i,t}
  - Variables include share of population living in urban area, its square, five-year average growth rate of housing price, homeownership rate, country fixed effect α_i, time fixed effect μ_t.
- Additional controls (credit gaps and expected output growth) tested and found not significant for explaining the residual term.
- For China forecast (2019–2029), impact of the level of urbanization on RIY is set to zero because the coefficient of the share of population living in urban area is not significant.

### Key regression estimates (selected figures from Table 3)
- Urban share in population: examples of reported coefficients: 0.493, -9.419***, -4.901**, 15.41, -1.263, -4.844.
- (Urban share in population)^2: coefficient -12.91* (standard error 7.433).
- Δ urban share in population: reported coefficients include -0.823, 3.637, 1.514, 1.798, 1.008, 0.192.
- Past five year growth of real housing prices: coefficients reported include 0.0270***, 0.0185**, 0.0195**, 0.0186*, 5.61e-05, -0.00732.
- Homeownership rate: coefficient 0.00799 (standard error 0.00854).
- Observations: 460, 122, 230 across different specifications.
- R-squared values across columns: 0.024, 0.418, 0.503, 0.507, 0.706, 0.483.
- Significance notation: *** p<0.01, ** p<0.05, * p<0.1.

### Empirical findings and comparative evidence
- Comparison with the United States (2004–2016):
  - Both countries had broadly stable consumption-to-GDP ratios; divergence in RIY dynamics did not arise from this term.
  - Deflator-to-user-cost ratio did not explain divergence.
  - The third channel—growth of consumption relative to growth of the cost of residential structures—explains China’s increase in RIY relative to the U.S.: China experienced faster consumption growth during 2004–16.
- Residual analysis:
  - Cross-country regression suggests a higher share of urban population can put downward pressure on RIY (possibly due to reduced land supply for urban housing development).
  - Regression shows a positive coefficient on housing price growth for the residual term; given moderating house price growth in China over the past decade, speculative demand is unlikely to explain the RIY boom.
  - No robust cross-country evidence of an independent credit effect on the residual term; credit may act indirectly via boosting consumption.
- Interpretation:
  - Increase in China’s RIY over the last decade is related to structural changes that raised consumption share of GDP rather than urbanization or speculation per se.

### Projections for China’s long-run residential investment (2019–2029)
- Main projection results:
  - Current residential investment level of 9 percent of GDP is projected to moderate to around 6 percent by 2024.
  - Residential investment’s contribution to real GDP growth is projected to decline from currently about half percent of GDP to -0.1 percent of GDP by 2024.
- Numerical assumptions and calibration:
  - Consumption growth is projected to moderate from 14 percent during the residential investment boom in 2010-15 to less than 9 percent by 2024.
  - The real user cost of structures is assumed to grow at around 5 percent between 2019 and 2029, slightly lower than the average of 5.4 percent between 2009 and 2018.
  - This rise in real user cost dominates the direct positive impact of rebalancing on residential investment.
  - Depreciation rate δ_S is chosen as 3.5 percent per year (following Lian (2018)).
- Implication:
  - If the “long-term mechanism” takes hold and policies balance supply and demand while curbing speculation, the residential housing market should be determined mostly by fundamental demand and supply factors, making the above projections more likely.

### Policy implications and recommendations
- If the projected decline in RIY reflects rebalancing and a transition to slower but more sustainable growth, the government should avoid preventing such a decline.
- Specific recommendations:
  - Move away from setting high annual growth targets that can foster an undesirable focus on boosting real estate activity through credit.
  - Implement pro-rebalancing policies such as:
    - A more progressive tax system.
    - Higher spending on social transfers.
    - Promotion of social and rental housing.
    - Liberalization of land supply.
  - These measures could boost consumption and ameliorate income inequality, support housing demand, improve affordability, and lead to a more gradual adjustment in the residential real estate market.
- Caution: Given housing’s importance to the Chinese economy, a sharp decline in residential investment should be avoided.

### Theoretical foundation and urbanization channel
- Benchmark Cobb-Douglas case:
  - Under Cobb-Douglas preferences, households spend a fixed portion of consumption expenditure on housing services; residual term defined from decomposition of log(RI/Y).
- Zero elasticity of substitution (Leontief) case:
  - When substitutability between land and structures is low, the residual term is no longer constant.
  - Leontief production: f(h(L_t,S_t), Z_t) = (min(L_t, S_t))^{1−γ} (Z_t)^γ.
  - Optimal condition: (P_t^L − P_{t+1}^L R_t + P_t^S − (1−δ) P_{t+1}^S R_t) S_t = (1−γ) P_t^C C_t.
  - If P_{i,t}^L − P_{i,t+1}^L R_{i,t} grows faster than P_{i,t}^S − (1−δ) P_{i,t+1}^S R_{i,t}, then log(ε_{i,t}) should exhibit a downward trend.
- Urbanization channels:
  - Faster urbanization (migration) increases demand for housing and is associated with stronger residential investment.
  - A countervailing channel: as the relative importance of land versus structures increases, θ_S may decline, implying a smaller ratio of residential investment over GDP — an urbanization-induced negative effect.
  - Empirical and historical evidence (Glaeser, Gyourko and Saks (2005); Lian (2018); Davis and Palumbo (2008)) indicate land’s role in housing supply has risen over time and tends to be higher in more regulated or coastal areas, affecting RIY dynamics.

*Prepared by Ding Ding and Weicheng Lian; IMF Working Paper WP/18/261, Asia and Pacific Department.*

### Section 1

### The Long-Run Trend of Residential Investment in China

### Introduction
- Residential housing investment grew from about 4 percent of GDP in 1997 to the peak of 15 percent of GDP in 2014, with residential investment accounting for more than two thirds of total housing investment.
- By cross-country standards, China’s current residential investment as a share of GDP is well above that of countries with similar real household income levels (adjusted for the purchasing power parity and inflation).
- Market distortions increasing susceptibility to price misalignment and overbuilding:
  - Supply-side distortion: local governments’ control over land supply and reliance on land sales to finance spending.
  - Demand-side distortion: housing attractive as an investment instrument due to a history of robust capital gains, high savings, low real deposit interest rates, lack of alternative financial assets, capital account restrictions, and rapid increase in mortgage loans as banks seek to maintain asset expansion amid slower corporate credit growth.
  - Government tendency to boost housing activity during downturns, creating expectation that housing is “too important to fail.”
- The announced “long-term mechanism” for the residential housing market aims to ensure sustained and stable development by balancing supply and demand and curbing speculation. Key elements include:
  - Increasing land supply in cities facing stiff upward pressure by converting rural collective land and urban commercial land to residential land.
  - Enhancing affordability by developing the rental market and expanding social housing including through the shanty town renovation program.
  - Reducing property market pressure in mega cities by developing urban clusters and satellite cities with improved public services and transportation.
  - Maintaining prudential standards on mortgage lending to help contain speculation.
  - Introducing a recurrent property value tax.
- Many underlying drivers behind past strong housing growth—strong household income growth and urbanization—face secular slowdown as China transitions to a “new normal” and experiences demographic changes.
- Focus of the paper: examine underlying drivers of housing investment as a share of GDP using cross-country evidence and project the long-run trend for China using a two-step approach that controls for endogeneity of house prices and structural changes of the economy. The approach focuses on the residential investment-to-GDP ratio (RIY) rather than residential investment levels.

### Methodology
- Two-step approach:
  1. Decompose residential investment to GDP ratios into three terms plus a residual based on an accounting identity (Davis and Heathcote (2007) framework).
  2. Conduct cross-country regression analysis on the residual term to identify drivers and predict long-run equilibrium for China.
- Decomposition identity (notations retained from source):
  - RIYt ≜ ItS / Yt = PtS (St − (1 − δS)St−1) / Yt = PtCCt / Yt × PtS / d tS × (1 − ((1 − δS)Pt−1CCt−1 / PtCCt d tS d t−1S)) eεt
  - Definitions and interpretations in the source:
    - dtS = PtS − (1 − δS) Pt+1S R t is the user cost of housing structures.
    - St is the residential structures stock; change in structures stock equals residential investment in national accounts.
    - Yt is nominal GDP; PtCCt is nominal consumption; PtS is the deflator of residential investment; δS is depreciation rate of residential structure; εt is the residual.
- Economic interpretation of the three main terms:
  - The consumption-to-GDP ratio:
    - Housing services are durable and not easily substitutable; higher income raises spending on housing-related consumption (renovation, moving up the property ladder).
    - Consumption-to-GDP ratio increased from 48 percent in 2010 to 54 percent in 2017 and is projected to rise to 58 percent by 2024 according to IMF staff’s latest projection (IMF 2018).
    - The rebalancing toward consumption is associated with the rise in residential investment-to-GDP ratio.
  - The deflator-to-user-cost ratio of residential structures:
    - Increases in expected returns to owning a house (lower interest rates, stronger risk appetite) raise the price deflator relative to the cost of structures, increasing RIY.
    - The decomposition uses the deflator of structures rather than house prices; structure prices are mainly determined by productivity differentials between construction and the rest of the economy.
  - The growth of consumption relative to the growth of the cost of residential structures:
    - If the cost of structures grows faster than consumption, downward pressure on RIY follows.
    - The construction sector is labor intensive; user-cost-of-structures is sensitive to labor costs.
    - Annual nominal wage growth in China has remained above 10 percent since the early 2000s (Das and N’Diaye 2013), plausibly contributing to rising cost of residential structures.
- The residual term (difference between RIY and the sum of the three terms) can reflect:
  - Urbanization (change in urban share in population): could increase construction for migrants but a larger urban share could reduce RIY due to land supply restrictions.
  - Expectations about future housing price and income growth, including speculative demand for housing.
- Cross-country panel regression of the residual term:
  - Sample covers 34 countries spanning 1980 to 2016.
  - Results (summarized in the paper and appendices) indicate urbanization and speculation played limited roles in driving China’s RIY boom.

### Empirical Findings and Comparative Evidence
- Comparison with the United States (2004–2016):
  - Both countries had broadly stable consumption-to-GDP ratios; divergence in RIY dynamics did not arise from this term.
  - The deflator-to-user-cost ratio did not explain divergence.
  - The third channel—growth of consumption relative to growth of the cost of residential structures—explains China’s increase in RIY relative to the U.S.: China experienced faster consumption growth during 2004–16.
- Residual analysis results:
  - Cross-country regression suggests a higher share of urban population can put downward pressure on RIY (possibly due to reduced land supply for urban housing development).
  - The regression shows a positive coefficient on housing price growth for the residual term; given moderating house price growth in China over the past decade, speculative demand is unlikely to explain the RIY boom.
  - No robust cross-country evidence of an independent credit effect on the residual term; credit may act indirectly via boosting consumption.
- Interpretation:
  - The increase in China’s RIY over the last decade is related to structural changes that raised consumption share of GDP rather than urbanization or speculation per se.

### Projection of China’s Long-Run Residential Investment
- Main projection results (from abstract and Section III):
  - Current residential investment level of 9 percent of GDP is projected to moderate to around 6 percent by 2024.
  - Residential investment’s contribution to real GDP growth is projected to decline from currently about half percent of GDP to -0.1 percent of GDP by 2024 (described in the paper as “slightly negative” and explicitly reported as -0.1 percent).
- Key drivers of the projected decline:
  - Moderation of consumption growth relative to growth of the costs of residential structures (third term in the decomposition).
  - Envisaged increases in labor costs due to demographic changes, raising user cost of structures.
  - A somewhat slower pace of rebalancing (consumption share rising more modestly going forward).
- Implication of the “long-term mechanism” taking hold:
  - If policies balance supply and demand and curb speculation, the residential housing market should be determined mostly by fundamental housing demand and supply factors, making the above projections more likely.

*Prepared by Ding Ding and Weicheng Lian; IMF Working Paper WP/18/261, Asia and Pacific Department.*

### Section 2

### wp18261 - Section 2

### Decomposition method and main drivers
- The residential investment to GDP ratio is decomposed into several terms capturing distinct drivers: rebalancing, urbanization, housing market speculation, and divergence between price and cost of residential structures.
- The decomposition controls for endogeneity often neglected in “partial equilibrium” settings.
- Income growth and economic rebalancing are identified as important forces for the relative GDP contribution of residential investment.

### Empirical specification for the residual term
- Residual regression (equation (2)) estimated:
  - ε_{i,t} = α_i + μ_t + β_1 (Pop_{i,t}^Urban / Pop_{i,t}) + β_2 (Pop_{i,t}^Urban / Pop_{i,t})^2 + β_3 ΔPop_{i,t}^Urban / Pop_{i,t} + ρ g_{i,t−5→t}^{House price} + φ ω_{i,t} + v_{i,t}
  - Variables: share of population living in urban area (Pop^Urban/Pop), past five-year average growth rate of housing price (g_{i,t−5→t}^{House price}), homeownership rate (ω_{i,t}), country fixed effect (α_i), time fixed effect (μ_t).
- As AEs and EMDEs are at different urbanization stages, column (5) reports estimates for EMDEs only.
- For the China forecast (2019–2029), the impact of the level of urbanization on residential investment to GDP ratio is set to zero because the coefficient of the share of population living in urban area is not significant.
- Additional controls (credit gaps and expected output growth) were tested and found not significant for explaining the residual term (results not reported).

### Key regression estimates (Table 3)
- Urban share in population: coefficients reported across specifications include 0.493, -9.419***, -4.901**, 15.41, -1.263, -4.844 (standard errors shown in table).
- (Urban share in population)^2: coefficient -12.91* (standard error 7.433).
- Δ urban share in population: coefficients reported include -0.823, 3.637, 1.514, 1.798, 1.008, 0.192 (standard errors shown in table).
- Past five year growth of real housing prices: coefficients 0.0270***, 0.0185**, 0.0195**, 0.0186*, 5.61e-05, -0.00732 (standard errors shown in table).
- Homeownership rate: coefficient 0.00799 (standard error 0.00854).
- Observations: 460 (columns with full sample), 122 (EMDEs in one column), 230 in another; R-squared values: 0.024, 0.418, 0.503, 0.507, 0.706, 0.483 across columns.
- Significance notation: *** p<0.01, ** p<0.05, * p<0.1.

### Projections for China (2019–2029) and numerical assumptions
- Projection summary:
  - Consumption growth is projected to moderate from 14 percent during the residential investment boom in 2010-15 to less than 9 percent by 2024.
  - The real user cost of structures is assumed to grow at around 5 percent between 2019 and 2029, which is slightly lower than the average of 5.4 percent between 2009 and 2018.
  - This rise in real user cost dominates the direct positive impact of rebalancing on residential investment.
- Projected path of China’s residential investment to GDP ratio:
  - Decline from currently around 9 percent of GDP to 6 percent of GDP by 2024.
- Calibration detail:
  - Following Lian (2018), the depreciation rate δ_S is chosen as 3.5 percent per year.

### Policy implications and recommendations
- If the projected decline in residential investment to GDP reflects rebalancing and a transition to slower but more sustainable growth, the government should avoid preventing such a decline.
- Specific policy recommendations:
  - Move away from setting high annual growth targets that can foster an undesirable focus on boosting real estate activity through credit.
  - Implement pro-rebalancing policies such as:
    - A more progressive tax system.
    - Higher spending on social transfers.
    - Promotion of social and rental housing.
    - Liberalization of land supply.
  - These measures could:
    - Boost consumption and ameliorate income inequality, supporting housing demand.
    - Improve affordability and indirectly support rebalancing, leading to a more gradual adjustment in the residential real estate market.
- Caution: Given housing’s importance to the Chinese economy, a sharp decline in residential investment should be avoided.

### Theoretical foundation (Appendix III) — key mechanisms
- Benchmark Cobb-Douglas preference case:
  - Household utility U defined over housing services h(L_t, S_t) (produced from land L_t and structures S_t) and other goods Z_t.
  - Under Cobb-Douglas, households spend a fixed portion of consumption expenditure on housing services; a residual term is defined from the decomposition of log(RI/Y).
- Zero elasticity of substitution (Leontief) case:
  - When substitutability between land and structures is low, the residual term is no longer constant.
  - Leontief production: f(h(L_t,S_t), Z_t) = (min(L_t, S_t))^{1−γ} (Z_t)^γ.
  - Optimal condition: (P_t^L − P_{t+1}^L R_t + P_t^S − (1−δ) P_{t+1}^S R_t) S_t = (1−γ) P_t^C C_t.
  - The ratio of residential investment over output RI/Y decomposes into terms involving structure and land prices, consumption deflator and output, and adjustment terms (see full expressions in the source).
  - If P_{i,t}^L − P_{i,t+1}^L R_{i,t} grows faster than P_{i,t}^S − (1−δ) P_{i,t+1}^S R_{i,t}, then log(ε_{i,t}) should exhibit a downward trend.
- Policy relevance of the residual:
  - Once log(ε_{i,t}) is not constant, housing policies and supply constraints (e.g., zoning) can affect it. Overly tight man-made housing supply constraints can reduce growth in housing quantity.

### Urbanization channel and interpretation
- Multiple channels link urbanization to the residential investment to GDP ratio:
  - Faster urbanization (migration) increases demand for housing and is associated with stronger residential investment.
  - A less-recognized channel: as the relative importance of land versus structures increases, θ_S may decline, implying a smaller ratio of residential investment over GDP — an urbanization-induced negative effect.
- Empirical and historical references:
  - Studies (Glaeser, Gyourko and Saks (2005); Lian (2018)) suggest land’s role in housing supply has risen over time in the United States, partly due to tighter housing regulations.
  - Cross-MSA data (Davis and Palumbo (2008)) indicate land’s share in housing value tends to be higher in coastal areas.

*Source: wp18261 - Section 2.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp18261.pdf_