## INTRODUCTION (IMF Working Paper — Household Debt and Borrower-Based Measures in Finland, INTRODUCTION section and chapter 3.4)

## Source details

**Canonical URL:** [INTRODUCTION (IMF Working Paper — Household Debt and Borrower-Based Measures in Finland, INTRODUCTION section and chapter 3.4)](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023262-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023262-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023262-print-pdf.pdf.json)

---

### Context and motivation
- Systemic risks such as sharply rising indebtedness and risk taking can accumulate and amplify adverse financial shocks; vulnerabilities in the housing sector repeatedly pose significant concerns.
- Macroprudential policy tools available in the household sector include sectoral capital requirements and caps on loan-to-value ratios (LTV), debt-to-income ratios (DTI), and debt-service-to-income ratios (DSTI).
- Household debt-to-disposable income ratio in Finland rose from 111 percent in 2010 to 133 percent in 2021.
- Finland’s maximum LTV (loan-to-collateral, LTC) ratio has been reduced to 85 percent, except for first-time home buyers; income-related borrower-based measures like DTI are not currently available as macroprudential tools despite recent proposals.

### Research question and approach
- Core question:
  - Would introducing a cap on the DTI ratio be useful in further mitigating risks associated with mortgage loans in Finland?
- Method:
  - Quantitative comparison of LTV caps and DTI caps on household consumption responses to negative economic scenarios using a simulation based on a heterogeneous agent household model.
- Rationale:
  - Theoretical heterogeneous-agent model permits counterfactuals and captures distributional consequences across income groups.
- Modeling simplifications:
  - Focus on household decision problem without market clearing; interest rates and housing prices are fixed.

### Main high-level finding
- LTV limits result in a larger and more persistent drop in consumption than DTI limits.
  - This outcome is mainly driven by middle-income households.
  - Under LTV, households can borrow large amounts relative to collateral value regardless of income, leading to higher leverage and larger debt burdens.
  - Under DTI, borrowing is limited by current income (affordability), reducing leverage and mitigating consumption contraction in adverse shocks.
- Policy implication:
  - Although DTI caps may be unpopular with lower income households because they limit borrowing capacity, a DTI cap is useful to mitigate potential consumption contraction in recessions and can complement LTV caps to stabilize consumption.

### Literature links (concise)
- Theoretical and empirical literature finds LTV tightening can be more contractionary than DTI tightening under some conditions; this paper contrasts by analyzing usefulness of tools without tightening under downturn scenarios and finds DTI cap more useful than LTV cap.
- Finnish-specific studies provide mixed evidence; this paper contributes by analyzing negative economic scenarios with a heterogeneous-agent model and finding DTI cap more useful in mitigating consumption reduction than LTV cap.

---

### MODEL

### Framework and assumptions
- Continuous time heterogeneous agent model; interest rate and housing prices are fixed.
- Agents indexed by liquid assets b, housing h, and labor productivity z.
- Labor productivity is idiosyncratic, uninsurable, and follows an exogenous three-state Markov process z_t ∈ {z1, z2, z3}.
- Two asset types: liquid assets and housing. Liquid assets can be used to save or borrow at interest rate r; liquid asset borrowings must respect borrowing limits determined by either LTV or DTI.
- Housing choice restricted to either no house (h_t = 0) or owning housing of at least minimum size h_min: h_t ∈ {0, [h_min, ∞)}.
- Household instantaneous utility is quasi-linear: q = c + f(h), with f(h) = η(h − h_min)^α; overall utility U(c,h) = u(q) = q^{1−σ}/(1−σ) with CRRA elasticity 1/σ and discount rate ρ.
- Budget constraint: ḃ_it + p_t ḣ_it = z_it + r b_it − c_it.

### Borrowing constraints (macroprudential tools)
- LTV cap:
  - Borrowing limit tied to collateral value: −b_it ≤ θ p_t h_it, with θ ∈ [0,1].
  - Down payment must be greater than the 1 − θ fraction of housing value.
  - Define φ = 1/(1 − θ). Admissible housing choices under LTV: H_LTV(a,z) = {h: p h ≤ φ a} ∩ {0, [h_min, ∞)}.
- DTI cap:
  - Borrowing limit tied to current income: −b_it ≤ γ z_it, with constant γ.
  - High-income households can access larger borrowing limits.
  - Admissible housing choices under DTI: H_DTI(a,z) = {h: p h ≤ γ z + a} ∩ {0, [h_min, ∞)}.
- Tools are not imposed simultaneously in the model; scenarios compare either LTV limit or DTI limit.

### Equilibrium characterization (steady state)
- Stationary equilibrium characterized by Hamilton-Jacobi-Bellman (HJB) and Kolmogorov Forward (KF) equations.
- HJB (value function v(a,z), k ∈ {LTV, DTI}):
  - ρ v(a,z) = max_{c,h ∈ H_k} U(c,h) + v_a(a,z) (z + r (a − p h) − c) + ∑_{z′≠z} λ(z,z′) {v(a,z′) − v(a,z)}.
- Using quasi-linear utility and q = c + F(a,z) separation:
  - ρ v(a,z) = max_q u(q) + v_a(a,z) (z + F(a,z) − q) + ∑_{z′≠z} λ(z,z′) {v(a,z′) − v(a,z)}.
  - F(a,z) = max_{h ∈ H_k} f(h) + r (a − p h).
- Kolmogorov forward equation for stationary density g(a,z):
  - 0 = d/da [ s(a,z) g(a,z) ] + ∑_{z′≠z} { −λ(z,z′) g(a,z) + λ(z′,z) g(a,z′) }.
  - Optimal saving policy s(a,z) = z + r (a − p h) − c.

### Computational notes and simplifications
- Continuous-time Poisson switching for productivity with arrival rate λ(z,z′).
- If households cannot afford minimum housing, h = 0; if constrained by borrowing limits, housing choice determined by LTV or DTI; otherwise choose h_opt.
- Macroprudential constraints are imposed continuously (not only at origination).

---

### 3.4 Solutions in Transition Dynamic — Calibration and Simulation

### Income process and shock implementation
- Productivity (income) discretized into three states with values:
  - z1 = 0.05
  - z2 = 0.2
  - z3 = 0.4
- Baseline transition probabilities (symmetric, no direct low↔high jumps):
  - λLM = 0.057
  - λMH = 0.063
- Negative shock experiment:
  - λLM increased from 0.057 to 0.102 to calibrate a steady-state increase in the low-income share by 3 percent (a stress-test level used in IMF (2023)).
  - The shock is implemented as an increased likelihood of moving to the low-income state from middle income for three years; afterwards transition probabilities return to baseline.
- Households anticipate the three-year recession once the shock at t=0 is realized; idiosyncratic income risk continues.

### Key parameter choices (Table 1 values)
- σ = 2
- α = 0.26
- η = 0.20
- ρ = 0.057
- θ = 0.85
- γ = 5
- h_min = 1
- r = 0.025
- p = 1
- z1 = 0.05
- z2 = 0.2
- z3 = 0.4
- λLM = 0.057
- λMH = 0.063
- Note: θ = 0.85 corresponds to a maximum LTV of 85 percent; γ = 5 implies households can borrow up to 5 times current annual income.

### Simulation setup and outcomes of interest
- Negative economic scenario: unexpected increase in transition probability from middle to low income for three years.
- Outcomes of interest:
  - consumption responses relative to steady state,
  - housing policy functions,
  - consumption policy functions,
  - household distributions across wealth and income states.

---

### Simulation results — Housing, Consumption, Distribution, and Policy Comparison

### Housing and consumption policy patterns
- Housing policy:
  - LTV scenario: housing choice exhibits three regions driven by minimum size restriction, the 15 percent down-payment implied by LTV cap, and an optimal housing region independent of assets (quasi-linear utility). Housing choice does not depend on income state under LTV because borrowing is capped by collateral size.
  - DTI scenario: housing choice also has three regions but clearly depends on income state because borrowing capacity scales with income; high-income households can afford larger housing even with similar asset levels.
- Consumption policy:
  - Consumption increases with income and assets in both LTV and DTI scenarios.
- Distribution:
  - Lower-income households are more likely to be constrained by LTV and DTI limits; higher-income households tend to hold optimal housing.

### Quantitative consumption response to the negative shock
- Peak consumption declines at t=0:
  - DTI scenario: around -3.2 percent
  - LTV scenario: around -3.8 percent
- Persistence at t=6:
  - DTI scenario: around -1.5 percent
  - LTV scenario: around -2.5 percent
- Debt service-to-income ratios:
  - LTV scenario: 0.29
  - DTI scenario: 0.09
- Explanation:
  - The LTV scenario exhibits higher leverage; higher debt service-to-income forces larger consumption reductions when aggregate income falls.

### Income-level contributions to aggregate consumption decline (in percent)
- Low income: LTV = 1.5, DTI = 2.5
- Middle income: LTV = 83.5, DTI = 82.7
- High income: LTV = 15.0, DTI = 14.8
- Interpretation:
  - Middle-income households account for over 80 percent of the contraction in both scenarios; they drive the bulk of the aggregate consumption decline.

### Policy implications and mechanisms
- Main finding:
  - Imposing a DTI cap mitigates the reduction in aggregate consumption following an unexpected negative income shock relative to an LTV cap.
- Mechanism:
  - LTV allows borrowing proportional to collateral irrespective of income, encouraging higher leverage and larger mortgages; this amplifies consumption declines in recessions because debt service burdens increase.
  - DTI constrains borrowing by income, limiting highly leveraged households and thereby cushioning consumption in downturns.
- Policy takeaway:
  - DTI can be effective in mitigating consumption contractions during recessions by limiting the number of highly leveraged households; LTV alone may be less effective at preventing high leverage across income groups.

### Caveats and limitations
- Model simplicity:
  - The heterogeneous-agent partial equilibrium model may not capture full complexity of the Finnish economy; results are illustrative of mechanisms rather than precise forecasts.
- General equilibrium effects are ignored:
  - Reduced consumption can lower firm output and wages.
  - Interest-rate changes and housing price feedbacks are not modeled and could amplify or alter results.
- Policy environment nuance:
  - Finnish authorities issued a nonbinding recommendation (June 2022, in force January 1, 2023) to originators to keep DSTI under stressed conditions at or below 60 percent; this could limit leverage even without explicit DTI regulation.
- Welfare and long-run effects:
  - Welfare and long-run utility effects of DTI in normal times are not fully assessed; imposing DTI may alter home purchase decisions and household utility, requiring more comprehensive calibration.

---

### Appendix I — Robustness to alternative calibration

### Alternative calibration
- LTV parameter θ = 0.80 (households can use 80 percent of housing value as collateral).
- DTI parameter γ = 6 (households can borrow up to 6 times their income).
- Interpretation: LTV is tighter while DTI is looser compared to the baseline.

### Main robustness result
- Even with this different calibration, DTI is still effective in mitigating the decline in consumption.
- Robustness note:
  - Similar results (larger consumption contraction in LTV scenario) can be obtained under other parameter settings in interest rates as long as the calibration strategy is the same.

*Source: IMF Working Paper — Household Debt and Borrower-Based Measures in Finland (INTRODUCTION section and chapter "3.4 Solutions in Transition Dynamic" and Appendix I).*

### INTRODUCTION __________________________________________________________________ 3

### INTRODUCTION

### Context and motivation
- Systemic risks such as sharply rising indebtedness and risk taking can accumulate and amplify adverse financial shocks; vulnerabilities in the housing sector repeatedly pose significant concerns.
- Macroprudential policy tools available in the household sector include sectoral capital requirements and caps on loan-to-value ratios (LTV), debt-to-income ratios (DTI), and debt-service-to-income ratios (DSTI).
- Household debt-to-disposable income ratio in Finland rose from 111 percent in 2010 to 133 percent in 2021.
- Finland’s maximum LTV (loan-to-collateral, LTC) ratio has been reduced to 85 percent, except for first-time home buyers; income-related borrower-based measures like DTI are not currently available as macroprudential tools despite recent proposals.

### Research question and approach
- Core question: Would introducing a cap on the DTI ratio be useful in further mitigating risks associated with mortgage loans in Finland?
- Method: Quantitative comparison of LTV caps and DTI caps on household consumption responses to negative economic scenarios using a simulation based on a heterogeneous agent household model.
- Rationale for method:
  - Theoretical model allows counterfactual analysis not feasible with past data series.
  - Heterogeneous agent model incorporates household inequality and distributional consequences, useful for assessing potential impacts on lower income households and first-time buyers.
- Modeling simplification: focus on household decision problem without market clearing; interest rates and housing prices are fixed.

### Main finding (high-level)
- LTV limits result in a larger and more persistent drop in consumption than DTI limits.
  - This outcome is mainly driven by middle-income households.
  - Under LTV, households can borrow large amounts relative to collateral value regardless of income, leading to higher leverage and larger debt burdens.
  - Under DTI, borrowing is limited by current income (affordability), reducing leverage and mitigating consumption contraction in adverse shocks.
- Policy implication: Although DTI caps may be unpopular with lower income households because they limit borrowing capacity, a DTI cap is useful to mitigate potential consumption contraction in recessions and can complement LTV caps to stabilize consumption.

---

### Literature links (concise)
- Theoretical literature examines macroprudential tools in house price boom-bust episodes; relevant studies: Alpanda and Zubairy (2017), Chen et al. (2020), Ferrero, Harrison, and Nelson (2018), Greenwald (2018), Justiniano et al. (2019), Ingholt (2022), Lambertini, Mendicino, and Punzi (2013).
  - Chen et al. (2020): LTV tightening more contractionary than DTI tightening when debt is high and monetary policy cannot accommodate.
  - This paper contrasts by analyzing usefulness of tools without tightening under downturn scenarios and finds DTI cap more useful than LTV cap.
- Empirical literature finds macroprudential tools effective in reducing household credit and consumption (e.g., Alam et al. (2019)); this paper uses a theoretical simulation to investigate mechanisms during financial turmoil.
- Finnish-specific studies:
  - Kärkkäinen and Nyholm (2021): replacing LTV with DTI has only minor long-run macro effects in a two-agent model.
  - Eerola, Lyytikäinen, and Ramboer (2021): LTV limit impact and potential DTI binding for young, single, highly educated first-time buyers in urban areas.
  - This paper’s contribution: analysis of benefits in negative economic scenarios using a heterogeneous agent model, finding DTI cap more useful in mitigating consumption reduction than LTV cap.

---

### MODEL

### Framework and assumptions
- Continuous time heterogeneous agent model focused on household decision problem; interest rate and housing prices are fixed.
- Agents indexed by liquid assets b, housing h, and labor productivity z.
- Labor productivity is idiosyncratic, uninsurable, and follows an exogenous three-state Markov process z_t ∈ {z1, z2, z3}.
- Two asset types: liquid assets and housing. Liquid assets can be used to save or borrow at interest rate r; liquid asset borrowings must respect borrowing limits determined by either LTV or DTI.
- Housing choice restricted to either no house (h_t = 0) or owning housing of at least minimum size h_min: h_t ∈ {0, [h_min, ∞)}.
- Household instantaneous utility is quasi-linear: q = c + f(h), with f(h) = η(h − h_min)^α; overall utility U(c,h) = u(q) = q^{1−σ}/(1−σ) with CRRA elasticity 1/σ and discount rate ρ.
- Budget constraint: ḃ_it + p_t ḣ_it = z_it + r b_it − c_it, where p_t is housing price and over-dot denotes time derivative.

### Household problem specifics
- Net worth a_it = b_it + p_t h_it.
- Quasi-linear utility separates housing choice from consumption-saving problem, facilitating computational solution.
- Optimal housing (unconstrained) given by: h_opt = h_min + (η α / (r p))^α.
- Computational solution uses upwind scheme following Achdou et al. (2022).

### Borrowing constraints (macroprudential tools)
- LTV cap:
  - Borrowing limit tied to collateral value: −b_it ≤ θ p_t h_it, with θ ∈ [0,1].
  - Down payment must be greater than the 1 − θ fraction of housing value.
  - Define φ = 1/(1 − θ). Admissible housing choices under LTV: H_LTV(a,z) = {h: p h ≤ φ a} ∩ {0, [h_min, ∞)}.
- DTI cap:
  - Borrowing limit tied to current income: −b_it ≤ γ z_it, with constant γ.
  - High-income households can access larger borrowing limits.
  - Admissible housing choices under DTI: H_DTI(a,z) = {h: p h ≤ γ z + a} ∩ {0, [h_min, ∞)}.
- Tools are not imposed simultaneously in the model; scenarios compare either LTV limit or DTI limit.

### Equilibrium characterization (steady state)
- Stationary equilibrium characterized by Hamilton-Jacobi-Bellman (HJB) and Kolmogorov Forward (KF) equations.
- HJB (value function v(a,z), k ∈ {LTV, DTI}):
  - ρ v(a,z) = max_{c,h ∈ H_k} U(c,h) + v_a(a,z) (z + r (a − p h) − c) + ∑_{z′≠z} λ(z,z′) {v(a,z′) − v(a,z)}.
- Using quasi-linear utility and q = c + F(a,z) separation:
  - ρ v(a,z) = max_q u(q) + v_a(a,z) (z + F(a,z) − q) + ∑_{z′≠z} λ(z,z′) {v(a,z′) − v(a,z)}.
  - F(a,z) = max_{h ∈ H_k} f(h) + r (a − p h).
- Kolmogorov forward equation for stationary density g(a,z):
  - 0 = d/da [ s(a,z) g(a,z) ] + ∑_{z′≠z} { −λ(z,z′) g(a,z) + λ(z′,z) g(a,z′) }.
  - Optimal saving policy s(a,z) = z + r (a − p h) − c.

### Computational notes and model simplifications
- Households switch productivity states according to a Poisson process with arrival rate λ(z,z′).
- If households cannot afford minimum housing, h = 0; if constrained by borrowing limits, housing choice determined by LTV or DTI; otherwise choose h_opt.
- Macroprudential constraints are imposed continuously (not only at origination) in the model; calibration and negative income shock scenarios imply this may be costly for consumption in the DTI cap, yet DTI remains useful to mitigate consumption drops.

---

*IMF Working Paper — Household Debt and Borrower-Based Measures in Finland (INTRODUCTION section).*

### 3.4 Solutions in Transition Dynamic

### 3.4 Solutions in Transition Dynamic

### Model and transition dynamics
- The dynamic Hamilton–Jacobi–Bellman (HJB) equation and Kolmogorov–Forward (KF) equations describe household optimal choices over assets, housing, and consumption under idiosyncratic income risk and aggregate shocks. The model assumes:
  - The aggregate shock is unexpected at t=0; once realized households know the future path of aggregate variables.
  - The only ongoing uncertainty after the shock is idiosyncratic income shocks governed by discrete productivity states and transition probabilities.
  - Housing choice enters via collateral and housing utility; a minimum housing size constraint is present in the baseline calibration.

### Calibration and income process
- Productivity (income) is discretized into three states: low z1, middle z2, and high z3.
- Income state values used: z1 = 0.05, z2 = 0.2, z3 = 0.4.
- The baseline transition probabilities (symmetric, no direct low↔high jumps) are:
  - λLM (between low and middle) = 0.057
  - λMH (between middle and high) = 0.063
- For the negative shock experiment, λLM is increased from 0.057 to 0.102 to calibrate a steady-state increase in the low-income share by 3 percent (a stress-test level used in IMF (2023)).
- The shock is implemented as an increased likelihood of moving to the low-income state from middle income for three years; afterward transition probabilities return to baseline.

### Key parameter choices (Table 1 values)
- σ (Intertemporal elasticity of substitution) = 2
- α (Preference on housing) = 0.26
- η (Preference on housing) = 0.20
- ρ (Discount factor) = 0.057
- θ (LTV parameter) = 0.85
- γ (DTI parameter) = 5
- h_min (Minimum amount of housing) = 1
- r (Interest rate) = 0.025
- p (Housing price) = 1
- z1 (Productivity for low income) = 0.05
- z2 (Productivity for middle income) = 0.2
- z3 (Productivity for high income) = 0.4
- λLM (Transition probability between low and middle income) = 0.057
- λMH (Transition probability between middle and high income) = 0.063

Additional calibration notes:
- LTV parameter θ = 0.85 corresponds to a maximum LTV of 85 percent in Finnish mortgage regulation (down payment at least 15 percent).
- DTI parameter γ = 5 implies households can borrow up to 5 times current annual income; comparable to a recommendation of up to 450 percent in gross annual income from a Ministry of Finance working group.
- The housing preference parameter α is set to target a share of constrained households of 0.40 in the LTV scenario.
- η is set to match a housing wealth-to-income ratio of 4.1 percent.

### Simulation setup
- Negative economic scenario: unexpected increase in transition probability from middle to low income for three years, mimicking recessionary unemployment rises.
- Households anticipate the three-year recession once the shock at t=0 is realized; idiosyncratic income risk continues.
- Outcomes of interest: consumption responses relative to steady state, housing policy functions, consumption policy functions, and household distributions across wealth and income states.

### Simulation results — housing, consumption, and distributional patterns
- Housing policy:
  - LTV scenario: housing choice exhibits three regions driven by minimum size restriction, the 15 percent down-payment implied by LTV cap, and an optimal housing region independent of assets (quasi-linear utility). Housing choice does not depend on income state under LTV because borrowing is capped by collateral size.
  - DTI scenario: housing choice also has three regions but clearly depends on income state because borrowing capacity scales with income; high-income households can afford larger housing even with similar asset levels.
- Consumption policy: consumption increases with income and assets in both LTV and DTI scenarios.
- Distribution: households endogenously choose home ownership for utility; lower-income households are more likely to be constrained by LTV and DTI limits; higher-income households tend to hold optimal housing.

### Quantitative consumption response to the negative shock (Figure 3)
- Peak consumption declines at t=0:
  - DTI scenario: around -3.2 percent
  - LTV scenario: around -3.8 percent
- Persistence at t=6:
  - DTI scenario: around -1.5 percent
  - LTV scenario: around -2.5 percent
- Explanation: the LTV scenario exhibits higher leverage; debt service-to-income ratios are:
  - LTV scenario: 0.29
  - DTI scenario: 0.09
  - Higher interest-relief burden under LTV forces larger consumption reductions when aggregate income falls.

### Income-level contributions to consumption decline (policy-function contributions)
- Contributions (in percent) to aggregate consumption decline:
  - Low income: LTV = 1.5, DTI = 2.5
  - Middle income: LTV = 83.5, DTI = 82.7
  - High income: LTV = 15.0, DTI = 14.8
- Interpretation: middle-income households account for over 80 percent of the contraction in both scenarios. Middle-income households reduce consumption both by anticipating higher low-income risk (endogenous policy-function adjustments) and when shocks are realized.

### Policy implications and comparison of LTV vs DTI
- Main finding: imposing a DTI cap mitigates the reduction in aggregate consumption following an unexpected negative income shock relative to an LTV cap.
- Mechanism:
  - LTV allows borrowing proportional to collateral irrespective of income, encouraging higher leverage and larger mortgages; this amplifies consumption declines in recessions because debt service burdens increase.
  - DTI constrains borrowing by income, limiting highly leveraged households and thereby cushioning consumption in downturns.
- Policy takeaway: DTI can be effective in mitigating consumption contractions during recessions by limiting the number of highly leveraged households; LTV alone may be less effective at preventing high leverage across income groups.

### Caveats and limitations
- Model simplicity: the heterogeneous-agent partial equilibrium model may not capture full complexity of the Finnish economy; results are illustrative of mechanisms rather than precise forecasts.
- General equilibrium effects are ignored:
  - Reduced consumption can lower firm output and wages.
  - Interest-rate changes and housing price feedbacks are not modeled and could amplify or alter results.
- Policy environment nuance:
  - Finnish authorities issued a nonbinding recommendation (June 2022, in force January 1, 2023) to originators to keep DSTI under stressed conditions at or below 60 percent; this could limit leverage even without explicit DTI regulation.
- Welfare and long-run utility effects of DTI in normal times are not fully assessed; imposing DTI may alter home purchase decisions and household utility, requiring more comprehensive calibration to household housing preferences.

### Conclusion
- The calibrated heterogeneous-agent model, using Finnish data and a three-state income discretization, shows that an unexpected three-year rise in the probability of moving to low income generates a larger and more persistent consumption decline under an LTV cap than under a DTI cap.
- DTI cap reduces leverage and therefore mitigates the aggregate consumption contraction following negative income shocks, with middle-income households driving most of the decline.
- Further research should incorporate general equilibrium feedbacks and a fuller welfare analysis of DTI effects in normal times.

*Source: IMF staff estimates, from chapter "3.4 Solutions in Transition Dynamic" of the provided content unit.*

### Appendix I. Results with Other Calibration

### Appendix I. Results with Other Calibration

### Calibration change
- LTV parameter 휃 = 0.80 (households can use 80 percent of housing value as collateral).
- DTI parameter 훾 = 6 (households can borrow up to 6 times their income).
- Interpretation: LTV is tighter while DTI is looser compared to the baseline scenario.

### Main results
- Even with this different calibration, DTI is still effective in mitigating the decline in consumption.
- Figure A-1 (in the source) illustrates consumption responses under the different calibration and supports the above result.

### Robustness note
- Even if other parameter settings in interest rates are used, similar results can be obtained (larger consumption contraction in LTV scenario) as long as the calibration strategy is the same.

*Source: Appendix I. Results with Other Calibration, IMF Working Paper chapter.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023262-print-pdf.pdf_