## Systemwide Liquidity Stress Testing Tool (IMF Working Paper — wpiea2022252-print-pdf)

## Source details

**Canonical URL:** [Systemwide Liquidity Stress Testing Tool (IMF Working Paper — wpiea2022252-print-pdf)](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022252-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022252-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022252-print-pdf.pdf.json)

---

### Executive summary: nature, limits, and proposed tool
- Definition: systemic liquidity crisis — multiple financial institutions simultaneously face liquidity stress.
- Historical forms: massive multi-bank runs; market runs (post-GFC); COVID-19 freeze of cashflows and massive liquidation of safe assets (e.g., U.S. treasury market).
- Literature focus: game-theoretic bank-run models; post-GFC emphasis on "run on repos", collateral channels, fire sales, interaction of market and funding liquidity, leverage cycle, cash hoarding.
- Monitoring and limitations:
  - Regulators use quantity and price/spread indicators and sector-specific stress tests.
  - Systemwide liquidity stress tests limited by data constraints, need to capture interconnectedness (banks, NBFIs, NFCs, foreign investors), behavioral uncertainty, resource-intensive granular data processing, and difficulty of standardization.
- Proposed solution:
  - A simple Excel-based systemwide liquidity stress testing tool relying mostly on BSA data to perform sector-by-sector cashflow liquidity tests (LCR-like) and link them through direct financial linkages.
  - Emphasis on liquidity contagion: one sector’s liquid buffers are another sector’s funding instruments.
  - Avoids modeling second-round effects to limit model risk.
  - Purpose: high-level vulnerability assessment to identify critical propagation channels and guide deeper, granular analyses.

### Tool design, scope, and methodological principles
- Scope and data:
  - Primarily uses sector-aggregate BSA data; can be augmented with financial statements, supervisory reporting, and proxies.
  - Applicable across many countries, including EMDEs and AEs, and executable with few resources in Excel.
  - Not suited to capture idiosyncratic-to-systemic amplification requiring agent-level models (e.g., ABMs) or transaction-level collateral channels and margin calls without richer data.
- Four main steps of systemwide liquidity stress tests:
  1. Map key sectors and financial linkages; visualize network and sectoral balance sheet structure.
  2. Formulate cashflow liquidity stress tests at sector-aggregate level.
  3. Gauge contagion under behavioral assumptions (pecking orders of liquidation).
  4. Consider central bank liquidity policy options to assess resilience with and without support.
- Key modeling choices:
  - Sector cashflows: inflows (operations, investment income, debt rollover, credit line drawdowns) vs outflows (operations, debt service, margin calls, deposit outflows).
  - Liquid assets: cash and highly-liquid instruments; apply valuation haircuts under stress.
  - Contagion driven by pecking order of asset liquidation rather than a single pro-rata assumption; examine a range of behavioral assumptions to mitigate model risk.

### BOP shocks, foreign liquidity, and the central bank LOLR role (Box 1)
- Taxonomy of liquidity: Domestic vs Foreign; Private vs Public. Both currencies possible.
- Foreign liquidity is "outside money" for the economy and can change net system liquidity via BOP flows (non-resident withdrawals repatriate money and remove net liquidity).
- Central bank LOLR capacity for FX is constrained by international reserves; collateral constraints limit borrowing in international markets.
- Policy implications:
  - Economies without a reserve currency need international reserves to prepare for systemwide liquidity stress.
  - Tool can estimate potential private-sector FX liquidity needs under scenarios and compare with international reserves; not a comprehensive reserve adequacy assessment but a supplement to broader frameworks.

### Liquid asset buffer structure, contagion mechanics, and pecking orders (Box 2)
- Liquid asset categories and contagion:
  - LC assets (domestic issuers): contagion to domestic private institutions; valuation losses from LC depreciation.
  - FX assets (domestic issuers): contagion domestically; no valuation losses from LC depreciation.
  - Global safe assets (FX, outside money): no domestic contagion if sold to foreign agents; no valuation loss from LC depreciation.
- Central bank contagion channels:
  - Reserve deposit outflows reduce central bank liabilities and international reserves.
  - FX market intervention and FX liquidity provision to banks reduce net international reserves.
- Pecking order matters materially:
  - If IFs/NFCs liquidate domestic bank deposits first, banks experience deposit outflows and can pass stress to central bank.
  - If they liquidate global safe assets first (sold to foreigners), contagion can be limited.
- Methodological choice: examine multiple pecking orders rather than assume pro-rata liquidation.

### Illustrative EM hypothetical: sectoral balance sheets and calibration (Box 2 details)
- NFC balance-sheet highlights (selected rows, in millions of USD; percentages preserved as stated):
  - Cash and cash equivalent: 420 (LC) / 400 (FX) = 820 total; 39% of TA
  - Bank deposits, domestic: 300 (LC) / 200 (FX) = 500 total; 24% of TA
  - Bank deposits, external: 0 (LC) / 200 (FX) = 200 total; 9% of TA
  - Total TA: 1,720 (LC) / 400 (FX) = 2,120 total
  - NFCs: nearly half of liabilities in FX; 60 percent of FX borrowings are from external sources; NFC liquid assets about 40 percent of total assets; NFC liquid FX assets about 40 percent of FX borrowings.
- Banks balance-sheet highlights (selected rows, in millions of USD; percentages preserved as stated):
  - Cash + CB reserve: 120 (LC) / 280 (FX) = 400 total; 25% of TA
  - Resident Deposits: 620 (LC) / 500 (FX) = 1,120 total; 69% of TA
  - Excess reserve: 58 (LC) / 174 (FX) = 232 total; 14% of TA
  - International bonds: 0 (LC) / 90 (FX) = 90 total
  - Total TA: 886 (LC) / 734 (FX) = 1,620 total
  - Memo: Reserve requirement (FX): 20 percent of FX customer deposits; Reserve requirement (LC): 10 percent of LC customer deposits.
- Central bank balance-sheet highlights (selected rows, in millions of USD):
  - International reserves: 0 (LC) / 370 (FX) = 370 total; 60% of TA
  - FX deposit abroad: 0 (LC) / 35 (FX) = 35 total
  - Bank reserves: 28 (LC) / 28 (FX) = 56 total
  - Total TA: 250 (LC) / 370 (FX) = 620 total; Total Liability: 290 (LC) / 330 (FX) = 620 total
  - Banks’ reserves are 45 percent of the central bank's total liabilities; excess reserves nearly 30 percent of total liabilities and three-quarters of international reserves.
- Illustrative stress-test calibration parameters (as stated):
  - FX deposit reserve requirement, %: 20
  - Exchange rate depreciation rate, %: 30
  - Banks: nonresident deposit outflows, % of stock: 50
  - Banks: rollover of maturing external debt from parent banks, % of maturing debt: 90
  - Banks: rollover of maturing external borrowing from unrelated banks, % of maturing debt: 0
  - Banks: rollover of maturing external bonds, % of maturing bonds: 0
  - Banks: rollover rate of maturing borrowing from domestic central banks and banks, % of maturing exposures: 100
  - Banks: household panic deposit run, % of stock: 0, 5
    - Note: 5% is the LCR assumption for stable retail deposits. 10% is the LCR assumption for less stable deposits.
  - Banks: NFC panic deposit run, % of remaining deposits after using them to fill NFC liquidity gap: 0, 10
  - NFCs: rollover of domestic bank loans, % of maturing loans: 50
    - Note: 50% is 50% net inflows from maturing customer loans...
  - NFCs: rollover of FX borrowing, excluding domestic bank loans, % of maturing debt: 0
  - Sovereign: liquidation rate of LC sovereign debt held by foreign investors, % of outstanding debt: 30
  - Central bank: FX market intervention, USD: 5 million

### Illustrative stress-test outcomes and sensitivities (Box 2 and Section 1)
- Key illustrative quantitative outcomes:
  - NFCs: have 200 million FX liquid assets in domestic banks and 200 million in foreign banks.
  - If NFCs use domestic deposits first, banks suffer US$80 million deposit outflows to fill NFCs’ liquidity gap.
  - With a 50 percent outflow of non-resident deposits, total deposit outflows without panic are US$95 million.
  - Outflows related to sales of LC sovereign bonds by foreign investors are US$92 million.
  - Net cash outflows from operations total US$24 million.
  - A domestic customer deposit run due to panic totals US$27 million.
  - Banks’ overall FX liquidity gap is US$16 million, which is absorbed by NIR in the illustrative example.
  - Both GIR and NIR remain positive in the worst-case scenario considered.
- Pecking order and panic effects:
  - NFCs withdraw 40 percent of domestic liquid assets if they use local deposits first; withdrawals decline to zero if they use foreign deposits first.
  - Banks’ excess reserve withdrawal rate declines from 100 to 60 percent under alternative pecking orders.
  - Pecking-order effects on the central bank balance sheet are larger than panic-deposit-run effects under standard LCR assumptions.
  - If panic-related deposit outflow rate exceeds 15½ percent, runs would have a severer impact than pecking-order differences.
- Sensitivity to banks’ customer loan refinancing rates:
  - Tested refinancing rates: 90, 50, and 30 percent.
  - Lower refinancing rates shift liquidity from NFCs to banks and increase NFCs’ deposit withdrawal rate; net effects on banks’ liquidity are positive in the example.
  - Changing refinancing rates has a much smaller impact than changing NFCs’ pecking order because NFCs hold more foreign assets.

### FX liquidity, policy experiments, and central bank options (Section 1 and Boxes 2–3)
- Policy levers analyzed:
  - Reserve requirement adjustments: lowering reserve requirement by four percentage points to 16 percent in the example eliminates need for central bank liquidity support.
  - Central bank liquidity support vs reserve requirement cut:
    - Liquidity support replaces NIR with domestic FX claims on banks and does not change central bank balance sheet size; reserve requirement cut reduces central bank liability and balance sheet size and raises NIR in the example.
  - Other safety-net tools: temporary release of macroprudential buffers, systemic emergency liquidity facilities with expanded collateral and counterparties, asset purchase programs, bilateral swap lines, multilateral contingent financing (IMF credit lines), private sector credit lines.
- Prudential design considerations:
  - HQLA composition matters: global safe assets can be higher quality in FX stress; Basel LCR treats foreign sovereigns as Level 2 with limits (up to 40 percent of HQLA) and haircuts (15 percent) which may be undesirable during capital outflow/dollar funding shocks.
  - Consider an FX LCR and adjust HQLA definitions and valuation haircuts where appropriate; align reserve requirements' currency composition with FX-LCR design.
  - Avoid policies that force banks to hold excessive liquidity due to misalignment between reserve requirements and HQLA definitions.
- Moral hazard and preventive measures:
  - Robust prudential measures reduce moral hazard from ex-post liquidity support.
  - Where central bank LOLR capacity is limited (no reserve-currency issuer), preventive measures (limits on FX positions, holding global safe assets, improved reporting) and building international reserves are important.

### Limitations, extensions, and recommended analytical approach (Box 3)
- Limitations:
  - Sector-aggregate BSA data omit micro-level amplification, collateral channels, encumbered assets, off-balance-sheet items, and derivatives margin calls.
  - Transaction-level data are resource-intensive and often confidential.
  - Second-round effects (fire-sale price impacts) are hard to estimate reliably; tool intentionally avoids detailed modeling of second-round effects to limit model risk.
- Recommended approach:
  - Use the Excel-based tool as a "roughly right" high-level vulnerability assessment.
  - Examine multiple behavioral and parameter assumptions to identify critical drivers of results.
  - If key risks are identified in NBFIs, NFCs, or data-limited markets, prioritize building granular data and more advanced models for targeted analyses.
  - If extending to second-round effects, apply a range of models/coefficients to capture model and parameter uncertainty rather than a single narrow specification.

*IMF Working Paper — Systemwide Liquidity Stress Testing Tool (wpiea2022252-print-pdf)*

### Executive Summary ......................................................................................................

### Executive Summary

### Nature and forms of systemic liquidity crises
- A systemic liquidity crisis occurs when multiple financial institutions simultaneously face liquidity stress.
- Historical forms:
  - Massive bank runs on multiple or systemically important banks when corporates and households experienced profound liquidity shocks or panic.
  - Market runs (post-GFC) affecting everyone trading in distressed markets and likely to spill over to other markets.
  - COVID-19 example: freezing economic activity and cashflows among households and corporates; massive liquidation of safe assets caused turmoil even in the safest asset markets, such as U.S. treasury markets.

### Literature, monitoring, and limitations
- Rich academic literature exists on theory of systemic liquidity crises, including game theoretic models from historical bank run experiences.
- Post-GFC studies focused on amplification channels: "run on repos", collateral channels, fire sales, interaction of market and funding liquidity, the leverage cycle, and cash hoarding.
- Regulators monitor liquidity risks using quantity and price/spread indicators and conduct sector-specific liquidity stress tests for banks and investment funds.
- Limitations for systemwide liquidity stress tests:
  - Data constraints and analytical challenges.
  - Need to incorporate interconnectedness among banks, NBFIs, non-financial sectors, and foreign investors.
  - Behavioral factors are important but hard to pin down and can change results noticeably.
  - Some granular data require resource-intensive processing.
  - Hard to standardize tests because liquidity events can differ across systems.

### Proposed tool: Systemwide liquidity stress testing (Excel-based)
- Purpose and scope:
  - A simple Excel-based “systemwide liquidity stress testing” tool using a limited data environment, mostly BSA data.
  - Focuses on systemwide liquidity stress—aggregate liquidity shocks to the system.
  - Formulates sector-by-sector cashflow liquidity stress tests akin to typical bank liquidity stress tests and links them to gauge domestic contagion through direct financial linkages.
- Core features and concepts:
  - Emphasizes liquidity contagion across sectors because one sector’s liquid asset buffers are another sector’s funding instruments.
  - Highlights the importance of assessing alternative behavioral and parameter assumptions to mitigate model risk.
  - Consciously avoids modeling second-round effects because of concerns about model risk.
- Scope and limitations:
  - High-level vulnerability assessment useful for identifying critical liquidity risk propagation channels for further granular analysis.
  - Primarily relies on sector-aggregate BSA data; limited scope to capture idiosyncratic shocks that become systemic via agent-level amplification.
  - Applicable to numerous countries and executable with few resources.

### Applications and policy relevance
- Uses and policy linkages:
  - Can inform macroprudential policies, crisis management frameworks, and priorities for improving risk analyses and data.
  - Ex-ante preventive policies and ex-post safety net tools are both vital.
  - Once critical behavioral patterns are identified, risk monitoring, prudential tools, and crisis management tools can be designed to incentivize stability-supporting behavior.
  - Can set priorities for closing data gaps and expanding risk monitoring to include NBFIs and NFCs if systemically important.
  - Can quantify potential liquidity support needs and the extent to which prudential measures should be relaxed under shocks, helping central banks avoid unnecessarily generous liquidity support (e.g., blanket guarantees).
- FX/BOP shock assessment:
  - When scenarios include balance of payment (BOP) shocks, the tool can help estimate potential foreign exchange (FX) liquidity needs from the private sector and compare them with the stock of international reserves.
  - Relevant for economies without a reserve currency where central bank capacity to act as lender of last resort (LOLR) is constrained by international reserves.
  - Not intended as a comprehensive reserve adequacy assessment; should be used to supplement broader frameworks for assessing reserve adequacy.

### Integration and prior use
- The tool and similar methods have been applied in IMF work across several countries:
  - Exercises focused on domestic liquidity contagion and on the impact of foreign liquidity loss on financial institutions and central bank balance sheets.
- Future uses:
  - Strengthen integration of Financial Sector Assessment Programs (FSAPs) and Article IV consultations.
  - Enrich policy dialogue under the Integrated Policy Framework (IPF).

*IMF Working Paper — Systemwide Liquidity Stress Testing Tool (Executive Summary)*

### Introduction

### Introduction

### Background
- Systemic liquidity crisis: occurs when multiple financial institutions simultaneously face liquidity stress; can take numerous forms depending on financial system structure.
- Historical manifestations:
  - Massive bank runs on multiple or systemically important banks when corporates and households experience profound liquidity shocks or panic.
  - Runs in financial markets (GFC example; IMF 2011).
  - Stress in investment and money market funds and contagion, including dollar funding shocks to banks engaging in cross-border activities (IMF 2019a).
  - COVID-19: systemic liquidity crisis caused by freezing economic activity and cashflows among households and corporate, with spillovers to financial institutions (including through credit lines, Kapan and Minoiu, 2021); substantial sales of liquid assets caused unusual turmoil even in the U.S. treasury market (Vissing-JØrgensen, 2021).
- Mechanisms amplifying market dislocation: fire sales, leverage cycle, cash hoarding, network effects; market runs tend to be systemic and spill over to other markets.
- Non-bank financial institutions (NBFIs):
  - Generally more exposed to financial market risks because of reliance on market funding or fickle investments (investment funds), and assets primarily marketable instruments.
  - More prone to market liquidity and asset valuation change risks than banks; for some NBFIs (asset managers) liquidity risks and propagation through interconnectedness are more relevant than solvency risks.
  - Financial market infrastructures (FMIs) such as central clearing counterparties (CCPs) face risks from liquidity stress of clearing members via jumps in margin calls; example: U.K. pension funds suffered sudden jumps in margin calls due to sharp valuation losses of gilts in October 2022.
  - Market stress can be caused by non-financial market participants (highly-leveraged real estate developers or commodity firms) when prices change drastically.
- Theoretical and empirical literature:
  - Microeconomic bank-run theory: Diamond and Dybvig (1985) for idiosyncratic bank runs; Allen and Gale (1998, 2000, 2004), Morris and Shin (1998), Holmstrom and Tirole (1998, 2010), Freixas, Parigi, and Rochet (2000), Rochet and Vives (2004) on systemic runs and uncertainty-driven amplification.
  - Post-GFC focus: run on repos and collateral channels (Gorton and Metrick 2012); fire sales; interaction of market and funding liquidity (Brunnermeier and Pedersen 2008); leverage cycle (Adrian and Shin 2010); cash hoarding and network effects (Allen and Gale 2004; Allen and others 2009).
- Current monitoring and limitations:
  - Regulators, central banks, and international organizations identify and monitor key liquidity markets and conduct sector-specific microprudential stress tests (FSAPs include “systemic liquidity assessment” describing key markets, participants, liquidity flows, and infrastructure and crisis management frameworks).
  - Academic liquidity measures include trade size, trade volume, trading frequency, bid-ask spreads, yield curve noise; examples: Amihud (2002), Benos and Zikes (2016), Kyle's lambda, Cont and Schaanning (2017), Duffie (2010) and Chen and Duffie (2021).
  - NBFI supervisors’ stress tests tend to focus on sectoral resilience without systemwide interlinkages.
- Data constraints limiting systemwide liquidity stress tests:
  - Need to incorporate interconnectedness among banks, NBFIs, foreign investors, and non-financial sectors (IMF 2021).
  - Cross-sectoral financial exposures and transaction data (activity-based data from FMIs) are critical but often lacking historically; post-crisis initiatives (G20 DGI, G20 2009) encouraged sector-aggregate data collection (BSA by IMF (IMF 2015); who-to-whom data by ECB; flow-of-funds).
  - Comprehensive institution-level, security-level, liquidity and capital buffer data often do not exist and are treated with confidentiality and administrative firewalls even within regulators.
- Technical and modeling challenges even when data exist:
  - Transaction-level data are “big” and resource-intensive to process.
  - Liquidity stress test results depend on hard-to-model behaviors and contagion effects; fire-sales and asset price distress channels are difficult to robustly estimate; models are often highly stylized and focus on single behavioral channels.
  - Systemic liquidity events vary across systems, making standardization difficult.

### Contribution of This Paper
- Proposes a simple Excel-based “systemwide liquidity stress testing” tool primarily using BSA data.
- Defines systemwide liquidity stress as a sub-category of systemic liquidity stress caused by aggregate liquidity shocks to the system, such as economy-wide earnings shocks (COVID-19) and balance of payment (BOP) shocks (Box 1).
- Tool features:
  - Formulates sector-by-sector cashflow liquidity stress tests akin to bank liquidity stress tests (e.g., Basel III liquidity coverage ratio—LCR).
  - Links sector tests to gauge possibility of domestic contagion through network of direct financial linkages.
  - Highlights transfer of stress: sector A may be resilient individually but can transfer stress to sector B if liquidated assets include liabilities issued by sector B, making the joint system vulnerable.
- Applicability:
  - Can assess liquidity spillover risks from NBFIs and non-financial sector depending on BSA data detail (can separate insurers, pension funds, asset managers, other NBFIs).
  - Can assess spillovers from NFCs and sovereigns to local financial institutions if BSA captures exposures reported by regulated financial institutions.
  - Acknowledges BSA data limitations: often miss instrument detail and linkages not involving domestic financial institutions; supplemented with financial statements, supervisory reporting, and proxy assumptions can yield indicative vulnerabilities.
- Role in analysis hierarchy:
  - Intended as a high-level vulnerability assessment to identify critical liquidity risk propagation channels for deeper investigation with granular data.
  - BSA-alone is insufficient for market liquidity risks from margin calls or off-balance-sheet items (bank credit lines) and not suited for idiosyncratic-to-systemic amplification that requires micro-level behavior and richer models (e.g., agent-based models, ABMs).
  - Benefits: applicable across many countries including EMDEs and AEs as BSA data availability grows; low-resource Excel implementation; easily enhanced to incorporate institution- and instrument-level data without moving to complex ABMs.
- Use for policy and crisis management:
  - Inform macroprudential policies and crisis management frameworks to prepare for potential systemwide liquidity events.
  - Encourage prudential requirements for institutions to self-insure against systemwide liquidity stress ex-ante; prioritize holding and using global safe assets during BOP-related stress to mitigate domestic spillovers.
  - Reinforce central bank role as lender of last resort (LOLR) within contingent crisis management frameworks designed to avoid moral hazard via appropriate collateral, valuation, haircuts, and terms.
  - Stress test can estimate potential liquidity injection needs under various macrofinancial and behavioral assumptions to guide whether releasing existing liquidity buffers suffices or additional ELA is needed; central bank may need to expand eligible collateral and counterparties if liquidity stress originates outside banking sector.
  - Macroprudential supervisors may need expanded legal powers to request data across financial supervisors and broader economic agents (including NFCs) when relevant for systemic risks.
- Application to reserve adequacy and FX liquidity:
  - For central banks without reserve currencies, the tool can estimate potential FX liquidity needs from the private sector under crisis scenarios and compare with international reserves.
  - If international reserves are insufficient, policy options include tighter prudential requirements, building international reserves, bilateral central bank swaps, multilateral contingent financing (including IMF credit lines), and private sector credit lines where accessible.
  - Tool is not a comprehensive reserve adequacy assessment; full assessments should incorporate other external stability factors (e.g., import coverage) as in IMF’s ARA framework (IMF 2016a); the tool can supplement international reserve needs assessments and aligns conceptually with the foreign currency liquidity data under IMF’s SDDS (IMF 2013a).
- Existing applications and integration with IMF work:
  - Tools and similar methods have been applied in several countries: Philippines (COVID-19 loan moratoria effects on bank-NFC liquidity), Luxembourg (bank-investment fund linkages), Andorra, Mauritius, Sweden (impact of foreign liquidity loss on institutions and central bank balance sheets), 2018 Armenia FSAP (reserve requirement design based on systemwide FX liquidity), and other TA to EMs and LICs.
  - Recent FSAPs increasingly incorporate systemwide liquidity analysis, enriching the tool with granular/confidential data:
    - Philippines FSAP: supplemented BSA with institution-level bank and NFC data.
    - 2021 Chile FSAP: used institution-level bank and some NBFI data to show institution-level vulnerabilities from pension portfolio reallocation.
    - 2022 Mexico FSAP: used granular BSA including institution-level bank and NBFI exposures to local sovereign bonds (cash, repo, derivatives) and detailed liquid asset buffer information to assess capital outflow scenarios and margin-call-driven liquidity needs.
  - Potential to strengthen integration of FSAP and Article IV surveillance and support integrated policy framework (IPF, IMF 2020).
- Organization of the paper:
  - Explains the tool conceptually and demonstrates application using illustrative EMDE data focusing on NFC-bank-central bank FX liquidity linkages.
  - Focuses on aggregate FX liquidity stress from capital outflow, relevant for EMDEs and advanced economies without a reserve currency, as a key channel linking external and financial stability.
  - Discusses roles of prudential and crisis management policies and how the stress test informs them quantitatively.
  - Provides details of some country cases before concluding.

*IMF WORKING PAPERS Systemwide Liquidity Stress Testing Tool — Introduction*

### Box 1. Why Are BOP Shocks and Foreign Liquidity Special in Systemwide Liquidity Analysis?

### Box 1. Why Are BOP Shocks and Foreign Liquidity Special in Systemwide Liquidity Analysis?

### Definitions and taxonomy of liquidity
- Liquidity in an open economy grouped into four categories by source:
  - Domestic Liquidity: liquidity created by domestic agents (local currency and other financial claims).
  - Foreign Liquidity: financial claims issued by foreign agents.
  - Private Liquidity: created by the private sector (bank loans, commercial papers, bonds, and others).
  - Public Liquidity: created by the public sector (currency and sovereign bonds).
- Both domestic and foreign liquidity can be denominated in local and foreign currencies.
- Public/private distinction follows literature on private and public liquidity (Holmstrom and Tirole 1998; Tirole 2008, 2011).

### Why foreign liquidity and BOP flows matter for net system liquidity
- Only foreign liquidity channeled by BOP flows can add or remove net liquidity from the economy; domestic agents alone cannot create net liquidity in the system because:
  - The net supply of domestic liquidity is zero: somebody's financial assets are somebody else's liabilities.
  - Example: deposit outflows between domestic banks or purchases of domestic bonds/equity reallocate liquidity but do not change total gross liquidity.
  - If non-residents withdraw deposits and repatriate them, the money is lost from the economy, causing an aggregate liquidity shock.
- Theoretical framing: inside vs outside money
  - Inside money: IOUs among agents within the system; net supply zero.
  - Outside money: deposits from outside the system (foreign liquidity when the system is the whole economy) that can change net supply and cause aggregate, non-diversifiable shocks.

### Central bank LOLR role and foreign liquidity as outside money
- In closed-economy formulations, public money (central bank liquidity facility and public debt) is outside money and can mitigate contractions of private liquidity by adding positive net supply.
- Extending to open economies: foreign liquidity functions as outside money and is a potential source of LOLR facilities.
- Constraints on foreign liquidity provision:
  - Collateral constraints limit borrowing from international markets to pledgeable assets (tradable goods).
  - If tradable goods are narrowly defined as existing stock of past foreign liquidity, pledgeable assets equal foreign assets held by residents.
  - The local central bank’s capacity to act as LOLR in FX is limited to international reserves.
- Policy implication:
  - Economies without a reserve currency need to carry certain levels of international reserves to prepare for potential systemwide liquidity stress.
  - Reserve currency issuers are interpreted as having abundant pledgeable assets and less concern about access to international capital markets in crises.
  - EMDEs (emerging market and developing economies) typically have fewer pledgeable assets and may face sudden loss of foreign liquidity despite benefiting from international capital markets in normal times.
- Note from literature: it may not be optimal to hold excessive international reserves to fully self-insure against foreign liquidity dry-up risks.

### FX liquidity stress versus currency mismatch
- Currency mismatch: amount of FX assets and liabilities is uneven across all maturities → solvency risk monitored via net open FX limits.
- FX liquidity risk: arises from maturity mismatch of FX assets and liabilities → liquidity risk can exist even without currency mismatch if FX assets mature much later than FX liabilities.

### Implications for systemwide liquidity stress testing (overview)
- Four main steps of systemwide liquidity stress tests:
  1. Understand key sectors and financial linkages; visualize network structure of the financial system and sectoral balance sheet structure to identify vulnerabilities and match macro-financial liquidity stress scenarios.
  2. Formulate cashflow liquidity stress testing at sector-aggregate level for systemically important sectors (similar to bank liquidity stress tests).
  3. Gauge contagion effects from one sector to another under behavioral assumptions—this contagion step differentiates systemwide tests from institution-specific tests.
  4. Consider impact of central bank liquidity policy options to determine economy-wide resilience with and without policy support.

### Designing macro-financial scenarios that generate aggregate liquidity shocks
- Sources of aggregate liquidity shocks include both foreign and domestic agents:
  - Foreign: withdrawal of local deposits by foreign investors; selling of domestic securities; non-refinancing of maturing debt; external debt covenants triggering early repayment; conversion of proceeds to hard currency causes FX outflows.
  - Domestic: portfolio reallocation to foreign assets (including cash hoarding), margin calls from exposures to foreign derivatives, reductions in remittances, changes in asset prices, exchange rates, and interest rates.
  - Higher foreign interest rates and LC depreciation increase FX cash outflows for debt service; commodity price declines reduce export proceeds.
- Scenario design approaches:
  - Macro scenario approach:
    - Use a functioning macrofinancial model to produce quantitative scenarios including policy, exchange, and market interest rates, public finance forecasts, government cashflows, and trade projections.
    - Supplement with empirical models when macro models do not include necessary parameters.
    - Statistical adjustment may be needed when stress horizon (e.g., one-month) differs from macro horizon (e.g., annual averages).
  - Statistical and historical approach:
    - Rely on historical data to obtain empirical distributions matched to stress-test horizon (e.g., weekly parameter changes for a one-week test).
    - Pick assumptions to match desired likelihood of stress; example language preserved exactly: "If the tester takes the worst two percentile points of weekly interest changes (i.e., the value at risk, VaR) of one-sided tail 2 percent), the test aims at one week per year event."
    - Include long-enough history ideally covering crisis events; use peer experiences if domestic history lacks stress events.

### Sectoral cashflow liquidity stress tests (cashflows, liquid assets, haircuts)
- Cashflow stress test structure:
  - Compare cash inflows and outflows within a test horizon and test whether liquid asset buffers cover net cashflow gaps.
  - Cash inflows include: revenue from operations (sales, fees, salary, remittance income, tax revenue), investment income (dividends, debt service income, realized gains/losses), financing (rollover of existing debt, drawdown of credit lines, new financing if available).
  - Cash outflows include: operational expenses (payments for imports), investment expenses (debt service, dividend payments, margin calls, responding to called credit lines), financing (deposit outflows for banks, contingent financings like debt covenants and credit lines).
  - Liquid assets include cash and highly-liquid cash-like instruments (bank deposits, sovereign securities, highly-rated private sector bonds).
- Valuation and haircuts:
  - Liquid asset values decline under market distress; higher risk premiums reduce bond valuations (haircuts).
  - Exchange rate depreciation reduces value of LC-denominated liquid assets when expressed in FX.
  - In typical bank FX liquidity stress tests, a "100 percent haircut" is applied to LC assets and excludes them fully from liquid asset buffers.
  - When additional data available, define “rapidly usable” liquid assets by excluding required reserves and encumbered assets pledged as collateral.

### Contagion mechanics and pecking order of liquidation
- Contagion can occur even if the sector directly affected by a BOP shock initially withstands the shock (example: banks and investment funds linkages).
- Transmission channels:
  - Foreign investor sales reduce asset valuations and cause margin calls.
  - LC depreciation reduces LC asset valuation in FX terms.
  - IFs with sufficient domestic bank deposits may withstand outflows but transfer liquidity shock to banks via deposit withdrawals.
  - Banks may transfer stress to IFs by liquidating IF shares.
- Strength of contagion depends substantially on pecking order of asset liquidation:
  - No contagion if IFs first liquidate global safe assets rather than domestic bank deposits.
  - No contagion from banks to IFs if banks liquidate global safe assets, central bank reserves, and sovereign bonds before selling IF shares.
- Methodological choice:
  - The paper examines a range of outcomes under multiple pecking orders of asset liquidation rather than assuming a single behavioral rule (e.g., pro-rata liquidation).

### Example and illustrative considerations
- Global safe assets include deposits at the central bank and sovereign securities of reserve currency issuers that preserve liquidity and value during global distress.
- Network visualization and BSA matrix help identify sectors affected by various BOP shocks and the propagation pathways across balance sheets.

*IMF Working Papers — Box 1. Why Are BOP Shocks and Foreign Liquidity Special in Systemwide Liquidity Analysis?*

### Box 2. Liquid Asset Buffer Structure and Contagion Effects

### Box 2. Liquid Asset Buffer Structure and Contagion Effects

### Liquid asset buffer categories and contagion effects
- Local currency assets, domestic issuers (inside money, accepted only for domestic transactions)
  - Contagion to other domestic private-sector institutions (through, for example, bank deposit withdrawal, sales of securities, redemption of investment funds, repos, contingent financing with domestic agents)
  - Possible contagion to the local central bank (even without ELA) through domestic banks' LC reserve deposits and access to standing liquidity facilities with LC securities as collateral
  - Valuation losses from local currency depreciation
- FX assets, domestic issuers (inside money, accepted only for domestic transactions)
  - Contagion to other domestic private-sector institutions (through, for example, bank deposit withdrawal, sales of securities, repos, the redemption of investment funds, contingent financing with domestic agents)
  - Possible contagion to the local central bank (even without ELA) in fully or partially dollarized economies through domestic banks' FX reserve deposits and access to standing liquidity facilities with FX securities as collateral
  - No valuation losses from local currency depreciation
- Global safe assets (FX assets, outside money, accepted for cross-border and domestic transactions)
  - No contagion to other domestic private-sector institutions if sold to foreign agents, creating an offsetting capital inflow
  - No contagion to the local central bank unless it takes foreign securities as collateral for standing liquidity facilities
  - No valuation losses from local currency depreciation

### Central bank contagion channels and implications
- Liquidity demand from banks
  - Reserve deposits (excess reserves and, often, required reserves) are a significant portion of banks’ high-quality liquid assets where Basel III LCR is implemented.
  - Even if banks maintain required LCR ratios using excess reserves, the central bank experiences an outflow of reserve deposits (contagion effects).
  - Banks can fill liquidity gaps by using standing central bank facilities with qualifying collaterals.
- FX liquidity provision through intervention
  - LC asset sales by foreign investors create FX liquidity outflows when they convert LC proceeds to FX for repatriation; if foreign investors exchange with local banks, this creates FX outflows from banks.
  - If banks use FX reserve deposits, the liquidity demand spills over to the central bank.
  - If the central bank intervenes in FX markets to limit volatility, FX liquidity demand spills over to the central bank.
- Policy-relevant uses of analysis
  - Quantify percentage points reduction of reserve requirements needed to restore banks’ liquidity balance if banks exhaust all liquidity other than required reserves.
  - Inform central banks’ international reserve adequacy for financial stability by linking FX liquid assets held by the central bank (mostly global assets) to international reserve assessments.

### Illustrative example: system characteristics (hypothetical EM data)
- Systemic importance and linkages
  - Banks and NFCs have large financial footprints (size measured as sum of gross financial assets and liabilities).
  - Asset managers are small but critical conduits between foreign investors and domestic government bond markets.
- Direct exposure to capital outflow shocks
  - NFCs are mostly exposed to foreign investors, followed by AMs and banks.
  - Exposures are mostly in FX, except AMs who channel foreign investors into LC government bonds.
  - All foreign liabilities are in FX except for investments in LC government bonds.
- Domestic transmission channels
  - NFCs' FX liquid asset buffer includes substantial global safe assets, followed by domestic FX deposits; redemption from AMs can cause domestic financing challenges to the government.
  - Banks' FX liquid assets are mostly FX reserve deposits; stress transmits to the central bank and reduces international reserves.
  - The government lacks additional global safe assets; international reserves are the only FX liquid asset buffers of the broad public sector.
- Additional channels
  - Households hold substantial FX deposits at local banks and global safe assets (including FX cash at home).
  - Bank behavior (reallocating away from LC assets and domestic FX bank deposits to global safe assets) can amplify capital outflows and stress.

### Sectoral balance sheet highlights (illustrative figures)
- NFCs
  - NFCs have nearly half of liabilities in FX; 60 percent of FX borrowings are from external sources.
  - NFC liquid assets (cash and cash equivalent) about 40 percent of total assets.
  - NFC liquid FX assets are about 40 percent of FX borrowings.
  - NFC balance sheet (selected rows, in millions of USD):
    - Cash and cash equivalent: 420 (LC) / 400 (FX) = 820 total; 39% of TA
    - Bank deposits, domestic: 300 (LC) / 200 (FX) = 500 total; 24% of TA
    - Bank deposits, external: 0 (LC) / 200 (FX) = 200 total; 9% of TA
    - Total TA: 1,720 (LC) / 400 (FX) = 2,120 total
- Banks
  - Highly dollarized: nearly half of bank assets and liabilities in FX.
  - Banks' FX assets are about 95 percent of FX liabilities.
  - About 60 percent of FX assets are loans (customer and interbank loans).
  - Banks' balance sheet (selected rows, in millions of USD):
    - Cash + CB reserve: 120 (LC) / 280 (FX) = 400 total; 25% of TA
    - Resident Deposits: 620 (LC) / 500 (FX) = 1,120 total; 69% of TA
    - Excess reserve: 58 (LC) / 174 (FX) = 232 total; 14% of TA
    - International bonds: 0 (LC) / 90 (FX) = 90 total
    - Total TA: 886 (LC) / 734 (FX) = 1,620 total
  - Memo:
    - Reserve requirement (FX): 20 percent of FX customer deposits
    - Reserve requirement (LC): 10 percent of LC customer deposits
- Central bank
  - Banks’ reserves are 45 percent of the central bank's total liabilities; excess reserves nearly 30 percent of total liabilities and three-quarters of international reserves.
  - Only 10 percent of GIR is the central bank’s own NIR under the paper's definition.
  - Central bank balance sheet (selected rows, in millions of USD):
    - International reserves: 0 (LC) / 370 (FX) = 370 total; 60% of TA
    - FX deposit abroad: 0 (LC) / 35 (FX) = 35 total
    - Bank reserves: 28 (LC) / 28 (FX) = 56 total; 45% of liabilities noted elsewhere
    - Total TA: 250 (LC) / 370 (FX) = 620 total; Total Liability: 290 (LC) / 330 (FX) = 620 total

### Stress testing design and scenario (illustrative example)
- Main stress sources
  - Stress from FX debt service of NFCs, capital outflows from household and non-resident deposits, and foreign investors' demand for FX to repatriate sales of LC investment.
- Key scenario assumptions
  - Domestic currency depreciation: 30 percent
  - Lower-quality domestic liquid assets (LC and FX issued by domestic agents) lose value via haircuts; FX reserves at central bank and global safe assets maintain full value.
  - Test horizon: one month (LCR benchmark usable)
- Stress test focus
  - Central bank, banks, and NFCs
  - Excludes second-round impacts (no feedback of fire sales on asset prices/exchange rates)
  - Main outputs: decline of liquid assets and their composition; size of liquidated assets by asset type; additional financing needs if a sector runs out of all liquid assets
- Pecking order options
  - NFCs: liquidate domestic FX assets first then global safe assets, or vice versa
  - Banks: may liquidate pro-rata across asset types; benchmark assumption: banks not allowed to liquidate required reserves
  - Policy experiments include effects of relaxing reserve requirements

### Additional stress testing parameters and calibration approaches
- Four parameter types: macro scenario, granular data, behavioral, liquidity-related policy
- Calibration approaches summarized (Table 5 outline)
  - Macro scenario: use macroeconomic models, empirical/statistical approaches, prorating for shorter horizons when needed
  - Granular data: instrument-by-counterpart, remaining maturities, contingent financing; use proxies if unavailable
  - Behavioral: refinancing/rollover rates, deposit withdrawal rates, pecking orders — hard to pin down; use Basel III LCR benchmarks and range of assumptions
  - Liquidity-related policy: central bank policy rate, reserve requirements, FX market intervention, ELA — test ranges and hypothetical policy experiments
- Consistency requirement across sectors for some parameters (e.g., refinancing assumptions between banks and NFCs)
- Benchmark assumptions about inter-financial-institution exposures: contractual inflows received 100 percent and refinanced, resulting in 0 net cash inflows (benchmark template)

### Stress testing parameter assumptions used in the example (Table 6)
- FX deposit reserve requirement, %: 20
- Exchange rate depreciation rate, %: 30
- Banks: nonresident deposit outflows, % of stock: 50
- Banks: rollover of maturing external debt from parent banks, % of maturing debt: 90
- Banks: rollover of maturing external borrowing from unrelated banks, % of maturing debt: 0
- Banks: rollover of maturing external bonds, % of maturing bonds: 0
- Banks: rollover rate of maturing borrowing from domestic central banks and banks, % of maturing exposures: 100
- Banks: household panic deposit run, % of stock: 0, 5
  - Note: 5% is the LCR assumption for stable retail deposits. 10% is the LCR assumption for less stable deposits.
- Banks: NFC panic deposit run, % of remaining deposits after using them to fill NFC liquidity gap: 0, 10
- NFCs: rollover of domestic bank loans, % of maturing loans: 50
  - Note: 50% is 50% net inflows from maturing customer loans, which reflect a 0% default rate and 50% rollover rate, among other possibilities.
- NFCs: rollover of FX borrowing, excluding domestic bank loans, % of maturing debt: 0
- Sovereign: liquidation rate of LC sovereign debt held by foreign investors, % of outstanding debt: 30
- Central bank: FX market intervention, USD: 5 million

### Results and interpretation (illustrative outcomes)
- Figure 6 structure and scenarios
  - Panels 1-4: assume no customer deposit run due to confidence loss
  - Panels 5-6: include a panic deposit run
  - Panels 1-2 and 5-6: NFCs use domestic bank deposits first and banks use excess reserve first
  - Panels 3-4: NFCs/banks use other liquid FX assets first
  - In all cases, required reserves decline when banks experience FX customer deposit outflows subject to reserve requirements
  - If banks lack sufficient usable liquid FX assets ("gap"), they ask for central bank liquidity support; central bank assumed to provide full support and post-stress international reserves reflect total systemwide liquidity shortage
  - FX market intervention and FX liquidity support for banks reduce net international reserves
- Key quantitative illustrative findings
  - NFCs have 200 million FX liquid assets in domestic banks and 200 million in foreign banks.
  - If NFCs use domestic deposits first, banks suffer from US$80 million deposit outflows to fill NFCs’ liquidity gap.
  - With a 50 percent outflow of non-resident deposits, total deposit outflows without panic are US$95 million.
  - Outflows related to sales of LC sovereign bonds by foreign investors are US$92 million.
  - Net cash outflows from operations total US$24 million.
  - A domestic customer deposit run due to panic totals US$27 million.
  - Combined effects: deposit outflows and sales of LC sovereign bonds are the most significant drivers reducing FX liquidity of banks and the central bank when NFCs use domestic bank deposits first.
- Role of pecking order and panic runs
  - Pecking order of NFCs substantially alters financial stability impact because NFCs hold substantial foreign assets.
  - Central bank holds two-thirds of foreign FX liquid assets; NFCs hold most of the remaining third.
  - Without panic, NFCs withdraw 40 percent of domestic liquid assets if they use local deposits first; withdrawal rates decline to zero if NFCs use foreign deposits first.
  - Banks’ excess reserve deposit withdrawal rate declines from 100 to 60 percent under alternative pecking orders.
  - Pecking order effects on the central bank balance sheet are larger than panic deposit-run effects under standard LCR assumptions.
  - If panic-related deposit outflow rate exceeds 15½ percent, runs would have a severer impact than pecking-order differences.

*Source: IMF Working Paper — Box 2. Liquid Asset Buffer Structure and Contagion Effects (illustrative example and analysis).*

### 1. FX liquid

### 1. FX liquid assets

### Bank FX cash flow analysis: key outcomes
- Banks run out of usable FX liquid asset buffers and need FX liquidity support only when there is a panic run.
- Banks’ overall FX liquidity gap is US$16 million, which is absorbed by NIR.
- Both GIR and NIR remain positive in the worst-case scenario considered, indicating the central bank can withstand the stress in this example.
- Alternative central bank actions:
  - Lowering the reserve requirement by four percentage points to 16 percent eliminates the need for liquidity support.
  - Post-stress GIR is the same whether the central bank uses liquidity support or reserve requirement adjustment.
  - NIR is higher with reserve requirement adjustment because it reduces the central bank's liability and balance sheet size.
  - Liquidity support replaces NIR with domestic assets (FX claims to domestic banks) and does not change central bank balance sheet size; this causes opposite monetary policy impacts compared with reserve requirement adjustment.

### Sensitivity to banks’ customer loan refinancing rates
- Tested refinancing rates: 90, 50, and 30 percent.
- Lower refinancing rates:
  - Shift liquidity from NFCs to banks and increase NFCs’ deposit withdrawal rate.
  - Net effects on banks' liquidity are positive in the example; banks' asset liquidation rates decline with lower customer loan refinancing rates.
- Impact on international reserves:
  - Remains the same when NFCs and banks liquidate domestic assets first, because banks exhaust all excess reserves even with 0 percent rollover rates.
  - When banks retain some excess reserves after stress, lower refinancing rates help conserve international reserves.
- Overall: changing refinancing rates has a much smaller impact on the financial sector's liquidity than changing NFCs’ pecking order of asset liquidation, because NFCs have a stronger liquidity position in the considered stress and have much more foreign assets than banks.

### Sectoral distribution of FX liquidity and liabilities (summary statements)
- About half of NFC FX liquid assets and liabilities are foreign, while most banks' exposures are vis-à-vis domestic sectors.
- All of the central bank's FX liquid assets are foreign, whereas most of its liabilities are with domestic banks.
- Two-thirds of FX foreign liquid assets are the central bank's international reserves, and the rest are mostly NFCs'.
- Banks hold a majority of LC liquid assets.
- Seventy percent of FX foreign liabilities are of NFCs, followed by banks.
- Banks have the largest share in domestic FX liabilities; both NFCs and the central bank hold notable shares of the rest.

### Figure-based assumptions highlighted
- One-month horizon with a 30 percent LC depreciation rate.
- Reserve requirement in baseline example: 20 percent for FX deposits.
- Foreign investors liquidate 10 percent of their LC sovereign bond holdings.
- Interbank and central bank exposures are repaid and refinanced fully, generating no net cash flows.
- Note definitions: GIR = gross international reserve; NIR = net international reserve; NFC = non-financial corporates; LC = local currency; FX = foreign exchange.

### Policy discussion: preventive and safety-net measures
- Preventive (ex-ante) measures:
  - Microprudential tools: Basel minimum (Pillar 1) and additional Pillar 2 requirements tailored to each bank’s risk profile.
  - Macroprudential tools: countercyclical capital buffer (CCyB), systemic risk buffers (SRBs), limits on FX position, reporting requirements for external borrowings and hedging.
  - Aim: limit exposures to systemic liquidity risks and increase buffers to absorb shocks.
- Safety-net (ex-post) tools:
  - Releasing macroprudential buffers temporarily (allowing banks to go below required LCR levels).
  - Lowering reserve requirements.
  - Systemic and emergency liquidity support (temporary liquidity facilities with expanded collateral and sometimes wider counterparts).
  - Asset purchase programs.
  - FX liquidity provision often backstopped with bilateral central bank swaps.
  - Some emerging market economies have used FX intervention and capital flow measures when facing capital outflow pressures.
- Moral hazard considerations:
  - Robust prudential measures are critical to avoid moral hazards associated with liquidity support.
  - Central bank liquidity support is now widely expected when systemic liquidity stress materializes; recent crises (e.g., March 2020) illustrate new liquidity facilities introduced in crisis periods.
- Role of preventive measures where central bank LOLR capacity is limited:
  - FX-liquidity support of central banks without reserve-currency status is often limited to international reserves; bilateral swaps may supplement but are often available only to advanced economies and major emerging markets.
  - Preventive measures become more important: limit exposures to systemic FX liquidity risks and encourage building liquidity buffers (self-insurance).
  - Data show limits on FX position are the most frequently used macroprudential measures among EMDEs.

### Specific prudential design considerations and recommendations
- Liquidity Coverage Ratio (LCR):
  - Global safe assets (foreign sovereign securities) can substantially limit liquidity spillovers and may be "higher quality" than domestic liquid assets for FX stress.
  - Basel LCR treats foreign sovereign securities as Level 2 assets with limits (up to 40 percent of HQLA) and valuation haircuts (15 percent), which may be undesirable when key risks are capital outflows and dollar funding shocks.
  - Consider changing the definition of HQLA if national authorities introduce an FX LCR.
  - Modify prohibitions that bar banks from holding foreign securities to allow investing in international-reserve-qualifying global safe assets where appropriate.
- Reserve requirements:
  - Currency composition of reserve requirements matters for resilience against FX liquidity stress.
  - Many EMDEs set FX reserve requirements that, if satisfied by LC reserve deposits at the central bank, could be drawn down when banks face FX liquidity stress.
  - FX reserves could be part of HQLA in FX LCR in jurisdictions meeting certain preconditions.
  - If FX reserve requirements are met with LC deposits, those reserves would not count as FX HQLA under FX LCR; central banks may need to change reserve requirements to be filled in FX to align measures.
  - Imposing FX LCR and reserve requirements without alignment could force banks to hold excessive liquidity; design details must be coordinated.

*Source: IMF Working Paper — "Systemwide Liquidity Stress Testing Tool"*

### Box 3. Structure of HQLA and Reserve Requirements

### Box 3. Structure of HQLA and Reserve Requirements

### Relationship between HQLA and reserve requirements
- In many EMs, the structure of HQLA is closely related to the details of reserve requirements.
- When short-term money markets (including treasury bills) are underdeveloped, central banks often manage liquidity using reserve requirements and central bank papers; both create central banks' liabilities to banks, which are, in turn, banks' liquid assets.
- In dollarized economies, the details of FX reserve requirements affect the HQLA and central bank balance sheet structure.
- Footnotes and survey findings:
  - According to the IMF survey of monetary operations and instruments database (MOID), 83 out of 125 respondent central banks had FX reserve requirements as of 2018.
  - As of end-2017, 38 countries had liquidity requirements by currency, including FX LCR (IMF macroprudential policy database).

### Country examples and specific institutional arrangements
- Armenia
  - Highly dollarized economy with a fixed exchange rate regime and a history of significant external shocks and severe depreciation pressures.
  - Reserve requirements for FX deposits are filled in domestic currency.
  - The 2018 FSAP (IMF 2018) recommended considering the FX deposit reserve requirement as a prudential rather than a monetary policy tool and starting filling it in FX to build more FX liquidity buffer.
- Philippines
  - Post-1997 reforms tightened prudential regulation to limit exposures to FX and FX liquidity risks.
  - Banks were required to limit their open FX positions to US$50 million (as of end-2019) irrespective of bank size.
  - Separate licensing required to create a "foreign currency deposit unit" with its own FX liquidity requirements; central bank recognized limitations of its FX liquidity support capacity.
  - As of 2021, the FX liquidity requirement was abolished, as the country adopted Basel III LCR.
- Türkiye
  - Early 2010s: introduced the reserve options mechanism (ROM) allowing banks to fill domestic currency deposit reserve requirements in FX and gold, but not vice versa—encouraging banks to self-insure against FX liquidity shocks while elevating contagion risk to the central bank.
  - The share of banks’ FX deposits over total central bank foreign assets stood at about 57 percent as of end-2019 and has been nearly 70 percent several times since 2017.
  - The policy boosted gross GIR but not NIR. The quality of international reserves deteriorated even more during the COVID-19 crisis, and the NIR has been negative as of fall 2021.
  - Analytical central bank balance sheet source: Türkiye‘s central bank.

### Systemwide liquidity stress testing tool: implications for reserve policy and surveillance
- The tool can quantify:
  - Potential amount for liquidity support.
  - Potential temporary reduction of FX LCR and reserve requirements.
- If ex-post support measures are undesirably high, the tool can help regulators:
  - Identify which prudential measures should be tightened and by how much.
  - Identify measures to enhance LOLR firepower.
- Surveillance scope:
  - May need expansion to include NBFIs and non-financial sectors as needed.
  - Several market stress events caused by non-financial market participants (e.g., corporate bond market stress from highly-leveraged real estate developers such as Evergrande; Enron in 2001; Wirecard in 2020).
  - Example: U.K. pension funds suffered from a sudden jump in margin calls due to sharp valuation loss of gilts in October 2022.
- Identifying key sectors and behaviors:
  - The tool can identify bank behaviors, contingency financing plans, and inform resolution planning and living wills.
  - NBFIs and non-financial sectors could be systemically important; macroprudential risk monitoring and powers might need to be expanded to include them.
  - Illustrative example: mobilizing liquid foreign assets held by NFCs is critical because NFCs have about one-third of the nation’s liquid foreign assets following the central bank; macroprudential authorities might need to extend risk monitoring to include some large NFCs.

### Monetary, liquidity, and external stability tradeoffs in FX stress
- BOP-related systemic FX liquidity stress entails monetary, liquidity, and external stability issues and policy tools.
- Potential conflict in fixed exchange rate systems:
  - Size of central bank liability and balance sheet tied to international reserves.
  - If banks withdraw excess reserves, international reserves and central bank balance sheet size fall.
  - Where the exchange rate system does not allow sterilization or other operations to expand central bank money, contraction of international reserves leads to monetary tightening—necessitating defense of the exchange rate at the cost of financial stability.
- Even with flexible exchange rates, tradeoffs between exchange rate stability from capital outflows and financial stability often lead EMDE central banks to tighten policy rates procyclically to avoid currency crises first (Jacome and others 2011), with exceptions observed during the COVID-19 crisis.
- Figure 9 scenario: central bank required to back 100 percent of its liability in international reserves (simplified case); assumes no sterilization operation. (Figure source: Author.)

### Conclusions on the tool, limitations, and principles
- Purpose: Proposes a simple Excel-based systemwide liquidity stress testing tool to assess vulnerability of a financial system and economy against systemic liquidity risks.
- Rationale: Developing reliable systemic liquidity stress testing models across institutions and sectors is challenging because of data constraints and behavioral factors; the tool simplifies analysis by focusing on economic-sector-aggregate data and behavior.
- Key limitations:
  - Relies on sector-aggregated data, assuming away micro-level amplification mechanisms (e.g., cash hoarding at institution levels).
  - Granular supervisory or FMI-collected transaction data are necessary to assess:
    - Collateral channels for secured funding like repo.
    - Effects of encumbered assets.
    - Liquidity needs from off-balance sheet items (like credit lines).
    - Margin calls from derivatives transactions.
  - These channels are particularly relevant in AEs with developed money markets and complex NBFI sectors and require advanced modeling.
  - Some countries lack sufficiently granular BSA data; proxy data may be needed (central bank and aggregate bank balance sheets, government financial statistics, BOP, international investment position, external debt data).
- Core principles that remain applicable:
  - Aggregate liquidity shocks spill over across sectors because agents’ liquid assets are somebody else’s funding.
  - Test results depend critically on agents’ behavior—especially the pecking order of liquid asset use. Many existing experiments assume pro-rata liquidation, but the pecking order can create substantial differences in cross-sector contagion effects.
  - Macro-financial assumptions matter: acute liquidity events usually last only for a short period because central banks typically take measures; short-term volatility could be severely underestimated with a standard VaR approach relying on recent history or past crises.
- Approach to second-round effects:
  - The paper focused on a "roughly right" tool and avoided assessing second-round effects from asset fire sales to prioritize robustness to model uncertainty.
  - Extending the tool to capture second-round effects is possible but may not provide reliable estimates without extensive empirical analyses; estimating reliable price-quantity relationships during extreme tail events is challenging.
  - Recommendation: build a "robust" stress testing approach incorporating model and parameter uncertainty of second-round effects by applying a range of potential models/coefficients rather than narrowing in on specific values.
  - Supporting quote and context: during the GFC, some markets and investors experienced a 25 standard deviation shock for multiple days—a 25-standard deviation event should occur only once in every 13.7 billion years assuming standard normal distribution (Haldane, 2009).
- Policy utility:
  - A "roughly right" stress testing tool can inform prudential policies, crisis management frameworks, and priorities to improve risk analyses and data.
  - Exercises that examine multiple behavioral and parameter assumptions can identify key assumptions with large influence on results; once identified, risk monitoring, prudential tools, and crisis management tools can be adjusted to incentivize agents to act in support of financial stability.
  - If high-level vulnerability assessments indicate key risks arise from NBFIs, NFCs, or data-limited markets, resources should be allocated to develop data and tools for those critical assessments.

*Source: Box 3. Structure of HQLA and Reserve Requirements, from the IMF Working Paper "Systemwide Liquidity Stress Testing Tool".*

### References

### wpiea2022252-print-pdf - References

### Liquidity, Funding, and Market Liquidity
- Adrian, T., and H.S. Shin. 2010. “Liquidity and Leverage.” Journal of Financial Intermediation, 19(3), 418-37.
- Amihud, Y. 2002. “Illiquidity and Stock Returns: Cross-Section and Time-Series Effects.” Journal of Financial Markets, (5)1: 31-56.
- Brunnermeier, M.K., and L.H. Pedersen. 2008. “Market Liquidity and Funding Liquidity.” The Review of Financial Studies, 22(6),2201-38.
- Bagehot, W., 1873. Lombard Street: A Description of the Money Market. New York: Scribner, Armstrong & Company.
- Fleming, M. 2003. “Measuring Treasury Market Liquidity.” FRBNY Economic Policy Review.
- Roll, R. 1984. “A Simple Implicit Measure of the Effective Bid-Ask Spread in an Efficient Market.” The Journal of Finance.
- Tirole, J. 2011. “Illiquidity and All Its Friends.” Journal of Economic Literature, 49(2), 287-325.
- Tirole, J. 2008. “Liquidity Shortages: Theoretical Underpinnings.” Banque de France Financial Stability Review: Special Issue on Liquidity, 11, 53-63.
- Vissing-Jorgensen, A., 2021. The treasury market in spring 2020 and the response of the federal reserve. Journal of Monetary Economics, 124, pp.19-47.
- Benos, E., and F. Zikes. 2016. “Liquidity Determinants in the UK Gilt Market.” Staff Working Paper No. 600. Bank of England, London.
- Hüser, A.C., C. Lepore, and L. Veraart. 2021. “How Do Secured Funding Markets Behave under Stress? Evidence from the Gilt Repo Market.” Staff Working Paper No. 910, Bank of England, London.
- Hüser, A.C., C. Lepore, and L. Veraart. 2021. “How does the Repo Market Behave under Stress? Evidence from the COVID-19 Crisis.” IMF Working Paper, International Monetary Fund, Washington, D.C.

### Central Bank Intervention, Policy Instruments, and Reserves
- Allen, F., E. Carletti, and D. Gale. 2009. “Interbank Market Liquidity and Central Bank Intervention.” Journal of Monetary Economics, 56(5), 639-52.
- Alper, K., H. Kara, and M. Yorukoglu. 2013. “Reserve Options Mechanism.” Central Bank Review, Vol. 13. Central Bank of the Republic of Türkiye.
- Carlson, M., B. Duygan-Bump, and W. Nelson. 2015. “Why Do We Need Both Liquidity Regulations and a Lender of Last Resort? A Perspective from Federal Reserve Lending during the 2007-09 US Financial Crisis.” BIS Working Paper No. 493. Bank for International Settlements, Basel, Switzerland.
- Goldber g, L., C. Kennedy, and J. Miu. 2010. “Central Bank Dollar Swap Lines and Overseas Dollar Funding Costs.” FRBNY Staff Report No. 29, Federal Reserve Bank of New York.
- Moreno, R. 2010. “Central Bank Instruments to Deal with the Effects of the Crisis on Emerging Market Economies.” BIS Papers No. 54. Bank for International Settlements, Basel, Switzerland.
- Federal Reserve Board. 2013. Quarterly Report on Federal Reserve Balance Sheet Developments, March.
- International Monetary Fund (IMF). 2013a. International Reserves and Foreign Currency Liquidity: Guidelines for A Data Template. Washington, D.C.: International Monetary Fund.
- International Monetary Fund (IMF). 2016a. “Guidance Note on the Assessment of Reserve Adequacy and Related Considerations.” IMF Policy Paper. International Monetary Fund, Washington, D.C.
- Irving Fisher Committee (IFC) on Central Bank Statistics. 2016. “Development of Financial Sectoral Accounts.” IFC Working Paper No.15. Bank for International Settlements, Basel, Switzerland.
- Ishi, K., M. Stone, and E. Yehoue. 2009. “Unconventional Central Bank Measures for Emerging Economies.” IMF Working Paper WP/09/226. International Monetary Fund, Washington, D.C.
- Oura, H. 2013. “Uses and Risks with Foreign Currency Liquidity Assistance in Systemic Crisis—Is It a Good Idea for Russia? 2013 IMF Article IV Consultation to Russian Federation, Selected Issues.” IMF Country Report No. 13/311. International Monetary Fund, Washington, D.C.

### Stress Testing, Macroprudential Policy, and Supervision
- Bouveret, A., 2017. “Liquidity Stress Tests for Investment Funds: A Practical Guide.” IMF Working Paper WP/17/226. International Monetary Fund, Washington, D.C.
- Committee on Payments and Market Infrastructures (CPMI) and Board of the International Organization of Securities Commissions (IOSCO). 2018. Framework for Supervisory Stress Testing of Central Counterparties (CCPs).
- European Banking Authority (EBA). 2018. EBA Report on Liquidity Measures under Article 509(1) of the CRR.
- European Central Bank (ECB). 2018. “Systemic Liquidity Concept, Measurement and Macroprudential Instruments.” Occasional Paper Series No. 214. ECB Task Force on Systemic Liquidity.
- Fache Rousová, L., A. Jukonis, E. Letizia, and M. Gravanis. 2020. “Derivatives-Related Liquidity Risk Facing Investment Funds.” Financial Stability Review, 1. European Central Bank, Frankfurt.
- Han, F., and M. Leika. 2019. “Integrating Solvency and Liquidity Stress Tests: The Use of Markov Regime-Switching Models.” IMF Working Paper WP/19/250. International Monetary Fund, Washington, D.C.
- International Monetary Fund (IMF). 2014a. “Staff Guidance Note on Macroprudential Policy.” IMF Policy Paper. International Monetary Fund, Washington, D.C.
- International Monetary Fund (IMF). 2014b. “Staff Guidance Note on Macroprudential Policy: Detailed Guidance on Instruments.” IMF Policy Paper. International Monetary Fund, Washington, D.C.
- International Monetary Fund (IMF). 2014c. “Staff Guidance Note on Macroprudential Policy: Considerations for Low Income Countries.” IMF Policy Paper. International Monetary Fund, Washington, D.C.
- International Monetary Fund (IMF). 2015. “Balance Sheet Analysis in Fund Surveillance.” IMF Policy Paper. International Monetary Fund, Washington, D.C.
- International Monetary Fund (IMF). 2021a. 2021 “Financial Sector Assessment Program Review—Background Paper on Quantitative Analysis.” IMF Policy Paper No. 2021/041. International Monetary Fund, Washington, D.C.
- International Monetary Fund (IMF). 2021b. “2021 Comprehensive Surveillance Review—Overview Paper.” IMF Policy Paper No. 2021/027. International Monetary Fund, Washington, D.C.
- International Monetary Fund (IMF). 2021c. “2021 Financial Sector Assessment Program Review—Towards a More Stable and Sustainable Financial System.” IMF Policy Paper No. 2021/040. International Monetary Fund, Washington, D.C.
- International Monetary Fund (IMF). 2021d. “2021 Comprehensive Surveillance Review—Background Paper on Systemic Risk and Macroprudential Policy Advice in Article IV Consultations.” IMF Policy Paper No. 2021/033. International Monetary Fund, Washington, D.C.
- International Monetary Fund (IMF). 2011. Global Financial Stability Report (April), Chapter 2, “How to Address the Systemic Part of Liquidity Risk.” Washington, D.C.: International Monetary Fund.
- Dobler, M., S. Gray, D. Murphy, and B. Radzewicz. 2016. “The Lender of Last Resort Function after the Global Financial Crisis.” IMF Working Paper WP/16/10. International Monetary Fund, Washington, D.C.
- Daokui, D., and P. Tucker, 2014. “Triennial Surveillance Review-External Study—Risks and Spillovers.” IMF Policy Paper. International Monetary Fund, Washington, DC.

### Contagion, Fire Sales, and Systemic Risk Transmission
- Allen, F., and D. Gale. 1998. “Optimal Financial Crises.” The Journal of Finance, 53(4), 1245-84.
- Allen, F., and D. Gale. 2000. “Financial Contagion.” Journal of Political Economy, 108(1), 1-33.
- Allen, F., and D. Gale. 2004. “Financial Fragility, Liquidity, and Asset Prices.” Journal of the European Economic Association, 2(6), 1015-48.
- Caballero, R.J., E. Farhi, and P.O. Gourinchas. 2017. “The Safe Assets Shortage Conundrum.” Journal of Economic Perspectives, 31(3),29-46.
- Calomiris, C.W., and J.R. Mason. 2003. “Fundamentals, Panics, and Bank Distress during the Depression.” American Economic Review, 93(5),1615-47.
- Cont, R., and L. Wagalath. 2016. “Fire Sales Forensics: Measuring Endogenous Risk.” Mathematical Finance, 26(4), 835-66.
- Cont, R., and E. Schaanning. 2017. “Fire Sales, Indirect Contagion and Systemic Stress Testing.” Working Paper.
- Jotikasthira, C., C. Lundblad, and T. Ramadorai. 2012. “Asset Fire Sales and Purchases and the International Transmission of Funding Shocks. The Journal of Finance, 67(6), 2015-50.
- Pelizzon, L., M.G. Subrahmanyam, D. Tomio, and J. Uno. 2016. “Sovereign Credit Risk, Liquidity, and ECB Intervention: Deus Ex Machina?” Journal of Financial Economics.
- Liu, A., M. Paddrik, S.Y. Yang, and X. Zhang. 2020. “Interbank Contagion: An Agent-based Model Approach to Endogenously Formed Networks.” Journal of Banking & Finance, 112.
- Paddrik, M., S. Rajan, and H.P. Young. 2020. “Contagion in Derivatives Markets.” Management Science, 66(8), 3603-16.
- Kapan, T. and Minoiu, C., 2021. Liquidity insurance vs. credit provision: Evidence from the COVID-19 crisis. Credit Provision: Evidence from the COVID-19 Crisis (September 30, 2021).

### Market Structure, Trading, and Microstructure Modeling
- Amihud, Y. 2002. “Illiquidity and Stock Returns: Cross-Section and Time-Series Effects.” Journal of Financial Markets, (5)1: 31-56.
- Lee, C.M., and M.J. Ready (n.d.). 1991. “Inferring Trade Direction from Intraday Data.” The Journal of Finance, 46(2), 733.
- Paddrik, M., R. Hayes, A. Todd, S. Yang, P. Beling, and W. Scherer. 2012. “An Agent- Based Model of the E-Mini S&P 500 Applied to Flash Crash Analysis.” In 2012 IEEE Conference on Computational Intelligence for Financial Engineering & Economics (CIFEr), 1-8. March. Institute of Electrical and Electronics Engineers, Piscataway, NJ.
- Lagos, R. 2010. “Inside and Outside Money.” In S.N. Durlauf and L.E. Blume (editors), Monetary Economics. The New Palgrave Economics Collection. London: Palgrave Macmillan.
- Duffie, D. 2010. “Presidential Address: Asset Price Dynamics with Slow‐Moving Capital.” The Journal of Finance, 65(4),1237-67.
- Roll, R. 1984. “A Simple Implicit Measure of the Effective Bid-Ask Spread in an Efficient Market.” The Journal of Finance.
- Levels, A., R. de Sousa van Stralen, S. Kroon Patrescu, and I. van Lelyveld. 2018. “CDS Market Structure and Risk Flows: The Dutch Case.” De Nerderlandsche Bank (DNB) Working Paper No. 592.
- Benos, E., and F. Zikes. 2016. “Liquidity Determinants in the UK Gilt Market.” Staff Working Paper No. 600. Bank of England, London.

### Derivatives, CCPs, Clearing, and Market Infrastructure
- Aramonte, S., and W. Huang. 2019. “Costs and Benefits of Switching to Central Clearing.” BIS Quarterly Review, December.
- Faruqui, U., W. Huang, and E. Takats. 2018. “Clearing Risks in OTC Derivatives Markets: The CCP-Bank Nexus.” BIS Quarterly Review, December.
- Huang, W., A. Menkveld, and S. Yu. 2019. “Central Counterparty Exposures in Stressed Markets.” BIS Working Paper No. 833. Bank for International Settlements, Basel, Switzerland.
- CPMI and IOSCO. 2018. Framework for Supervisory Stress Testing of Central Counterparties (CCPs).
- Financial Stability Board (FSB). 2017. Guidance on Central Counterparty Resolution and Resolution Planning.
- Financial Stability Board (FSB). 2021. COVID-19 Support Measures: Extending, amending, and ending.

### Additional Theory, Historical, and Cross-Border Perspectives
- Diamond, D.W., and P.H. Dybvig. 1983. “Bank Runs, Deposit Insurance, and Liquidity.” Journal of Political Economy, 91(3), 401-19.
- Goodfriend, M., and R. King. 1988. “Financial Deregulation, Monetary Policy and Central Banking.” In Williams Haraf and Rose Marie Kushmeider (editors), Restructuring Banking and Financial Services in America. Lanham, MD: AEI Studies 481, American Enterprise Institute.
- Gorton, G. 1988. “Banking Panics and Business Cycles.” Oxford Economic Papers, 40(4), 751-81.
- Gorton, G., S. Lewellen, and A. Metrick. 2012. “The Safe-Asset Share.” American Economic Review, 102(3),101-06.
- Gorton, G., and A. Metrick. 2012. “Securitized Banking and the Run on Repo.” Journal of Financial Economics, 104(3), 425-51.
- Morris, S., and H.S. Shin. 1998. “Unique Equilibrium in a Model of Self-Fulfilling Currency Attacks.” American Economic Review, 587-97.
- Kiyotaki, N., and J. Moore. 1997. “Credit Cycles.” Journal of Political Economy, 105(2), 211-48.
- Laeven, L., and F. Valencia. 2018. “Systemic Banking Crises Revisited.” IMF Working Paper No. 18/206. International Monetary Fund, Washington, D.C.
- Vieira, F., M. Holland, and M. Resende. 2012. “Financial Dollarization and Systemic Risks: New Empirical Evidence.” Journal of International Money and Finance.
- Basso, H., O. Calvo-Gonzalez, and M. Jurgilas. 2011. “Financial Dollarization: The Role of Foreign-Owned Banks and Interest Rates.” Journal of Banking and Finance, 35.
- Jotikasthira, C., C. Lundblad, and T. Ramadorai. 2012. “Asset Fire Sales and Purchases and the International Transmission of Funding Shocks. The Journal of Finance, 67(6), 2015-50.
- Goldberg, L., C. Kennedy, and J. Miu. 2010. “Central Bank Dollar Swap Lines and Overseas Dollar Funding Costs.” FRBNY Staff Report No. 29, Federal Reserve Bank of New York.

*wpiea2022252-print-pdf - References*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022252-print-pdf.pdf_