## wpiea2021023-print-pdf — References: key findings on TCRs and MNC investment

## Source details

**Canonical URL:** [wpiea2021023-print-pdf — References: key findings on TCRs and MNC investment](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021023-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021023-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021023-print-pdf.pdf.json)

---

### I. INTRODUCTION — key findings on TCRs and MNC investment
- Intra-company borrowing and lending is a common channel for MNCs to shift profits between countries to minimize overall tax liability.
- Empirical evidence confirms internal capital structures of MNCs respond strongly to CIT rate differentials.
- Thin-capitalization rules (TCRs) limit international debt shifting by specifying a threshold (safe-haven ratio) for internal debt relative to equity beyond which interest expenses are no longer deductible.
- Main empirical findings reported:
  - At an average CIT rate of 27 percent, a TCR reduces investment of multinational affiliates by 20 percent, on average.
  - The negative investment effect of TCRs depends critically on the CIT rate in the country where the TCR is introduced and is stronger at higher CIT rates.
  - The negative effect increases non-linearly in the firm leverage ratio: affiliates with internal debt above the safe-haven ratio are most affected.
  - TCRs with a stricter safe-haven ratio have more pronounced effects in reducing MNC investment.
  - A one standard deviation increase in the TCRScore would increase the marginal effect of the CIT rate by about two thirds in countries with a TCR.
- Paper scope and sample:
  - Micro-level dataset containing information on MNC affiliates in 34 countries during 2006-2014.
  - Analysis includes MNCs headquartered in over 60 countries between 2006 and 2014.

### II. THIN-CAPITALIZATION RULES — definition, mechanisms, and measurement
- Mechanism:
  - Debt shifting: MNCs borrow in high-tax countries (interest deductible) financed by loans from parent/affiliates in low-tax countries (interest taxable), reducing global tax liability without changing overall group debt exposure.
  - TCRs are partial restrictions denying interest deductibility beyond a fixed level of debt or interest; they define a safe-haven ratio for internal debt/equity.
- Variation across countries:
  - TCRs differ in safe-haven ratios and in enforcement strictness; some trigger automatic disallowance, others apply discretion (e.g., comparison to debt ratios of unrelated firms).
  - Empirical evidence shows TCRs constrain internal leverage and reduce debt shifting; when targeted at total debt ratio they can reduce total group leverage.
- IMF FAD database coverage and variables:
  - Database provides information on three aspects of TCRs in 62 countries: (1) introduction year, (2) whether the TCR restricts related-party debt only or all debt, and (3) the safe-haven ratio.
  - Two empirical measures used in analysis:
    - A discrete indicator TCR_kt that equals one for all years following introduction of some TCR on internal debt in country k.
    - A tightness measure TCRScore_kt based on the safe-haven ratio σ_kt, defined as:
      - TCRScore_kt = σ_kt / (1 + σ_kt).
    - Interpretations:
      - If safe-haven ratio σ_kt = 0 (complete denial of interest deduction), then TCRScore_kt = 0.
      - If σ_kt is very large (limitation not binding for most firms), TCRScore_kt converges to 1.
      - Overall, higher TCRScore implies smaller scope of the TCR in limiting internal debt.
  - In the dataset, the value of the TCRScore ranges between 0.5 and 0.8 during the sample period of 2006-2014.

### III. Institutional changes during 2006–2014 — countries introducing related-party TCRs
- Seven countries in the dataset introduced TCRs on related-party debt during 2006-2014:
  - China (2008)
  - Czech Republic (2007)
  - Finland (2013, with effective implementation in 2014)
  - France (2007)
  - Greece (2010)
  - Norway (2014)
  - Sri Lanka (2006)

### Theoretical considerations and empirical specification
- Two channels through which TCRs affect MNC affiliate investment:
  - Denial of interest deductibility raises cost of capital for affiliates with debt above the safe-haven ratio → ceteris paribus reduces investment.
  - MNCs may replace debt with equity to avoid international double taxation; replacement can reduce overall tax burden but raises cost of capital → reduces investment.
- Expected interaction: the higher the statutory corporate income tax (CIT) rate, the larger the investment-reducing effect of a TCR.
- Hypotheses:
  - (a) Hypothesis 1. TCRs reduce investment in affiliates of MNCs; this effect rises in the CIT rate and in the strictness of the TCR.
  - (b) Hypothesis 2. The negative investment effect of TCRs increases in the leverage ratio of MNC affiliates.
- Main empirical specification (eq. (2)):
  - Dependent variable: Investmentikt = Iikt / Kik,t−1 (current-year investment in fixed tangible assets divided by lagged capital stock in tangible assets).
  - Key regressors: CITkt (statutory CIT rate), TCRkt (dummy = 1 for years after TCR introduction), CITkt × TCRkt interaction.
  - Expected signs: β1 (CIT) negative; β3 (CIT × TCR) negative.
- Identification and controls:
  - Firm fixed effects ai and year fixed effects dt.
  - Firm-level controls Xikt: log(Sales), cash flow rate, profitability (EBIT/sales), sales growth rate, leverage ratio (total debt to capital), all ratio variables winsorized at top and bottom 1 percentile.
  - Country-level controls Zkt: GDP per capita, GDP growth rate, population, unemployment rate.
  - Preferred specification includes industry-year and country-industry fixed effects.

### Data
- Unbalanced panel: 397,216 observations for 75,154 unique companies in 34 countries for years 2006 to 2014.
- Sample constructed from unconsolidated financial statements of non-financial, non-utility affiliates in ORBIS (Bureau van Dijk).
- MNC affiliate definition: ultimate parent in a different country and owns at least 50% of shares.
- Macroeconomic variables: IMF World Economic Outlook database.
- Tax rates: Oxford University CBT Database.
- Key descriptive statistics (selection from Table 1):
  - Investment spending (in log): Mean 12.33, Std Dev 2.54, Median 12.43, P10 9.08, P90 15.43.
  - Fixed tangible assets (in log): Mean 13.84, Std Dev 2.56, Median 13.97, P10 10.45, P90 16.99.
  - Investment rate (Invt/Kt−1): Mean 0.53, Std Dev 1.29, Median 0.16, P10 -0.05, P90 1.22.
  - Leverage Ratio: Mean 0.65, Std Dev 0.30, Median 0.64, P10 0.29, P90 0.94.
  - CIT rate: Mean 0.27, Std Dev 0.05, Median 0.28, P10 0.19, P90 0.33.

### Empirical results — Average effects of TCRs
- Main findings:
  - Estimated coefficient on CIT rate is negative in columns (1)-(3).
  - Interaction CITkt × TCRkt is negative and statistically significant when controlling for country-level characteristics (columns (2) and (3)), supporting Hypothesis 1.
  - Standalone TCR dummy shows no significant effect in countries with zero CIT.
- Quantitative magnitudes:
  - At an average CIT rate of 27 percent, a TCR reduces the investment rate by 0.11.
  - Evaluated at the mean asset level of $21.3 mil, this corresponds to a reduction in investment by 20.8 percent.
  - At a CIT rate of 40 percent, MNC investment would fall by 37.7 percent.
- Cost-of-capital comparison:
  - Calculated increase in cost of capital for affected firm: from 0.026 to 0.0435 → increase of 67 percent.
  - With investment elasticities ranging between -0.25 and -1, affected firms’ investment would fall between 16 and 67 percent.
  - Average effect of 20.8 percent consistent with a relatively high elasticity in the literature.
- CIT marginal effects:
  - From column (3), a one-percentage point increase in the CIT rate reduces the investment rate by 0.00692 in countries with no TCR → evaluated at mean asset level implies a 1.2 percent reduction in MNC investment.
  - In countries with a TCR, a one-percentage point increase in the CIT rate reduces the investment rate by 0.014 (=0.00692+0.00686) → equivalent to a 2.4 percent reduction in MNC investment.
- Robustness:
  - Results quantitatively similar when controlling for transfer-pricing regulations and when using log investment as dependent variable.

### Empirical results — Heterogeneous (differential) effects
- Leverage heterogeneity (Table 3, quartiles based on pre-TCR leverage for implementing countries):
  - TCR negative investment effect is more pronounced for highly-levered firms.
  - No significant effect in bottom quartile; coefficient magnitude increases (non-monotonically) across higher quartiles.
  - Semi-elasticities of MNC investment with respect to statutory CIT rate in countries with TCRs:
    - Quartile 2: 2.5
    - Quartile 3: 3.1
    - Quartile 4: 2.3
  - Evaluated at sample mean CIT rate of 27 percent, an introduction of TCR would reduce MNC investment by 0.6, 12.2, 21,7 and 9.8 percent in the 1-4 quartile of leverage ratio, respectively.
- Scope/strictness heterogeneity (TCRScore and safe-haven ratio):
  - TCRScore decreases in strictness of the TCR; interaction TCRScorekt × CITkt has a positive and significant coefficient at the 5 percent level.
  - Combined with main TCR × CIT interaction (negative and significant at 1 percent), results imply: for one standard deviation increase in TCRScore (0.17, roughly corresponding to an increase in the safe-haven ratio of 1:1 to 2:1), the marginal effect of the CIT rate increases by about two thirds in the presence of a TCR.

### Conclusions and policy implications
- Core conclusions:
  - Introduction of a TCR has a negative effect on MNC investment.
  - The negative effect rises in the statutory CIT rate.
  - The negative effect rises with the degree of affiliate leverage and with the strictness of the TCR safe-haven ratio.
- Policy implications:
  - Negative investment effects from TCRs can make governments more reluctant to introduce TCRs unilaterally.
  - Governments may prefer more lenient TCR regulations to mitigate adverse effects on investment.

*Italic: Content derived from wpiea2021023-print-pdf (source PDF provided).*

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

### wpiea2021023-print-pdf - References .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .  .

### I. INTRODUCTION — key findings on TCRs and MNC investment
- Intra-company borrowing and lending is a common channel for MNCs to shift profits between countries to minimize overall tax liability.
- Empirical evidence confirms internal capital structures of MNCs respond strongly to CIT rate differentials (citations in source).
- Thin-capitalization rules (TCRs) are implemented to limit international debt shifting by specifying a threshold (safe-haven ratio) for internal debt relative to equity beyond which interest expenses are no longer deductible.
- Main empirical findings reported:
  - At an average CIT rate of 27 percent, a TCR reduces investment of multinational affiliates by 20 percent, on average.
  - The negative investment effect of TCRs depends critically on the CIT rate in the country where the TCR is introduced and is stronger at higher CIT rates.
  - The negative effect increases non-linearly in the firm leverage ratio: affiliates with internal debt above the safe-haven ratio are most affected.
  - TCRs with a stricter safe-haven ratio have more pronounced effects in reducing MNC investment.
  - A one standard deviation increase in the TCRScore would increase the marginal effect of the CIT rate by about two thirds in countries with a TCR.
- Paper scope and sample:
  - Micro-level dataset containing information on MNC affiliates in 34 countries during 2006-2014.
  - Analysis expands on prior work by including MNCs headquartered in over 60 countries between 2006 and 2014 (period of elevated efforts to tackle profit shifting).

### II. THIN-CAPITALIZATION RULES — definition, mechanisms, and measurement
- Mechanism:
  - Debt shifting: MNCs borrow in high-tax countries (interest deductible) financed by loans from parent/affiliates in low-tax countries (interest taxable), reducing global tax liability without changing overall group debt exposure.
  - TCRs are partial restrictions denying interest deductibility beyond a fixed level of debt or interest; they define a safe-haven ratio for internal debt/equity.
- Variation across countries:
  - TCRs differ in safe-haven ratios and in enforcement strictness; some trigger automatic disallowance, others apply discretion (e.g., comparison to debt ratios of unrelated firms).
  - Empirical evidence shows TCRs constrain internal leverage and reduce debt shifting; when targeted at total debt ratio they can reduce total group leverage.
- IMF FAD database coverage and variables:
  - Database provides information on three aspects of TCRs in 62 countries: (1) introduction year, (2) whether the TCR restricts related-party debt only or all debt, and (3) the safe-haven ratio.
  - Two empirical measures used in analysis:
    - A discrete indicator TCR_kt that equals one for all years following introduction of some TCR on internal debt in country k.
    - A tightness measure TCRScore_kt based on the safe-haven ratio σ_kt, defined as:
      - TCRScore_kt = σ_kt / (1 + σ_kt).
    - Interpretations:
      - If safe-haven ratio σ_kt = 0 (complete denial of interest deduction), then TCRScore_kt = 0.
      - If σ_kt is very large (limitation not binding for most firms), TCRScore_kt converges to 1.
      - Overall, higher TCRScore implies smaller scope of the TCR in limiting internal debt.
  - In the dataset, the value of the TCRScore ranges between 0.5 and

### III. Institutional changes during 2006–2014 — countries introducing related-party TCRs
- Seven countries in the dataset introduced TCRs on related-party debt during 2006-2014:
  - China (year of introduction: 2008)
  - Czech Republic (2007)
  - Finland (2013, with effective implementation in 2014)
  - France (2007)
  - Greece (2010)
  - Norway (2014)
  - Sri Lanka (2006)

*Italic: Content derived from wpiea2021023-print-pdf (source PDF provided).*

### 0.8 during the sample period of 2006-2014.

### wpiea2021023-print-pdf - 0.8 during the sample period of 2006-2014.

### Theoretical considerations and empirical specification
- Two channels through which thin capitalization rules (TCRs) affect multinational corporation (MNC) affiliate investment:
  - Denial of interest deductibility raises cost of capital for affiliates with debt above the safe-haven ratio → ceteris paribus reduces investment.
  - MNCs may replace debt with equity to avoid international double taxation; replacement can reduce overall tax burden but raises cost of capital → reduces investment.
- Expected interaction: the higher the statutory corporate income tax (CIT) rate, the larger the investment-reducing effect of a TCR.
- Hypotheses:
  - (a) Hypothesis 1. TCRs reduce investment in affiliates of MNCs; this effect rises in the CIT rate and in the strictness of the TCR.
  - (b) Hypothesis 2. The negative investment effect of TCRs increases in the leverage ratio of MNC affiliates.
- Main empirical specification (eq. (2)):
  - Dependent variable: Investmentikt = Iikt / Kik,t−1 (current-year investment in fixed tangible assets divided by lagged capital stock in tangible assets).
  - Key regressors: CITkt (statutory CIT rate), TCRkt (dummy = 1 for years after TCR introduction), CITkt × TCRkt interaction.
  - Expected signs: β1 (CIT) negative; β3 (CIT × TCR) negative.
- Identification and controls:
  - Firm fixed effects ai and year fixed effects dt.
  - Firm-level controls Xikt: log(Sales), cash flow rate, profitability (EBIT/sales), sales growth rate, leverage ratio (total debt to capital), all ratio variables winsorized at top and bottom 1 percentile.
  - Country-level controls Zkt: GDP per capita, GDP growth rate, population, unemployment rate.
  - Preferred specification includes industry-year and country-industry fixed effects.

### Data
- Unbalanced panel: 397,216 observations for 75,154 unique companies in 34 countries for years 2006 to 2014.
- Sample constructed from unconsolidated financial statements of non-financial, non-utility affiliates in ORBIS (Bureau van Dijk).
- MNC affiliate definition: ultimate parent in a different country and owns at least 50% of shares.
- Macroeconomic variables: IMF World Economic Outlook database.
- Tax rates: Oxford University CBT Database.
- Key descriptive statistics (selection from Table 1):
  - Investment spending (in log): Mean 12.33, Std Dev 2.54, Median 12.43, P10 9.08, P90 15.43.
  - Fixed tangible assets (in log): Mean 13.84, Std Dev 2.56, Median 13.97, P10 10.45, P90 16.99.
  - Investment rate (Invt/Kt−1): Mean 0.53, Std Dev 1.29, Median 0.16, P10 -0.05, P90 1.22.
  - Leverage Ratio: Mean 0.65, Std Dev 0.30, Median 0.64, P10 0.29, P90 0.94.
  - CIT rate: Mean 0.27, Std Dev 0.05, Median 0.28, P10 0.19, P90 0.33.

### Empirical results — Average effects of TCRs
- Main findings (Table 2, dependent variable: investment per $ fixed asset; column references preserved in source):
  - Estimated coefficient on CIT rate is negative in columns (1)-(3).
  - Interaction CITkt × TCRkt is negative and statistically significant when controlling for country-level characteristics (columns (2) and (3)), supporting Hypothesis 1.
  - Standalone TCR dummy shows no significant effect in countries with zero CIT.
- Quantitative magnitudes:
  - At an average CIT rate of 27 percent, a TCR reduces the investment rate by 0.11.
  - Evaluated at the mean asset level of $21.3 mil, this corresponds to a reduction in investment by 20.8 percent.
  - At a CIT rate of 40 percent, MNC investment would fall by 37.7 percent.
- Cost-of-capital comparison:
  - Calculated increase in cost of capital for affected firm: from 0.026 to 0.0435 → increase of 67 percent.
  - With investment elasticities ranging between -0.25 and -1, affected firms’ investment would fall between 16 and 67 percent.
  - Average effect of 20.8 percent consistent with a relatively high elasticity in the literature.
- CIT marginal effects:
  - From column (3), a one-percentage point increase in the CIT rate reduces the investment rate by 0.00692 in countries with no TCR → evaluated at mean asset level implies a 1.2 percent reduction in MNC investment.
  - In countries with a TCR, a one-percentage point increase in the CIT rate reduces the investment rate by 0.014 (=0.00692+0.00686) → equivalent to a 2.4 percent reduction in MNC investment.
- Robustness:
  - Results quantitatively similar when controlling for transfer-pricing regulations and when using log investment as dependent variable.

### Empirical results — Heterogeneous (differential) effects
- Leverage heterogeneity (Table 3, quartiles based on pre-TCR leverage for implementing countries):
  - TCR negative investment effect is more pronounced for highly-levered firms.
  - No significant effect in bottom quartile; coefficient magnitude increases (non-monotonically) across higher quartiles.
  - Semi-elasticities of MNC investment with respect to statutory CIT rate in countries with TCRs:
    - Quartile 2: 2.5
    - Quartile 3: 3.1
    - Quartile 4: 2.3
  - Evaluated at sample mean CIT rate of 27 percent, an introduction of TCR would reduce MNC investment by 0.6, 12.2, 21,7 and 9.8 percent in the 1-4 quartile of leverage ratio, respectively.
- Scope/strictness heterogeneity (TCRScore and safe-haven ratio):
  - TCRScore decreases in strictness of the TCR; interaction TCRScorekt × CITkt has a positive and significant coefficient at the 5 percent level.
  - Combined with main TCR × CIT interaction (negative and significant at 1 percent), results imply: for one standard deviation increase in TCRScore (0.17, roughly corresponding to an increase in the safe-haven ratio of 1:1 to 2:1), the marginal effect of the CIT rate increases by about two thirds in the presence of a TCR.

### Conclusions and policy implications
- Core conclusions:
  - Introduction of a TCR has a negative effect on MNC investment.
  - The negative effect rises in the statutory CIT rate.
  - The negative effect rises with the degree of affiliate leverage and with the strictness of the TCR safe-haven ratio.
- Policy implications:
  - Negative investment effects from TCRs can make governments more reluctant to introduce TCRs unilaterally.
  - Governments may prefer more lenient TCR regulations to mitigate adverse effects on investment.

*Source: Excerpt from IMF working paper (wpiea2021023-print-pdf) — empirical analysis and results for 2006-2014 sample.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021023-print-pdf.pdf_