## 1phlea2022001

## Source details

**Canonical URL:** [1phlea2022001](https://www.imf.org/-/media/files/publications/cr/2022/english/1phlea2022001.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2022/english/1phlea2022001.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2022/english/1phlea2022001.pdf.json)

---

### Executive summary — context and overview
- The Philippines is categorized as one of the world’s most vulnerable countries to climate change and natural disasters, especially typhoons.
- All major cities and most of the population reside on the coastline, including the metropolitan Manila area where about 60 percent of economic activities take place.
- The Philippines experiences 20 typhoons per year on average.
- Exposure to transition risk is concentrated in the coal-based power generation sector; the share of coal-based power generation increased to nearly 50 percent in 2016 from a negligible amount in the 1990s.
- The country’s policies as of end-2020 are not entirely consistent with the Paris Agreement long-term temperature goal but "appear to put the country closer to achieving the goal than advanced economies and several large emerging market economies."

### Climate change stress testing approach (methodology and scenarios)
- FSAP developed a macro-scenario stress testing method linking three sub-components:
  - Climate scenario: PAGASA study and IPCC RCPs to estimate changes in typhoon frequency and intensity (windspeed and precipitation) by mid-21st century; RCP 8.5 chosen for main exercise.
  - Disaster scenario: World Bank CAT model (AIR Worldwide support) translating typhoon impacts into loss of physical capital (damage rates).
  - Macrofinancial scenario: Philippines-specific DSGE model mapping lost physical capital (treated as capital depreciation shocks) and correlated productivity shocks into macroeconomic outcomes.
- Scenarios and timing:
  - "Current scenarios" start in late 2020; "future scenarios" occur sometime in the mid-21st century.
  - For each disaster likelihood, three sub-scenarios: baseline without disaster, once in 25-years disaster, and once in 500-years disaster.
  - Two baselines considered: with pandemic (October 2020 WEO) and without pandemic (January 2020 WEO) to assess compound risk.
- CAT modeling details:
  - Hazard module generated 10,000 simulated events per climate parameterization; five climate models → 50,000 simulated typhoons.
  - Vulnerability module produced damage rate = value of lost physical capital / stock of existing physical capital (percent).
  - For stress testing, the 90th percentile future loss estimates were used conservatively.

### Key quantitative findings — macroeconomic impacts
- Damage rates and peak GDP declines (Table 3 figures preserved exactly):
  - Return period (years) / Damage rate of capital: Current — Damage rate of capital: Future — Peak GDP decline: Current — Peak GDP decline: Future
    - 2 — 0.52 — 0.76 — -0.73 — -1.10
    - 10 — 0.52 — 0.76 — -0.73 — -1.10
    - 25 — 0.92 — 1.10 — -1.31 — -1.61
    - 50 — 1.42 — 1.98 — -2.04 — -2.90
    - 100 — 2.23 — 3.12 — -3.28 — -4.50
    - 250 — 3.68 — 5.70 — -5.36 — -8.39
    - 500 — 5.03 — 8.47 — -7.39 — -12.28
    - Annual average — 0.23 — 0.31 — ... — -0.12
- Main macro outcomes:
  - 10- and 25-year return period typhoons give less than one standard deviation shock to GDP under both current and future scenarios.
  - A 500-year return period typhoon’s impact on real GDP is comparable to that of COVID-19 (COVID-19 peak GDP decline reported as -14.49 in memo items).
  - Incremental impact of climate change on steady-state annual average GDP growth: -0.12 percentage points per year.
- Contextual benchmarks:
  - Philippines’ potential growth rate estimated at 6½ percent per year as of 2020.
  - One standard deviation of annual GDP growth rate (1980–2019) is 3⅓ percent.

### Key quantitative findings — banking sector solvency impacts
- Test design and scope:
  - Solvency stress test over a 3-year horizon; quasi-static balance sheet assumption; banks included: 46 UKBs representing 92 percent of the banking system by assets; starting position: December 2019, solo basis.
  - Risk channels included: credit risk (PDs, LGDs), market risk (MTM for HFT/AFS; credit approach for HtM), net interest income gap analysis, pre-impairment income shocks; operational risks excluded.
  - Regulatory parameters: Tax rate = 30%; LGD = 68 percent for UKBs; cure rate = 10 percent for UKBs with extra 18 percent on New NPLs in 2021; Basel II standardized approach.
- Climate change alone:
  - Climate change visibly reduces bank capital ratio only in 500-year tail events; decline described as "small at one percentage point."
  - Difference of capital ratios between baseline and a 500-year typhoon:
    - Maximum 0.2 percentage points in the current scenario.
    - Maximum 0.9 percentage points in the future scenario.
  - Policy rate cuts can produce valuation gains on sovereign bond assets (~25 percent of bank assets) that may raise bank capital immediately after the shock.
- Compound shocks (pandemic + typhoon):
  - Joint shock substantially increases impacts:
    - In the current scenario: additional impact of typhoons increases by 2 percentage points for a 25-year typhoon and 5⅓ percentage points for a 500-year typhoon.
    - In the future scenario: joint shock increases additional impact by 2.2 percentage points (25-year) and nearly 8⅔ percentage points (500-year) compared to normal time baseline.
    - For 500-year events with the pandemic, the difference between current and future scenarios rises to 4½ percentage points versus one percentage point under normal time baseline.

### DSGE model design and calibration (macrofinancial mapping)
- Model features:
  - Philippines-specific simplified DSGE with capital depreciation shocks representing physical capital losses, adjustment costs to rebuild capital, and correlated persistent productivity shocks.
  - CAT-model damage rates fed as depreciation shocks; empirical calibration assumes roughly one-third of GDP impact from capital destruction and two-thirds from persistent TFP shock.
  - Shock processes and data mapping:
    - Model shocks include consumption preference, investment adjustment cost, TFP, price elasticity, monetary policy, aggregate corporate risk, and idiosyncratic/default risk.
    - Mapped time series: Real GDP growth, Investment growth, Employment growth, GDP deflator/Inflation, Policy Rate, Stock Price Growth, Lending Spread.
  - Calibrated parameters (selection preserved exactly):
    - α = 0.30; β = 0.99; δ = 0.025; h = 0.75; σ = 1.00; φ = 1.00; τ_n = 4.00; τ = 0.20; μ = 0.40; γ_D = 0.03; θ_ss = 0.95.
- Operational assumptions for stress testing:
  - For future scenario, bank balance sheet structure and baseline macro scenario are assumed the same as 2020 to identify marginal effects of climate change.
  - Disaster modeled as a large capital depreciation shock in the first quarter followed by recovery; no additional tail disasters during the projection.

### Uncertainties, limitations, and interpretation
- Multiple layers of uncertainty:
  - Climate model uncertainty dominates for mid-21st century projections; internal variability matters in near term; scenario differences grow by end-century.
  - CAT model limitations: focuses on typhoons and floods; does not include sea-level rise, drought, heatwaves, or storm-surge/precipitation channels in this exercise.
  - Damage estimates rely on exposure data (location and value of physical capital) which may have significant margins of error in EMDEs.
  - Macrofinancial translation subject to substantial model uncertainty; macro scenario approach lacks sectoral and regional precision relative to micro-level approaches.
- Modeling omissions that may understate impacts:
  - Excludes flood damages, agricultural losses, health impacts, sea-level rise, transition risks, and potential increases in LGD from declines in property collateral values.
  - Operational bank disruptions (destroyed branches, ATMs, payment outages) not modeled.
  - GDP losses could be larger if key infrastructure (roads, ports, airports, utilities) is destroyed causing broader spillovers.
- Nature of inference:
  - Results are exploratory; under "deep" uncertainty it is inappropriate to draw strong prudential policy implications from a single small exercise.
  - Stress tests are intended to illuminate ranges of outcomes and vulnerabilities rather than yield point-estimate probabilities.

### Policy discussion and high-priority recommendations
- High-priority institutional and supervisory actions (timing coded ST = within one year; MT = one to three years):
  - Perform in-depth environmental risk assessments on banks and insurers. Timing: MT. Agency: BSP, IC.
  - Improve information collection, monitoring of risk metrics, and stress test capacity for climate and environmental risks. Timing: MT. Agency: BSP, IC, government.
  - Issue detailed guidelines on environmental risk management, stress testing and disclosures. Timing: MT. Agency: BSP, SEC.
- Additional recommendations and focus areas:
  - Increase granularity of banking-sector exposure data by industries and locations; collect LTV data to assess real estate loan risk.
  - Further strengthen disclosure requirements to enrich firms’ and financial institutions’ exposures to climate change risks.
  - Foster inter-agency collaboration for physical risk assessment (including PAGASA, DOE, DA).
  - For transition risk: rebalance long-term energy plans away from coal toward renewables, noting potential electricity price increases; implement the moratorium on new coal-based power plants announced by DOE in late 2020.
  - Explicitly incorporate deep model and scenario uncertainty into further analytical work; consider a variety of models and calibrations (Robust Decision Making/RDM approaches) rather than relying on single-scenario predict-then-act rules.

*Source: PHILIPPINES — INTERNATIONAL MONETARY FUND, excerpts from IMF FSAP chapter 1phlea2022001.*

### EXECUTIVE SUMMARY __________________________________________________________________________ 4

### EXECUTIVE SUMMARY

### Introduction and Context
- The Philippines is categorized as one of the world’s most vulnerable countries to climate change and natural disasters, especially typhoons.
- All major cities and most of the population reside on the coastline, including the metropolitan Manila area where about 60 percent of economic activities take place.
- The Philippines experiences 20 typhoons per year on average.
- Exposure to transition risk is concentrated in the coal-based power generation sector and the government’s licensing policy to build new power plants.
- The country’s current policies are still short of achieving the goal set by the Paris Agreement, but "appear to put the country closer to achieving the goal than advanced economies and several large emerging market economies."

### Climate Change Stress Testing Approach (methodology and scenarios)
- The FSAP developed a new macro-scenario stress testing method linking three sub-components:
  - Climate scenario: based on an existing study about changes in typhoon intensity and frequency.
  - Disaster scenario: based on the World Bank’s catastrophe (CAT) risk model for the Philippines, translating typhoon impacts into loss of physical capital.
  - Macrofinancial scenario: uses a Dynamic Stochastic General Equilibrium (DSGE) model calibrated for the Philippines to map lost physical capital into economic outcomes (damaging capital stock and productivity).
- Innovation: first to link typhoon CAT models to macro scenario stress tests of banks, explicitly accounting for climate change effects by considering current disaster risks and potential future development under a chosen climate scenario.
- Scenarios constructed and evaluated:
  - "Current scenarios" that start in late 2020 and "future scenarios" that occur sometime in the mid-21st century.
  - For each disaster scenario, three sub-scenarios: baseline without disaster, with once in 25-years disaster, and with once in 500-years disaster.
  - Two baselines considered: with pandemic (October 2020 World Economic Outlook, WEO, forecast) and without pandemic (January 2020 WEO forecast) to assess compound risk.

### Key Quantitative Findings (macroeconomic and banking impacts)
- GDP impacts:
  - Typhoons with 10- and 25-year return periods give less than one standard deviation shock to GDP under both current and future scenarios.
  - The impact of a 500-year return period typhoon on real GDP is comparable to that of COVID-19.
  - Incremental impact of climate change on steady-state annual average GDP growth: -0.12 percentage points per year.
- Bank solvency impacts:
  - Without other extreme tail events, climate change reduces bank capital ratio visibly only in 500-year tail events; decline is "small at one percentage point."
  - Difference of capital ratios between baseline and a 500-year return period typhoon:
    - Maximum 0.2 percentage points in the current scenario.
    - Maximum 0.9 percentage points in the future scenario.
  - Compound shocks (pandemic + typhoon) substantially increase impacts:
    - Joint shock strengthens the additional impact of typhoons by 2.2 and nearly 8⅔ percentage points compared to the normal time baseline case.
    - For 25-year return period events, the effects of climate change remain small.
    - For 500-year return period events with the pandemic, the difference between current and future scenarios rises to 4½ percentage points compared to one percentage point under normal time baseline.

### Interpretation and Limitations
- Results are exploratory and should be interpreted with caution due to "significant uncertainty over scenarios and models."
- Stress tests for climate change face multiple layers of model and scenario uncertainty, often not quantifiable; some uncertainties are characterized as "deep" uncertainty or ambiguity (e.g., inability to put probability distributions over Representative Concentration Pathways).
- Under deep uncertainty, it is inappropriate to draw strong prudential policy implications from a single small exercise.

### Policy Discussion and Recommendations (summary)
- The FSAP emphasizes the need for further exploratory work, more granular data, and inter-agency collaboration (including PAGASA, DOE, DA) for physical risk assessment.
- For transition risk, focus on rebalancing long-term energy plans away from coal toward renewables, acknowledging potential electricity price increases.
- For banking sector data and disclosure:
  - The Bangko Sentral ng Pilipinas (BSP) should increase granularity of exposure data by industries and locations.
  - Further strengthen disclosure requirements to enrich firms’ and financial institutions’ exposures to climate change risks.
- Take deep model and scenario uncertainty explicitly into further analytical work and consider a variety of models and scenarios.

### High-Priority Recommendations (from Table 1)
- Addressing Resilience, Integrity, and Effectiveness — Climate Change, Environment Risks and Supervision:
  - Perform in-depth environmental risk assessments on banks and insurers. Timing: MT. Agency: BSP, IC.
  - Improve information collection, monitoring of risk metrics, and stress test capacity for climate and environmental risks. Timing: MT. Agency: BSP, IC, government.
  - Issue detailed guidelines on environmental risk management, stress testing and disclosures. Timing: MT. Agency: BSP, SEC.
- Timing definitions provided: Short-term (ST) = within one year; medium-term (MT) = one to three years.

*Source: PHILIPPINES — INTERNATIONAL MONETARY FUND, EXECUTIVE SUMMARY.*

### 7.      Exposures to transition risk are also relevant and appear to be concentrated in the

### 7.      Exposures to transition risk are also relevant and appear to be concentrated in the coal-based power generation sector.

### Transition risk and power sector exposure
- The Philippines’ current policies are not entirely consistent with the long-term temperature goal of the Paris Agreement as of end-2020, while they are consistent with the 2009 Copenhagen 2°C goal.
- The share of coal-based power generation increased from a negligible amount in the 1990s’ to nearly 50 percent in 2016.
- Policy action needed to achieve the Paris Agreement target includes implementing the moratorium on new coal-based power plants announced by the Department of Energy in late 2020 (Ahmed and Brown, 2020).

### Rationale for focusing on physical risks
- The FSAP focused on the risk assessment of physical risks primarily because substantial existing studies facilitate the analysis.
- The Philippines is relatively better placed to achieve the Paris Agreement than many advanced economies and several large emerging market economies that are well below achieving the 2009 Copenhagen 2°C goal (Climate Action Tracker).
- Given frequent typhoons, substantial research exists on global warming’s implications for typhoons in the Philippines and expected damages to infrastructure, enabling tailored physical risk scenarios based on solid science.

### Financial stability channel and sectoral focus
- The exercise focused on solvency of banks because banks dominate the Philippines’ financial system.
- The insurance sector, especially the catastrophe insurance market, is underdeveloped partly due to limited technical capacity of the insurance regulator.
- Insurers can generally adjust premiums annually; thus, large impacts on insurers are expected to be small in capital terms, shifting costs to consumers or the public sector if premiums rise significantly.
- Therefore, the prominent linkages between financial stability and climate change lie in the banking sector.

### New macro-scenario stress testing method for physical risks
- The FSAP developed a macro-scenario stress testing method for banks to assess physical risks; scenarios consist of three sub-components: climate, disaster, and macrofinancial scenarios.
  - Climate scenario: based on government study and IPCC Representative Concentration Pathways (RCP); estimates changes in typhoon distribution (frequency and intensity measured by windspeed and precipitation) in the Philippines by the mid-21st century for a given global scenario.
  - Disaster scenario: based on the WB’s catastrophe risk (CAT) model for the Philippines; translates typhoon impacts into loss of physical capital.
  - Macrofinancial scenario: uses a DSGE model calibrated for the Philippines to calculate the impact of lost physical capital on the economy.
- The impact on bank solvency is then estimated using the same approach as standard macro scenario stress tests for banks.

### Novelty relative to existing literature and practice
- Climate risk stress tests to date have focused mostly on transition risks in advanced economies; physical-risk bank stress tests are relatively limited, especially in EMDEs.
- This FSAP is the first, to the authors’ knowledge, to link typhoon CAT models to macro scenario stress tests of banks.
- CAT models consider extreme tail events such as once in 100, 250, and 500 years events (return periods).
- Historical-observation-based disaster scenarios tend to underestimate disaster risk because recorded history is usually much shorter than 250 or 500 years (example: category five in 2013 was a once in 30-year event in terms of losses because it affected mostly rural areas).

### Uncertainties and limitations of the stress test
- The exercise faces multiple layers of model and scenario uncertainties and limitations with existing tools:
  - Uncertainties from climate models:
    - Wide range of globally accepted climate scenarios with different GHG emissions trajectories (Rogelj and others, 2018).
    - Climate science model uncertainty about precise effects of global warming on local climate phenomena and potential tipping points (Good and others, 2014).
  - Limitation of catastrophe models:
    - Exercise focused only on risks to physical capital from typhoons and did not include broader physical risks (sea-level rise, drought, heatwave) due to CAT model limitations.
    - Typical CAT models handle only typhoons and floods and each at a time; damages measured by value of destructed buildings and infrastructure.
    - CAT model accuracy depends on data on physical capital location and value, which can contain significant margin of error in EMDEs with scant disaster insurance coverage.
  - Macrofinancial impact translation:
    - Translating physical capital losses to macrofinancial impact is subject to substantial model uncertainties.
    - The macro scenario approach captures general equilibrium effects well but lacks sectoral and regional precision compared with micro-level approaches.

### Analytical emphasis and intended use
- The exercises emphasize a range of possible outcomes rather than point estimates, acknowledging “deep” uncertainty that cannot be meaningfully probabilified (e.g., probability over future RCP scenarios).
- Scenario analysis and stress tests help produce a range of outcomes based on alternative scenarios and key parameter assumptions.
- Given uncertainties and limitations, the climate change stress test is not intended for prudential regulation decisions; it is exploratory and aimed to inform academic and broad policy discussions.
- Policy recommendations therefore focus on expanding data, enhancing disclosures, and strengthening capacity to analyze financial stability risks associated with climate change rather than immediate prudential measures.

### Overview of stress test workflow and scope
- Four key steps in the FSAP stress test for physical risks:
  1. Climate scenario
     - Climate science models, IPCC and the PAGASA (with UK Met Office).
     - Global climate scenario: RCP 8.5, mid-21st century using Global Climate Models (GCM).
     - Local climate scenario: typhoon frequency and intensity (windspeed).
  2. Disaster scenario
     - CAT models (World Bank and AIR Worldwide).
     - Hazard model: simulating paths of typhoons for a given local climate scenario.
     - Engineering: physical capital loss estimating using data for the location of assets.
  3. Macrofinancial scenario
     - FSAP-developed DSGE model calibrated for the Philippines.
     - Translate physical capital loss to output losses and impact on other macrofinancial variables.
     - Start in 2020 (current scenario) and at some point in the mid-21st century (future scenario).
  4. Bank solvency test
     - Standard FSAP macro scenario solvency stress test model (excl. operational risk).
     - For future scenario, assume the same balance sheet structure and baseline macro scenario as the current scenario. Two baseline scenarios (with and without pandemic).
- Scenario coverage limitations:
  - Climate change effects considered via typhoon windspeed only (PAGASA study mid-century projection 2036–2065) and not via storm surge or precipitation changes.
  - Disaster scenarios quantify direct physical capital losses from typhoon wind but not floods or other channels (e.g., agricultural, health).

### Solvency focus and event types considered
- Both extreme tail events and incremental events are considered, but solvency tests focus on extreme events (tail risks).
- Incremental changes were expressed as reductions in steady-state economic growth rates, but steady-state impact was small compared to the baseline potential economic growth rate; hence solvency tests concentrated on tail risks.

*Source: Selected excerpts from the IMF FSAP chapter on climate-related financial risks for the Philippines.*

### 21.      We developed a Philippines-specific DSGE model to account for physical capital losses

### 21. We developed a Philippines-specific DSGE model to account for physical capital losses explicitly

### Model design and rationale
- Developed a simplified single-country DSGE model with:
  - capital depreciation shocks representing physical capital losses,
  - adjustment costs to rebuild capital,
  - parameters calibrated using historical data of the economy.
- Rationale:
  - CAT risk models produce value of lost physical capital stock in percent of existing capital; interpreting that as a capital depreciation shock allows direct incorporation into a macro model.
  - The Global Macrofinancial Model (GFM) used in many FSAPs does not include capital depreciation shocks because it focuses on multi-country real and financial spillovers and various macrofinancial linkages.

### Bank solvency test design and scenarios
- Stress test horizon: 3-years upon a disaster.
- Timing of disaster scenarios:
  - “Current scenarios” that start in late 2020.
  - “Future scenarios” that take place at some point in the mid-21st century.
- For each disaster scenario analyzed:
  - A typhoon causes large economic damages in the first quarter of the projection period, followed by recovery (no more additional tail disasters).
  - Three sub-scenarios examined: baseline without disaster, once in 25-years disaster, and once in 500-years disaster.
- Comparison approach:
  - The effect of climate change measured as the difference between impacts implied by the current and future probability distributions of typhoons.
  - For future scenarios, the bank balance sheet and baseline economic projections are assumed to be the same as 2020 to identify marginal effects of climate change, keeping other conditions equal.
- Pandemic consideration:
  - Two baselines for current and future scenarios: with and without the pandemic.
  - Single disaster risk uses pre-pandemic forecast (January 2020 WEO).
  - Compound risk with pandemic uses October 2020 WEO that incorporates the effects of COVID-19.
  - Rationale: COVID-19 increases the likelihood of double tail risk—pandemic and disaster—substantially.

### Operational risk scope
- The bank solvency test does not account for operational risks from disasters (standard FSAP practice).
- Operational impacts noted but excluded due to data needs:
  - Potential disruptions: destroyed branch offices or ATMs, payment system outages.
  - Accurate assessment would require detailed information on banks’ contingency planning and insurance for disruptions.

### Climate scenario selection and uncertainty
- Climate scenario used: RCP 8.5 (high-emission scenario).
- RCP context from IPCC AR5:
  - RCP 2.6 (stringent mitigation),
  - RCP 4.5 and RCP 6.0 (intermediate),
  - RCP 8.5 (very high GHG emissions).
  - RCP 8.5 leads global temperatures by 2100 to 4.0-6.1°C above pre-industrial levels.
- Justification for RCP 8.5:
  - Physical risks tend to rise in high emission and temperature scenarios; RCP 8.5 is appropriate because it is more likely to increase physical risks.
- Climate model uncertainty notes:
  - Climate model uncertainty dominates over climate scenario uncertainty for the mid-21st century used in the PAGASA exercise.
  - Drivers of uncertainty over time:
    - First couple of decades: internal variability (chaotic weather variation).
    - By mid-century: differences across climate models become main source of uncertainty.
    - Towards end of century: scenario differences start to dominate.
  - Implication: For mid-21st century projections, differences between RCPs are small and largely explained by model uncertainties.

### PAGASA projections on typhoon characteristics
- PAGASA’s projections (based on five RCM and GCM simulations) suggest:
  - On average, fewer but slightly more intense typhoons in the future.
  - Simulation outcomes:
    - Three models suggest significant decreases in tropical cyclone frequency; two suggest little change.
    - Four models agree on projected increase in tropical cyclone intensity (maximum sustained windspeed) overall; two show significant increases.
  - Year-to-year variability will remain high.
  - For improved country-level accuracy, downscaling GCM outputs with a Regional Climate Model at higher 10-25 km2 resolution is required.
- Model uncertainty caveat:
  - Existing studies find low confidence in geographical distribution of GCMs at the ocean basin level; downscaling and model validation against historical data are necessary.

### Disaster scenario construction (CAT modeling)
- CAT modeling modular approach used (hazard → vulnerability → financial/impact):
  - Hazard module simulates number and intensity (windspeed) of hurricanes/typhoons.
  - Vulnerability module overlays hazard outputs on exposure data (location and value of private and public buildings and infrastructure) to estimate damage severity.
  - Standard CAT outputs: Annual Average Loss (AAL) and Exceedance Probability (EP) curve.
- For this exercise:
  - Approach aligned with World Bank DRFI program with technical support from Air Worldwide.
  - Hazard module generated 10 thousand simulated events based on detailed historical typhoon catalogs for each climate parameterization.
  - Five climate models used, yielding 50 thousand simulated typhoons (10 thousand each for five climate models).
  - Vulnerability module produced damage rate = value of lost physical capital / stock of existing physical capital (in percent).
  - Instead of directly calculating insured losses, the damage rate was fed into the macrofinancial DSGE model as capital stock depreciation shock to capture general-equilibrium effects.

### Recalibration for future climate conditions
- Future damage scenarios recalibrated following Ranger and Niehörster (2012):
  - Used PAGASA’s future number of typhoons and intensities from each of the five climate models.
  - Hazard module in the CAR model generated 10 thousand typhoons using these future parameters.
  - Vulnerability and macrofinancial modules remained the same as in the current scenario.

### Key empirical findings from CAT outputs
- Damage likely to increase in the future despite projected declines in total number of typhoons:
  - Non-linear relationship between wind-speed and damages implies more intense storms produce higher damages.
  - Increases in typhoon risk for majority of climate scenarios observed for:
    - medium events (once in 50 years),
    - tail events (once in 250 years).
- Distributional results:
  - Figure of estimated physical capital losses from 50 thousand simulated typhoons shows wide ranges of potential losses, especially for tail events with longer return periods.
  - For each return period, future median losses are higher than current median losses (future median corresponds to about 25 percentile losses in the future scenarios).
  - The exercise used the 90th percentile future loss estimates for the stress test to be conservative (more severe).
  - Result highlighted: the damage rate (share of the value of lost physical capital over the value of existing stock in percent, using the 90th percentile future loss estimates) nearly doubles in the future scenario for the tail events (500 year return period).

### Macrofinancial consideration of systemic impact
- Whether a disaster becomes a systemic macrofinancial shock depends on size and diversification of the economy.
- Literature examples and evidence:
  - Central American study: a one standard deviation increase in a hurricane’s intensity leads to a decrease in total GDP per capita growth of 0.9–1.6 percent and a 3 percent decrease in total income (Ishizawa and Miranda 2016).
  - Reconstruction after major disasters can last 5 to 10 years, magnifying impacts on GDP and consumption.
  - Hurricane Katrina (U.S., 2005): macro-level impact on the whole U.S. economy was limited; Federal Reserve Board continued tightening monetary policy in 2005 and 2006. Local impacts and displacement were substantial but national employment and income impacts were small and transitory.
    - Katrina displacement example: about 400,000 people displaced.
- Implication for the Philippines:
  - Philippines is more geographically and economically diverse than small island countries but much smaller than the U.S.
  - It is not clear whether disasters could cause a systemic impact without a model-based analysis; hence the use of the country-specific DSGE model with capital depreciation shocks.

*Source: 1phlea2022001 - 21. We developed a Philippines-specific DSGE model to account for physical capital losses explicitly*

### 35.      To fully account for the economic impact of intense typhoons, including potentially

### To fully account for the economic impact of intense typhoons, including potentially long-lasting amplification effects

### Channels of economic impact considered
- Immediate physical destruction
  - Shock interpreted as a one-time capital depreciation shock using damage rates from the CAT risk model.
  - CAT risk model covers buildings and infrastructure; damage rate is applied to all capital stock as an approximation.
- Long-lasting decline in total factor productivity (TFP)
  - Reconstruction frictions (clean-up, mismatch of usable roads/buildings, short-run inability to reallocate capital) cause TFP to decline in addition to capital loss.
  - Empirical finding cited: roughly one-third of the impact on GDP stems from capital destruction, and two-thirds from accompanying TFP shock (Bakkensen and Barrage 2018).
- Time for reconstruction (financial, regulatory, technical constraints)
  - Financial: delays in transferring resources from central government to local authorities; phased private rebuilding spreads costs over years.
  - Regulatory: long-term planning for “build back better”; political processes and procurement can delay reconstruction by months.
  - Technical: sectoral capacity constraints (e.g., debris removal equipment, skilled labor) cause “demand surge” visible in construction wages and prices.

### Modeling approach and calibration (RBC model application)
- Model incorporates three channels:
  - Immediate physical destruction
    - Shock z_(d,t) applied to depreciation rate δ/z_(d,t); a decline in z_(d,t) raises depreciation.
    - Shock calibrated so depreciation increases identical in magnitude to modelled damages.
  - Long-lasting decline in TFP
    - GDP production function: y_t = 〖k_(t-1)〗^α (z_(y,t) n_t )^(1-α).
    - Shock z_(y,t) calibrated to decline in magnitude twice as much as the increase in depreciation. Example: if depreciation increased from 2.5% to 6.5% (by four percentage points), z_(y,t) would fall from 1 to 0.92.
  - Time for reconstruction
    - Habits in consumption and investment adjustment costs lengthen adjustment.
    - z_(d,t) is short-lived; z_(y,t) has long persistence (allocation of capital and TFP eventually returns to pre-disaster level).
- Adaptation and mitigation not explicitly modeled
  - Measures (higher-quality infrastructure, developed insurance markets, contingent financing) could limit damages but are not incorporated due to data limitations and uncertainty.

### Key scenarios used
- Baselines
  - Baseline with COVID-19: October 2020 WEO.
  - Baseline without pandemic: January 2020 WEO.
- Adverse scenarios
  - Four adverse scenarios for each baseline: once in 25- and 500-years typhoon using current and future disaster risks.
  - Incremental change scenario (climate-change-only incremental effect) not stress tested because effect on GDP is minimal.

### Table 3 — Disasters’ impact on capital and GDP (selected figures preserved exactly as in source)
- Return period (years)/episodes — Damage rate of capital: Current scenario — Damage rate of capital: Future scenario — Peak GDP decline: Current scenario — Peak GDP decline: Future scenario
  - 2 — 0.52 — 0.76 — -0.73 — -1.10
  - 10 — 0.52 — 0.76 — -0.73 — -1.10
  - 25 — 0.92 — 1.10 — -1.31 — -1.61
  - 50 — 1.42 — 1.98 — -2.04 — -2.90
  - 100 — 2.23 — 3.12 — -3.28 — -4.50
  - 250 — 3.68 — 5.70 — -5.36 — -8.39
  - 500 — 5.03 — 8.47 — -7.39 — -12.28
  - Annual average — 0.23 — 0.31 — ... — -0.12
- Memo items: Actual shocks to real GDP growth rate in past crises (selected)
  - Global financial crisis — ... — ... — -2.90 — ...
  - Asian financial crisis — ... — ... — -5.69 — ...
  - COVID-19 — ... — ... — -14.49 — ...
  - 1980s political turmoil — ... — ... — -16.50 — ...

### Main simulation results and interpretation
- Potential growth and volatility context
  - Philippines’ potential growth rate estimated at 6½ percent per year as of 2020.
  - One standard deviation of annual GDP growth rate (1980–2019) is 3⅓ percent.
- GDP impacts
  - 10- and 25-year return period typhoons: shocks to GDP are less than one standard deviation and smaller than shocks during the Global Financial Crisis or Asian Financial Crisis.
  - Tail events: severity and difference between current and future scenarios grow markedly for tail events; a 500-year typhoon impact on real GDP is comparable to COVID-19.
  - Incremental impact of climate change on steady-state annual average GDP growth rate: -0.12 percentage points per year.
- Bank solvency and capital impacts
  - Stress test details
    - Standard FSAP solvency stress test over three-year horizon; quasi-static balance sheet assumption; gross exposures grow with nominal GDP; capital buffers built only via retained earnings.
  - Climate change alone
    - Climate change may not cause extreme financial instability on its own.
    - Bank capital ratio decline is visible only in 500-year tail events; decline small at one percentage point.
    - Policy rate cuts can produce valuation gains on sovereign bond assets (~25 percent of bank assets) that may raise bank capital immediately after the shock.
    - Difference of capital ratios between baseline and a 500-year typhoon: maximum 0.2 percentage points (current scenario) and 0.9 percentage points (future scenario).
  - Compound risks (typhoon + pandemic)
    - Joint materialization of typhoons and pandemic substantially increases marginal effects:
      - In current scenario: additional impact of typhoons increases by 2 percentage points for a 25-year typhoon and 5⅓ percentage points for a 500-year typhoon.
      - In future scenario: joint shock increases additional impact by 2.2 percentage points (25-year) and nearly 8⅔ percentage points (500-year) compared to normal time baseline.
    - For 500-year events, difference between current and future scenario with the pandemic rises to 4½ percentage points versus one percentage point under normal time baseline.
- Cautionary limits of the exercise
  - Excludes many climate-related risks: damages from storm precipitation, floods, agricultural losses, sea-level rise impacts, transition risk.
  - Does not account for increased LGD from declines in property collateral values due to data limitations.
  - GDP losses could be larger if key infrastructure (roads, ports, airports, utilities) are destroyed, causing broader spillovers.
  - Overall bank capital impact could be larger if all climate risks and firm- or industry-level channels are included.

### Policy discussion: Decision making under “deep” uncertainty
- Nature of uncertainty
  - Climate science and financial-stability climate stress testing face “deep” uncertainty where agents may not be able to assign probabilities to outcomes (ambiguity).
  - Traditional predict-then-act (expected value) approaches may be inadequate under deep uncertainty.
- Scenario analysis and Robust Decision Making (RDM)
  - Scenario-based stress tests are useful to illuminate low-probability high-impact tail events.
  - RDM treats models as mappings from assumptions to consequences; runs models numerous times with varied calibrations/assumptions to identify vulnerabilities and find robust policies across many assumption combinations.
  - RDM and other systematic methods can help authorities design adaptation and mitigation policies under deep uncertainty.

_Italic: Source — Author’s estimation, excerpted from the IMF chapter on Philippines climate risk and stress testing._

### 51.      The deep uncertainty with climate change means that we should be cautious when

### 1phlea2022001 - 51.      The deep uncertainty with climate change means that we should be cautious when

### Deep uncertainty and cautious interpretation
- The stress tests are subject to deep uncertainty at multiple levels: climate scenario, climate model for the Philippines, CAT risk model, macrofinancial model, and stress testing model.
- Conclusion: drawing prudential policy implications at this stage is premature.
- Recommended next step: explicitly explore sensitivity of climate change stress tests vis-à-vis alternative models and parameter assumptions at each step of the test.

### Recommendation (paragraphs 52–53)
- Focus areas for authorities:
  - More exploratory work to examine the potential impact of climate change on financial stability.
  - For physical risks: inter-agency collaboration (including PAGASA, DOE, and DA) is critical.
  - BSP and the Insurance Commission should tap into existing climate science resources.
  - For transition risks: primary source appears to be the power generation sector and the need to rebalance long-term energy plans from coal-based power plants to renewable sources; this transition could be accompanied by electricity price increases.
- Treatment of model and scenario uncertainty:
  - The issue of deep model and scenario uncertainty should be taken into these works explicitly.
  - Consider a variety of models and scenarios; the FSAP exercise could be one of the first steps.
- Data improvements to increase confidence in stress test results:
  - Increase granularity of banking-sector exposure data by industries and locations.
  - Collect LTV data to assess risks from real estate loans (irrespective of whether the BSP uses it as a prudential tool or not).
  - Further strengthen disclosure requirements to enrich information on firms’ and financial institutions’ exposures to climate change risks.
  - Such information would be helpful to identify micro-level channels linking climate change and financial stability.

### Climate change stress test of bank solvency — institutional perimeter and design (summary)
- Institutions included:
  - 46 UKBs
- Market share:
  - 92 percent of the banking system by assets.
- Data and baseline date:
  - Supervisory data (balance sheet and income statements).
  - Started position: December 2019
  - Data on a ‘solo basis.’

### Channels of risk propagation and methodology (section 2)
- Type of test:
  - Solvency stress test—the same as macro scenario stress tests, including risk factors, satellite models, and various micro and behavioral assumptions (such as LGD) following IMF solvency stress test workbox.
- Climate scenario:
  - Constructed by combining climate science, catastrophe risk, and macrofinancial models.
- Risk factors:
  - Physical risks—extreme weather events (typhoons) affecting bank solvency through macrofinancial impact.
  - Credit Risk: Satellite models per bank type to estimate loan losses. Regression model for logit transformed PDs. Regressors include the lagged dependent variable and contemporaneous and lagged macroeconomic variables: GDP growth, short term rates, term spread, unemployment, stock price, and exchange rate.
  - Market risk: valuation losses for HFT and AFS securities calculated using a Mark to Market (MTM) approach. Valuation losses for held-to-maturity (HtM) securities calculated using a credit risk approach. As sensitivity analysis, an MTM approach is used for HtM securities.
  - Net interest income: gap analysis based on granular asset/liability structure broken into funding sources and time to re-pricing buckets. Interest margin shocks vary per scenario.
  - Pre-impairment income for banks: income in the absence of shocks is assumed to stay at the level observed for 2019 with the additional feature that non-performing loans will not generate any income.
- Test horizon:
  - Three years upon a severe disaster.
  - Current scenario: A typhoon shock materializes in 2020, using end-2019 balance sheet data and distribution of typhoons (frequency and severity) as of 2020.
  - Future scenario: A typhoon shock materializes sometime in mid-21st century, using end-2019 balance sheet data and distribution of typhoons as of mid-21st century.

### Climate scenario formulation (section 3)
- Components:
  - Climate scenario: provides future distribution of typhoon intensity (windspeed) and frequencies for the Philippines using climate science model, based on a given global climate scenario. FSAP scenarios are based on the 2018 exercise by the Philippine Atmospheric, Geophysical, and Astronomical Services Administration. It shows the impact of IPCC global scenario RCP 8.5—a high greenhouse gas emission scenario with little global-scale mitigation policies and technological advancement. The global temperature would increase by 1.4-2.6°C above the 1986-2005 levels. For the Philippines, such climate change is likely to increase the average intensity (windspeed) of typhoons but reduce the average number of typhoons hitting the country.
  - Disaster scenario: estimates typhoon damages (physical capital losses in percent of existing stock) for a given likelihood (25-500 return period). For a given climate scenario, ten thousand possible losses from disasters are simulated using the catastrophe-risk model developed under the World Bank disaster risk financing and insurance program with the Government of the Philippines. Sea-level rise and flood effects are not included. For each return period, the mission uses the losses at the 90th percentile of the ten thousand simulation results for the future scenario (the losses under the current scenario roughly correspond to the 25th percentile of the simulation results).
  - Macrofinancial scenario: constructed by a staff-developed Dynamic Stochastic General Equilibrium (DSGE) model calibrated for the Philippines. Physical damage is modeled as a capital depreciation shock, which also causes correlated productivity shocks in line with empirical findings in disaster/infrastructure economics (⅓ of the impact comes from short-lived capital depreciation and ⅔ from persistent productivity shocks). The model includes adjustment costs with investment, slowing capital accumulation and recovery.

### Tail shocks and baseline economic scenarios (section 4)
- Two baseline economic scenarios before applying typhoon shocks:
  - Normal time: January 2020 WEO forecast (before COVID-19 shock), in line with the potential growth rate of 6½ percent.
  - Pandemic: October 2020 WEO forecast that includes the impact of COVID-19.
- Scenarios consider realization of two extreme events—pandemic and severe typhoons.
- Typhoon severity:
  - Measured by value of destroyed physical capital in percent of existing capital (for both current and future scenarios), for 25-500 return periods (i.e., once in 25-500-years event).
  - Insurance industry reference: insurers price disaster insurance aiming at losses under 100-year return period. They examine adequacy of reserves for tail events (250-500 years return period).

### Risks, buffers, and behavioral assumptions (section 5)
- Risks/factors assessed:
  - Credit risk (provision costs)
  - Market risk, including FX risk
  - Stress on pre-provision profits, including interest margin
- Behavioral adjustments and balance-sheet assumptions:
  - Balance sheet growth assumption: Quasi-Static—balance sheet size/GDP remains constant.
  - Balance sheet composition remains constant over the stress test horizon.
  - Banks can only accumulate capital through retained earnings.
  - Banks pay dividends only if net income after taxes is positive, with dividend payout ratio consistent with individual banks’ 2019 ratios for UKBs.
  - Tax rate: 30% (corporate tax rate for banks)

### Regulatory and market-based parameters (section 6)
- Calibration of risk parameters:
  - PDs: proxies based on actual and estimated new NPL flows over performing loans. PDs for banks with limited credit information taken as the weighted average PD of the rest of the institutions.
  - LGDs: 68 percent for UKBs based on average historical provision coverage ratio (provisions/NPLs).
  - Cure rates with respect to NPL(t-1): 10 percent for UKBs based on historical averages. In addition, an extra cure rate of 18 percent on New NPLs (t-1) is assumed for UKBs in 2021, given portfolio concentration to the Manila metropolitan area (99% of total loans).
- Regulatory and accounting standards:
  - Basel II standardized approach.
  - RWAs evolve with credit growth, net of increases in provisions. RWAs further adjusted by the new NPLs that are not provisioned (to reach the weight of 150% required by regulation).

### Reporting format for results (section 7)
- Output presentation:
  - The level of real GDP and aggregate CAR chart for five years for current and future scenarios showing the range of potential outcomes of the baselines (without disaster shocks) and stress scenarios (with typhoon shock ranging from once in 25-500-year severity).

### Appendix II — Macrofinancial Model: overview and key elements
- Model origin and structure:
  - Simplified version of Lipinsky and Miescu (2020), adapted for the Philippines.
  - Agents: households, firms, banks, monetary authority, and the government.
- Households:
  - Maximize expected lifetime utility E0 Σ β^t U_t with internal habit formation h>0.
  - Utility depends on consumption c_t and labor I_t with parameters σ, φ, and demand shock z_c,t.
  - Budget constraint: I_t + d_t = w_t I_t + R_d,t-1 π_t d_t-1 + Π_i,t + Π_y,t (notation preserved as in source).
  - First-order conditions for c_t, n_t, d_t, i_t explicitly stated in original formulation.
- Firms:
  - Invest q_t k_t in productive assets k_t; finance with commercial loans l_t and equity n_F,t = q_t k_t − l_t.
  - Gross return on assets: R_k,t+1 = r_k,t+1 + (1−δ) q_t+1.
  - Firms pay share Γ_t+1 of earnings to the bank and face taxes and cashflow specified in original equations.
- Banks:
  - Intermediaries between households and firms; accept deposits d_t and extend commercial loans l_t; equity n_B,t = l_t − d_t.
  - Receive share Γ_t+1 of earnings from firms but incur default costs R_k,t+1 k_t μ Δ_t+1.
  - Basel II capital constraint imposed to shield the bank from default (constraint equation preserved in source).
  - Optimization problem: maximize cashflow subject to Basel II capital constraint and firms’ participation constraint (objective function and constraints provided explicitly).
- Key modeling features and choices:
  - Income of firms and banks taxed at rate τ>0; households don’t pay taxes on deposits.
  - Commercial loans chosen optimally trading off benefit of financial intermediation τ( R_k,t+1 k_t )(Γ_t+1 − μ Δ_t+1) versus cost of default −R_k,t+1 k_t μ Δ_t+1 (1−τ).
  - Γ_t+1 and Δ_t+1 depend on ε_t+1^* and on θ_t and π_t+1 in the mapping specified in the source.
  - Idiosyncratic/default risk shock: σ_F,t affects tail mass and defaults (Christiano, Motto and Rostagno (2014) reference retained as in source).
- Solution characteristics (selected results):
  - Bank provides commercial loans l_t such that return equals cost; first-order condition includes factor X_t+1 which embeds default cost −μ Δ(ε_t+1^*) and benefit of intermediation λ_F,t V θ_t [Γ(ε_t+1)−μ Δ(ε_t+1)].
  - Bank chooses equity n_B,t internalizing that more equity loosens the Basel II capital constraint; first-order condition for I_B,t given explicitly.
  - Bank chooses commercial loan rates R_l,t such that X_t+1 is maximized and invariant to changes in the default threshold; first-order condition for R_l,t provided in the source.
- Stock price gaps:
  - Firms accumulate net worth n_F,t out of retained earnings, distributing share D of earnings and receiving fixed equity injection ω.
  - Investment I_F,t and target net worth I_F,t^* relations given; stock price gap defined as sss = (q_t − q_t^*)/q_t^* (notation preserved).

*Source: IMF staff text from the content unit provided.*

### 13.      In the optimal solution (흏흏풏풏

### 13.      In the optimal solution (흏흏풏풏

### Optimal solution (summary)
- In the optimal solution (푭푭,풕풕:  흀흀 푭푭,풕풕 =ퟏퟏ), the participation constraint of firms determines 풏풏푭푭,풕풕.
- The optimality condition with respect to 풍풍풕 determines 풌풌풕 and hence 푙푙푡 = 푞푞푡 푘푘푡 − 퐼퐼퐹,푡.
- The Basel II capital constraint determines 푑푑푡 and hence 퐼퐼퐵퐵,푡 = 푙푙푡 − 푑푑푡.
- The optimality condition with respect to 퐼퐼퐵퐵,푡 together with the optimality condition of households with respect to 푑푑푡 imply 휆휆푡푉푉 > 0.

### Realized (suboptimal) solution
- 풏풏푭푭,풕풕 is given through the exogenous law of motion.
- The participation constraint of firms then determines 푘푘푡 and hence 푙푙푡 = 푞푞푡 푘푘푡 − 퐼퐼퐹,푡.
- The optimality condition with respect to 푙푙푡 determines 휆휆퐹퐹,푡푡, which is time varying.
- 푑푑푡 and 휆휆푡푉푉 are determined as above.

### Financial sector representative and market clearing
- Financial sector agent maximizes lifetime utility E0 ∑∞ t=0 βt U t FSS where
  - U t FSS = z I,t [I FSS,t − h I FSS,t−1]^(1−σ)/(1−σ)
- Financial sector aggregates income across firms and the bank.
- Financial sector budget constraint:
  - 퐼퐼퐹퐹퐹퐹,푡 + q_t k_t − d_t = [R_k,t k_{t−1} (1−μ Δ_t) − R_d,t−1/π_t d_{t−1}] (1−τ)
- Stochastic discount factor:
  - M_{t+1} = β (λ_{FSS,t+1} / λ_{FSS,t})
- λ_{FSS,t} is the Lagrange multiplier associated to the budget constraint.

### Final good firms, prices, and monetary policy
- Final good firms:
  - Rent capital k_{i,t−1} from firms, hire workers n_{i,t} from households, and choose prices p_{i,t} subject to quadratic adjustment cost:
    - (φ/2) [P_{i,t}/P_{i,t−1} − 1]^2 y_t = (φ/2) [π_t s_{i,t}/s_{i,t−1} − 1]^2 y_t
    - π_t ≡ P_t / P_{t−1}; s_{i,t} ≡ P_{i,t} / P_t
    - η_t = η_ss z_η,t
  - Profits per firm (Π_{y,i,t}):
    - Π_{y,i,t} = s_{i,t} y_{i,t} − r_t k_{i,t−1} − w_t I_i,t − (φ/2)[π_t s_{i,t}/s_{i,t−1} − 1]^2 y_t
  - Production:
    - y_{i,t} = k_{i,t−1}^α [z_y,t I_{i,t}]^{1−α}
  - Aggregate relation:
    - y_{i,t} = s_{i,t} − η_t y_t (as given in text: 푦푦_{푖푖,푡푡} = 푠푠_{푖푖,푡푡} − 휂휂_{푡푡} 푦푦_{푡푡})
- Optimization problem for firm (as presented) leads to first-order conditions in symmetric equilibrium.

### Symmetric equilibrium first-order conditions (as given)
- v_t = 1 − s_{m,t}
- ∂/∂k_{i,t−1}: r_t = s_{m,t} α y_t / k_{t−1}
- ∂/∂I_{i,t}: w_t = s_{m,t} (1 − α) y_t / I_t
- ∂/∂s_{i,t}: s_{m,t} = 1 − (1/η_t) [1 − φ π_t (π_t − 1) + φ β E_t( (λ_{t+1}/λ_t) (π_{t+1}/π_{t+1}) (π_{t+1} − 1) y_t / y_{t−1} )]

### Monetary policy (Taylor rule)
- Monetary authority sets interest rates:
  - R_{d,t} / R_{d,ss} = (R_{d,t−1} / R_{d,ss})^{γ_R} [ (π_t / π_{ss}) ]^{1−γ_R} [ (y_t / y_{ss}) ]^{γ_π} [ y_I σ_I ε_{I,t} ]^{1−γ_R}
  - (Presented in the source as a Taylor-type rule with persistence and responses to inflation and output, including a monetary policy shock term ε_{I,t}.)

### Simple RBC special case (without inflation)
- Final good firms rent capital k_{t−1} and hire workers I_t from households.
- Objective simplifies to I_mmm { y_t − r_t k_{t−1} − w_t I_t }
- Production: y_t = k_{t−1}^α [ z_y,t I_t ]^{1−α}
- First-order conditions:
  - ∂/∂k_{t−1}: r_t = α y_t / k_{t−1}
  - ∂/∂I_t: w_t = (1 − α) y_t / I_t

### Shocks and shock processes
- Model shocks (three macro, two monetary-policy related, two financial):
  - Consumption preference shock z_{c,t}:
    - ln(z_{c,t}) = ρ_c ln(z_{c,t−1}) + σ_c ε_{c,t}
  - Investment adjustment cost shock z_{i,t}:
    - ln(z_{i,t}) = ρ_i ln(z_{i,t−1}) + σ_i ε_{i,t}
  - TFP shock z_{y,t}:
    - ln(z_{y,t}) = ρ_y ln(z_{y,t−1}) + σ_y ε_{y,t}
  - Price elasticity shock z_{η,t} (η_t = η_ss z_{η,t}):
    - ln(z_{η,t}) = ρ_η ln(z_{η,t−1}) + σ_η ε_{η,t}
  - Monetary policy shock in the Taylor rule: σ_I ε_{I,t}
  - Aggregate corporate risk shock z_{θ,t} (θ_t = θ_ss z_{θ,t}):
    - ln(z_{θ,t}) = ρ_θ ln(z_{θ,t−1}) + σ_θ ε_{θ,t}
  - Idiosyncratic corporate risk shock σ_{F,t}:
    - ln(σ_{F,t}) = [1−ρ_σF] σ_{F,ss} + ρ_σF ln(σ_{F,t−1}) + σ_σF ε_{σF,t}

### Data used for estimation and mapping of shocks
- Three macroeconomic time series: Real GDP growth; Investment growth; Employment growth.
- Two monetary policy series: GDP deflator/Inflation; Policy Rate.
- Two financial series: Stock Price Growth; Lending (corporate credit) Spread.
- Mapping table (as provided in source):
  - TFP Shock → Real GDP Growth
  - Investment Adjustment Cost Shocks → Investment Growth
  - Demand Shock → Employment Growth
  - Monetary Policy Shock → GDP Deflator/Inflation
  - Elasticity of Substitution between Goods Shock → Policy Rate
  - Aggregate/Value-at-Risk Shock → Stock Price Growth
  - Idiosyncratic/Default Risk Shock → Lending Spread

### Calibrated parameters (before shocks)
- 훼: power on capital in production function = 0.30
- 훽: discount rate = 0.99
- 휹: depreciation rate of capital = 0.025
- ℎ: habit parameter = 0.75
- 휎: power on consumption = 1.00
- 휙: curvature on disutility of labor = 1.00
- 휏푛: disutility weight on labor = 4.00
- τ: corporate tax rate = 0.20
- 휇: loss-given-default = 0.40
- γ_D: payout rate = 0.03
- ω: household equity injection; such that in steady state 휆휆퐹퐹,푡푡 = 1 (value not numerically specified)
- 휃_ss: steady state value-at-risk = 0.95

*Italic source attribution: Content summarized from the supplied IMF chapter/section text.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2022/english/1phlea2022001.pdf_