## 1. Prediction Power of the Fiscal Reaction Function

## Source details

**Canonical URL:** [1. Prediction Power of the Fiscal Reaction Function](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020269-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020269-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020269-print-pdf.pdf.json)

---

### Introduction
- The Dutch fiscal balance is projected to deteriorate by an unprecedented magnitude due to fiscal measures and automatic stabilizers responding to the COVID-19 pandemic.
- Key observed and projected fiscal figures:
  - Fiscal deficit projected by the Dutch Ministry of Finance in its June update: 8.7 percent of GDP in 2020 (from a surplus of 1.7 percent in 2019).
  - Public debt projected to rise to 63 percent of GDP in 2020 (from below 50 percent in 2019).
  - Some forecasts (released after this analysis) project a fiscal deficit of 7.6 percent of GDP, and public debt at 60 percent of GDP in 2020.
- Motivation and key questions:
  - Assess risks to fiscal and debt sustainability under alternative recovery scenarios and uncertain pandemic evolution.
  - How much fiscal room remains to continue support without compromising sustainability? What medium-term fiscal adjustment paths would safeguard public finances while supporting potential growth?

### Debt dynamics and interest payments
- Limitations of standard identity:
  - Standard identity: Debt_t - Debt_{t-1} = r_{t-1} * Debt_{t-1} + primary deficit_t.
  - Standard approach assumes full rollover each period and often uses average historical interest rates rather than interest rates on new issuance; ignores that only a fraction of debt matures each period.
- Augmented debt dynamics features:
  - Assume a constant fraction (1 − γ) of existing debt matures each period.
  - For simulations, all debt instruments assumed maturity of 5 years, implying γ = 0.8.
  - Closed-form debt dynamics (as in source): B_t = R_t B_{t−1} + (R_t − γ)/(1−γ) D_t ⇔ B_t = (1 + r_t) B_{t−1} + (1−γ + r_t)/(1−γ) D_t.
- Interest payments modeling:
  - Interest payments recursion (as in source): IP_t = r_{t−1} [ (1−γ) B_{t−2} + D_{t−1} ] + γ IP_{t−1}.
  - New issuances pay interests at their new rates; outstanding past issuances pay at their original rates.
- Financing accounting definitions preserved:
  - F_t = (1−γ) B_{t−1} + D_t.
  - N_t = F_t / H_t ⇔ N_t H_t = F_t.
  - R_t = γ H_t + (1−γ) H_t ⇔ H_t = (1−γ)/(R_t − γ).
  - B_t = γ B_{t−1} + N_t.

### Fiscal reaction function — specification and predictive performance
- Specification estimated over 1980–2018 (annual data):
  - d_t = 0.49 d_{t-1} − 0.49 b_{t-1} + 0.022 y_{t-1} − 0.32 p_{t-1} + ε_t
  - Standard errors in parentheses: (0.09), (0.1), (0.01), (0.14) respectively.
  - Variables: d (primary deficit in percent of GDP), y (real GDP growth), b (debt-to-GDP ratio), p (inflation).
- Findings:
  - All coefficients statistically significant at standard thresholds with expected signs.
  - The estimated reaction function broadly reflects Dutch fiscal policy patterns with a relatively small residual.
  - Forward-looking simulations also consider alternative specifications (e.g., more strongly countercyclical responses).

### Macro (non-fiscal) framework, VAR, and shocks identification
- Core macro variables and estimation:
  - Real GDP growth, inflation, and the interest rate on public debt used as core macro variables.
  - Unrestricted VAR estimated using quarterly data over 1999Q1-2019Q3; starts with two lags and is pruned for parsimony.
  - Transformations: first differences of logs for y and p; r included as difference from a linear trend to ensure stationarity.
- Interest rate data constraint:
  - Due to limited Dutch long-term bond data, German long-term bond interest rate used as reference; Netherlands rate modeled as DEU rate plus a risk premium (RP).
- Distinction between cyclical and structural shocks:
  - Focus on cyclical shocks over the 10-year forecast horizon.
  - Decompose variables X into deviations from long-term trends estimated with a Kalman filter; estimate cyclical shock vector Ω via equation (6) using quarterly data over 1980Q1-2019Q3.

### Simulation algorithm and shock draws
- Stochastic shock generation and implementation:
  - At each forecast period t+f, sets of three shocks (real GDP growth, inflation, interest rate) are drawn randomly from historical estimated shocks.
  - For each t+f, 5000 sets of shocks are drawn and applied to baseline forecasts.
  - Shocks are added to trend values: x_{t+f} = x_{t+f} + ω_{t+f}, where ω is an estimated shock.
  - This preserves historical correlations and VAR-implied persistence; macro variables are forecasted quarterly and annualized to feed the fiscal framework.
- Distributional features and tail probabilities reported:
  - 5 percent probability that real GDP growth falls below −4 percent over the forecast horizon.
  - 5 percent probability that interest rate reaches 4 percent.
  - 15 percent probability that inflation turns negative.
  - Magnitude of potential downturns could be larger than observed in the GFC, though with low probability.
- Outputs:
  - Fan charts represent frequency distributions of simulated fiscal and debt trajectories; color shading corresponds to half-deciles of the distribution.

### Risks to debt and fiscal sustainability — main findings
- Simulation setup notes:
  - Simulations use 2022 as the starting point (post-pandemic scenarios), while fiscal policy paths begin in 2020 and diverge from 2021 onward; hence initial fiscal positions for simulations (starting 2022) are not identical.
  - Use of a 10-year forecast horizon and a wide set of shock combinations increases scenario coverage.
- Overall assessment:
  - Risks to debt and fiscal sustainability appear to be contained.
  - Even under adverse scenarios, a high estimated probability that debt remains below the IMF’s debt sustainability risk threshold.
  - Containment attributed largely to pre-pandemic strength of the fiscal position and explicit modeling of effective interest rates and debt maturity structure.

### Baseline forecasts, scenarios, and assumptions
- Baseline inputs:
  - Baseline forecasts built around the April 2020 IMF WEO projections adjusted to reflect information available as of June 2020.
  - As of June 2020, external forecasts project a headline fiscal deficit ranging between 6.4 and 13 percent of GDP in 2020.
  - As of June 2020, external forecasts project government debt at 61 to 64 percent of GDP in 2020.
- Scenario design (three fiscal policy paths):
  - Optimistic scenario: private consumption and investment recover quickly and strongly; sharp short-term fiscal consolidation.
  - Weaker recovery scenario: weaker-than-expected recovery of domestic demand; more gradual phase-out of fiscal support.
  - Pessimistic recovery scenario: more persistent pandemic impact; prolonged fiscal policy measures required.
- Two layers of uncertainty:
  - Layer 1: three scenarios with distinct baseline paths for key fiscal variables capturing alternative pandemic outcomes.
  - Layer 2: within each scenario, thousands of random simulations using shocks drawn from historical Netherlands experience prior to the pandemic.

### Scenario results and key statistics
- Medium-term budget balance target scenario (A):
  - Assumptions and outcomes:
    - Strong recovery by end-2020 and sharp rebound in 2021.
    - Balanced budget in 2025.
    - Debt-to-GDP ratio falls slightly below the 60 percent SGP limit over five years.
  - Probabilities and projections:
    - Probability to breach 3 percent deficit threshold over forecast horizon is 5 percent (or below).
    - Baseline debt-to-GDP ratio declines gradually, falling below 50 percent by 2030.
    - Probability that debt exceeds 60 percent of GDP falls to about 10 percent by end of forecast horizon.
    - Probability that debt-to-GDP exceeds 70 percent is below 5 percent.
  - Selected exact trajectory (Great lockdown, medium-term budget balance scenario):
    - Real GDP growth: 2019: 1.8; 2020: −7.7; 2021: 5.0; 2022: 2.0; 2023: 1.7; 2024: 1.5; 2025: 1.5.
    - Inflation: 2019: 2.7; 2020: 0.5; 2021: 1.2; 2022: 1.4; 2023: 1.8; 2024: 1.9; 2025: 2.0.
    - Headline fiscal balance (% GDP): 2019: 1.7; 2020: −10.1; 2021: −4.0; 2022: −2.2; 2023: −1.3; 2024: −0.5; 2025: 0.1.
    - Debt-to-GDP ratio: 2019: 48.3; 2020: 62.3; 2021: 62.7; 2022: 62.7; 2023: 61.9; 2024: 60.3; 2025: 58.1.
- Medium-term debt objective scenario (B):
  - Target debt-to-GDP ratio of about 68 percent in 2025 (peak observed in 2014).
  - Probability to breach 3 percent of GDP deficit limit over 2025-2030 is about 25 percent.
  - There is a 15 to 35 percent probability that government debt will fall below 60 percent of GDP by 2030.
  - Probability that debt-to-GDP exceeds 84 percent by 2030 is only 5-10 percent.
  - Selected trajectory (debt objective scenario):
    - Headline fiscal balance (% GDP): 2019: 1.7; 2020: −10.1; 2021: −6.1; 2022: −4.3; 2023: −3.4; 2024: −2.6; 2025: −2.1.
    - Debt-to-GDP ratio: 2019: 48.3; 2020: 62.3; 2021: 64.8; 2022: 66.9; 2023: 68.1; 2024: 68.4; 2025: 68.1.
- Protracted fiscal consolidation scenario (C):
  - More gradual consolidation calibrated to be similar to post-GFC trend adjustment.
  - Baseline fiscal deficit most likely remains above 3 percent of GDP by end of forecast horizon.
  - Approximately 25 percent probability that fiscal deficit falls below 3 percent between 2025-2030.
  - Baseline debt-to-GDP ratio increases to around 85 percent by 2030.
  - Probability that debt falls below 60 percent over forecast horizon is below 5 percent.
  - Probability of exceeding 100 percent of GDP is 10 percent by 2030.
  - Even in this adverse scenario, in 90 percent of simulations Dutch debt ratios remained below 2019 levels observed in France (99 percent) and Belgium (98.5 percent) and below the average for G20 advanced economies in 2019 (113 percent).
  - Selected trajectory (protracted consolidation scenario):
    - Headline fiscal balance (% GDP): 2019: 1.7; 2020: −10.1; 2021: −8.0; 2022: −6.4; 2023: −5.6; 2024: −4.7; 2025: −4.2.
    - Debt-to-GDP ratio: 2019: 48.3; 2020: 62.3; 2021: 66.7; 2022: 70.9; 2023: 74.0; 2024: 76.2; 2025: 77.8.
- Summary probabilities across scenarios:
  - Probability of exceeding debt level of around 70 percent of GDP at least once in next ten years:
    - Below 10 percent under the budget balance scenario.
    - Around 55 percent under the debt objective scenario.
  - Probability fiscal deficit exceeds 5 percent of GDP (largest deficit observed during GFC and European debt crisis):
    - Around 1 percent under the budget balance scenario.
    - 5 percent under the debt objective scenario.

### Methodology and model specifics
- Simulation approach:
  - Thousands of stochastic simulations drawing shocks from historical Dutch experience pre-pandemic; fan charts from hundreds/thousands of random drawings.
- Debt dynamics and accounting preserved as in source (variables and exact equations listed under "Debt dynamics and interest payments").
- Kalman filter augmentation:
  - Equations (2), (3), (4) augmented with Kalman filter to simulate observed series and minimize errors ε(X).
  - Netherlands rate modeled as DEU rate plus a risk premium (RP) due to limited NLD long-term bond yield data.
- Interest payments recursion explicitly used in simulations (equation (7) in source): IP_t = r_{t−1} [ (1−γ) B_{t−2} + D_{t−1} ] + γ IP_{t−1}.

### Policy implications and recommendations
- Overall conclusion: Risks to Dutch fiscal and debt sustainability remain low across scenarios, including under severe stress.
- Policy guidance:
  - If COVID-19 impact is short-lived, fiscal policy should continue to focus on medium-to-long-term targets and maintaining a strong fiscal position to strengthen resilience.
  - The SGP deficit and debt limits remain useful benchmarks over the medium-term.
  - Space would remain available over at least two successive legislatures to allow fiscal policy to support structural investment in the medium-term.
  - Moderate fiscal deficits that support long-term growth would not jeopardize long-term fiscal sustainability if deficits support long-term growth and real GDP growth exceeds the real cost of additional government debt.
  - Targeted measures to enhance productivity growth would boost potential output growth and contribute to long-term fiscal sustainability.

*Source: wpiea2020269-print-pdf*

### 1. Prediction Power of the Fiscal Reaction Function ________________________________8

### 1. Prediction Power of the Fiscal Reaction Function

### Introduction
- The Dutch fiscal balance is projected to deteriorate by an unprecedented magnitude due to fiscal measures and automatic stabilizers responding to the COVID-19 pandemic.
- Key observed and projected fiscal figures:
  - Fiscal deficit projected by the Dutch Ministry of Finance in its June update: 8.7 percent of GDP in 2020 (from a surplus of 1.7 percent in 2019).
  - Public debt projected to rise to 63 percent of GDP in 2020 (from below 50 percent in 2019).
  - Some forecasts (released after this analysis) project a fiscal deficit of 7.6 percent of GDP, and public debt at 60 percent of GDP in 2020.
- Context and policy background:
  - The Netherlands entered the crisis with a strong fiscal position and fiscal space built after the global financial crisis (GFC) and the European debt crisis.
  - Automatic stabilizers and a domestic fiscal framework have allowed limited discretionary fiscal policy historically, while taking into account long-term trends such as population aging.
- Motivation for analysis:
  - The Great Lockdown raises the need to assess risks to fiscal and debt sustainability under alternative recovery scenarios and uncertain pandemic evolution.
  - Key questions: How much fiscal room remains to continue support without compromising sustainability? What medium-term fiscal adjustment paths would safeguard public finances while supporting potential growth?

### Debt dynamics and interest payments
- Modeling improvements over the standard equation (1):
  - Standard identity: Debt_t - Debt_{t-1} = r_{t-1} * Debt_{t-1} + primary deficit_t (equation 1).
  - Limitations of the standard approach: assumes full rollover each period and often uses average historical interest rates rather than interest rates on new issuance; ignores that only a fraction of debt matures each period.
- Augmented debt dynamics:
  - Assumes a constant fraction (1 - γ) of existing debt matures each period.
  - Debt dynamics formalized as (equation 2):
    - 1/(1+γ) * [ (1+ r_{t-1}) * B_{t-1} + primary deficit_t ]  (presented in text as a function with parameters B, r, D, γ).
  - For simulations, all debt instruments assumed maturity of 5 years, implying γ = 0.8.
- Interest payments modeling:
  - Interest payments (IP) are derived to reflect effective interest rates tied to the timing and rates of debt issuances.
  - Interest payments at t are a function of interest payments at t-1, past issued debt, and primary deficit (equation 3):
    - IP_t = (1+γ)^{-1} [ IP_{t-1} + r * B_{t-1} + D_{t-1} ]  (presented in text as relationship showing dependence on past variables).
  - New issuances pay interests at their new rates; outstanding past issuances pay at their original rates.

### Fiscal reaction function
- Purpose:
  - Captures fiscal policy responses to macroeconomic changes given debt and fiscal sustainability objectives.
- Specification and estimation:
  - Primary fiscal deficit (d) is modeled as a function of its own lag, lagged debt-to-GDP ratio (b), real GDP growth (y), and inflation (p).
  - Estimated over 1980–2018 using annual data; all coefficients statistically significant at standard thresholds with expected signs.
  - Estimated reaction function (equation 4) presented in the source:
    - d_t = 0.49 d_{t-1} - 0.49 b_{t-1} + 0.022 y_{t-1} - 0.32 p_{t-1} + ε_t
    - Standard errors reported in parentheses: (0.09), (0.1), (0.01), (0.14) respectively.
  - Variables d, y, b, and p denote primary deficit in percent of GDP, real GDP growth, debt-to-GDP ratio, and inflation, respectively.
- Predictive performance:
  - The estimated reaction function broadly reflects Dutch fiscal policy patterns with a relatively small residual (figure referenced).
  - The forward-looking simulation also considers alternative specifications (e.g., more strongly countercyclical responses).

### Macro (non-fiscal) framework — identifying shocks and VAR
- Key macro variables and methodology:
  - Real GDP growth, inflation, and the interest rate on public debt are used as core macro variables.
  - An unrestricted VAR is estimated to capture interactions and generate consistent projections.
  - VAR estimated using quarterly data over 1999Q1-2019Q3; specification starts with two lags and non-significant terms are dropped to produce a parsimonious model.
- VAR structure and estimated coefficients (equation 5):
  - VAR system reported with coefficient matrices and standard errors in parentheses as shown in source text.
  - Transformations: first differences of logs for y and p; r included as difference from a linear trend to ensure stationarity.
  - Interest rate data constraint: due to limited Dutch long-term bond data, German long-term bond interest rate used as reference.
- Distinction between cyclical and structural shocks:
  - Focus is on cyclical shocks over the 10-year forecast horizon.
  - Decompose variables X into deviations from long-term trends estimated with a Kalman filter; estimate cyclical shock vector Ω via equation (6) using quarterly data over 1980Q1-2019Q3.

### Simulation algorithm and shock draws
- Stochastic shock generation:
  - At each forecast period t+f, sets of three shocks (real GDP growth, inflation, interest rate) are drawn randomly from historical estimated shocks in equation (6).
  - For each t+f, 5000 sets of shocks are drawn and applied to baseline forecasts.
  - Shocks are added to trend values: x_{t+f} = x_{t+f} + ω_{t+f}, where ω is an estimated shock.
  - This preserves historical correlations and VAR-implied persistence; macro variables are forecasted quarterly and annualized to feed the fiscal framework.
- Distributional features and tail probabilities noted in simulations:
  - There is a 5 percent probability that real GDP growth falls below -4 percent over the forecast horizon.
  - There is a 5 percent probability that interest rate reaches 4 percent.
  - There is a 15 percent probability that inflation turns negative.
  - The magnitude of potential downturns could be larger than observed in the GFC, though with low probability.
- Simulation outputs:
  - Fan charts represent frequency distributions of simulated fiscal and debt trajectories; color shading corresponds to half-deciles of the distribution.

### Risks to debt and fiscal sustainability — main findings
- Simulation starting point and horizon:
  - Simulations use 2022 as the starting point (post-pandemic scenarios), while fiscal policy paths begin in 2020 and diverge from 2021 onward; hence initial fiscal positions for simulations (starting 2022) are not identical.
- Overall assessment:
  - Risks to debt and fiscal sustainability appear to be contained.
  - Even under adverse scenarios, simulations show a high estimated probability that debt remains below the IMF’s debt sustainability risk threshold.
  - Containment of risk is attributed largely to the pre-pandemic strength of the fiscal position.
- Modeling strengths relevant to findings:
  - Use of a 10-year forecast horizon (vs. standard 5 years) and a wider range of shock combinations.
  - Fiscal framework explicitly models effective interest rates related to new debt issuance and debt maturity structure, enhancing sensitivity analysis to interest rate changes.

*Source: wpiea2020269-print-pdf - 1. Prediction Power of the Fiscal Reaction Function*

### 15.      The baseline forecasts are built around the April 2020 IMF WEO projections

### 15.      The baseline forecasts are built around the April 2020 IMF WEO projections

### Baseline and data inputs
- Baseline forecasts built around the April 2020 IMF WEO projections adjusted to reflect information available as of June 2020.
- As of June 2020, external forecasts project a headline fiscal deficit ranging between 6.4 and 13 percent of GDP in 2020.
- As of June 2020, external forecasts project government debt at 61 to 64 percent of GDP in 2020.
- Macroeconomic (non-fiscal) forecasts are derived from the June 2020 IMF WEO Update.
- Medium-term fiscal projections cover 2021–2025 and are built around alternative envisaged scenarios.

### Scenario design and assumptions
- Three fiscal policy paths (scenarios) reflect contingency on pandemic impact and differing needs for fiscal support:
  - Optimistic scenario: private consumption and investment recover quickly and strongly; sharp short-term fiscal consolidation.
  - Weaker recovery scenario: weaker-than-expected recovery of domestic demand; more gradual phase-out of fiscal support.
  - Pessimistic recovery scenario: more persistent pandemic impact; prolonged fiscal policy measures required.
- Framework does not explicitly model feedback loops from fiscal policy to economic activity; purpose is to assess range of fiscal sustainability outcomes.
- Two layers of uncertainty modeled:
  - Layer 1: three scenarios with distinct baseline paths for key fiscal variables capturing alternative pandemic outcomes.
  - Layer 2: within each scenario, thousands of random simulations using shocks drawn from historical Netherlands experience prior to the pandemic; successive positive and negative shocks can push fiscal variables away from baseline.

### Scenario results and key statistics
- Medium-term budget balance target scenario (A):
  - Assumes short-lived COVID-19 impact; strong recovery by end-2020 and sharp rebound in 2021.
  - Tax deferrals lifted earlier; 2020 revenue shortfalls mostly recovered in 2021.
  - Balanced budget in 2025.
  - Debt-to-GDP ratio falls slightly below the 60 percent SGP limit over five years.
  - Simulations: probability to breach 3 percent deficit threshold over forecast horizon is 5 percent (or below).
  - Baseline debt-to-GDP ratio declines gradually, falling below 50 percent by 2030.
  - Probability that debt exceeds 60 percent of GDP falls to about 10 percent by end of forecast horizon.
  - Probability that debt-to-GDP exceeds 70 percent is below 5 percent.
- Medium-term debt objective scenario (B):
  - Target debt-to-GDP ratio of about 68 percent in 2025 (peak observed in 2014).
  - Baseline: high probability headline fiscal balance remains below 3 percent deficit in short term.
  - Probability to breach 3 percent of GDP deficit limit over 2025-2030 is about 25 percent.
  - There is a 15 to 35 percent probability that government debt will fall below 60 percent of GDP by 2030.
  - Probability that debt-to-GDP exceeds 84 percent (pre-pandemic Euro Area average) by 2030 is only 5-10 percent.
- Protracted fiscal consolidation scenario (C):
  - More gradual consolidation calibrated to be similar to post-GFC trend adjustment.
  - Baseline fiscal deficit most likely remains above 3 percent of GDP by end of forecast horizon.
  - Approximately 25 percent probability that fiscal deficit falls below 3 percent between 2025-2030.
  - Baseline debt-to-GDP ratio increases to around 85 percent by 2030.
  - Probability that debt falls below 60 percent over forecast horizon is below 5 percent.
  - Probability of exceeding 100 percent of GDP is 10 percent by 2030.
  - Even in this adverse scenario, in 90 percent of simulations Dutch debt ratios remained below 2019 levels observed in France (99 percent) and Belgium (98.5 percent) and below the average for G20 advanced economies in 2019 (113 percent).
- Summary probabilities across all scenarios (Figure 6 highlights):
  - Probability of exceeding debt level of around 70 percent of GDP at least once in next ten years:
    - Below 10 percent under the budget balance scenario.
    - Around 55 percent under the debt objective scenario.
  - Probability fiscal deficit exceeds 5 percent of GDP (largest deficit observed during GFC and European debt crisis):
    - Around 1 percent under the budget balance scenario.
    - 5 percent under the debt objective scenario.
- Tabulated historical and projected values (selected exact figures from Appendix 3):
  - Great lockdown (2019–2025) baseline real GDP growth (selected): 2019: 1.8; 2020: -7.7; 2021: 5.0; 2022: 2.0; 2023: 1.7; 2024: 1.5; 2025: 1.5.
  - Great lockdown inflation (selected): 2019: 2.7; 2020: 0.5; 2021: 1.2; 2022: 1.4; 2023: 1.8; 2024: 1.9; 2025: 2.0.
  - Medium-term budget balance scenario headline fiscal balance (% GDP): 2019: 1.7; 2020: -10.1; 2021: -4.0; 2022: -2.2; 2023: -1.3; 2024: -0.5; 2025: 0.1.
  - Medium-term budget balance scenario debt-to-GDP ratio: 2019: 48.3; 2020: 62.3; 2021: 62.7; 2022: 62.7; 2023: 61.9; 2024: 60.3; 2025: 58.1.
  - Medium-term debt objective scenario headline fiscal balance (% GDP): 2019: 1.7; 2020: -10.1; 2021: -6.1; 2022: -4.3; 2023: -3.4; 2024: -2.6; 2025: -2.1.
  - Medium-term debt objective scenario debt-to-GDP ratio: 2019: 48.3; 2020: 62.3; 2021: 64.8; 2022: 66.9; 2023: 68.1; 2024: 68.4; 2025: 68.1.
  - Protracted consolidation scenario headline fiscal balance (% GDP): 2019: 1.7; 2020: -10.1; 2021: -8.0; 2022: -6.4; 2023: -5.6; 2024: -4.7; 2025: -4.2.
  - Protracted consolidation scenario debt-to-GDP ratio: 2019: 48.3; 2020: 62.3; 2021: 66.7; 2022: 70.9; 2023: 74.0; 2024: 76.2; 2025: 77.8.

### Methodology and model specifics
- Simulation approach:
  - Thousands of stochastic simulations drawing shocks from historical Dutch experience pre-pandemic.
  - Fan charts constructed from hundreds/thousands of random drawings (equation (6) used for shock estimation).
- Debt dynamics and accounting (variables and exact equations preserved from source):
  - Variables: B (Nominal public debt, end of period); D (Nominal primary deficit); IP (Nominal interest payments); F (Gross financing needs); N (New debt issuance); R (Interest factor, i.e. 1+r); γ (Share of past debt that does not have to be rolled-over).
  - Financing needs: F_t = (1−γ) B_{t−1} + D_t    (1)
  - New debt issuance: N_t = F_t / H_t ⇔ N_t H_t = F_t    (2)
  - Bond price relationship: R_t = γ H_t + (1−γ) H_t  ⇔  H_t = (1−γ)/(R_t − γ)    (3)
  - Debt stock recursion: B_t = γ B_{t−1} + N_t    (4)
  - Debt dynamics closed form: B_t = R_t B_{t−1} + (R_t − γ)/(1−γ) D_t  ⇔  B_t = (1 + r_t) B_{t−1} + (1−γ + r_t)/(1−γ) D_t    (5)
  - Interest payments recursion: IP_t = r_{t−1} [ (1−γ) B_{t−2} + D_{t−1} ] + γ IP_{t−1}    (7)
- Bond pricing derivation and limiting expression for H_t given R_t and γ are presented in Appendix 1.
- Kalman filter augmentation:
  - Equations (2), (3), (4) augmented with Kalman filter to simulate observed series and minimize errors ε(X).
  - Due to limited Dutch long-term bond yield data, German (DEU) long-term interest rates used as reference; Netherlands (NLD) rate modeled as DEU rate plus a risk premium (RP).

### Policy implications and recommendations
- Overall conclusion: Risks to Dutch fiscal and debt sustainability remain low across scenarios, including under severe stress.
- If COVID-19 impact is short-lived, fiscal policy objectives should continue to focus on medium-to-long-term targets and maintaining a strong fiscal position to strengthen resilience.
- The SGP deficit and debt limits remain useful benchmarks over the medium-term.
- Space would remain available over at least two successive legislatures to allow fiscal policy to support structural investment in the medium-term.
- Moderate fiscal deficits (as opposed to returning to past fiscal surpluses) would not jeopardize long-term fiscal sustainability if deficits support long-term growth and real GDP growth exceeds the real cost of additional government debt.
- Targeted measures to enhance productivity growth would boost potential output growth and contribute to long-term fiscal sustainability.

*Source: wpiea2020269-print-pdf - 15.      The baseline forecasts are built around the April 2020 IMF WEO projections*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020269-print-pdf.pdf_