## 1. Possible Central Bank Losses in Different Scenarios ## Source details **Canonical URL:** [1. Possible Central Bank Losses in Different Scenarios](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2013/_041813.pdf) ## Other formats - [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2013/_041813.pdf.md) - [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2013/_041813.pdf.json) --- ### Overview and scope - Background to the main Board paper, “The Role and Limits of Unconventional Monetary Policy.” - Focus: bond purchase programs and experience of the United States Federal Reserve (Fed), the Bank of England (BOE) and the Bank of Japan (BOJ). - Section A (excerpted): evaluates whether bond purchase programs were effective at decreasing bond yields and through which channels they operated. - Main finding (Section A): bond purchases significantly decreased long-term bond yields; effects most often transmitted through the signaling channel, with duration and scarcity channels also important; evidence of diminishing returns is less clear cut. ### Measurement and empirical approach - Effectiveness measured using event studies controlling for surprises (the “bang for the buck”: change in yield per unit surprise). - Surprise measure: daily changes in the one-year-ahead futures contract on the three-month Libor. - Rationale points preserved exactly from source. - Samples: - U.S. and U.K. pre-crisis: January 3, 2000 to July 8, 2007. - Japan pre-crisis: 2001 to 2006. - Post-crisis: from the date of first bond purchase announcement to January 2013. - Regression: Δy_t = a + b S_t + e_t, with S_t the surprise measured as the change in the one-year ahead futures on 3 month Libor; regressions estimated over event-day samples only. ### Event-study regression key statistics (response to a monetary policy easing: one basis point decrease in one-year-ahead futures on three-month Libor) - Fixed Income - Gov Bond Yield 2yr: - U.S. Pre -0.70***; U.S. Post -0.53*** - U.K. Pre -0.75***; U.K. Post -0.63*** - Japan Pre -0.46***; Japan Post -0.25*** - Gov Bond Yield 10yr: - U.S. Pre -0.52***; U.S. Post -1.14*** - U.K. Pre -0.49***; U.K. Post -0.70*** - Japan Pre -1.36***; Japan Post -0.82*** - MBS Yield 15yr: - U.S. Pre -0.72***; U.S. Post -0.88*** - Agency Yield 10 yr: - U.S. Pre -0.47***; U.S. Post -1.52*** - BBB-AAA Corp Spread: - U.S. Pre -0.03; U.S. Post 0.25*** - U.K. Pre 0.04 - Japan Pre 1.52***; Japan Post -0.08 - Inflation and output - Breakeven Inflation 5yr: - U.S. Pre -0.09*; U.S. Post 0.05 - U.K. Pre -0.20***; U.K. Post -0.18** - Japan Pre 0.08 - Stock Returns: - U.S. Pre 0.01; U.S. Post 0.02 - U.K. Pre -0.05***; U.K. Post -0.05 - Japan Pre -0.05; Japan Post -0.34 - NEER: - U.S. Pre 0.00; U.S. Post -0.01* - U.K. Pre 0.00; U.K. Post -0.02 - Japan Pre -0.31; Japan Post 0.08 - Tail Risk - VIX: - U.S. Pre -0.13*; U.S. Post 0.00 - U.K. Pre 0.11; U.K. Post 0.24 - Japan Pre -0.19; Japan Post 2.02 - FX Risk Reversals (first principal component): - U.S. Pre -0.01; U.S. Post -0.05*** - U.K. Pre -0.05**; U.K. Post -0.07* - Japan Pre -0.01; Japan Post -0.23 - Notes: - One-day (or where noted two-day) changes in asset prices in response to a monetary policy easing (negative surprise). - ***, **, and * indicate significance at the 1 percent, 5 percent, and 10 percent levels, respectively. - Bold fonts in original indicate post-crisis coefficients statistically different from pre-crisis; table source: Bloomberg, Datastream and IMF Staff estimates. - April 2013 QQME announcement for Japan excluded because meaningful surprise could not be extracted from near-zero one-year ahead futures on Libor. ### Key empirical findings (text summary) - Government yields: - Bond purchases, on average, decreased government bond yields. - Effects appear similar in magnitude (for comparable surprises) to conventional rate cuts in the pre-crisis period. - In the U.S., the effect is statistically larger in the post-crisis period than in the pre-crisis period. - Inflation and output: - Effects on equity prices, five-year ahead breakeven inflation rates, and nominal exchange rates are mostly insignificant when using the one-year ahead futures surprise measure. - Equities and expected inflation rates tend to increase significantly when the target factor (current futures) is used as regressor. - Apparent increase in corporate bond risk premia in the U.S. after bond purchases may reflect flight-to-safety dynamics. - Overall, weak results on equity prices support the view that effects on real output from asset purchases are not clear-cut. - Tail risk: - Asset purchases appear to have decreased tail risks of a severe economic crisis in the U.S. and U.K., evidenced by effects on FX risk reversals and monthly inflation surveys (decrease in left-skewness/risk of deflation after announcements). - In Japan, bond purchases did not seem to affect tail risks during QEP or CME. ### Transmission channels and testable hypotheses - Two approaches to disentangle channels: - Model-based: decompose yield curve into risk-neutral expectations of future short rates (signaling channel) and term premia (duration/scarcity channels). - Model-free: use testable asset-price reaction hypotheses to infer channel relevance. - Testable hypotheses: - Signaling channel: larger impact on medium-term vs long-term maturities; 5-year yields should decrease as much as, or more than, 10-year yields. - Duration channel: affects bonds of all maturities by decreasing price of risk; predicts 10-year yields react more than 5-year, and 5-year more than 2-year. - Scarcity channel: only the interest rate of the purchased assets should react most strongly; purchased-maturity yield should react more than other yields and more than a same-maturity OIS contract. ### Transmission channels and empirical evidence (additional) - Channels are not mutually exclusive; e.g., purchases of 10-year MBS may affect 10-year MBS rates (scarcity) and also five-year bonds (signaling). - Evidence mostly points to the signaling channel, though scarcity and duration channels occasionally materialized. - In the U.S. and Japan the signaling channel is prevalent in all programs except in Maturity Extension Program (MEP, “Operation Twist”). - Operation Twist and APP2 in the U.K. worked through the duration channel. - Scarcity effects are rejected for Operation Twist and APP2 since the 10-year OIS and 10-year government yield decreased by the same amount. - Scarcity appears in LSAP1A and LSAP3 (U.S.) and APP1 (U.K.). - U.K. results (APP1 and APP2) point to greater market segmentation than in the U.S. ### Relative importance of channels over time - Relative to pre-crisis rate cuts, bond purchases extended the signaling channel and gave much greater prominence to portfolio rebalancing channels (duration and scarcity together). - Kim and Wright (2005) decomposition (U.S.): - Prior to the crisis: main channel was signaling (predominance of expectations of future short yields component—ST). - Post-crisis: portfolio rebalancing channels significantly grew in importance over all maturities (higher coefficient on term premia component). - Post-crisis: signaling channel was extended in time, with increasing marginal effects up to seven years ahead. ### Diminishing returns and announcement surprises - Evidence of some diminishing returns: - Figure 5 shows bond purchase announcements had similar effects on long-term yields when normalized by surprise (∆y_t/∆S_t), implying “bang for the buck” remained relatively constant. - In the U.S., surprise of subsequent announcements decreased notably over time. - Interpretations offered: - Surprises greater when times are most uncertain. - Most effects may have been felt early, limiting scope for additional easing through similar announcements. - Announcements of significantly different programs, or amid worsening conditions, could have greater effects. ### Macroeconomic effects of bond purchase programs: approach and caveats - Goals: - Quantify macroeconomic impact of bond purchases on output growth and inflation using U.S. and U.K. data. - Analyze relative effects of signaling versus portfolio rebalancing channels. - Caveats: - Long and variable lags between long-term yields, GDP growth, and inflation; crisis likely perturbed historical relationships. - Credit channel may be impaired during deleveraging. - Results focus on U.S. and U.K. and may not generalize. ### VAR specification and identification - VAR variables: inflation (π_t), annualized GDP growth (y_t), short-term interest rate (SR_t: three-month treasury yield), term spread (SP_t: 10-year government bond minus three-month treasury yield). - Lag length: two. - Structural shocks identified: supply, demand, monetary policy (short-term rate), spread shock (term spread shock representing bond purchase effects). - Identification restrictions summary: - Monetary policy shock: raises SR, compresses SP, lowers GDP growth and inflation. - Demand shock: raises inflation, output, and short-term interest rates. - Negative supply shock: raises inflation, lowers output growth. - Spread shock: lowers inflation and output growth, assumed zero contemporaneous impact on SR. - Benchmark samples: - U.S.: 2001Q4 to 2012Q2. - U.K.: 2001Q1 to 2012Q2. - Short-term rate restricted to move within a +/- 5bp band for four quarters in rolling-sample robustness exercises. ### VAR results: effects on growth and inflation - Key results: - The relationship between the term spread and both growth and inflation varies over time; instability existed before the crisis. - A reduction of the term spread increases GDP growth and inflation, but effects appear substantial and short lived, disappearing after a few quarters. - Effects are more persistent for inflation than for GDP growth. - Findings broadly consistent with Peseran and Smith (2012) and Kapetanios and others (2012). ### Which channel is strongest for macroeconomic impact? - Mapping: - ST shock ↔ signaling channel. - TP shock ↔ portfolio rebalancing channels (duration or scarcity). - Identification method: - Use OIS one-year rate as an instrument for ST: - OIS relevance: highly correlated with ST. - OIS validity: short maturity, little term premium, not correlated with TP. - Procedure: regress ST on one-year OIS to obtain ST_hat (ST ෢ ), use ST ෢ in VAR to compare TP vs ST ෢ shocks. - Identification assumptions: - ST ෢ shock: positive shock decreases output and inflation, no initial impact on SR at the ZLB. - TP shock: assumed no contemporaneous impact on ST ෢ . - Result: - A shock to expected future short rates (ST) has approximately a two times larger impact on GDP and inflation than a shock to the term premium (TP). - Lower long rates increase inflation and GDP growth substantially but short-livedly, with effects on inflation more persistent than on GDP. ### Spillover effects on emerging-market asset prices - Early AE asset purchase announcements buoyed global asset prices by decreasing tail risk; effects diminished once markets normalized. - Event study details: - Window: two-day returns over January 1, 2003 to December 31, 2012. - Assets: money, bond, equity, and foreign exchange markets. - Countries: 23 AEs and 11 EMEs. - Surprises measured by one-year-ahead futures on Libor. - Two-day window chosen to account for differing time zones; estimates pooled using Generalized Least Squares with robust standard errors for panel regressions with cross-sectional dependence. - Key insight: - Direct impact of an AE monetary surprise on a foreign asset price is often magnified by indirect spillovers via third countries and simultaneous domestic asset price movements. - Example: direct effect of early U.S. bond purchase announcements on EME bond yields appears to have almost doubled by indirect spillovers through other advanced, systemic economies. ### Methodology for controlling “typical” behavior and event-study limitations - Two-stage approach: - Stage 1: Structural VAR for four systemic AEs (Germany, Japan, U.K., U.S.) identifying shocks through heteroscedasticity. - Stage 2: Use those shocks as inputs into a similar system for each small open economy to uncover typical correlations among domestic assets. - Event study limitations: - Only reveal immediate market reaction; revisions over time not captured. - Announcements convey both policy decision and assessment of current conditions, complicating interpretation. ### Spillover findings (summary) - Spillovers largest when domestic risks are most acute. - U.S. announcements: - Early announcements: widespread reductions in bond yields and money market rates; rises in equity prices; appreciation of currencies vis-à-vis the U.S. dollar. - Later announcements: more muted effects or negative effects on equities and bond yields. - U.K. announcements: similar basic pattern to U.S. - Emerging markets: general appreciation of EME currencies against the U.S. dollar and British pound following early announcements; later announcements had more modest impact. - Japan announcements: led to falling foreign equity prices and broad depreciation of foreign currencies vis-à-vis the yen. - OMT announcement by the ECB: - Raised bond yields in the core of the euro area and much of the rest of the world. - Significantly lowered bond yields in the euro area periphery. - Reduction in tail risks in the euro area led to a generalized rally in global equity markets. - Cross-study evidence: other studies mostly find early AE bond purchase announcements had the largest spillovers; effects diminished once markets normalized. Examples preserved exactly in source. ### Inflation and debt reduction in advanced economies - General assessment: - Higher inflation could help reduce public debt-to-GDP ratio somewhat but could hardly solve the debt problem alone and would raise significant challenges and risks. - Difficult to create higher inflation (Japan’s experience cited). - Un-anchoring of inflation expectations could increase long-term real interest rates, distort resource allocation, reduce economic growth, and hurt lower–income households. - Channels through which higher inflation could reduce debt: 1. Seigniorage. 2. Debt erosion. 3. Primary balance effects. - Seigniorage estimate: - One additional point of inflation would raise seigniorage for the sample by about 0.12 percent of GDP annually. - Raising inflation from WEO baseline projections to 6 percent for five years would generate cumulative seigniorage revenue of about 2½ percentage points of GDP. - Country-specific estimates vary from less than one percent (Canada) to about 5 percent (Japan). - Debt erosion estimate and caveats: - Same increase in inflation, assuming constant debt maturity structure, no impact on growth, and full Fisher effect, would reduce average net debt-to-GDP ratio by less than 10 percentage points by the end of the period for most countries (other than Japan and Italy, where effect would be larger). - Erosion effect drops rapidly after five years as securities are reissued at higher interest rates. - Reductions may be overestimated because assumptions omit effects of inflation on output growth, debt structure, or real interest rates. - Risks and distributional impacts: - Reliance on inflation could lead to fiscal dominance and un-anchoring of expectations. - Un-anchoring could increase liquidity, currency, and interest rate risk in government debt portfolios. - Higher inflation would be regressive and could harm lower–income households. - Financial repression is difficult to enforce and could cause collateral damage. - Output costs of restoring inflation to moderate levels would be substantial based on advanced-economy experience in the 1980s. ### Risks of exit from bond purchase programs and central bank losses - Central bank balance sheets grew in size and maturity following bond purchase programs. - Two implications investigated: 1. Could central banks generate losses from tightening policy rates? 2. How will the transmission channel be affected by the excess liquidity? - Risk of central bank losses: - Maturity mismatch: assets longer maturity; liabilities shifted from currency to reserve balances on which interest is paid. - Upon exit, asset returns unchanged while interest on liabilities would increase with the policy rate, potentially generating negative cash flows and losses. - In net present value terms, these losses are similar to selling long-term assets outright. - Losses are relevant mainly if they put pressure on central bank independence; losses per se should not constrain monetary policy effectiveness. - Pre-crisis vs post-2008: - Pre-crisis AE central bank balance sheets: mostly currency; positive cash flows and remitted surplus income to government. - Post-2008: reserve balances now dominate liabilities; central banks still run a profit because returns on assets exceed reserve remuneration rates (which are unusually low). - Exposure to interest-rate increases: - A sharp increase in short-term rates might inflict losses as interest on liabilities could exceed coupon payments. - FRB and BOE are probably most at risk (BOE has a government indemnity against losses on its QE portfolio). - BOJ has shorter asset maturity profile and lower proportion of bonds; faces less interest-rate risk. - Measurement and accounting: - NPV of losses could appear as revaluation losses, realized if bonds are sold, or unrealized if portfolios are marked to market. - Not all central banks mark portfolios to market; cash flow loss is the same whether portfolios are marked to market or not. - Converting NPV to fiscal pressures requires allocation and maturity details beyond scope. ### ECB balance sheet and direct interest rate risk - ECB faces relatively little direct interest rate risk because "the bulk of its loan assets are linked to its short-term policy rate." - Shrinking ECB balance sheet may be difficult because "those commercial banks currently borrowing from the ECB may not easily be able to repay loans on maturity." - Draining surplus liquidity via OMOs could reduce net income if yield on OMOs exceeds yield earned on lending, "assuming a positively-sloped yield curve," particularly if OMOs are longer maturity. - SMP purchases described as "small relative to balance sheet size" and not considered in loss estimates. - Lending rates are "linked to the short-term OMO rate, regardless of maturity, at the moment." Draining operations "could be longer maturity—several months or even years—and therefore further out on the yield curve." ### Potential central bank losses from bond holdings (valuation framework and scenarios) - Losses depend on: - "(i) the size of central banks’ bond holdings and the extent of any additional purchases;" - "(ii) the maturity of bonds purchased;" - "(iii) the severity of the interest rate scenario." - Price-sensitivity valuation formula: ∆P/P = ½ C∆i^2 - D∆i (second-order Taylor approximation). - Scenarios notes (preserved exactly from source): - Scenario I: parallel shift +100 bps in yields. - Scenario II: +400bps in short-term and up to +225bps in long-term yields (Japan +200 and +125bps, respectively). - Scenario III: +600bps in short-term and up to +375bps in long-run yields (Japan +400bps and +167bps). - "No Expansion B/S" assumes a static balance sheet (B/S). - "The Fed B/S expansion assumes purchases of UST and MBS until end-2013 of $85bn per month." - "The BOE B/S expansion assumes an addition of £75bn to its Asset Purchase Facility in the long range (+7 years) segment." - "The BOJ B/S expansion accounts for the April 4, 2013 announcement of qualitative and quantitative easing of ¥100 trillion JGB purchases in 2013-14 at a longer maturity (going from slightly below 3 years to about 7 years)." - Central bank losses from asset purchases are "based on bond valuations as of February 6, 2013 using price sensitivity to interest rate changes, multiplied by the central banks’ balance sheet exposure, and expressed as a percent of 2012 GDP." - Figure 1 indicates "the net present value of losses, as a percent of GDP, could be substantial, even for modest interest rate increases." - Carpenter and others (2013) arrives at maximum (unrealized) losses of about 1.6 percent in 2017 (as a percent of 2012 GDP), assuming asset purchases continue into 2013. ### Credit risk and banking-system exposures - Credit risk on bond portfolios is "contained by the relatively small proportion of non-government securities purchased." - ECB potentially faces credit risk on lending to the banking system for financial stability purposes. - In a "benign" tightening (in response to higher inflation from growth), non-performing loans should fall; but weak banks in lagging eurozone areas could come under further pressure in a rising rate environment. ### Transmission of policy and exit considerations - Exiting large-scale accommodation moves central banks from excess-liquidity operations to higher policy rates and resumption of OMOs. - With substantial excess reserves, arbitrage "will be less effective" in transmitting the policy rate to other market prices; expectations and inflation impact "will be less smooth." - Central banks "will want to drain excess reserves, at the appropriate time, rather than simply relying on interest on excess reserves." - Market response during exit "will not be as finely tuned" as pre-crisis and may be less predictable for months or years. - Tightening—by selling securities or raising short-term rates—"could lead to sharp movements in the longer-term yield curve" and "reduced market liquidity until expectations stabilize." - Higher rates may "induce an appreciation of the currency," complicating policy and fueling uncertainty over future short rates. - Example: if the U.S. yield curve moves from "(three months–10bp; 10-years 2 percent; 30 years 3.2 percent) to a flat 4 percent, bonds in the 10–30 year range could lose 14 to 16 percent of their value." ### Ways to reduce exit impact on long rates (policy recommendations) - Use clear forward guidance to manage expectations of future short rates. - Focus liquidity-draining operations on short-term instruments, "such as three or six month repos," early in tightening. - When resorting to asset sales, pre-announce them to dampen movements in term premia. - In the U.S., the Fed could operate with a wider range of counterparties than pre-crisis to help overcome market segmentation, though transmission through a larger and more diverse group of counterparties is uncertain. - Emphasis that "expectations, rather than central bank policy, will continue to drive longer-term yields." *Source: UNCONVENTIONAL MONETARY POLICIES—RECENT EXPERIENCE AND PROSPECTS—BACKGROUND PAPER, April 18, 2013 (International Monetary Fund).* ### 1. Possible Central Bank Losses in Different Scenarios ___________________________________ 22 ### 1. Possible Central Bank Losses in Different Scenarios ### Overview and scope - Paper provides background to the main Board paper, “The Role and Limits of Unconventional Monetary Policy.” - Divided into five sections, mostly focused on bond purchase programs and the experience of the United States Federal Reserve (Fed), the Bank of England (BOE) and the Bank of Japan (BOJ). - Section A (excerpted here) evaluates whether bond purchase programs were effective at decreasing bond yields and through which channels they operated. - Main finding summary (Section A): bond purchases significantly decreased long-term bond yields; effects most often transmitted through the signaling channel, with duration and scarcity channels also important; evidence of diminishing returns is less clear cut. ### Measurement and empirical approach - Effectiveness measured using event studies that control for surprises (the “bang for the buck”: change in yield per unit surprise). - Surprise measure: daily changes in the one-year-ahead futures contract on the three-month Libor. - Rationale: - Variance in one-year ahead futures remains relatively stable from pre- to post-crisis despite the ZLB; current futures variance drops dramatically at the ZLB. - One-year ahead futures account for both target and path surprises; current futures more closely correlated with the target factor. - Futures on Libor have good forecasting power. - Sample: all official and scheduled central bank announcement days, divided into pre-crisis and post-crisis: - U.S. and U.K. pre-crisis: January 3, 2000 to July 8, 2007. - Japan pre-crisis: 2001 to 2006 (to match QEP duration). - Post-crisis: from the date of first bond purchase announcement to January 2013. - Regression specification: Δy_t = a + b S_t + e_t, with S_t the surprise measured as the change in the one-year ahead futures on 3 month Libor; regressions estimated over event-day samples only. ### Event-study regression key statistics (response to a monetary policy easing: one basis point decrease in one-year-ahead futures on three-month Libor) - Fixed Income - Gov Bond Yield 2yr: - U.S. Pre -0.70***; U.S. Post -0.53*** - U.K. Pre -0.75***; U.K. Post -0.63*** - Japan Pre -0.46***; Japan Post -0.25*** - Gov Bond Yield 10yr: - U.S. Pre -0.52***; U.S. Post -1.14*** - U.K. Pre -0.49***; U.K. Post -0.70*** - Japan Pre -1.36***; Japan Post -0.82*** - MBS Yield 15yr: - U.S. Pre -0.72***; U.S. Post -0.88*** - Agency Yield 10 yr: - U.S. Pre -0.47***; U.S. Post -1.52*** - BBB-AAA Corp Spread: - U.S. Pre -0.03; U.S. Post 0.25*** - U.K. Pre 0.04 - Japan Pre 1.52***; Japan Post -0.08 - Inflation and output - Breakeven Inflation 5yr: - U.S. Pre -0.09*; U.S. Post 0.05 - U.K. Pre -0.20***; U.K. Post -0.18** - Japan Pre 0.08 - Stock Returns: - U.S. Pre 0.01; U.S. Post 0.02 - U.K. Pre -0.05***; U.K. Post -0.05 - Japan Pre -0.05; Japan Post -0.34 - NEER: - U.S. Pre 0.00; U.S. Post -0.01* - U.K. Pre 0.00; U.K. Post -0.02 - Japan Pre -0.31; Japan Post 0.08 - Tail Risk - VIX: - U.S. Pre -0.13*; U.S. Post 0.00 - U.K. Pre 0.11; U.K. Post 0.24 - Japan Pre -0.19; Japan Post 2.02 - FX Risk Reversals (first principal component): - U.S. Pre -0.01; U.S. Post -0.05*** - U.K. Pre -0.05**; U.K. Post -0.07* - Japan Pre -0.01; Japan Post -0.23 - Notes: - One-day (or where noted two-day) changes in asset prices in response to a monetary policy easing (negative surprise). - ***, **, and * indicate significance at the 1 percent, 5 percent, and 10 percent levels, respectively. - Bold fonts in original indicate post-crisis coefficients statistically different from pre-crisis; table source: Bloomberg, Datastream and IMF Staff estimates. - April 2013 QQME announcement for Japan excluded because meaningful surprise could not be extracted from near-zero one-year ahead futures on Libor. ### Key empirical findings (text summary) - Government yields: - Bond purchases, on average, decreased government bond yields. - Effects appear similar in magnitude (for comparable surprises) to conventional rate cuts in the pre-crisis period. - In the U.S., the effect is statistically larger in the post-crisis period than in the pre-crisis period. - Inflation and output: - Effects on equity prices, five-year ahead breakeven inflation rates, and nominal exchange rates are mostly insignificant when using the one-year ahead futures surprise measure. - Equities and expected inflation rates tend to increase significantly when the target factor (current futures) is used as regressor, consistent with prior literature. - Apparent increase in corporate bond risk premia in the U.S. after bond purchases may reflect flight-to-safety dynamics. - Overall, weak results on equity prices support the view that effects on real output from asset purchases are not clear-cut. - Tail risk: - Asset purchases appear to have decreased tail risks of a severe economic crisis in the U.S. and U.K., evidenced by effects on FX risk reversals and corroborating evidence from monthly inflation surveys (decrease in left-skewness/risk of deflation after announcements). - In Japan, bond purchases did not seem to affect tail risks during QEP or CME, consistent with Japan not facing the same crisis-level prospects as the U.S. and U.K. in 2008–09. ### Transmission channels and testable hypotheses - Two approaches to disentangle channels: - Model-based: decompose yield curve into risk-neutral expectations of future short rates (signaling channel) and term premia (duration/scarcity channels). - Model-free (used in this paper): use testable asset-price reaction hypotheses to infer channel relevance. - Testable hypotheses for channels: - Signaling channel: - Should have larger impact on medium-term maturities relative to long-term maturities. - After announcement, 5-year bond yields should decrease as much as, or more than, 10-year yields. - Duration channel: - Affects bonds of all maturities by decreasing the price of risk; larger effects on longer maturities. - Predicts 10-year yields react more than 5-year yields, and 5-year yields more than 2-year yields. - Scarcity channel: - Only the interest rate of the purchased assets should react most strongly. - If 10-year bonds are purchased, 10-year yields should react more than any other yield, and more than a 10-year OIS contract (which has lower “use value” and is less accepted as collateral). *Source: UNCONVENTIONAL MONETARY POLICIES—RECENT EXPERIENCE AND PROSPECTS—BACKGROUND PAPER, April 18, 2013 (International Monetary Fund).* ### 8. It remains to be said that channels are not mutually exclusive of one another. For ### _041813 - 8. It remains to be said that channels are not mutually exclusive of one another. For ### Transmission channels and empirical evidence - Channels are not mutually exclusive: e.g., purchases of 10-year MBS may most directly affect 10-year MBS rates (scarcity), but five-year bonds can also react and may decrease by more than 10-year bonds (signaling). - Evidence mostly points to the signaling channel, though the scarcity and duration channels occasionally played important roles. - Staff estimates of changes in yields of key bonds are displayed in Figure 2 in Appendix I (changes standardized by the announcement’s surprise, ∆y_t/∆S_t, where S_t is one-year ahead futures on Libor). - In the U.S. and Japan the signaling channel is prevalent in all programs except in Maturity Extension Program (MEP, “Operation Twist”). - Operation Twist and the second wave of the Asset Purchase Program (APP2) in the U.K. worked through the duration channel (longer-term maturities affected most). - Scarcity effects are rejected for Operation Twist and APP2 since the 10-year OIS and 10-year government yield decreased by the same amount. - Scarcity appears in LSAP1A and LSAP3 (U.S.) and APP1 (U.K.), where purchased assets decreased substantially more than other assets (purchased assets are marked by an asterisk in the figure). - U.K. results (APP1 and APP2) point to greater market segmentation than in the U.S. ### Relative importance of channels over time - Relative to pre-crisis rate cuts, bond purchases extended the signaling channel and gave much greater prominence to the portfolio rebalancing channel (duration and scarcity together). - Using the Kim and Wright (2005) decomposition of the U.S. government bond yield curve: - Prior to the crisis: main channel was signaling (predominance of expectations of future short yields component—ST). - Post-crisis: portfolio rebalancing channels significantly grew in importance over all maturities (higher coefficient on term premia component). - Post-crisis: signaling channel was considerably extended in time, having increasing marginal effects up to seven years ahead. ### Diminishing returns and announcement surprises - Evidence points to some degree of diminishing returns to bond purchase programs: - Figure 5 in Appendix I shows bond purchase announcements had similar effects on long-term yields through time when normalized by the surprise factor (∆y_t/∆S_t), implying the “bang for the buck” remained relatively constant. - In the U.S., the role of signaling versus portfolio rebalancing channels seems relatively constant over time, but the surprise of subsequent announcements decreased notably over time. - Interpretations: - Surprises are greater when times are most uncertain. - Most effects may have been felt early, in anticipation of later programs, limiting scope for additional easing through similar announcements. - Announcements of significantly different programs (in size and scope), or made amid worsening economic conditions, could have greater effects. ### Macroeconomic effects of bond purchase programs: approach and key caveats - Goals of the section: - Quantify macroeconomic impact of bond purchases on output growth and inflation using U.S. and U.K. data. - Analyze relative effects of signaling versus portfolio rebalancing channels on growth and inflation. - Caveats: - Relationship between long-term yields, GDP growth, and inflation is subject to long and variable lags; crisis likely perturbed historical relationships. - Credit channel may be impaired during deleveraging. - Results focus on the U.S. and U.K. and cannot be easily generalized. ### VAR specification and identification - VAR model includes four variables: inflation (π_t), annualized GDP growth (y_t), short-term interest rate (SR_t: three-month treasury yield), term spread (SP_t: 10-year government bond minus three-month treasury yield). - Lag length set to two. - Four structural shocks identified: supply, demand, monetary policy (short-term rate), and spread shock (shock to term spread representing effects of bond purchase programs flattening the yield curve). - Key identification restrictions (textual summary of Table 2): - Monetary policy shock: raises SR, compresses SP, lowers GDP growth and inflation. - Demand shock: raises inflation, output, and short-term interest rates. - Negative supply shock: raises inflation, lowers output growth. - Spread shock: lowers inflation and output growth, assumed to have zero contemporaneous impact on the short-term interest rate (to identify a “pure” spread shock as a flattening without a cut in short-term rates). - Benchmark samples: - U.S.: 2001Q4 to 2012Q2. - U.K.: 2001Q1 to 2012Q2. - Note: benchmark sample dictated by availability of OIS data used as an instrument for ST (start in 2001Q4 for U.S. and 2001Q1 for U.K.). - Short-term rate restricted to move within a +/- 5bp band for four quarters after the initial shock in rolling-sample robustness exercises. ### VAR results: effects on growth and inflation - Key results: - The relationship between the term spread and both growth and inflation varies over time; instability existed before the crisis. - A reduction of the term spread increases GDP growth and inflation, but effects appear substantial and short lived, disappearing after a few quarters. - Effects are more persistent for inflation than for GDP growth. - These findings are broadly consistent with Peseran and Smith (2012) and Kapetanios and others (2012). ### Which channel is strongest for macroeconomic impact? - Conceptual mapping: - Shock to expected future short rates (ST) ↔ signaling channel. - Shock to term premia (TP) ↔ portfolio rebalancing channels (duration or scarcity). - Identification challenge: ST and TP shocks affect inflation, output, and short-term rates similarly and thus cannot be identified jointly in the baseline equation. - Novel identification method: - Use OIS rate with a one-year maturity as an instrument for ST: - Relevance: OIS is highly correlated with ST (both capture expectations of future short rates). - Validity: OIS is short maturity and thus contains little term premium, so it is not correlated with TP. - Procedure: regress ST on the one-year OIS and a constant, obtain predicted ST_hat (ST ෢ ) exogenous to TP, and use ST ෢ in the VAR to compare the effects of shocks to TP and ST ෢ . - Identification assumptions for this exercise: - ST ෢ shock: positive shock decreases output and inflation, has no initial impact on SR (short-term rate) at the ZLB. - TP shock: assumed to have no contemporaneous impact on ST ෢ . - Result: - A shock to expected future short rates (ST) has approximately a two times larger impact on GDP and inflation than a shock to the term premium (TP). - Otherwise, results remain that lower long rates increase inflation and GDP growth substantially but short-livedly, with effects on inflation more persistent than on GDP. ### Spillover effects on emerging-market asset prices - Early announcements of asset purchases in advanced economies buoyed global asset prices by decreasing the tail risk of a severe recession; effects diminished once markets normalized. - This section focuses on immediate impact of UMP announcements on asset prices in emerging economies using event studies. - Event study details: - Window: two-day returns over January 1, 2003 to December 31, 2012. - Asset markets considered: money, bond, equity, and foreign exchange markets. - Countries: 23 advanced economies (AEs) and 11 emerging market economies (EMEs) listed in the source. - Surprises measured by one-year-ahead futures on Libor. - Two-day window chosen to account for differing time zones; country-specific estimates pooled across subgroups using Generalized Least Squares with robust standard errors for panel regressions with cross-sectional dependence. - Key insight: - Direct impact of a monetary surprise on a foreign asset price is often magnified by indirect spillovers via third countries and simultaneous domestic asset price movements. - Example: direct effect of early U.S. bond purchase announcements on bond yields in EMEs appears to have almost doubled by indirect spillovers through simultaneous asset price movements in other advanced, systemic economies. *Prepared by IMF staff as presented in the supplied chapter excerpt.* ### 27. A two-stage approach is used to control for this “typical” behavior. First, the ### _041813 - 27. A two-stage approach is used to control for this “typical” behavior. First, the ### Methodology - A two-stage approach is used to control for “typical” behavior. - Stage 1: Examine transmission of shocks between bond yields, equity prices, exchange rates, and money market rates within and between four systemic financial markets (Germany, Japan, the U.K., and the U.S.) in a simultaneous manner. - A structural VAR is estimated by identifying the shocks through heteroscedasticity. - Stage 2: Use the corresponding underlying shocks as inputs into a similar system for each small open economy of interest to uncover typical correlations among domestic assets. - Example: The model for Brazil accounts for contemporaneous financial markets shocks from the four systemic AEs as well as interactions across Brazilian asset prices. ### Limitations of the event study approach - Event studies advantages: - Help avoid contamination by nonpolicy-related news when examining knock-on responses to bond purchase announcements. - Spillover effects expected to rapidly transmit between liquid and highly integrated financial markets through portfolio rebalancing and expectation channels. - Event studies limitations: - Only reveal the immediate market reaction; revisions as views evolve over time are not captured. - Announcements convey both a policy decision and an assessment of current economic conditions, complicating interpretation. ### Spillover findings from bond purchase announcements - General result: - Spillovers from bond purchases are largest when domestic risks are most acute. - U.S. announcements: - Early U.S. bond purchase announcements led to major global financial market rallies: - Widespread reductions in bond yields and money market rates (Figures 13 and 16). - Rises in equity prices (Figure 14). - Appreciation of currencies vis-à-vis the U.S. dollar (Figure 15). - Later announcements had either more muted effects on all foreign assets, or even negative effects on equities and bond yields. - U.K. announcements: - Similar basic pattern to U.S.; exchange rate effect similar to early U.S. announcements. - Emerging markets: - General appreciation of emerging-market currencies against the U.S. dollar and British pound following early announcements (Figure 15). - Later announcements had a more modest impact on exchange rates. - Japan announcements: - Asset purchase announcements led to falling foreign equity prices and broad depreciation of foreign currencies vis-à-vis the Japanese yen. - Possibly reflects market repricing of yen-denominated assets as the 2008–09 financial crisis mostly affected the rest of the world. - Market-conditions dependence: - Spillover effects of early bond purchase programs appear largest; spillovers were smaller once markets normalized. - Lower spillover effects from later programs may also reflect smaller announcement surprises. ### Notable case: OMT announcement by the ECB - The OMT announcement: - Raised bond yields in the core of the euro area and in much of the rest of the world. - Significantly lowered bond yields in the euro area periphery. - The reduction in tail risks in the euro area led to a generalized rally in global equity markets. ### Pre-crisis vs. post-crisis and cross-study evidence - Pre-crisis easing and post-crisis non-UMP announcements: - Sign and size of bond market spillovers were generally similar. - Evidence from other asset markets is more mixed. - Pre-crisis period: U.S. monetary policy had larger and more generalized financial market spillovers than other countries. - Expansionary U.S. monetary policy shocks tended to boost domestic financial conditions in the rest of the world: bond yields generally fell while equity prices rose together with money market rates in AEs. - Other studies: - Mostly find early AE bond purchase announcements had the largest spillovers, buoying asset prices globally by decreasing tail risk of a severe recession; effects diminished once markets normalized. - Significant spillover effects on bond yields and currency in EMEs, with larger estimated effects from LSAP1 than from LSAP2. - Examples cited: Neely (2012) finds US quantitative easing lowered bond rates in other AEs by 20–80 basis points and the value of U.S. dollar by 4–11 percentage points. ### Inflation and debt reduction in advanced economies - General assessment: - Higher inflation could help reduce public debt-to-GDP ratio somewhat but could hardly solve the debt problem alone and would raise significant challenges and risks. - Difficult to create higher inflation (Japan’s experience cited). - Un-anchoring of inflation expectations could increase long-term real interest rates, distort resource allocation, reduce economic growth, and hurt lower–income households. - Channels through which higher inflation could reduce debt: 1. Seigniorage: governments capture real resources by base money creation. 2. Debt erosion: inflation erodes the real value of the debt; impact depends on maturity structure and currency denomination of debt and interest rate response. 3. Primary balance effects: e.g., if tax brackets are not indexed under a progressive income tax. - Seigniorage quantitative estimate: - One additional point of inflation would raise seigniorage for the sample by about 0.12 percent of GDP annually. - Raising inflation from WEO baseline projections to 6 percent for five years would generate cumulative seigniorage revenue of about 2½ percentage points of GDP. - Country-specific estimates vary from less than one percent (Canada) to about 5 percent (Japan). - Debt erosion quantitative estimate and caveats: - Same increase in inflation, assuming constant debt maturity structure, no impact on growth, and full Fisher effect, would reduce average net debt-to-GDP ratio by less than 10 percentage points by the end of the period for most countries (other than Japan and Italy, where the effect would be larger). - Erosion effect drops rapidly after five years as securities are reissued at higher interest rates. - These reductions may be overestimated because assumptions omit effects of inflation on output growth, debt structure, or real interest rates. - Risks and distributional impacts: - Reliance on inflation could lead to fiscal dominance and un-anchoring of expectations, undermining credibility of inflation-control frameworks. - Un-anchoring could increase liquidity, currency, and interest rate risk in government debt portfolios, making them more crisis prone. - Higher inflation would be regressive: a tax on bondholders but disproportionately affecting lower-income households with limited access to indexed assets. - Introducing financial repression could keep interest rates low but would be difficult to enforce and could cause collateral damage. - Output costs of restoring inflation to moderate levels would be substantial based on advanced-economy experience in the 1980s. ### Risks of exit from bond purchase programs - Central bank balance sheets: - Have grown in size and maturity following recent bond purchase programs. - Two implications investigated: 1. Could central banks generate losses from tightening policy rates? 2. How will the transmission channel be affected by the excess liquidity? - Risk of central bank losses: - Maturity mismatch: assets have become longer maturity while liabilities shifted from currency to reserve balances on which interest is paid. - Upon exit, asset returns unchanged while interest paid on liabilities would increase with the policy rate, potentially generating negative cash flows and losses. - In net present value terms, these losses are similar to selling long-term assets outright. - Losses are relevant mainly to the extent they could put pressure on central bank independence; losses per se should not constrain monetary policy effectiveness. - Pre-crisis position: - AE central bank balance sheets consisted mostly of currency; central banks generated positive cash flows and remitted surplus income to the government. - Post-2008 changes: - Balance sheet expansion since 2008, particularly in the U.K. and U.S., means reserve balances—on which interest is paid—now dominate liabilities. - For the moment, central banks still run a profit because returns on asset portfolios are higher than reserve remuneration rates (which currently are unusually low). - Exposure to interest-rate increases: - A sharp increase in short-term rates might inflict losses as interest payments on liabilities could exceed coupon payments on securities. - The Federal Reserve Bank (FRB) and Bank of England (BOE) are probably most at risk in this respect (BOE has a government indemnity against losses sustained on its QE portfolio). - Bank of Japan (BOJ) has a shorter asset maturity profile and a lower proportion of bonds in its balance sheet total, thus faces less interest-rate risk. - Measurement and accounting: - Net Present Value (NPV) of likely future losses could show up as revaluation losses, realized if bonds are sold to drain excess reserves, or unrealized if portfolios are marked to market. - Not all central banks mark portfolios to market; cash flow loss is the same whether portfolios are marked to market or not. - Converting NPV into fiscal pressures requires knowledge of allocation of bond purchases over time and maturities and accounting for portfolio dynamics, which is beyond the scope of the section. *UNCONVENTIONAL MONETARY POLICIES—RECENT EXPERIENCE AND PROSPECTS—BACKGROUND PAPER* ### 43. The ECB faces relatively little direct interest rate risk, as the bulk of its loan assets are ### _041813 - 43. The ECB faces relatively little direct interest rate risk, as the bulk of its loan assets are ### ECB balance sheet and direct interest rate risk - The ECB faces relatively little direct interest rate risk because "the bulk of its loan assets are linked to its short-term policy rate." - Shrinking the ECB balance sheet may be difficult because "those commercial banks currently borrowing from the ECB may not easily be able to repay loans on maturity." - Draining surplus liquidity via open market operations (OMOs) could reduce net income if the yield on liquidity-draining OMOs exceeds the yield earned on lending, "assuming a positively-sloped yield curve," particularly if draining operations are of longer maturity. - The ECB’s new bond portfolio (SMP purchases) is described as "small relative to balance sheet size" and "is not considered in the estimation of losses discussed below." - Lending rates are "linked to the short-term OMO rate, regardless of maturity, at the moment." Draining operations "could be longer maturity—several months or even years—and therefore further out on the yield curve." ### Potential central bank losses from bond holdings (valuation framework and magnitudes) - Losses depend on: - "(i) the size of central banks’ bond holdings and the extent of any additional purchases;" - "(ii) the maturity of bonds purchased;" - "(iii) the severity of the interest rate scenario." - Price-sensitivity valuation formula used: ∆P/P = ½ C∆i^2 - D∆i (a second-order Taylor approximation where ∆P/P denotes relative bond price changes, ∆i denotes changes in yields, C denotes convexity and D modified duration). - Projections are based on various interest rate scenarios and current and future bond purchase strategies. - Figure 1 indicates "the net present value of losses, as a percent of GDP, could be substantial, even for modest interest rate increases." - Comparison to other work: Carpenter and others (2013) "arrives at maximum (unrealized) losses of about 1.6 percent in 2017 (as a percent of 2012 GDP), assuming asset purchases continue into 2013." Notes to scenarios and balance-sheet assumptions (as stated in the source notes) - Scenario I: parallel shift +100 bps in yields. - Scenario II: +400bps in short-term and up to +225bps in long-term yields (Japan +200 and +125bps, respectively). - Scenario III: +600bps in short-term and up to +375bps in long-run yields (Japan +400bps and +167bps). - "No Expansion B/S" assumes a static balance sheet (B/S). - "The Fed B/S expansion assumes purchases of UST and MBS until end-2013 of $85bn per month." - "The BOE B/S expansion assumes an addition of £75bn to its Asset Purchase Facility in the long range (+7 years) segment." - "The BOJ B/S expansion accounts for the April 4, 2013 announcement of qualitative and quantitative easing of ¥100 trillion JGB purchases in 2013-14 at a longer maturity (going from slightly below 3 years to about 7 years)." - Central Bank losses from asset purchases are "based on bond valuations as of February 6, 2013 using price sensitivity to interest rate changes, multiplied by the central banks’ balance sheet exposure, and expressed as a percent of 2012 GDP." ### Credit risk and banking-system exposures - Credit risk on bond portfolios is "contained by the relatively small proportion of non-government securities purchased by these central banks." - The ECB "potentially faces credit risk on its lending to the banking system for financial stability purposes." - In a "benign" scenario (tightening in response to higher inflation from economic growth), "non-performing loans should fall and bank balance sheets should improve," but "some areas of the eurozone may lag in economic recovery." - Weak banks in lagging areas "could come under further pressure in a rising rate environment: weak banks may not be able to pass on to weak customers the rising costs of financing their balance sheets." ### Transmission of policy and exit considerations - Exiting from large-scale accommodation will move central banks from operating with excess liquidity—"where market rates are guided by the interest paid on excess reserves"—to higher policy rates and "eventual resumption of OMOs to guide short-term market rates." - In the presence of substantial excess reserve balances, arbitrage "will be less effective" in transmitting the policy rate to other market prices, implying market expectations and inflation impact "will be less smooth." - Consequently, central banks "will want to drain excess reserves, at the appropriate time, rather than simply relying on interest on excess reserves." - Market response during exit "will not be as finely tuned during the exit as pre-crisis, as the market response will be less predictable initially (possibly for several months or even years)." - A move to tightening—by running down securities portfolios or raising short-term rates—"could lead to sharp movements in the longer-term yield curve" and "reduced market liquidity until expectations stabilize." - Higher rates may "induce an appreciation of the currency, further complicating policy and fueling uncertainty over future short rates." - Example provided: if the U.S. yield curve moves from "(three months–10bp; 10-years 2 percent; 30 years 3.2 percent) to a flat 4 percent, bonds in the 10–30 year range could lose 14 to 16 percent of their value." ### Ways to reduce exit impact on long rates (policy recommendations) - Use clear forward guidance to help manage expectations of future short rates. - Focus liquidity-draining operations on short-term instruments, "such as three or six month repos," at least early in the tightening cycle. - When resorting to asset sales, pre-announce them to dampen movements in term premia. - In the U.S., the Fed could operate with a wider range of counterparties than pre-crisis to help overcome market segmentation, though "it may be less clear how transmission through a larger and more diverse group of counterparties will work." - Emphasis that "expectations, rather than central bank policy, will continue to drive longer-term yields." *Source: UNCONVENTIONAL MONETARY POLICIES—RECENT EXPERIENCE AND PROSPECTS—BACKGROUND PAPER (excerpt).* ### Appendix II. Supplementary Figures and Tables to Section B ### Appendix II. Supplementary Figures and Tables to Section B ### Instability of Coefficients (Figure 6) - Source: IMF staff estimates. - Note: Each point represents the sum of the first and second lag coefficients of the spread from the reduced form VAR. The coefficients are obtained using a roling sample with a fixed window. For example, 2012Q2 represents the sample 2001Q4-2012Q2 for US and 2000Q4-2012Q1 for UK. - Charts/labels shown: - U.K. Instability of Coefficients: vertical axis values shown: -2, -1.5, -1, -0.5, 0, 0.5, 1; horizontal axis quarter labels: 2001Q1 2002Q1 2003Q1 2004Q1 2005Q1 2006Q1 2007Q1 2008Q1 2009Q1 2010Q1 2011Q1 2012Q1; series labels: COEF_GDP_UK COEF_INF_UK - U.S. Instability of Coefficients: vertical axis values shown: -1, -0.8, -0.6, -0.4, -0.2, 0, 0.2, 0.4, 0.6, 0.8; horizontal axis quarter labels: 2001Q1 2002Q1 2003Q1 2004Q1 2005Q1 2006Q1 2007Q1 2008Q1 2009Q1 2010Q1 2011Q1 2012Q1; series labels: COEF_GDP_US COEF_INF_US ### Responses to 25 bps Spread Shock (Figures 7 and 8) - Source: IMF staff estimates. - United States (Figure 7): responses shown for Inflation, GDP, Short-term Interest Rate, Spread with month/time-step axis labeled 1 through 12 (displayed as 123456789101112). Displayed series: Median, 75th percentile, 25th percentile. Vertical axis markers: - Inflation: -0.2, 0, 0.2, 0.4, 0.6, 0.8, 1 - GDP: -1, -0.5, 0, 0.5, 1, 1.5, 2, 2.5, 3, 3.5 - Short-term Interest Rate: -0.06, -0.04, -0.02, 0, 0.02, 0.04, 0.06, 0.08, 0.1, 0.12, 0.14, 0.16 - Spread: -0.35, -0.3, -0.25, -0.2, -0.15, -0.1, -0.05, 0 - United Kingdom (Figure 8): responses shown for Inflation, GDP, Short-term Interest Rate, Spread with month/time-step axis labeled 1 through 12 (123456789101112). Displayed series: Median, 75th percentile, 25th percentile. Vertical axis markers: - Inflation: -0.15, -0.1, -0.05, 0, 0.05, 0.1, 0.15, 0.2, 0.25, 0.3 - GDP: -0.4, -0.3, -0.2, -0.1, 0, 0.1, 0.2, 0.3, 0.4 - Short-term Interest Rate: -0.06, -0.04, -0.02, 0, 0.02, 0.04, 0.06, 0.08 - Spread: -0.3, -0.25, -0.2, -0.15, -0.1, -0.05, 0, 0.05 ### Responses to 25 bps Short-term (ST) Shock (Figures 9 and 10) - Source: IMF staff estimates. - United States (Figure 9): responses shown for Inflation, GDP, Short-term Interest Rate, TP, ST with month/time-step axis labeled 1 through 12 (123456789101112). Displayed series: Median, 75th percentile, 25th percentile. Vertical axis markers: - Inflation: -1, -0.5, 0, 0.5, 1, 1.5, 2, 2.5, 3 - GDP: -2, -1, 0, 1, 2, 3, 4, 5, 6 - Short-term Interest Rate: -1.2, -1, -0.8, -0.6, -0.4, -0.2, 0, 0.2, 0.4, 0.6 - TP: -1.2, -1, -0.8, -0.6, -0.4, -0.2, 0, 0.2 - ST: -0.5, -0.4, -0.3, -0.2, -0.1, 0, 0.1, 0.2, 0.3 - United Kingdom (Figure 10): responses shown for Inflation, GDP, Short-term Interest Rate, TP, ST with month/time-step axis labeled 1 through 12 (123456789101112). Displayed series: Median, 75th percentile, 25th percentile. Vertical axis markers: - Inflation: 0, 0.2, 0.4, 0.6, 0.8, 1, 1.2, 1.4 - GDP: -2, -1, 0, 1, 2, 3, 4, 5 - Short-term Interest Rate: -1.2, -1, -0.8, -0.6, -0.4, -0.2, 0, 0.2 - TP: -0.45, -0.4, -0.35, -0.3, -0.25, -0.2, -0.15, -0.1, -0.05, 0, 0.05, 0.1 - ST: -0.4, -0.35, -0.3, -0.25, -0.2, -0.15, -0.1, -0.05, 0 ### Responses to 25 bps Other Shock (Figures 11 and 12) - Source: IMF staff estimates. - United States (Figure 11): responses shown for Inflation, GDP, Short term interest rate, ST, TP with month/time-step axis labeled 1 through 12 (123456789101112). Displayed series: Median, 75th percentile, 25th percentile. Vertical axis markers: - Inflation: -0.2, 0, 0.2, 0.4, 0.6, 0.8, 1 - GDP: -0.6, -0.4, -0.2, 0, 0.2, 0.4, 0.6, 0.8, 1, 1.2, 1.4 - Short term interest rate: -0.1, -0.05, 0, 0.05, 0.1, 0.15, 0.2, 0.25, 0.3, 0.35, 0.4, 0.45 - ST: -0.02, 0, 0.02, 0.04, 0.06, 0.08, 0.1, 0.12, 0.14, 0.16 - TP: -0.3, -0.25, -0.2, -0.15, -0.1, -0.05, 0, 0.05 - United Kingdom (Figure 12): responses shown to 25 bps TP Shock for Inflation, GDP, Short-term Interest Rate, ST, TP with month/time-step axis labeled 1 through 12 (123456789101112). Displayed series: Median, 75th percentile, 25th percentile. Vertical axis markers: - Inflation: -0.2, 0, 0.2, 0.4, 0.6, 0.8, 1 - GDP: -2, -1, 0, 1, 2, 3, 4 - Short-term Interest Rate: -0.4, -0.2, 0, 0.2, 0.4, 0.6, 0.8, 1 - ST: -0.1, -0.05, 0, 0.05, 0.1, 0.15, 0.2, 0.25, 0.3, 0.35 - TP: -0.3, -0.25, -0.2, -0.15, -0.1, -0.05, 0, 0.05, 0.1 ### Two-Day Changes in Asset Prices and Rates (Tables 6–9) - Sources: Datastream, Bloomberg, and staff calculations. - Table 6. Two-Day Change in 10-Year Bond Rates: categorical/timeline entries showing directional responses (arrows) by monetary policy announcement and region/market grouping. Examples preserved exactly from source: - United States: Pre-UMP↓↓↓↓↓↓; Post-UMP↓↓↑ ↓↑↓↓; LSAP1A↓↓↓ ↓↓↓; LSAP1B↓↓↓ ↓; LSAP2↓↓↓ ↓↓; TWIST↑↑↓; LSAP3↑↑↓ - United Kingdom: Pre-UMP↓↓ ↓↓↓↓; Post-UMP↑↑↑ ↑; APP1↓↓ ↓; APP2↑↑ ↑↑ ↑; FLS↓↓↓ - Japan: Pre-UMP↓↓↓↓↓↓↓; Post-UMP↑↑; CME↓↓↓↓↓↓ - Emerging/Other regions: entries include directional strings and ellipses as shown in source. - Table 7. Two-Day Change in the Rate of Return of the Stock Market: directional responses by announcement and region. Examples preserved exactly: - United States: Pre-UMP↑↑↑ ↑; Post-UMP↓↓↓↓↓; LSAP1A↑↑↑↑↑↑; LSAP1B↑↑; LSAP2↓↓↓ ↓↓↓↓; TWIST↑↑; LSAP3↑↑ - United Kingdom: Pre-UMP↓↓↓↓↓ ↓; Post-UMP↓↓↓↓ ↓; APP1↓↓ ↓ ↓ ↓ ↓; APP2↑↑ ↑↑↑↑; FLS↑↑ ↑ - Japan: Pre-UMP↑↑; Post-UMP; CME↓↓↓↓↓↓↓ - Table 8. Two-Day Change in the Rate of Return of Cross Bilateral Foreign Exchange: directional responses by announcement and region. Examples preserved exactly: - United States: Pre-UMP↓; Post-UMP↓↓↓ ↓↓↓↑; LSAP1A↓↓ ↓↓↓↓; LSAP1B↓↓↓ ↓↓↓↓; LSAP2↓↓ ↓↓↓; TWIST↓↓↓↑; LSAP3↓↓↓ - United Kingdom: Pre-UMP↓↓↓↓ ↓ ↓; Post-UMP↑; APP1↓↓↓↓ ↓ ↓; APP2↓↑; FLS↓↓ - Japan: Pre-UMP↑↑↑; Post-UMP↓; CME↑ - Table 9. Two-Day Change in Money Market Rates: directional responses by announcement and region. Notes include country group compositions: - Note 1/ Core EA: Austria, Belgium, Finland, France, and Netherlands. - Note 2/ Periphery EA: Greece, Ireland, Italy, Spain, and Portugal. - Note 3/ European Safe Havens: Denmark and Switzerland. - Note 4/ Inflation Targeters: Australia, Canada, New Zealand, Norway, and Sweden. - Note 5/ Latin America: Brazil and Mexico. - Note 6/ Asia: Indonesia, Malaysia, South Korea, and Thailand. - Note 7/ Europe: Czech Republic and Poland. - Note 8/ Other EMs: Russia, South Africa, and Turkey. - Directional examples preserved exactly: - United States: Pre-UMP↑↑ ↑↑↑↑; Post-UMP↓; LSAP1A↓↓↓↓; LSAP1B↑; LSAP2; TWIST; LSAP3 - United Kingdom: Pre-UMP↓↑ ↓↓↓; Post-UMP; APP1↓; APP2; FLS - Euro Area: Pre-UMP↑↑↑; Post-UMP↓; OMT↓ - Japan: Pre-UMP↓↓↓↓↓↓; Post-UMP; CME ### Foreign Responses: Bond Yields, Equity, FX, and Money Market Rates (Figures 13–16) - Sources: Datastream, Bloomberg, and staff calculations. - Figure 13. Foreign Bond Yields Responses: cumulative bps scales and region labels shown. U.S., U.K., EA, JP announcement series and program labels (LSAP1A, LSAP1B, LSAP2, TWIST, LSAP3, APP1, APP2, FLS, OMT, CME) are plotted against vertical ranges including: - Scales include: -60, -50, -40, -30, -20, -10, 0, 10, 20, 30, 40 and -30, -20, -10, 0, 10, 20, 30, 40, 50, 60 in separate panels. - Figure 14. Foreign Equity Prices Response: cumulative percent scales and region labels shown. Scales include -5 through 5 and 0.0 through 1.4 in various panels; program labels LSAP1A, LSAP1B, LSAP2, TWIST, LSAP3, APP1, APP2, FLS, OMT, CME are used in respective announcement panels. - Figure 15. Relevant Cross Foreign Exchange Rates Response: cumulative percent scales and region labels shown. Scales include -5 through 3, -3.5 through 1.5, 0.0 through 0.4, and 0.0 through 1.8 across panels; program labels LSAP1A, LSAP1B, LSAP2, TWIST, LSAP3, APP1, APP2, FLS, OMT, CME are used. - Note: A rise (fall) in the cumulative 2-day window rate of return of the cross FX indicates a depreciation (appreciation) of the foreign currency vis-à-vis the relevant bilateral FX rate (e.g., USD, GBP, EUR, or JPY). - Figure 16. Foreign Money Market Rates Responses: cumulative bps scales and region labels shown. Scales include -60 through 10, -15 through 15, -20 through 40, and 0 through 3 across panels; program labels LSAP1A, LSAP1B, LSAP2, TWIST, LSAP3, APP1, APP2, FLS, OMT, CME are used. *Appendix II. Supplementary Figures and Tables to Section B — UNCONVENTIONAL MONETARY POLICIES—RECENT EXPERIENCE AND PROSPECTS—BACKGROUND PAPER — INTERNATIONAL MONETARY FUND* --- _Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/np/pp/eng/2013/_041813.pdf_