## _wp0520 - 1. Correlations of Tariffs Over Time

## Source details

**Canonical URL:** [_wp0520 - 1. Correlations of Tariffs Over Time](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0520.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0520.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0520.pdf.json)

---

### Introduction: scope and key conclusions
- Paper examines impact of India’s 1991 trade liberalization on industry wage premiums using micro-level NSSO data (1980–2000) combined with industry trade protection data.
- Main empirical finding: a strong and negative relationship between changes in trade policy (tariffs) and changes in industry wage differentials; robust to instrumenting for tariffs and to inclusion of nontariff barrier measures.
- Distributional implication: because tariff reductions were relatively larger in sectors with a higher proportion of unskilled workers and these sectors experienced an increase in relative wages, unskilled workers experienced an increase in incomes relative to skilled workers—suggesting trade liberalization decreased wage inequality in India.

### Background of India’s trade liberalization
- Pre-1991 regime: import-substitution strategy with high nominal tariffs and extensive nontariff barriers (NTBs) and complex import licensing.
- Balance of payments crisis in 1991: foreign currency assets less than US$1 billion, just enough to cover two weeks of imports.
- IMF-supported program (Stand-By-Arrangement) in August 1991 tied to structural reforms; policy changes included elimination of state trading monopolies, easing import licensing, removing export restrictions, allowing foreign investment in previously reserved sectors, and full convertibility of domestic currency on foreign exchange transactions.
- EXIM policy 1992–97: eliminated all import licensing lists and established a “negative” list; by April 2002 all remaining quantitative restrictions were removed.

### Quantitative description of policy change and protection levels
- Manufacturing average tariff: increased from 86 percent in 1980–81 to 117 percent in 1990–91, then declined to 39 percent in 1999–2000.
- Import coverage ratio (measure of NTBs): declined from 82 percent in 1990–91 to 17 percent in 1999–2000.
- Cross-industry dispersion of tariffs: standard deviation of the tariff rate was 0.23 in 1980–81; standard deviation dropped to 0.05 in 1999–2000.
- Intertemporal pair-wise correlations for tariffs since 1990–91 range from 0.42 to 0.96; Indian intertemporal tariff correlation is significantly lower than U.S. example (U.S. correlation between 1972 and 1988 ≈ 0.98).

### Theoretical channels and predictions
- Competitive models with short-run labor immobility (specific-factors) predict trade liberalization reduces returns to the sector-specific factor where tariffs fall more.
- Heterogeneous-worker models: tariffs can change relative wages via compositional effects if trade affects worker composition.
- If industry differentials reflect compensating differentials, trade liberalization should have no effect on wage premiums.
- Imperfect competition introduces channels via changes in capital or labor rents.
- Unions may extract protection rents in employment rather than wages; liberalization could affect employment instead of relative wages depending on union seniority structures.
- Firm-level productivity increases from liberalization may raise industry wages if productivity gains are passed through.

### Empirical strategy and estimation
- Two-stage industry wage premium methodology:
  - First stage (equation (1)): ln(w_ijt) regressed on worker characteristics H_ijt and industry dummies I_ijt to estimate industry wage premiums wp_jt. First-stage regressions estimated separately for each NSS round; normalized wage premiums expressed as deviations from employment-weighted average.
  - Exact standard errors for normalized premiums calculated using Haisken-DeNew and Schmidt (1997) two-step restricted least squares procedure.
  - Second stage (first-differenced, equation (2)): Δwp_jt regressed on ΔT_jt (change in tariffs), ΔD_jt (changes in other trade-related variables), year indicators π_t, and an error term ε_jt. First differencing controls for unobserved industry-specific heterogeneity.
- Estimation uses weighted least squares with weights equal to the inverse of the standard error of the first-stage wage premium.

### Data: trade policy and household surveys
- Trade data source: Das (2003); covers 72 three-digit manufacturing industries (NIC-1987) for 1980–81 to 1999–2000.
- NSSO household survey rounds used: 1983–84 (38th round), 1987–88 (43rd round), 1993–94 (50th round), 1999–2000 (55th round).
- Sample: urban manufacturing workers, ages 15–65, labor force participants reporting positive weekly earnings.
- Wage measure: weekly earnings in rupees deflated by the consumer price index from the International Financial Statistics.
- Education categories based on completed years of schooling:
  - primary or less: at most 5 years of schooling
  - middle or secondary: 6–11 years of schooling
  - higher secondary or more: at least 12 years of schooling

### Key empirical results
- First-stage wage regressions (industry dummies plus worker characteristics):
  - Example (1999–2000): R-squared excluding industry dummies = 0.50; including industry dummies increases R-squared to 0.55.
  - Industry indicators thus account for 5 percent of the total variation in log weekly earnings in 1999–2000.
  - Across survey rounds, industry indicators explain about 4 to 7 percent of the variation in log weekly earnings.
- Structure of industry wage differentials:
  - Large differences in wages across industries exist for seemingly similar workers.
  - The structure of industry wage differentials changed over time.
  - Labor market rigidities are a plausible explanation for the existence of wage premiums in India.
- Impact of trade policy on wages:
  - The change in tariffs is strongly and negatively associated with changes in industry wage premiums.
  - Result is robust to instrumental variable estimation and to controlling for nontariff barriers (import coverage ratios).
  - Sectors with larger tariff reductions tended to be those with higher proportions of unskilled workers; those sectors experienced increases in relative wages.

### Annex: inter-industry wage premiums (72 industries)
- Wage premiums expressed as deviations from employment-weighted average; standard errors calculated by Haisken-DeNew and Schmidt (1997).
- Joint significance of wage premiums: p-value = 0.00 in all years.
- Standard deviations of industry wage premiums (by year): 0.2782, 0.3376, 0.2826, 0.24.
- Example estimates:
  - Manufacture of fertilizer and pesticides (industry code = 301): 0.314 (1983–84).
  - Weaving and finishing of cotton khadi (industry code = 232): −0.084 (1983–84).
  - Example difference: switching from leather footwear to khadi implies a decline of 40 percent in weekly earnings (0.314-(-0.084)).

### Time variation and correlations of premiums
- The structure of wage premiums changed over time.
- Year-year correlations of industry wage premiums (Table 3):
  - 1983–84 with 1987–88 = 0.4788; with 1993–94 = 0.3648; with 1999–00 = 0.2619.
  - 1987–88 with 1993–94 = 0.4767; with 1999–00 = 0.4033.
  - 1993–94 with 1999–00 = 0.4337.
- Correlation between 1983–84 and 1999–2000 reported as 0.26; between 1987–88 and 1999–2000 reported as 0.40.
- Comparisons: U.S. and Brazil studies report correlation coefficients of order 0.9; Indian premiums are much less correlated over time.

### Tests of explanations for wage premiums
- Labor mobility / rigidities:
  - India ranked forty-fifth for degree of labor market flexibility in GCR (1998).
  - Industrial Disputes Act of 1947: firms employing 100 or more workers require government permission before laying off or retrenching workers (1982 amendment).
  - Regression of industry employment share on tariff rates, industry and year indicators: coefficient = 0.001 and statistically insignificant — no evidence of significant employment sensitivity to trade shocks.
- Labor quality explanation:
  - Adding measured human capital controls reduces standard deviation of industry wage differentials by 5 to 6 percentage points; nevertheless standard deviation remains substantial and ranges from 24–34 percent.
  - Authors note longitudinal data would be required for more conclusive evidence.
- Occupation-level correlations:
  - Correlations of industry wage differentials across six occupation groups are much lower than U.S. results; cannot reject labor quality/compensating differentials/efficiency wages explanations based on cross-occupation correlations.
  - Selected occupation correlation examples: occupation 6 vs 5 = 0.0264; 6 vs 4 = 0.1244; 5 vs 4 = 0.1420.
  - Occupation codes: 1=professional; 2=clerical; 3=service; 4=skilled; 5=production; 6=unskilled.
- Unions:
  - For 1993–94, non-union workers dispersion = 0.389 and union workers dispersion = 0.340.
  - Correlation between wage premiums for union and non-union workers in India = 0.225 (compared to 0.6–0.8 in the United States).
  - Institutional notes: Trade Union Act of 1926 allows any seven workers to register a union; multiple unions and outsider control noted; Industrial Disputes Act of 1947 gives the state power in dispute settlement.

### Preliminary evidence linking trade policy and industry characteristics
- Raw scatter relationships:
  - Tariff reduction (1983–84 to 1999–2000) positively related to initial tariffs in 1980: coefficient = 0.66, standard error = 0.09, statistically significant at 1 percent, number of observations = 72.
  - Tariff reduction positively related to share of unskilled workers in 1983: coefficient = 0.71, standard error = 0.25, statistically significant at 1 percent, number of observations = 65. (Unskilled defined as <12 years schooling.)
  - Tariff reductions greatest in low wage industries; relationship between tariff reduction and wage premium in 1983–84: coefficient = −0.19, standard error = 0.12 (statistically significant at 15 percent), number of observations = 65.
  - Scatter of changes in tariffs vs changes in wage premiums across consecutive periods: coefficient = −0.10, standard error = 0.02, statistically significant at 5 percent, number of observations = 208.

### Second-stage regressions: tariffs and wage premiums
- Sample: 72 industries with tariff information.
- Selected specification results (Table 4):
  - Specification I (first differenced): Tariff rate coefficient = −0.174**, standard error = (0.07), year indicators Yes, first differencing Yes, industry indicators No, Number of observations = 2092.
  - Specification II (levels, no industry indicators): Tariff rate coefficient = −0.435***, standard error = (0.10), year indicators Yes, industry indicators No, Number of observations = 8128.
  - Specification III (levels, with industry indicators): Tariff rate coefficient = −0.153**, standard error = (0.07), year indicators Yes, industry indicators Yes, first differencing No, Number of observations = 1281.
- Interpretation example: coefficient of −0.17 in Specification I implies reducing tariffs from 50 percent to 0 percent increases average wage in that sector by 8.5 percent (0.17x0.5).

### Controlling for non-tariff barriers and import measures
- Nontariff barrier measure: import coverage ratio.
- Selected Table 5 coefficients:
  - Tariff rate: −0.198** (s.e. 0.0790) in Column I; −0.386*** (s.e. 0.1000) in Column II; −0.146** (s.e. 0.0720) in Column III.
  - Import coverage ratio coefficients reported as 0.0000 (s.e. (0.0010)) and overall import coverage ratio is statistically insignificant.
  - Import penetration ratio coefficients reported as −0.129, 0.114, 0.081 with respective standard errors (0.1010), (0.1180), (0.1190).
- Tariff coefficient remains negative and statistically significant (at least at 5 percent) in all specifications.

### Instrumental variables strategy and results
- Instruments: interactions of foreign exchange reserves with tariff rates in 1980 and with share of unskilled workers in 1983.
- First-stage (Table 6B) selected results:
  - Tariff rate in 1980 interacted with foreign exchange reserves: coefficient = −0.295*** (s.e. 0.0790).
  - Share of unskilled workers in 1983 interacted with foreign exchange reserves: coefficient = −0.312* (s.e. 0.1840).
  - Import coverage ratio coefficient = 0.001* (s.e. 0.0003).
  - Import penetration ratio coefficient = 0.0820 (s.e. 0.0860).
  - R-squared = 0.65; F-statistic for joint significance of instruments = 13.1, p-value = 0.
  - Number of observations = 168; year indicators Yes; first differencing Yes.
- Second-stage IV (Table 6A) selected result:
  - Tariff rate instrumented: coefficient = −0.577*** (s.e. 0.2260), first differencing Yes, year indicators Yes, number of observations = 168.
  - Import coverage ratio = −0.0002 (s.e. 0.0004).
  - Import penetration ratio = −0.109 (s.e. 0.0990).
- Over-identifying restrictions: tests fail to reject at 1 percent and 5 percent levels in reported specifications; chi-squared statistics cited in notes (e.g., 0.1056 and 0.2992 for two exclusions).

### Robustness checks
- Inclusion of gross fixed capital formation by sector (logs) from Annual Survey of Industries (2002).
- Table 7 selected results:
  - Nominal rate of protection: −0.16** (s.e. 0.07) in Column I; −0.20** (s.e. 0.08) in Column II — first differencing Yes, year indicators Yes.
  - Import coverage ratio coefficient = −0.01 (s.e. 0.01).
  - Import penetration ratio coefficient = −0.13 (s.e. 0.10).
  - Gross fixed capital formation (logs) coefficients = −0.01 (s.e. (0.01)) in reported columns.
  - Number of observations = 199 (Column I) and 167 (Column II).
- Conclusion: tariff coefficient similar to previous tables; negative correlation between tariffs and wage premiums not driven by capital accumulation measure.

### Discussion and conclusions
- Main empirical findings:
  - Increasing protection in a sector lowers wages in that sector; sectors with largest tariff reductions saw wages increase relative to economy-wide average.
  - IV results produce a larger negative tariff coefficient than non-IV first-differenced estimates.
  - Tariff reductions were relatively larger in sectors with a higher proportion of unskilled workers, and these sectors experienced increases in relative wages.
- Implication for wage inequality:
  - Because sectors with higher shares of unskilled workers experienced larger tariff reductions and increases in relative wages, trade liberalization is associated with decreased wage inequality in India.
- Comparisons with other countries:
  - Results contrast with studies on Colombia, Brazil, and Mexico (which found positive or insignificant relationships) but are similar in sign to Gaston and Trefler (1994) for U.S. manufacturing.
- Authors’ interpretive remarks:
  - Negative relationship consistent with liberalization-induced productivity increases at firm level that are passed onto industry wages.
  - No single model fully explains all facts; labor market rigidities are a plausible explanation for existence of wage premiums in India.

*Source: _wp0520 - 1. Correlations of Tariffs Over Time (content unit extracted from the supplied IMF PDF).*

### 1. Correlations of Tariffs Over Time ..................................................................................2

### _wp0520 - 1. Correlations of Tariffs Over Time ..................................................................................2

### Introduction: scope and key conclusions
- Paper examines impact of India’s 1991 trade liberalization on industry wage premiums using micro-level NSSO data (1980–2000) combined with industry trade protection data.
- Main empirical finding: a strong and negative relationship between changes in trade policy (tariffs) and changes in industry wage differentials; robust to instrumenting for tariffs and to inclusion of nontariff barrier measures.
- Distributional implication: because tariff reductions were relatively larger in sectors with a higher proportion of unskilled workers and these sectors experienced an increase in relative wages, unskilled workers experienced an increase in incomes relative to skilled workers—suggesting trade liberalization decreased wage inequality in India.

### Background of India’s trade liberalization
- Pre-1991 regime: import-substitution strategy with high nominal tariffs and extensive nontariff barriers (NTBs) and complex import licensing.
- Balance of payments crisis in 1991: foreign currency assets less than US$1 billion, just enough to cover two weeks of imports.
- IMF-supported program (Stand-By-Arrangement) in August 1991 tied to structural reforms; policy changes included elimination of state trading monopolies, easing import licensing, removing export restrictions, allowing foreign investment in previously reserved sectors, and full convertibility of domestic currency on foreign exchange transactions.
- EXIM policy 1992–97: eliminated all import licensing lists and established a “negative” list; by April 2002 all remaining quantitative restrictions were removed.

### Quantitative description of policy change and protection levels
- Manufacturing average tariff: increased from 86 percent in 1980–81 to 117 percent in 1990–91, then declined to 39 percent in 1999–2000.
- Import coverage ratio (measure of NTBs): declined from 82 percent in 1990–91 to 17 percent in 1999–2000.
- Cross-industry dispersion of tariffs: standard deviation of the tariff rate was 0.23 in 1980–81; standard deviation dropped to 0.05 in 1999–2000.
- Intertemporal pair-wise correlations for tariffs since 1990–91 range from 0.42 to 0.96; Indian intertemporal tariff correlation is significantly lower than U.S. example (U.S. correlation between 1972 and 1988 ≈ 0.98).

### Theoretical channels and predictions
- Competitive models with short-run labor immobility (specific-factors) predict trade liberalization reduces returns to the sector-specific factor where tariffs fall more.
- Heterogeneous-worker models: tariffs can change relative wages via compositional effects if trade affects worker composition.
- If industry differentials reflect compensating differentials, trade liberalization should have no effect on wage premiums.
- Imperfect competition introduces channels via changes in capital or labor rents.
- Unions may extract protection rents in employment rather than wages; liberalization could affect employment instead of relative wages depending on union seniority structures.
- Firm-level productivity increases from liberalization may raise industry wages if productivity gains are passed through.

### Empirical strategy and estimation
- Two-stage industry wage premium methodology:
  - First stage (equation (1)): ln(w_ijt) regressed on worker characteristics H_ijt and industry dummies I_ijt to estimate industry wage premiums wp_jt. First-stage regressions estimated separately for each NSS round; normalized wage premiums expressed as deviations from employment-weighted average.
  - Exact standard errors for normalized premiums calculated using Haisken-DeNew and Schmidt (1997) two-step restricted least squares procedure.
  - Second stage (first-differenced, equation (2)): Δwp_jt regressed on ΔT_jt (change in tariffs), ΔD_jt (changes in other trade-related variables), year indicators π_t, and an error term ε_jt. First differencing controls for unobserved industry-specific heterogeneity.
- Estimation uses weighted least squares with weights equal to the inverse of the standard error of the first-stage wage premium.

### Data: trade policy and household surveys
- Trade data source: Das (2003); covers 72 three-digit manufacturing industries (NIC-1987) for 1980–81 to 1999–2000.
- NSSO household survey rounds used: 1983–84 (38th round), 1987–88 (43rd round), 1993–94 (50th round), 1999–2000 (55th round).
- Sample: urban manufacturing workers, ages 15–65, labor force participants reporting positive weekly earnings.
- Wage measure: weekly earnings in rupees deflated by the consumer price index from the International Financial Statistics.
- Education categories based on completed years of schooling:
  - primary or less: at most 5 years of schooling
  - middle or secondary: 6–11 years of schooling
  - higher secondary or more: at least 12 years of schooling

### Key empirical results
- First-stage wage regressions (industry dummies plus worker characteristics):
  - Example (1999–2000): R-squared excluding industry dummies = 0.50; including industry dummies increases R-squared to 0.55.
  - Industry indicators thus account for 5 percent of the total variation in log weekly earnings in 1999–2000.
  - Across survey rounds, industry indicators explain about 4 to 7 percent of the variation in log weekly earnings.
- Structure of industry wage differentials:
  - Large differences in wages across industries exist for seemingly similar workers.
  - The structure of industry wage differentials changed over time.
  - Labor market rigidities are a plausible explanation for the existence of wage premiums in India.
- Impact of trade policy on wages:
  - The change in tariffs is strongly and negatively associated with changes in industry wage premiums.
  - Result is robust to instrumental variable estimation and to controlling for nontariff barriers (import coverage ratios).
  - Sectors with larger tariff reductions tended to be those with higher proportions of unskilled workers; those sectors experienced increases in relative wages.

### Interpretation and implications
- Evidence suggests trade liberalization in India led to decreased wage inequality between unskilled and skilled workers during the liberalization period studied.
- The Indian 1991 reform episode provides a natural experiment due to the exogenous, IMF-conditioned design and the predetermined target tariff rates, which limited policymakers’ responsiveness to industry lobbying and supports causal interpretation.

*Source: _wp0520 - 1. Correlations of Tariffs Over Time (content unit extracted from the supplied IMF PDF).*

### Annex Table 1 shows the inter-industry wage premiums for the 72 three digit industries for

### _wp0520 - Annex Table 1 shows the inter-industry wage premiums for the 72 three digit industries for

### Overview of inter-industry wage premiums
- Wage premiums are expressed as deviations from the employment weighted average wage premium.
- Standard errors are calculated by Haisken-DeNew and Schmidt (1997) procedure.
- The wage premiums are jointly statistically significant at 1 percent level (p-value = 0.00) in all the years.
- Standard deviations of industry wage premiums range from 0.24 to 0.34 across years (reported standard deviations: 0.2782, 0.3376, 0.2826, 0.24).
- Example estimates:
  - Manufacture of fertilizer and pesticides (industry code = 301): 0.314 (1983–84).
  - Weaving and finishing of cotton khadi (industry code = 232): −0.084 (1983–84).
  - Difference example: switching from leather footwear to khadi implies a decline of 40 percent in weekly earnings (0.314-(-0.084)).

### Time variation and correlations
- The structure of wage premiums changed over time.
- Year-year correlations of industry wage premiums (Table 3):
  - Correlation between 1983–84 and 1999–2000: 0.26.
  - Correlation between 1987–88 and 1999–2000: 0.40.
- Comparisons:
  - Correlation coefficients for the United States and Brazil are of the order of 0.9 in comparable studies; Indian premiums are much less correlated over time, suggesting significant structural change.
- Table 3 correlation matrix entries preserved as presented:
  - 1983–84 with 1987–88 = 0.4788; with 1993–94 = 0.3648; with 1999–00 = 0.2619.
  - 1987–88 with 1993–94 = 0.4767; with 1999–00 = 0.4033.
  - 1993–94 with 1999–00 = 0.4337.

### Tests of explanations for wage premiums
- Labor mobility / rigidities:
  - India has evidence of labor market rigidities; India is ranked forty-fifth for degree of labor market flexibility in GCR (1998).
  - Industrial Disputes Act of 1947: firms employing 100 or more workers require government permission before laying off or retrenching workers (1982 amendment).
  - Regression of industry employment share on tariff rates, industry and year indicators: coefficient = 0.001 and statistically insignificant — no evidence of significant employment sensitivity to trade shocks.
- Labor quality explanation:
  - Adding measured human capital controls (education, age, age squared) reduces standard deviation of industry wage differentials by 5 to 6 percentage points; nevertheless standard deviation remains substantial and ranges from 24–34 percent.
  - Authors note longitudinal data would be required for more conclusive evidence.
- Occupation-level correlations:
  - Correlations of industry wage differentials across six occupation groups (Annex Table 2) are much lower than U.S. results; cannot reject labor quality/compensating differentials/efficiency wages explanations based on cross-occupation correlations.
  - Table 2 correlation matrix entries (selected): correlations among occupations are generally low (e.g., occupation 6 vs 5 = 0.0264; 6 vs 4 = 0.1244; 5 vs 4 = 0.1420).
  - Occupation codes: 1=professional; 2=clerical; 3=service; 4=skilled; 5=production; 6=unskilled.
- Unions:
  - For 1993–94, non-union workers dispersion = 0.389 and union workers dispersion = 0.340.
  - Correlation between wage premiums for union and non-union workers in India = 0.225 (compared to 0.6–0.8 in the United States).
  - Institutional notes: Trade Union Act of 1926 allows any seven workers to register a union; multiple unions and outsider control are noted; Industrial Disputes Act of 1947 gives the state power in dispute settlement.

### Preliminary evidence linking trade policy and industry characteristics
- Raw scatter relationships:
  - Tariff reduction (1983–84 to 1999–2000) positively related to initial tariffs in 1980: coefficient = 0.66, standard error = 0.09, statistically significant at 1 percent, number of observations = 72.
  - Tariff reduction positively related to share of unskilled workers in 1983: coefficient = 0.71, standard error = 0.25, statistically significant at 1 percent, number of observations = 65. (Unskilled defined as <12 years schooling.)
  - Tariff reductions greatest in low wage industries; relationship between tariff reduction and wage premium in 1983–84: coefficient = −0.19, standard error = 0.12 (statistically significant at 15 percent), number of observations = 65.
  - Scatter of changes in tariffs vs changes in wage premiums across consecutive periods: coefficient = −0.10, standard error = 0.02, statistically significant at 5 percent, number of observations = 208.

### Second-stage regressions: tariffs and wage premiums
- Sample: 72 industries with tariff information.
- Specification summaries (Table 4):
  - Specification I (first differenced): Tariff rate coefficient = −0.174**, standard error = (0.07), year indicators Yes, first differencing Yes, industry indicators No, Number of observations = 2092.
  - Specification II (levels, no industry indicators): Tariff rate coefficient = −0.435***, standard error = (0.10), year indicators Yes, industry indicators No, Number of observations = 8128.
  - Specification III (levels, with industry indicators): Tariff rate coefficient = −0.153**, standard error = (0.07), year indicators Yes, industry indicators Yes, first differencing No, Number of observations = 1281.
- Interpretation: negative coefficient implies increasing protection lowers wages in that industry. Example: coefficient of −0.17 in Specification I implies reducing tariffs from 50 percent to 0 percent increases average wage in that sector by 8.5 percent (0.17x0.5).

### Controlling for non-tariff barriers and import measures (Table 5)
- Nontariff barrier measure: import coverage ratio.
- Table 5 reported coefficients (selected):
  - Tariff rate: −0.198** (s.e. 0.0790) in Column I; −0.386*** (s.e. 0.1000) in Column II; −0.146** (s.e. 0.0720) in Column III.
  - Import coverage ratio coefficients reported as 0.0000 in one column with parentheses (0.0010) in another; overall import coverage ratio is statistically insignificant in specifications.
  - Import penetration ratio coefficients reported as −0.129, 0.114, 0.081 with respective standard errors (0.1010), (0.1180), (0.1190).
- Tariff coefficient remains negative and statistically significant (at least at 5 percent) in all specifications.

### Instrumental variables strategy and results (Tables 6A, 6B)
- Instruments constructed as interactions of foreign exchange reserves with:
  - Tariff rates in 1980.
  - Share of unskilled workers in 1983.
- Rationale: foreign exchange reserves evolution correlated with post-1991 tariff changes; industry-specific interactions generate time-varying instruments.
- First-stage regression (Table 6B):
  - Tariff rate in 1980 interacted with foreign exchange reserves: coefficient = −0.295*** (s.e. 0.0790).
  - Share of unskilled workers in 1983 interacted with foreign exchange reserves: coefficient = −0.312* (s.e. 0.1840).
  - Import coverage ratio coefficient = 0.001* (s.e. 0.0003).
  - Import penetration ratio coefficient = 0.0820 (s.e. 0.0860).
  - R-squared = 0.65; F-statistic for joint significance of instruments = 13.1, p-value = 0.
  - Number of observations = 168; year indicators Yes; first differencing Yes.
- Second-stage IV regression (Table 6A):
  - Tariff rate instrumented: coefficient = −0.577*** (s.e. 0.2260), first differencing Yes, year indicators Yes, number of observations = 168.
  - Import coverage ratio = −0.0002 (s.e. 0.0004).
  - Import penetration ratio = −0.109 (s.e. 0.0990).
- Over-identifying restrictions:
  - Tests fail to reject over-identifying restrictions at 1 percent and 5 percent levels in reported specifications; chi-squared statistics reported in notes (e.g., 0.1056 and 0.2992 for two exclusions).

### Robustness checks (Table 7)
- Inclusion of gross fixed capital formation by sector (in logs) from Annual Survey of Industries (2002).
- Table 7 results:
  - Nominal rate of protection: −0.16** (s.e. 0.07) in Column I; −0.20** (s.e. 0.08) in Column II — first differencing Yes, year indicators Yes.
  - Import coverage ratio coefficient = −0.01 (s.e. 0.01).
  - Import penetration ratio coefficient = −0.13 (s.e. 0.10).
  - Gross fixed capital formation (in logs) coefficients reported as −0.01 and −0.01 (s.e. (0.01) each).
  - Number of observations = 199 (Column I) and 167 (Column II).
- Conclusion: tariff coefficient similar to previous tables; negative correlation between tariffs and wage premiums not driven by capital accumulation measure.

### Discussion and conclusions
- Main empirical findings:
  - Increasing protection in a sector lowers wages in that sector; sectors with largest tariff reductions saw wages increase relative to economy-wide average.
  - IV results produce a larger negative tariff coefficient than non-IV first-differenced estimates.
  - Tariff reductions were relatively larger in sectors with a higher proportion of unskilled workers, and these sectors experienced increases in relative wages.
- Implication for wage inequality:
  - Since sectors with higher shares of unskilled workers experienced larger tariff reductions and increases in relative wages, trade liberalization is associated with decreased wage inequality in India.
- Comparisons with other countries:
  - Results contrast with studies on Colombia, Brazil, and Mexico (which found positive or insignificant relationships) but are similar in sign to Gaston and Trefler (1994) for U.S. manufacturing.
- Authors’ interpretive remarks:
  - Negative relationship consistent with liberalization-induced productivity increases at firm level that are passed onto industry wages.
  - No single model fully explains all facts; labor market rigidities are a plausible explanation for existence of wage premiums in India.

### Key reported statistics and figures (exact values preserved)
- Joint significance of wage premiums: p-value = 0.00.
- Standard deviations of industry wage premiums: 0.2782, 0.3376, 0.2826, 0.24 (across years).
- Correlations of premiums:
  - 1983–84 vs 1999–00 = 0.26 (Table 3 reported 0.2619).
  - 1987–88 vs 1999–00 = 0.40 (Table 3 reported 0.4033).
- Earnings regression top coefficients and diagnostics (Table 2 highlights):
  - Constant and other coefficients shown across years; R-squared with industry indicators: 0.59, 0.56, 0.5, 0.55; Number of observations: 9,309; 9,083; 8,570; 7,855 (by year).
- Scatter relationships:
  - Tariff reduction vs pre-liberalization tariffs: coefficient = 0.66 (se=0.09), observations = 72.
  - Tariff reduction vs share of unskilled workers: coefficient = 0.71 (s.e. = 0.25), observations = 65.
  - Tariff reduction vs wage premium in 1983–84: coefficient = −0.19 (s.e. = 0.12), observations = 65.
  - Differences in wage premiums vs differences in tariffs: coefficient = −0.10 (s.e. = 0.02), observations = 208.
- Second-stage and IV estimates (select):
  - Table 4 Specification I tariff coefficient = −0.174** (s.e. (0.07)), observations = 2092.
  - Table 6A IV tariff coefficient = −0.577*** (s.e. (0.2260)), observations = 168.
  - First-stage instrument coefficients: −0.295*** (s.e. 0.0790) and −0.312* (s.e. 0.1840); first-stage R-squared = 0.65; F-statistic = 13.1, p-value = 0.

*Source: _wp0520 - Annex Table 1 shows the inter-industry wage premiums for the 72 three digit industries for*

### REFERENCES

### _wp0520 - REFERENCES

### References

- Aghion, P., R. Burgess, S. Redding, and F. Zilibotti, 2003, “The Unequal Effects of Liberalization: Theory and Evidence from India,” (unpublished; London School of Economics, London).  
- Annual Survey of Industries, 2002, “Database on the Industrial Sector in India: 1973–74 to 1997–98”, Mumbai: Economic and Political Weekly Research Foundation.  
- Attanasio, O., P. Goldberg, and N. Pavcnik, 2004, “Trade Reforms and Wage Inequality in Colombia,” Journal of Development Economics, Vol. 74, No. 4, pp. 331–66.  
- Blom, A., P. Goldberg, N. Pavcnik, and N. Schady, 2004, “Trade Liberalization and Industry Wage Structure: Evidence from Brazil,” World Bank Economic Review, forthcoming.  
- Das, D.K., 2003, “Quantifying Trade Barriers: Has Protection Declined Substantially in Indian Manufacturing? ICRIER,” Indian Council for Research on International Economic Relations Working Paper No. 105, New Delhi.  
- Dickens, W.T., and L.F. Katz, 1987, “Inter-Industry Wage Differentials and Theories of Wage Discrimination,” NBER Working Paper No. 2271 (Cambridge, Massachusetts: National Bureau of Economic Research).  
- Dutt, P. 2003, “Labor Market Outcomes and Trade Reforms: The Case of India,” in The Impact of Trade on Labor: Issues, Perspectives and Experiences from Developing Asia, ed. by R. Hasan and D. Mitra (Haarlem: North-Holland/Elsevier Science B.V., pp. 1–45).  
- Fallon, P.R. and R.E. Lucas, 1993, “Job Security Regulation and the Dynamic Demand for Industrial Labor in India and Zimbabwe”, Journal of Development Economics, Vol. 40, No. 2, pp. 241–275.  
- Feliciano, Z., 2001, “Workers and Trade Liberalization: The Impact of Trade Reforms in Mexico on Wages and Employment,” Industrial and Labor Relations Review, Vol. 55 No. 1, pp. 95–115.  
- Frankel, J. and Romer, D., 1999. “Does Trade Cause Growth?” American Economic Review, Vol. 89 (June), pp. 379–399.  
- Gaston, N., and D. Trefler, 1994, “Protection, Trade and Wages: Evidence from U.S. Manufacturing.” Industrial and Labor Relations Review, Vol. 47, No. 4, pp. 574−593.  
- Goldberg, P. and N. Pavcnik, 2004, “Trade, Wages, and the Political Economy of Trade Protection: Evidence from the Colombian Trade Reforms,” Journal of International Economics, forthcoming.  
- - 42 -  
- Grossman, G.M., 1984, “International Competition and the Unionized Sector,” Canadian Journal of Economics, Vo. 17, No. 3, pp. 541–556.  
- Grossman, G.M. and E. Helpman, 1994, “Protection for Sale,” American Economic Review, Vol. 84, No. 4, pp. 833–850.  
- Haisken-DeNew, J.P. and C.M. Schmidt, 1997, “Inter-Industry and Inter-Region Wage Differentials: Mechanism and Interpretation,” Review of Economics and Statistics, Vol. 79, No. 3, pp. 516–21.  
- Hanson, G. and A. Harrison, 1999, “Who Gains from Trade Reform? Some Remaining Puzzles,” Journal of Development Economics, Vol. 59 (June), pp. 125–54.  
- Harrison, A., 1994, “Productivity, Imperfect Competition and Trade Reform: Theory and Evidence,” Journal of International Economics, Vol. 36, Nos. 1−2, pp. 53–73.  
- Katz, L.F., L.H. Summers, R.E. Hall, C.L. Schultze and R.H. Topel, 1989, “Industry Rents: Evidence and Implications,” Brookings Papers on Economic Activity. Microeconomics, Vol. 1989, pp. 209–90. Brookings Institution.  
- Krishna, P. and D. Mitra, 1998, “Trade Liberalization, Market Discipline and Productivity Growth: New Evidence from India,” Journal of Development Economics, Vol. 56, No. 2 , pp. 447–462.  
- Krueger, A.B. and L.H. Summers, 1988, “Efficiency Wages and the Inter-Industry Wage Structure,” Econometrica, Vol. 56, pp. 259–293.  
- McDonald, I.M. and R.M. Solow, 1981, “Wage Bargaining and Unemployment,” American Economic Review, Vol. 71 (December), pp. 896–908.  
- National Sample Survey Organization (NSSO), 1983–84, 1987–88, 1993–94, 1999−2000, “Employment and Unemployment, Schedule 10.0”, Ministry of Statistics and Program Implementation, Government of India.  
- Pavcnik, N., 2000, “Trade Liberalization, Exit and Productivity Improvements. Evidence From Chilean Plants,” Review of Economic Studies, Vol. 69, pp. 245–76.  
- Revenga, A., 1997, ‘Employment and Wage Effects of Trade Liberalization: The Case of Mexican Manufacturing,” Journal of Labor Economics, Vol. 15, pp. S20–43.  
- Topalova, P., 2004, “Trade Liberalization and Firm Productivity: The Case of India,” IMF Working Paper 04/28 (Washington: International Monetary Fund).  
- World Economic Forum, 1998, Global Competitiveness Report (Geneva).

*Source: _wp0520 - REFERENCES*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp0520.pdf_