## wp1890

## Source details

**Canonical URL:** [wp1890](https://www.imf.org/-/media/files/publications/wp/2018/wp1890.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp1890.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp1890.pdf.json)

---

### Overview and motivation
- House prices in Hong Kong SAR increased by 145 percent in real terms since mid-2007.
- Household credit-to-GDP ratio increased by about 20 percent over the same period.
- Hong Kong Monetary Authority implemented eight rounds of macroprudential tightening since October 2009.
- Macroprudential measures:
  - Reductions in the maximum loan-to-value (LTV) ratio and debt servicing ratio (DSR).
  - Current maximum LTV ratio: 60 percent for the mass market and 50 percent for the luxury market.
    - 50 percent LTV ratio applies to properties whose value is above HKD 10 million.
  - Maximum DSR ratio is 50 percent.
  - Tighter caps apply to borrowers whose main income is not Hong Kong SAR based; with multiple mortgages; purchasing non-owner occupied properties; or receiving mortgages based on net worth instead of income.
- Stricter risk weights applied for residential mortgages on banks.
- Housing-related tax measures:
  - Ad Valorem Stamp Duty tax: several increases since 2013, with a current highest flat rate of 15 percent.
  - Buyer’s Stamp Duty tax: introduced in October 2012 targeting non-resident and company buyers, with a rate of 15 percent.
  - Special Stamp Duty tax: implemented in November 2010 targeting properties resold within 36 months or less, with a highest rate of 20 percent.

### Contribution and key empirical findings
- Approach:
  - Estimated Dynamic Stochastic General Equilibrium (DSGE) model for Hong Kong SAR.
  - Small open economy real business cycle model extended with a housing sector, financial frictions, and foreign investors.
  - Allows policy levers: limits to LTV ratios and stamp duty taxes affecting domestic and foreign purchases.
  - Estimation via Bayesian methods on Hong Kong SAR data for 1996Q1–2017Q2.
- Counterfactual simulation (holding LTV and stamp duty taxes at pre-October 2009 levels):
  - House prices would have been 10.5 percent higher at the end of 2017.
  - Household credit-GDP ratio would have been 14 percent higher at the end of 2017.
  - LTV cap is more effective at containing leverage.
  - Stamp duty taxes are more effective at containing house prices.

### Relation to existing literature
- Empirical findings consistent with prior studies:
  - ECM and VAR approaches find macroprudential and stamp duty policies slowed house price appreciation and that house prices react with lags to LTV changes.
  - Specific findings noted:
    - Craig and Hua (2011) and Rabanal and Shukla (2018): ECM approaches find macroprudential and stamp duty policies slowed house price appreciation.
    - Nabar and Ahuja (2011): VAR approach finds house prices react with lags to changes in LTV ratios.
    - He (2014): LTV caps constrain household leverage but do not have sustained effects on housing prices; transaction taxes more effective at restraining price growth.
  - Related DSGE work: Funke and Paetz (2016) evaluate LTV and stamp duty tax rules in terms of welfare (not empirically).

### Paper organization
- Section 2: The model.
- Section 3: Estimation procedure.
- Section 4: Impulse responses to housing and policy shocks.
- Section 5: Counterfactual simulation without policy changes since 2009.
- Section 6: Concluding remarks.

### Model structure (high-level)
- Baseline: small open economy RBC model with flexible prices; world interest rate taken as given.
- Policy instruments:
  - LTV cap on new loans.
  - Ad Valorem Stamp Duty and Buyer’s Stamp Duty taxes on transactions and foreign purchases.
- Agents and components:
  - Savers: consume non-durables and housing, supply labor, accumulate capital, access international capital markets, provide credit to borrowers; utility includes external habit and housing preference shock ξD t.
  - Borrowers: consume non-durables and housing, supply labor, do not accumulate capital, more impatient (βB < β), borrowing constrained by housing collateral; loans are long-term with fraction κ refinanced each period so only new credit flow is subject to current LTV cap.
  - Foreign buyers: account for about 5 percent of purchases on average; face Buyer’s Stamp Duty τBSD t; foreign demand follows an Euler condition reacting to QD t, τBSD t, and world interest rate R* t.
  - Housing supply: assumed fixed at level D̄.
  - Consumption goods producers: Cobb-Douglas technology using capital and hours from both household types.
  - Fiscal policy: government rebates stamp duty revenues lump-sum to households; government does not accumulate debt.
- Shock processes:
  - Five exogenous macroeconomic disturbances: TFP, world interest rate, domestic demand, foreign demand, and credit shock.
  - Three policy shocks: LTV ratio, Ad-Valorem Stamp Duty, Buyer’s Stamp Duty.
  - All macro disturbances except the credit shock follow AR(1) processes in logs; credit shock is iid Normal. Policy shocks follow AR(1)-type processes centered on their steady values.

### Data used for estimation
- Sample: 1996Q1–2017Q2.
- Series and sources:
  - Real GDP growth (Y t / Y t−1): Real GDP in 2015 HK Dollars, Hong Kong SAR Census and Statistics Department.
  - Real house prices (QD t): Property Market Price Index, Private Domestic All Classes, Hong Kong SAR Rating and Valuation Department; deflated by Hong Kong SAR CPI in IMF’s World Economic Outlook Database.
  - Household Credit / GDP ratio (L t / Y t): Loans and Advances by Professional and Private Individuals, Hong Kong Monetary Authority, divided by Nominal GDP from the Hong Kong SAR Census and Statistics Department.
  - World real interest rate (R* t): 3-month Libor rate deflated by CPI (divided by 4 to convert to quarterly).
  - LTV ratio (LT V t): average of lowest and highest LTV cap for typical borrower, Hong Kong Monetary Authority.
  - Ad Valorem Stamp Duty tax (τADV t): average of lowest and highest ad valorem stamp duty tax, Hong Kong SAR Inland Revenue Department.
  - Buyer’s Stamp Duty tax (τBSD t): 15 percent starting in 2012Q4, zero otherwise, Hong Kong SAR Inland Revenue Department.
- Transformations: Real GDP growth, household credit-to-GDP ratio, and real house prices are logged and demeaned. LTV and stamp duty tax series are not transformed.

### Calibrated parameters (values and rationale)
- Calibrated parameter values:
  - β (Discount factor for savers): 0.997
  - βB (Discount factor borrowers): 0.99
  - α (Capital share of output): 0.33
  - δ (Depreciation rate): 0.025
  - ε (Fraction of savers): 0.55
  - γ (Fraction of consumption in utility): 0.8
  - D∗/ ̄D (Foreign ownership housing stock): 0.05
  - LTV (Steady-state LTV): 0.7
  - ̄τADV (Steady-state Ad Valorem Stamp Duty Tax): 0.03
  - ̄τBSD (Steady-state Buyer’s Stamp Duty Tax): 0
  - ρLTV (AR(1) coefficient for LTV shock): 0.9999
  - ρτADV (AR(1) coefficient for ADV shock): 0.9999
  - ρτBSD (AR(1) coefficient for BSD shock): 0.9999
- Rationale and interpretation:
  - β=0.997 chosen to match average real interest rate during the sample; βB=0.99 reflects borrower impatience.
  - α=0.33 and δ=0.025 are standard in RBC literature.
  - ε=0.55 calibrated to fraction of residential properties without a mortgage during the sample; noting that including renters would imply ε calibrated to 0.30.
  - γ set via 1−γ=0.2 to reflect housing importance; γ=0.8 is about double Iacoviello and Neri (2010) for the United States.
  - D∗/ ̄D set to 0.05 based on International Monetary Fund (2018).
  - Steady-state policy instrument values set to their 2009 (pre-measures) values: LTV cap at 0.7, Ad-Valorem Stamp Duty at 0.03, Buyer’s Stamp Duty at 0.
  - AR(1) coefficients on policy shocks set to 0.9999 so agents expect changes to be permanent while avoiding non-stationarity.

*Source: IMF Working Paper (content from "1. Introduction" and model overview sections of wp1890).*

### 1.    Introduction .....................................................................................................

### 1. Introduction

### Overview and motivation
- House prices in Hong Kong SAR increased by 145 percent in real terms since mid-2007.
- The household credit-to-GDP ratio increased by about 20 percent over the same period.
- In response, the Hong Kong Monetary Authority implemented eight rounds of macroprudential tightening since October 2009.
- Macroprudential measures include reductions in the maximum loan-to-value (LTV) ratio and debt servicing ratio (DSR).
  - Current maximum LTV ratio: 60 percent for the mass market and 50 percent for the luxury market.
  - More specifically, the 50 percent LTV ratio applies to properties whose value is above HKD 10 million.
  - Maximum DSR ratio is 50 percent.
  - Tighter caps apply to borrowers (i) whose main income is not Hong Kong SAR based, (ii) with multiple mortgages, (iii) purchasing non-owner occupied properties, and (iv) receiving mortgages based on net worth instead of income.
- Authorities also applied stricter risk weights for residential mortgages on banks.
- Housing-related tax measures:
  - Ad Valorem Stamp Duty tax: several increases since 2013, with a current highest flat rate of 15 percent.
  - Buyer’s Stamp Duty tax: introduced in October 2012 targeting non-resident and company buyers, with a rate of 15 percent.
  - Special Stamp Duty tax: implemented in November 2010 targeting properties resold within 36 months or less, with a highest rate of 20 percent.

### Contribution and key empirical findings
- The paper evaluates the impact of housing market related policies using an estimated Dynamic Stochastic General Equilibrium (DSGE) model for Hong Kong SAR.
- Model features:
  - Small open economy real business cycle model extended with a housing sector, financial frictions, and foreign investors that can purchase the domestic housing stock.
  - Allows for housing market policies in the form of limits to LTV ratios and stamp duty taxes that affect domestic and foreign purchases.
  - Estimation using Bayesian methods on Hong Kong SAR data for the period 1996Q1–2017Q2.
- Counterfactual simulation result (holding LTV and stamp duty taxes at pre-October 2009 levels):
  - Without these policies, house prices would have been 10.5 percent higher at the end of 2017.
  - Household credit-GDP ratio would have been 14 percent higher at the end of 2017.
  - LTV cap is more effective at containing leverage.
  - Stamp duty taxes are more effective at containing house prices.

### Relation to existing literature
- Empirical studies using alternative approaches generally find housing policies affect house price growth:
  - Craig and Hua (2011) and Rabanal and Shukla (2018): ECM approaches find macroprudential and stamp duty policies slowed house price appreciation.
  - Nabar and Ahuja (2011): VAR approach finds house prices react with lags to changes in LTV ratios.
  - He (2014): LTV caps constrain household leverage but do not have sustained effects on housing prices; transaction taxes more effective at restraining price growth.
- Closely related DSGE work: Funke and Paetz (2016) estimate a DSGE model for Hong Kong SAR and evaluate LTV and stamp duty tax rules in terms of welfare (but not empirically).

### Paper organization
- Section 2: The model.
- Section 3: Estimation procedure.
- Section 4: Impulse responses to housing and policy shocks.
- Section 5: Counterfactual simulation without policy changes since 2009.
- Section 6: Concluding remarks.

### Model structure (high-level)
- Baseline: small open economy RBC model with flexible prices; world interest rate taken as given.
- Policy instruments: LTV cap on new loans; Ad Valorem Stamp Duty and Buyer’s Stamp Duty taxes on transactions and foreign purchases.
- Key agents and components:
  - Savers: consume non-durables and housing, supply labor, accumulate capital, access international capital markets, and provide credit to borrowers; utility includes external habit in consumption and housing preference shock ξD t.
  - Borrowers: consume non-durables and housing, supply labor, do not accumulate capital, more impatient (βB < β), borrowing constrained by housing collateral; loans are long-term with fraction κ refinanced each period so only new credit flow is subject to current LTV cap.
  - Foreign buyers: account for about 5 percent of purchases on average; face Buyer’s Stamp Duty τBSD t; foreign demand follows an Euler condition reacting to QD t, τBSD t, and the world interest rate R* t.
  - Housing supply: assumed fixed at level D̄ (capturing limited land availability in Hong Kong SAR).
  - Consumption goods producers: Cobb-Douglas technology using capital and hours from both household types.
  - Fiscal policy: government rebates stamp duty revenues lump-sum to households; government does not accumulate debt.
- Shock processes:
  - Five exogenous macroeconomic disturbances: TFP, world interest rate, domestic demand, foreign demand, and credit shock.
  - Three policy shocks: LTV ratio, Ad-Valorem Stamp Duty, Buyer’s Stamp Duty.
  - All macro disturbances except the credit shock follow AR(1) processes in logs; credit shock is iid Normal. Policy shocks follow AR(1)-type processes centered on their steady values.

### Data used for estimation
- Sample: 1996Q1–2017Q2.
- Series:
  - Real GDP growth (Y t / Y t−1): Real GDP in 2015 HK Dollars, Hong Kong SAR Census and Statistics Department.
  - Real house prices (QD t): Property Market Price Index, Private Domestic All Classes, Hong Kong SAR Rating and Valuation Department; deflated by Hong Kong SAR CPI in IMF’s World Economic Outlook Database.
  - Household Credit / GDP ratio (L t / Y t): Loans and Advances by Professional and Private Individuals, Hong Kong Monetary Authority, divided by Nominal GDP from the Hong Kong SAR Census and Statistics Department.
  - World real interest rate (R* t): 3-month Libor rate deflated by CPI (divided by 4 to convert to quarterly).
  - LTV ratio (LT V t): average of lowest and highest LTV cap for typical borrower, Hong Kong Monetary Authority.
  - Ad Valorem Stamp Duty tax (τADV t): average of lowest and highest ad valorem stamp duty tax, Hong Kong SAR Inland Revenue Department.
  - Buyer’s Stamp Duty tax (τBSD t): 15 percent starting in 2012Q4, zero otherwise, Hong Kong SAR Inland Revenue Department.
- Transformations: Real GDP growth, household credit-to-GDP ratio, and real house prices are logged and demeaned. LTV and stamp duty tax series are not transformed.

*Source: IMF Working Paper (content from "1. Introduction" and model overview sections of wp1890).*

### 3.2  Calibrated Parameters

### 3.2  Calibrated Parameters

### Calibrated parameter values
- β (Discount factor for savers): 0.997
- βB (Discount factor borrowers): 0.99
- α (Capital share of output): 0.33
- δ (Depreciation rate): 0.025
- ε (Fraction of savers): 0.55
- γ (Fraction of consumption in utility): 0.8
- D∗/ ̄D (Foreign ownership housing stock): 0.05
- LTV (Steady-state LTV): 0.7
- ̄τADV (Steady-state Ad Valorem Stamp Duty Tax): 0.03
- ̄τBSD (Steady-state Buyer’s Stamp Duty Tax): 0
- ρLTV (AR(1) coefficient for LTV shock): 0.9999
- ρτADV (AR(1) coefficient for ADV shock): 0.9999
- ρτBSD (AR(1) coefficient for BSD shock): 0.9999

### Calibration choices and rationale
- Discount factors: savers set to 0.997 to match the average real interest rate during the sample period; borrowers set to 0.99 to reflect impatience.
- Capital share and depreciation: set to values common in the real business cycle literature (α=0.33, δ=0.025).
- Fraction of savers (ε=0.55): calibrated to the fraction of residential properties in Hong Kong SAR that do not have a mortgage during the sample period (see Hong Kong Monetary Authority (2017)). Note: the model only considers households that own a residential property; if renting versus owning were included, fraction of savers would be calibrated to 0.30.
- Weight of housing in utility set via 1−γ=0.2 (implying γ=0.8) to reflect the importance of housing expenditure in Hong Kong SAR; this is about double the value used by Iacoviello and Neri (2010) for the United States.
- Fraction of housing stock owned by foreign investors set to 5 percent, based on International Monetary Fund (2018).
- Steady-state policy instrument values set to their 2009 values (pre-measures): LTV cap at 70 percent, Ad-Valorem Stamp Duty at 3 percent, Buyer’s Stamp Duty at 0 percent.
- AR(1) coefficients on all policy shocks set to a number arbitrarily close to one (but smaller than one) so agents expect policy changes to be permanent while avoiding non-stationarity.

### Notes on interpretation
- The AR(1) coefficients are chosen very close to one so that once LTV or stamp duty tax is changed, agents expect the change to be permanent; assuming a much smaller AR(1) would imply agents are surprised every period to keep the policy instrument constant as in the data.
- The calibration is conditional on the model considering only owner-occupied households; alternative model specifications (e.g., including renting) would change calibrated fractions.

*Source: Staff estimates.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp1890.pdf_