## _wp0814 - 1. Trade Liberalization Episodes (Measure of Liberalization Dates Based on Li (2004))

## Source details

**Canonical URL:** [_wp0814 - 1. Trade Liberalization Episodes (Measure of Liberalization Dates Based on Li (2004))](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0814.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0814.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0814.pdf.json)

---

### Introduction and research question
- Primary research question: did past liberalization episodes in developing countries lead to a deterioration of their trade balance?
- Theoretical background: no clear theoretical prediction for the effect of a tariff change on the trade balance; effect depends on real wages, exchange rates, various elasticities, degree of capital mobility, and whether the tariff shock is perceived as temporary or permanent.
- Empirical motivation: limited cross-country empirical studies on the impact of trade liberalization on the overall trade balance for developing countries.

### Data, measures, and sample
- Two measures of trade liberalization dates used:
  - First measure: based on Li (2004), extended to 2004 using UNCTAD’s TRAINS and IMF’s TPID databases.
    - Liberalization episode: continuous and accumulated tariff reduction by at least 35 percent.
    - Once overall tariff level reaches 10 percent or lower, further cuts not considered a liberalization episode.
    - Non-tariff barrier ratings (open, moderate, restrictive) considered; reductions in NTBs usually accompanied by large tariff cuts.
    - For countries with multiple liberalization episodes, a dummy = 1 after end of last recorded episode (exception: China assigned dummy = 1 after 1998).
  - Second measure: Wacziarg and Welch (2003).
    - Liberalization date: date after which all Sachs and Warner (1995) openness criteria are continuously met.
    - A country classified as closed if it displays at least one characteristic: (i) average tariff rates of 40% or more; (ii) nontariff barriers covering 40% or more of trade; (iii) black market exchange rate depreciated by 20% or more relative to official rate on average; (iv) state monopoly on major export; (v) socialist economic system.
    - Relied on country case studies in many instances; covers a substantially larger sample.
- Sample coverage:
  - First sample: 39 developing countries for the period 1970–2004.
  - Second sample: 77 developing countries for the period 1970–2001.
- The two measures are significantly and positively correlated, with a correlation coefficient of 0.57 (for countries in which they overlap).

### Summary statistics: imports, exports, trade balance before and after liberalization
- Using the extended Li measure (1970–2004):
  - Cross-country average Imports/GDP: increased from 23.8% to 30.6%.
  - Cross-country average Exports/GDP: increased from 19.5% to 24.1%.
  - Cross-country average Trade balance/GDP: changed from -4.3% to -6.5%.
  - Before<After counts: Imports 33, Exports 28, Trade balance 15.
  - Before>After counts: Imports 4, Exports 9, Trade balance 22.
  - Narrative counts:
    - 33 countries saw import-to-GDP ratio increase versus 4 countries experiencing a decline.
    - 28 countries saw export-to-GDP ratio increase versus 9 with a reduction.
    - 22 experienced deterioration of trade balance after liberalization and 15 experienced improvement.
- Using the Wacziarg-Welch measure (1970–2001):
  - Average Imports/GDP: increased from 25.1% to 29.9%.
  - Average Exports/GDP: increased from 18.5% to 20.4%.
  - Average Trade balance/GDP: changed from -6.5% to -9.5%.
  - Before<After counts: Imports 47, Exports 40, Trade balance 21.
  - Before>After counts: Imports 15, Exports 22, Trade balance 41.
  - Narrative counts:
    - 47 out of 62 developing countries had higher import-to-GDP ratios after liberalization.
    - 40 countries experienced export ratio increases and 22 decreases.
    - 41 out of 62 countries experienced a worsening of their trade balance.

### Regression analysis — specification and data
- Estimated dynamic panel equation (symbols as in source):
  - TBit = α + β1 TBit-1 + β2 libit + β3 ˆyit + β4 *ˆyit + β5 ˆreerit + β6 ˆTOTit + β7 fisrit + iui + vit
    - TB = trade balance; lib = trade liberalization dummy; ˆy and *ˆy = domestic and foreign real GDP growth; ˆreer = change in (log) real exchange rate; ˆTOT = change in terms of trade; fisr = fiscal balance to GDP ratio; iui = country fixed effects; vit = disturbance term.
- Data sources:
  - Trade, GDP, fiscal balance: IMF’s International Financial Statistics (IFS).
  - Terms of trade: IMF’s WEO.
  - Foreign real GDP growth: weighted growth rates of export market countries (weights = market country’s 1990 share of home country’s total exports); bilateral trade data from IMF’s Direction of Trade Statistics.
  - Real exchange rate: geometric weighted average of bilateral real exchange rates; major trading partners included if trade share > 10% (China excluded due to incomplete CPI data).
- Import and export regressions use log(imports/GDP) and log(exports/GDP) to preserve sample size.

### Impact of trade liberalization on imports — key regression results
- Table 3a (Extended Li measure, 39 countries, 1202 observations) — selected coefficients (exact values preserved):
  - Lagged dependent var.: 0.778*** (Fixed effects); GMM: 0.897***, 0.854***, 0.883***, 0.859***.
  - Trade liberalization: 0.082*** (Fixed effects); GMM: 0.037***, 0.047***, 0.041*, 0.043**.
  - Domestic GDP growth: 0.005*** (Fixed effects); GMM: 0.005**, 0.004, 0.005*, 0.004.
  - Change in real effective exchange rate: -0.115*** (Fixed effects); GMM: -0.141***, -0.127***, -0.135**, -0.124**.
  - Changes in terms of trade: -0.001*** (Fixed effects); GMM: -0.001***, -0.002**, -0.002**, -0.002**.
  - Fiscal balance/GDP: 0.003* (Fixed effects); GMM: 0.001, 0.00, 0.001, 0.003 (insignificant in GMM).
  - Arellano-Bond AR(2) p values: 0.36–0.37 across GMM columns.
  - Hansen test p value: 1.00 for reported specifications.
  - Interpretation: trade liberalization leads to higher imports (positive and significant across specifications); domestic growth raises import-to-GDP ratio; real exchange rate appreciation and improved terms of trade reduce imports.
- Table 3b (Wacziarg-Welch measure, 77 countries, 2039 observations) — selected coefficients:
  - Lagged dependent var.: 0.767*** (Fixed effects); GMM: 0.811***, 0.793***, 0.812***, 0.787***.
  - Trade liberalization: 0.061*** (Fixed effects); GMM: 0.069***, 0.074***, 0.070***, 0.077***.
  - Domestic GDP growth: 0.003*** (Fixed effects); GMM: 0.003*, 0.002, 0.002, 0.001.
  - Change in real effective exchange rate: -0.124*** (Fixed effects); GMM: -0.127**, -0.120**, -0.126**, -0.118**.
  - Changes in terms of trade: -0.001*** (Fixed effects); GMM: -0.001**, -0.001**, -0.001**, -0.001**.
  - Fiscal balance/GDP: -0.001 (Fixed effects); GMM: -0.001 (insignificant).
  - Arellano-Bond AR(2) p values: 0.43–0.46.
  - Hansen test p values: 1.00, 0.93 across specifications.
  - Interpretation: results broadly similar to Table 3a; trade liberalization dummy positive and significant at the 1% level in all specifications; estimated coefficients larger than in Table 3a.

### Impact of trade liberalization on exports — key regression results
- General patterns:
  - Higher foreign growth and terms of trade improvement lead to higher exports; real exchange rate appreciation lowers exports.
- Trade liberalization dummy:
  - Positive and significant at the 5% or 10% level in all regressions except the fixed effects regression when the Wacziarg-Welch dates are used.
  - Coefficients from export regressions (one-step GMM (2, 3)): 0.030 and 0.036 for the two measures, respectively.
  - Corresponding import coefficients (one-step GMM (2, 3)): 0.047 and 0.074, respectively.
  - Interpretation: liberalization may lead to higher import growth than export growth, possibly leading to a deterioration in the overall trade balance.
- Select coefficients from Table 4a (Extended Li measure, 39 countries, 1203 observations):
  - Trade liberalization: 0.040*** (Fixed effects); GMM: 0.025**, 0.030**, 0.026**, 0.033*.
  - Foreign GDP growth: 0.007*, 0.011***, 0.010***, 0.013***, 0.011***.
  - Change in real effective exchange rate: -0.225***, -0.217***, -0.199**, -0.207**, -0.219**.
  - Changes in terms of trade: 0.003*** across specifications.
- Select coefficients from Table 4b (Wacziarg-Welch measure, 77 countries, 2055 observations):
  - Trade liberalization: 0.019 (Fixed effects); GMM: 0.028**, 0.036**, 0.031**, 0.042**.
  - Foreign GDP growth: 0.005, 0.008***, 0.007**, 0.008***, 0.007**.
  - Change in real effective exchange rate: -0.226***, -0.224***, -0.214***, -0.226***, -0.214***.
  - Changes in terms of trade: 0.003*** across specifications.

### Impact of trade liberalization on the trade balance — key regression results
- Using the Li measure (Table 5a, Extended Li measure, 1970–2004; 39 countries, 1202 observations):
  - Control variables:
    - Domestic GDP growth: negative and significant.
    - Foreign GDP growth: positive, significant in fixed effects and one-step GMM.
    - Change in real effective exchange rate: negative but insignificant.
    - Changes in terms of trade: consistently positive and significant (0.086***, 0.089***, 0.083***, 0.080***, 0.079***).
    - Fiscal balance/GDP: positive but significant only in one-step GMM (0.090* and 0.116* in some specs).
  - Trade liberalization dummy:
    - Fixed effects: -0.951***.
    - GMM regressions: negative but insignificant (-0.395, -0.588, -4.591, -0.259).
    - Conclusion: little support that liberalization has a negative and significant impact on the overall trade balance when using the Li measure.
  - Lagged dependent var.: 0.694***, 0.883***, 0.787***, 0.919***, 0.742*** across specifications.
- Using the Wacziarg-Welch measure (Table 5b, 1970–2001; 77 countries, 2039 observations):
  - Control variables:
    - Higher domestic GDP growth leads to deterioration of the trade balance (e.g., -0.106***, -0.079**, -0.077*, -0.083**, -0.085**).
    - Higher foreign GDP growth improves the trade balance (e.g., 0.217***, 0.214***, 0.237***, 0.211***, 0.227***).
    - Real exchange rate appreciation tends to deteriorate the trade balance (e.g., -1.887***, -1.910, -2.033*, -1.953, -2.235**).
    - Changes in terms of trade: positive and significant (0.076***, 0.082***, 0.079***, 0.082***, 0.078***).
    - Fiscal balance/GDP: positive, significant only in fixed effects (0.071*** in column (1)).
  - Trade liberalization dummy:
    - Negative and significant in all specifications except the standard two-step GMM estimate.
    - Coefficients: -1.260*** (Fixed effects), -0.844**, -1.300***, -0.843, -1.249***.
    - One-step GMM (2,3) estimate: -1.30 (an immediate worsening of the trade balance-to-GDP ratio of 1.3 percent after liberalization).
    - Comparison: the -1.30 estimate is substantially smaller than Santos-Paulino and Thirlwall (2004) estimates of -2.52 and -3.57.
  - Lagged dependent var.: 0.637***, 0.842***, 0.752***, 0.839***, 0.754*** across specifications.
- Robustness and coverage checks:
  - Re-running trade balance regressions using Wacziarg-Welch dates but limited to the 39 Li countries yields broadly similar results to Table 5b (trade liberalization dummy negative and significant in all specs except the standard two-step GMM).
  - Suggests differences between Tables 5a and 5b are more likely from differences in the measure of liberalization dates than from country coverage differences.

### Concluding remarks — summary of empirical findings
- Robust and consistent evidence using both measures that trade liberalization in developing countries promotes both imports and exports.
- Impact on the overall trade balance is mixed and depends on the liberalization date measure:
  - Using the extended Li (2004) measure: little evidence of a statistically significant negative impact on the overall trade balance.
  - Using the Wacziarg-Welch measure: some evidence that liberalization worsens the trade balance, though not robust across all estimation specifications and quantitatively smaller than some previous estimates (e.g., Santos-Paulino and Thirlwall (2004)).

*Section 3 of _wp0814 (PDF filename: _wp0814 - section 3. Finally, section 4 provides some concluding remarks.)*

### 1.   Trade Liberalization Episodes (Measure of Liberalization Dates Based on Li (2004)) .......... 6

### _wp0814 - 1.   Trade Liberalization Episodes (Measure of Liberalization Dates Based on Li (2004)) .......... 6

### Introduction and Research Question
- Many developing countries have substantially liberalized their trade regime over the past three decades, either unilaterally or as part of multilateral initiatives; trade barriers remain high in many developing countries.
- Primary research question: did past liberalization episodes in developing countries lead to a deterioration of their trade balance?
- Theoretical background (Ostry and Rose (1992)): no clear theoretical prediction for the effect of a tariff change on the trade balance; the effect depends on real wages, exchange rates, various elasticities, degree of capital mobility, and whether the tariff shock is perceived as temporary or permanent.
- Empirical motivation: limited cross-country empirical studies on the impact of trade liberalization on the overall trade balance for developing countries.

### Related empirical literature (select findings cited)
- Studies finding positive impact of liberalization on imports: Melo and Vogt, 1984; Bertola and Faini, 1991; Santos-Paulino, 2002a.
- Mixed findings on exports: improvements found in Ahmed, 2000; Thomas et al., 1991; Santos-Paulino, 2002b; little evidence in Greenaway and Sapsford, 1994; Jenkins, 1996.
- Cross-country studies on trade balance:
  - Ostry and Rose (1992): no statistically significant effect (mostly OECD datasets).
  - UNCTAD (1999): significant negative relationship for 15 developing countries, 1970–1995.
  - Santos-Paulino and Thirlwall (2004): sample of 22 developing countries, 1972–1997, found liberalization stimulated export growth but raised import growth by more, worsening the overall trade balance.
- Fiscal concern: reductions in tariff revenue—often a major revenue source for developing countries—are a common policy concern. Baunsgaard and Keen (2005) find that low-income countries have mostly not been able to offset reductions in trade tax revenues by increasing their domestic tax revenues.

### Data, measures, and sample
- A key empirical constraint is lack of systematic data measuring dates of trade liberalization.
- This paper uses two recently compiled data sets establishing trade liberalization dates that cover a large sample of developing countries for a long period.
- Sample coverage:
  - First sample: 39 developing countries for the period 1970–2004.
  - Second sample: 77 developing countries for the period 1970–2001.
- Paper structure: Section 2 discusses the two measures of liberalization dates and reports some stylized facts; estimation results follow in subsequent sections.

### Main empirical findings (as reported)
- Trade liberalization leads to faster import growth.
- Trade liberalization leads to faster export growth.
- Evidence on the overall trade balance is mixed:
  - Using the first measure of trade liberalization dates, there is little evidence that trade liberalization worsens the trade balance.
  - Using the second measure of liberalization dates, there is some evidence that liberalization leads to a deterioration of the trade balance, although this finding is not robust to alternative estimation specifications.

*Source: _wp0814 - 1.   Trade Liberalization Episodes (Measure of Liberalization Dates Based on Li (2004)) .......... 6*

### section 3. Finally, section 4 provides some concluding remarks.

### _wp0814 - section 3. Finally, section 4 provides some concluding remarks.

### II. TWO MEASURES OF TRADE LIBERALIZATION DATES
- First measure: based on Li (2004), extended for the 39 developing countries in her data set to 2004 using UNCTAD’s TRAINS database and IMF’s TPID database.
  - A trade liberalization episode is identified if there is a continuous and accumulated tariff reduction by at least 35 percent (example: a tariff reduction from 15% to 9.75%).
  - Once a country’s overall tariff level reaches 10 percent or lower, it is regarded as open and a further tariff cut, even by more than 35 percent, is not considered a liberalization episode.
  - IMF TPID non-tariff barrier ratings (open, moderate, restrictive) are considered; reductions in non-tariff barriers are usually accompanied by large tariff cuts.
  - For countries with multiple liberalization episodes, a trade liberalization dummy is defined that takes the value of one after the end of the last recorded liberalization episode and zero beforehand. (One exception made: China assigned liberalization dummy = 1 after 1998 due to dramatic 1990s opening up though strict application of rules would make dummy zero.)
- Second measure: Wacziarg and Welch (2003)
  - Liberalization date defined as the date after which all Sachs and Warner (1995) openness criteria are continuously met.
  - A country is classified as closed if it displays at least one characteristic: (i) average tariff rates of 40% or more; (ii) nontariff barriers covering 40% or more of trade; (iii) black market exchange rate depreciated by 20% or more relative to official rate on average; (iv) state monopoly on major export; (v) socialist economic system.
  - Data limitations meant reliance on country case studies in many instances.
  - Advantage: covers a substantially larger sample of developing countries.
- Comparison of the two measures:
  - Many identified dates are very close across measures (example: Argentina 1992 in first measure vs. 1991 in Wacziarg and Welch).
  - For countries with multiple episodes, Wacziarg-Welch date sometimes closer to the first episode (example: Chile — Li measure: episodes 1974–79 and 1985–92, first liberalization dummy = 1 starting 1993; Wacziarg-Welch identifies 1976 as open, missing later reversal).
  - In some cases, dates differ substantially (example: Indonesia — Li identifies liberalization era 1985–1996; Wacziarg-Welch classifies Indonesia as open from 1970).
  - The two measures are significantly and positively correlated, with a correlation coefficient of 0.57 (for countries in which they overlap).

### Summary statistics: Imports, Exports, Trade Balance Before and After Liberalization
- Using the extended Li measure (1970–2004) — Table 2a (cross-country averages and counts preserved exactly):
  - Cross-country average Imports/GDP increased from 23.8% to 30.6%.
  - Cross-country average Exports/GDP increased from 19.5% to 24.1%.
  - Cross-country average Trade balance/GDP changed from -4.3% to -6.5%.
  - Before<After counts: Imports 33, Exports 28, Trade balance 15.
  - Before>After counts: Imports 4, Exports 9, Trade balance 22.
  - Narrative findings: 33 countries saw import-to-GDP ratio increase versus 4 countries experiencing a decline; 28 countries saw export-to-GDP ratio increase versus 9 with a reduction; 22 countries experienced deterioration of trade balance after liberalization and 15 experienced improvement.
- Using the Wacziarg-Welch measure (1970–2001) — Table 2b (averages preserved exactly):
  - Average Imports/GDP increased from 25.1% to 29.9%.
  - Average Exports/GDP increased from 18.5% to 20.4%.
  - Average Trade balance/GDP changed from -6.5% to -9.5%.
  - Before<After counts: Imports 47, Exports 40, Trade balance 21.
  - Before>After counts: Imports 15, Exports 22, Trade balance 41.
  - Narrative findings: 47 out of 62 developing countries had higher import-to-GDP ratios after liberalization; 40 countries experienced export ratio increases and 22 decreases; 41 out of 62 countries experienced a worsening of their trade balance.

### III. REGRESSION ANALYSIS — Specification and data
- Dependent variable: trade balance over GDP (dynamic panel).
- Estimated equation (symbols preserved as in source):
  - TBit = α + β1 TBit-1 + β2 libit + β3 ˆyit + β4 *ˆyit + β5 ˆreerit + β6 ˆTOTit + β7 fisrit + iui + vit
  - Where: TB = trade balance; lib = trade liberalization dummy; ˆy and *ˆy = domestic and foreign real GDP growth respectively; ˆreer and ˆTOT = change in (log) real exchange rate and terms of trade; fisr = fiscal balance to GDP ratio; iui = time-invarying country-specific effects; vit = disturbance term.
- Data sources:
  - Trade, GDP, fiscal balance: IMF’s International Financial Statistics (IFS).
  - Terms of trade: IMF’s WEO.
  - Foreign real GDP growth: weighted growth rates of export market countries; weights = market country’s 1990 share of home country’s total exports; bilateral trade data from IMF’s Direction of Trade Statistics.
  - Real exchange rate: geometric weighted average of bilateral real exchange rates; formula preserved from source; major trading partners included if trade share > 10% (China excluded due to incomplete CPI data).
- Import and export analyses:
  - Dependent variables: log of import-to-GDP ratio and log of export-to-GDP ratio to preserve sample size (use of volumes would reduce sample due to missing price data).
  - Using ratios in level yields broadly similar results (source note).

### III.B. Impact of trade liberalization on imports — Regression results
- Table 3a: Using extended Li measure (39 countries, 1202 observations)
  - Fixed effects benchmark (column 1) and system GMM estimates (columns 2–5) reported.
  - Key coefficient signs and significance (preserved exactly as in Table 3a):
    - Lagged dependent var.: 0.778*** (Fixed effects) ; GMM estimates: 0.897***, 0.854***, 0.883***, 0.859*** (with standard errors reported in table).
    - Trade liberalization: 0.082*** (Fixed effects) ; GMM: 0.037***, 0.047***, 0.041*, 0.043**.
    - Domestic GDP growth: 0.005*** (Fixed effects) ; GMM: 0.005**, 0.004, 0.005*, 0.004.
    - Change in real effective exchange rate: -0.115*** (Fixed effects) ; GMM: -0.141***, -0.127***, -0.135**, -0.124**.
    - Changes in terms of trade: -0.001*** (Fixed effects) ; GMM: -0.001***, -0.002**, -0.002**, -0.002**.
    - Fiscal balance/GDP: 0.003* (Fixed effects) ; GMM: 0.001, 0.00, 0.001, 0.003 (becomes insignificant in GMM).
  - Tests:
    - Arellano-Bond test for AR(2) in first differences (p value): 0.36–0.37 (values reported across GMM columns).
    - Hansen test of joint validity of instruments (p value): 1.00 for reported specifications.
  - Interpretation and robustness:
    - Trade liberalization leads to higher imports (positive and significant across specifications).
    - Domestic growth raises import-to-GDP ratio, suggesting an income elasticity larger than one in some specifications.
    - Real exchange rate appreciation and improved terms of trade (lower import prices) reduce imports (price elasticity lower than one).
    - Fiscal balance positive sign in fixed effects is puzzling; becomes insignificant in GMM.
    - System GMM (Blundell and Bond 1998) used for consistent estimates in dynamic panel; both one-step and two-step presented; instrument sets varied (all available lags vs. lags t-2 and t-3 labeled GMM(2, 3)).
    - Arellano-Bond and Hansen tests indicate no evidence against specification in reported models.
- Table 3b: Using Wacziarg-Welch measure (77 countries, 2039 observations)
  - Key coefficients preserved exactly as in Table 3b:
    - Lagged dependent var.: 0.767*** (Fixed effects) ; GMM: 0.811***, 0.793***, 0.812***, 0.787***.
    - Trade liberalization: 0.061*** (Fixed effects) ; GMM: 0.069***, 0.074***, 0.070***, 0.077***.
    - Domestic GDP growth: 0.003*** (Fixed effects) ; GMM: 0.003*, 0.002, 0.002, 0.001.
    - Change in real effective exchange rate: -0.124*** (Fixed effects) ; GMM: -0.127**, -0.120**, -0.126**, -0.118**.
    - Changes in terms of trade: -0.001*** (Fixed effects) ; GMM: -0.001**, -0.001**, -0.001**, -0.001**.
    - Fiscal balance/GDP: -0.001 (Fixed effects) ; GMM: -0.001 (insignificant across specifications).
  - Tests:
    - Arellano-Bond AR(2) p values: 0.43–0.46.
    - Hansen test p values: 1.00, 0.93 across specifications.
  - Interpretation:
    - Results broadly similar to Table 3a.
    - Trade liberalization dummy is positive and significant at the 1% level in all specifications; estimated coefficients larger than those in Table 3a.
    - Fiscal balance becomes negative as expected (though insignificant) in this larger sample.
- Additional methodological notes:
  - Concerns about instrument proliferation addressed by reporting GMM estimates using limited lags (t-2 and t-3).
  - Robust standard errors reported; two-step estimates use Windmeijer (2005) correction.

*Italic source attribution: Section 3 of _wp0814 (PDF filename: _wp0814 - section 3. Finally, section 4 provides some concluding remarks.)*

### 0.074 vs. 0.047 in Table 3a.

### _wp0814 - 0.074 vs. 0.047 in Table 3a.

### C. Impact of trade liberalization on exports
- Main pattern: higher foreign growth and terms of trade improvement lead to higher exports; real exchange rate appreciation lowers exports.
- Trade liberalization dummy:
  - Positive and significant at the 5% or 10% level in all regressions except the fixed effects regression when the Wacziarg-Welch trade liberalization dates are used.
  - Coefficients from export regressions (one-step GMM (2, 3)): 0.030 and 0.036 for the two measures of liberalization dates, respectively.
  - Corresponding coefficients from import regressions (one-step GMM (2, 3)): 0.047 and 0.074, respectively.
  - Interpretation: liberalization may lead to higher import growth than export growth, possibly leading to a deterioration in the overall trade balance.
- Select coefficients from Table 4a (The Extended Li measure, 1970–2004):
  - Trade liberalization: 0.040*** (Fixed effects), 0.025**, 0.030**, 0.026**, 0.033* (GMM specifications).
  - Foreign GDP growth: 0.007*, 0.011***, 0.010***, 0.013***, 0.011***.
  - Change in real effective exchange rate: -0.225***, -0.217***, -0.199**, -0.207**, -0.219**.
  - Changes in terms of trade: 0.003*** across specifications.
  - Sample: No. of countries 39; No. of obs. 1203.
- Select coefficients from Table 4b (The Wacziarg-Welch measure, 1970–2001):
  - Trade liberalization: 0.019 (Fixed effects), 0.028**, 0.036**, 0.031**, 0.042** (GMM specifications).
  - Foreign GDP growth: 0.005, 0.008***, 0.007**, 0.008***, 0.007**.
  - Change in real effective exchange rate: -0.226***, -0.224***, -0.214***, -0.226***, -0.214***.
  - Changes in terms of trade: 0.003*** across specifications.
  - Sample: No. of countries 77; No. of obs. 2055.

### D. Impact of trade liberalization on the trade balance
- Using the Li measure (Table 5a, Extended Li measure, 1970–2004):
  - Control variables:
    - Domestic GDP growth: negative and significant.
    - Foreign GDP growth: positive, significant in fixed effects and one-step GMM.
    - Change in real effective exchange rate: negative but insignificant.
    - Changes in terms of trade: consistently positive and significant (0.086***, 0.089***, 0.083***, 0.080***, 0.079***).
    - Fiscal balance/GDP: positive but significant only in one-step GMM (0.090* and 0.116* in some specs).
  - Trade liberalization dummy:
    - Fixed effects: -0.951***.
    - GMM regressions: negative but insignificant (-0.395, -0.588, -4.591, -0.259).
    - Conclusion: little support that liberalization has a negative and significant impact on the overall trade balance when using the Li measure.
  - Selected figures from Table 5a:
    - Lagged dependent var.: 0.694***, 0.883***, 0.787***, 0.919***, 0.742***.
    - No. of countries 39; No. of obs. 1202.
- Using the Wacziarg-Welch measure (Table 5b, 1970–2001):
  - Control variables:
    - Higher domestic GDP growth leads to deterioration of the trade balance (e.g., -0.106***, -0.079**, -0.077*, -0.083**, -0.085**).
    - Higher foreign GDP growth improves the trade balance (e.g., 0.217***, 0.214***, 0.237***, 0.211***, 0.227***).
    - Real exchange rate appreciation tends to deteriorate the trade balance (e.g., -1.887***, -1.910, -2.033*, -1.953, -2.235**).
    - Changes in terms of trade: positive and significant (0.076***, 0.082***, 0.079***, 0.082***, 0.078***).
    - Fiscal balance/GDP: positive, significant only in fixed effects (0.071*** in column (1)).
  - Trade liberalization dummy:
    - Negative and significant in all specifications except the standard two-step GMM estimate.
    - Coefficients: -1.260*** (Fixed effects), -0.844**, -1.300***, -0.843, -1.249***.
    - One-step GMM (2,3) estimate: -1.30 (an immediate worsening of the trade balance-to-GDP ratio of 1.3 percent after liberalization).
    - Comparison: the -1.30 estimate is substantially smaller than Santos-Paulino and Thirlwall (2004) estimates of -2.52 and -3.57.
  - Selected figures from Table 5b:
    - Lagged dependent var.: 0.637***, 0.842***, 0.752***, 0.839***, 0.754***.
    - No. of countries 77; No. of obs. 2039.
- Robustness and coverage checks:
  - Re-running trade balance regressions using Wacziarg-Welch dates but limited to the 39 Li countries yields broadly similar results to Table 5b (trade liberalization dummy negative and significant in all specs except standard two-step GMM).
  - Suggests differences between Tables 5a and 5b are more likely from differences in the measure of liberalization dates than from country coverage differences.

### IV. Concluding remarks — summary of empirical findings
- Robust and consistent evidence using both measures that trade liberalization in developing countries promotes both imports and exports.
- Impact on the overall trade balance is mixed and depends on the liberalization date measure:
  - Using the extended Li (2004) measure: little evidence of a statistically significant negative impact on the overall trade balance.
  - Using the Wacziarg-Welch measure: some evidence that liberalization worsens the trade balance, though not robust across all estimation specifications and quantitatively smaller than some previous estimates (e.g., Santos-Paulino and Thirlwall (2004)).

*Source: _wp0814 - 0.074 vs. 0.047 in Table 3a.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp0814.pdf_