## Macro Effects of Formal Adoption of Inflation Targeting — Section 1–4 (wpiea2023007-print-pdf)

## Source details

**Canonical URL:** [Macro Effects of Formal Adoption of Inflation Targeting — Section 1–4 (wpiea2023007-print-pdf)](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023007-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023007-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023007-print-pdf.pdf.json)

---

### Abstract — core findings
- Study examines impact of formal adoption of inflation targeting (IT) on inflation, growth and anchoring of inflation expectations in advanced economies and emerging markets and developing economies (EMDEs).
- Early adopters of IT (pre-2000) all saw declines in inflation rates following adoption; IT adopters since then enjoyed such success in only about half the cases.
- On average there is not much difference between IT and non-IT countries in mean inflation, inflation volatility and the extent of inflation anchoring.
- Cannot rule out that apparent success of IT may be due to "regression to the mean".
- Country-level analysis using the Synthetic Control Method (SCM) shows IT adoption delivers significant inflation gains in about a third of the cases.
- Limited support that adoption of IT systematically leads to poorer growth outcomes.
- Concluding implication: belief that IT adoption will be sufficient to keep inflation in check cannot be taken for granted.

### Introduction — background and competing views
- New Zealand adopted IT in 1989; its annual inflation rate averaged nearly 12 percent in the three years prior and about 3 percent in the three years after adoption.
- Four other advanced economies—Canada, UK, Australia and Sweden—followed early; next six adopters in the 1990s included emerging markets such as Brazil, Colombia and Poland.
- Figure reported: average inflation rates in the 3 years following IT adoption dropped in all eleven early adopters compared to the three years prior.
- Proponents’ cited benefits (as summarized in the paper):
  - Lower inflation.
  - Stable (more anchored) inflation.
  - Better growth performance.
  - Lower sacrifice ratio.
  - Wider benefits: more transparent and coherent policymaking, increased accountability, greater attention to long-run considerations.
- Skeptical views (as summarized in the paper):
  - Global structural factors and globalization may account for pervasive declines in inflation.
  - Regression to the mean: countries with unusually high inflation tend to see declines regardless of IT adoption.
  - Concern that IT may harm growth or reduce attention to real outcomes; some evidence IT resulted in lower output growth in developing economies.

### Data and empirical approach
- Panel comprises a total of 190 countries: 24 classified as advanced economies (AEs) and the remainder as emerging markets and developing economies (EMDEs).
- Annual data on inflation and GDP from the World Bank (Ha et al., 2019) and IMF’s International Financial Statistics.
- Inflation forecasts from Consensus Forecasts (IMF WEO forecasts give similar results).
- Methods used:
  - Panel data framework of Levin et al. (2004) and Choi et al. (2018) to test whether inflation surprises lead to changes in inflation expectations (anchoring).
  - Panel test following Ball and Sheridan to examine regression to the mean by including pre-IT inflation levels.
  - Country-by-country comparison using the Synthetic Control Method (SCM) to compare IT adopters to a synthetic cohort that shared pre-adoption profiles.
- Mean inflation by decade (Table 1):
  - Period 1990-99: All 5.3, AE 3.0, EMDEs 5.9
  - Period 2000-09: All 4.7, AE 2.3, EMDEs 5.2
  - Period 2010-19: All 3.4, AE 1.3, EMDEs 3.7
- Aggregate trends:
  - Median inflation in advanced economies trended down for the last three decades after the 1970s–early-1980s run-up.
  - Median inflation in EMDEs remained high through the 1990s but trended down thereafter (with a flare-up during the Global Financial Crisis).
  - The gap between median inflation in AEs and EMDEs, which increased to 10 percent by the 1990s, has since closed substantially.

### Synthesis of Section 1 results and interpretation
- Little difference on average between IT and non-IT countries in mean inflation, inflation volatility, or extent of inflation anchoring.
- Inflation surprises have little effect on inflation expectations in both IT and non-IT countries; for countries with annual inflation rates exceeding 20 percent, modest evidence that non-IT countries have less inflation anchoring than IT countries.
- SCM country-level analysis: IT adoption yields significant inflation gains in a subset of countries (about a third) but not in the majority.
- Regression to the mean remains a powerful explanation for apparent effects of IT across both advanced economies and EMDEs.
- Growth impacts: panel analysis finds little average difference between IT and non-IT countries; SCM shows modest evidence that larger inflation declines following IT correspond to larger output declines in those countries.
- Overall conclusion: formal adoption of IT is neither necessary nor sufficient for attaining low inflation outcomes; causal identification is challenging because the multidecade decline in inflation coincided with widespread formal adoption of IT and because persistent low inflation itself can anchor expectations without a formal IT framework.

### Inflation Targeting Adoption and Immediate Effects (Section 2 highlights)
- Excluding the ECB and its constituent countries there are a total of 33 countries identified as full-fledged inflation targeters, of which 9 are advanced economies; Kazakhstan and Argentina dropped from analysis due to data issues.
- First test: compare mean inflation rates in the three years prior to and the three years after IT adoption:
  - All 11 adopters in the first decade (1989-1999) saw post-adoption declines in mean inflation rates.
  - Only half of the 22 subsequent adopters saw post-adoption declines.
- Strong relationship between initial inflation at time of IT adoption and the post-IT decline in inflation: countries with high initial inflation rates saw bigger post-adoption declines.
- Structural-break analysis (Zivot-Andrews Test) in a five-year window around adoption:
  - Using full 1960-2019 sample: 4 of the 11 early adopters had breaks close to year of adoption, compared with 3 of the 22 later adopters.
  - Restricting sample to post-1990: 6 early adopters and 6 later adopters had breaks close to adoption year.
- Selected country data (Table 2 rows preserved):
  - New Zealand 1989 11.7 3.2 ↓ 1988 2012
  - Canada 1991 4.6 1.2 ↓ 1983 1995
  - UK 1992 7.1 2.4 ↓ 1981 2014
  - Australia 1993 3.8 3.1 ↓ 1991 2000
  - Colombia 1999 19.3 7.9 ↓ 1999 1999
  - United States 2012 1.5 1.1 ↓ 1982 2009
  - Japan 2013 -0.3 1.1 1981 1999
  - India 2016 7.7 5.0 ↓ 1979 2008
- Zivot-Andrews Test used to identify year for structural breaks; full sample 1960–2019.

### IT vs. Non-IT Countries: Level and Volatility (preserve values)
- Table 3: Average Inflation in IT and Non-IT Country Groups
  - Advanced Economies
    - 1990-99: Number of Countries IT 5 Non-IT 22; Mean Inflation IT 2.0 Non-IT 3.1; Median Inflation IT 2.3 Non-IT 2.7
    - 2000-09: Number of Countries IT 7 Non-IT 19; Mean Inflation IT 2.8 Non-IT 2.1; Median Inflation IT 2.2 Non-IT 2.2
    - 2010-19: Number of Countries IT 7 Non-IT 19; Mean Inflation IT 1.8 Non-IT 1.2; Median Inflation IT 1.4 Non-IT 1.5
  - Emerging Markets & LDCs
    - 1990-99: Number of Countries IT 6 Non-IT 123; Mean Inflation IT 7.2 Non-IT 5.4; Median Inflation IT 5.1 Non-IT 8.4
    - 2000-09: Number of Countries IT 19 Non-IT 14; Mean Inflation IT 4.7 Non-IT 5.0; Median Inflation IT 5.0 Non-IT 4.7
    - 2010-19: Number of Countries IT 22 Non-IT 13; Mean Inflation IT 4.0 Non-IT 3.5; Median Inflation IT 3.1 Non-IT 3.1
- Table 4: Average Inflation Volatility (standard deviation)
  - Advanced Economies
    - 1990-99: IT 1.5 Non-IT 2.3
    - 2000-09: IT 2.2 Non-IT 1.4
    - 2010-19: IT 1.1 Non-IT 1.2
  - Emerging Markets & LDCs
    - 1990-99: IT 3.2 Non-IT 4.4
    - 2000-09: IT 3.0 Non-IT 4.1
    - 2010-19: IT 2.7 Non-IT 3.4
- Summary:
  - Striking similarity in the pattern of decline in average inflation rates between IT and non-IT countries for both groups.
  - Over 2000-19, average inflation rates have been quite similar across regimes; if anything, inflation has been lower in the non-IT regime on average.
  - Inflation volatility declined in both regimes over time; IT countries have lower inflation volatility (Table 4).

### Inflation Anchoring (preserve tables and estimates)
- Table 5: Mean Inflation Expectations (3-, 5-, 10-years ahead)
  - Advanced Economies
    - 1990-99 IT: 3-yrs 2.5 5-yrs 2.4 10-yrs 2.4; Non-IT: 3-yrs 2.5 5-yrs 2.5 10-yrs 2.5
    - 2000-09 IT: 3-yrs 2.0 5-yrs 2.0 10-yrs 1.9; Non-IT: 3-yrs 1.9 5-yrs 1.9 10-yrs 1.9
    - 2010-19 IT: 3-yrs 1.9 5-yrs 1.9 10-yrs 1.8; Non-IT: 3-yrs 1.8 5-yrs 1.8 10-yrs 1.7
  - Emerging Markets
    - 1990-99 IT: 3-yrs 6.0 5-yrs 5.5 10-yrs 5.2
    - 2000-09 IT: 3-yrs 3.7 5-yrs 3.5 10-yrs 3.5; Non-IT: 3-yrs 3.4 5-yrs 3.3 10-yrs 3.3
    - 2010-19 IT: 3-yrs 3.7 5-yrs 3.6 10-yrs 3.5; Non-IT: 3-yrs 3.0 5-yrs 2.9 10-yrs 2.9
- Table 6: Standard Deviation of Inflation Expectations (3-, 5-, 10-years ahead)
  - Advanced Economies
    - 1990-99 IT: 3-yrs 2.5 5-yrs 2.4 10-yrs 2.4; Non-IT: 3-yrs 2.5 5-yrs 2.5 10-yrs 2.5
    - 2000-09 IT: 3-yrs 2.0 5-yrs 2.0 10-yrs 1.9; Non-IT: 3-yrs 1.9 5-yrs 1.9 10-yrs 1.9
    - 2010-19 IT: 3-yrs 1.9 5-yrs 1.9 10-yrs 1.8; Non-IT: 3-yrs 1.8 5-yrs 1.8 10-yrs 1.8
  - Emerging Markets
    - 1990-99 IT: 3-yrs 6.0 5-yrs 5.5 10-yrs 5.2
    - 2000-09 IT: 3-yrs 3.7 5-yrs 3.5 10-yrs 3.5; Non-IT: 3-yrs 3.4 5-yrs 3.3 10-yrs 3.3
    - 2010-19 IT: 3-yrs 3.7 5-yrs 3.6 10-yrs 3.5; Non-IT: 3-yrs 3.0 5-yrs 2.9 10-yrs 2.9
- Formal anchoring test (Levin et al., 2004; Choi et al., 2018):
  - ∆π^e_t+n = α + θ_j + β_1 Inflation Surprise + β_2 Median Inflation + ε_t
  - Inflation Surprise defined as deviation from 3-period moving average (inflation in period t minus 3-period moving average at t-1).
- Table 7: Determinants of 3-Year Ahead Inflation Expectations (estimates preserved)
  - Advanced Economies
    - IT: Inflation Surprise 0.00 (0.01); Median CPI Inflation 0.054 (0.03); Constant -0.12 (0.05); R-Square 0.033
    - Non-IT: Inflation Surprise 0.00 (0.00); Median CPI Inflation 0.12*** (0.02); Constant -0.33*** (0.04); R-Square 0.089
  - Emerging Economies
    - IT: Inflation Surprise 0.11 (0.07); Median CPI Inflation 0.05 (0.04); Constant -0.33 (0.14); R-Square 0.13
    - Non-IT: Inflation Surprise 0.00 (0.00); Median CPI Inflation 0.06* (0.02); Constant -0.35* (0.11); R-Square 0.02
- Main finding: inflation surprises have no impact on inflation expectations in either the IT or the non-IT regime for the 3-year horizon.
- High Inflation Economies (at least one year of inflation above 20 percent) — Table 8 estimates preserved:
  - 3-Year Ahead Inflation Expectations
    - IT: Inflation Surprise 0.03 (0.02); Median CPI Inflation 0.03 (0.02); Constant -0.08 (0.06); R-Square 0.01
    - Non-IT: Inflation Surprise 0.11 (0.07); Median CPI Inflation 0.02 (0.43); Constant 1.26 (1.68); R-Square 0.00
  - 5-Year Ahead Inflation Expectations
    - IT: Inflation Surprise 0.01 (0.02); Median CPI Inflation 0.03* (0.01); Constant -0.15** (0.04); R-Square 0.01
    - Non-IT: Inflation Surprise 0.10 (0.07); Median CPI Inflation -0.42 (0.65); Constant 4.03 (2.55); R-Square 0.00
- Interpretation: modest evidence that inflation expectations are more anchored in IT adopters in high-inflation EMs; coefficient estimates larger in non-IT countries than in IT countries, though not significantly different from zero.

### Inflation and Growth in IT vs. Non-IT Countries: Cross-Country Results (preserve estimates)
- Differences-in-differences approach following Ball and Sheridan:
  - Regression: X_post − X_pre = α_o + α_1 D + α_2 X_pre + ε
  - Inclusion of X_pre controls for regression to the mean.
- Replication and annual vs quarterly data (Table 9):
  - Ball & Sheridan (quarterly) column (1): IT Dummy -2.19* (0.88); Constant -1.77** (0.52); R-Square 0.21; N 20
  - Our Computation (annual) column (2): IT Dummy -2.32* (1.03); Constant -1.79** (0.58); R-Square 0.23; N 20
  - Ball & Sheridan with X_pre (quarterly) column (3): IT Dummy -0.55 (0.35); Inflation (Pre) -0.74*** (0.08); Constant 1.12*** (0.32); R-Square 0.90
  - Our Computation with X_pre (annual) column (4): IT Dummy -1.14** (0.30); Inflation (Pre) -0.77*** (0.04); Constant 1.44*** (0.24); R-Square 0.93
- Extended sample through 2019 and EMDEs (Table 10):
  - Advanced Economies
    - Column (1) IT Dummy -1.02 (1.30); Constant -3.04** (0.97); R-Square 0.03; N 23
    - Column (2) with X_pre IT Dummy 0.00 (0.27); Inflation (Pre) -0.87*** (0.04); Constant 1.19*** (0.23); R-Square 0.97; N 23
  - Emerging Markets & Developing Countries
    - Column (3) without X_pre IT Dummy -4.75*** (1.28); Constant -1.74*** (0.45); R-Square 0.12; N 110
    - Column (4) with X_pre IT Dummy -0.78 (0.79); Inflation (Pre) -1.01*** (0.05); Constant 2.76*** (0.41); R-Square 0.77; N 110
- Interpretation: For EMDEs, the large estimated impact of IT without X_pre (a -4.75 percentage point reduction) falls to -0.78 when controlling for X_pre and is no longer statistically different from zero — regression to the mean explains apparent success.
- Regression-to-the-mean illustrated: higher initial inflation rate associated with greater subsequent decline in inflation (Figure 4 described).

### Growth Impacts (preserve estimates)
- Median growth shows little difference between IT and non-IT groups for both advanced and emerging economies (Figure 5 described).
- Table 11: Impact of IT on Growth
  - Advanced Economies
    - Column (1) IT Dummy 0.79 (0.61); Constant -0.86* (0.33); R-Square 0.08; N 23
    - Column (2) with Lagged Growth IT Dummy 0.49 (0.41); Lagged Growth -0.80** (0.22); Constant 1.44*** (0.52); R-Square 0.51; N 23
  - Emerging Markets
    - Column (3) IT Dummy -0.48 (0.91); Constant 0.17 (0.61); R-Square 0.00; N 110
    - Column (4) with Lagged Growth IT Dummy -0.01 (0.43); Lagged Growth -1.01*** (0.05); Constant 4.09*** (0.31); R-Square 0.86; N 110
- Conclusion: IT dummy is not significantly different from zero for growth outcomes; scant evidence that IT affects growth.

### Identification, selection bias, and robustness
- Ball & Sheridan specification yields unbiased estimator of treatment effect only under strong assumption that structural inflation dynamics are the same across targeters and non-targeters.
- Evidence of selection bias in adoption of IT noted in literature; Heckman two-stage procedure used to correct for selection bias in related work.
- Authors include robustness checks: annual vs quarterly data, extended sample to 2019, exclusion of high inflation economies.

### Country-Level Analysis: Synthetic Control Method (SCM) — examples and results (Section 3)
- SCM compares a treatment country with a synthetic cohort (weighted composite of non-adopters) matched on pre-treatment outcomes.
- Illustrative cases:
  - Mexico: pre-treatment gyrations similar; after adoption of IT, Mexico had noticeably lower inflation than its counterpart.
  - Poland: pre-treatment match close; after adoption of IT, Poland’s inflation is much lower than its synthetic counterpart.
  - United Kingdom and Israel: adoption of IT does not seem to have affected the inflation outcome; UK tracks synthetic; Israel shows more volatility but no clear improvement.
- Table 12 — selected country-level SCM results (Gap and P values preserved):
  - Mexico: Inflation RMSPE 7.84; Gap -6.84; T-test P value 0.00. Growth RMSPE 3.23; Gap -1.30; T-test P value 0.08.
  - Poland: Inflation RMSPE 2.28; Gap -3.40; T-test P value 0.00. Growth RMSPE 0.78; Gap 0.52; T-test P value 0.67.
  - Israel: Inflation RMSPE 3.35; Gap 0.20; T-test P value 0.74. Growth RMSPE 2.83; Gap -3.56; T-test P value 0.42.
  - United Kingdom: Inflation RMSPE 1.08; Gap -1.54; T-test P value 0.68. Growth RMSPE 2.00; Gap 0.08; T-test P value 0.83.
  - Indonesia: Inflation RMSPE 14.19; Gap 2.93; T-test P value 0.02. Growth RMSPE 5.40; Gap 0.57; T-test P value 0.00.
  - Ghana: Inflation RMSPE 5.96; Gap 2.99; T-test P value 0.23. Growth RMSPE 0.89; Gap 4.67; T-test P value 0.01.
  - Peru: Inflation RMSPE 1.52; Gap 0.59; T-test P value 0.25. Growth RMSPE 2.35; Gap 2.40; T-test P value 0.02.
  - Philippines: Inflation RMSPE 2.49; Gap 0.51; T-test P value 0.78. Growth RMSPE 2.14; Gap 0.72; T-test P value 0.00.
- Notes on Table 12:
  - "Gap is defined as actual - synthetic inflation/growth averaged over 5 years post IT years."
  - "For countries that adopted IT in early to mid-2010s we have a fairly small post-IT sample size."
  - "The P values are for the t-test to check for any statistical difference between the synthetic & the actual for the entire post IT period."
  - "We perform the two-side t-test here with the null hypothesis that there is no difference between IT and its synthetic counterpart."

### Aggregate and subsample SCM T-tests (5-year post IT)
- Aggregate two-sided T-test P values (actual vs. synthetic, 5-year post-IT):
  - 5-year post IT (21 Countries): 0.24 (inflation), 0.71 (growth).
  - Advanced Economies: 0.07 (inflation), 0.24 (growth).
  - Emerging Markets: 0.77 (inflation), 0.21 (growth).

### Summary of country-level findings
- Inflation:
  - In 8 of the 23 cases, IT-adoption is associated with lower average inflation compared with the synthetic cohort.
  - In 3 cases (Colombia, Mexico, Poland), the difference is statistically significant (using a cut-off of p ¡ 0.05).
  - In the remaining 15 cases, there is either no difference or average inflation is higher in the IT adopter than in the synthetic cohort.
- Growth:
  - In 11 of the 23 cases, IT-adoption is associated with higher average real GDP growth compared with the synthetic cohort.
  - Significantly higher growth in 4 cases (Ghana, Indonesia, Peru, Philippines).
  - In the other 12 cases, there is either no difference or average growth is lower in the IT adopter than in the synthetic cohort.
- Overall interpretation: IT-adoption appears neither necessary nor sufficient for better macroeconomic outcomes.

### Visualization and correlation evidence
- Figure 7 described:
  - Panel A (Inflation): inflation outcomes are lower in IT-adopters in only a small number of cases.
  - Panel B (Growth): growth outcomes are split about evenly; growth is just as likely to be lower in the IT-adopter as higher.
- Figure 8 described:
  - Shows a modest positive correlation: IT-adopting countries with better inflation outcomes than their synthetic cohorts also tended to have worse growth outcomes than their synthetic cohorts.
  - This correlation is noted as a "glimmer of evidence" for concerns that a single-minded focus on inflation may come at the expense of growth.

### Methodological and interpretive notes
- SCM used for country-level counterfactuals; two-sided T-tests compare means of actual and synthetic outcomes over the 5-year post-adoption period under the null of no difference.
- SCM results complement panel methods; authors note potential publication biases in the empirical IT literature and discuss selection issues in adoption.
- Robustness checks include alternative data frequencies, extended sample, and exclusion of high-inflation economies.

### Conclusions and policy implications (section conclusions summarized)
- Main findings:
  1. Though early IT-adopters saw inflation declines post adoption, only half of the 22 subsequent adopters saw post-adoption inflation declines.
  2. No difference between IT-adopters and other countries in average level and volatility of inflation; no difference in expected inflation and no difference in anchoring of inflation between the two groups.
  3. Regression to the mean remains a plausible explanation for assumed benefits of IT-adoption, applicable to EMDEs and advanced economies.
  4. Country-level SCM comparison finds little evidence that adoption improves macroeconomic performance.
- Interpretation and implications:
  - Central banks and IFIs should critically evaluate claims about the benefits of formal IT adoption given potential groupthink and publication biases.
  - Formal adoption of IT is neither necessary nor sufficient for beneficial inflation and growth outcomes; focus should shift to understanding why some countries achieved better outcomes and what practices can be transferred.
  - Results do not constitute a full cost-benefit analysis; potential advantages of IT not considered here and adherence to IT can lead to policy mistakes if inflation objectives crowd out other goals.
  - Alternative explanations for the great moderation in inflation, such as demographic changes and globalization, deserve serious consideration.

*Source: IMF Working Paper WP/23/7 — Sections 1–4 (wpiea2023007-print-pdf)*

### Section 1

### Macro Effects of Formal Adoption of Inflation Targeting — Section 1

### Abstract — core findings
- Study examines impact of formal adoption of inflation targeting (IT) on inflation, growth and anchoring of inflation expectations in advanced economies and emerging markets and developing economies (EMDEs).
- Early adopters of IT (pre-2000) all saw declines in inflation rates following adoption; IT adopters since then enjoyed such success in only about half the cases.
- On average there is not much difference between IT and non-IT countries in mean inflation, inflation volatility and the extent of inflation anchoring.
- Cannot rule out that apparent success of IT may be due to "regression to the mean".
- Country-level analysis using the Synthetic Control Method (SCM) shows IT adoption delivers significant inflation gains in about a third of the cases.
- Limited support that adoption of IT systematically leads to poorer growth outcomes.
- Concluding implication: belief that IT adoption will be sufficient to keep inflation in check cannot be taken for granted.

### Introduction — background and competing views
- New Zealand adopted IT in 1989; its annual inflation rate averaged nearly 12 percent in the three years prior and about 3 percent in the three years after adoption.
- Four other advanced economies—Canada, UK, Australia and Sweden—followed early; next six adopters in the 1990s included emerging markets such as Brazil, Colombia and Poland.
- Figure reported: average inflation rates in the 3 years following IT adoption dropped in all eleven early adopters compared to the three years prior.

- Proponents’ cited benefits:
  - Lower inflation (Bernanke et al., 2000; Mishkin and Schmidt-Hebbel, 2002; Mishkin, 2002).
  - Stable (more anchored) inflation (King, 2002).
  - Better growth performance (Bernanke et al., 2000).
  - Lower sacrifice ratio (Gonçalves and Carvalho, 2009; Huang et al., 2019).
  - Wider benefits: more transparent and coherent policymaking, increased accountability, greater attention to long-run considerations (Bernanke and Mishkin, 1997; Bernanke et al., 2000).

- Skeptical views:
  - Global structural factors and globalization may account for pervasive declines in inflation, reducing the share of credit attributable to IT (Rogoff, 2003; Forbes, 2019).
  - Regression to the mean: countries with unusually high inflation tend to see declines regardless of IT adoption (Ball and Sheridan, 2004).
  - Concern that IT may harm growth or reduce attention to real outcomes (Meyer, 2002; Rivlin, 2002; Blanchard, 2003; Friedman, 2003); some evidence IT resulted in lower output growth in developing economies (Brito and Bystedt, 2010).

### Data and empirical approach
- Panel comprises a total of 190 countries: 24 classified as advanced economies (AEs) and the remainder as emerging markets and developing economies (EMDEs).
- Annual data on inflation and GDP from the World Bank (Ha et al., 2019) and IMF’s International Financial Statistics.
- Inflation forecasts from Consensus Forecasts (IMF WEO forecasts give similar results).
- Methods used:
  - Panel data framework of Levin et al. (2004) and Choi et al. (2018) to test whether inflation surprises lead to changes in inflation expectations (anchoring).
  - Panel test following Ball and Sheridan to examine regression to the mean by including pre-IT inflation levels.
  - Country-by-country comparison using the Synthetic Control Method (SCM) to compare IT adopters to a synthetic cohort that shared pre-adoption profiles.

- Mean inflation by decade (Table 1):
  - Period 1990-99: All 5.3, AE 3.0, EMDEs 5.9
  - Period 2000-09: All 4.7, AE 2.3, EMDEs 5.2
  - Period 2010-19: All 3.4, AE 1.3, EMDEs 3.7

- Aggregate trends:
  - Median inflation in advanced economies trended down for the last three decades after the 1970s–early-1980s run-up.
  - Median inflation in EMDEs remained high through the 1990s but trended down thereafter (with a flare-up during the Global Financial Crisis).
  - The gap between median inflation in AEs and EMDEs, which increased to 10 percent by the 1990s, has since closed substantially.

### Synthesis of results and interpretation (from Section 1)
- Little difference on average between IT and non-IT countries in mean inflation, inflation volatility, or extent of inflation anchoring.
- Inflation surprises have little effect on inflation expectations in both IT and non-IT countries; for countries with annual inflation rates exceeding 20 percent, modest evidence that non-IT countries have less inflation anchoring than IT countries.
- SCM country-level analysis: IT adoption yields significant inflation gains in a subset of countries (about a third) but not in the majority.
- Regression to the mean remains a powerful explanation for apparent effects of IT across both advanced economies and EMDEs.
- Growth impacts: panel analysis finds little average difference between IT and non-IT countries; SCM shows modest evidence that larger inflation declines following IT correspond to larger output declines in those countries—providing limited support for the concern that inflation gains may come at the expense of output.
- Overall conclusion for Section 1: formal adoption of IT is neither necessary nor sufficient for attaining low inflation outcomes; causal identification is challenging because the multidecade decline in inflation coincided with widespread formal adoption of IT and because persistent low inflation itself can anchor expectations without a formal IT framework.

*Source: IMF Working Paper WP/23/7 — Section 1*

### Section 2

### Section 2

### Inflation Targeting Adoption and Immediate Effects
- Excluding the ECB and its constituent countries there are a total of 33 countries identified as full-fledged inflation targeters, of which 9 are advanced economies. Issues with the data lead to dropping Kazakhstan and Argentina from the analysis.
- First test: compare mean inflation rates in the three years prior to and the three years after IT adoption.
  - All 11 adopters in the first decade (1989-1999) saw post-adoption declines in mean inflation rates.
  - Only half of the 22 subsequent adopters saw post-adoption declines.
- Strong relationship between initial inflation at time of IT adoption and the post-IT decline in inflation: countries with high initial inflation rates saw bigger post-adoption declines.
- Structural-break analysis (Zivot-Andrews Test) in a five-year window around adoption:
  - Using full 1960-2019 sample: 4 of the 11 early adopters had breaks close to year of adoption, compared with 3 of the 22 later adopters.
  - Restricting sample to post-1990: 6 early adopters and 6 later adopters had breaks close to adoption year.
- Table 2 (selected rows, preserved values):
  - 1 New Zealand 1989 11.7 3.2 ↓ 1988 2012
  - 2 Canada 1991 4.6 1.2 ↓ 1983 1995
  - 3 UK 1992 7.1 2.4 ↓ 1981 2014
  - 4 Australia 1993 3.8 3.1 ↓ 1991 2000
  - 11 Colombia 1999 19.3 7.9 ↓ 1999 1999
  - 27 United States 2012 1.5 1.1 ↓ 1982 2009
  - 28 Japan 2013 -0.3 1.1 1981 1999
  - 33 India 2016 7.7 5.0 ↓ 1979 2008
- Note: Zivot Andrews Test used to identify year for structural breaks, looking for breaks in both trend and intercept. Full sample 1960–2019.

### IT vs. Non-IT Countries: Level and Volatility
- Table 3: Average Inflation in IT and Non-IT Country Groups (preserve values)
  - Advanced Economies
    - 1990-99: Number of Countries IT 5 Non-IT 22; Mean Inflation IT 2.0 Non-IT 3.1; Median Inflation IT 2.3 Non-IT 2.7
    - 2000-09: Number of Countries IT 7 Non-IT 19; Mean Inflation IT 2.8 Non-IT 2.1; Median Inflation IT 2.2 Non-IT 2.2
    - 2010-19: Number of Countries IT 7 Non-IT 19; Mean Inflation IT 1.8 Non-IT 1.2; Median Inflation IT 1.4 Non-IT 1.5
  - Emerging Markets & LDCs
    - 1990-99: Number of Countries IT 6 Non-IT 123; Mean Inflation IT 7.2 Non-IT 5.4; Median Inflation IT 5.1 Non-IT 8.4
    - 2000-09: Number of Countries IT 19 Non-IT 14; Mean Inflation IT 4.7 Non-IT 5.0; Median Inflation IT 5.0 Non-IT 4.7
    - 2010-19: Number of Countries IT 22 Non-IT 13; Mean Inflation IT 4.0 Non-IT 3.5; Median Inflation IT 3.1 Non-IT 3.1
- Table 4: Average Inflation Volatility (standard deviation)
  - Advanced Economies
    - 1990-99: IT 1.5 Non-IT 2.3
    - 2000-09: IT 2.2 Non-IT 1.4
    - 2010-19: IT 1.1 Non-IT 1.2
  - Emerging Markets & LDCs
    - 1990-99: IT 3.2 Non-IT 4.4
    - 2000-09: IT 3.0 Non-IT 4.1
    - 2010-19: IT 2.7 Non-IT 3.4
- Summary findings:
  - Striking similarity in the pattern of decline in average inflation rates between IT and non-IT countries for both groups.
  - Over 2000-19, average inflation rates have been quite similar across regimes; if anything, inflation has been lower in the non-IT regime on average.
  - Inflation volatility declined in both regimes over time, but IT countries have lower inflation volatility (Table 4).

### Inflation Anchoring
- Table 5: Mean Inflation Expectations (3-, 5-, 10-years ahead) (preserve values)
  - Advanced Economies
    - 1990-99 IT: 3-yrs 2.5 5-yrs 2.4 10-yrs 2.4; Non-IT: 3-yrs 2.5 5-yrs 2.5 10-yrs 2.5
    - 2000-09 IT: 3-yrs 2.0 5-yrs 2.0 10-yrs 1.9; Non-IT: 3-yrs 1.9 5-yrs 1.9 10-yrs 1.9
    - 2010-19 IT: 3-yrs 1.9 5-yrs 1.9 10-yrs 1.8; Non-IT: 3-yrs 1.8 5-yrs 1.8 10-yrs 1.7
  - Emerging Markets
    - 1990-99 IT: 3-yrs 6.0 5-yrs 5.5 10-yrs 5.2
    - 2000-09 IT: 3-yrs 3.7 5-yrs 3.5 10-yrs 3.5; Non-IT: 3-yrs 3.4 5-yrs 3.3 10-yrs 3.3
    - 2010-19 IT: 3-yrs 3.7 5-yrs 3.6 10-yrs 3.5; Non-IT: 3-yrs 3.0 5-yrs 2.9 10-yrs 2.9
- Table 6: Standard Deviation of Inflation Expectations (3-, 5-, 10-years ahead) (preserve values)
  - Advanced Economies
    - 1990-99 IT: 3-yrs 2.5 5-yrs 2.4 10-yrs 2.4; Non-IT: 3-yrs 2.5 5-yrs 2.5 10-yrs 2.5
    - 2000-09 IT: 3-yrs 2.0 5-yrs 2.0 10-yrs 1.9; Non-IT: 3-yrs 1.9 5-yrs 1.9 10-yrs 1.9
    - 2010-19 IT: 3-yrs 1.9 5-yrs 1.9 10-yrs 1.8; Non-IT: 3-yrs 1.8 5-yrs 1.8 10-yrs 1.8
  - Emerging Markets
    - 1990-99 IT: 3-yrs 6.0 5-yrs 5.5 10-yrs 5.2
    - 2000-09 IT: 3-yrs 3.7 5-yrs 3.5 10-yrs 3.5; Non-IT: 3-yrs 3.4 5-yrs 3.3 10-yrs 3.3
    - 2010-19 IT: 3-yrs 3.7 5-yrs 3.6 10-yrs 3.5; Non-IT: 3-yrs 3.0 5-yrs 2.9 10-yrs 2.9
- Preliminary interpretation:
  - For advanced economies, limited material difference in mean inflation expectations between inflation targeters and non-targeters.
  - Inflation expectations (short-, medium-, long-run) moderated substantially over last three decades for advanced and emerging economies.
  - Non-inflation targeters have had somewhat lower average inflation expectations in the last decade compared to inflation targeters.
  - Standard deviation (anchoring proxy): non-inflation targeters have lower standard deviation in inflation expectations for advanced economies; for emerging markets non-inflation targeters exhibit slightly higher standard deviation than IT peers.
- Formal anchoring test (Levin et al., 2004; Choi et al., 2018): estimate
  - ∆π^e_t+n = α + θ_j + β_1 Inflation Surprise + β_2 Median Inflation + ε_t
  - Inflation Surprise defined as deviation from 3-period moving average (inflation in period t minus 3-period moving average at t-1).
- Table 7: Determinants of 3-Year Ahead Inflation Expectations (preserve estimates and significance)
  - Advanced Economies
    - IT: Inflation Surprise 0.00 (0.01); Median CPI Inflation 0.054 (0.03); Constant -0.12 (0.05); R-Square 0.033
    - Non-IT: Inflation Surprise 0.00 (0.00); Median CPI Inflation 0.12*** (0.02); Constant -0.33*** (0.04); R-Square 0.089
  - Emerging Economies
    - IT: Inflation Surprise 0.11 (0.07); Median CPI Inflation 0.05 (0.04); Constant -0.33 (0.14); R-Square 0.13
    - Non-IT: Inflation Surprise 0.00 (0.00); Median CPI Inflation 0.06* (0.02); Constant -0.35* (0.11); R-Square 0.02
- Main finding: inflation surprises have no impact on inflation expectations in either the IT or the non-IT regime for the 3-year horizon.
- High Inflation Economies (at least one year of inflation above 20 percent):
  - Table 8 results (preserve estimates)
    - 3-Year Ahead Inflation Expectations
      - IT: Inflation Surprise 0.03 (0.02); Median CPI Inflation 0.03 (0.02); Constant -0.08 (0.06); R-Square 0.01
      - Non-IT: Inflation Surprise 0.11 (0.07); Median CPI Inflation 0.02 (0.43); Constant 1.26 (1.68); R-Square 0.00
    - 5-Year Ahead Inflation Expectations
      - IT: Inflation Surprise 0.01 (0.02); Median CPI Inflation 0.03* (0.01); Constant -0.15** (0.04); R-Square 0.01
      - Non-IT: Inflation Surprise 0.10 (0.07); Median CPI Inflation -0.42 (0.65); Constant 4.03 (2.55); R-Square 0.00
  - Modest evidence that inflation expectations are more anchored in IT adopters in high-inflation EMs; coefficient estimates larger in non-IT countries than in IT countries, though not significantly different from zero.

### Inflation and Growth in IT vs. Non-IT Countries: Cross-Country Results
- Differences-in-differences approach following Ball and Sheridan:
  - Regression: X_post − X_pre = α_o + α_1 D + α_2 X_pre + ε
  - Inclusion of X_pre controls for regression to the mean.
- Replication and annual vs quarterly data:
  - Table 9 (preserve estimates):
    - Ball & Sheridan (quarterly) column (1): IT Dummy -2.19* (0.88); Constant -1.77** (0.52); R-Square 0.21; N 20
    - Our Computation (annual) column (2): IT Dummy -2.32* (1.03); Constant -1.79** (0.58); R-Square 0.23; N 20
    - Ball & Sheridan with X_pre (quarterly) column (3): IT Dummy -0.55 (0.35); Inflation (Pre) -0.74*** (0.08); Constant 1.12*** (0.32); R-Square 0.90
    - Our Computation with X_pre (annual) column (4): IT Dummy -1.14** (0.30); Inflation (Pre) -0.77*** (0.04); Constant 1.44*** (0.24); R-Square 0.93
  - Conclusion: use of annual data does not materially change BS results; inclusion of X_pre sharply attenuates estimated IT impact.
- Extended sample through 2019 and EMDEs (Table 10, preserve estimates):
  - Advanced Economies
    - Column (1) IT Dummy -1.02 (1.30); Constant -3.04** (0.97); R-Square 0.03; N 23
    - Column (2) with X_pre IT Dummy 0.00 (0.27); Inflation (Pre) -0.87*** (0.04); Constant 1.19*** (0.23); R-Square 0.97; N 23
  - Emerging Markets & Developing Countries
    - Column (3) without X_pre IT Dummy -4.75*** (1.28); Constant -1.74*** (0.45); R-Square 0.12; N 110
    - Column (4) with X_pre IT Dummy -0.78 (0.79); Inflation (Pre) -1.01*** (0.05); Constant 2.76*** (0.41); R-Square 0.77; N 110
  - Interpretation: For EMDEs, the large estimated impact of IT without X_pre (a -4.75 percentage point reduction) falls to -0.78 when controlling for X_pre and is no longer statistically different from zero — regression to the mean explains apparent success.
- Regression-to-the-mean illustrated: higher initial inflation rate associated with greater subsequent decline in inflation (Figure 4).

### Growth Impacts
- Median growth shows little difference between IT and non-IT groups for both advanced and emerging economies (Figure 5).
- Table 11: Impact of IT on Growth (preserve estimates)
  - Advanced Economies
    - Column (1) IT Dummy 0.79 (0.61); Constant -0.86* (0.33); R-Square 0.08; N 23
    - Column (2) with Lagged Growth IT Dummy 0.49 (0.41); Lagged Growth -0.80** (0.22); Constant 1.44*** (0.52); R-Square 0.51; N 23
  - Emerging Markets
    - Column (3) IT Dummy -0.48 (0.91); Constant 0.17 (0.61); R-Square 0.00; N 110
    - Column (4) with Lagged Growth IT Dummy -0.01 (0.43); Lagged Growth -1.01*** (0.05); Constant 4.09*** (0.31); R-Square 0.86; N 110
- Conclusion: IT dummy is not significantly different from zero for growth outcomes; scant evidence that IT affects growth.

### Identification, Selection Bias, and Robustness
- Ball & Sheridan specification yields unbiased estimator of treatment effect only under strong assumption that structural inflation dynamics are the same across targeters and non-targeters (Geraats, 2013).
- Evidence of selection bias in adoption of IT noted in literature (Gonçalves and Carvalho, 2009); Heckman two-stage procedure used to correct for selection bias in related work.
- Authors include robustness checks: annual vs quarterly data, extended sample to 2019, exclusion of high inflation economies.

### Country-Level Analysis: Synthetic Control Method (SCM)
- SCM compares a treatment country with a synthetic cohort (weighted composite of non-adopters) matched on pre-treatment outcomes.
- SCM used to supplement panel evidence by estimating country-specific impacts of IT adoption.
- Illustration: Figure 6 examines four cases — two where IT appears to have delivered good outcomes and two where it did not make a difference.

*Italic: Source — wpiea2023007-print-pdf - Section 2 (IMF working paper content provided)*

### Section 3

### wpiea2023007-print-pdf - Section 3

### Synthetic Control Examples and Visual Evidence
- Mexico vs. synthetic cohort (Panel A, left-hand side): pre-treatment gyrations are similar; after adoption of IT, Mexico had noticeably lower inflation than its counterpart.
- Poland vs. synthetic cohort (Panel A, right-hand side): pre-treatment inflation evolution in treatment and control groups is remarkably close; after adoption of IT, Poland’s inflation is much lower than that of its synthetic counterpart.
- United Kingdom and Israel (Panel B): adoption of IT does not seem to have affected the inflation outcome. In the UK, inflation evolution essentially tracks the synthetic cohort over the whole period. In Israel, inflation is more volatile than in the synthetic control but shows no clear IT-related improvement.

### Table 12 — Country-level SCM Results (Inflation and Growth)
- Table 12 reports for each IT-adopting country:
  - RMSPE between actual and synthetic inflation in the pre-adoption period (smaller indicates better match; in most cases RMSPE is below 3).
  - Gap: actual minus synthetic average inflation in the five years post-IT adoption.
  - P values from a two-sided T-test of the difference in means (null: no difference between actual and synthetic).
  - Analogous RMSPE, gap, and p-values for growth.
- Selected exact results from Table 12 (Gap and P values preserved as presented):
  - Mexico: Inflation RMSPE 7.84; Gap -6.84; T-test P value 0.00. Growth RMSPE 3.23; Gap -1.30; T-test P value 0.08.
  - Poland: Inflation RMSPE 2.28; Gap -3.40; T-test P value 0.00. Growth RMSPE 0.78; Gap 0.52; T-test P value 0.67.
  - Israel: Inflation RMSPE 3.35; Gap 0.20; T-test P value 0.74. Growth RMSPE 2.83; Gap -3.56; T-test P value 0.42.
  - United Kingdom: Inflation RMSPE 1.08; Gap -1.54; T-test P value 0.68. Growth RMSPE 2.00; Gap 0.08; T-test P value 0.83.
  - Indonesia: Inflation RMSPE 14.19; Gap 2.93; T-test P value 0.02. Growth RMSPE 5.40; Gap 0.57; T-test P value 0.00.
  - Ghana: Inflation RMSPE 5.96; Gap 2.99; T-test P value 0.23. Growth RMSPE 0.89; Gap 4.67; T-test P value 0.01.
  - Peru: Inflation RMSPE 1.52; Gap 0.59; T-test P value 0.25. Growth RMSPE 2.35; Gap 2.40; T-test P value 0.02.
  - Philippines: Inflation RMSPE 2.49; Gap 0.51; T-test P value 0.78. Growth RMSPE 2.14; Gap 0.72; T-test P value 0.00.
  - Notes as presented:
    - "Gap is defined as actual - synthetic inflation/growth averaged over 5 years post IT years."
    - "For countries that adopted IT in early to mid-2010s we have a fairly small post-IT sample size."
    - "The P values are for the t-test to check for any statistical difference between the synthetic & the actual for the entire post IT period."
    - "We perform the two-side t-test here with the null hypothesis that there is no difference between IT and its synthetic counterpart."

### Aggregate and Subsample T-tests (5-year post IT)
- Aggregate two-sided T-test results for 5-year post-IT period (actual vs. synthetic) presented as P values:
  - 5-year post IT (21 Countries): 0.24 (inflation), 0.71 (growth).
  - Advanced Economies: 0.07 (inflation), 0.24 (growth).
  - Emerging Markets: 0.77 (inflation), 0.21 (growth).

### Summary of Country-level Findings (as reported)
- Inflation:
  - In 8 of the 23 cases, IT-adoption is associated with lower average inflation compared with the synthetic cohort.
  - In 3 cases (Colombia, Mexico, Poland), the difference is statistically significant (using a cut-off of p ¡ 0.05).
  - In the remaining 15 cases, there is either no difference or average inflation is higher in the IT adopter than in the synthetic cohort.
- Growth:
  - In 11 of the 23 cases, IT-adoption is associated with higher average real GDP growth compared with the synthetic cohort.
  - Significantly higher growth in 4 cases (Ghana, Indonesia, Peru, Philippines).
  - In the other 12 cases, there is either no difference or average growth is lower in the IT adopter than in the synthetic cohort.
- Overall interpretation: IT-adoption appears neither necessary nor sufficient for better macroeconomic outcomes.

### Visualization and Correlation Evidence
- Figure 7 (described): shows gaps between IT-adopters and synthetic cohorts:
  - Panel A (Inflation): inflation outcomes are lower in IT-adopters in only a small number of cases.
  - Panel B (Growth): growth outcomes are split about evenly; growth is just as likely to be lower in the IT-adopter as higher.
- Figure 8 (described): shows correlation between inflation gap and output (growth) gap:
  - Reports a modest positive correlation: IT-adopting countries with better inflation outcomes than their synthetic cohorts also tended to have worse growth outcomes than their synthetic cohorts.
  - This correlation is noted as a "glimmer of evidence" for concerns that a single-minded focus on inflation may come at the expense of growth.

### Methodological and Interpretive Notes
- The authors use the Synthetic Control Method (SCM) for country-level counterfactuals and perform two-sided T-tests comparing means of actual and synthetic outcomes over the 5-year post-adoption period.
- T-tests are constructed under the null hypothesis of no difference between the two sample means.
- The SCM results are part of a broader analysis that also uses other methods earlier in the paper to provide a fuller picture of IT impacts.
- The authors cite concerns from meta-analyses that the empirical literature on IT may suffer from publication biases favoring beneficial effects of IT.

### Conclusions and Policy Implications (section conclusions summarized)
- Main findings:
  1. Though early IT-adopters saw inflation declines post adoption, only half of the 22 subsequent adopters saw post-adoption inflation declines (Section II).
  2. No difference between IT-adopters and other countries in average level and volatility of inflation; no difference in expected inflation and no difference in anchoring of inflation between the two groups (Section III).
  3. Regression to the mean remains a plausible explanation for assumed benefits of IT-adoption, applicable to EMDEs and advanced economies (Section IV).
  4. Country-level SCM comparison finds little evidence that adoption improves macroeconomic performance (Section V).
- Interpretation and implications:
  - Central banks and IFIs should critically evaluate claims about the benefits of formal IT adoption given potential groupthink and publication biases.
  - Formal adoption of IT is neither necessary nor sufficient for beneficial inflation and growth outcomes; focus should shift to understanding why some countries achieved better outcomes and what practices can be transferred.
  - Results do not constitute a full cost-benefit analysis; potential advantages of IT not considered here and adherence to IT can lead to policy mistakes if inflation objectives crowd out other goals.
  - Alternative explanations for the great moderation in inflation, such as demographic changes and globalization, deserve serious consideration.

*Source: wpiea2023007-print-pdf - Section 3*

### Section 4

### Section 4

### References cited in Section 4
- International Monetary Fund.
- Meyer, L. H. (2002).  Inflation targets and inflation targeting.The North American Journal of Economics and Finance, 13(2):147–162.
- Mishkin, F. S. (2002). Inflation targeting: Lessons from international experience.
- Mishkin, F. S. and Schmidt-Hebbel, K. (2002).  One decade of inflation targeting in the world: What do we know and what do we need to know. InInflation Targeting: Design, Performance, Challenges, Central Bank of. Citeseer.
- Rivlin, A. (2002). Comment on us monetary policy in the 1990s.American Economic Policy in the 1990s.
- Roger, M. S. (2009). Inflation targeting at 20-achievements and challenges.
- Rogoff,  K. (2003).   Globalization and global disinflation.Economic Review-Federal Reserve Bank of Kansas City, 88(4):45–80.
- Staff, I. M. F. (2011).IMF Performance in the Run-up to the Financial and Economic Crisis: IMF Surveillance in 2004-07. International Monetary Fund.
- Zivot, E. and Andrews, D. W. K. (2002).   Further evidence on the great crash, the oil-price shock, and the unit-root hypothesis.Journal of business & economic statistics, 20(1):25–44.

*Source: wpiea2023007-print-pdf - Section 4*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023007-print-pdf.pdf_