## wpiea2022199-print-pdf

## Source details

**Canonical URL:** [wpiea2022199-print-pdf](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022199-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022199-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022199-print-pdf.pdf.json)

---

### Introduction
- The paper investigates the potential revenue impact of different tax policy changes using the Tax Policy Reform Database (TPRD).
- Revenue responses depend on many factors, including the type of tax system, institutional capacity, state of the economy, and the nature of the tax policy change itself.
- Estimating revenue responses requires granular information on the nature of the change: tax instrument (e.g., VAT or personal income tax), type of change (e.g., tax base, tax rate), timing, and size.
- Baseline shocks are identified in the year when a tax change is announced.

### The Tax Policy Reform Database (TPRD v.4.0)
- Documents main features of tax policy changes for six tax types: personal income tax (PIT), corporate income tax (CIT), value-added and sale taxes (VAT), excise (EXE), social security contributions (SSC), and property tax (PRO).
- Features recorded include: type of change (base, rate changes), timing (exact announcement and implementation dates), significance (major or minor change), and size of rate changes when available.
- Coverage and sample:
  - Database documents 3,462 tax policy measures.
  - Average of 4.7 tax measures per country per year.
  - Covers 23 advanced and emerging market economies; empirical analysis constructed for 21 countries (China and India excluded due to data limitations).
  - Time coverage extended from 2014 to 2019 for most countries; overall empirical period 1990-2018.
- Composition of measures:
  - About 60 percent are tax base changes.
  - The majority of tax base changes consists of decreases in tax liabilities (about ⅔ of all tax base changes).
  - Rate changes are more balanced between rate increases and decreases.
  - More than 80 percent of all identified tax measures are defined as “major” tax changes or reforms in a single year, with a majority introduced as part of a package.
  - About 80 percent of tax rate measures in the database contain information on the size of the announced change.
- Implementation timing:
  - About 52 percent of all tax changes documented are announced and implemented within the same year.
  - 41 percent are implemented with a one-year lag.
  - The remaining 7 percent are implemented on a longer horizon.

### Constructing Tax Policy Shock Measures and Indicators
- Major tax reforms counted within a year (announcement year) to construct tax change indicator 푇푇퐼퐼 for each country.
- Coding procedure:
  - Each individual reform coded as 푎푎 ∈ {1, −1, 푖푖푖푖 퐼퐼퐼퐼퐼퐼}.
  - When multiple tax changes announced in the same year, aggregate to produce 푇푇퐼퐼 = ∑푎푎 for each reform type, tax type, country and announcement year.
- Aggregate TCI total row (aggregate across years and tax types) shown as:
  - Total -117 -130 -171 -63 5 36 -16 20 11 50 8 2 (corresponding to CIT base, CIT rate, PIT base, PIT rate, VAT base, VAT rate, SSC base, SSC rate, EXE base, EXE rate, PRO base, PRO rate).
- Correlations and distributions:
  - CIT base broadening correlated with CIT rate cuts (correlation of 0.35).
  - PIT base changes positively and significantly associated with same-direction rate changes.
  - PIT, CIT and VAT together account for the largest share in major policy changes and about 85 percent of both rate decrease and base narrowing.

### Macro Variables, Data Coverage, and Notes
- Macro variables sourced from the OECD: total tax revenues, unemployment, GDP, and government consumption expenditure.
- Analysis uses the ratio of tax revenues to GDP rather than nominal tax revenues.
- Total tax revenue data include social security contributions and trade related revenues; analysis covers domestic tax revenues but trade-related taxes could not be excluded due to data limitations.
- Sample focuses on 21 countries from TPRD v.4.0; China and India excluded due to lack of macro data for 1990-2018.

### Econometric Method
- Method: Jordà (2005)’s local projections method.
- Estimated regression (notation preserved):
  - 푦푦푖푖,푡푡+ℎ − 푦푦푖푖,푡푡−1 = 훼훼푖푖 + 훾훾푡푡 + 휆휆ℎ 푇푇퐼퐼푖푖,푡푡 + ∑훽훽푗푗ℎ Δ푦푦푖푖,푡푡−푗푗3 푗푗=1 + 휃휃ℎ′ 푿푿푖푖,푡푡 + 휀휀푖푖,푡푡+ℎ
  - Outcome: 푦푦푖푖,푡푡+ℎ − 푦푦푖푖,푡푡−1 (tax revenue change from 푡푡−1 to 푡푡+ℎ).
  - 푇푇퐼퐼푖푖,푡푡 is the tax policy change indicator.
  - 푿푿푖푖,푡푡 includes controls: GDP growth, unemployment and government expenditure.
  - Both country fixed effects 훼훼푖푖 and time effects 훾훾푡푡 included.
  - Same-tax contemporaneous base changes controlled for when estimating rate change effects, and vice versa.
- Estimation details:
  - Regression estimated separately for each tax type and by rate vs base changes.
  - Benchmark regression captures response from year 0 (announcement year) to year 5.
  - OLS estimation.
  - Impulse response functions reported with both 90 percent and 68 percent confidence intervals.
  - Standard errors computed using Driscoll-Kraay (1998) formula.

### Baseline Results and Key Empirical Findings
- General:
  - A one-step increase corresponds to one additional major rate hike or base broadening announced within a year.
  - Size and temporal profile of revenue yield vary across taxes and change types (rate vs base).
- Corporate Income Tax (CIT):
  - Announcements of CIT increases associated with a somewhat transitory rise in tax collection.
  - Tax revenues respond on impact to announcements of CIT base-broadening measures, with collection peaking one year after announcement.
  - Revenue response to CIT rate changes is more uncertain.
- Personal Income Tax (PIT):
  - Announcements of PIT increases associated with relatively large and persistent increases in the tax-to-GDP ratio.
  - Gains more significant and persistent after announcements of PIT base-broadening measures rather than PIT rate hikes.
- Value-Added Tax (VAT):
  - Announcements of VAT increases generally associated with significant improvements in tax collection.
  - VAT rate increases more likely to be followed by significant and protracted increases in the tax-to-GDP ratio than VAT base-broadening announcements.
  - Caveat: VAT base-broadening announcements are a small share (about ¼ of all documented VAT changes).
- Social Security Contributions (SSC):
  - Announcements of rises in SSC associated with sizable surges in the tax-to-GDP ratio.
  - Tax collection responds more significantly to announcements of SSC rate increases than to SSC base-broadening measures.
- Excises (EXE):
  - Announcements of increases in EXE generally followed by large and persistent increases in tax collection, especially for base-broadening measures.
  - Likely reflects expansion of products subject to excise to address negative externalities.

### Asymmetry, Margins, and Robustness
- Asymmetry:
  - Revenue impact is asymmetric: statistical significance changes depending on direction of tax changes (increases/decreases in tax liabilities).
  - Positive reactions to announcements of rate hikes or base broadening in VAT, EXE, or SSC; unlikely responses to rate cuts or base narrowing for same taxes.
  - Asymmetric response also observed for rate increases in PIT and PRO; symmetric responses observed for CIT rate or base changes and for PIT and PRO base changes.
- Intensive vs. Extensive margins:
  - Intensive-margin analysis restricts to 659 tax rate changes (excludes rate changes documented as ranges or in non–percentage-point units).
  - Table 3 means and sds (EM = extensive margin; IM = intensive margin) preserved:
    - CIT: EM mean -0.17; EM sd 0.67; IM mean -0.98; IM sd 5.43
    - PIT: EM mean -0.11; EM sd 0.70; IM mean -0.49; IM sd 4.39
    - VAT: EM mean 0.05; EM sd 0.50; IM mean 0.03; IM sd 3.43
    - SSC: EM mean 0.04; EM sd 0.44; IM mean 0.20; IM sd 3.80
    - EXE: EM mean 0.03; EM sd 0.24; IM mean 1.35; IM sd 14.65
    - PRO: EM mean 0.01; EM sd 0.18; IM mean 0.38; IM sd 5.50
  - Both IM and EM standardized to zero mean and unit standard deviation for each tax type.
  - Comparison findings:
    - Response of tax-to-GDP to CIT and SSC rate hikes and to some extent VAT based on IM cannot be distinguished from EM (confidence bands overlap).
    - For EXE and PRO, impact on revenues is higher on intensive than extensive margins.
    - For PIT, impact is mostly felt on the extensive margin.
- Robustness:
  - Results broadly hold when including minor tax policy changes and when identifying tax shocks using implementation as opposed to announcement years.
  - Including minor changes leads to somewhat smaller estimates of tax gains following announcements.
  - When shocks are dated by implementation years:
    - Tax-to-GDP ratio less responsive to changes in VAT and EXE (anticipation effects).
    - Tax collection appears less responsive to increases in SSC and PRO.
    - CIT and PIT implementation-year impacts suggest taxpayers may better understand impact of rate changes on liabilities compared to base changes.

### Methodological Caveats and Limitations
- Baseline results based on qualitative indicators that record presence of a tax change (extensive margins) but not magnitude; therefore baseline results do not provide estimates of tax multipliers nor of their relative differences across tax heads.
- TPRD provides some information on motivations for some tax changes but not consistently across time and countries; not used in identifying tax shocks.
- Econometric strategy does not control for compositional effects for a given type of change (number of +1 and -1 within a year); about 25% (A_A) and 20% (B_B) of multi-change aggregates lose information when reduced to three-state variables.
- Standard errors and confidence intervals:
  - Impulse responses reported with 90 percent and 68 percent confidence intervals.
  - Driscoll-Kraay standard errors used (robust to heteroscedasticity, autocorrelation and cross-country dependence).
- Alternative specification including minor tax policy changes as additional controls (TC_minor defined with step size ±0.5) does not change main results for major tax policy changes.

### Conclusions and Future Research Directions
- Revenue yields depend on tax type and whether the change is a rate or base measure; rate changes tend to be more transitory in PIT, CIT, EXE, and PRO announcements.
- VAT and SSC rate hikes show relatively larger effects on collections.
- Revenue impacts are often asymmetric by direction of policy change.
- Quantitative information (intensive margins) generally corroborates qualitative findings and improves estimate precision.
- Future research planned to:
  - Deepen analysis of asymmetric effects of tax changes.
  - Further exploit quantitative information in TPRD v.4.0 to precisely estimate revenue yields (e.g., controlling for revenue share of a particular tax type).
  - Advance the debate on macroeconomic effects of different tax changes (e.g., tax multipliers).

*IMF Working Papers — Cross-Country Evidence on the Revenue Impact of Tax Reforms (wpiea2022199-print-pdf).*

### Introduction ...........................................................................................................

### wpiea2022199-print-pdf - Introduction ...........................................................................................................

### Introduction
- The paper investigates the potential revenue impact of different tax policy changes using the Tax Policy Reform Database (TPRD).
- Revenue responses depend on many factors, including the type of tax system, institutional capacity, state of the economy, and the nature of the tax policy change itself.
- Estimating revenue responses requires granular information on the nature of the change: tax instrument (e.g., VAT or personal income tax), type of change (e.g., tax base, tax rate), timing, and size.

### The Tax Policy Reform Database (TPRD)
- The TPRD documents main features of tax policy changes for six tax types: personal income tax (PIT), corporate income tax (CIT), value-added and sale taxes (VAT), excise (EXE), social security contributions (SSC), and property tax (PRO).
- Features recorded include: type of change (base, rate changes), timing (exact announcement and implementation dates), significance (major or minor change), and size of rate changes when available.
- The database covers 23 advanced and emerging market economies over the last four decades.
- Countries included: Australia, Austria, Brazil, Canada, China, Czech Republic, Denmark, France, Greece, Germany, India, Ireland, Italy, Japan, Korea, Luxembourg, Mexico, Poland, Portugal, Spain, Turkey, United Kingdom and the United States.

### Constructing Tax Policy Shock Measures
- New measures of tax policy shocks are constructed for the period 1990-2018 for 21 countries.
- China and India are excluded from the empirical analysis due to data limitations.
- Both major and minor tax policy changes are considered; focus is on major shocks since these account for about 80 percent of all documented tax changes in the database.
- Baseline shocks are identified in the year when a tax change is announced.

### Econometric Method
- The impact of tax policy changes on tax revenues is assessed using Jordà (2005)’s local projections method.

### Baseline Results and Main Empirical Findings
- Revenue yield of tax policy changes varies significantly across tax types and types of changes.
- Tax rate changes generally have a more transitory revenue impact than tax base changes for most taxes.
- Base broadening changes in PIT, CIT, EXE, and PRO have on average a more significant and long-lasting impact on tax collection than rate changes.
- Rate hikes have relatively more significant effects on VAT and SSC measures.
- Results broadly hold when:
  - including minor tax policy changes in the analysis, and
  - identifying tax shocks using implementation as opposed to announcement years.
- The revenue impact is asymmetric: statistical significance changes depending on the direction of tax changes considered (increases/decreases in tax liabilities).
- Responsiveness of tax revenues to announcement of tax rate changes does not materially change when using qualitative information on tax shocks (extensive margins) versus using the size of announced rate changes (intensive margin).

### Caveats and Limitations
- Baseline results are based on qualitative indicators that record the presence of a tax change (extensive margins) but not its magnitude.
- Consequently, baseline results do not provide estimates of tax multipliers nor of their relative differences across tax heads.
- The TPRD provides some information on motivations for some tax changes (e.g., CIT changes aimed at promoting investment), but this information is not consistently available across time and countries and thus is not used in identifying tax shocks.

### Relation to Existing Literature
- Mertens and Ravn (2012): unanticipated and anticipated tax shocks have distinct macroeconomic effects due to taxpayer anticipation.
- Gechert and Groß (2019): tax base-broadening measures are less harmful to economic growth than tax hikes.
- Dabla-Norris and Lima (2018): during fiscal consolidations, tax base-broadening measures lead to smaller output and employment declines compared to rate increases.
- Mertens and Ravn (2013): PIT increases are most effective in raising revenues, CIT increases approximately revenue neutral for the United States.
- Kawano and Slemrod (2016): CIT base changes can offset revenue effects of CIT rate changes for OECD countries.
- Earlier CIT studies (Clausing (2007), Devereux (2007)) obtain similar results.
- This study focuses on revenue impacts while accounting for anticipation effects and contemporaneous changes in taxes; macroeconomic effects of TPRD v.4.0 tax policy changes are left to future research.

### Paper Structure
- Section 2 describes main features of the updated TPRD and discusses the strategy to obtain an indicator of annual tax policy changes for each tax type.

*IMF Working Papers — Cross-Country Evidence on the Revenue Impact of Tax Reforms (Introduction).*

### Section 3 describes the econometric strategy to derive the revenue yield of various tax reforms. Section 4

### wpiea2022199-print-pdf - Section 3 describes the econometric strategy to derive the revenue yield of various tax reforms. Section 4

### The Tax Policy Reform Database (TPRD v.4.0)
- TPRD v.4.0 documents 3,462 tax policy measures.
- Average of 4.7 tax measures per country per year.
- Composition of measures:
  - About 60 percent are tax base changes.
  - The majority of tax base changes consists of decreases in tax liabilities (about ⅔ of all tax base changes).
  - Rate changes are more balanced between rate increases and decreases.
- More than 80 percent of all identified tax measures are defined as “major” tax changes or reforms in a single year, with a majority introduced as part of a package.
  - Among major reforms:
    - Decreases in the tax base introduced as part of a policy package in a single year account for almost 20 percent of all identified tax measures.
    - Single-year tax base increases in policy packages account for 12 percent of total identified measures.
    - Major rate increases introduced as part of a package in a single year represent 11 percent of all identified rate measures.
- About 80 percent of tax rate measures in the database contain information on the size of the announced change (generally expressed in percentage points).
- Time coverage extended from 2014 to 2019 for most countries; classification extended to all tax types.

### Constructing the Tax Change Indicator (TC / TCI)
- Major tax reforms are counted within a year following Gechert and Groß (2019) to construct a tax change indicator variable 푇푇퐼퐼 for each country using the announcement year to date each tax change.
- Coding procedure:
  - Each individual reform coded as 푎푎 ∈ {1, −1, 푖푖푖푖 퐼퐼퐼퐼퐼퐼}.
  - When multiple tax changes announced in the same year, aggregate to produce 푇푇퐼퐼 = ∑푎푎 for each reform type, tax type, country and announcement year.
- Heatmap and summary:
  - Table 2 provides annual TCI values by tax and by year (CIT, PIT, VAT, SSC, EXE, PRO; base vs rate).
  - Total row (aggregate across years and tax types) shown as: Total -117 -130 -171 -63 5 36 -16 20 11 50 8 2 (corresponding to CIT base, CIT rate, PIT base, PIT rate, VAT base, VAT rate, SSC base, SSC rate, EXE base, EXE rate, PRO base, PRO rate).
- Correlations:
  - CIT base broadenings are moderately correlated with CIT rate cuts (correlation of 0.35).
  - PIT base changes are positively and significantly associated with same-direction rate changes.
- Distribution:
  - PIT, CIT and VAT together account for the largest share in major policy changes.
  - The same three taxes account for about 85 percent of both rate decrease and base narrowing.

### Macro Variables and Data Coverage
- Total tax revenues, unemployment, GDP, and government consumption expenditure sourced from the OECD.
- Study uses the ratio of tax revenues to GDP rather than nominal tax revenues.
- Note on total tax revenue data: Total tax revenue data include social security contributions and trade related revenues; analysis covers domestic tax revenues but trade-related taxes could not be excluded due to data limitations.
- Sample:
  - Analysis focuses on 21 countries from TPRD v.4.0.
  - China and India excluded due to lack of macro data for 1990-2018.
- Implementation timing:
  - About 52 percent of all tax changes documented are announced and implemented within the same year, 41 percent are implemented with a one-year lag, and the remaining 7 percent are implemented on a longer horizon.

### Econometric Method
- Method: Jordà (2005)’s local projections method.
- Estimated regression (notation preserved):
  - 푦푦푖푖,푡푡+ℎ − 푦푦푖푖,푡푡−1 = 훼훼푖푖 + 훾훾푡푡 + 휆휆ℎ 푇푇퐼퐼푖푖,푡푡 + ∑훽훽푗푗ℎ Δ푦푦푖푖,푡푡−푗푗3 푗푗=1 + 휃휃ℎ′ 푿푿푖푖,푡푡 + 휀휀푖푖,푡푡+ℎ
  - Outcome: 푦푦푖푖,푡푡+ℎ − 푦푦푖푖,푡푡−1 (tax revenue change from 푡푡−1 to 푡푡+ℎ).
  - 푇푇퐼퐼푖푖,푡푡 is the tax policy change indicator.
  - 푿푿푖푖,푡푡 includes controls: GDP growth, unemployment and government expenditure.
  - Both country fixed effects 훼훼푖푖 and time effects 훾훾푡푡 included.
  - Same-tax contemporaneous base changes controlled for when estimating rate change effects, and vice versa.
- Estimation details:
  - Regression estimated separately for each tax type and by rate vs base changes.
  - Benchmark regression captures response from year 0 (announcement year) to year 5.
  - OLS estimation.
  - Impulse response functions reported with both 90 percent and 68 percent confidence intervals.
  - Standard errors computed using Driscoll-Kraay (1998) formula (robust to heteroscedasticity, autocorrelation and cross-country dependence).

### Impact on Tax Revenues — Baseline Results and Key Findings
- General description:
  - A one-step increase corresponds to one additional major rate hike or base broadening announced within a year.
  - Size and temporal profile of revenue yield vary across taxes and across change types (rate vs base).
- Corporate Income Tax (CIT):
  - Announcements of CIT increases are associated with a somewhat transitory rise in tax collection.
  - Tax revenues respond on impact to announcements of CIT base-broadening measures, with collection peaking one year after announcement.
  - Revenue response to CIT rate changes is more uncertain.
  - Possible explanations: quicker firm adaptation to visible rate changes; CIT base changes often less visible and can affect collection on impact; rarity of CIT rate changes increases uncertainty.
- Personal Income Tax (PIT):
  - Announcements of PIT increases are associated with relatively large and persistent increases in the tax-to-GDP ratio.
  - Gains in tax collection appear more significant and persistent after announcements of PIT base-broadening measures rather than PIT rate hikes.
  - Interpretation: PIT has a relatively broad base in most sample countries; visibility of rate vs base changes may affect tax avoidance and behavioral responses.
- Value-Added Tax (VAT):
  - Announcements of VAT increases are generally associated with significant improvements in tax collection.
  - VAT rate increases are more likely to be followed by significant and protracted increases in the tax-to-GDP ratio than VAT base-broadening announcements.
  - Caveat: number of VAT base-broadening announcements in the sample is small (about ¼ of all documented VAT changes).
- Social Security Contributions (SSC):
  - Announcements of rises in SSC are associated with sizable surges in the tax-to-GDP ratio.
  - Tax collection responds more significantly to announcements of SSC rate increases than to SSC base-broadening measures.
  - Possible cause: recalibration of SSC base in many countries (e.g., increases in share of wages subject to SSC) combined with exemptions for some categories (e.g., exemptions for young, low-skilled workers).
- Excises (EXE):
  - Announcements of increases in EXE are generally followed by large and persistent increases in tax collection, especially for base-broadening measures.
  - Likely reflects expansion of the spectrum of products subject to excise in many countries to address negative externalities (e.g., pollution and health).

*Source: IMF working paper content (Section excerpts provided).*

### introduction of an excise on sugar-sweetened beverages in Mexico in October 2013).

### wpiea2022199-print-pdf - introduction of an excise on sugar-sweetened beverages in Mexico in October 2013

### Key Findings on Revenue Impact of Tax Reforms
- Revenue yield of tax policy changes varies substantially across taxes and types of changes.
- Tax rate measures generally show more transitory effects on tax revenues than tax base changes; this is the case in the sample for announcements of PIT, CIT, EXE, and PRO changes.
- Rate hikes have relatively more significant effects on tax collection in the case of VAT and SSC measures.
- The revenue impact of most tax policy changes is asymmetric: statistical significance changes depending on the direction of tax changes (expected increases/decreases in tax liabilities).
- Using quantitative information (intensive margins) for tax rate changes produces estimated revenue yields that are not materially different from those obtained using qualitative information (extensive margins), while generally improving the precision of estimates (narrower confidence bands).

### Data and Coding Notes
- Analysis uses the Tax Policy Reform Database v.4.0 (TPRD v.4.0).
- Minor tax policy changes represent about 20 percent of all documented changes in the TPRD v.4.0.
- About 50 percent of all major tax changes are implemented in the years following the announcement year.
- For coding of major/minor changes: the five-state tax policy change indicator 푇푇퐼퐼_퐴퐴퐴퐴퐴퐴 takes values ±2 for major changes, ±1 for minor changes, and 0 if changes fully cancel out or there is no change.
- The intensive-margin analysis restricts to 659 tax rate changes (excludes rate changes documented as ranges or in non–percentage-point units).

### Major vs. Minor Tax Policy Changes (Accounting for Both)
- Inclusion of minor tax changes does not change findings qualitatively but leads to somewhat smaller estimates of tax gains following announcements.
- One-step increase in the shock indicator captures both a minor policy change from 0 to 1 and a major change from 1 to 2; results using the −2 to 2 shock indicator should be interpreted with caution.

### Announcement Year vs. Implementation Year (Fiscal Foresight)
- Baseline identification uses announcement years to minimize fiscal foresight issues.
- When shocks are dated by implementation years:
  - Tax-to-GDP ratio is less responsive to changes in VAT and EXE, pointing to significant anticipation effects.
  - Tax collection appears less responsive to increases in SSC and PRO.
  - Impact of CIT and PIT changes based on implementation years seems to partially support that taxpayers may better understand the impact of rate changes on their tax liabilities compared to base changes.

### Directional Asymmetry of Tax Changes
- TC variable split into TCI (base-broadening or rate-increasing) and TCD (base-narrowing or rate-decreasing) to test asymmetry.
- Findings indicate asymmetric revenue responses for most tax changes:
  - Positive reactions to announcements of rate hikes or base broadening in VAT, EXE, or SSC.
  - Unlikely responses to announcements of rate cuts or base narrowing for the same taxes.
  - Asymmetric response also observed for rate increases in PIT and PRO.
  - Symmetric responses observed for CIT rate or base changes and for PIT and PRO base changes.

### Intensive vs. Extensive Margins (Size of Rate Changes)
- Table 3 (means and standard deviations for coded changes) reports:
  - CIT: EM mean -0.17; EM sd 0.67; IM mean -0.98; IM sd 5.43
  - PIT: EM mean -0.11; EM sd 0.70; IM mean -0.49; IM sd 4.39
  - VAT: EM mean 0.05; EM sd 0.50; IM mean 0.03; IM sd 3.43
  - SSC: EM mean 0.04; EM sd 0.44; IM mean 0.20; IM sd 3.80
  - EXE: EM mean 0.03; EM sd 0.24; IM mean 1.35; IM sd 14.65
  - PRO: EM mean 0.01; EM sd 0.18; IM mean 0.38; IM sd 5.50
- Both IM and EM are standardized to zero mean and unit standard deviation for each tax type.
- Comparison of a one-standard-deviation tax rate increase:
  - Response of tax-to-GDP to CIT and SSC rate hikes and to some extent VAT based on IM cannot be distinguished from EM (blue and red confidence bands fully overlap).
  - Revenue impact of PIT, excise, and property tax rate increases broadly consistent across margins, though significance and persistence may be weaker on IM than EM for some taxes.
  - For EXE and PRO, impact on revenues is higher on intensive than extensive margins.
  - For PIT, impact is mostly felt on the extensive margin.

### Methodological Remarks
- Baseline estimates use announcement-year identification and local projections.
- Confidence intervals displayed at 68 percent and 90 percent levels in figures; Driscoll-Kraay standard errors used for Figure 7 (68% confidence levels reported).
- The econometric strategy does not control for compositional effects for a given type of change (number of +1 and -1 within a year). Future research will investigate compositional effects using category details in TPRD v.4.0.
- Alternative specification: including minor tax policy changes as additional controls (TC_minor defined with step size ±0.5) does not change main results for major tax policy changes.

### Conclusions and Future Research Directions
- Revenue yields depend on tax type and whether the change is a rate or base measure; rate changes tend to be more transitory in PIT, CIT, EXE, and PRO announcements.
- VAT and SSC rate hikes show relatively larger effects on collections.
- Revenue impacts are often asymmetric by direction of policy change.
- Quantitative information (intensive margins) generally corroborates qualitative findings and improves estimate precision.
- Future research planned to:
  - Deepen analysis of asymmetric effects of tax changes.
  - Further exploit quantitative information in TPRD v.4.0 to precisely estimate revenue yields (e.g., controlling for revenue share of a particular tax type).
  - Advance the debate on macroeconomic effects of different tax changes (e.g., tax multipliers).

*IMF WORKING PAPERS Cross-Country Evidence on the Revenue Impact of Tax Reforms — INTERNATIONAL MONETARY FUND*

### Appendix A

### Appendix A

### Distribution of Tax Changes by Tax Type, Reform Change, and Category (Table A.1)
- Source: Tax Policy Reform Database, OECD, IBFD.
- Summary counts and averages (tax type / change type / category — number of observations, count of country years, average number of measures):
  - PIT: 11994802.5
    - RATE: 3732361.6
      - Statutory rates: 89771.2
      - Top rate: 95861.1
      - Bottom rate: 58551.1
      - Surcharges: 57541.1
      - Capital gains: 27261.0
      - Dividends: 11111.0
      - Other: 36331.1
    - BASE: 8264102.0
      - Standard relief: 2642091.3
      - Child relief: 77721.1
      - Capital gains: 53471.1
      - Interest relief: 42411.0
      - SSC, pension, insurance relief: 35331.1
      - Other relief: 3552821.3
  - CIT: 10434522.3
    - RATE: 3662461.5
      - Statutory rates: 55451.2
      - Top rate: 95871.1
      - SMEs: 33331.0
      - Surcharges: 51441.2
      - Capital gains: 36331.1
      - Dividends: 14131.1
      - Other: 82741.1
    - BASE: 6773761.8
      - R&D promotion: 48421.1
      - Investment promotion: 2261861.2
      - Loss-carry rules: 25231.1
      - Capital gains: 73661.1
      - Thin capitalization: 31301.0
      - Other base changes: 2742101.3
  - VAT: 3912481.6
    - RATE: 2411701.4
      - Standard rate: 89851.0
      - Reduced rate: 32291.1
      - Other: 1201011.2
    - BASE: 1501181.3
      - Exemptions on food: 981.1
      - Exemptions on medical supply: 1111.0
      - Other base changes: 1401161.2
  - SSC: 3671852.0
    - RATE: 2081261.7
      - Employee: 93801.2
      - Employer: 74691.1
      - Other: 41351.2
    - BASE: 1591061.5
      - Employee: 59521.1
      - Employer: 71611.2
      - Other: 29271.1
  - EXE: 3291781.8
    - RATE: 2751541.8
      - Alchool products: 41411.0
      - Tobacco: 60571.1
      - Oil products: 98841.2
      - Other: 76681.1
    - BASE: 54421.3
      - Alchool products: 221.0
      - Tobacco: 551.0
      - Oil products: 23221.0
      - Other: 24221.1
  - PRO: 1331041.3
    - RATE: 37351.1
      - Real estate: 21201.1
      - Financial assets: 661.0
      - Other: 10101.0
    - BASE: 96801.2
      - Real estate: 43371.2
      - Financial assets: 18171.1
      - Other: 35331.1
  - Grand Total: 34627354.7 (Total changes)

### Detailed Information of Tax Changes by Type, Reform Change and Category (Table A.2)
- Source: Tax Policy Reform Database, OECD, IBFD.
- Table presents disaggregated counts of observations, counts of country years, and average number of measures across multiple partitions (quantitative/qualitative, increase/decrease).
- Selected exact entries and aggregates preserved as in source (examples drawn verbatim):
  - PIT: 112861.3210444.83222051.6270188
  - RATE (PIT block): 112 86 1.3210444.8322 2051.6
  - Statutory rates (PIT): 17171.054331.671601.233
  - BASE aggregates and reliefs (PIT): 255179 1.45715011.48264102.0; Standard relief: 68621.1196464.3; Standard relief row: 264209 1.3
  - CIT aggregates and RATE breakdowns: CIT88 801.1235475.0 323227 1.4269; RATE 88801.1235475.03232271.417151.126
  - VAT aggregates: VAT130961.48434 2.52141491.499881.178 392.01771371.3; RATE130961.484342.52141491.418181.099
  - SSC aggregates: SSC92631.540221.8132891.5125751.7110402.82351361.7; RATE92 631.540221.8 13289 1.552341.5
  - EXE aggregates: EXE 97591.620141.4117701.7180 1061.732211.5212124 1.7; RATE97591.620141.4117701.7142811.8
  - PRO aggregates: PRO16161.01181.427261.053481.153321.7106861.2; RATE16161.01181.42726 1.0651.2431.3
  - Grand Total (detailed row): 5352672.06005012.011354602.59964302.313315424.623276613.5
- Table distinguishes:
  - Increase / Decrease / IncreaseDecrease
  - Quantitative information / Total quantitative changes
  - Qualitative information / Total qualitative changes

### Correlation Matrix of Tax Types (Table A.3)
- Variables and key correlations (note labels: “bi” = base broadening, “bd” = base narrowing, “ri” = rate hike, “rd” = rate cut). Significance indicators: *** p<0.01, ** p<0.05, * p<0.1.
- Selected correlations (preserved exactly as presented):
  - CIT_bi with CIT_bi: 1
  - CIT_bd with CIT_bi: -0.13
  - CIT_ri with CIT_bi: 0.22
  - CIT_rd with CIT_bi: 0.35*
  - PIT_bi with PIT_bi: 1
  - PIT_bd with PIT_bi: 0.17
  - PIT_ri with PIT_bi: 0.46**
  - PIT_rd with PIT_bi: 0.23
  - VAT_bi with VAT_bi: 1
  - VAT_bd with VAT_bi: -0.12
  - VAT_ri with VAT_bi: 0.18
  - VAT_rd with VAT_bi: -0.27
  - SSC_bi with SSC_bi: 1
  - SSC_bd with SSC_bi: -0.10
  - SSC_ri with SSC_bi: 0.29
  - SSC_rd with SSC_bi: -0.09
  - EXE_bi with EXE_bi: 1
  - EXE_bd with EXE_bi: -0.10
  - EXE_ri with EXE_bi: -0.14
  - EXE_rd with EXE_bi: -0.11
  - PRO_bi with PRO_bi: 1
  - PRO_bd with PRO_bi: -0.09
  - PRO_ri with PRO_bi: 0.07
  - PRO_rd with PRO_bi: -0.09

### Distribution of Major Tax Rate Measures During 1990 – 2018 (Figure A.1)
- Bars and counts shown (verbatim counts from figure labels):
  - # of rate measures: 886
  - # of rate measures with intensive margins: 739
  - # of rate measures in pps: 714
  - # of rate measures in pps ("from xx to xx" obs. excluded): 659
  - # of rate cut in pps ("from xx to xx" obs. excluded): 361
  - # of rate hike in pps ("from xx to xx" obs. excluded): 298
- Caption: our sample

### Appendix B — Coding Major and Minor Tax Policy Changes
- Overview:
  - The coding scheme addresses nine possible cases where at least two same-year tax changes are recorded; single policy change cases are skipped.
  - Tax change announcement year is used to date each tax change.
- Step T1 to T2: Coding the Direction and Magnitude of a Change
  - Two functions defined:
    - a_a codes all directions of changes as ±1 (regardless of magnitude) and stores output in column a_a of T2.
    - b_b codes a_a as ±1 if a tax change is considered major and 0 otherwise; output stored in b_b.
  - Equation (B.1) (text preserved in structure):
    - a_a = { 1, -1, [conditional text as in source] }
    - b_b = { 1, -1, 0, [conditional text as in source] }
- Step T2 to T3: Aggregation
  - Aggregate multiple same-year changes into T3 columns A_A and B_B:
    - A_A = ∑ a_a
    - B_B = ∑ b_b
- Step T3 to T4: Dimension Reduction
  - Transform A_A and B_B into three-state variables A_A* and B_B* in T4 using equation (B.2):
    - A_A* = { -1 if A_A<0; 1 if A_A>0; 0 if A_A=0 }
    - B_B* = { -1 if B_B<0; 1 if B_B>0; 0 if B_B=0 }
  - Note on information loss:
    - On average, about 25% of tax changes in A_A that are > 1 or < -1 are replaced by 1 and -1, respectively.
    - About 20% for B_B.
    - Footnote: The coded tax changes after the step of T2 to T3 consist of 1941 observations, of which 486 observations are less than – 1 or greater than 1 for A_A and 413 for B_B.
- Step T4 to T5: Constructing the Five-State Tax Change Variable
  - Final five-state variable TTI_AAAA = A_A* + B_B* (labels preserved as in source).
  - Applying this procedure across tax reform, tax type, country and year generates full set of TTI_AAAA indicators.

*Cross-Country Evidence on the Revenue Impact of Tax Reforms — Working Paper No. WP/2022/199*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022199-print-pdf.pdf_