## wpiea2020061-print-pdf

## Source details

**Canonical URL:** [wpiea2020061-print-pdf](https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020061-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2020/english/wpiea2020061-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2020/english/wpiea2020061-print-pdf.pdf.json)

---

### 2.1 Firm-level data
- Data source and sample:
  - 2008-2015 waves of the Colombian Annual Survey of Manufacturers (Encuesta Annual Manufacturera or EAM); survey covers virtually all firms in manufacturing with at least 10 employees.
  - Sample size: 9,110 firms.
- Key variables and definitions:
  - Investment = sum of expenditures on new and used machinery and office equipment (investment into structures, buildings, and land ignored).
  - Investment rate = Investment divided by total fixed assets.
  - Main outcome: change in the investment rate between 2011 and 2010 (∆Investment), trimmed at the 1st and 99th percentiles within each sector.
  - Available firm-level variables: expenditures on different types of capital, sales, employment, fixed assets, 4-digit ISIC industry code.
- Descriptive statistics:
  - Average ∆Investment (2011 vs 2010): -0.3 percentage points.
  - Standard deviation of ∆Investment: 12.12.
  - 10th percentile ∆Investment: -10.4 percentage points.
  - 90th percentile ∆Investment: 9.57 percentage points.
  - Average investment rate in 2010: 5.75 percentage point.
  - Average investment rate in 2011: 5.43 percentage points.
  - Share of firms not investing in a given year: more than 25 percent.
  - 90th percentile investment rates: 17.24 (2010) and 16.09 (2011).
  - Employment distribution: median firm has 25 employees; mean is 74; standard deviation is 145.
  - Log of sales (in pesos) mean: 14.47.
  - Log of fixed assets (in pesos) mean: 13.42.
  - Importer status: 22% of firms were importers in 2010; 21% in 2011; 4% of firms became importers in 2011.

### 2.2 Tariff measures
- Tariff data and aggregation:
  - Source: Felbermayr et al. (2018), HS 6-digit MFN tariffs for Colombia.
  - Sectors: 33 sectors defined analogously to the 2008 OECD input-output Table for Colombia.
  - Output tariff for sector s: TO_s,t = 1/N_s ∑_{h_s ∈ S} T_{h_s,t}.
  - Output tariffs also computed separately for capital goods (TO,C_s,t) and other goods (TO,¬C_s,t).
  - Input tariffs constructed as weighted averages of output tariffs using input-expenditure shares w_{s,s'} from the 2008 input-output table:
    - TI_s,t = ∑_{s'} w_{s,s'} TO_{s',t}
    - TI,C_s,t = ∑_{s'} w_{s,s'} TO,C_{s',t}
    - TI,¬C_s,t = ∑_{s'} w_{s,s'} TO,¬C_{s',t}
- Sector exposure statistics:
  - Example: most exposed sector faced input tariff rate of 12% between 2008 and 2010, dropping to 8% in 2011; least exposed sector had almost no change in 2011.
  - Average sector reduction in input tariffs (2011 vs 2010): 3.14 percentage points.
  - Average reduction in capital goods tariffs: 1.03 percentage points; standard deviation: 1.23.
  - 10th percentile change in capital goods tariffs: -2.92 percentage points.
  - Least exposed sector change in capital goods tariffs: -0.04 percentage points.
  - Average reduction in tariffs on other inputs: 3.1 percentage points.
  - Average reduction in output tariffs: 4.59 percentage points.

### 2.3 Alternative input tariff measures (robustness)
- Alternative construction:
  - Based on Colombian import transactions (DANE) at importer-HS10 good-origin-month level; each importer matched to an ISIC 4-digit sector.
  - For each observed 4-digit sector s̃, import expenditure shares sh_{s̃,h_s,t} used to construct:
    - T̃I_{s̃,t} = ∑_{h_s∈Ω} sh_{s̃,h_s,t} T_{h_s,t}
    - T̃I,C_{s̃,t} = ∑_{h_s∈Ω_C} sh_{s̃,h_s,t} T_{h_s,t}
    - T̃I,¬C_{s̃,t} = ∑_{h_s∈Ω_¬C} sh_{s̃,h_s,t} T_{h_s,t}
- Summary statistics (selected from Table 3):
  - ∆ ̃T_I_2011: mean -4.511, p10 -7.597, p25 -5.030, p50 -4.463, p75 -3.563, p90 -2.516, sd 1.668, Observations 110.
  - ∆ ̃T_I,C_2011: mean -0.436, p10 -0.861, p25 -0.531, p50 -0.249, p75 -0.140, p90 -0.0319, sd 0.497.

### 3 Trade liberalization and Investment Rates

#### 3.1 Baseline empirical approach
- Cross-sectional specifications (t = 2011):
  - ∆Investment_i = α + β1 ∆TI_{s(i)} + Xγ + ε_i, with controls X including lagged logs of fixed assets and sales.
  - Extended: ∆Investment_i = α + β1 ∆TI,C_{s(i),t} + β2 ∆TI,¬C_{s(i),t} + β3 ∆TO_{s(i),t} + Xγ + ε_i.

#### Baseline findings (Table 4)
- Overall input tariff reduction:
  - A one percentage point stronger exposure to the reduction in overall tariffs is associated with a 0.12 percentage points increase in the investment rate (coefficient -0.116 reported; not statistically significant at conventional levels).
- Capital goods input tariffs:
  - A one percentage point stronger exposure to capital goods input tariff reduction is associated with a 0.377 percentage points increase in the investment rate (column (2); coefficient -0.377 ∗∗∗, std. error 0.055).
  - For the average firm with 2010 investment rate 5.75 percent, a one percentage point stronger reduction in capital goods input tariffs increases investment rate to 6.127 percent (a 6.6 percent increase).
  - Sector with largest exposure faced a reduction of 3.03 percentage points in capital goods tariffs; least affected sector faced a reduction of 0.03 percentage points.
  - Based on regression, firms in the most exposed sector increased investment by around 1.14 percentage point in 2011; average firm would see increase from 5.75 to 6.89 (an almost 20% increase).
  - Coefficient on change in capital goods input tariffs ranges from -0.377 to -0.370 across specifications (columns (2)-(4)), indicating stability.
- Other input tariffs:
  - Change in other input tariffs has a negative but not statistically significant effect on investment; economically small.
- Output tariffs:
  - Decline in output tariffs decreases investment rate of domestic firms (consistent with crowding out), but effect is not statistically significant and is economically tiny.

#### Instrumental variable checks (Table 5)
- Instrument: level of the capital goods tariff rate in 2010 used to instrument the change in the tariff rate.
- OLS: one percentage point larger tariff rate in 2010 raised the investment rate by 0.19 percentage points in 2011 (capital input tariffs 2010 coefficient 0.193 ∗∗∗, std. error 0.055).
- IV regression where reduction instrumented with level in 2010: IV coefficient -0.373 ∗∗∗ (std. error 0.079), very similar to baseline.
- First-stage F-statistic: 24.54 (exceeds Stock and Yogo weak instrument threshold).
- Conclusion: OLS and IV coefficients statistically not different; proceed with OLS for efficiency.

#### 3.2 Dynamic effects
- Dynamic specification regresses Investment_{i,t} − Investment_{i,2010} on 2011–2010 changes in tariffs for t = 2008, 2009, 2011, 2012, 2013, 2014, 2015, 2016.
- Placebo: changes in investment rate between 2008–2010 and 2009–2010 are not significantly associated with exposure to capital goods input tariffs.
- Year-by-year effects:
  - Coefficient for 2011 vs 2010 equals -0.37 for change in investment rate (consistent with Table 4).
  - Coefficient remains negative for 2012 and 2013, not statistically significant in 2013.
  - After 2013, effect fluctuates around 0.
- Interpretation: more-exposed firms increased capital stock relative to others, leading to capital deepening and an increase in labor productivity.

#### 3.3 Heterogeneity across firms (size effects)
- Mechanism: larger firms more likely to import; tariff reduction can lower fixed costs and induce importing.
- Size-interaction specification: quartiles by employment or sales.
- Key heterogeneous effects (Table 6):
  - Employment quartiles:
    - Effect of change in capital goods tariffs on investment is negative for the smallest firms (statistically significant only when sales used as size indicator).
    - Medium-large firms (third quartile) benefit most; largest firms also benefit but less than medium-large.
    - For third quartile firm exposed to a 1 percentage point decline in capital goods input tariffs, investment increases by 0.66 percentage point more compared to smallest firms; first quartile increase is 0.02 percentage point.
  - Sales quartiles:
    - First quartile firm increases investment by 0.367 percentage point more in response to a 1 percentage point decline in capital goods tariffs.
    - Medium-large firm (third quartile) increases investment by 0.781 percentage point for the same tariff decline.
    - Fourth quartile firms do not benefit more than smallest firms.
  - Sector fixed effects:
    - Including sector fixed effects absorbs sector-level heterogeneity; interaction coefficients remain robust and stable.

### 3.4 Import Entry
- Probit evidence on extensive margin:
  - Dependent variable: Import Entry = 1 if firm not importing in 2010 but starts importing in 2011.
  - Reduction in overall input tariffs increases probability to start importing in 2011, but coefficient not statistically significant (Column (1), Table 7).
  - Splitting tariff changes:
    - Firms exposed to stronger reduction in capital goods tariffs are more likely to start importing (∆Capital Input Tariffs coefficients: -0.0560 ∗, -0.0546 ∗∗, -0.0597 ∗ across columns).
    - No similar effect for change in tariffs for other (non-capital) inputs.
    - Firms more exposed to a reduction in output tariffs are also more likely to become importers (∆Output Tariffs coefficient -0.0361 ∗∗ in Column (4)).
  - Quantitative marginal effect: average marginal effect of a one percentage point reduction in capital goods input tariffs on the probability of import entry is 0.005 (Column (4), Table 7).

### 3.5 Robustness
- Additional firm-level controls (Table 8):
  - Successive inclusion of controls (lagged ln fixed assets, lagged ln sales, lagged ln TFP, ∆ln fixed assets, ∆ln sales) leaves ∆Capital Input Tariffs coefficient ranging from -0.372 to -0.376.
  - Adds confidence that omitted firm-level controls are unlikely to alter baseline result.
- Alternative tariff measures (Table 11):
  - Replacing sector-weighted tariffs with DANE-based alternative tariffs confirms baseline findings:
    - Firms exposed to stronger decline in overall input tariffs, non-capital goods input tariffs and output tariffs do not significantly change investment rate.
    - Larger exposure to capital goods input tariff cuts has a statistically and economically strong effect on ∆Investment (e.g., ∆Capital Input Tariff coefficient -0.587 ∗∗∗ in column (2)).

### 4 Trade liberalization and Employment
- Theoretical context:
  - Effects depend on factor intensity of sectors facing tariff declines and substitutability between labor and capital or intermediates.
  - Prior estimates: substitution elasticities often below one; Oberfield and Raval (2014) estimate 0.84 for Colombian manufacturing.
- Empirical employment results (Table 9 and Table 10):
  - Aggregate: larger exposure to decline in input tariffs associated with a reduction in workers between 2010 and 2011 (Column (1), Table 9), but input-type heterogeneity matters.
  - By input type:
    - Reduction in non-capital good input tariffs leads to a decline in number of workers (consistent with labor and intermediate inputs as substitutes).
    - Reduction in capital goods tariffs associated with an increase in number of employees (consistent with labor and capital as complements).
  - Heterogeneity by worker type (Table 10):
    - Coefficient on capital goods tariffs is around 50% higher for manual (production) workers than for administrative workers.
    - Effect on administrative workers not statistically significant.
  - Persistence (dynamic evidence, Figure 5):
    - Increase in production workers remains significant for four years in sectors more exposed to reduction in capital goods tariffs.
    - After four years, difference between more and less exposed sectors no longer statistically significant.
  - Interpretation: production workers and capital are complements for about four years, then effectively independent (Cobb-Douglas case).

### 5 Conclusion
- Core findings:
  - Output tariffs have no significant effect on investment.
  - Decline in capital goods tariffs may substantially boost investment.
  - Employment responses heterogeneous:
    - Reduction in capital goods tariffs associated with higher employment of production workers.
    - Reduction in input tariffs on non-capital goods associated with declines in employment.
  - Firms deterred from importing by fixed costs benefit most from reductions in capital goods tariffs.
- Policy implications:
  - Effects of tariff reductions depend on which tariffs are cut:
    - Reduction in capital goods tariffs can significantly stimulate investment.
    - Reduction in tariffs on other inputs and output tariffs do not affect investment but can help boost productivity (citing Amiti and Konings, 2007; Topalova and Khandelwal, 2011).
  - Heterogeneous firm responses imply targeted considerations: firms facing fixed-cost barriers to importing and those reliant on capital goods may gain disproportionately from cuts in capital goods tariffs.

*Source: wpiea2020061-print-pdf*

### 2.1  Firm-level data

### 2.1  Firm-level data

### Firm-level dataset and variable construction
- Data source: 2008-2015 waves of the Colombian Annual Survey of Manufacturers (Encuesta Annual Manufacturera or EAM); survey covers virtually all firms in manufacturing with at least 10 employees.
- Key firm-level variables available: expenditures on different types of capital, sales, employment, fixed assets, 4-digit ISIC industry code.
- Investment definition: sum of expenditures on new and used machinery and office equipment (investment into structures, buildings, and land ignored).
- Investment rate: Investment divided by total fixed assets.
- Main outcome variable: change in the investment rate between 2011 and 2010 (∆Investment), trimmed at the 1st and 99th percentiles within each sector.
- Sample size: 9,110 firms.

### Summary statistics and distributions
- Average change in the investment rate between 2011 and 2010: -0.3 percentage points.
- Standard deviation of the change in investment rate: 12.12.
- 10th percentile change in investment rate: -10.4 percentage points.
- 90th percentile change in investment rate: 9.57 percentage points.
- Average investment rate in 2010: 5.75 percentage point.
- Average investment rate in 2011: 5.43 percentage points.
- Share of firms not investing in a given year: more than 25 percent.
- 90th percentile investment rates: 17.24 (2010) and 16.09 (2011).
- Employment distribution: median firm has 25 employees; mean is 74; standard deviation is 145.
- Log of sales (in pesos) mean: 14.47.
- Log of fixed assets (in pesos) mean: 13.42.
- Importer status: 22% of firms were importers in 2010; 21% in 2011; 4% of firms became importers in 2011.

---

### 2.2  Tariff measures

### Sectoral tariff data and aggregation
- Tariff data source: Felbermayr et al. (2018), HS 6-digit MFN tariffs for Colombia.
- Sector aggregation: 33 sectors defined analogously to the 2008 OECD input-output Table for Colombia.
- Output tariff for sector s:
  - TO_s,t = 1/N_s ∑_{h_s ∈ S} T_{h_s,t} (simple average of HS 6-digit tariffs for goods produced by sector s).
- Output tariffs also computed separately for capital goods (TO,C_s,t) and other goods (TO,¬C_s,t) using sets S_C and S_¬C and counts N_C_s and N_¬C_s.
- Input tariffs constructed as weighted averages of output tariffs using input-expenditure shares w_{s,s'} from the 2008 input-output table:
  - TI_s,t = ∑_{s'} w_{s,s'} TO_{s',t}
  - TI,C_s,t = ∑_{s'} w_{s,s'} TO,C_{s',t}
  - TI,¬C_s,t = ∑_{s'} w_{s,s'} TO,¬C_{s',t}
- Measures capture access to cheaper inputs and allow differential responses to cuts in capital goods tariffs versus other inputs.

### Key sector exposure statistics
- Example plotted in Figure 3: most exposed sector faced input tariff rate of 12% between 2008 and 2010, dropping to 8% in 2011; least exposed sector had almost no change in 2011.
- Average sector reduction in input tariffs (2011 vs 2010): 3.14 percentage points.
- Average reduction in capital goods tariffs: 1.03 percentage points; standard deviation: 1.23.
- 10th percentile change in capital goods tariffs: -2.92 percentage points.
- Least exposed sector change in capital goods tariffs: -0.04 percentage points.
- Average reduction in tariffs on other inputs: 3.1 percentage points.
- Average reduction in output tariffs: 4.59 percentage points.

---

### 2.3  Alternative input tariff measures (robustness)
- Alternative tariffs computed from Colombian import transactions (DANE) at importer-HS10 good-origin-month level; each importer matched to an ISIC 4-digit sector observed in the manufacturing survey.
- For each observed 4-digit sector s̃, import expenditure shares on HS 6-digit goods denoted sh_{s̃,h_s,t} are calculated and used to construct:
  - T̃I_{s̃,t} = ∑_{h_s∈Ω} sh_{s̃,h_s,t} T_{h_s,t}
  - T̃I,C_{s̃,t} = ∑_{h_s∈Ω_C} sh_{s̃,h_s,t} T_{h_s,t}
  - T̃I,¬C_{s̃,t} = ∑_{h_s∈Ω_¬C} sh_{s̃,h_s,t} T_{h_s,t}
- Table 3 presents summary statistics for these alternative tariff measures.

---

### 3  Trade liberalization and Investment Rates

### 3.1  Baseline empirical approach
- Basic cross-sectional specification (for t=2011):
  - ∆Investment_i = α + β1 ∆TI_{s(i)} + Xγ + ε_i
  - Controls X include lagged logs of fixed assets and sales.
- Extended specification separating capital goods and other input tariffs and output tariffs:
  - ∆Investment_i = α + β1 ∆TI,C_{s(i),t} + β2 ∆TI,¬C_{s(i),t} + β3 ∆TO_{s(i),t} + Xγ + ε_i

### Baseline findings (Table 4)
- Exposure to overall input tariff reduction:
  - A one percentage point stronger exposure to the reduction in overall tariffs is associated with a 0.12 percentage points increase in the investment rate (coefficient not statistically significant at conventional levels).
- Capital goods input tariffs:
  - A one percentage point stronger exposure to capital goods input tariff reduction is associated with a 0.377 percentage points increase in the investment rate (column (2)).
  - For the average firm with 2010 investment rate 5.75 percent, a one percentage point stronger reduction in capital goods input tariffs increases investment rate to 6.127 percent (a 6.6 percent increase).
  - Sector with largest exposure faced a reduction of 3.03 percentage points in capital goods tariffs; least affected sector faced a reduction of 0.03 percentage points.
  - Based on regression, firms in the most exposed sector increased investment by around 1.14 percentage point in 2011; average firm would see increase from 5.75 to 6.89 (an almost 20% increase).
- Other input tariffs:
  - Change in other input tariffs has a negative but not statistically significant effect on investment.
  - Average effect economically small and statistically insignificant.
- Output tariffs:
  - Decline in output tariffs decreases investment rate of domestic firms (consistent with crowding out), but effect is not statistically significant and is economically tiny.
- Stability:
  - Coefficient on change in capital goods input tariffs ranges from -0.377 to -0.370 across specifications (columns (2)-(4)), indicating stability.

### Instrumental variable checks (Table 5)
- Instrument: level of the capital goods tariff rate in 2010 used to instrument the change in the tariff rate.
- OLS of investment rate on level of capital goods tariff rate in 2010: one percentage point larger tariff rate in 2010 raised the investment rate by 0.19 percentage points in 2011.
- IV regression where reduction in tariff is instrumented with level in 2010: IV coefficient very similar and not statistically different from baseline coefficient in column (2).
- First-stage F-statistic: 24.54 (exceeds Stock and Yogo weak instrument threshold).
- Conclusion: OLS and IV coefficients statistically not different; proceed with OLS for efficiency.

---

### 3.2  Dynamic effects
- Dynamic specification (Equation 12) regresses Investment_{i,t} − Investment_{i,2010} on 2011–2010 changes in tariffs, for t = 2008, 2009, 2011, 2012, 2013, 2014, 2015, 2016.
- Placebo test: changes in investment rate between 2008–2010 and 2009–2010 are not significantly associated with exposure to capital goods input tariffs.
- Year-by-year effects:
  - Coefficient for 2011 vs 2010 equals -0.37 for change in investment rate (consistent with Table 4).
  - Coefficient remains negative for 2012 and 2013, but is not statistically significant in 2013.
  - After 2013, effect fluctuates around 0.
- Interpretation: firms more exposed to the decline in capital goods tariffs increased capital stock more than other firms, leading to capital deepening and an increase in labor productivity.

---

### 3.3  Heterogeneity across firms (size effects)
- Hypothesis: larger firms are more likely to import and face fixed costs of entering import markets; tariff reduction can lower fixed costs and induce importing.
- Size interaction regression (quartiles by employment or sales), equation for t=2011:
  - ∆Investment_i = α + β1 ∆TI,C_{s(i),t} + ∑_{q=2}^{4} I_q × β_{q2} ∆TI,C_{s(i),t} + ∑_{q=2}^{4} I_q × β_{q3} + Xγ + ε_i
  - I_1..I_4 denote size quartiles (I_1 smallest, I_4 largest).
- Key heterogeneous effects (Table 6):
  - Using employment quartiles:
    - Effect of change in capital goods tariffs on investment is negative for the smallest firms (both sales- and employment-based size), but statistically significant only when sales used as size indicator.
    - Firms in second quartile benefit more but effect not statistically significant.
    - Medium-large firms (third quartile) benefit most; largest firms also benefit but less than medium-large.
    - For third quartile firm exposed to a 1 percentage point decline in capital goods input tariffs, investment increases by 0.66 percentage point more compared to smallest firms; first quartile increase is 0.02 percentage point.
  - Using sales quartiles:
    - First quartile firm increases investment by 0.367 percentage point more in response to a 1 percentage point decline in capital goods tariffs.
    - Medium-large firm (third quartile) increases investment by 0.781 percentage point for the same tariff decline.
    - Fourth quartile firms do not benefit more than smallest firms.
- Sector fixed effects:
  - Including sector fixed effects (Columns (2) and (4)) absorbs sector-level heterogeneity; interaction term results remain robust.
  - Interaction coefficients are stable when fixed effects included, suggesting confounding sector-level factors are minor.
- Conclusion: medium-large firms (third quartile) benefit most from capital goods tariff reduction, consistent with tariff cuts reducing fixed costs of importing and enabling medium-large firms to realize import-related gains.

*Source: wpiea2020061-print-pdf - 2.1  Firm-level data*

### 3.4  Import Entry

### 3.4  Import Entry

### Findings on extensive margin of importing
- Estimated a probit regression where dummyImportEntry = 1 if firm not importing in 2010 but starts importing in 2011, and zero otherwise.
- A reduction in overall input tariffs increases the probability to start importing in 2011, but the coefficient is not statistically significant (Column (1), Table 7).
- When splitting tariff changes:
  - Firms exposed to a stronger reduction in capital goods tariffs are more likely to start importing.
  - No similar effect is found for the change in tariffs for other (non-capital) inputs.
  - Firms more exposed to a reduction in output tariffs are also more likely to become importers.
- Quantitative effect reported: the average marginal effect of a one percentage point reduction in capital goods input tariffs on the probability of a positive outcome is 0.005 (Column (4), Table 7).

### 3.5  Robustness

### Additional firm-level controls
- Conducted robustness tests by successively adding controls (Table 8):
  - Column (1): no controls — base result confirmed.
  - Column (2): adds lagged log of fixed assets.
  - Column (3): adds lagged log of sales and lagged log of total factor productivity.
  - Column (4): adds change in the log of fixed assets and sales between 2011 and 2010 (following Kalemli-Ozcan et al. (2018)).
- Base coefficient on the main specification varies only from -0.372 to -0.376 across these specifications.
- Because adding controls affects the coefficient only marginally, omitted firm-level controls such as Tobin’s Q or leverage are unlikely to significantly alter the baseline result.
- The main variable of interest appears uncorrelated with observed firm-level characteristics, suggesting it is likely uncorrelated with unobserved characteristics that could bias the result.

### Alternative tariff measure
- Used an alternative measure of input tariffs based on previous import volumes from Colombia’s statistical authority (DANE) (see subsection 2.3 for construction details).
- Confirmed baseline findings using the alternative measure (Table 11):
  - Firms exposed to a stronger decline in overall input tariffs, non-capital goods input tariffs and output tariffs do not significantly change their investment rate more than other firms.
  - Larger exposure to capital goods input tariff cuts has a statistically and economically strong effect on the change in the investment rate.

### 4  Trade Liberalization and Employment

### Theoretical context
- Effects of tariff reduction on returns to labor and capital depend on factor intensity of sectors facing tariff declines (Heckscher-Ohlin perspective).
- Prior evidence:
  - Attanasio et al. (2004): decline in output tariffs not associated with labor reallocation but associated with declines in industry wage premiums.
  - Karabarbounis and Neiman (2013): labor and capital are substitutes; decline in price of capital led to substitution away from labor and a decline in labor share.
  - Chirinko (2008), Grossman et al. (2017), Raval (2014): elasticity of substitution often below one, typically smaller in short-run than long-run.
  - Oberfield and Raval (2014): substitution between labor and capital of 0.84 for average manufacturing sector in Colombia.
  - Chan (2017) and Hummels et al. (2014): intermediate inputs and labor are substitutes (though complementarity is possible).

### Empirical employment results
- Estimation strategy: same baseline approach, testing whether firms more exposed to reductions in (i) output tariffs, (ii) capital good input tariffs, and (iii) non-capital good input tariffs show different responses in number of employees (Table 9).
- Aggregate effect:
  - Larger exposure to the decline in input tariffs is associated with a reduction in workers between 2010 and 2011 (Column (1), Table 9), but heterogeneity by good type matters.
- By input type:
  - Reduction in non-capital good input tariffs leads to a decline in number of workers — consistent with models where labor and intermediate inputs are substitutes; firms stop producing intermediate inputs in-house as imported intermediates become cheaper.
  - Reduction in capital goods tariffs is associated with an increase in number of employees — consistent with labor and capital being complements (elasticity between labor and capital lower than unity).
- Heterogeneity by worker type (Table 10):
  - The coefficient on capital goods tariffs is around 50% higher for manual (production) workers than for administrative workers.
  - The effect on administrative workers is not statistically significant.
- Persistence of employment effects:
  - Replacing change in investment by change in log production employees (regression analogous to Equation 12) and plotting dynamic response (Figure 5) shows:
    - Increase in production workers remains significant for four years in sectors more exposed to reduction in capital goods tariffs.
    - After four years the difference between more and less exposed sectors is not statistically significant anymore.
  - Interpretation: production workers and capital are complements for four years, but are independent thereafter (Cobb-Douglas case).

### 5  Conclusion

### Summary of core findings
- Output tariffs have no significant effect on investment.
- Decline in capital goods tariffs may substantially boost investment.
- Employment responses are heterogeneous:
  - Reduction in capital goods tariffs associated with higher employment of production workers.
  - Reduction in input tariffs on non-capital goods associated with declines in employment.
- Firms that refrain from importing due to fixed costs benefit most from reductions in capital goods tariffs, as lower costs help make importing profitable.

### Policy implications
- The effect of a tariff reduction depends on which kind of tariffs are cut:
  - Reduction in capital goods tariffs can significantly stimulate investment.
  - Reduction in tariffs on other inputs and output tariffs do not affect investment but can help boost productivity (citing Amiti and Konings, 2007; Topalova and Khandelwal, 2011).
- Heterogeneous firm responses imply targeted considerations: firms facing fixed-cost barriers to importing and those reliant on capital goods may gain disproportionately from cuts in capital goods tariffs.

*Source: wpiea2020061-print-pdf - 3.4  Import Entry*

### References

### wpiea2020061-print-pdf - References

### References
- Aghion, Philippe, Nick Bloom, Richard Blundell, Rachel Griffith, and Peter Howitt, “Competition and innovation: An inverted-U relationship,” The Quarterly Journal of Economics, 2005, 120(2), 701–728.
- Amiti, Mary and Jozef Konings, “Trade liberalization, intermediate inputs, and productivity: Evidence from Indonesia,” American Economic Review, 2007, 97(5), 1611–1638.
- Antras, Pol, Teresa C Fort, and Felix Tintelnot, “The margins of global sourcing: Theory and evidence from us firms,” American Economic Review, 2017, 107(9), 2514–64.
- Atalay, Enghin, “How important are sectoral shocks?,” American Economic Journal: Macroeconomics, 2017, 9(4), 254–80.
- Attanasio, Orazio, Pinelopi K Goldberg, and Nina Pavcnik, “Trade reforms and wage inequality in Colombia,” Journal of development Economics, 2004, 74(2), 331–366.
- Autor, David, David Dorn, and Gordon H Hanson, “The China syndrome: Local labor market effects of import competition in the United States,” American Economic Review, 2013, 103(6), 2121–68.
- Baqaee, David Rezza and Emmanuel Farhi, “Productivity and Misallocation in General Equilibrium.,” Technical Report, National Bureau of Economic Research 2017.
- Bernard, Andrew B, J Bradford Jensen, Stephen J Redding, and Peter K Schott, “Global firms,” Journal of Economic Literature, 2018, 56(2), 565–619.
- Bøler, Esther Ann, Andreas Moxnes, and Karen Helene Ulltveit-Moe, “R&D, international sourcing, and the joint impact on firm performance,” American Economic Review, 2015, 105(12), 3704–39.
- Bustos, Paula, “Trade liberalization, exports, and technology upgrading: Evidence on the impact of MERCOSUR on Argentinian firms,” American Economic Review, 2011, 101(1), 304–40.
- Chan, Mons, “How Substitutable are Labor and Intermediates?,” Unpublished working paper. University of Minnesota, 2017.
- Chirinko, Robert S, “σ: The long and short of it,” Journal of Macroeconomics, 2008, 30(2), 671–686.
- Eaton, Jonathan and Samuel Kortum, “Trade in capital goods,” European Economic Review, 2001, 45(7), 1195–1235.
- Estevadeordal, Antoni and Alan M Taylor, “Is the Washington Consensus dead? Growth, openness, and the great liberalization, 1970s–2000s,” Review of Economics and Statistics, 2013, 95(5), 1669–1690.
- Felbermayr, Gabriel J, Feodora Teti, and Erdal Yalcin, “On the profitability of trade deflection and the need for rules of origin,” 2018.
- Fernandes, Ana M., “Trade policy, trade volumes and plant-level productivity in Colombian manufacturing industries,” Journal of International Economics, 2007, 71(1), 52 – 71.
- Goldberg, Pinelopi Koujianou, Amit Kumar Khandelwal, Nina Pavcnik, and Petia Topalova, “Imported intermediate inputs and domestic product growth: Evidence from India,” The Quarterly Journal of Economics, 2010, 125(4), 1727–1767.
- Grossman, Gene M, Elhanan Helpman, Ezra Oberfield, and Thomas Sampson, “Balanced growth despite Uzawa,” American Economic Review, 2017, 107(4), 1293–1312.
- Gutiérrez, Germán and Thomas Philippon, “Declining Competition and Investment in the US,” 2017.
- Halpern, László, Miklós Koren, and Adam Szeidl, “Imported inputs and productivity,” American Economic Review, 2015, 105(12), 3660–3703.
- Hsieh, Chang-Tai and Peter J Klenow, “Relative prices and relative prosperity,” American Economic Review, 2007, 97(3), 562–585.
- Hummels, David, Rasmus Jørgensen, Jakob Munch, and Chong Xiang, “The wage effects of offshoring: Evidence from Danish matched worker-firm data,” American Economic Review, 2014, 104(6), 1597–1629.
- Jones, Charles I, “Economic growth and the relative price of capital,” Journal of Monetary Economics, 1994, 34(3), 359–382.
- Kalemli-Ozcan, Sebnem, Luc Laeven, and David Moreno, “Debt Overhang, Rollover Risk, and Corporate Investment: Evidence from the European Crisis,” Technical Report, National Bureau of Economic Research 2018.
- Karabarbounis, Loukas and Brent Neiman, “The global decline of the labor share,” The Quarterly journal of economics, 2013, 129(1), 61–103.
- Kugler, Maurice and Eric Verhoogen, “Prices, plant size, and product quality,” The Review of Economic Studies, 2011, 79(1), 307–339.
- Lian, Weicheng, Natalija Novta, Evgenia Pugacheva, Yannick Timmer, and Petia B Topalova, “The Price of Capital Goods: A Driver of Investment Under Threat,” IMF Working Paper, 2019.
- Long, J Bradford De and Lawrence H Summers, “Equipment investment and economic growth,” The Quarterly Journal of Economics, 1991, 106(2), 445–502.
- Long, J Bradford De and Lawrence H Summers, “How strongly do developing economies benefit from equipment investment?,” Journal of Monetary Economics, 1993, 32(3), 395–415.
- Melitz, Marc J, “The impact of trade on intra-industry reallocations and aggregate industry productivity,” Econometrica, 2003, 71(6), 1695–1725.
- Muendler, Marc-Andreas, “Trade, technology and productivity: a study of brazilian manufacturers 1986-1998,” 2004.
- Oberfield, Ezra and Devesh Raval, “Micro data and macro technology,” Technical Report, National Bureau of Economic Research 2014.
- Pavcnik, Nina, “Trade liberalization, exit, and productivity improvements: Evidence from Chilean plants,” The Review of Economic Studies, 2002, 69(1), 245–276.
- Peter, Alessandra and Cian Ruane, “The Aggregate Importance of Intermediate Input Substitutability,” Job market paper, 2017.
- Pierce, Justin R and Peter K Schott, “Investment responses to trade liberalization: Evidence from US industries and establishments,” Journal of International Economics, 2018, 115, 203–222.
- Raval, Devesh R, “The micro elasticity of substitution and non-neutral technology,” The RAND Journal of Economics, 2014.
- Restuccia, Diego and Carlos Urrutia, “Relative prices and investment rates,” Journal of Monetary Economics, 2001, 47(1), 93–121.
- Topalova, Petia and Amit Khandelwal, “Trade liberalization and firm productivity: The case of India,” Review of Economics and Statistics, 2011, 93(3), 995–1009.
- Torres, Mauricio and Germán Romero, “Efectos de la reforma estructural arancelaria en la protección efectiva arancelaria de la economía colombiana,” Cuadernos de Economía, 2013, 32(59), 265–303.

### Tables — Key Statistics and Findings
- Table 1: Descriptive Statistics (selected)
  - ∆Investment: mean -0.0023, p50 -0.0210, p25 0.0172, p75 0.122
  - investment_2010: mean 0.0571, p50 0.0116, p25 0.0629, sd 0.109
  - investment_2011: mean 0.0548, p50 0.0121, p25 0.0619, sd 0.103
  - employees: 75.4625 mean, p50 12, p25 70, p75 148.9
  - ln_sales: mean 14.78, p50 14.46, p25 13.48, p75 15.80, sd 1.745
  - ln_fixed_assets: mean 13.68, p50 13.41, p25 12.27, p75 14.89, sd 2.096
  - importer_2010: mean 0.221, p50 0, p75 0.415
  - importer_2011: mean 0.206, p50 0, p75 0.405
  - import_entry: mean 0.0393, p50 0, p75 0.194
  - Observations: 8498

- Table 2: Descriptive Statistics – Reduction in Tariffs (selected)
  - ∆T_I_2011: mean -0.942, p10 -2.007, p25 -1.449, p50 -0.783, p75 -0.300, p90 -0.0170, sd 0.849, Observations 132
  - ∆T_I,C_2011: mean -0.296, p10 -0.820, p25 -0.143, p50 -0.0336, p75 -0.00439, p90 -0.000347, sd 0.727
  - ∆T_I,¬C_2011: mean -0.830, p10 -1.834, p25 -1.274, p50 -0.589, p75 -0.250, p90 -0.0110, sd 0.857
  - ∆T_O_2011: mean -4.416, p10 -7.364, p25 -6.391, p50 -3.735, p75 -2.781, p90 -1.900, sd 2.094

- Table 3: Alternative Tariff Measure (selected)
  - ∆ ̃T_I_2011: mean -4.511, p10 -7.597, p25 -5.030, p50 -4.463, p75 -3.563, p90 -2.516, sd 1.668, Observations 110
  - ∆ ̃T_I,C_2011: mean -0.436, p10 -0.861, p25 -0.531, p50 -0.249, p75 -0.140, p90 -0.0319, sd 0.497

- Table 4: Baseline regression of ∆Investment on tariff measures (columns (1)-(4))
  - Column (1): ∆Input Tariffs coefficient -0.116 (std. error 0.119); Observations 9110; Controls Yes
  - Column (2): ∆Capital Input Tariffs coefficient -0.377 ∗∗∗ (std. error 0.055); Observations 9110; Controls Yes
  - Column (3): ∆Capital Input Tariffs coefficient -0.371 ∗∗∗ (std. error 0.052); ∆Other Input Tariffs -0.0478 (0.036); Observations 9110; Controls Yes
  - Column (4): ∆Capital Input Tariffs coefficient -0.370 ∗∗∗ (std. error 0.051); ∆Other Input Tariffs -0.0547 (0.055); ∆Output Tariffs 0.00553 (0.040); Observations 9110; Controls Yes
  - Note: standard errors clustered at the sector level; ∗, ∗∗, ∗∗∗ represent 5%, 1%, and 0.1% significance levels respectively.

- Table 5: Instrumental Variable Regression (columns (1)-(3))
  - Column (1) OLS: Capital Input Tariffs 2010 coefficient 0.193 ∗∗∗ (0.055); Observations 9110
  - Column (2) OLS: ∆Capital Input Tariffs coefficient -0.386 ∗∗∗ (0.056); Observations 9110
  - Column (3) IV: ∆Capital Input Tariffs coefficient -0.373 ∗∗∗ (0.079); Observations 9110
  - Note: IV instruments change in capital goods tariffs between 2011 and 2010 with its level in 2010; standard errors clustered at sector level.

- Table 6: Interaction with Size Quartiles (Employees and Sales)
  - Baseline ∆Capital Input Tariffs coefficients: -0.0155 (0.141) and -0.367 ∗∗∗ (0.123) in columns shown
  - 2nd quartile×∆Capital Input Tariffs: -0.384 (0.261), -0.385 (0.261), 0.342 (0.408), 0.342 (0.406)
  - 3rd quartile×∆Capital Input Tariffs: -0.648 ∗∗ (0.285), -0.647 ∗∗ (0.287), -0.421 ∗∗ (0.170), -0.423 ∗∗ (0.171)
  - 4th quartile×∆Capital Input Tariffs: -0.429 ∗∗ (0.160), -0.428 ∗∗ (0.162), -0.0617 (0.122), -0.0625 (0.120)
  - Observations 9110; Controls Yes; Columns (2) and (4) include sector fixed effects.

- Table 7: Import Entry - Probit Regression (dependent: Import Entry)
  - ∆Input Tariffs coefficient -0.0206 (0.032) in column (1)
  - ∆Capital Input Tariffs coefficients: -0.0560 ∗ (0.029), -0.0546 ∗∗ (0.026), -0.0597 ∗ (0.032) across columns
  - ∆Other Input Tariffs: -0.00964 (0.028) and 0.0329 (0.028) in columns indicated
  - ∆Output Tariffs coefficient -0.0361 ∗∗ (0.018) in column (4)
  - Observations 9110; Controls Yes; standard errors clustered at sector level.

- Table 8: Baseline with Additional Controls (dependent: ∆Investment)
  - ∆Capital Input Tariffs coefficients: -0.375 ∗∗∗ (0.049), -0.372 ∗∗∗ (0.050), -0.373 ∗∗∗ (0.051), -0.376 ∗∗∗ (0.050) across columns (1)-(4)
  - ∆Other Input Tariffs: -0.0918 ∗∗ (0.042), -0.0544 (0.053), -0.0493 (0.055), -0.0755 (0.056)
  - ∆Output Tariffs: 0.00478 (0.037), 0.00516 (0.038), 0.00807 (0.041), 0.0144 (0.040)
  - lagged ln(Fixed Assets): -0.0806 (0.061), -0.138 (0.152), -0.0364 (0.146)
  - lagged ln(Sales): 0.0806 (0.153), 0.0082 (0.149)
  - lagged ln(TFP): -0.157 (0.146), -0.143 (0.146)
  - ∆ln(Fixed Assets): 0.907 ∗∗∗ (0.225)
  - ∆ln(Sales): -0.0465 (0.209)
  - Observations: 9105, 9105, 9105, 9105

- Table 9: The effect of tariffs on employment (dependent: ∆Employment)
  - ∆Tariffs coefficient 1.258 ∗∗∗ (0.426) in column (1)
  - ∆Capital Input Tariffs coefficients: -0.666 (0.401), -0.847 ∗∗∗ (0.218), -0.877 ∗∗∗ (0.243) across columns
  - ∆Other Input Tariffs coefficients: 1.428 ∗∗∗ (0.359), 1.718 ∗∗∗ (0.424)
  - ∆Output Tariffs -0.233 (0.164)
  - Observations 8954; Controls Yes; additional specifications include lagged logs of revenue TFP and changes in logs as in Kalemli-Ozcan et al. (2018).

- Table 10: Employment by type (Production vs Administrative)
  - ∆Tariffs: 1.244 ∗∗ (0.478) for Production in column (1); 0.934 ∗∗ (0.320) for Admin in column (2)
  - ∆Capital Input Tariffs: -0.829 ∗ (0.411), -0.516 (0.495), -1.011 ∗∗∗ (0.243), -0.652 (0.416), -1.051 ∗∗∗ (0.267), -0.664 (0.383) across columns
  - ∆Other Input Tariffs: 1.439 ∗∗∗ (0.353), 1.065 ∗∗∗ (0.310), 1.824 ∗∗∗ (0.398), 1.183 ∗∗ (0.413)
  - ∆Output Tariffs: -0.308 ∗ (0.166), -0.0938 (0.185)
  - Observations: 8960 (Production columns) and 8961 (Admin columns); Controls Yes.

- Table 11: Baseline Regression using Alternative Tariff Measures (dependent: ∆Investment)
  - ∆Input Tariffs coefficient 0.0515 (0.038) in column (1)
  - ∆Capital Input Tariff coefficient -0.587 ∗∗∗ (0.169) in column (2)
  - ∆Capital Input Tariffs coefficients: -0.536 ∗∗∗ (0.171), -0.572 ∗∗∗ (0.179) across columns (3)-(4)
  - ∆Other Input Tariffs: 0.0421 (0.045), 0.0789 (0.060)
  - ∆Output Tariffs: -0.0840 (0.065)
  - Observations 8849; Controls Yes; standard errors clustered at sector level.

### Figures — Notes and Captions
- Figure 1: Distribution of MFN Tariffs changes in Colombia in 2011
  - Note: this figure plots the density of changes in Colombian MFN tariff rates. Source: Felbermayr et al. (2018).
- Figure 2: Distribution of MFN Tariffs changes in Colombia in 2011
  - Note: this figure plots the density of changes in Colombian MFN tariff rates between 2011 and 2010 against the level of tariffs in 2010. Source: Felbermayr et al. (2018).
- Figure 3: Evolution of Input Tariffs Over Time
  - Note: plots the evolution of input tariffs defined in subsection 2.2 over time for two sectors. The high-exposure sector experienced the biggest reduction in input tariffs in 2011 while the low-exposure sector experienced the lowest decline. Source: Felbermayr et al. (2018).
- Figure 4: Dynamic Response of Investments to Capital Goods Input Tariffs cut
  - Note: plots estimated coefficients of Equation 12. Left-hand side: difference between investment rates in year t (plotted on horizontal axis) and investment rate in 2010. Right-hand side: measure of reduction in capital goods input tariffs in 2011 (subsection 2.2).
- Figure 5: Dynamic Response of Production Workers to Capital Goods Input Tariffs cut
  - Note: plots estimated coefficients of Equation 12. Left-hand side: difference between log number of production workers in year t (plotted on horizontal axis) and log number of production workers in 2010. Right-hand side: measure of reduction in capital goods input tariffs in 2011 (subsection 2.2).

*Source: wpiea2020061-print-pdf - References*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2020/english/wpiea2020061-print-pdf.pdf_