## Appendix I. MATR Data Construction

## Source details

**Canonical URL:** [Appendix I. MATR Data Construction](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022001-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022001-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022001-print-pdf.pdf.json)

---

### Contents overview
- Appendix covers construction of MATR and lists: Appendix Table A1.1-A1.6 and Appendix Figures A1.1–A1.3.
- Related appendices: Appendix II (Other Data), Appendix III (MATR and the Literature), Appendix IV (Robustness Checks and Extensions).

### Introduction and motivation
- Measuring restrictiveness of trade policy faces aggregation challenges: tariff schedules with more than 5,000 individual tariff lines; specific or ad valorem tariffs; tariffs that differ by exporter; difficulty weighting individual goods; complex non-tariff barriers (quotas, licenses, regulatory requirements).
- Theoretically ideal measures: “Trade Restriction Indices” (TRIs) (Anderson and Neary (2005)) map policies to a uniform tariff producing the same welfare effect, but TRIs are not widely available.
- Desiderata for a practical measure: simple; based on plausible, quantitative, objective policy inputs; timely and easily updateable; available for many countries over a substantial period; covering both tariff and non-tariff barriers.

### A New Measure: MATR — definition and components
- MATR aggregates up to 22 AREAER binary variables; each sub-indicator = 1 if a restriction is present in a country-year, = 0 otherwise.
- Theoretical range: 0–22; empirical range: between 2 and 21.
- Underlying AREAER categories used (headings and subheadings preserved):
  - II. Exchange measures
    - II.A. Restrictions and/or multiple currency practices
    - II.B. Exchange measures imposed for security reasons
  - IV. Restrictions to payments
    - IV.A. Prescription of currency requirements
    - IV.B. Payments arrangements
    - IV.C. Administration of control
    - IV.D. Payment arrears
    - IV.F. Controls on exports and imports of banknotes
  - VII. Import Restrictions
    - VII.A. Foreign exchange budget
    - VII.B. Financing requirements for imports
    - VII.C. Documentation requirements for release of forex for imports
    - VII.D. Import licenses and other nontariff measures
    - VII.E. Import taxes and/or tariffs
    - VII.F. State Import Monopoly
  - VIII. Export Restrictions
    - VIII.A. Repatriation requirements
    - VIII.B. Financing requirements
    - VIII.C. Documentation requirements
    - VIII.D. Export licenses
    - VIII.E. Export taxes
  - IX. Payments for Invisibles Restrictions
    - IX.A. Payments for Invisibles, Transfers & Current Transfers
  - X. Proceeds for Invisibles Restrictions
    - X.A. Repatriation requirements on Proceeds
    - X.A.1. Surrender Requirements on Proceeds
    - X.B. Restrictions on use of funds
- Rationale: sections VII and VIII likely most important, but other instruments matter; each fundamental is absolute (unity = presence, zero = absence).

### MATR construction and coding rules (AREAER-based)
- Primary source: IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER), extended back to 1949 via hard-copy examination and gap-filling.
- Coding rules:
  - Restriction = 1, absence of restriction = 0, missing/unavailable = n/a.
  - Dating convention: use AREAER country text dates; when future announced change is stated, use announced date.
  - Granularity: collect data as granular as reasonably possible (e.g., VII_D_1 and VII_D).
  - Frequency and status quo: check year-to-year changes via country text; absence of change indicates status quo (carry backward/forward when no information).
  - Category nesting: unity in a sub-component implies unity in higher-level aggregates (example category coIV_A_1_b_1 → coIV_A_1_b → coIV_A_1 → coIV_A).
- Large-change identification threshold: more than 1.64 (i.e., > |2σ| in practice).

### Properties, coverage, and illustrative patterns
- Coverage:
  - MATR covers an unbalanced panel of 157 countries annually between 1949 and 2019.
  - Coverage expands from ~30 economies in 1949 to >100 in 1973 and over 150 by 2000.
- Advantages:
  - Includes tariffs and a broad set of non-tariff barriers.
  - Avoids normalization issues by aggregating binary components.
  - Regularly updated and essentially unaffected by missing granular data because missing entries can be filled using AREAER annual change entries.
- Limitations:
  - Ad hoc and lacks a clear theoretical interpretation.
  - Codes presence/absence but not intensity or efficacy.
  - Aggregate measure rather than weighted average of disaggregated microdata.
- Stylized facts and cross-sections:
  - No strong relationship between MATR and natural logarithm of real GDP in 2016.
  - AEs and EMDEs began comparable, liberalized from early 1970s, little change since early 2000s; liberalizations more dramatic for AEs.
  - Regional patterns: Europe typically lowest; Africa highest.
  - Country examples: United States low MATR; China high MATR; Indonesia more restrictions than United Kingdom; Russia more than Canada and Germany; India and South Africa numerous restrictions throughout; dramatic increases in Venezuela and reductions for United Kingdom and Hong Kong.

### Alternative aggregations and sensitivity analysis
- Baseline: unweighted sum of up to 22 binary variables.
- Alternatives considered:
  - Standard factor analysis: first eigenvalue = 6.16; second and third eigenvalues = .72 and .64; first-factor loadings positive for 21 of 22 fundamentals.
  - First principal component following Chinn and Ito (2008).
  - Variant aggregations:
    - Variant 1: restrictive version summing eleven trade-related variables (VII.A–VII.F and VIII.A–VIII.E), principle range 0–11.
    - Variant 4: most inclusive, adds 27 more fundamentals (all sub-indicators).
    - Variants 2 and 3 intermediate.
  - MATR is highly correlated with all six variants tested.

### Main empirical methodology
- Estimation method: Jordà’s (2005) local projection method with OLS for horizons k = 0,...,5.
- Sample: unbalanced sample of 157 countries from 1949 to 2019.
- Controls: country fixed effects ({α}), time fixed effects ({γ}), lagged changes in MATR ({ϑ}), lagged output ({θ}).
- Inference: Driscoll and Kraay (1998) standard errors to account for cross-sectional and time dependence.
- One-standard deviation in MATR = 0.82 changes in the index.

### Key empirical findings on output and other outcomes
- Aggregate output responses:
  - A one-standard deviation increase in MATR (comparable to Thailand in 2000) is associated with:
    - Immediate reduction in output by 0.2 percent.
    - Reduction in output by 0.7 percent five years after.
  - Effect is highly significant statistically and economically.
  - The medium-term output effect is almost twice the medium-term output effect of one-standard deviation increase in tariff rates found in Furceri et al. (2021).
- Large-change episodes:
  - Large-change definition: changes in MATR > |2σ|.
  - Episodes identified: 385 episodes (123 increases; 262 liberalizations), majority between 1989 and 2008.
  - Crisis-filtered sample (exclude episodes preceded or succeeded by crises within one year): 162 episodes.
  - Narrative-filtered exogenous sample: 58 episodes (7 increases; 51 liberalizations).
  - Output declines for large/exogenous episodes:
    - ~0.35 percent decline in output 5 years after major restriction (large-change sample).
    - Almost 0.4 percent decline 5 years after in narrative-exogenous sample excluding structural reforms.
- Component and subsample results:
  - Effects significant in both AEs and EMDEs, larger in EMDEs.
  - In AEs: results driven mainly by export and import restrictions (import restriction statistically significant).
  - In EMDEs: drivers include invisibles, exports, imports, and payments (all statistically significant).
- Alternative dependent variables / channels:
  - Labor productivity: declines by about 1.5 percent after five years following an increase in trade barriers.
  - Investment: falls following increased trade barriers.
  - Exports and imports: both fall after increases in protectionism.
  - Consumption: falls by around 1 percent after five years.
  - Inequality: increases in trade restrictions are associated with a reduction in inequality.
- Interpretation: trade restrictions resemble shocks to productivity of the tradeables sector; reduced-form estimates indicate meaningful adverse economic effects but do not permit full welfare analysis.

### Non-linearities and heterogeneity
- Complementarity hypothesis:
  - Effect of increase in non-tariff restrictions is larger when tariffs are high.
- GVC hypothesis:
  - Larger adverse effects for countries importing larger share of intermediate inputs (greater Global Value Chain participation).
- Empirical implementation:
  - Smooth transition model: F(zit) = exp(−zit)/(exp(−zit)+exp(zit)) with zit normalized to zero mean and unit variance.
  - Coefficients β1k and β2k capture impacts in low-z and high-z states.
  - Equation (2) estimated with OLS; standard errors clustered at country level.

### Identification, exogeneity, and robustness checks
- Narrative search:
  - Manual search of official national documents, IMF/World Bank/OECD/WTO reports, policy papers, academic publications, and AREAER reports to identify exogenous MATR changes and classify episodes.
  - Appendix Table A1.5 lists exogenous episodes with descriptions and source links.
- Discrete-variable specification:
  - Coding: 1 for major restriction, -1 for major liberalization, 0 otherwise — results similar though smaller than baseline continuous-change estimates.
- Granger causality test:
  - No evidence that past GDP growth helps predict large changes in MATR; p-value ≈ 0.76 (null that GDP growth Granger causes large changes in MATR).
- Robustness:
  - Results robust to alternative non-linear specifications; robustness checks and alternative measures shown in Appendix Figures A4.1–A4.6.

### Selected data sources and sample coverage
- Core macro series: Penn World Table (PWT10.0) for GDP, productivity, employment, investment, consumption.
- Trade flows and balances: World Economic Outlook (April 2021).
- Gini: Standardized World Income Inequality Database (SWIID).
- Regulation and job protection: Alesina et al. (2019).
- GVC: UNCTAD EORA database.
- Tariffs: Furceri et al. (2021).
- Country classification: IMF World Economic Outlook.
- Sample and time coverage summary:
  - Unbalanced sample of 157 countries, 1949–2019.
  - MATR begins in 1949; Trade Freedom begins in 1995.

### Appendix highlights and select tables
- Appendix Table A1.3: Examples of AREAER narrative records (2016) for II, IV, VII, VIII, IX, and X categories with verbatim excerpts (e.g., Angola, The Bahamas, Cameroon, Denmark, Egypt, Guinea, Bangladesh, Guinea-Bissau, India, Kazakhstan, Laos, Macedonia, Niger, Barbados, Eritrea, Fiji, Croatia, Liberia, Sri Lanka, Suriname, Ukraine, China).
- Appendix Table A1.5: Enumerated episodes of exogenous changes in MATR with country, year, sign (Liberalization / Increase), textual description, and source links (selected entries include Angola 2018 Liberalization; Austria 1955 Liberalization; Bahrain 2004 Increase in restrictions; Belarus 2006 Liberalization; Bosnia and Herzegovina 1997 and 2001 Liberalizations; Botswana 1999 Liberalization; Costa Rica 1966 Increase; Costa Rica 2011 Liberalization; etc.).
- Appendix Table A1.6: Compact ISO3-coded listing of exogenous episodes “Not Part of Broader Reform Packages” (examples):
  - AGO 2018 -1
  - AUT 1955 -1
  - AUT 1989 -1
  - BGR 2006 -1
  - BHR 2004 1
  - BIH 1997 -1
  - CRI 1966 1
  - CRI 2011 -1
  - JPN 2003 1
  - LBR 1986 1
  - MWI 1997 1
  - PHL 1966 1
  - UKR 2018 -1
- Appendix Table A2.2 (Other Data): Database coverage and indicator counts and years (examples):
  - Penn World Table (PWT 10.0): 213 1950–2019 Real GDP; 212 1950–2019 Employment; 211 1950–2019 Productivity; 212 1950–2019 Investment; 213 1950–2019 GDP Deflator Index; 212 1950–2019 Household Consumption.
  - World Economic Outlook (April 2021): 172 1950–2019 Trade Balance; 178 1950–2019 Exports volume; 178 1950–2019 Imports volume.
  - Furceri, Hannan, Ostry and Rose (2021): 168 1960–2014 Tariffs.
  - UNCTAD EORA Database: 158 1990–2018 GVC.
  - World Bank (WDI): 209 1960–2019 Total Population; 195 1960–2019 Trade %GDP; 190 1990–2019 Real GDP per capita PPP; 187 1988–2018 Tariff rate applied weighted mean all products.
- Appendix Figures A4.1–A4.6: Robustness and heterogeneity IRFs — cumulative IRFs after one standard deviation increase in MATR with 90 percent confidence intervals and Driscoll-Kraay standard errors; figure A4.6 focuses on Large Changes in MATR defined as > |2σ| and coding 1 ≡ increase in restrictions, -1 ≡ liberalization, 0 ow.

### Summary conclusions and suggested directions for future research
- MATR is a simple, policy-based aggregate measure of trade restrictions covering both tariffs and a wide set of non-tariff barriers for 157 countries, 1949–2019.
- Main empirical conclusion: increases in MATR are associated with large and persistent declines in GDP and declines in labor productivity, investment, trade flows, and consumption; complementary effects with tariffs and larger impacts for countries with higher GVC participation.
- Limitations: theoretically ad hoc; aggregates presence/absence not intensity; does not replace micro-founded TRIs but complements existing measures.
- Suggested future research directions:
  - Consider impediments to FDI in addition to goods and services.
  - Consider lagged values or moving averages analogous to Chinn and Ito (2008) rather than contemporaneous values only.
  - Develop a rigorous theoretical model to understand general equilibrium and welfare consequences of trade restrictions.

*Source: wpiea2022001-print-pdf - Appendix I. MATR Data Construction.*

### Appendix I. MATR Data Construction .....................................................................................

### Appendix I. MATR Data Construction

### Contents overview
- Appendix I. MATR Data Construction ............................................................................................................. 29
- Appendix Table A1.1 List of Economies ....................................................................................................... 31
- Appendix Table A1.2 Selected Economies and MATR Values .................................................................... 32
- Appendix Table A1.3 Examples of Narrative Records in AREAER, 2016 .................................................... 33
- Appendix Table A1.4 Large Changes in MATR, With and Without Recessions. Recession years are shown 
in black italic. ................................................................................................................................................ 36
- Appendix Table A1.5 Exogenous Changes in MATR ................................................................................... 39
- Appendix Table A1.6 Exogenous Changes in MATR, Not Part of Broader Reform Packages .................... 47
- Appendix Figure A1.1: MATR country coverage over time, by region ......................................................... 48
- Appendix Figure A1.2: MATR changes over time, 1976-2016 ..................................................................... 48
- Appendix Figure A1.3: Perturbations of MATR ............................................................................................ 49

### Related appendices referenced
- Appendix II. Other Data ....................................................................................................................................... 50
- Appendix Table A2.1 Other Trade Indicators ............................................................................................... 50
- Appendix Table A2.2 Other Data ................................................................................................................. 52
- Appendix III. MATR and the Literature ................................................................................................................ 55
- Appendix IV. Robustness Checks and Extensions ............................................................................................. 58
- Appendix Figure A4.1: Response of (log) GDP to Changes in MATR, by income group ............................. 58
- Appendix Figure A4.2: Response of (log) GDP to Changes in MATR: Robustness Checks ....................... 59
- Appendix Figure A4.3: Response of (log) GDP to Changes in MATR: Alternative Measures ..................... 60
- Appendix Figure A4.4: Response of (log) GDP to Changes in MATR, tariff complementarity (%) .............. 61
- Appendix Figure A4.5: Response of (log) GDP to Changes in MATR, role of GVCs (%) ............................ 61
- Appendix Figure A4.6: Response of (log) GDP to Large Changes in MATR (%) ........................................ 62

*Source: wpiea2022001-print-pdf - Appendix I. MATR Data Construction.*

### References .............................................................................................................

### wpiea2022001-print-pdf - References .............................................................................................................

### Introduction
- Measuring the restrictiveness of a country’s trade policy is important for empirical assessments, trade negotiations, and economic rescue programs.
- Aggregation challenges include: tariff schedules with more than 5,000 individual tariff lines; tariffs that are specific or ad valorem; tariffs that differ by exporter; difficulty weighting individual goods; and complex non-tariff barriers (quotas, licenses, regulatory requirements).
- Theoretically ideal measures: “Trade Restriction Indices” (TRIs) (Anderson and Neary (2005)) which map policies to a uniform tariff producing the same welfare effect, but TRIs are not widely available.
- This paper proposes a practical alternative measure guided by desiderata: simple; based on plausible, quantitative, objective policy inputs; timely and easily updateable; available for many countries over a substantial period; covering both tariff and non-tariff barriers.

### A New Measure of Aggregate Trade Restrictions (MATR)
- MATR is constructed using the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER) combining:
  - AREAER online database (available from 1999 onwards).
  - Narrative accounts in printed AREAER country-year reports (from 1949 onwards).
- Key properties:
  - Simple, plausible, quantitative, easily updated.
  - Based solely on policy-relevant measures of trade policy.
  - Covers an unbalanced panel of 157 countries annually between 1949 and 2019.
  - Strongly correlated with existing measures of openness and trade policy.
  - More comprehensive and more granular than many existing indicators.
  - Robust to minor methodological perturbations.

### Construction and Components of MATR
- MATR aggregates up to 22 AREAER binary variables; each sub-indicator is scored 1 if a restriction is present in a country-year, 0 otherwise.
- In principle MATR ranges between 0 and 22; in practice MATR varies between 2 and 21.
- Underlying AREAER categories used (headings and subheadings preserved):
  - II. Exchange measures
    - II.A. Restrictions and/or multiple currency practices
    - II.B. Exchange measures imposed for security reasons
  - IV. Restrictions to payments
    - IV.A. Prescription of currency requirements
    - IV.B. Payments arrangements
    - IV.C. Administration of control
    - IV.D. Payment arrears
    - IV.F. Controls on exports and imports of banknotes
  - VII. Import Restrictions
    - VII.A. Foreign exchange budget
    - VII.B. Financing requirements for imports
    - VII.C. Documentation requirements for release of forex for imports
    - VII.D. Import licenses and other nontariff measures
    - VII.E. Import taxes and/or tariffs
    - VII.F. State Import Monopoly
  - VIII. Export Restrictions
    - VIII.A. Repatriation requirements
    - VIII.B. Financing requirements
    - VIII.C. Documentation requirements
    - VIII.D. Export licenses
    - VIII.E. Export taxes
  - IX. Payments for Invisibles Restrictions
    - IX.A. Payments for Invisibles, Transfers & Current Transfers
  - X. Proceeds for Invisibles Restrictions
    - X.A. Repatriation requirements on Proceeds
    - X.A.1. Surrender Requirements on Proceeds
    - X.B. Restrictions on use of funds
- Rationale: sections VII (Import Restrictions) and VIII (Export Restrictions) are likely most important, but other instruments matter for protectionism.
- Each fundamental is absolute (unity reflects presence of a trade barrier, zero its absence), not relative.

### Properties, Coverage, and Data Features
- Coverage expansion:
  - Coverage increases from about 30 economies in 1949 to more than 100 in 1973, and over 150 by 2000.
- Advantages:
  - Includes tariffs and a broad set of non-tariff barriers.
  - Avoids normalization issues by aggregating binary components.
  - Regularly updated and available for many economies from 1949 through 2019.
  - Essentially unaffected by missing granular data because missing entries can be filled using AREAER annual change entries.
- Limitations:
  - Ad hoc and lacks a clear theoretical interpretation.
  - Codes presence/absence of restrictions but not intensity or efficacy of measures.
  - Intrinsically aggregate rather than a weighted average of disaggregated microdata.

### Graphical Portrait and Stylized Facts
- No strong relationship between MATR and the natural logarithm of real GDP (size of the economy) in 2016.
- Time-series patterns:
  - Advanced economies (AEs) and emerging market and developing economies (EMDEs) began in comparable situations, liberalized from early 1970s, and have done little since early 2000s; liberalizations were more dramatic for AEs.
  - Regional patterns: European countries typically have the lowest restrictions; African countries the highest.
  - MATR moves little over a typical year; stability is visible across decades.
- Examples from country panels:
  - United States: low MATR throughout period.
  - China: high MATR throughout period.
  - Indonesia: more restrictions than United Kingdom.
  - Russia: more restrictions than Canada and Germany.
  - India and South Africa: numerous restrictions throughout.
  - Dramatic changes observed in Venezuela (increases) and reductions for United Kingdom and Hong Kong.
- Cross-sectional relationships (1996 examples):
  - Richer countries tend to have fewer trade restrictions (MATR negatively correlated with log real GDP per capita).
  - Smaller countries tend to be more open and have fewer restrictions (MATR positively related to log population in the description).
  - MATR is strongly correlated with the World Bank’s mean weighted applied tariff rate.
  - MATR is negatively correlated with de facto trade openness (ratio of exports and imports to output).
- Caveat: these are primarily bivariate visual relationships; Table 1 (in source) reports multivariate regressions controlling for year fixed effects, log size and income.

### Sensitivity Analysis and Alternative Aggregations
- Baseline MATR: simple unweighted sum of up to 22 binary variables.
- Alternative aggregation methods considered:
  - Standard factor analysis to extract the first principal factor from the 22 fundamentals.
    - First eigenvalue: 6.16; second and third eigenvalues: .72 and .64.
    - First-factor loadings are positive for 21 of the 22 fundamentals.
  - First principal component following Chinn and Ito (2008).
  - Variants based on different subsets of AREAER fundamentals:
    - Variant 1: restrictive version summing eleven trade-related variables (VII.A through VII.F and VIII.A through VIII.E) ranging in principle from 0-11.
    - Variant 4: most inclusive, adds 27 more fundamentals (all sub-indicators).
    - Variants 2 and 3 are intermediate.
  - MATR is highly correlated with all six variants tested (see Figure A1.3 and Table 2 in source).

### The Economic Effects of Trade Restrictions
- Estimation approach:
  - Use Jordà’s (2005) local projection method to estimate responses of GDP and other outcomes (investment, productivity, employment, consumption, trade flows, inequality) to changes in MATR.
- Main empirical findings:
  - Trade restrictions are associated with large and persistent declines in GDP.
  - A one-standard deviation increase in the index (example: Thailand in 2000) is associated with:
    - Reduction in the level of GDP by about 0.2 percent in the year of the change in MATR.
    - Reduction in the level of GDP by about 0.7 percent five years later.
  - These effects are statistically significant and economically sizeable.
  - The impact is almost twice the medium-term output effect of the one-standard deviation increase in tariff rates found in Furceri et al. (2021).
- Additional findings:
  - Complementarity between tariff and non-tariff restrictions: the effect of an increase in non-tariff restrictions is larger when tariffs are high.
  - Larger adverse effects for countries that import a larger share of intermediate inputs (greater participation in Global Value Chains).
- Identification concerns and robustness:
  - Narrative source limitations: AREAER narratives do not always allow separation of policy changes motivated by cyclical stabilization from exogenous policy shifts.
  - To address endogeneity, the authors conduct an extensive narrative search (official national documents, IMF/World Bank/OECD/WTO reports, policy papers, academic publications) to identify dates of major MATR changes and classify exogenous changes.
  - Results using exogenous measures confirm that trade restrictions harm economic activity.

*Source: wpiea2022001-print-pdf - References .............................................................................................................*

### 1. Novy’s (2012) measure of Trade Costs, a “comprehensive all-inclusive measure ... providing an alternative

### wpiea2022001-print-pdf - 1. Novy’s (2012) measure of Trade Costs, a “comprehensive all-inclusive measure ... providing an alternative

### Comparison with alternative indicators
- Alternative indicators discussed and their scope:
  - Novy’s (2012) measure of Trade Costs, with export weights.
  - World Economic Forum’s 2016 Index of Trade Enablement, covering 136 economies and available for 2012, 2014, and 2016.
  - Quinn’s measure of Current Account Financial Openness.
  - A World Bank TRI (2009) using Kee et al. (2009) methodology.
  - A World Bank TRI using only tariffs for 2009.
  - A World Bank TRI for the service sector.
- Correlation findings:
  - In each case, the correlation between alternative indicators and MATR is correctly signed.
  - Table 1 indicates correlations are significantly different from zero except for some TRI measures (potentially due to small sample size).
- Specific comparisons:
  - Heritage Foundation’s Trade Freedom:
    - Trade Freedom is available from 1995; MATR begins in 1949.
    - Trade Freedom and MATR are strongly (negatively) correlated at multiple points in the sample.
    - Both are imperfect measures and can serve as instrumental variables for each other.
  - Chinn-Ito (2008) kaopen:
    - Kaopen measures international mobility of capital; MATR focuses on flows of goods and services.
    - Kaopen is the first principal component of four transformed AREAER-based variables; MATR sums up to 22 AREAER fundamentals.
    - MATR and kaopen are substantially different series but strongly negatively related.

### Potential applications of MATR
- Rationale:
  - More imperfect measures are better than fewer; MATR is proposed as a helpful addition because of substantial coverage over time.
- Examples where MATR might have been useful:
  - Ilzetzki, Mendoza and Végh (2010): splitting samples by openness — MATR could improve identification distinguishing policy-driven openness from low openness for other reasons.
  - Rodrik (1998) and Alesina and Wacziarg (1998): studies linking openness to government size might benefit from distinguishing artificial vs. natural barriers using MATR.
  - Hall and Jones (1999) and Sachs and Warner (1995): studies of policy-driven trade openness as determinant of income and growth could use MATR to address criticisms focused on mismeasured trade barriers.
  - Empirical work that employs separate measures of goods/services mobility and capital mobility (e.g., Fratzscher (2012), Eichengreen, Park and Shin (2012), Milesi-Ferretti and Tille (2011)) could benefit from a simple, broad, policy-based analogue to Chinn-Ito for trade (MATR).

### Empirical methodology for assessing MATR’s effect on output
- Estimation approach:
  - Local projection method of Jordà (2005) with OLS for horizons k = 0,...,5.
  - Equation (1) estimated for an unbalanced sample of 157 countries from 1949 to 2019.
  - Country fixed effects ({α}) and time fixed effects ({γ}) included.
  - ∆R denotes the change in MATR (increasing with restrictions).
  - Lagged changes in MATR and lagged output included as controls ({ϑ}, {θ}).
  - Driscoll and Kraay (1998) standard errors applied to account for cross-sectional and time dependence.
- Identification and robustness:
  - Baseline does not distinguish exogenous vs. endogenous changes in trade barriers; sensitivity checks focus on major episodes of liberalization/tightening.
  - Alternative specifications and robustness checks described in Appendix 4.
  - Panel cointegration tests reject the null that residual of equation (1) is non-stationary.
  - One-standard deviation in MATR is equivalent to 0.82 changes in the index.

### Main results on output (aggregate)
- Impulse response summary:
  - A one-standard deviation increase in MATR (comparable to Thailand in 2000) is associated with:
    - Immediate reduction in output by 0.2 percent.
    - Reduction in output by 0.7 percent five years after.
  - The effect is highly significant statistically and economically.
  - The medium-term output effect is almost twice the medium-term output effect of one-standard deviation increase in tariff rates found in Furceri et al. (2021).
  - Results are consistent with simulation results from a sectoral CGE model with input-output linkages (Caliendo, Feenstra, Romalis and Taylor, 2017).
- Subsample findings:
  - Effects of reducing trade barriers are statistically significant in both advanced economies (AEs) and emerging market and developing economies (EMDEs), but larger in EMDEs.
  - Component drivers:
    - In AEs: results driven mainly by export and import restrictions (import restriction statistically significant).
    - In EMDEs: results driven mainly by invisibles, exports, imports, and payments (all statistically significant).

### Potential channels of influence (alternative dependent variables)
- Variables examined: labor productivity, employment, inequality, trade balance, investment, consumption, exports, imports.
- Key channel findings:
  - Labor productivity declines by about 1.5 percent after five years following an increase in trade barriers.
  - Investment falls following increased trade barriers, consistent with reduced competition and lower incentives to invest.
  - Both imports and exports fall after increases in protectionism.
  - Consumption falls by around 1 percent after five years.
  - Increases in trade restrictions are associated with a reduction in inequality (consistent with trade liberalization generating winners and losers and redistribution effects).
- Interpretation:
  - Broad characteristics indicate trade restrictions resemble shocks to the productivity of the tradeables sector.
  - Reduced-form estimates do not permit full welfare analysis but indicate meaningful adverse economic effects.

### Components of MATR and alternative aggregations
- MATR composed of five main components:
  - (i) exchange measures
  - (ii) payment restrictions
  - (iii) import restrictions
  - (iv) export restrictions
  - (v) payment for invisibles
- Findings:
  - Increases in any component are associated with a decline in output.
  - Effects are not statistically different across components.

### Non-linearities and heterogeneity
- Theoretical hypotheses tested:
  - Complementarity between tariff and non-tariff restrictions: effect of non-tariff increases larger when tariffs are higher.
  - Larger effects for countries that import a larger share of intermediate inputs (greater GVC participation).
- Empirical specification:
  - Smooth transition model (equation (2)) with weight function F(zit) = exp(−zit)/(exp(−zit)+exp(zit)) and zit normalized to zero mean and unit variance.
  - Coefficients β1k and β2k capture output impact in low-z and high-z states, respectively.
- Estimation:
  - Equation (2) estimated with OLS and standard errors clustered at the country level.
  - Time and country sample reduced compared to equation (1) due to data availability for z.
- Conclusion:
  - Results obtained by estimating equation (2) confirm theoretical predictions (detailed analogues to Figure 11 reported in appendices).

### Data sources (summary)
- Core series:
  - GDP, labor productivity (GDP/employment), employment, investment, consumption: Penn World Table (PWT10.0).
  - Exports, imports, trade balance: World Economic Outlook (April 2021).
  - Gini coefficient: Standardized World Income Inequality Database (SWIID).
  - Regulation measures and job protection legislation: Alesina et al. (2019).
  - Global Value Chains (GVC): UNCTAD EORA database.
  - Tariffs: Furceri et al. (2021).
  - Country classification (income groups and regions): IMF World Economic Outlook.
- Sample and time coverage:
  - Unbalanced sample of 157 countries from 1949 to 2019.
  - MATR begins in 1949; Trade Freedom begins in 1995.

*IMF WORKING PAPERS A Measurement of Aggregate Trade Restrictions and their Economic Effects — content excerpt*

### Appendix 4). The decline in output following a one-standard deviation increase in MATR (excluding the tariff

### wpiea2022001-print-pdf - Appendix 4). The decline in output following a one-standard deviation increase in MATR (excluding the tariff

### Key empirical findings on MATR and output
- The decline in output following a one-standard deviation increase in MATR (excluding the tariff component) is larger when tariffs are high than when tariff rates are low, indicating complementarity effects of protectionism: non-tariff barriers hurt more when tariffs are higher.
- The effect of MATR is larger for countries with high GVC participation.
- Major trade restrictions are associated with significant output declines:
  - Around 0.35 percent decline in output 5 years after a major restriction (baseline large-change sample).
  - Output falls by almost 0.4 percent 5 years after when using exogenous changes identified via narrative approach and excluding structural reforms.

### Identification of large and plausibly exogenous MATR episodes
- Large-change criterion:
  - Episodes where changes in MATR exceed their average by at least two standard deviations (in practice, where MATR changes by more than 1.64).
  - This identifies 385 episodes: 123 of increased restrictions and 262 of liberalization; the majority occurred between 1989 and 2008.
- Crisis-filtered sample:
  - Restrict to episodes not preceded or succeeded by economic and financial crises in a one-year interval → 162 episodes (listed in Appendix 1, Table A1.4).
- Narrative-filtered exogenous sample:
  - Manual search of official national documents, IMF/WTO/World Bank reports, AREAER reports, trade institutes, policy papers and other online sources to find episodes with absence of discussion of counter-acting shocks or desire to close an output gap.
  - Produces 58 episodes: 7 increases in restrictions and 51 liberalizations (see Table A1.5 of Appendix 1).
  - Many of these episodes are associated with ideological/political changes, EU or WTO membership events, or import-substitution strategies (example: Costa Rica 1966).
- Further cleaning:
  - Exclude episodes that occurred during an IMF stabilization program and those associated with major product, domestic and external finance, and labor market reforms (dates from Alesina et al. 2020; see Table A.1.6).

### Robustness and causality checks
- Results from discrete-variable specification (1 for major restriction, -1 for major liberalization, 0 otherwise) are similar to but smaller than baseline continuous-change estimates (see Figure 14).
- Granger causality test:
  - No evidence that past GDP growth helps predict large changes in MATR; p-value for null that GDP growth Granger causes large changes in MATR is about 0.76.
- Non-linear specifications:
  - Results are robust to alternative non-linear specifications, e.g., (a) interaction between change in MATR and the level of tariff (GVC participation), or (b) interactions between change in MATR and dummies denoting quartiles of the country’s characteristics.

### MATR measure properties and summary conclusions
- MATR construction and coverage:
  - MATR covers an unbalanced sample of 157 countries annually between 1949 and 2019.
  - MATR is strongly correlated with existing measures of openness and trade policy and is more comprehensive than existing measures.
- Nature and limitations:
  - MATR is based on quantitative, aggregate trade policy inputs including tariffs and a host of non-tariff barriers drawn from transparent sources.
  - MATR is theoretically ad hoc and constructed from aggregate data (not a weighted average of disaggregated data); it is presented as a complementary, not replacement, measure of aggregate trade restrictions.
- Suggested directions for future research:
  - Consider impediments to FDI in addition to goods and services.
  - Consider lagged values or moving averages analogous to Chinn and Ito (2008) rather than contemporaneous values only.
  - Develop a rigorous theoretical model to understand general equilibrium and welfare consequences of trade restrictions.

### MATR data construction and coding rules (AREAER-based)
- Primary source:
  - IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER), extended back to 1949 via hard-copy examination and gap-filling when appropriate.
- Focus categories (and sub-components) from AREAER:
  - II (Exchange Measures);
  - VII (Imports and Import Payments): including (i) foreign exchange budget; (ii) financing requirement for imports; (iii) documentation requirements for release of foreign exchange for imports; (iv) import licenses and other nontariff measures; (v) import taxes and/or tariffs; (vi) state import monopoly.
  - VIII (Exports and Export Proceeds): including (i) repatriation requirements; (ii) financing requirements; (iii) documentation requirements; (iv) export licenses; (v) export taxes and/or tariffs.
  - IX (Payments for Invisible Transactions and Current Transfers);
  - IX & X (Payments and Proceed from Invisible Transactions and Current Transfers).
- Coding rules:
  - Restriction = 1, absence of restriction = 0, missing/unavailable = n/a.
  - Dating convention: AREAER country text dates are used; when country text states a future announced change, the announced date is used.
  - Granularity: collect data as granular as reasonably possible (e.g., VII_D_1 and VII_D).
  - Frequency: check for year-to-year changes via country text descriptions.
  - Absence of change indicates status quo (carry backward/forward when no information).
- Example of category nesting:
  - Category coIV_A_1_b_1 is a sub-component of coIV_A_1_b, itself a sub-component of coIV_A_1, a sub-component of coIV_A; unity in a sub-component implies unity in all higher-level aggregates.

### Select numerical and sample figures reported
- Large-change identification threshold: more than 1.64 (i.e., > |2σ| in practice).
- Episodes identified by threshold: 385 episodes (123 increases; 262 liberalizations).
- Episodes after excluding crises within one year: 162 episodes.
- Episodes after narrative filtering for exogeneity: 58 episodes (7 increases; 51 liberalizations).
- Output responses:
  - ~0.35 percent decline 5 years after major restriction (large-change sample).
  - Almost 0.4 percent decline 5 years after in narrative-exogenous sample excluding structural reforms.
- Coverage: 157 countries, 1949–2019.
- Granger causality test p-value: about 0.76 (null that GDP growth Granger causes large MATR changes).

*IMF WORKING PAPERS — A Measurement of Aggregate Trade Restrictions and their Economic Effects (Appendix excerpts as provided).*

### Appendix Table A1.3 Examples of Narrative Records in AREAER, 2016

### Appendix Table A1.3 Examples of Narrative Records in AREAER, 2016

### II. Exchange measures
- II.A. Restrictions and/or multiple currency practices — 2016, Report Page 22, Angola  
  - “Restrictions on the making of payments and transfers for current international transactions. ... [including]: (i) limits on the availability of foreign exchange for invisible transactions, such as travel, medical or educational allowances; and (ii) limits on unrequited transfers to foreign-based individuals and institutions. In addition, Angola maintained three exchange restrictions ... from: (i) the discriminatory application of the 0.015 percent stamp tax on foreign exchange operations; (ii) the operation of the priority list for access to US dollars at the official exchange rate; and (iii) a special tax of 10 percent on transfers to non-residents under contracts for foreign technical assistance or management services. Angola also maintains three multiple currency practices ...”

- II.B. Exchange measures imposed for security reasons — 2016, Report Page 252, The Bahamas  
  - “Restrictions on Bahamas-licensed banks and financial institutions regarding transactions with: (i) the Democratic Republic of the Congo; (ii) the Islamic Republic of Iran; (iii) the Democratic People’s Republic of Korea; and (iv) Osama bin Laden, the Al-Qaida organization ...”

### IV. Restrictions to payments
- IV.A. Prescription of currency requirements — 2016, Report Pages 617-618, Cameroon  
  - "the monetary unit is the CFA franc, the sole official currency and legal tender in all CEMAC members, and all transactions among residents are required to be settled in domestic currency"

- IV.B. Payments arrangements — 2016, Report Page 944, Denmark  
  - "Denmark had payments arrangements since it was bound by regional arrangements, as Denmark was a member of the EU."

- IV.C. Administration of control — 2016, Report Page 1017, Egypt  
  - "Egypt had administration of control since “Banks [were] authorized to execute foreign exchange transactions within the framework of a general authorization ...”"

- IV.D. Payment arrears — 2016, Report Page 1312, Guinea  
  - "Guinea had payments arrears “with various financial institutions"

- IV.F. Controls on exports and imports of banknotes — 2016, Report Page 316, Bangladesh  
  - "Residents and nonresidents may take out up to Tk 5,000 in domestic currency."

### VII. Import Restrictions
- VII.A. Foreign exchange budget — 2016, Report Page 1328, Guinea-Bissau  
  - "The country had a foreign exchange budget, meaning an a priori allocation of a certain amount of foreign exchange for the importation of specific types of goods.."

- VII.B. Financing requirements for imports — 2016, Report Page 1497, India  
  - "The country had a financing requirement for imports (minimum financing and advance payment requirements) though “The RBI allows requests from exporters through their AD Category-I banks to offset export receivables against import payables of the same foreign buyer and supplier, subject to certain terms and conditions.”"

- VII.C. Documentation requirements for release of forex for imports — 2016, Report Pages 1681-1682, Kazakhstan  
  - “... importers ... [were obligated to] submit to the bank a foreign trade contract or other corroborating document.”

- VII.D. Import licenses and other nontariff measures — 2016, Report Page 1812, Laos  
  - "The country required import licenses for 25 categories of goods, mostly for quality control, safety, or animal quarantine, but some (e.g., for cement) to limit the overall level of imports."

- VII.E. Import taxes and/or tariffs — 2016, Report Page 1941, Macedonia  
  - "The country had an average unweighted tariff rate for industrial products was 6.1 percent and for agricultural products 15.7 percent."

- VII.F. State Import Monopoly — 2016, Report Page 2359, Niger  
  - "The government of Niger, through the Société Nigérienne des Produits Pétroliers, had a monopoly on hydrocarbon imports."

### VIII. Export Restrictions
- VIII.A. Repatriation requirements — 2016, Report Page 341, Barbados  
  - "The country had repatriation requirements on export proceeds "Proceeds must be repatriated within six months.""

- VIII.B. Financing requirements — 2016, Report Page 1163, Eritrea  
  - "\"Exports may be made under LCs -Letters of Credit- on an advance-payment or consignment basis.\""

- VIII.C. Documentation requirements — 2016, Report Page 1210, Fiji  
  - "The country required a confirmation of receipt of export proceeds six months after export of goods."

- VIII.D. Export licenses — 2016, Report Page 931, Croatia  
  - "\"Exports are free of licensing requirements, except certain products (e.g., weapons, narcotics, dual-use items) for which permits must be obtained.\""

- VIII.E. Export taxes — 2016, Report Page 1970, Liberia  
  - "\"Export taxes are imposed on processed and unprocessed goods at rates of zero and 2.5 percent, respectively; on diamonds at a rate of 3 percent; and on iron ore at a rate of 4.5 percent.\""

### IX. Payments and X. Proceeds for Invisibles Restrictions
- IX.A. Payments for Invisibles, Transfers & Current Transfers — 2016, Report Page 3121, Sri Lanka  
  - "\"Indicative limits and bona fide tests are applied to all these transactions.\""

- X.A. Repatriation requirements on Proceeds — 2016, Report Page 3198, Suriname  
  - "The country had repatriation requirements on invisible proceeds."

- X.A.1. Surrender Requirements on Proceeds — 2016, Report Page 3578, Ukraine  
  - "The country had a 75 percent surrender requirement"

- X.B. Restrictions on use of funds — 2016, Report Page 767, China  
  - "The country had restriction on use of funds from invisibles"

*Source: Appendix Table A1.3, AREAER narrative examples, 2016 (from the supplied PDF content).*

### Appendix Table A1.5 Exogenous Changes in MATR

### Appendix Table A1.5 Exogenous Changes in MATR

### Enumerated episodes (selected entries from the table)
- Angola — 2018 — Liberalization  
  - The measure is created to facilitate long-term investment and not to solve short term economic issues: "On August 10, 2018, the Angolan Government enacted a private investment law aimed at facilitating investment. The law removed the previous requirement that foreign investors identify a local partner with a 35 percent stake prior to investing in priority sectors, thereby allowing foreign investors to own investments in their entirety. The law also eliminated minimum levels of foreign direct investment and established firm sunset clauses for tax incentives. In addition to changes to the investment legal framework, the government created the Agency for Private Investment and Exports Promotion, a state-run agency with the goal of facilitating investment and export processes."  
  - Link: https://ustr.gov/sites/default/files/files/reports/2021/2021NTE.pdf (p.16)

- Austria — 1955 — Liberalization  
  - Austrian Independence Treaty is signed.  
  - Link: https://2001-2009.state.gov/r/pa/ho/time/lw/107185.htm

- Austria — 1989 — Liberalization  
  - Austria applied for the EC in this year.  
  - Link: https://www.austria.org/austria-in-the-eu

- Bahrain — 2004 — Increase in restrictions  
  - Bahrain signs a Foreign Trade Agreement with the United States.  
  - Link: https://ustr.gov/about-us/policy-offices/press-office/press-releases/archives/2004/september/united-states-and-bahrain-sign-free-tra

- Belarus — 2006 — Liberalization  
  - Exchange Rate measures lifted. Previously, foreign exchange controls were in place to limit imports in the context of balance-of-payments problem.  
  - Link: https://hrcak.srce.hr/18634

- Belgium — 1990 — Liberalization  
  - Part of reforms related to The Schengen Agreement on the elimination of border checks, signed in 1985.  
  - Link: https://www.schengenvisainfo.com/schengen-agreement/

- Bosnia and Herzegovina — 1997 — Liberalization  
  - The country adopted a new law on Foreign Trade Policy (1997) focused on liberalizing laws. The laws aimed at improving existing and future economic collaboration between BIH and its entities, neighboring and other states and international organizations.  
  - Link: https://www.researchgate.net/publication/297550578_Non-tariff_barriers_and_their_impact_on_trade_flows_within_CEFTA_2006_The_case_study_of_Bosnia_and_Herzegovina

- Bosnia and Herzegovina — 2001 — Liberalization  
  - Part of EU ascension road map.  
  - Link: http://pdc.ceu.hu/archive/00003190/01/economic_chalenges_for_bosnia_and_herzegovina_on_the_road_to_membership_in_the_eu.pdf

- Botswana — 1999 — Liberalization  
  - The liberalization is part of an export-led strategy for the country that started in 1998. "Botswana has shown its commitment to promoting diversification of its economy through export-led industrialization (World Trade Organization, 1998b)."  
  - Link: https://www.researchgate.net/publication/287111055_The_Evolution_of_Trade_Policy_in_Botswana (p.23)

- Bulgaria — 2006 — Liberalization  
  - Reforms related to EU accession.  
  - Link: https://oxfordre.com/politics/view/10.1093/acrefore/9780190228637.001.0001/acrefore-9780190228637-e-508

- Costa Rica — 1966 — Increase in restrictions  
  - The country started a long-term plan to develop an Import Substitution Industrialization (ISI) to enhance growth in the country.  
  - Link: http://www.fao.org/3/I8308EN/i8308en.pdf

- Costa Rica — 2011 — Liberalization  
  - Foreign Trade Agreements with the European Union China and Singapore entered into force.  
  - Links: http://www.fao.org/3/I8308EN/i8308en.pdf ; https://www.econstor.eu/handle/10419/57577

- Croatia — 2003 — Liberalization  
  - Croatia signed its ascension to the Central European Union Free Trade Agreement in December 2002.  
  - Link: http://www.mvep.hr/en/foreign-politics/multilateral-relationsold/central-european-free-trade-agreement-(cefta)/

- Eritrea — 1996 — Liberalization  
  - Continuity of the trade liberalization process of Eritrea over the system it inherited from Ethiopia.  
  - Link: https://www.imf.org/external/pubs/ft/scr/1996/cr9666.pdf

- France — 1967 — Liberalization  
  - The European Communities signs the final Act of the General Agreement on Tariffs and Trade (GATT) multilateral negotiations (Kennedy round).  
  - Link: https://europa.eu/european-union/about-eu/history/1960-1969/1967_en

(Entries continue in the source for many other countries covering different years, signs, episode descriptions, and links.)

### Appendix Table A1.6 — Exogenous Changes in MATR, Not Part of Broader Reform Packages (ISO3, Year, Sign)
- AGO 2018 -1
- AUT 1955 -1
- AUT 1989 -1
- BGR 2006 -1
- BHR 2004 1
- BIH 1997 -1
- BIH 2001 -1
- BLR 2006 -1
- BWA 1999 -1
- CRI 1966 1
- CRI 2011 -1
- DEU 1958 -1
- ERI 1996 -1
- ESP 1989 -1
- FRA 1967 -1
- GAB 1968 1
- GMB 1989 -1
- GMB 1990 -1
- HRV 2003 -1
- HTI 1996 -1
- HUN 1998 -1
- HUN 2000 -1
- ISR 1996 -1
- JPN 2003 1
- KOR 1995 -1
- KOR 2016 -1
- LBR 1986 1
- MOZ 1995 -1
- MWI 1988 -1
- MWI 1997 1
- MWI 2003 -1
- MYS 2014 -1
- NAM 1995 -1
- NER 1996 -1
- NLD 1953 -1
- NLD 1954 -1
- NLD 1977 -1
- NLD 1992 -1
- PER 1960 -1
- PHL 1960 -1
- PHL 1966 1
- PRY 2007 -1
- SDN 1999 -1
- SVK 1990 -1
- SVK 1996 -1
- SVN 2003 -1
- SWE 1984 -1
- SWE 1989 -1
- SWE 1996 -1
- TJK 2002 -1
- UGA 1998 -1
- UKR 2018 -1

### Notes on content scope and structure from the source
- Appendix Table A1.5 provides country-level episodes of exogenous changes in MATR with: Country, Year, Sign (Liberalization / Increase in restrictions), a textual description of the episode, and a link to source documentation.
- Appendix Table A1.6 provides a compact ISO3-coded listing of episodes “Not Part of Broader Reform Packages” as three-column rows: Country Code (ISO3), Year of the Episode, Sign of the Change.

*Source: Appendix Table A1.5 and Appendix Table A1.6 from the PDF "A Measurement of Aggregate Trade Restrictions and their Economic Effects" (IMF Working Papers).*

### Appendix Table A2.2 Other Data

### Appendix Table A2.2 Other Data

### Database coverage and indicators
- Penn World Table (PWT 10.0)
  - 213 1950–2019: Real GDP. Constant 2017 prices
  - 212 1950–2019: Employment (no. of people employed)
  - 211 1950–2019: Productivity (Real GDP/ Employment)
  - 212 1950–2019: Investment. Constant 2017 prices
  - 213 1950–2019: GDP Deflator Index. 2017=100
  - 212 1950–2019: Household Consumption. Constant 2017 prices
- World Economic Outlook (April 2021)
  - 172 1950–2019: Trade Balance (Exports (real, LCU)-Imports (real, LCU)/ Real GDP)
  - 178 1950–2019: Volume of total exports of goods, USD
  - 178 1950–2019: Volume of total imports of goods, USD
- World Economic Outlook (WEO)
  - 196 1995–2020: Real GDP forecasted in that year Oct. WEO
- Furceri, Hannan, Ostry and Rose (2021)
  - 168 1960–2014: Tariffs
- UNCTAD EORA Database
  - 158 1990–2018: GVC
- World Bank (WDI Indicators)
  - 209 1960–2019: Total Population
  - 195 1960–2019: Trade %GDP
  - 195 1960–2019: Imports of G&S %GDP
  - 190 1990–2019: Real GDP per capita PPP
  - 187 1988–2018: Tariff rate applied weighted mean all products
- Systemic Peace: Polity IV project
  - 180 1950–2020: polity 2: Democracy-Autocracy Index
  - 180 1950–2020: Constraints on Executive (higher=more)
- The Standardized World Income Inequality Database (SWIID)
  - 194 1960–2019: Inequality in disposable (GINI net)
- Alesina et al. 2020
  - 90 1973–2014: Structural Reforms

### Index of Economic Freedom — Trade Freedom component
- Trade freedom (TF) is one component of “open markets” within the Heritage Foundation’s Index of Economic Freedom (IEF); TF is graded on a scale of 0 to 100.
- TF is described as a composite measure of the absence of tariff and non-tariff barriers that affect imports and exports of goods and services.
- Trade freedom score inputs:
  - The trade-weighted average tariff rate
  - Non-tariff barriers (NTBs)
- Scoring equation (as presented):
  - Trade Freedom_i = (((Tariff_max – Tariff_i) / (Tariff_max – Tariff_min)) * 100) – NTB_i
  - Tariff_min = 0 percent; Tariff_max = 50 percent
- NTB penalty scale (points subtracted from base score):
  - 20 — NTBs are used extensively across many goods and services and/or act to effectively impede a significant amount of international trade.
  - 15 — NTBs are widespread across many goods and services and/or act to impede a majority of potential international trade.
  - 10 — NTBs are used to protect certain goods and services and impede some international trade.
  - 5 — NTBs are uncommon, protecting few goods and services, and/or have very limited impact on international trade.
  - 0 — NTBs are not used to limit international trade.
- NTB categories considered in the penalty include:
  - Quantity restrictions (import quotas; export limitations; voluntary export restraints; import–export embargoes and bans; countertrade, etc.)
  - Price restrictions (antidumping duties; countervailing duties; border tax adjustments; variable levies/tariff rate quotas)
  - Regulatory restrictions (licensing; domestic content and mixing requirements; sanitary and phytosanitary standards (SPSs); safety and industrial standards regulations; packaging, labeling, and trademark regulations; advertising and media regulations)
  - Investment restrictions (exchange and other financial controls)
  - Customs restrictions (advance deposit requirements; customs valuation procedures; customs classification procedures; customs clearance procedures)
  - Direct government intervention (subsidies and other aid; government industrial policy and regional development measures; government-financed research and other technology policies; national taxes and social insurance; competition policies; immigration policies; government procurement policies; state trading, government monopolies, and exclusive franchises)
- Example: Botswana received a trade freedom score of 79.7; its weighted average tariff of 5.2 percent would have yielded a score of 89.7, but NTBs reduced the score by 10 points.
- Data sourcing and grading practices:
  - Uses the most recently reported weighted average tariff rate for a country from primary sources; may be reviewed if updated reliable information exists.
  - Prioritizes World Bank weighted average applied tariff rates; if unavailable, uses average applied tariff rate; if unavailable, uses weighted average or simple average of MFN tariff rates.
  - In rare cases where duties and customs revenues are unavailable, uses data on international trade taxes or an estimated effective tariff rate.
- Listed priority sources for trade policy scores (unless otherwise noted): World Bank, World Development Indicators 2012; World Trade Organization, Trade Policy Review, 1995–2012; Office of the U.S. Trade Representative, 2012 National Trade Estimate Report on Foreign Trade Barriers; World Bank, Doing Business 2011 and 2012; U.S. Department of Commerce, Country Commercial Guide, 2008–2012; Economist Intelligence Unit, Country Commerce, 2009–2012; World Bank, Data on Trade and Import Barriers: Trends in Average Applied Tariff Rates in Developing and Industrial Countries, 1981–2010; and official government publications of each country.
- Authors of this IMF appendix note difficulty obtaining underlying NTB data sources from the IEF authors despite repeated contact.

### Appendix III — MATR and the literature (summary)
- Conceptual taxonomy of aggregate trade restrictions:
  - a) openness (e.g., ratio of exports and imports to output)
  - b) trade costs (includes natural and artificial costs)
  - c) artificial trade barriers (policy-driven)
- Limitations of common measures:
  - Openness is affected by national tastes, technologies, institutions, and geography and thus conflates factors beyond trade policy.
  - Trade costs measures (e.g., Anderson and van Wincoop; Waugh and Ravikumar) include both natural and artificial costs; Novy (2012) produces a bilateral trade-costs measure intended to be comprehensive (natural + artificial), larger in scope than MATR.
  - Tariff-only measures are too narrow because they ignore NTBs; NTB-only measures are too narrow because they ignore tariffs and other policy elements.
  - Trade Restrictiveness Indices (TRIs) are theoretically attractive (Anderson and Neary) but difficult to implement due to data requirements; available TRIs are limited across countries and time.
  - Kee, Nicita and Olarreaga (2009) produce TRIs for 78 countries combining tariffs and ad-valorem equivalents of NTBs, but are limited in time coverage (different years between 2000 and 2004).
- Positioning MATR:
  - MATR is more narrowly focused than openness or trade costs, and broader than average tariff rate or NTB coverage ratio.
  - Advantages claimed for MATR:
    - a) simple
    - b) based on plausible, relevant policy inputs
    - c) quantitative and objective
    - d) timely and easily updateable
    - e) available for a large number of countries
    - f) available for a substantial period of time
    - g) covers both tariff and non-tariff barriers
  - Main substantive issue: theoretically ad hoc
  - Difference from many literature measures: not built from the ground up (not an aggregation of industry- or product-level measures)
  - Conclusion: MATR is a useful complementary measure of aggregate trade restrictions, not a panacea.

### Appendix IV — Robustness checks and extensions (figure summaries)
- Appendix Figure A4.1: Response of (log) GDP to Changes in MATR, by income group
  - Cumulative IRFs after one standard deviation increase in MATR; shaded area is 90 percent confidence interval; Driscoll-Kraay standard errors.
- Appendix Figure A4.2: Response of (log) GDP to Changes in MATR: Robustness Checks
  - Cumulative IRFs after one standard deviation increase in MATR; shaded area is 90 percent confidence interval; Driscoll-Kraay standard errors.
- Appendix Figure A4.3: Response of (log) GDP to Changes in MATR: Alternative Measures
  - Cumulative IRFs after one standard deviation increase in MATR; shaded area is 90 percent confidence interval; Driscoll-Kraay standard errors.
  - Variant 1 includes sum of: VII_A ;VII_B ;VII_C ;VII_D ;VII_E ;VII_F ;VIII_A ;VIII_B ;VIII_C ;VIII_D and VIII_E.
  - Variant 2 includes sum of: IV_A ;IV_B ;IV_C ;IV_D ;IV_E ;IV_F ;VII_A ;VII_B ;VII_C ;VII_D ;VII_E ;VII_F; IX_A_1; VIII_A ;VIII_B ;VIII_C ;VIII_D ;VIII_E ;X_A and X_B.
  - Variant 3 includes sum of: II_A ;II_B ;IV_A ;IV_B ;IV_C ;IV_D ;IV_E ;IV_F ;VII_A ;VII_B ;VII_C ;VII_D ;VII_E ;VII_F; IX_A_1 ;IX_A_2 ;IX_A_3 ;IX_A_4 ;IX_A_5 ;IX_A_6 ;IX_A_7 ;VIII_A ;VIII_B ;VIII_C ;VIII_D and VIII_E.
  - Variant 4 includes sum of: III_F ;III_G ;II_A ;II_B ;IV_A_1 ;IV_A_2 ;IV_B_1 ;IV_B_2 ;IV_C ;IV_D_1 ;IV_D_2 ;IV_E_1; IV_E_2 ;IV_F_1 ;IV_F_2 ;VII_B_1 ;VII_A ;VII_B_2 ;VII_B_3 ;VII_C_1 ;VII_C_2 ;VII_C_3 ;VII_C_4 ;VII_C_5 ;VII_D_1; VII_D_2 ;VII_D_3 ;VII_D_4 ;VII_D_5 ;VII_E_1 ;VII_F ;VIII_A_1 ;VIII_C_1 ;VIII_B ;VIII_C_2 ;VIII_C_3 ; VIII_C_4; VIII_C_5; VIII_D_1 ;VIII_D_2 ;VIII_E_1 ;VIII_E_2 ;IX_A_1 ;IX_A_2 ;IX_A_3 ;IX_A_4 ;IX_A_5 ;IX_A_6 ;IX_A_7 ;X_B; X_A_1_a and X_A_1_b.
- Appendix Figure A4.4: Response of (log) GDP to Changes in MATR, tariff complementarity (%)
  - Cumulative IRFs after one standard deviation increase in MATR; shaded area is 90 percent confidence interval; Driscoll-Kraay standard errors.
  - Heterogenous effects for tariffs calculated using interaction of smooth transition function of tariffs with MATR.
- Appendix Figure A4.5: Response of (log) GDP to Changes in MATR, role of GVCs (%)
  - Cumulative IRFs after one standard deviation increase in MATR; shaded area is 90 percent confidence interval; Driscoll-Kraay standard errors.
  - Heterogenous effects for GVC calculated using interaction of smooth transition function of GVC with MATR.

*Source: wpiea2022001-print-pdf - Appendix Table A2.2 Other Data*

### Appendix Figure A4.6: Response of (log) GDP to Large Changes in MATR (%)

### Appendix Figure A4.6: Response of (log) GDP to Large Changes in MATR (%)

### Figure description
- Cumulative IRFs after one standard deviation increase in MATR; shaded area is 90 percent confidence interval; Driscoll-Kraay standard errors.
- Title: Response of (log) GDP to Large Changes in MATR (%)

### Definition of "Large changes in MATR"
- Large changes in MATR defined as changes in index > |2σ| for all countries and time.
- Coding rule: 1 ≡ increase in restrictions; -1 ≡ liberalization; 0 ow.

### Estimation and inference details
- Impulse responses are cumulative and computed after a one standard deviation increase in MATR.
- Confidence bands shown are 90 percent confidence interval.
- Standard errors are Driscoll-Kraay standard errors.

### Key terms and notation preserved exactly
- MATR
- IRFs
- one standard deviation
- 90 percent confidence interval
- Driscoll-Kraay standard errors
- > |2σ|
- 1 ≡ increase in restrictions
- -1 ≡ liberalization
- 0 ow

*Source: Appendix Figure A4.6, A Measurement of Aggregate Trade Restrictions and their Economic Effects, Working Paper No. WP/22/1*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022001-print-pdf.pdf_