## wpiea2019078 - References (excerpt)

## Source details

**Canonical URL:** [wpiea2019078 - References (excerpt)](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019078.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wpiea2019078.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019078.pdf.json)

---

### I. Introduction and research objective
- Corporate dynamism via firm entry and exit is essential for competitiveness, innovation and growth; establishing a conducive ecosystem for firm survival is necessary for sustainable private sector development and socioeconomic cohesion.
- Research gap: "to the best of our knowledge, however, there is no paper analyzing the impact of taxation on firms’ survival prospects."
- Main objective: "shed light on how taxation affects the probability of firm survival across countries."
- Empirical approach: hazard models on a dataset covering more than 4 million nonfinancial firms from 21 countries and 21.5 million firm-year observations over the period 1995–2015.
- Main explanatory variable: firm-specific effective marginal tax rate (EMTR) estimated for each firm rather than the statutory CIT rate.
- Sample summary facts:
  - Average failure rate among nonfinancial firms from 167 countries in Orbis: 3.5 percent of active firms during the 1995-2015 period.
  - Average failure rate in advanced economies: 4 percent.
  - Main sample after cleaning: 4,026,648 unique firms from 21 countries (14 advanced and 7 developing) with a total of 21,450,725 firm-year observations over the period from 1995 to 2015.
- High-level finding: "a lower level of effective marginal tax rate increases survival probability among nonfinancial firms."
- Heterogeneity: the effect of taxation on firm survival is "significantly greater in developing countries than advanced economies."
- Nonlinear finding: "a nonlinear relationship between the firm-specific EMTR and the probability of corporate failure, which implies that taxation becomes a detriment to firm survival at higher levels."

### II. Data overview, cleaning, and sample construction
- Data source: Orbis database (harmonized firm-level financial data including balance sheets and income statements).
- Country selection and cleaning rules:
  - Threshold of 10,000 annual observations per country.
  - Filter out firms with negative values of total assets, employment, operating revenues, and short-term loans and long-term debt in any given year (following Gal (2013) and Kalemli-Ozcan and others (2015)).
  - Restricted sample to firms and countries with information required to compute EMTR and TFP.
  - Exclude 1 percent of observations on both tails of the distribution of variables to minimize extreme outliers.
- Geographic and sectoral coverage: uneven across 21 countries and 10 nonfinancial sectors grouped by NACE; "the substantial majority is concentrated in Europe, accounting for 95 percent of nonfinancial firms covered in our sample."

### III. EMTR estimation methodology
- Conceptual definition:
  - EMTR = (p* − r) / p*, where p* is expected pre-tax (gross) rate of return and r is expected post-tax (net) rate of return on a marginal investment.
  - p* = I/Q − δ (minimum rate of return covering acquisition cost Q and economic depreciation δ).
  - r is equivalent to nominal market interest rate i minus inflation rate π (common across firms within a country and time).
- Interpretation: EMTR is "a consolidated indicator of the various tax factors that affect the cost of capital relative to a normal rate of return."
- Estimation inputs and parameter values (assumed common across firms within a country/time):
  - p* = 0.2
  - r = 0.05
  - π = 0.02
  - Economic depreciation (δ) by asset type:
    - machinery: 0.1225
    - buildings: 0.0361
    - land and inventories: 0
    - intangible assets: 0.15
- Estimation note: firm-level EMTRs calculated using company-specific information plus country-level parameters including statutory CIT rate, depreciation rules, inflation, nominal interest rate, and real before-tax return on equity.
- Rationale: forward-looking EMTR preferred over backward-looking measures (backward-looking effective rates "are inherently prone to endogeneity inducing biased estimation results").

### IV. TFP estimation methodology
- Purpose: incorporate efficiency differences across enterprises into survival probability analysis.
- Production function:
  - Cobb-Douglas with constant returns to scale:
    - y_it = β_k k_it + β_l l_it + μ_i + μ_t + ε_it  (logs)
  - Real capital stock via perpetual inventory method:
    - k_it = k_{i,t-1}(1−δ_{i,t-1}) + I_it (net fixed investment in real terms, using industry-specific price deflators)
  - TFP defined as the residual from the production function.
- Estimation technique: Levinsohn and Petrin (2003) using intermediate inputs as proxies.
  - First stage: estimate β̂_l by OLS and approximate φ_i(k_it, m_it) with a third-order polynomial in k_it and m_it.
  - Second stage: estimate ω̂_i using:
    - ω̂_it = φ̂_it − β_k* k_it − β_m* m_it
  - Residual computed and interacted with instruments (real capital stock and material costs) using moment conditions.

### V. Empirical findings on firm survival and determinants
- Main tax-related result:
  - "The tax burden—measured by the firm-specific EMTR—exerts an adverse effect on companies’ survival prospects. In other words, a lower level of EMTR increases the survival probability among firms in our sample."
  - Effect is statistically and economically important and "remains robust when we partition the sample into country subgroups."
- Baseline coefficient estimates (dependent variable: probability of failure; coefficients reported):
  - Age: -0.297*** (All), -0.296*** (Advanced), -0.214*** (Emerging) [0.002; 0.002; 0.008]
  - Size: -0.035*** (All), -0.044*** (Advanced), 0.054*** (Emerging) [0.002; 0.002; 0.005]
  - Profitability: -1.494*** (All), -1.803*** (Advanced), -0.741*** (Emerging) [0.017; 0.023; 0.013]
  - Capital Intensity: -0.089*** (All), -0.095*** (Advanced), -0.061*** (Emerging) [0.001; 0.001; 0.004]
  - TFP: -0.130*** (All), -0.088*** (Advanced), -0.418*** (Emerging) [0.003; 0.003; 0.007]
  - Leverage: 0.401*** (All), 0.361*** (Advanced), 0.143*** (Emerging) [0.008; 0.009; 0.028]
  - EMTR: 3.951*** (All), 3.943*** (Advanced), 3.927*** (Emerging) [0.097; 0.113; 0.404]
- Sample sizes and failure counts (baseline):
  - # of observations: 12,877,294 (All), 11,719,946 (Advanced), 1,157,348 (Emerging)
  - # of firms: 2,555,686 (All), 2,288,437 (Advanced), 267,249 (Emerging)
  - # of failures: 446,983 (All), 412,702 (Advanced), 34,281 (Emerging)
- Nonlinear taxation effects (EMTR and EMTR^2 included):
  - All Countries: EMTR -5.913*** [0.436]; EMTR^2 20.122*** [0.904]
  - Advanced: EMTR -6.976*** [0.354]; EMTR^2 22.182*** [0.626]
  - Emerging: EMTR 4.064*** [0.802]; EMTR^2 -0.242 [1.576]
  - Interpretation: For All and Advanced samples, negative EMTR linear term and positive EMTR^2 imply taxation becomes detrimental to survival at higher EMTR levels; Emerging sample patterns less clear.
- Heterogeneity results:
  - By age and size (EMTR coefficients):
    - Young firms: EMTR 1.458*** [0.149]; # observations 5,394,715; # firms 1,707,355; # failures 235,760
    - Old firms: EMTR 6.494*** [0.149]; # observations 7,482,579; # firms 1,371,568; # failures 211,223
    - Small firms: EMTR 3.472*** [0.185]; # observations 2,344,254; # firms 879,179; # failures 126,153
    - Large firms: EMTR 4.132*** [0.192]; # observations 3,662,299; # firms 739,288; # failures 92,031
  - Sectoral EMTR coefficients (coefficient [std. error]; # observations):
    - (1) Accommodation and food services: 1.551*** [0.252]; 869,025
    - (2) Construction: 1.224*** [0.267]; 2,135,058
    - (3) Information technology: 8.701*** [0.744]; 414,096
    - (4) Manufacturing: 8.667*** [0.259]; 2,972,510
    - (5) Professional and administrative activities: 5.681*** [0.336]; 1,181,842
    - (6) Real estate: -0.055 [0.387]; 369,748 (not significant)
    - (7) Transportation and storage: 7.928*** [0.497]; 591,338
    - (8) Utilities: 3.183*** [1.201]; 110,736
    - (9) Wholesale and retail trade: 3.145*** [0.167]; 4,232,941
  - Interpretation: EMTR positive and statistically significant across all sectors except real estate; larger magnitudes in capital-intensive sectors (information technology, manufacturing).
- Robustness checks:
  - Inclusion of personal income tax (PIT) rate: EMTR effect remains negative and significant.
  - Alternative methodologies: accelerated failure time (AFT) models produce consistent findings.
  - Multilevel mixed-effects parametric survival model with fixed and random effects aligns with baseline Cox results.

### VI. Data, descriptive statistics, and sample composition
- Sectoral composition (Table 1: Number of Firms; Percent):
  - Accommodation and Food Services: 338,592; 8.4
  - Construction: 699,047; 17.4
  - Information Technology: 157,500; 3.9
  - Manufacturing: 710,620; 17.7
  - Mining: 10,885; 0.3
  - Professional and Administrative Activities: 503,135; 12.5
  - Real Estate: 148,467; 3.7
  - Transportation and Storage: 192,739; 4.8
  - Utilities: 33,920; 0.8
  - Wholesale and Retail Trade: 1,231,743; 30.6
  - Total: 4,026,648; 100.0
- Sample period: 1995-2015
- Panel: 4,026,648 firms from 21 countries
- Selected descriptive statistics (Table 2 — selected):
  - Age [log]: Obs. 20.91 ml; Mean 2.14; Std. Dev. 0.95 — Advanced: Obs. 18.23 ml; Mean 2.21; Std. Dev. 0.95 — Developing: Obs. 2.68 ml; Mean 1.68; Std. Dev. 0.87
  - Size [log]: Obs. 21.39 ml; Mean 13.55; Std. Dev. 2.33 — Advanced: Obs. 18.49 ml; Mean 13.61; Std. Dev. 2.33 — Developing: Obs. 2.90 ml; Mean 13.17; Std. Dev. 2.33
  - Profitability: Obs. 20.64 ml; Mean 0.01; Std. Dev. 0.21 — Advanced: Obs. 17.84 ml; Mean 0.01; Std. Dev. 0.16 — Developing: Obs. 2.80 ml; Mean 0.01; Std. Dev. 0.40
  - Capital Intensity: Obs. 21.45 ml; Mean 10.19; Std. Dev. 2.20 — Advanced: Obs. 18.54 ml; Mean 10.15; Std. Dev. 2.19 — Developing: Obs. 2.91 ml; Mean 10.41; Std. Dev. 2.26
  - Leverage: Obs. 17.02 ml; Mean 0.18; Std. Dev. 0.22 — Advanced: Obs. 15.32 ml; Mean 0.18; Std. Dev. 0.23 — Developing: Obs. 1.70 ml; Mean 0.10; Std. Dev. 0.19
  - TFP [log]: Obs. 21.45 ml; Mean 7.20; Std. Dev. 1.12 — Advanced: Obs. 18.54 ml; Mean 7.19; Std. Dev. 1.11 — Developing: Obs. 2.91 ml; Mean 7.26; Std. Dev. 1.17
  - EMTR: Obs. 16.39 ml; Mean 0.28; Std. Dev. 0.07 — Advanced: Obs. 14.82 ml; Mean 0.29; Std. Dev. 0.06 — Developing: Obs. 1.57 ml; Mean 0.20; Std. Dev. 0.07
- Appendix Table A1 — country subtotals:
  - Advanced Economies subtotal:
    - Number of firms: 3,410,389
    - Percent of total number of firms: 84.7
    - Number of observations: 18,542,420
    - Percent of total number of observations: 86.4
  - Emerging Economies subtotal:
    - Number of firms: 616,259
    - Percent of total number of firms: 15.3
    - Number of observations: 2,908,305
    - Percent of total number of observations: 13.6
  - Total:
    - Number of firms: 4,026,648
    - Percent of total number of firms: 100.0
    - Number of observations: 21,450,725
    - Percent of total number of observations: 100.0
  - Selected country entries (exact figures):
    - France: 739,297 firms; 18.4 percent; 3,755,091 observations; 17.5 percent
    - Italy: 774,463 firms; 19.2 percent; 4,186,215 observations; 19.5 percent
    - Spain: 831,494 firms; 20.7 percent; 5,285,465 observations; 24.6 percent
    - Korea: 190,408 firms; 4.7 percent; 680,180 observations; 3.2 percent
    - Portugal: 261,315 firms; 6.5 percent; 1,383,457 observations; 6.5 percent

### VII. Econometric methodology and robustness
- Survival definition: firm failed if status is receivership, liquidation, or dissolved.
- Treatment of panel features: left truncation and right censoring accounted for; year of first appearance used as time at risk; exclude observations when a firm drops out of the database.
- Baseline model: Cox proportional hazard in discrete time using complementary log-log specification:
  - Hazard: ℎi(Xiit) = 1 − exp{−exp(Xiit′ tβ + θ(t))}
  - Extended with unobserved heterogeneity μi: ℎi(Xiit) = 1 − exp{−exp(Xiit′β + θ(t)) + μi}
- Covariates: age, size, profitability, leverage, capital intensity, productivity; sector and country fixed effects.
- Estimation: random effects panel methods for binary dependent variable; robust standard errors clustered at the firm level.
- Robustness model coefficients (Appendix Table A2 — selected EMTR and other model estimates):
  - EMTR across models:
    - Gompertz (PH): 9.366*** [0.105]
    - Exponential (PH/AFT): 7.986*** [0.058]
    - Log-logistic (AFT): 8.560*** [0.058]
    - PIT: 7.072*** [0.067]
    - Mixed Effects: 5.578*** [0.030]
  - PIT rate in PIT model only: 0.032*** [0.000]
  - Model sample sizes and diagnostics (exact figures):
    - # of observations: Gompertz (PH) 11,639,304; Exponential (PH/AFT) 11,639,304; Log-logistic (AFT) 11,639,304; PIT 11,401,501; Mixed Effects 11,639,304
    - # of firms: Gompertz (PH) 2,380,710; Exponential (PH/AFT) 2,380,710; Log-logistic (AFT) 2,380,710; PIT 2,297,867; Mixed Effects 2,380,710
    - # of failures: Gompertz (PH) 667,550; Exponential (PH/AFT) 667,550; Log-logistic (AFT) 667,550; PIT 429,747; Mixed Effects 667,550
    - Wald tests and log-likelihoods reported per model in source.
- Note on inference: Standard errors reported in brackets. *** p<0.01, ** p<0.05, * p<0.1.

### VIII. Main conclusions and policy implications
- Main conclusions:
  - Corporate tax burden (firm-specific EMTR) exerts a highly significant adverse effect on firms’ survival prospects; lower EMTR increases survival probability.
  - Relationship is nonlinear: taxation becomes particularly detrimental to survival at higher EMTR levels.
  - Adverse effects of taxation on firm survival appear stronger in developing countries than in advanced economies, though data constraints limit analysis.
  - Firm characteristics matter: older and larger firms have lower failure risk; higher profitability reduces failure risk; higher leverage increases failure risk; capital intensity and higher TFP reduce failure risk.
- Policy implications and recommendations:
  - "Tax systems can be designed better to improve efficiency and boost investment that foster innovation and job creation."
  - Adopt a "coherent and fair approach to taxation" to reduce legal uncertainty and distortions in resource allocation.
  - Tax policy and administration should aim to:
    - cut the costs of compliance,
    - facilitate entrepreneurship and innovation,
    - encourage alternative sources of financing by particularly addressing the corporate debt bias.
  - Policymakers should focus not merely on reducing statutory CIT rates but on "leveling the playing field for all firms by rationalizing differentiated tax treatments across sectors, capital asset types and sources of financing."
  - EMTR relevance: "the EMTR holds a special key by influencing firms’ investment decisions and the probability of survival over time."

*Source: wpiea2019078 PDF chapter/section (Orbis-based firm-level analysis, tables and results as provided in the source content).*

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

### wpiea2019078 - References

### I. Introduction and research objective
- Corporate dynamism via firm entry and exit is essential for competitiveness, innovation and growth; establishing a conducive ecosystem for firm survival is necessary for sustainable private sector development and socioeconomic cohesion.
- Research gap identified: "to the best of our knowledge, however, there is no paper analyzing the impact of taxation on firms’ survival prospects."
- Main objective: "shed light on how taxation affects the probability of firm survival across countries."
- Key empirical approach: hazard models applied to a comprehensive dataset covering more than 4 million nonfinancial firms from 21 countries and 21.5 million firm-year observations over the period 1995–2015.
- Main explanatory variable: firm-specific effective marginal tax rate (EMTR) estimated for each firm rather than using the statutory corporate income tax (CIT) rate.
- Sample summary facts:
  - Average failure rate among nonfinancial firms from 167 countries in Orbis: 3.5 percent of active firms during the 1995-2015 period.
  - Average failure rate in advanced economies: 4 percent.
  - Main sample after cleaning: 4,026,648 unique firms from 21 countries (14 advanced and 7 developing) with a total of 21,450,725 firm-year observations over the period from 1995 to 2015.
- Key high-level finding: "a lower level of effective marginal tax rate increases survival probability among nonfinancial firms."
- Heterogeneity: the effect of taxation on firm survival is "significantly greater in developing countries than advanced economies."
- Nonlinear finding: "a nonlinear relationship between the firm-specific EMTR and the probability of corporate failure, which implies that taxation becomes a detriment to firm survival at higher levels."

### II. Data overview, cleaning, and sample construction
- Data source: Orbis database (harmonized firm-level financial data including balance sheets and income statements).
- Country selection and cleaning rules:
  - Threshold of 10,000 annual observations per country.
  - Filter out firms with negative values of total assets, employment, operating revenues, and short-term loans and long-term debt in any given year (following Gal (2013) and Kalemli-Ozcan and others (2015)).
  - Restricted sample to firms and countries with information required to compute EMTR and TFP.
  - Exclude 1 percent of observations on both tails of the distribution of variables to minimize extreme outliers.
- Geographic and sectoral coverage note: sample is unevenly distributed across 21 countries and 10 nonfinancial sectors grouped by NACE; "the substantial majority is concentrated in Europe, accounting for 95 percent of nonfinancial firms covered in our sample."

### III. EMTR estimation methodology
- Conceptual definition:
  - EMTR = (p* − r) / p*, where p* is expected pre-tax (gross) rate of return and r is expected post-tax (net) rate of return on a marginal investment.
  - p* = I/Q − δ (minimum rate of return covering acquisition cost Q and economic depreciation δ).
  - r is equivalent to nominal market interest rate i minus inflation rate π (common across firms within a country and time).
- Interpretation: EMTR is "a consolidated indicator of the various tax factors that affect the cost of capital relative to a normal rate of return."
- Estimation inputs and parameter values (assumed common across firms within a country/time):
  - p* = 0.2
  - r = 0.05
  - π = 0.02
  - Economic depreciation (δ) by asset type:
    - machinery: 0.1225
    - buildings: 0.0361
    - land and inventories: 0
    - intangible assets: 0.15
- Estimation note: firm-level EMTRs calculated using company-specific information plus country-level parameters including statutory CIT rate, depreciation rules, inflation, nominal interest rate, and real before-tax return on equity.
- Rationale: forward-looking EMTR preferred over backward-looking measures (backward-looking effective rates "are inherently prone to endogeneity inducing biased estimation results").

### IV. TFP estimation methodology
- Purpose: incorporate efficiency differences across enterprises into survival probability analysis.
- Production function setup:
  - Standard Cobb-Douglas with constant returns to scale:
    - y_it = β_k k_it + β_l l_it + μ_i + μ_t + ε_it  (logs)
  - Real capital stock (k_it) computed via perpetual inventory method:
    - k_it = k_{i,t-1}(1−δ_{i,t-1}) + I_it (net fixed investment in real terms, using industry-specific price deflators)
  - TFP defined as the residual from the production function.
- Estimation technique to address simultaneity:
  - Levinsohn and Petrin (2003) algorithm using intermediate inputs (energy or material costs) as proxies.
  - First stage: estimate β̂_l by OLS and approximate φ_i(k_it, m_it) with a third-order polynomial in k_it and m_it.
  - Second stage: estimate ω̂_i for values of β_k* and β_m* using:
    - ω̂_it = φ̂_it − β_k* k_it − β_m* m_it
  - Residual computed and interacted with instruments (real capital stock and material costs) using moment conditions that capital stock in period t is driven by prior-period investment, and lagged material costs uncorrelated with current shock.

### V. Empirical findings on firm survival and determinants
- Main tax-related result:
  - "The tax burden—measured by the firm-specific EMTR—exerts an adverse effect on companies’ survival prospects. In other words, a lower level of EMTR increases the survival probability among firms in our sample."
  - This effect is statistically and economically important and "remains robust when we partition the sample into country subgroups."
- Heterogeneous effects:
  - Taxation effect on survival is "significantly greater in developing countries than advanced economies."
  - Nonlinear relationship: taxation "becomes a detriment to firm survival at higher levels."
- Other firm-level determinants (findings consistent with prior literature):
  - Age and size: "older and larger firms experiencing a lower risk of failure."
  - Profitability: probability of failure diminishes with degree of profitability.
  - Indebtedness: probability of failure increases with level of indebtedness.
  - Capital intensity and TFP: "capital intensity and TFP play a significant role in reducing the probability of failure."

### VI. Policy implications and recommendations
- High-level policy message: "Tax systems can be designed better to improve efficiency and boost investment that foster innovation and job creation."
- Specific policy guidance preserved from source:
  - Adopt a "coherent and fair approach to taxation" to reduce legal uncertainty and distortions in resource allocation.
  - Tax policy and administration should aim to:
    - cut the costs of compliance,
    - facilitate entrepreneurship and innovation,
    - encourage alternative sources of financing by particularly addressing the corporate debt bias.
  - Policymakers should focus not merely on reducing statutory CIT rates but on "leveling the playing field for all firms by rationalizing differentiated tax treatments across sectors, capital asset types and sources of financing."
- EMTR relevance: "the EMTR holds a special key by influencing firms’ investment decisions and the probability of survival over time."

*Source: wpiea2019078 - References (excerpt provided from the IMF PDF content unit)*

### 2000. In terms of sectoral coverage, the dataset is based on the NACE classification of economic

### wpiea2019078 - 2000. In terms of sectoral coverage, the dataset is based on the NACE classification of economic

### Data and sample
- Coverage: nonfinancial sectors excluding agriculture, public administration and defense, activities of extraterritorial organizations and bodies, and activities of households as employers and for own use.
- Sectoral composition (Table 1: Number of Firms; Percent):
  - Accommodation and Food Services: 338,592; 8.4
  - Construction: 699,047; 17.4
  - Information Technology: 157,500; 3.9
  - Manufacturing: 710,620; 17.7
  - Mining: 10,885; 0.3
  - Professional and Administrative Activities: 503,135; 12.5
  - Real Estate: 148,467; 3.7
  - Transportation and Storage: 192,739; 4.8
  - Utilities: 33,920; 0.8
  - Wholesale and Retail Trade: 1,231,743; 30.6
  - Total: 4,026,648; 100.0
- Sample period: 1995-2015
- Panel: 4,026,648 firms from 21 countries
- Variables and measurements:
  - Age: log of years since inception
  - Size: log of total assets
  - Profitability: profits after tax / total assets
  - Capital intensity: capital per worker
  - Leverage: short-term and long-term debt / total assets
  - Firm-level nominal variables deflated using industry-specific deflators from OECD STAN (advanced economies) and UN National Accounts (developing countries)
  - Real capital stock: perpetual inventory method

### Descriptive statistics (Table 2 — selected)
- Observations and means (All / Advanced / Developing) for key variables:
  - Age [log]: Obs. 20.91 ml; Mean 2.14; Std. Dev. 0.95 — Advanced: Obs. 18.23 ml; Mean 2.21; Std. Dev. 0.95 — Developing: Obs. 2.68 ml; Mean 1.68; Std. Dev. 0.87
  - Size [log]: Obs. 21.39 ml; Mean 13.55; Std. Dev. 2.33 — Advanced: Obs. 18.49 ml; Mean 13.61; Std. Dev. 2.33 — Developing: Obs. 2.90 ml; Mean 13.17; Std. Dev. 2.33
  - Profitability: Obs. 20.64 ml; Mean 0.01; Std. Dev. 0.21 — Advanced: Obs. 17.84 ml; Mean 0.01; Std. Dev. 0.16 — Developing: Obs. 2.80 ml; Mean 0.01; Std. Dev. 0.40
  - Capital Intensity: Obs. 21.45 ml; Mean 10.19; Std. Dev. 2.20 — Advanced: Obs. 18.54 ml; Mean 10.15; Std. Dev. 2.19 — Developing: Obs. 2.91 ml; Mean 10.41; Std. Dev. 2.26
  - Leverage: Obs. 17.02 ml; Mean 0.18; Std. Dev. 0.22 — Advanced: Obs. 15.32 ml; Mean 0.18; Std. Dev. 0.23 — Developing: Obs. 1.70 ml; Mean 0.10; Std. Dev. 0.19
  - TFP [log]: Obs. 21.45 ml; Mean 7.20; Std. Dev. 1.12 — Advanced: Obs. 18.54 ml; Mean 7.19; Std. Dev. 1.11 — Developing: Obs. 2.91 ml; Mean 7.26; Std. Dev. 1.17
  - EMTR: Obs. 16.39 ml; Mean 0.28; Std. Dev. 0.07 — Advanced: Obs. 14.82 ml; Mean 0.29; Std. Dev. 0.06 — Developing: Obs. 1.57 ml; Mean 0.20; Std. Dev. 0.07
- Note: Source: Orbis; authors’ calculation.

### Econometric methodology
- Objective: impact of taxation (firm-specific effective marginal tax rate, EMTR) on firms’ survival probability.
- Survival definition: firm failed if status is receivership, liquidation, or dissolved.
- Treatment of panel features: left truncation and right censoring accounted for; year of first appearance used as time at risk; exclude observations when a firm drops out of the database.
- Baseline model: Cox proportional hazard in discrete time using complementary log-log specification:
  - Hazard specification: ℎi(Xiit) = 1 − exp{−exp(Xiit′ tβ + θ(t))}
  - Extended with unobserved heterogeneity μi: ℎi(Xiit) = 1 − exp{−exp(Xiit′β + θ(t)) + μi}
- Covariates: age, size, profitability, leverage, capital intensity, productivity; sector and country fixed effects.
- Estimation: random effects panel methods for binary dependent variable; robust standard errors clustered at the firm level.

### Key empirical results — baseline (Table 3)
- Main finding: EMTR exerts a positive and highly significant coefficient on the probability of failure (i.e., higher EMTR → higher probability of failure; equivalently, lower EMTR increases survival probability).
- Selected coefficient estimates (dependent variable: probability of failure; coefficients reported):
  - Age: -0.297*** (All), -0.296*** (Advanced), -0.214*** (Emerging) [standard errors: 0.002; 0.002; 0.008]
  - Size: -0.035*** (All), -0.044*** (Advanced), 0.054*** (Emerging) [0.002; 0.002; 0.005]
  - Profitability: -1.494*** (All), -1.803*** (Advanced), -0.741*** (Emerging) [0.017; 0.023; 0.013]
  - Capital Intensity: -0.089*** (All), -0.095*** (Advanced), -0.061*** (Emerging) [0.001; 0.001; 0.004]
  - TFP: -0.130*** (All), -0.088*** (Advanced), -0.418*** (Emerging) [0.003; 0.003; 0.007]
  - Leverage: 0.401*** (All), 0.361*** (Advanced), 0.143*** (Emerging) [0.008; 0.009; 0.028]
  - EMTR: 3.951*** (All), 3.943*** (Advanced), 3.927*** (Emerging) [0.097; 0.113; 0.404]
- Sample sizes and fit:
  - # of observations: 12,877,294 (All), 11,719,946 (Advanced), 1,157,348 (Emerging)
  - # of firms: 2,555,686 (All), 2,288,437 (Advanced), 267,249 (Emerging)
  - # of failures: 446,983 (All), 412,702 (Advanced), 34,281 (Emerging)
  - Fixed effects: Yes (All columns)
  - Wald test: 85,784 (All), 82,950 (Advanced), 12,343 (Emerging)
  - Log-likelihood: -4,435,071 (All), -4,147,730 (Advanced), -281,918 (Emerging)
- Interpretation: older and larger firms have lower hazard; profitability reduces failure risk; leverage increases failure risk; capital intensity and higher TFP reduce failure risk.

### Nonlinear taxation effects (Table 4)
- Evidence of nonlinear relationship between EMTR and failure probability:
  - EMTR coefficient (with EMTR squared included):
    - All Countries: EMTR -5.913*** [0.436]; EMTR^2 20.122*** [0.904]
    - Advanced: EMTR -6.976*** [0.354]; EMTR^2 22.182*** [0.626]
    - Emerging: EMTR 4.064*** [0.802]; EMTR^2 -0.242 [1.576]
- Sample sizes and fit (same as baseline):
  - # of observations: 12,877,294; # of firms: 2,555,686; # of failures: 446,983
  - Wald test: 86,177 (All), 84,900 (Advanced), 12,363 (Emerging)
  - Log-likelihood: -4,434,475 (All), -4,147,067 (Advanced), -281,918 (Emerging)
- Interpretation:
  - For All and Advanced samples, negative linear EMTR term and positive EMTR^2 imply taxation becomes detrimental to survival at higher EMTR levels.
  - For Emerging sample, coefficients on EMTR and EMTR^2 differ in sign and are less clear, possibly due to fewer observations (267,249 firms) and institutional factors (lengthy bankruptcy procedures, weak rule of law).

### Heterogeneity and robustness
- Age and size heterogeneity (Table 5):
  - EMTR effect:
    - Young firms: EMTR 1.458*** [0.149]; # observations 5,394,715; # firms 1,707,355; # failures 235,760
    - Old firms: EMTR 6.494*** [0.149]; # observations 7,482,579; # firms 1,371,568; # failures 211,223
    - Small firms: EMTR 3.472*** [0.185]; # observations 2,344,254; # firms 879,179; # failures 126,153
    - Large firms: EMTR 4.132*** [0.192]; # observations 3,662,299; # firms 739,288; # failures 92,031
  - Finding: taxation has a significantly greater effect on probability of failure as firms age; adverse impact greater for large firms than small firms relative to baseline.
- Sectoral heterogeneity (Table 6; columns correspond to sectors):
  - EMTR coefficients by sector (coefficient [std. error]):
    - (1) Accommodation and food services: 1.551*** [0.252]
    - (2) Construction: 1.224*** [0.267]
    - (3) Information technology: 8.701*** [0.744]
    - (4) Manufacturing: 8.667*** [0.259]
    - (5) Professional and administrative activities: 5.681*** [0.336]
    - (6) Real estate: -0.055 [0.387] (not significant)
    - (7) Transportation and storage: 7.928*** [0.497]
    - (8) Utilities: 3.183*** [1.201]
    - (9) Wholesale and retail trade: 3.145*** [0.167]
  - # of observations by sector: 869,025; 2,135,058; 414,096; 2,972,510; 1,181,842; 369,748; 591,338; 110,736; 4,232,941 respectively.
  - Interpretation: EMTR positive and statistically significant across all sectors except real estate; magnitude ranges from 1.6 (accommodation and food services noted in text) to 8.7 (information technology and manufacturing), indicating greater tax impact in capital-intensive sectors.
- Additional robustness checks:
  - Inclusion of personal income tax (PIT) rate: effect of EMTR on firm survival remains negative and significant.
  - Alternative methodologies: accelerated failure time (AFT) models produce consistent findings.
  - Multilevel mixed-effects parametric survival model with fixed and random effects aligns with baseline Cox results.

### Main conclusions and policy implications
- Main conclusions:
  - Corporate tax burden (firm-specific EMTR) exerts a highly significant adverse effect on firms’ survival prospects; lower EMTR increases survival probability.
  - Relationship is nonlinear: taxation becomes particularly detrimental to survival at higher EMTR levels.
  - Adverse effects of taxation on firm survival appear stronger in developing countries than in advanced economies, though data constraints limit analysis.
  - Firm characteristics matter: older and larger firms have lower failure risk; higher profitability reduces failure risk; higher leverage increases failure risk; capital intensity and higher TFP reduce failure risk.
- Policy implications:
  - Well-designed tax systems can enhance efficiency and encourage investment, innovation, and job creation.
  - The EMTR captures differences in overall tax burden and discriminatory tax treatments across sectors and firm types; policy should address distortions.
  - Reforms in tax policy and revenue administration should:
    - Cut costs of compliance
    - Facilitate entrepreneurship and innovation
    - Encourage alternative sources of financing, particularly by addressing corporate debt bias
  - Policymakers should focus not only on reducing statutory CIT rates but also on rationalizing differentiated tax treatments across sectors, capital asset types, and sources of financing to level the playing field and reduce legal uncertainty.

*Source: wpiea2019078 PDF chapter/section (Orbis-based firm-level analysis, tables and results as provided in the source content).*

### REFERENCES

### wpiea2019078 - REFERENCES

### References (selected themes and representative citations)
- Firm survival and life cycle:
  - Agarwal, R., 1997, “Survival of firms Over the Product Life Cycle,” Southern Economic Journal, Vol. 63, pp. 571–584.
  - Agarwal, R., and D. Audretsch, 2001, “Does Entry Size Matter? The Impact of the Life Cycle and Technology on Firm Survival,” Journal of Industrial Economics, Vol. 49, pp. 21–43.
  - Agarwal, R., and M. Gort, 2002, “Firm and Product Life Cycles and Firm Survival,” American Economic Review, Vol. 92, pp. 184–190.
  - Audretsch, D., 1991, “New-Firm Survival and the Technological Regime,” Review of Economics and Statistics, Vol. 60, pp. 441-450.
  - Mata, J., and P. Portugal, 1994, “Life Duration of New Firms,” Journal of Industrial Economics, Vol. 42, pp. 227-245.
  - Helmers, C., and M. Rogers, 2010, “Innovation and the Survival of New Firms in the UK,” Review of Industrial Organization, Vol. 36, pp. 227-248.
  - Zingales, L., 1998, “Survival of the Fittest or the Fattest? Exit and Financing in Trucking Industry,” Journal of Finance, Vol. 53, pp. 905-938.

- Taxation, investment, and location decisions:
  - Aghion, P., Akcigit, U., Cage, J. and Kerr, W. 2016, “Taxation, Corruption, and Growth,” European Economic Review, Vol. 86, pp. 24–51.
  - Devereux, M., and R. Griffith, 1998, “The Taxation of Discrete Investment Choices,” IFS Working Paper No. 98/16.
  - Devereux, M., and R. Griffith, 2003, “Evaluating Tax Policy for Location Decisions,” International Tax and Public Finance, Vol. 10, pp. 107-126.
  - King, M., and D. Fullerton, 1984, The Taxation of Income from Capital (Chicago: University of Chicago Press).
  - International Monetary Fund, 2017, “Upgrading the Tax System to Boost Productivity,” Fiscal Monitor, Chapter II, April 2017.

- Financial constraints, investment, and firm dynamics:
  - Clementi, L. and H. Hopenhayn, 2006, “A Theory of Financing Constraints and Firm Dynamics,” Quarterly Journal of Economics, Vol. 54, pp. 229–265.
  - Hubbard, G., 1998, “Capital Market Imperfections and Investment,” Journal of Economic Literature, Vol. 36, pp. 193-225.
  - Gilchrist, S., and E. Zakrajsek, 2007, “Investment and the Cost of Capital: New Evidence from the Corporate Bond Market,” NBER Working Paper No. 1317.
  - Bridges, S., and A. Guariglia, 2008, “Financial Constraints, Global Engagement, and Firm Survival in the U.K.: Evidence from Micro Data,” Scottish Journal of Political Economy, Vol. 55, pp. 444-464.
  - Musso, P., and S. Schiavo, 2008, “The Impact of Financial Constraints on Firm Survival and Growth,” Journal of Evolutionary Economics, Vol. 15, pp. 135-149.

- Econometric and methodological sources:
  - Cox, D., 1972, “Regression Models and Life Tables,” Journal of the Royal Statistical Society, Vol. 34, pp. 187-202.
  - Cox, D., and D. Oakes, 1984, Analysis of Survival Data (London: Chapman and Hall).
  - Kleinbaum, D., and M. Klein, 2005, Survival Analysis: A Self-Learning Text (New York: Springer).
  - Levinsohn, J., and A. Petrin, 2003, “Estimating Production Functions Using Inputs to Control for Unobservables,” Review of Economic Studies, Vol. 70, pp. 317-341.
  - Orbe, J., E. Ferreira, and V. Núñez-Antón, 2002, “Comparing Proportional Hazards and Accelerated Failure Time Models for Survival Analysis,” Statistics in Medicine, Vol. 21, pp. 3493-3510.
  - Patel, K., Kay, R. and Rowell, L., 2006, “Comparing Proportional Hazards and Accelerated Failure Time Models: An Application in Influenza,” Pharmaceutical Statistics, Vol. 5, pp. 213-224.

### Appendix Table A1. List of Countries (key figures)
- Advanced Economies subtotal:
  - Number of firms: 3,410,389
  - Percent of total number of firms: 84.7
  - Number of observations: 18,542,420
  - Percent of total number of observations: 86.4
- Emerging Economies subtotal:
  - Number of firms: 616,259
  - Percent of total number of firms: 15.3
  - Number of observations: 2,908,305
  - Percent of total number of observations: 13.6
- Total:
  - Number of firms: 4,026,648
  - Percent of total number of firms: 100.0
  - Number of observations: 21,450,725
  - Percent of total number of observations: 100.0
- Selected country entries (exact figures as listed):
  - France: 739,297 firms; 18.4 percent; 3,755,091 observations; 17.5 percent
  - Italy: 774,463 firms; 19.2 percent; 4,186,215 observations; 19.5 percent
  - Spain: 831,494 firms; 20.7 percent; 5,285,465 observations; 24.6 percent
  - Korea: 190,408 firms; 4.7 percent; 680,180 observations; 3.2 percent
  - Portugal: 261,315 firms; 6.5 percent; 1,383,457 observations; 6.5 percent

### Appendix Table A2. Robustness Checks — Estimated coefficients and model statistics
- Variables and model coefficients (with standard errors in brackets)
  - Age:
    - Gompertz (PH): -0.396*** [0.003]
    - Exponential (PH/AFT): -0.306*** [0.002]
    - Log-logistic (AFT): -0.427*** [0.002]
    - PIT: -0.267*** [0.003]
    - Mixed Effects: -0.188*** [0.002]
  - Size:
    - Gompertz (PH): 0.053*** [0.002]
    - Exponential (PH/AFT): 0.046*** [0.002]
    - Log-logistic (AFT): 0.064*** [0.002]
    - PIT: 0.017*** [0.002]
    - Mixed Effects: 0.016*** [0.008]
  - Profitability:
    - Gompertz (PH): -1.397*** [0.029]
    - Exponential (PH/AFT): -1.413*** [0.030]
    - Log-logistic (AFT): -1.762*** [0.008]
    - PIT: -1.448*** [0.015]
    - Mixed Effects: -1.689*** [0.007]
  - Capital Intensity:
    - Gompertz (PH): -0.112*** [0.001]
    - Exponential (PH/AFT): -0.103*** [0.001]
    - Log-logistic (AFT): -0.088*** [0.001]
    - PIT: -0.098*** [0.001]
    - Mixed Effects: -0.088*** [0.001]
  - TFP:
    - Gompertz (PH): -0.160*** [0.003]
    - Exponential (PH/AFT): -0.155*** [0.003]
    - Log-logistic (AFT): -0.092*** [0.002]
    - PIT: -0.196*** [0.002]
    - Mixed Effects: -0.168*** [0.002]
  - Leverage:
    - Gompertz (PH): 0.188*** [0.011]
    - Exponential (PH/AFT): 0.189*** [0.010]
    - Log-logistic (AFT): 0.108*** [0.006]
    - PIT: 0.240*** [0.008]
    - Mixed Effects: 0.389*** [0.007]
  - EMTR:
    - Gompertz (PH): 9.366*** [0.105]
    - Exponential (PH/AFT): 7.986*** [0.058]
    - Log-logistic (AFT): 8.560*** [0.058]
    - PIT: 7.072*** [0.067]
    - Mixed Effects: 5.578*** [0.030]
  - PIT rate:
    - PIT model only: 0.032*** [0.000]

- Model sample sizes and diagnostics (exact figures)
  - # of observations:
    - Gompertz (PH): 11,639,304
    - Exponential (PH/AFT): 11,639,304
    - Log-logistic (AFT): 11,639,304
    - PIT: 11,401,501
    - Mixed Effects: 11,639,304
  - # of firms:
    - Gompertz (PH): 2,380,710
    - Exponential (PH/AFT): 2,380,710
    - Log-logistic (AFT): 2,380,710
    - PIT: 2,297,867
    - Mixed Effects: 2,380,710
  - # of failures:
    - Gompertz (PH): 667,550
    - Exponential (PH/AFT): 667,550
    - Log-logistic (AFT): 667,550
    - PIT: 429,747
    - Mixed Effects: 667,550
  - Cluster: Firm (all models)
  - Wald test:
    - Gompertz (PH): 111,685
    - Exponential (PH/AFT): 170,362
    - Log-logistic (AFT): 335,391
    - PIT: 116,671
    - Mixed Effects: 253,034
  - Log-likelihood:
    - Gompertz (PH): -1,761,534
    - Exponential (PH/AFT): -1,770,586
    - Log-logistic (AFT): -1,721,012
    - PIT: -4,202,494
    - Mixed Effects: -4,055,095

- Note: Standard errors are displayed in brackets. *** p<0.01, ** p<0.05, * p<0.1. PH denotes the proportional hazards model; AFT denotes the accelerated failure time model.

*Source: Orbis; authors’ calculation.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019078.pdf_