## wpiea2025048-print-pdf ## Source details **Canonical URL:** [wpiea2025048-print-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025048-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025048-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025048-print-pdf.pdf.json) --- ### Executive Summary — Background and contribution - Motivation: Apply High Frequency Identification (HFI) to a large developing economy, Brazil, focusing on daily market-based inflation expectations, the exchange rate, and sovereign risk premium (CDS). - Distinctive features: - Focus on market-based inflation expectations (high frequency), not surveys. - Use of data from a high-debt emerging economy (Brazil). - Contrast between “case study” OLS approaches and IV estimations exploiting heteroskedasticity; changes in interest rates around monetary policy meetings are not automatically treated as exogenous. - Relation to literature: Links to the “tight money paradox” literature (Sargent and Wallace (1981); Leeper (1991); Bianchi and Melosi (2019)). The paper finds no evidence of the tight-money paradox on average for Brazil in the sample and robustness subperiods. ### Institutional and fiscal context - Key institutional facts: - Real Plan stabilization in 1994; peg abandoned in 1999; inflation-targeting with floating exchange rate thereafter; formal central bank independence granted in 2019 (de facto earlier). - Long-term inflation expectations as of January 2025 were still 50 bps above the target. - Fiscal context: - Brazil characterized by one of the largest public debts among emerging economies, elevated deficits, and high real interest rates since the 1990s. - Simple estimates of the Bohn rule suggest passive fiscal policy is the norm. - Aggregate finding: Estimated coefficients indicate that a positive/negative interest rate surprise causes inflation expectations to decline/increase by a nontrivial amount. ### Data - Sample: all weeks between September 2009 and December 2024 (daily Wednesday-to-Thursday changes around Copom meetings). - Interest rate measure Δi: changes in inter-bank deposit rates (maturities: 30, 90, 180 and 360 days; also reported elsewhere as 1 month, 3 months, 6 months, 12 months). - Inflation expectations Δπe: derived from private market inflation-indexed instruments (benchmark 1 year ahead; also 2, 3 and 5 years). Measure equals expected return on nominal debt minus coupon of same-maturity inflation-indexed instruments (includes an inflation risk premium). - Alternative data: Central Bank weekly survey (FOCUS) used for robustness; weekly timing issues noted (participants update mostly on Fridays). - Other dependent variables: percentage change in BRL/USD exchange rate (ΔE: Wednesday-to-Thursday, positive implies depreciation), Wednesday-to-Thursday change in CDS risk premium (Δrisk). - Data sources: Sao Paulo Stock Exchange (BM&FBOVESPA), Bloomberg, Brazilian Financial and Capital Markets Association, Central Bank of Brazil (SGS). ### Identification strategy - Structural system: - Main estimating equation: Δπet = α + β Δit + ut. - Endogeneity: Δit = γ + δ Δπet + vt. If δ ≠ 0, OLS for β is inconsistent. - Identification through heteroskedasticity (Rigobon 2003; Rigobon and Sack 2004): - Partition sample into C (weeks with Copom meetings) and NC (weeks without Copom meetings). - Identifying assumptions: - (5) σvC > σvNC (variance of interest-rate shocks higher on Copom dates). - (6) σuC = σuNC (variance of inflation expectation shocks equal across partitions). - Estimators for β constructed from differences in variance-covariance matrices across partitions (ω12/ω11 or ω22/ω12). - Equivalent IV formulation: instrument zi constructed by normalizing Δi and flipping sign across partitions; first-stage (7) and reduced-form (8) expressions documented. - Tests: variance ratio tests (Rigobon methodology) compare variances in set C and set N for Δi, Δπe, ΔE, Δrisk. ### Tests of identifying assumptions (variance ratios) - Variance ratios (Ratio of variances C / N) with 99% CI: - Δi Maturity 12m: 2.33 [1.65; 3.42] - Δi Maturity 6m: 3.40 [2.42; 4.99] - Δi Maturity 3m: 4.53 [3.21; 6.65] - Δi Maturity 1m: 5.25 [3.73; 7.69] - Δπe 1 year: 1.23 [0.88; 1.81] - Δπe 2 years: 1.59 [1.13; 2.34] - Δπe 3 years: 1.33 [0.95; 1.96] - Δπe 5 years: 1.11 [0.79; 1.63] - ΔE: 0.92 [0.65; 1.35] - Δrisk: 0.85 [0.60; 1.24] - Interpretation: - Variances for Δi differ markedly across subsamples (supports condition (5)). - For Δπe, Δt, and ΔE we generally cannot reject equal variances (supports condition (6)), except for the 2-year inflation expectations. - Conclusion: identifying conditions hold in most cases and IV via heteroskedasticity should identify causal effects. ### OLS (naive and event-study) results - Correlation patterns: - Full-sample scatter: positive association between Δi and Δπe (naive interpretation could suggest higher rates increase inflation expectations). - Copom-only (case study) scatter: slope is negative but not statistically significant. - Table 3 OLS estimates (Δπe1year on Δi at maturities; coefficients with Newey-West standard errors): - Full Sample: - 1 month: β 0.16* (se) (0.07) - 3 months: β 0.25*** (se) (0.05) - 6 months: β 0.24*** (se) (0.03) - 12 months: β 0.21*** (se) (0.02) - Copom Sample: - 1 month: β -0.13 (se) (0.11) - 3 months: β -0.09 (se) (0.09) - 6 months: β -0.02 (se) (0.06) - 12 months: β 0.01 (se) (0.05) - Interpretation: - Naive full-sample OLS coefficients are positive and sometimes statistically significant; without correcting for endogeneity, this could be misinterpreted as evidence of the tight money paradox. - Copom-only OLS coefficients are negative or near-zero and not statistically significant; an event-study stopping here would suggest no clear impact. ### IV (heteroskedasticity-based) results and main findings - All IV estimates in Table 4 are negative with small standard-errors; majority of p-values < 1%. - Positive interest rate surprises at different maturities are associated with lower inflation expectations at all horizons. - Key IV coefficient estimates (Dependent variable = Δπ^e; N = 768): - Δi (1 month): -0.27*** (se 0.07) for 1 year; -0.64*** (se 0.07) for 2 years; -0.69*** (se 0.08) for 3 years; -0.67*** (se 0.07) for 5 years. - Δi (3 months): -0.26*** (se 0.07) for 1 year; -0.60*** (se 0.06) for 2 years; -0.69*** (se 0.07) for 3 years; -0.69*** (se 0.07) for 5 years. - Δi (6 months): -0.21*** (se 0.05) for 1 year; -0.45*** (se 0.05) for 2 years; -0.53*** (se 0.06) for 3 years; -0.67*** (se 0.08) for 5 years. - Δi (12 months): -0.20*** (se 0.06) for 1 year; -0.43*** (se 0.07) for 2 years; -0.50*** (se 0.07) for 3 years; -0.53*** (se 0.06) for 5 years. - Magnitude interpretation: a 100 basis point surprise in the interest rate leads to a reduction of between 0.2 to 0.3 percentage points in the 1-year ahead measure of inflation expectations. - Overall conclusion: IV uncovers a strong and consistent textbook-like impact—monetary tightenings lower inflation expectations on average in the 2009–2024 sample. ### Effect on exchange rate and credit risk (IV results) - Theoretical concern: higher interest rates could cause currency depreciation via higher default probability. - Empirical IV evidence (Table 5) does not support "higher rates → depreciation via increased risk" for Brazil. - Exchange rate (ΔE) IV estimates (Δß interest rate measure): - Δß (1 month): -5.64*** (se 0.91) - Δß (3 months): -5.10*** (se 0.75) - Δß (6 months): -3.93*** (se 0.63) - Δß (12 months): -3.42*** (se 0.69) - Interpretation: positive interest rate surprises cause an appreciation of the Real against the U.S. dollar. - ΔCDS (5-year CDS) IV estimates: - (1 month) 0.01 (se 0.06) - (3 months) -0.03 (se 0.05) - (6 months) -0.05 (se 0.04) - (12 months) -0.11* (se 0.05) - Conclusion: IV evidence shows no material increase in CDS following tightenings in the full 2009:2024 sample. ### Robustness tests — summary of key checks and results - Robustness check 1 — Survey-based inflation expectations (Table 6; Interest rate measure = Δß): - Δπ^e_survey (median of all participants): -0.60*** (se 0.17) for 1 month; -0.54*** (se 0.13) for 3 months; -0.43*** (se 0.11) for 6 months; -0.36*** (se 0.12) for 12 months. - Δπ^e_survey_top5 (top-5 forecasters): -0.25*** (se 0.10) for 1 month; -0.20*** (se 0.08) for 3 months; -0.13** (se 0.06) for 6 months; -0.09 (se 0.07) for 12 months. - Interpretation: survey-based estimates broadly confirm market-based findings. - Robustness check 2 — Excluding extreme swings in expected inflation (sample reduced to 689 observations; Table 7 highlights): - 3 months: Δπ^e 1y -0.07* (se 0.04); Δπ^e 2y -0.47*** (se 0.05); Δπ^e 3y -0.59*** (se 0.04); Δπ^e 5y -0.61*** (se 0.06); ΔE -2.00** (se 0.69). - 12 months: Δπ^e 1y -0.12*** (se 0.03); Δπ^e 2y -0.39*** (se 0.05); Δπ^e 3y -0.46*** (se 0.04); Δπ^e 5y -0.46*** (se 0.05); ΔE -0.94* (se 0.48). - Robustness check 3 — Moving-window sub-samples (six 10-year windows; 3-month interest rate): - Δπ^e 1y coefficients: 2009:2019 -0.25*** (se 0.06); 2010:2020 -0.13*** (se 0.05); 2011:2021 -0.17** (se 0.08); 2012:2022 -0.18** (se 0.09); 2013:2023 -0.21** (se 0.10); 2014:2024 -0.19* (se 0.11). - Δπ^e 5y coefficients: -0.61*** (se 0.06); -0.56*** (se 0.06); -0.79*** (se 0.09); -0.81*** (se 0.10); -0.85*** (se 0.11); -0.91*** (se 0.13). - ΔE/E coefficients: -2.85*** (se 0.73); -3.67*** (se 0.73); -5.50*** (se 0.97); -5.90*** (se 1.06); -7.70*** (se 1.25); -8.79*** (se 1.40). - Interpretation: out of 18 entries in Table 8, only one is borderline significant; evidence suggests monetary policy effectiveness has increased over time judging by these rows. - Robustness check 4 — Dropping Copom dates that coincide with FOMC (32 coincident meetings): - C-subsample with 90 data points; NC unchanged; N = 736 for each column in Table 9. - Table 9 IV regression highlights: - Δi (3 month): Δπ^e 1 year -0.35*** (se 0.07); 2 years -0.82*** (se 0.08); 3 years -0.89*** (se 0.08); 5 years -0.82*** (se 0.08); FX -5.90*** (se 0.95). - Δi (12 months): Δπ^e 1 year -0.27.*** (se 0.08); 2 years -0.65*** (se 0.11); 3 years -0.72*** (se 0.12); 5 years -0.71*** (se 0.11); FX -4.98*** (se 1.03). - Conclusion: main message remains in this smaller sample. ### Final remarks and key takeaways - Literature gaps: - HFI literature is too US-centric; emerging economy features (shallow credit markets, lack of fiscal credibility, risk premia) can impair transmission. - Inflation expectations have been largely neglected despite their importance. - Contribution: uses Brazilian data to assess the impact of monetary surprises on inflation expectations, bridging the two gaps. - Principal conclusions: - HFI can be successfully applied to an emerging, high-debt economy (Brazil) to identify causal effects on high-frequency inflation expectations, exchange rates, and CDS. - After correcting for endogeneity via heteroskedasticity-based IV, monetary tightenings lower inflation expectations on average in the 2009–2024 sample. - No average evidence in 2009–2024 that monetary tightenings increase sovereign risk premia or depreciate the exchange rate via higher probability of default. - Naive OLS and event-study approaches can give misleading inferences; addressing endogeneity is essential. *Source: wpiea2025048-print-pdf - 3.2 Effect of monetary surprises on inflation expectations using IV* ### Executive Summary ### Executive Summary ### Background and contribution - Motivation: High Frequency Identification (HFI) studies have improved understanding of monetary policy shocks, but HFI has rarely been deployed for Emerging Economies; this paper applies HFI to a large developing economy, Brazil. - Three distinctive features of this paper: - Focus on the impact of monetary policy on daily market-based inflation expectations (not survey data); the key variables— inflation expectations, the exchange rate, and sovereign risk premium (CDS)—are high frequency. - Use of data from a high-debt emerging economy (Brazil) rather than the usual U.S.-centric literature. - Changes in interest rates around monetary policy meetings are not automatically treated as exogenous shocks; the paper contrasts “case study” OLS approaches with IV estimations exploiting heteroskedasticity and documents different results. - Relation to literature: Connects to the “tight money paradox” literature (Sargent and Wallace (1981); Leeper (1991); Bianchi and Melosi (2019)), which emphasizes fiscal behavior in determining whether higher interest rates raise or lower inflation expectations. The paper finds no evidence of the tight-money paradox on average for Brazil in the sample and robustness subperiods. ### Institutional and fiscal context - Brazil’s institutional history noted: stabilization in 1994 (Real Plan), abandonment of the peg in 1999, inflation-targeting with a floating exchange rate thereafter; formal central bank independence granted in 2019 (de facto earlier). - Credibility indicator: long-term inflation expectations as of January 2025 were still 50 bps above the target. - Fiscal context: Brazil has one of the largest public debts among emerging economies, elevated deficits, and high real interest rates since the 1990s; simple estimates of the Bohn rule suggest passive fiscal policy is the norm. - Despite this background, estimated coefficients indicate that a positive/negative interest rate surprise causes inflation expectations to decline/increase by a nontrivial amount. ### Data - Sample: all weeks between September 2009 and December 2024 (daily Wednesday-to-Thursday changes around Copom meetings). - Interest rate measure Δi: changes in inter-bank deposit rates (maturities: 30, 90, 180 and 360 days; also reported elsewhere as 1 month, 3 months, 6 months, 12 months). - Inflation expectations Δπe: derived from private market inflation-indexed instruments (benchmark 1 year ahead; also 2, 3 and 5 years). Measure equals expected return on nominal debt minus coupon of same-maturity inflation-indexed instruments (note: includes an inflation risk premium). - Alternative: Central Bank weekly survey (FOCUS) used for robustness; weekly timing issues noted (participants update mostly on Fridays). - Other dependent variables: percentage change in BRL/USD exchange rate (ΔE: Wednesday-to-Thursday, positive implies depreciation), Wednesday-to-Thursday change in CDS risk premium (Δrisk). - Data sources referenced: Sao Paulo Stock Exchange (BM&FBOVESPA), Bloomberg, Brazilian Financial and Capital Markets Association, Central Bank of Brazil (SGS). ### Identification strategy - Main estimating equation: Δπet = α + β Δit + ut, with potential reverse causality and omitted variable bias; system includes Δit = γ + δ Δπet + vt. - Problem: if δ ≠ 0, OLS for β is inconsistent; naive OLS may be biased (formula for dβOLS in text). - Identification through heteroskedasticity (Rigobon 2003; Rigobon and Sack 2004): - Partition sample into C (weeks with Copom meetings) and NC (weeks without Copom meetings). - Identifying assumptions: - (5) σvC > σvNC (variance of interest-rate shocks higher on Copom dates). - (6) σuC = σuNC (variance of inflation expectation shocks equal across partitions). - Under these assumptions, difference in variance-covariance matrices across partitions yields estimators for β (ω12/ω11 or ω22/ω12). - Equivalent IV formulation constructed using normalized Δi and constructed instrument zi (sign flip across partitions); first-stage expression (7) and reduced-form (8) documented. - Tests of identifying assumptions: variance ratio tests (Rigobon methodology) compare variances in set C and set N for Δi, Δπe, ΔE, Δrisk. ### Tests of identification assumptions (variance ratios) - Variance ratios (Ratio of variances C / N) and 99% CI (as reported): - Δi Maturity 12m: 2.33 [1.65; 3.42] - Δi Maturity 6m: 3.40 [2.42; 4.99] - Δi Maturity 3m: 4.53 [3.21; 6.65] - Δi Maturity 1m: 5.25 [3.73; 7.69] - Δπe 1 year: 1.23 [0.88; 1.81] - Δπe 2 years: 1.59 [1.13; 2.34] - Δπe 3 years: 1.33 [0.95; 1.96] - Δπe 5 years: 1.11 [0.79; 1.63] - ΔE: 0.92 [0.65; 1.35] - Δrisk: 0.85 [0.60; 1.24] - Interpretation: variances for Δi differ markedly across subsamples (supporting condition (5)); for Δπe, Δt, and ΔE we cannot reject equal variances (supporting condition (6)) except for the 2-year inflation expectations. Thus the identifying conditions hold in most cases and IV via heteroskedasticity should identify causal effects. ### OLS (naive and event-study) results - Correlation patterns: - Full sample scatter suggests a positive association between Δi and Δπe (naive interpretation could suggest higher rates increase inflation expectations). - Copom-only (case study) scatter: slope is negative but not statistically significant. - Table 3 OLS estimates (Δπe1year on Δi measured at different maturities); coefficients and Newey-West standard errors: - Full Sample: - 1 month: β 0.16* (se) (0.07) - 3 months: β 0.25*** (se) (0.05) - 6 months: β 0.24*** (se) (0.03) - 12 months: β 0.21*** (se) (0.02) - Copom Sample: - 1 month: β -0.13 (se) (0.11) - 3 months: β -0.09 (se) (0.09) - 6 months: β -0.02 (se) (0.06) - 12 months: β 0.01 (se) (0.05) - Interpretation: - Naive full-sample OLS coefficients are positive and in some cases statistically significant, which—if uncorrected for endogeneity—could be (mis)interpreted as evidence of the tight money paradox. - Copom-only OLS coefficients are negative or near-zero and not statistically significant; an event-study stopping at this regression would suggest no clear impact of monetary policy on inflation expectations. ### IV (identification through heteroskedasticity) results and main findings - Using IV exploiting heteroskedasticity uncovers a strong and consistent textbook-like impact of monetary policy: - Estimated coefficients indicate that positive interest-rate surprises cause inflation expectations to decline, and negative surprises cause inflation expectations to increase (described qualitatively in the Executive Summary: "a positive/negative interest rate surprise causes inflation expectations to decline/increase by a nontrivial amount"). - All coefficients estimated using ID through heteroskedasticity are reported as highly statistically significant (textual statement). - Exchange rate and risk premium: - The analysis for 2009:2024 as a whole does not support the conjecture that tighter monetary policy increases the probability of default or leads to exchange rate depreciation via higher risk premia. The naive OLS yields a positive association between tightenings and risk premia, but IV evidence does not support that mechanism in the full sample. - The 2003–2008 period is an exception in other work (Goncalves and Guimaraes (2011) documented opposite results for that earlier period). ### Robustness and scope - The paper emphasizes that it estimates average effects; monetary policy could have been inefficient at specific moments within the sample even if the average effect is that monetary policy works. - Robustness tests using different sample periods corroborate the main finding that monetary policy works. - Additional robustness checks and extensions are documented in later sections (listed in the Contents): IV through heteroskedasticity estimates, exchange rate and risk IV estimates, survey measures of πe, trimming extreme inflation changes, moving window IV regressions, dropping FOMC dates, etc. ### Key takeaways - HFI can be successfully applied to an emerging, high-debt economy (Brazil) to identify the causal effect of monetary policy shocks on high-frequency inflation expectations, exchange rates, and CDS. - After correcting for endogeneity via heteroskedasticity-based IV, monetary tightenings lower inflation expectations on average in the 2009–2024 sample. - No average evidence in 2009–2024 that monetary tightenings increase sovereign risk premia or depreciate the exchange rate via higher probability of default. - Naive OLS and event-study approaches can give misleading inferences—case-study OLS and IV estimates differ meaningfully—highlighting the importance of addressing endogeneity. *Source: wpiea2025048-print-pdf - Executive Summary* ### 3.2 Effect of monetary surprises on inflation expectations using IV ### 3.2 Effect of monetary surprises on inflation expectations using IV ### Main IV estimates and interpretation - All IV estimates in Table 4 are negative and display small standard-errors; majority of p-values are smaller than 1%. - Positive interest rate surprises at different maturities are associated with lower inflation expectations at all horizons. - Key coefficient estimates from Table 4 (Dependent variable = Δπ^e): - Δi (1 month): -0.27*** (se 0.07) for 1 year; -0.64*** (se 0.07) for 2 years; -0.69*** (se 0.08) for 3 years; -0.67*** (se 0.07) for 5 years. N = 768. - Δi (3 months): -0.26*** (se 0.07) for 1 year; -0.60*** (se 0.06) for 2 years; -0.69*** (se 0.07) for 3 years; -0.69*** (se 0.07) for 5 years. N = 768. - Δi (6 months): -0.21*** (se 0.05) for 1 year; -0.45*** (se 0.05) for 2 years; -0.53*** (se 0.06) for 3 years; -0.67*** (se 0.08) for 5 years. N = 768. - Δi (12 months): -0.20*** (se 0.06) for 1 year; -0.43*** (se 0.07) for 2 years; -0.50*** (se 0.07) for 3 years; -0.53*** (se 0.06) for 5 years. N = 768. - Magnitude interpretation: using Table 4, a 100 basis point surprise in the interest rate leads to a reduction of between 0.2 to 0.3 percentage points in the 1-year ahead measure of inflation expectations. ### 3.3 Effect of monetary surprises on CDS and the exchange rate ### Findings on exchange rate and credit risk - Theoretical concern: an increase in interest rates could lead to currency depreciation via higher probability of default. - Empirical IV evidence does not support "higher rates → depreciation via increased risk" for Brazil. - Positive interest rate surprises cause an appreciation of the Brazilian currency (the Real) against the U.S. dollar and do not seem to affect the CDS materially. - Key IV estimates from Table 5 (Interest rate measure = Δß): - ΔE (exchange rate): Δß (1 month) -5.64*** (se 0.91); (3 months) -5.10*** (se 0.75); (6 months) -3.93*** (se 0.63); (12 months) -3.42*** (se 0.69). - ΔCDS: (1 month) 0.01 (se 0.06); (3 months) -0.03 (se 0.05); (6 months) -0.05 (se 0.04); (12 months) -0.11* (se 0.05). - Notes: coefficients reported use the 5-year CDS. Results with changes in 1-year CDS are available upon request. ### 3.4 Robustness tests ### Robustness check 1 — Survey-based inflation expectations - Replace market measures with Central Bank survey expectations (weekly frequency). - Table 6 results (Interest rate measure = Δß): - Δπ^e_survey (median of all participants): -0.60*** (se 0.17) for 1 month; -0.54*** (se 0.13) for 3 months; -0.43*** (se 0.11) for 6 months; -0.36*** (se 0.12) for 12 months. - Δπ^e_survey_top5 (top-5 forecasters): -0.25*** (se 0.10) for 1 month; -0.20*** (se 0.08) for 3 months; -0.13** (se 0.06) for 6 months; -0.09 (se 0.07) for 12 months. - Reassuring result: survey-based estimates using Brazilian data broadly confirm the market-based findings; out of 8 entries in Table 6, only one is not statistically significant. Coefficients are larger in the row using the median of all respondents. ### Robustness check 2 — Excluding extreme swings in expected inflation - Exclude all Δπ above the percentile 0.95 and below percentile 0.05, reducing sample to 689 observations. - Table 7 key estimates (Δi used; reporting selected entries): - 3 months: Δπ^e 1y -0.07* (se 0.04); Δπ^e 2y -0.47*** (se 0.05); Δπ^e 3y -0.59*** (se 0.04); Δπ^e 5y -0.61*** (se 0.06); ΔE -2.00** (se 0.69). - 12 months: Δπ^e 1y -0.12*** (se 0.03); Δπ^e 2y -0.39*** (se 0.05); Δπ^e 3y -0.46*** (se 0.04); Δπ^e 5y -0.46*** (se 0.05); ΔE -0.94* (se 0.48). - Number of observations is now 689. ### Robustness check 3 — Moving-window sub-samples - Six different 10-year moving windows with start dates in 2009, 2010, 2011, 2012, 2013 and 2014; use 3-month interest rate in all specifications. - Table 8 highlights: - Δπ^e 1y coefficients: 2009:2019 -0.25*** (se 0.06); 2010:2020 -0.13*** (se 0.05); 2011:2021 -0.17** (se 0.08); 2012:2022 -0.18** (se 0.09); 2013:2023 -0.21** (se 0.10); 2014:2024 -0.19* (se 0.11). - Δπ^e 5y coefficients: -0.61*** (se 0.06); -0.56*** (se 0.06); -0.79*** (se 0.09); -0.81*** (se 0.10); -0.85*** (se 0.11); -0.91*** (se 0.13). - ΔE/E coefficients: -2.85*** (se 0.73); -3.67*** (se 0.73); -5.50*** (se 0.97); -5.90*** (se 1.06); -7.70*** (se 1.25); -8.79*** (se 1.40). - Out of 18 entries in Table 8, only one is borderline significant. Over time monetary policy effectiveness appears to have increased judging by the second and third rows. ### Robustness check 4 — Dropping Copom dates that coincide with FOMC - From 2009 to end-2024, 32 Copom meetings coincide with FOMC meetings. - Dropping those common dates produces a C-subsample with 90 data points; the NC-subsample is unchanged. N reported in Table 9 is 736 for each column. - Table 9 key IV regression results (Δπ^e and ΔE/E against Δi): - Δi (3 month): Δπ^e 1 year -0.35*** (se 0.07); 2 years -0.82*** (se 0.08); 3 years -0.89*** (se 0.08); 5 years -0.82*** (se 0.08); FX -5.90*** (se 0.95). - Δi (12 months): Δπ^e 1 year -0.27.*** (se 0.08); 2 years -0.65*** (se 0.11); 3 years -0.72*** (se 0.12); 5 years -0.71*** (se 0.11); FX -4.98*** (se 1.03). - The main message of the paper remains in this smaller sample. ### 4 Final remarks - Literature gaps identified: - High-frequency identification (HFI) literature is too US-centric; emerging economy characteristics (shallow credit markets, lack of fiscal credibility, risk premia) can impair monetary transmission. - Inflation expectations have been largely neglected despite being a crucial determinant of actual inflation. - Contribution: this paper uses data from an emerging economy (Brazil) to assess the impact of monetary surprises on inflation expectations, bridging the two gaps noted above. - On the FTPL and Sargent and Wallace (1981) logic: the paper does not find systematic evidence that monetary policy backfires when fiscal policy is active and debt is elevated using Brazilian data. *Source: wpiea2025048-print-pdf - 3.2 Effect of monetary surprises on inflation expectations using IV* --- _Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025048-print-pdf.pdf_