## wpiea2024101-print-pdf ## Source details **Canonical URL:** [wpiea2024101-print-pdf](https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024101-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2024/english/wpiea2024101-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2024/english/wpiea2024101-print-pdf.pdf.json) --- ### Executive summary — overview - Study focus: EMDE central bank interventions during COVID-19 in core markets (money, securities, FX funding) aimed at combating market dysfunction (distinct from quantitative easing). - Methodology: empirical analysis (panel/local projections) across 74 countries covering 90 percent of global GDP, complemented by case studies. - Main conclusion: Interventions by EMDE central banks were in line with financial stability mandates and were effective at addressing market dysfunctions during the COVID-19 pandemic, often through large announcement effects. ### Timing and scale of impacts - Impacts strongest in the first two months after program announcement. - Effects diminished over time as other factors (e.g., slow return of foreign investors) adversely affected liquidity. - Transaction sizes varied across jurisdictions: significant in Chile and Indonesia; limited in the Philippines and South Africa. - In EMDEs there were cases where interbank market activity dried up and interbank interest rates fell relative to central bank policy rates; advanced economy markets proved more resilient. ### Effects on market segments — securities, money, FX - Securities markets: - Interventions stabilized prices; prices improved significantly from depressed March 2020 levels. - Evidence of improved liquidity: higher volumes traded and narrower bid-offer spreads. - Much of the impact came from “announcement effects.” - Positive spillovers from major central banks aided EMDE interventions; in some cases markets improved without EMDE transactions. - Money markets: - Large liquidity injections sometimes reduced interbank activity, though overall markets held up reasonably well. - EMDEs with more developed markets fared better. - Interventions generally mitigated negative repercussions from liquidity expansion. - FX funding markets: - Initial widening of bid-offer spreads and CIP deviations; post-intervention spreads contracted and CIP deviations reversed. - Federal Reserve swap line announcements were pivotal in normalizing functioning, reducing bid-offer spreads and reversing CIP deviations. - EMDE U.S. dollar operations moderately improved CIP deviations (greatest impact 10–15 days post-announcement) but were not estimated to significantly narrow bid-offer spreads on their own. - Announcement effects often drove effectiveness, even for smaller-scale operations. ### Non-bank financial institutions (NBFIs) and financial stability - COVID-19 exposed NBFI vulnerabilities: precautionary cash demands, investor runs, and loss of securities market liquidity. - EMDE central banks sometimes adjusted interventions to assist NBFIs, dealing with non-standard counterparties (examples: Chile, India). ### Exiting intervention programs - Exiting was more challenging in EMDEs than in AEs. - Advanced markets: support withdrawn relatively easily as liquidity improved. - EMDEs: exits complicated by expanded government financing needs and absent capital inflows; some EMDEs retained or expanded asset purchases (India, Indonesia). - Easier exits: standard, well-understood programs (short-term repo operations, FX swaps). - Program design facilitating self-liquidation recommended (pricing spreads, short-term operations, liquid securities). ### Policy conclusions and recommendations (high-level) - Specify objectives clearly, distinguishing market functioning from monetary accommodation and government financing; communicate exit strategies ex-ante tied to objectives. - Set realistic objectives acknowledging interventions may be small or short-lived and other fundamentals drive market activity. - Use liquidity-based triggers (not price-based); maintain operational flexibility with qualitative and quantitative assessments. - Interventions should be large enough to address dysfunction but account for fiscal dominance, moral hazard, and balance sheet risks. - Facilitate self-liquidation through pricing, short-duration instruments, and purchases of sufficiently liquid securities. - Prepare operationally and legally in advance: identify core markets, define dysfunction metrics, expand collateral and counterparty frameworks if needed. - Normalize by reducing balance sheets to sizes required for efficient monetary policy; reduce long-duration and higher credit risk assets accumulated during crises. *Source: IMF Working Papers — EMDE Central Bank Interventions during COVID-19 to Support Market Functioning (source content as provided).* ### Empirical approach and coverage - Coverage: 74 countries representing 90 percent of global GDP. - Markets analyzed: interbank money, government bond, and FX swaps/funding markets. - Liquidity metrics: price-based, volume-based, volatility-based indicators (bid-offer spreads, asset swap spreads, CIP deviations, Amihud ratio, trading volumes). - Data frequency: daily converted to weekly moving averages for money and government bond markets. - FX funding analysis: decomposition into global common factor and idiosyncratic factors via principal components. - Econometrics: pooled cross-sectional and time-series panel data with local projections (Jordà 2005); controls included Fed swap line announcements, country fixed effects, U.S. MOVE index, and VIX. ### Empirical findings — money markets (summary) - Turbulence was temporary and limited: increased interbank rate volatility and lower trading volumes. - Term repos and reserve requirement reductions: impulse responses largely negligible; most pronounced effects within 10–20 days post-announcement when present. - FX funding support announcements diminished interest rate volatility relative to trading volumes. - No detectable spillovers from interventions in other economies to EMDE money markets in panel estimates. - Interbank rates generally remained anchored to policy rates on average at repo announcements; reserve requirement cuts reduced interbank rates in some cases. ### Empirical findings — government securities markets (summary) - Bond market liquidity deteriorated sharply after the shock: wider bid-ask and asset swap spreads; lower trading volumes; increased Amihud ratio. - Asset purchase program (APP) announcements associated with: - Declines in government bond spreads to swaps. - Tighter bid-offer spreads. - Amihud ratio broadly unchanged in some specifications; turnover/market turnover unchanged in aggregate. - Timing: effects typically evident within 1–2 weeks after announcements. - No evidence that EMDE interventions caused adverse spillovers within EMDE bond market liquidity; global factors strongly influenced EMDE liquidity. ### Empirical findings — FX funding markets (summary) - U.S. dollar scarcity: larger CIP deviations and wider FX swap bid-offer spreads. - Variance decomposition: AE FX funding dislocations dominated by a global common factor; EMDEs had larger country-specific idiosyncratic drivers. - Federal Reserve swap line announcements: pivotal in reducing bid-offer spreads and reversing CIP deviations. - EMDE dollar-providing operations: moderate improvement in CIP deviations (greatest impact 10–15 days after announcements); bid-offer spread narrowing not statistically significant for local operations. - Global risk factors (VIX) had marginal/insignificant direct role in driving bid-offer spreads in FX funding markets during the period. ### Program sizes and announcement effects - Available EMDE FX intervention program sizes: maximum program sizes range from 0.1 to 3.8 percent of GDP; median program size 2 percent of GDP. - Much of program impact derived from announcement effects; large committed volumes often unnecessary to achieve market functioning results. ### Case study synthesis — cross-country highlights - Countries examined in depth: Chile, India, Indonesia, Philippines, Poland, South Africa. - Objectives across cases: - Supporting market functioning was a core objective in all cases. - Most programs also pursued monetary accommodation and government financing (SARB emphasized market functioning only). - Market focus: - Interbank money markets supported via short-term repos and lengthened maturities. - Government bond purchases prominent (typically secondary market); Indonesia used both primary and secondary purchases. - FX funding support prominent in Chile, Indonesia, India (often underutilized). - Some cases targeted non-government securities where market structure warranted (Chile bank bonds; Poland agency bonds). - Triggers: spikes in price volatility, reversals in non-resident flows, bond spread widening, increased liquidity demand, signs of NBFI fire-sales. - Transparency varied: Chile and India published results; South Africa conducted covert purchases; Indonesia and the Philippines did not publish immediate results. - Operational capacity constraints: Poland and the Philippines faced frictions implementing bond purchases. - Balance sheet expansions and risks (selected Table 3 figures, Feb–Dec 2020): - Chile: Change in Balance Sheet Size 88.6 %; Change in Net Local Currency Assets 13.9 % GDP; Change in Domestic Bond Holdings 1,242.7 %; 14.7 % GDP; 2,152.5 %; 14.7 % GDP. - India: 25.5 %; 5.9 % GDP; 35.7 %; 1.9 % GDP; 31.7 %; 1.6 % GDP. - Indonesia: 24.6 %; 4.0 % GDP; 75.7 %; 3.7 % GDP; 102.8 %; 3.4 % GDP. - Philippines: 38.5 %; 11.0 % GDP; 123.6 %; 6.9 % GDP; 512.3 %; 6.4 % GDP. - Poland*: 35.1 %; 8.0 % GDP; 119.5 %; 5.5 % GDP; N/A; 3.3 %. - South Africa: -4.7 %; -0.9 % GDP; 1.9 %; 0.1 % GDP; 404.5 %; 0.6 % GDP. - Average: 34.6 %; 7.0 % GDP; 266.5 %; 5.5 % GDP; N/A; 5.0 %. - Credit risk: generally low where purchases were government securities or lending collateralized/short-term; Chile’s bank bond accumulation represented higher credit risk. - Exit: AE exits quicker and smoother; EMDE exits more varied and prolonged, particularly for novel or long-duration asset purchase programs. ### Selected country case study highlights - Chile (BCCh): - Context: bank debt securities market almost double government debt market end-2019; BCCh prohibited from buying government debt until legal change in August 2020. - Measures: extended repo and FX swap maturities; corporate bonds included as collateral; bank bond purchase program announced March 20, 2020 (up to five-year maturities, limits per issuance); Special Asset Purchase Program totaling $8 billion; funding-for-lending scheme ~ $40 billion. - Effectiveness: restored key market functioning; Local Stress Index normalized relatively quickly; interbank trading volumes plummeted and interbank rate fell to corridor floor; some persistence in reduced interbank activity. - Exit: FX liquidity programs included end-dates and naturally liquidated; bank bond purchases exit was more prolonged and complex; BCCh maintained reinvestment programs into 2021 then stopped reinvestment of coupons/redemptions when policy rate rose mid-2021. - Indonesia (Bank Indonesia, BI): - Policy measures: FX reserve requirements cut from 8 to 4 percent; rupiah reserve requirements cut by 2.5 percent; bond repo maturities extended up to 12 months; FX swap auctions increased to daily frequency; “triple intervention” (FX NDFs, spot interventions, government bond purchases primary and secondary). - Quantities: total allocation under National Economic Recovery Program = 4.4 percent of GDP in 2020; primary market purchases = IDR 473.4 trillion in 2020 (IDR 76 trillion via “market mechanism”); BI financed about half of remaining deficit in 2020. - Effectiveness: exchange rate volatility fell in April–May 2020; bond markets stabilized from April 2020; BI purchases contributed to lower local bond yields and bid-ask spreads and higher trading volumes; excess reserves increased and interbank activity halved compared to 2019. - Exit and risks: bond purchases extended into 2022 and burden-sharing agreements extended; concerns about fiscal dominance and BI balance sheet risks; total public debt ~ 40 percent of GDP. - India (RBI): - Context: well-developed government bond market; pre-existing shadow banking stresses; policy rate reduced by 135 bps between Feb 2019–Mar 2020. - Measures: $2 billion FX swap auctions; policy rate cuts totaling 135 bps and corridor widened by 40 bps; scaled OMO and Special OMO purchases; 100-bps cut to cash reserve ratio; LTROs/TLTROs up to three years; Special Lending Facility for Mutual Funds. - Effectiveness: FX swap oversubscribed; term repos/LTROs oversubscribed; holdings of government bonds increased by more than 60 percent since pandemic start, reaching about 7.5 percent of GDP. - Exit: many LTROs repaid early; some facilities extended and TLTROs/floating rate on-demand facilities continued until December 2021; Government Securities Acquisition Program suspended October 2021 but OMOs continued. - Philippines (BSP): - Context: early-stage local currency markets; reserve requirements high at 12 percent; FX swap markets more developed. - Measures: reduced sterilization and reserve requirements; peso deposit and reverse repo auctions; government bond purchase window; advance dividend to government; PHP 300 billion repo to government; reduced sterilization operations. - Effectiveness: interest rates fell in April–May 2020; narrower bid-offer spreads and more two-way trading; interbank activity declined significantly and price discovery impaired; FX swap market remained relatively active. - Exit and communication: bond purchase window extended without specifying withdrawal criteria; ex post communications limited; operational readiness for bond purchases was initially lacking. - Poland (NBP): - Program size: total asset purchase program PLN 143 billion, or 5.4 percent of GDP. - Observations: purchases aided price discovery and supported secondary market; SOMO program remained active; lack of transparency on purchase pricing and on-and-off nature of purchases complicated exit and converted purchases into policy signaling tools. - Recommendations: secondary market purchases recommended during stress; primary market purchases only in rare circumstances with safeguards and high transparency. - South Africa (SARB): - Money market effects: banks’ holdings of government bonds rose from 17 percent (end-2019) to 23 percent (end-2020). - Bid-ask spreads: peak 10 basis points in March 2020; tightened to around 4–6 basis points by July 2020. - Yields: declined markedly in 2020Q2; yield curve remained steeper than pre-COVID. - Exit: purchases phased out as market functioning normalized in 2020Q3; limited transparency complicated exit and expectation management. ### Model estimates and selected regression results (Annex highlights) - Annex VIII (government bond markets) — selected 20-day ahead coefficients: - Intervention announcement: Bid/ask spread = -0.004; Asset swap spread = -15.69***; Amihud ratio = 0.71; Market turnover = 0.19. - VIX: 0.05*** (Bid/ask spread); -0.04 (Asset swap spread); 0.0003 (Amihud ratio); -0.01 (Market turnover). - MOVE: 0.01 (Bid/ask spread); -0.02 (Asset swap spread); 0.01*** (Amihud ratio); -0.003 (Market turnover). - Federal Reserve intervention announcement: -0.09 (Bid/ask spread); 6.11* (Asset swap spread); 0.03 (Amihud ratio); -0.18 (Market turnover). - USCRD: 0.02*** (Bid/ask spread); 0.11*** (Asset swap spread); 0.003 (Amihud ratio); -0.0004 (Market turnover). - Observations: 2,627 (Bid/ask spread and Asset swap spread); 1,751 (Amihud ratio and Market turnover). - R2: 0.21; 0.04; 0.33; 0.31 respectively. Significance notation: * p<0.1; ** p<0.05; *** p<0.01. - Annex IX (FX funding markets) — selected coefficients: - Federal Reserve swapline announcement: Bid/ask spread = 1.59**; Bid/ask spread (common) = 0.78***; CIP deviation = 1.48***; CIP deviation (common) = 0.84***; CIP deviation (unique) = 0.64*. - Local intervention announcement: Bid/ask spread = -0.2; Bid/ask spread (common) = -0.87***; CIP deviation = 0.26; CIP deviation (common) = -0.1. - Observations: 1,697 for all reported dependent variables. - R2 values: 0.21 (Bid/ask spread); 0.04 (Bid/ask spread common); 0.23 (Bid/ask spread unique); 0.03 (CIP deviation); 0.14 (CIP deviation common); 0.03 (CIP deviation unique). - Lagged dependent variables often strongly negative (e.g., Lagged dependent (1) = -0.50*** for Bid/ask spread). Significance notation as above. ### Practical implications and operational lessons - Interventions most effective when targeted at core markets that are normally liquid (money, government securities, FX funding). - Design for self-liquidation and exit: pricing, short-duration instruments, pre-announced quantity/duration targets where appropriate. - Transparency: ex-ante and ex-post transparency improves effectiveness and facilitates exit; excessive secrecy complicates market assessment and exit management. - Balance sheet and fiscal risks: monitor interest rate/sterilization risks, reassess capital buffers, consider burden-sharing or indemnities, and plan for cleanup and normalization. - Operational readiness: strengthen legal, operational, and counterparty/collateral frameworks in advance; build capacity for bond purchases and other less-familiar tools. - NBFI vulnerabilities: address proactively via regulation and ex-ante measures to reduce reliance on ad-hoc central bank interventions. *Source: IMF Working Papers — EMDE Central Bank Interventions during COVID-19 to Support Market Functioning (source content as provided).* ### Executive Summary ...................................................................................................... ### Executive Summary ### Overview of the study - The COVID-19 shock undermined the functioning of financial markets and resulted in unprecedented central bank interventions in advanced economies (AEs) and emerging market and developing economies (EMDEs) alike. - This paper looks at EMDE central bank interventions during the pandemic in core markets, namely money, securities, and FX funding markets. - Focus is on interventions aimed at combating market dysfunction as opposed to those providing monetary accommodation (i.e., quantitative easing). - The analysis combines empirical analysis with case studies to derive conclusions about effectiveness and to draw lessons about the design of future programs. ### Timing and scale of impacts - The impacts from EMDE central bank interventions were most strongly seen in the first two months after the announcement of a program. - These impacts diminished over time as other factors adversely impacted market liquidity, including the relatively slow return of foreign investors. - EMDE central banks did not always transact in large amounts. While in some jurisdictions the interventions were significant (Chile, Indonesia), in others, they were not (Philippines, South Africa). - Larger advanced economy markets with deep and liquid markets were more resilient to liquidity expansion (Europe, the U.K., and the U.S.) whereas in EMDEs there were cases where interbank market activity dried up and interbank interest rates fell relative to central bank policy rates. ### Effects on market segments - Securities markets: - Interventions were successful in stabilizing EMDE securities markets, especially when measured in terms of the impact on prices, which significantly improved from the depressed levels reached in March 2020. - Evidence of improved liquidity across a range of market functioning indicators, including volumes traded and the width of bid-offer spreads. - Much of the impact from central bank actions came from “announcement effects.” - Positive spillovers from the actions of major central banks calmed major markets and provided strong tailwinds assisting EMDE central banks’ own interventions; in some cases market conditions improved without EMDE central banks intervening themselves. - Money markets: - The scale of liquidity injections associated with interventions sometimes had adverse implications for interbank markets—although in general these markets held up well. - EMDEs with more developed markets tended to fare better during the crisis. - In general, EMDE central banks were successful in mitigating the negative repercussions on market functioning caused by significant liquidity expansion. - FX funding markets: - Initially, widening bid-offer spreads and covered interest parity (CIP) deviations were observed. - Post-intervention, spreads contracted and CIP deviations reversed, indicating easing U.S. dollar funding pressures. - The Federal Reserve’s swap line announcements played a major role in normalizing market functioning, contributing to the reduction of bid-offer spreads and promptly reversing CIP deviations. - EMDE central bank interventions moderately improved FX funding market liquidity by reducing CIP deviations, but did not significantly impact bid-offer spreads. - Overall, interventions—often even in smaller scales—proved effective due to announcement effects. ### Non-bank financial institutions (NBFIs) and financial stability - The COVID-19 shock highlighted financial stability risks arising from vulnerabilities in non-bank financial institutions (NBFIs). - NBFIs faced increased precautionary cash demands contributing to investor runs and were compounded by loss of liquidity in securities markets. - EMDE central banks had to adjust intervention approaches to assist NBFIs, sometimes dealing directly with entities that are not central bank counterparts during normal times (Chile, India). ### Exiting intervention programs - Exiting intervention programs proved more challenging for EMDE central banks than for counterparts in advanced economies. - In advanced markets, underlying liquidity improved sufficiently quickly that support became less relevant and could be withdrawn relatively easily. - In EMDEs, adverse impacts lingered due to expanded government financing needs and the absence of a resumption in capital inflows. - Some EMDE central banks needed to retain or even expand asset purchase programs (India, Indonesia), if only as a backstop against a shortfall in demand (Philippines, South Africa). - EMDEs found it easier to withdraw programs that were more standard and better understood by markets, for example, short-term repo operations and FX swaps. ### Overall assessment - Empirical analysis and case study reviews conclude that interventions by EMDE central banks were in line with their financial stability mandates, effectively addressing dysfunctions in core markets during the COVID-19 pandemic. ### Policy conclusions and recommendations - Intervention objectives should be well specified, particularly as regards addressing market dysfunction. For interventions with additional aims, such as monetary accommodation or government financing, clear communication should articulate how the program intends to address each objective. Exit strategies should be communicated ex-ante and tied to the achievement of the stated objectives. - Objectives should also be realistic, acknowledging that the impact of intervention may be small or short lived and that other fundamental factors also drive market activity. - Intervention triggers should be focused on metrics of liquidity and not prices. In challenging and fast-moving market conditions, operational flexibility, encompassing both qualitative and quantitative assessments, is essential for considering intervention triggers. - Interventions should be large enough to address the identified market dysfunction while taking into account the potentially significant risks of fiscal dominance, moral hazard, and financial risks to central bank balance sheets. - Where possible, program design should facilitate self-liquidation, such as through pricing (i.e., setting appropriate spreads) or involving short-term operations that roll off the balance sheet relatively quickly. Since interventions target core markets, the securities purchased should ultimately be sufficiently liquid to ensure they can be readily sold when market conditions stabilize. - EMDE central banks should be well-prepared so that programs can be launched quickly in the event of a shock. The financial markets relevant to financial stability and monetary transmission (core markets) should be identified in advance with methodologies for determining dysfunction and modalities for intervention established. This may entail preparations for the expansion of collateral and counterparty frameworks. - In some EMDE cases, interventions have significantly increased risks to central bank balance sheets, raising issues of policy solvency and operational independence. Normalization should entail reducing the balance sheet to a size no larger than required to efficiently implement monetary policy. Changes to the composition of balance sheets should reduce long-duration and higher credit risk assets accumulated during crisis periods. *Source: Executive Summary, wpiea2024101-print-pdf* ### 2020. During previous stress periods, it was primarily advanced economy (AE) central banks that introduced ### wpiea2024101-print-pdf - 2020. During previous stress periods, it was primarily advanced economy (AE) central banks that introduced ### H3: Purpose and scope - Examines EMDE central bank interventions during the COVID-19 period to extract lessons for future liquidity crises. - Focus: programs aimed at preserving market functioning (funding liquidity in local and foreign currency and improving market liquidity through asset purchases), distinct from quantitative easing programs that provided monetary accommodation. - Key empirical questions: - Were EMDE central banks effective in combating illiquidity? - To what extent did AE central bank interventions spill over into EMDE market liquidity? - Were EMDE program designs aligned with ideal features of well-designed programs? - How easy was exit from crisis programs in EMDEs relative to AEs? - How should central banks prepare for future market dysfunction? ### H3: Framework for well-designed interventions (features) - Objectives should align with central bank financial stability mandates: support monetary transmission, maintain flow of credit, mitigate fire-sale dynamics. - Target the largest, most interconnected markets central to transmission and pricing: in EMDEs typically money, government securities, and FX funding markets. - Triggers should be well-specified, tied to market functioning (not merely limiting price movements), and set with a high bar. - Program design should diagnose the problem and tailor tools accordingly; pricing and access should incentivize market resumption and facilitate exit (self-liquidating features, spreads, short-duration instruments). - Financial risks may require government indemnity; reputational risks managed via accountability and transparency. ### H3: Motivations and observed objectives - Both AE and EMDE central banks acted in the first half of 2020 in response to sharply deteriorating conditions: - AE actions: cut policy rates, abundant short-term liquidity, scaled asset purchases, reintroduced GFC-era programs. - EMDE actions: cut policy rates, term repos, lower reserve requirements, widen interest rate corridor, broaden collateral eligibility, FX swaps/repos/derivatives, government securities purchases (primary and secondary). - Three overlapping objectives often present: - Providing monetary accommodation (policy rate cuts, asset purchases, long-term lending). - Supporting market functioning (broadening liquidity frameworks, new facilities, asset purchases). - Maintaining flow of credit (government purchases, targeted term funding). - Interventions often targeted multiple objectives given the concurrent large negative shock to growth and inflation. ### H3: Institutional and structural constraints in EMDEs - Factors increasing EMDE challenges: - Less developed financial markets. - More concentrated investor bases. - Greater reliance on foreign investors; higher sensitivity to global factors. - Less robust macroeconomic and institutional policy frameworks. - Greater role of exchange rate in anchoring inflation expectations. - Higher country risk premiums and less resilient banking sectors. ### H3: Empirical approach and coverage - Coverage: interventions in core markets across 74 countries, representing 90 percent of global GDP. - Markets analyzed: interbank money, government bond, and FX swaps/funding markets. - Liquidity metrics: price-based, volume-based, volatility-based indicators (including bid-offer spreads, asset swap spreads, CIP deviations, Amihud ratio, trading volumes). - Data: daily (converted to weekly moving averages) for money and government bond markets; FX funding analysis decomposes indicators into global common factor and idiosyncratic factors using principal components. - Econometric method: pooled cross-sectional and time-series panel data with local projections (Jordà 2005); controls include Fed swap line announcements, country fixed effects, U.S. MOVE index, and VIX. ### H3: Empirical findings — Money markets - Money markets exhibited temporary and limited turbulence: interbank rate volatility increased and trading volumes declined. - Estimated impulse responses for term repos and reserve requirement reductions: - Liquidity-providing interventions had negligible effects overall; most impulse responses not significantly different from zero. - Most pronounced effects, if any, appeared within 10–20 days post-announcement. - FX funding support announcements contributed positively to money market resiliency, significantly diminishing interest rate volatility relative to trading volumes. - No detectable spillovers from interventions in other economies to EMDE money markets in the panel estimates. - Interbank rates generally remained anchored to policy rates on average when repo interventions were announced; reserve requirement cuts reduced interbank rates in some cases. - Conclusion: operational frameworks were resilient and interventions appeared well-tailored to meet increased liquidity demand. ### H3: Empirical findings — Government securities markets - EMDE bond market liquidity deteriorated sharply after the shock: wider bid-ask and asset swap spreads; lower trading volumes; increased Amihud ratio. - Asset purchase program (APP) announcements associated with: - Declines in government bond spreads to swaps. - Tighter bid-offer spreads. - Amihud ratio broadly unchanged after announcements in some specifications; turnover/market turnover unchanged in aggregate. - Timing: effects typically evident within 1–2 weeks after announcements. - No evidence that EMDE interventions caused adverse spillovers within EMDE bond market liquidity; global factors strongly influenced EMDE liquidity (high correlation of price-based measures across markets). - Impact on trading volumes ambiguous: some temporary boosts shortly after announcements but longer-term volumes driven by other factors. ### H3: Empirical findings — FX funding markets - U.S. dollar scarcity increased in EMDEs: larger CIP deviations and wider FX swap bid-offer spreads. - Variance decomposition: AE FX funding dislocations dominated by common global factor; EMDEs driven more by country-specific idiosyncratic factors. - Federal Reserve swap line announcements: - Pivotal in normalizing market functioning. - Largest effect on reducing bid-offer spreads and reversing CIP deviations. - EMDE central bank U.S. dollar-providing operations: - Contributed moderately to improving CIP deviations with greatest impact 10–15 days post-announcement. - Not estimated to significantly narrow bid-offer spreads on their own (coefficients and impulse responses not significantly different from zero for bid-offer spreads). - Global risk factors (VIX) played a marginal/insignificant direct role in driving bid-offer spreads in FX funding markets during the period. ### H3: Program sizes and announcement effects - Available data on EMDE FX intervention program sizes: maximum program sizes range from 0.1 to 3.8 percent of GDP; median program size 2 percent of GDP. - Much of the program impact appears to derive from announcement effects, so large committed volumes were often unnecessary to achieve market functioning results. ### H3: Case study synthesis (Chile, India, Indonesia, Philippines, Poland, South Africa) - Objectives: - Supporting market functioning was a core objective across all cases. - Most programs also pursued monetary accommodation and supporting government financing (exceptions: SARB focused solely on market functioning). - Market focus: - All central banks supported interbank money markets (short-term repos; lengthened maturities). - Government bond purchases prominent (typically secondary market); Indonesia used both primary and secondary purchases. - FX funding support prominent in Chile, Indonesia, India (often not heavily used). - Some interventions targeted non-government securities where market structure warranted (Chile bank bonds; Poland agency bonds). - Triggers: - Triggers aligned with financial stability considerations: spikes in price volatility, reversals in non-resident flows, bond spread widening, increased demand for liquidity, signs of fire-sale dynamics in NBFIs. - Transparency: - High ex-ante transparency on objectives and modalities across cases. - Ex-post transparency mixed: Chile and India published results; South Africa conducted covert purchases; Indonesia and the Philippines did not publish immediate results (Philippines’ purchases inferable later from market data). - Some central banks set end/review dates; exit criteria often unclear or evolved. - Operational capacity and risks: - Some central banks lacked established infrastructure for bond purchases (Poland, Philippines), leading to operational frictions. - Balance sheets expanded materially in most cases, increasing interest rate and sterilization risks, and raising exposure to governments. - Table 3 (Change in Balance Sheet Size; Feb–Dec 2020) reports: - Chile: Change in Balance Sheet Size 88.6 %; Change in Net Local Currency Assets 13.9 % GDP; Change in Domestic Bond Holdings 1,242.7 %; 14.7 % GDP; 2,152.5 %; 14.7 % GDP. - India: 25.5 %; 5.9 % GDP; 35.7 %; 1.9 % GDP; 31.7 %; 1.6 % GDP. - Indonesia: 24.6 %; 4.0 % GDP; 75.7 %; 3.7 % GDP; 102.8 %; 3.4 % GDP. - Philippines: 38.5 %; 11.0 % GDP; 123.6 %; 6.9 % GDP; 512.3 %; 6.4 % GDP. - Poland*: 35.1 %; 8.0 % GDP; 119.5 %; 5.5 % GDP; N/A; 3.3 %. - South Africa: -4.7 %; -0.9 % GDP; 1.9 %; 0.1 % GDP; 404.5 %; 0.6 % GDP. - Average: 34.6 %; 7.0 % GDP; 266.5 %; 5.5 % GDP; N/A; 5.0 %. - Credit risk generally low where purchases were of government securities or lending was collateralized/short-term; Chile’s accumulation of bank bonds represented higher credit risk. - Some burden-sharing or repayment arrangements mitigated risks in cases (Indonesia, Philippines repayment of BSP advances in 2022). - Money market impacts: - Large injections of reserves sometimes reduced interbank trading activity, especially in markets dominated by few banks (Chile, Indonesia, Philippines). - Transmission largely held up; some central banks adjusted operating frameworks (Indonesia moving toward floor implementation). - Exits: - AE central banks generally scaled back programs relatively quickly and smoothly, aided by prior experience (2008 GFC) and pre-established facilities. - EMDE exits were more varied and often took longer; mixing multiple objectives complicated exits (market functioning vs. monetary accommodation vs. government financing). - Tools more familiar to markets (short-term OMOs, FX swaps) were easier to scale back in EMDEs; novel or long-duration asset purchase programs were harder to exit. ### H3: Implications and policy recommendations for future programs - EMDE central banks can effectively support market functioning, but interventions typically have modest effects relative to fundamental drivers (global investor risk attitudes). - Focus interventions on core markets that are normally liquid (money, government securities, FX funding). - Design interventions to be sufficient for immediate post-shock support but aim to exit quickly; empirical results show intervention effectiveness most obvious for 10–20 days after shock. - Distinguish clearly between market functioning interventions and those for monetary accommodation or government financing to simplify exit and communication. - Develop and communicate exit criteria and indicators at the outset, including reinvestment approaches for accumulated instruments. - Transparency on operations and progress is important for effectiveness and exit management; too little transparency complicates market assessment of central bank support. - Account for balance sheet and fiscal risks: - Be mindful of interest rate/sterilization risks from long-duration domestic asset accumulation. - Reassess capital buffers and be prepared for recapitalization or dividend delays if central bank capital is depleted. - Consider burden-sharing frameworks where appropriate. - Prepare operationally and legally in advance: - Ensure flexible legal frameworks and operational readiness to implement interventions rapidly and transparently. - Strengthen operational capacity for bond purchases and other less-familiar tools. - Address NBFI liquidity risks proactively via regulation and ex-ante measures to reduce likelihood of ad-hoc intervention reliance. - Ensure interventions’ modalities, pricing, access, duration, and communication are tailored to incentivize market resumption and reduce moral hazard and fiscal dominance risks. *Italic source attribution: IMF Working Papers — EMDE Central Bank Interventions during COVID-19 to Support Market Functioning (source content as provided).* ### Annex I. Case Study: Central Bank of Chile ### Annex I. Case Study: Central Bank of Chile ### Context - BCCh intervened in FX markets in only four exceptional circumstances since 1999: 2001, 2002, 2019, and 2022. - The BCCh’s COVID-19 crisis response followed earlier extensive support measures implemented during social unrest in November 2019. - Chile’s bank debt securities market was almost double the government debt securities market at the end of 2019. - Mutual funds and pension funds are the most important providers of funding to the domestic banking system (mutual funds provide about half of time deposits) and pension funds are the largest buyers of bank bonds. - Foreign investors play a lesser role, though their holdings—particularly in government securities—had been increasing in recent years. - The BCCh was prohibited from buying government debt or financing public spending until a legal change in August 2020. ### Why Was Intervention Necessary? - Two episodes of market disruption occurred in short succession: domestic social unrest in late 2019 and the global COVID-19 shock in March 2020. - Market stress during late 2019: - Intraday FX volatility spiked, sovereign bond rates rose, and the Local Stress Index rose sharply. - NBFI portfolio reallocations away from banks and towards FX led to peso and FX liquidity shortages and stress on banks. - The BCCh offered FX swaps at a 200-bps margin above normal market rates and calmed markets by early 2020 without transacting significant volumes of swaps. - The outstanding amount of swaps peaked at $1.1 billion in late November/early December 2019. - Market stress during March 2020 (COVID-19): - Exchange rate and FX funding pressures from capital outflows; onshore U.S. dollar funding market pressures, though less than in the social unrest episode. - Challenges in maintaining effective monetary policy transmission: policy rates were cut close to the effective lower bound; higher risk premia and concerns about collateral and credit risk reduced banks’ willingness to lend; bank funding markets came under pressure with wider credit spreads. - Portfolio shifts in the non-bank financial sector: pension and mutual funds withdrew from bank funding markets due to mutual fund redemptions and Congress permitting extraordinary pension withdrawals, severely impacting banks because NBFIs contribute about half of total bank liabilities. - NBFI investment maturities shortened and dollarization within domestic investors’ portfolios increased, significantly reducing banking sector liquidity. - Sovereign bond market pressures were less acute because the government could use its sovereign wealth fund to finance COVID-19 programs and did not need a sudden increase in government securities issuance; BCCh purchases of government bonds were constitutionally prohibited until August 2020. - Pension withdrawals: initially a July 2020 bill allowed up to 10 percent withdrawals; two more withdrawals in December 2020 and April 2021—each another 10 percent—resulting in the equivalent of about 20 percent of GDP withdrawn from the pension system. ### What Was Done? - Initial measures: - The BCCh cut the policy rate and extended prior Repo and FX Swap programs, initially until January 9, 2021, with additional maturities and an increase in daily volumes. - Repo operations were also offered at 7- and 180-day maturities, in addition to the existing 30- and 90-day maturities. - FX swaps were offered at 90- and 180-day maturities, in addition to the existing 30-day maturities. - The window for possible FX interventions was extended to January 2021 (the BCCh ultimately did not intervene in the FX spot market and merely rolled over expiring forward contracts until June 2020). - Corporate bonds were included as collateral in BCCh liquidity facilities. - Bank-bond purchase program: - Announced on March 20, 2020: BCCh would buy bank bonds with up to five-year maturities at a premium to the local overnight indexed swap yield curve, based on the issuer’s credit rating. - Purchases were limited to 20 percent of issuance per bank (later extended to 30 percent) and conducted via auctions. - Additional measures to strengthen external financing and liquidity: - A second “Special Asset Purchase Program,” encompassing the purchase of bank bonds and BCCh securities, totaling $8 billion. - A funding-for-lending scheme was introduced; the overall funding-for-lending program reached about $40 billion in total. - A new two-year IMF Flexible Credit Line was approved in May 2020. - Access to the New York Federal Reserve’s FIMA Repo Facility was arranged in June 2020. - In July 2020, Chile expanded its existing currency swap line with China. - From January to October 2021, the BCCh accumulated additional international reserves. ### Were Interventions Effective? - Market functioning and liquidity: - The BCCh succeeded in restoring the functioning of key markets; the Local Stress Index and other indicators show conditions normalized relatively quickly—especially in the FX funding market. - Liquidity conditions stabilized and the functioning of the bank bond market was maintained, with positive spillovers into other markets. - Total bank debt outstanding fell as banks substituted private for central bank funding. - Turnover in bank and government bonds rose as shocks hit, then subsequently returned to normal levels. - Market participants noted that bank bond purchase programs helped corporate bond issuance and supported the government bond market. - Unintended effects and transmission: - The boost in excess reserves negatively impacted the interbank market: interbank trading volumes plummeted and the interbank rate dropped to the floor of the BCCh’s interest rate corridor. - Some interbank activity returned in the second half of 2020 but volumes remained well below pre-crisis levels. - Overall, monetary transmission remained adequate as the interbank rate realigned with the policy rate following the policy rate increase and as activity picked up from the second half of 2020. ### Exit Considerations - FX liquidity program design aided exit: - BCCh included end-dates for FX liquidity programs with each announcement, creating clear market expectations and a timeline for reviewing ongoing need. - The stock of non-deliverable forward contracts gradually fell, reaching zero by end-October 2020. - The volume of FX swaps was reduced to zero by end-June 2020 despite FX swaps being available until January 2021. - Pricing of FX funding support established facilities as backstops that naturally liquidated as market conditions normalized. - Exit from bank bond purchases was more prolonged and complex: - Bank bond purchases transitioned from crisis interventions to quantitative easing and served to facilitate ongoing pension fund withdrawals approved by the government, complicating the exit strategy. - BCCh gradually discontinued bank bond purchases as it phased out QE, but maintained a fixed stock of bank bonds supported by a new “Bank Bond Reinvestment Program” in January 2021. - When the policy rate was raised in mid-2021, the BCCh stopped the reinvestment of coupons and redemptions from bank bonds. - Structural and persistent effects: - Reduced activity in the interbank and money markets may persist due to increased structural liquidity from the Conditional Financing Facility for Increased Loans and the shrinking of the pension sector following withdrawals, which is likely to undermine money market activity for an extended period. *Source: Annex I. Case Study: Central Bank of Chile (wpiea2024101-print-pdf).* ### 3.75 percent, a historic low. ### 3.75 percent, a historic low. ### Policy measures implemented by Bank Indonesia (BI) - Reduced reserve requirements: - FX reserve requirements were cut from 8 to 4 percent. - Rupiah requirements were cut by 2.5 percent. - Increased liquidity provision: - The maturity of bond repos was extended up to 12 months. - More frequent FX swap auctions: - 1-, 3-, 6-, and 12-month auctions shifted from weekly to daily frequency. - A strategy of “triple intervention,” including: - FX intervention: Mainly through domestic non-deliverable forwards to manage exchange rate volatility and help protect international reserves. - Spot market FX interventions: Reflected in a $9 billion fall in net international reserves in March 2020. - Government bond purchases in the primary and secondary markets: - Initially bought around $10.8 billion worth of government bonds from foreign investors in the secondary market. - By late April 2020, secondary market purchases were replaced by primary market purchases using: - the “market mechanism,” where BI joined auctions as a non-competitive bidder, and - a “burden sharing agreement” with the Ministry of Finance, where BI participated in private placements at below-market interest rates. - Communication and risk management: - BI prioritized communications on overall intervention objectives (to preserve market stability) rather than operational modalities; ex-ante communication on size and exact timing of interventions was limited. - Four risk management principles for government bond purchases: (i) give priority to the market mechanism; (ii) consider the impact on inflation; (iii) purchase tradable and marketable bonds; and (iv) act as the buyer of last resort. - Quantities and financing: - Total allocation for the COVID-19 response under the National Economic Recovery Program amounted to 4.4 percent of GDP in 2020. - Overall, primary market purchases amounted to IDR 473.4 trillion in 2020, of which IDR 76 trillion were carried out under the “market mechanism.” - BI financed about half of the remaining deficit in 2020. - Reserve replenishment and external facilities: - BI replenished international reserves relatively quickly and supplemented reserves with several bilateral swap arrangements and repo lines, including accessing $60 billion through the Federal Reserve FIMA repo line in April 2020. ### Effectiveness of interventions and market outcomes - FX market: - Exchange rate volatility fell in April and May 2020; bid-ask spreads in the FX spot market had normalized by June 2020. - Decomposition of FX market pressures suggests the bulk of the effect was country-specific, indicating BI’s policies were effective in restoring stability. - Government securities market: - Markets started to stabilize in April 2020, though pressures persisted through mid-2020. - Bond market interventions mitigated the impact of increased bond supply and kept government bond rates from rising. - The share of non-resident investors in local currency government bonds fell, despite an increase in absolute holdings. - BI purchases contributed to a fall in local currency bond yields and bid-ask spreads, and to higher trading volumes in the secondary market. - Money market and liquidity: - Liquidity support pushed the interbank rate to the floor of the corridor and reduced interbank market activity. - Excess reserves increased substantially; BI sterilized part of the excess liquidity through increased reverse repo operations. - Transaction volume in the interbank market halved compared to 2019, though it fared relatively well compared to some other emerging markets. ### Exit strategy, persistence, and risks - Timeline and extensions: - Bond purchases were extended into 2022; the maximum quantity of bond purchases was increased as government funding pressures persisted. - Purchases under the “market mechanism” were extended in December 2020 to help finance the 2021 budget. - The burden-sharing agreement, initially set to expire at the end of 2020, was extended into 2022. - Motivations and trade-offs: - Extensions reflected authorities’ view that fiscal support to sectors most affected by COVID-19, funded through monetary financing, was more effective than monetary accommodation via lower policy rates. - The BI emphasized the temporary and extraordinary nature of the measures and intentions to reduce the budget deficit to a maximum of 3 percent of GDP in 2023. - Fiscal dominance and balance sheet risks: - The scale and extended timeframe of bond purchases raised concerns about fiscal dominance and risks to the BI balance sheet. - Total public debt remains manageable at about 40 percent of GDP. - The significant expansion of the BI’s securities holdings increased its financial risks and heightened the risk of fiscal dominance. - Policy rate decisions: - Authorities were concerned that further rate reductions would lead to a surge in capital outflows; however, BI ultimately cut the rate by an additional 0.25 bp in February 2021 to 3.5 percent. ### Key lessons and implications - EMDE vulnerabilities and toolkit expansion: - The Indonesian case highlights challenges to market functioning for an EMDE heavily reliant on non-resident financing. - BI successfully expanded its toolkit to manage FX volatility and used new tools in the government bond market, including primary and secondary market purchases. - Effectiveness and limits: - Interventions were effective in improving bond market liquidity, though liquidity conditions remained strained for some time as government financing needs increased. - BI’s interventions adversely impacted the money market by significantly increasing excess liquidity. - Required follow-up actions: - BI will need to realign the interbank rate with its policy rate and sterilize excess reserves to incentivize interbank market trading activity. - The larger and longer-duration government bond portfolio will require careful management of associated balance sheet risks and potential fiscal dominance risks. *IMF WORKING PAPERS EMDE Central Bank Interventions during COVID-19 to Support Market Functioning* ### References to the purchase program were removed from the NBP’s monetary policy statement, ### wpiea2024101-print-pdf - References to the purchase program were removed from the NBP’s monetary policy statement, ### Poland: asset purchase program and market interventions - References to the purchase program were removed from the NBP’s monetary policy statement, but the SOMO program remained active. - The total size of the asset purchase program amounted to PLN 143 billion, or 5.4 percent of GDP. - Operational and market observations: - Asset purchases aided price discovery and served as a reference for the secondary market during market stress. - Asset purchases are one factor in determining bond premia; their effectiveness in risk-premia compression depends on expectations and persistent market imbalances. - While large-scale purchases seemed to have influenced Poland’s bond risk premia over the medium run, their relative contribution vis-à-vis other factors, such as private credit dynamics, needs separate assessment. - Liquidity bottlenecks may occur in the NBFI sector; central banks may need to expand eligible counterparties to ensure effective liquidity provision and mitigate contagion and spillover risks. - Transparency enhancements—such as disclosing the range of prices for purchased bonds—would have further enhanced transparency and addressed concerns about market segmentation (notably regarding SOMO auctions in April 2020). - Secondary market purchases are recommended even during times of stress; primary market purchases might be considered to address short-term financing needs in rare circumstances and subject to safeguards and high transparency where legal constraints require workarounds. - Multiple objectives and lack of clear anchors can complicate exit. The on-and-off nature of bond purchases in 2021 converted SOMO purchases volume into a policy signaling tool and increased uncertainty. - More explicit forward guidance, announcing a predefined purchase limit and the anticipated duration of the program, could serve as a yardstick for both the NBP and market participants to aid exit preparation—even if targets are later revised. - Operational note: - The NBP did not accept floating rate notes; these had to be first switched to short-term bonds before being offered to NBP. The Ministry of Finance carried out a switching auction to provide fixed-rate paper for floating-rate paper. - The reserve ratio increase was effective from November 30. ### Poland: empirical indicators and market functioning (summary of figures) - Market reaction and indicators: - Poland’s bond curve reacted positively to NBP interventions, outperforming the Czech Republic’s during the same period. - Bid-offer spreads started to tighten following the first structural open market operations (SOMO) auction. - The impact of subsequent SOMO auctions on secondary market functioning is less clear, as NBP purchases were offset by a large bond placement by the Ministry of Finance (MoF). - NBP purchases in the secondary market absorbed a large portion of COVID-19 debt issuance and, coupled with the change in the banking liquidity structure, supported government bond valuations. - Metrics shown in figures (labels preserved as in source): - Bond market premia during COVID-19 shock (basis points). - Bond market liquidity during COVID-19 shock (basis points and billions of zloty). - NBP Purchases and market indicators (Percent and billions of zloty). - NBP purchase and MoF issuance (in billions of zloty). - NBP purchases and MoF issuances (billions of Polish zloty and percent). - 10-year government bond swap spread (basis points). ### India: Reserve Bank of India (RBI) case study — context and interventions - Context and vulnerabilities: - India’s bond market is well-developed; government securities are most common and serve as a credible benchmark. - The rupee is fully convertible for current account transactions but has tight capital account restrictions. - Money markets are relatively developed; the RBI operates an interest rate corridor system and uses repo operations, standing facilities, reserve requirements, and structural liquidity requirements. - Pre-COVID-19 vulnerabilities included stresses in the shadow banking sector that began in 2018; the RBI had been easing policy while the banking system experienced a structural liquidity surplus. - Between February 2019 and March 2020, the policy rate was reduced by 135 bps, and special OMOs were introduced. - In February 2020, liquidity was provided through new Long-Term Refinancing Operations (LTROs). - Why intervention was necessary: - Pre-existing financial-sector vulnerabilities and fiscal year-end seasonal liquidity deterioration amplified the COVID-19 shock. - Mutual fund industry redemptions led to forced sales of liquid assets and breakdowns in primary markets, increasing funding pressures for non-bank finance companies and contagion fears. - Market stress increased rapidly around mid-March, with government bond market liquidity dropping and long-end risk premia rising. - What was done (key measures): - FX-swap facility to provide dollars against the Indian rupee. - Reduction of the policy rate by 135 bps and widening of the policy corridor by 40 bps. - Scaling up of OMO and Special OMO purchases of government bonds. - Liquidity injections through existing term repos and a 100-bps cut to the cash reserve ratio. - Scaling up of existing LTROs, and new targeted LTROs providing funding for up to three years. - Special Lending Facility for Mutual Funds. - Transparency: - The RBI’s operations were very transparent ex ante and ex post: purchase limits and targeted securities for OMO and Special OMO auctions were announced in advance; detailed results (demand, allocation, average and cut-off yields) were published afterward. - The RBI published full results for LTRO and TLTRO auctions and in April 2021 started to disclose the OMO purchase envelope for the upcoming quarter at market participants’ request. ### India: effectiveness, exit, and lessons - Was intervention effective? - The RBI’s $2 billion FX swap auction was overbid by two times and contributed to a swift recovery of the CIP basis. - Term Repos and LTROs saw significant demand; the first TLTRO was more than twice oversubscribed. - Operations and reserve requirement cuts eased liquidity conditions and promoted a sustained reduction in money market risk premia from March 27. - Bond purchases at the end of March provided liquidity, aided price discovery, and helped normalize bid-offer spreads for more liquid securities. - A more decisive turnaround occurred after the special OMO operations on April 23; subsequent operations consistently impacted the yield curve and anchored risk premia amid elevated bond supply. - Targeted liquidity measures supported NBFIs and mutual funds; demand for some instruments fell short of capacity due to ample liquidity, but measures boosted market confidence and helped stabilize commercial paper and corporate bond spreads. - Primary markets recovered. - Money market and balance-sheet outcomes: - Interbank money market activity fell: call money market volumes fell by about 50 percent compared to pre-COVID-19 levels, while activity increased in other money market segments. - Credit growth stabilized in most sectors but had not yet recovered fully. - The RBI’s holdings of government bonds increased by more than 60 percent since the start of the pandemic and have now reached about 7.5 percent of GDP. - Exit strategy and challenges: - The RBI designed LTROs and TLTROs with pre-announced program targets and allowed counterparties to prepay early; almost all LTROs and a third of TLTROs were repaid early in September–October 2020. - The RBI struggled to fully exit from providing liquidity support: the bond purchase program was scaled up after the initial shock and new floating rate on-demand TLTRO facilities continued until December 2021. - Frequency of special OMOs increased from Q3 2020 to mitigate effects of increased government bond supply on long-term yields. - The Government Securities Acquisition Program was suspended in October 2021, but OMOs and special OMOs were retained and exit conditions remained uncertain even as market conditions normalized; the RBI was expected to tighten monetary policy. - Key lessons and implications: - Targeted liquidity provision can effectively address market dysfunction, especially in NBFI or corporate sectors. Quantity and duration-based targets plus prepayment options can facilitate exit. - Effectiveness for additional objectives (e.g., credit expansion) can be contingent on factors like credit guarantees and needs separate assessment. - Asset purchase programs with separate targets can boost effectiveness while assisting exit. Announcements on purchase quantities are beneficial for large-scale programs with objectives like risk-premia compression, moving the program from reactive to proactive. - Announcing quantity and duration targets served as a checkpoint for reassessment and facilitated exit guidance. Even where market expectations favored additional tranches, the bond market reaction to the exit announcement was contained. *Source: IMF Working Papers — EMDE Central Bank Interventions during COVID-19 to Support Market Functioning (content as provided in the source PDF).* ### Annex V. Case Study: Bangko Sentral ng Pilipinas ### Annex V. Case Study: Bangko Sentral ng Pilipinas ### Context - Philippine local currency financial markets are at a relatively early stage of development, with domestic banks holding a dominant position. Foreign banks are also active, particularly those from regional Asian market hubs such as Singapore. - The bond markets remain small, primarily comprised of government securities, with few non-bank institutional and foreign investors. - Money markets are undeveloped; recent initiatives include the government security repo market and issuance of Bangko Sentral ng Pilipinas (BSP) bills. - Recent reform of the operational framework, notably the introduction of an interest rate corridor, is expected to bolster money market activity. - Treasury bills are the most active segment. - Financial market infrastructures and regulatory frameworks are being reformed to facilitate more active markets. - FX spot and swap markets are relatively more developed, reflecting strong remittance and export flows; the implied peso interest rate derived from FX swaps is the most widely used money market benchmark. - FX controls, including on borrowing in foreign currency, have limited FX mismatches in banks, NBFIs, and corporates. - The BSP’s operational focus has been on sterilizing the large structural liquidity surplus arising from significant holdings of FX reserves. - Reserve requirements remain high at 12 percent by global standards. - The BSP has been transitioning to indirect instruments to sterilize liquidity through peso deposit auctions (since 2017), and more recently BSP bills. - An overnight lending facility is available to banks, though liquidity provision is seldom needed. ### Why Was Intervention Necessary? - Market conditions quickly deteriorated in March 2020, with indicators showing escalating financial stress due to a surge in the demand for precautionary liquidity. Indicators included: - A marked rise in short- and long-term interest rates. - An increase in bid-offer spreads in the government bond market (spreads for standard parcel sizes increased from 10 to 50 basis points). - Increased use of the BSP’s standing overnight liquidity facility. - Underbidding in the BSP’s deposit auctions in March 2020. - Settlement failures in government securities’ auctions. - Growing expectation that the government would need to scale up its financing to support the COVID-19 response made the government bond market one-sided. - The exchange rate depreciated and the equity market weakened as expectations of economic weakness reduced confidence. - To forestall fire-sales and halt a downward-spiral of confidence, the authorities decided to close the financial markets on March 17. ### What Was Done? - The BSP’s operational response focused on supporting the government by providing liquidity through reduced sterilization and government bond purchases in the secondary market. - Liquidity provision measures included: - Reducing both sterilization operations and reserve requirements to equip banks with more excess reserves. - Peso deposit and reverse repurchase auctions (regular operations). - Government bond purchases via a newly operationalized bond purchase window. - An advance dividend to the government. - A six-month repo line. - A 540 billion peso advance in October 2020; the latter two measures were subsequently extended. - Timeline highlights: - March 16-17, 2020: OMO sterilization cancelled; government securities and FX market closed on March 17. - March 18, 2020: Policy rate cut to 3.25%; rediscounting spread reduced to zero. - March 22-25, 2020: BSP lends PHP 300 billion via repo to government; upcoming OMO sterilization cancelled; reserve requirements reduced by 200 bps for large banks; BSP announces advance dividend to the government. - March 24, 2020: BSP announced opening of a government bond purchase window. - April 9, 2020: Range of government securities eligible for purchase widened; overnight reverse repo sterilization operations reduced. - June 2020: Monetary Board begins discussions to scale back liquidity injections; BSP cuts policy rate a further 50 bps. - July 2020: Monetary Board notes BSP is ready to begin scaling back liquidity provision; BSP cuts reserve requirements for smaller thrift and rural banks by 100 bps. - August–September 2020: Monetary Board holds policy rates and announces the beginning of BSP bill sterilization operations. - The BSP was not required to provide FX liquidity support; FX market liquidity remained robust after an initial depreciation, aided by uninterrupted remittance inflows and curtailed importers’ demand for FX due to lockdowns. - Major central banks’ prompt liquidity provision in their jurisdictions had positive spillovers to the Philippines. ### Was Intervention Effective? - The BSP’s actions quickly calmed markets: - Interest rates fell significantly over April and May, and the yield curve flattened. - Evidence of improved market functioning included narrower bid-offer spreads, more two-way bond trading, stronger support for government bond auctions, stronger demand for the BSP’s sterilization operations, and a decrease in requests at the BSP’s bond purchase window. - An adverse side effect: a significant decline in interbank money market activity due to the injected liquidity. Once precautionary liquidity demand subsided, few participants were short of liquidity, reducing trading opportunities and compromising price discovery. - Difficulty calculating an overnight reference rate occurred because many days had no significant trading volumes. - The Philippine Interbank Reference Rate indicated the FX swap market was larger than the interbank market and remained more actively traded during the post-intervention period. - While the BSP restored market functioning and provided liquidity to banks and the government, there were fewer indications of robust monetary accommodation as credit demand remained weak. A definitive conclusion on monetary accommodation is challenging due to the absence of a counterfactual during the severe macro shock. ### Communication and Exit Strategy - The BSP employed a multi-pronged communication approach: - Early in the stress period, the BSP extensively communicated objectives and modalities directly to market participants, supplemented by press releases from the Governor. - Detailed intervention methods were discussed directly with market participants rather than through public channels. - Ex post communication was less comprehensive: - The BSP did not publicize the results of its daily bond purchase window; participants could infer results from comprehensive post-trade data on dealing platforms. - For other operations (term deposits, reverse repurchases, bill operations), announcements were made regularly in line with pre-COVID-19 practices. - Scaling back support: - By June 2020 the Monetary Board concluded markets had stabilized sufficiently and began discussing scaling back liquidity support; decisions to increase sterilization and re-introduce BSP bills were publicly announced. - The BSP extended the bond purchase window beyond the initially announced six-month period without specifying criteria for withdrawal, communicating reductions in purchases informally and directly with market participants starting from August 2020. - Ongoing uncertainty led participants to continue regularly questioning the BSP on the facility’s availability. ### Key Lessons and Implications - Operational readiness: - The BSP lacked operational readiness to implement bond purchases rapidly. Preexisting instruments (term deposit auctions) were quickly scaled up, but bond purchases required mid-course operational adjustments. - Some operational modalities were chosen to facilitate faster implementation at the expense of transparency and possibly effectiveness (e.g., bond purchase window rather than an auction). - The BSP is exploring alternative modalities for future crises. - Intervention objectives and exit complexity: - Initial interventions targeted multiple objectives (backstopping markets, providing liquidity, monetary accommodation) simultaneously, but lack of clarity about which instruments addressed each objective complicated the scaling back of operations when markets normalized. - The bond purchase window, perceived as backstopping government debt auctions and perhaps providing monetary accommodation, became difficult to discontinue despite normalized bond market liquidity conditions. - Market structure constraints: - With a bank-focused financial system, money markets are likely to remain impaired until more liquidity is withdrawn; these markets are not well-developed and are less robust to structural liquidity changes from crisis interventions. - The peso implied FX swap market was more resilient, consistent with a more developed market and a more diverse set of participants. - Although the BSP scaled up sterilization since mid-2020, additional measures are likely needed to bring excess reserves closer to banks’ precautionary demand (as witnessed in the pre-COVID-19 period) for money market recovery. - Communication: - More comprehensive ex post communications, particularly for bond purchases, could have boosted effectiveness and facilitated exit. - Lack of transparency on bond purchase volumes and pricing likely reduced BSP’s effectiveness in supporting price discovery and made scaling back more challenging. - Ex post communications within an auction structure, and clearer communication of intervention strategy, objectives, and outcomes, could have reinforced overall effectiveness. *Source: Annex V. Case Study: Bangko Sentral ng Pilipinas (IMF working paper).* ### 2020. Term funding supported banks’ capacity to purchase government bonds, with banks’ holdings increasing ### 2020. Term funding supported banks’ capacity to purchase government bonds, with banks’ holdings increasing ### Market functioning and bond purchase program - Banks’ holdings of government bonds increased from 17 percent of the total outstanding (end-2019) to 23 percent (end-2020). - The SARB government bond purchase program improved market functioning despite its limited size. - Bid-ask spreads of benchmark bonds: - Peak of 10 basis points in March 2020. - Tightened gradually to around 4–6 basis points by July 2020. - These spreads are significantly wider than pre-COVID levels but consistent with other EMDE government bond markets. - Yields and yield curve: - Yields declined markedly in 2020Q2. - The yield curve remained steeper than before, suggesting the SARB did not target a yield level objective by continuing purchases. - Program objective and market reaction: - SARB emphasized market functioning as the primary objective rather than quantitative easing. - As purchases were phased out in line with normalizing market functioning in 2020Q3, SAGB yields remained relatively stable, indicating absence of significant speculative positioning for further bond purchases. - Purchase activity since July 2020: - No significant purchases have been conducted since July 2020, although the market support program was never officially withdrawn and SARB continued to stand ready to intervene. - Transparency was minimal during interventions; monthly SAGB holdings were published with some lag. ### Exit strategy and communication - Liquidity operations were gradually unwound as money market conditions normalized. - Standing facility rate adjustments (mid-August 2020): - Credit standing facility rate adjusted to the repo rate plus 100 basis points. - Deposit standing facility adjusted to the repo rate less 100 basis points. - Facility discontinuations and normalization: - The three-month term repo facility was discontinued in December 2020. - Intraday overnight supplementary repurchase operations were discontinued in February 2021. - The main Wednesday repurchase auction amount reverted to R56 billion in February 2021. - Communication and expectation management: - Minimal transparency created challenges in managing expectations and increased risk of misinterpretation of unexpected changes. - To mitigate risk, in the event of a bond redemption early in 2021, the SARB chose to roll over its SAGB holdings, avoiding a drop in its bond holdings; this action was inconsistent with stated program objectives but reinforced commitment to remain active in the market. ### Key lessons and implications - Unique aspects of the SARB experience: - ZAR exchange rate depreciated sharply, but no persistent FX funding pressures were evident in the FX swap market. - SARB could focus market support on domestic money and bond markets. - A concurrent sovereign credit rating downgrade amplified the shock and prompted unprecedented SAGB market intervention. - Objective and moral hazard: - The SARB’s sole objective was to restore market functioning while minimizing moral hazard. - Limited transparency constrained the market’s ability to understand central bank actions and complicated the exit strategy. - Consequences of limited transparency: - Uncertainty about duration and conditions of purchases made it nearly impossible to credibly commit to stopping purchases. - In the absence of measurable expectations, the SARB was forced to keep the program in place permanently to avert potential market turbulence from an unexpected termination. ### Annex VII–IX: Model estimates and empirical findings (selected results) - Annex VII (Money Markets): - Presents estimated models for announcements of repo operations and adjustments (reductions or easing) in reserve requirements on: - Normalized spread of the interbank rate relative to the policy rate. - Traded interbank market volumes. - Amihud ratio (weekly volatility of absolute daily changes in the interbank rate relative to daily average traded volumes). - Notes and definitions: - Daily traded volumes normalized by dividing daily volumes by average daily volumes over the previous year. - Difference between interbank and policy rate measured as absolute value and normalized by the average level over the preceding year. - Standard deviation of the difference between the interbank rate and the policy rate measured over a 10-day window. - Amihud ratio = standard deviation of the difference between the interbank rate and the policy rate over a 10-day window, divided by the 10-day moving average of daily traded volumes. - “MPR” denotes the level of the monetary policy rate. ActionFL and Action PVVPP indicate announcements of funding for lending and private sector asset purchases. - Annex VIII (Government Bond Markets) — selected panel regression estimates for 20-days ahead: - Dependent variables reported: Bid/ask spread, Asset swap spread, Amihud ratio, Market turnover. - Coefficients (with significance): - Intervention announcement: Bid/ask spread = -0.004; Asset swap spread = -15.69***; Amihud ratio = 0.71; Market turnover = 0.19. - VIX: 0.05*** (Bid/ask spread); -0.04 (Asset swap spread); 0.0003 (Amihud ratio); -0.01 (Market turnover). - MOVE: 0.01 (Bid/ask spread); -0.02 (Asset swap spread); 0.01*** (Amihud ratio); -0.003 (Market turnover). - Federal Reserve intervention announcement: -0.09 (Bid/ask spread); 6.11* (Asset swap spread); 0.03 (Amihud ratio); -0.18 (Market turnover). - USCRD: 0.02*** (Bid/ask spread); 0.11*** (Asset swap spread); 0.003 (Amihud ratio); -0.0004 (Market turnover). - Sample and fit: - Observations: 2,627 (Bid/ask spread and Asset swap spread); 1,751 (Amihud ratio and Market turnover). - R2: 0.21 (Bid/ask spread); 0.04 (Asset swap spread); 0.33 (Amihud ratio); 0.31 (Market turnover). - Adjusted R2: 0.20; 0.03; 0.32; 0.30 respectively. - Note definitions: - Amihud ratio defined as the standard deviation of the spread of the interbank rate to the policy rate divided by daily traded volumes. - VIX = first difference of the Chicago Board Option Exchange implied volatility index of S&P 500 futures. - MOVE = first difference of the U.S. Treasury bond futures volatility index. - USCRD = first difference of the credit spread of investment grade corporate bonds over the 10-year U.S. Treasury yield. - Significance notation: * p<0.1; ** p<0.05; *** p<0.01. - Annex IX (FX Funding Markets) — selected panel regression estimates and local projection impulse responses: - Model uses global common factor and idiosyncratic factors for deviations from covered interest parity (CIP) and bid-offer spreads. - Events covered include initiation of FX swap or FX lending operations and expansion/enhancement announcements of the Federal Reserve’s swap line network. - Selected coefficients (with significance) for dependent variables across common and unique components: - Lagged dependents show strong negative coefficients (e.g., Lagged dependent (1) = -0.50*** for Bid/ask spread). - Federal Reserve swapline announcement: - Bid/ask spread = 1.59**; Bid/ask spread (common) = 0.78***; Bid/ask spread (unique) = 0.51; CIP deviation = 1.48***; CIP deviation (common) = 0.84***; CIP deviation (unique) = 0.64*. - Local intervention announcement: - Bid/ask spread = -0.2; Bid/ask spread (common) = -0.87***; Bid/ask spread (unique) = 0.64; CIP deviation = 0.26; CIP deviation (common) = -0.1; CIP deviation (unique) = 0.24. - Sample and fit: - Observations = 1,697 for all reported dependent variables. - R2 values: 0.21 (Bid/ask spread); 0.04 (Bid/ask spread common); 0.23 (Bid/ask spread unique); 0.03 (CIP deviation); 0.14 (CIP deviation common); 0.03 (CIP deviation unique). - Adjusted R2 values: 0.2; 0.03; 0.22; 0.02; 0.13; 0.02 respectively. - Note definitions: - CIP = covered interest parity. - VIX = first difference of the Chicago Board Option Exchange implied volatility index of S&P 500 futures. - Significance notation: * p<0.1; ** p<0.05; *** p<0.01. *IMF Working Paper: EMDE Central Bank Interventions during COVID-19 to Support Market Functioning, Working Paper No. WP/2024/101* --- _Source: https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024101-print-pdf.pdf_