## Keeping It Simple–Efficiency Costs of Fixed Margin Regimes in Transfer Pricing (Introduction and Annexes)

## Source details

**Canonical URL:** [Keeping It Simple–Efficiency Costs of Fixed Margin Regimes in Transfer Pricing (Introduction and Annexes)](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022193-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022193-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022193-print-pdf.pdf.json)

---

### Introduction: purpose and scope
- Fundamental international tax reform initiatives are underway at global, regional, and national levels, including agreement on the introduction of a global minimum tax and allocation of some residual profits to market countries.
- For the foreseeable future, allocation of a large share of profits from multinational enterprises (MNEs) will continue to be determined by internal transfer prices in line with the arm’s length principle (ALP).
- Simplification of the ALP is interpreted as removing the requirement for case-specific analysis (for example, deemed pricing approaches, safe harbors, or fixed margin regimes), and can be seen as a form of presumptive taxation because the tax base is determined using indirect means.
- Simplified approaches use less information to reduce administrative and compliance costs but provide a less accurate approximation of the tax base, with detrimental effects on vertical and horizontal equity and potential impacts on sectoral production efficiency.

### Conceptual model: how presumptive rules affect allocation and efficiency
- Premise: In a pure profit-based taxation system (abstracting from profit shifting), a subsidiary’s idiosyncratic productivity dictates its demand for input factors, implying marginal products within an MNE group are equalized in equilibrium; this allocation is optimal.
- Effect of presumptive rules:
  - Using proxies (for example, percentage of assets as a proxy for taxable profit) induces subsidiary-specific variation in marginal effective tax rates.
  - A presumptive system imposes a higher marginal effective tax rate on less productive firms, reducing their demand for labor and capital relative to a profit-based system.
  - Result: transfer pricing simplification can alter the allocation of resources within and across MNE groups and independent competitors, reducing aggregate production.
- Key dependence: the extent of efficiency losses depends on the predictability of profits; where proxies (fixed assets, employees, or combinations of input factors) accurately predict true profits, presumptive systems will not reduce efficiency.
- Theoretical expressions and implications:
  - Presumptive tax base: E[πi | Xi] = πi + εi, with εi uncorrelated with the presumed base.
  - Net-of-tax profit under presumptive taxation: πi − E[πi | Xi] τi = Si (1 − δi τi) − r Ki (1 − τi), where δi = E[Ai | X]/Ai.
  - First-order condition for optimal capital allocation: ∂Si/∂Ki = r (1−τi) / (1−τi δi).
  - Presumptive taxation leads to misallocation of capital across subsidiaries and reduces aggregate factor productivity relative to profit-based taxation.
  - Efficiency losses increase with returns-to-scale parameter α (or with elasticity of substitution σ) and are quadratic in the tax rate τ.

### Empirical approach and data
- Data: firm-level balance sheets and profit and loss accounts of around 300,000 firms located in 22 countries and operating in 53 sub-sectors.
- Objective: assess (un)predictability of profits across sectors and quantify efficiency effects from transfer pricing simplification.
- Predictive models: incorporate flexible firm-specific and country-level predictors to approximate firm-specific tax bases under alternative scenarios.
- Two information scenarios in predictive regressions:
  - Full information: Xi includes fixed assets, total assets, cost of goods sold, number of employees, GDP, GDP per capita, GDP growth, and inflation.
  - Limited information: Xi includes fixed assets only (used for manufacturing subsector analysis).
- Sample details:
  - ORBIS-based sample of independent companies with positive profits, sales, fixed assets; drop entities with intangible assets > 10% of fixed assets and outliers relative to MNE size.
  - Baseline sample: about 343,000 entities observed 2011–2019.
  - Manufacturing subset: 66,000 independent entities in 22 NACE 2-digit sectors across 21 countries.
- Predictive regression summary (from Table 2):
  - Average observations: 32,968 (All industries, full info); 17,454 (Manufacturing, full info); 17,454 (Manufacturing, limited info).
  - Average Adj. R2: 0.748 (All industries, full info); 0.773 (Manufacturing, full info); 0.63 (Manufacturing, limited info).
  - Observations: predicted profit slightly above recorded profit on average (predicted ≈ 500,000 USD; recorded ≈ 460,000 USD).

### Findings on profit predictability and drivers of residual variation
- Predictability patterns:
  - Variation in unpredicted (residual) profit is lowest among entities engaged in trade, basic manufacturing, and some services.
  - Variation is highest among entities with head office functions and technical advisory services.
- Residual variation drivers:
  - Higher residual variation where firms in a sector enjoy, on average, elevated market power.
  - Residual variation is higher in low-income countries; possible explanations include larger productivity differentials, less competition, slower propagation of technologies, measurement problems, less stringent financial reporting, or weaker administrative capacity and enforcement.
- Regression results (N = 3,950; Table 4 highlights):
  - log(GdpPc) coefficients: Model 1 −0.330***, Model 2 −0.493***, Model 3 −0.457***, Model 4 −0.294***.
  - log(Predicted profit margin) coefficients: ~1.490*** (Model 2), 1.363*** (Model 3), 1.285*** (Model 4).
  - log(Mark-up) coefficients: 1.138*** (Model 3) and 0.989*** (Model 4).
  - Adjusted R2 increases from 0.055 (Model 1) to 0.452 (Model 4 with industry FE).

### Quantifying efficiency costs: scenarios and results
- Two scenarios inspired by current tax practices:
  - Full information scenario: akin to benchmarking analysis, using a broad range of firm-specific and country-level variables to approximate firm-specific tax bases.
  - Limited information scenario: only fixed assets used as a firm-level predictor for manufacturing firms (replicating fixed margins).
- Measured efficiency losses:
  - In the full information scenario, efficiency losses already range between 0.25 and 2.2 percent of current productivity.
  - In manufacturing, simplification more than doubles sectoral inefficiency on average — from 0.4 percent in the full information scenario to over 1 percent in the simplified (fixed-asset) scenario.
  - Sectoral differences are large; for example, sectoral efficiency losses are smallest in manufacturing of basic metals.
- Firm-specific tax wedge distribution (Table 3):
  - Baseline sample, full information: Min. −0.468; Median 0.036; Mean 0.012; Max. 0.377; Variance 0.035.
  - Manufacturing sample, full information: Min. −0.431; Median 0.039; Mean 0.009; Max. 0.291; Variance 0.028.
  - Manufacturing sample, limited information: Min. −0.380; Median 0.090; Mean 0.202; Max. 1.970; Variance 0.35.
- Illustrative calculation:
  - With a tax rate of 25 percent and elasticity of substitution σ = 3, approximate efficiency losses could amount to 0.3 percent (0.003 = 3*025^2/2*0.035) under the full information scenario in the baseline sample.
- Direct estimates (Figure 2 summary):
  - Under full information scenario (benchmarking akin to ALP practice), estimated reduction in total factor productivity ranges between 0.3 percent and 2.3 percent across sectors (assuming σ = 3 and average tax rate τ = 0.23).
  - Losses smallest for trading activities; larger losses observed in legal services, R&D, management and consulting—sectors where benchmarking is less common and simplification is not advisable.
- Incremental costs of simplification (manufacturing subsectors, Figure 3 summary):
  - Using fixed assets alone (limited information) raises efficiency losses relative to full information benchmarking in many manufacturing subsectors.
  - Differences between full- and limited-information scenarios approximate incremental efficiency costs of fixed margin simplification; some subsectors (e.g., apparel) show notably larger costs when fixed assets are a poorer profit proxy.
  - Allowing mark-ups to vary across subsectors increases estimated efficiency losses, suggesting σ = 3 may overestimate market power for manufacturers in the sample.

### Compliance burdens, simplification options, and examples
- Transfer pricing-related compliance cost drivers (documentation, clearances and rulings, mutual agreement procedures) account for about 60 percent of all compliance costs (EU, 2011), equivalent to 0.14 percent and 0.33 of turnover.
- In an average OECD country, transfer pricing compliance is estimated at between 3.9 percent and 12.7 percent of corporate tax collection, amounting to 0.1 to 0.4 percent of GDP.
- Costs and adverse effects may be more pronounced in developing countries, where inconsistent application of the ALP can result in significant base erosion and revenue losses, undesirable litigation, and greater taxpayer uncertainty.
- Simplification mechanisms:
  - Relaxing the degree of comparability aimed for.
  - Aggregating and standardizing analytical steps for specific sectors.
  - Prescribing expected returns under tax legislation for transactions/entities in scope (deemed pricing, opt-in/opt-out safe harbors, fixed margin regimes).
- Country examples of simplified approaches:
  - Cameroon: central purchasing fees deductible up to 5% of the cost of goods purchased centrally.
  - Côte d’Ivoire: interest expense cap at two basis points above the central bank’s lending rate.
  - Brazil: allows only CUP, resale price and cost-plus methods with predetermined margins; noted vulnerabilities and ongoing alignment work with OECD.
  - Dominican Republic: APAs with predetermined gross (net) margins and a minimum effective tax rate of 2% for accommodation providers.
  - Mexico: safe harbor for foreign-owned manufacturers with deemed return on assets or operating expenses; mark-up differentiated since 2016; opt-out APA abolished in 2021.
  - India: opt-in safe harbor (2013, revised 2017) for specified activities; limited take-up due to rates seen as too high.
- Observers’ and business proposals:
  - PCT (2017) argues (rebuttable) safe harbors appropriate for manufacturing, sales, distribution, and many service providers; suggests using administrative data to design broad/regional safe harbors for developing economies.
  - Picciotto (2018) suggests focusing on sectors of elevated economic importance and opt-out safe harbors placing burden of proof on taxpayers.
  - Hofman and Riedel (2018) argue presumptive outcomes above the “true” arm’s length result acceptable if additional tax cost does not exceed firms’ compliance cost savings.
  - Business proposals suggest formularized overlays referencing an MNE group’s overall performance.

### Risks, mitigation, and design considerations
- Risks and limitations:
  - Mechanical aggregate approaches do not eliminate abuse risks via aggressive tax planning.
  - May lead to economic double taxation when adopted unilaterally.
  - Can generate efficiency costs from strategic responses to qualify.
  - Meaningful deviations from OECD TP Guidelines (e.g., disallowing taxpayers’ right to rebut profit margins) risk international scrutiny (Brazil example).
- Mitigation examples:
  - Mexico negotiated agreements between jurisdictions to avoid double taxation via MAP; India pursued similar approaches.
  - Small developing countries can implement well-researched sectoral benchmarks successfully (Dominican Republic tourism example).
- Design considerations:
  - Focus simplified regimes on sectors where observable inputs predict profitability well (low residual variation) and where activities are routine (limited intangible contributions).
  - Ensure scope, margin levels, and opt-in/opt-out rules reflect expected benefits (deemed profits, reduced documentation, lower litigation and double taxation risk) to encourage taxpayer participation.
  - Monitor for strategic responses and potential tax planning; consider transitional or negotiated agreements to address international allocation issues.
  - Sector- and country-heterogeneity matters: standardized fixed margins across diverse markets may be problematic; sector- and region-specific benchmarks or administrative-data-based regional safe harbors may be preferable.

### Methodological notes and annex findings
- Annex I — Descriptive statistics (values in thousands of USD):
  - Baseline sample, N=1846243:
    - Profit: Min. 0, Median 58, Mean 461, Max. 5183, Variance 1086286.
    - Predicted profit: Min. 1, Median 96, Mean 499, Max. 4911, Variance 1011934.
  - Manufacturing sample, N=384009:
    - Profit: Min. 1, Median 125, Mean 836, Max. 8588, Variance 3086555.
    - Predicted profit: Min. 2, Median 208, Mean 889, Max. 7843, Variance 2667236.
    - Predicted profit (limited info): Min. 3, Median 297, Mean 890, Max. 6420, Variance 1946092.
- Annex II — Estimation of firm-specific mark-ups:
  - Firm-specific mark-ups computed following De Loecker and others (2020).
  - Formula provided: μ_i = θ^L_s * S_i / w^L_i (A.1) (symbols and definitions as in source).
  - Definitions and implementation:
    - θ^L_s is the elasticity of output with respect to intermediate inputs in sector s.
    - S_i are sales.
    - w^L_i are intermediate inputs; "cost of goods sold" used as the measure for intermediate inputs at the firm level.
    - Industry-specific elasticities θ^L_s are obtained at the 2-digit NACE level using panel regression methods by regressing the logarithm of turnover on firm-specific variables, including cost of goods sold, and firm- and time-specific fixed effects.

### Conclusion: trade-offs and policy guidance
- Assessing desirability of tax simplification options requires contrasting associated costs with potential gains.
- There are large asymmetries across sectors in the reliability of different profit predictions.
- Where efficiency losses from predicting profit are comparatively large, benchmarking to determine arm’s length pricing is less common practice and simplification to avoid case-by-case assessments is not advisable.
- Efficiency costs are moderate in several sectors; this supports exploring widening simplified approaches for those sectors but does not suffice alone to determine application.
- Simplified approaches can increase welfare if reductions in compliance and administrative costs and other benefits outweigh resource misallocation.
- Country experiences suggest benefits likely greater for developing economies, including increased taxpayer certainty, lower compliance and administrative cost, and reduced scope for tax avoidance.
- An incremental, activity-targeted approach at country or regional level—initially targeting very specific activities—is recommended given sectoral heterogeneity.

*Keeping It Simple–Efficiency Costs of Fixed Margin Regimes in Transfer Pricing, Annex 2 and Annexes I–II (from Working Paper No. WP/2022/193)*

### Introduction ...........................................................................................................

### wpiea2022193-print-pdf - Introduction ...........................................................................................................

### Introduction: purpose and scope
- Fundamental international tax reform initiatives are underway at global, regional, and national levels, including agreement on the introduction of a global minimum tax and allocation of some residual profits to market countries.
- For the foreseeable future, allocation of a large share of profits from multinational enterprises (MNEs) will continue to be determined by internal transfer prices in line with the arm’s length principle (ALP).
- Simplification of the ALP is interpreted as removing the requirement for case-specific analysis (for example, deemed pricing approaches, safe harbors, or fixed margin regimes), and can be seen as a form of presumptive taxation because the tax base is determined using indirect means.
- Simplified approaches use less information to reduce administrative and compliance costs but provide a less accurate approximation of the tax base, with detrimental effects on vertical and horizontal equity and potential impacts on sectoral production efficiency.

### Conceptual model: how presumptive rules affect allocation and efficiency
- Premise: In a pure profit-based taxation system (abstracting from profit shifting), a subsidiary’s idiosyncratic productivity dictates its demand for input factors, implying marginal products within an MNE group are equalized in equilibrium; this allocation is optimal.
- Effect of presumptive rules:
  - Using proxies (for example, percentage of assets as a proxy for taxable profit) induces subsidiary-specific variation in marginal effective tax rates.
  - A presumptive system imposes a higher marginal effective tax rate on less productive firms, reducing their demand for labor and capital relative to a profit-based system.
  - Result: transfer pricing simplification can alter the allocation of resources within and across MNE groups and independent competitors, reducing aggregate production.
- Key dependence: the extent of efficiency losses depends on the predictability of profits; where proxies (fixed assets, employees, or combinations of input factors) accurately predict true profits, presumptive systems will not reduce efficiency.

### Empirical approach and data
- Data: firm-level balance sheets and profit and loss accounts of around 300,000 firms located in 22 countries and operating in 53 sub-sectors.
- Objective: assess (un)predictability of profits across sectors and quantify efficiency effects from transfer pricing simplification.
- Predictive models: incorporate flexible firm-specific and country-level predictors to approximate firm-specific tax bases under alternative scenarios.
- Findings on profit predictability:
  - Variation in unpredicted (residual) profit is lowest among entities engaged in trade, basic manufacturing, and some services.
  - Variation is highest among entities with head office functions and technical advisory services.
- Drivers of residual variation:
  - Higher residual variation where firms in a sector enjoy, on average, elevated market power.
  - Residual variation is higher in low-income countries, potentially reflecting larger productivity differentials, less competition, slower propagation of technologies, measurement problems, less stringent financial reporting, or weaker administrative capacity and enforcement.

### Quantifying efficiency costs: scenarios and results
- Two scenarios inspired by current tax practices:
  - Full information scenario: akin to benchmarking analysis, using a broad range of firm-specific and country-level variables to approximate firm-specific tax bases.
  - Limited information scenario: only fixed assets used as a firm-level predictor for manufacturing firms (replicating fixed margins).
- Measured efficiency losses:
  - In the full information scenario, efficiency losses already range between 0.25 and 2.2 percent of current productivity.
  - In manufacturing, simplification more than doubles sectoral inefficiency on average — from 0.4 percent in the full information scenario to over 1 percent in the simplified (fixed-asset) scenario.
  - Sectoral differences are large; for example, sectoral efficiency losses are smallest in manufacturing of basic metals.

### Contributions and policy-relevant implications
- Empirical implications:
  - Moderate efficiency costs of simplification are found in a wider range of sectors than currently under consideration, including some basic manufacturing, suggesting some scope for broader application of simplified approaches.
  - Variation in profit margins within a sector or country matters as much as average values for production efficiency and revenue.
  - Costs of tax simplification depend on market factors, limiting scope for standardized cross-country solutions to determine specific returns.
  - Residual variation of profits is more pronounced in lower income economies, though its efficiency impact may be partially mitigated by more concentrated market power in these economies.
- Relation to literature:
  - Estimated efficiency losses align with prior assessments of tax-induced efficiency losses from factor misallocation (Restuccia and Rogerson, 2008; Fajgelbaum and others, 2019).
  - Distinction: a presumptive tax system induces a negative correlation between effective tax rates and productivity, resulting in lower productivity losses than when variation in effective tax burden is uncorrelated with productivity.

*Source: IMF Working Paper — Introduction section of wpiea2022193-print-pdf*

### 3.9 percent and 12.7 percent of corporate tax collection, amounting to 0.1 to 0.4 percent of GDP in an average

### Keeping It Simple–Efficiency Costs of Fixed Margin Regimes in Transfer Pricing

### Compliance and Revenue Burdens
- Transfer pricing-related compliance cost drivers (documentation, clearances and rulings, mutual agreement procedures) account for about 60 percent of all compliance costs (EU, 2011), equivalent to 0.14 percent and 0.33 of turnover.  
- In an average OECD country, transfer pricing compliance is estimated at between 3.9 percent and 12.7 percent of corporate tax collection, amounting to 0.1 to 0.4 percent of GDP.  
- Costs and adverse effects may be more pronounced in developing countries, where inconsistent application of the arm’s length principle (ALP) can result in significant base erosion and revenue losses, undesirable litigation, and greater taxpayer uncertainty (IMF and OECD, 2017).

### Simplification Options and Examples
- Simplification can be achieved by:
  - Relaxing the degree of comparability aimed for.
  - Aggregating and standardizing analytical steps for specific sectors.
  - Prescribing expected returns under tax legislation for transactions/entities in scope of a simplification rule (deemed pricing, opt-in/opt-out safe harbors, fixed margin regimes).
- Practical examples of simplified approaches with predetermined sectoral benchmarks (from Table 1):
  - Cameroon: central purchasing fees deductible up to 5% of the cost of goods purchased centrally.
  - Côte d’Ivoire: interest expense cap at two basis points above the central bank’s lending rate.
  - Brazil: allows only CUP, resale price and cost-plus methods with predetermined margins; noted vulnerabilities and ongoing alignment work with OECD.
  - Dominican Republic: APAs with predetermined gross (net) margins and a minimum effective tax rate of 2% for accommodation providers.
  - Mexico: safe harbor for foreign-owned manufacturers with deemed return on assets or operating expenses; mark-up differentiated since 2016; opt-out APA abolished in 2021.
  - India: opt-in safe harbor (2013, revised 2017) for specified activities; limited take-up due to rates seen as too high.
- Observers’ proposals:
  - PCT (2017) argues (rebuttable) safe harbors appropriate for manufacturing, sales, distribution, and many service providers; suggests using administrative data to design broad/regional safe harbors for developing economies.
  - Picciotto (2018) suggests focusing on sectors of elevated economic importance and opt-out safe harbors placing burden of proof on taxpayers.
  - Hofman and Riedel (2018) argue presumptive outcomes above the “true” arm’s length result acceptable if additional tax cost does not exceed firms’ compliance cost savings.
  - Business proposals suggest formularized overlays referencing an MNE group’s overall performance.

### Risks and Limitations of Simplification
- Simplification is not a panacea:
  - Mechanical aggregate approaches do not eliminate abuse risks via aggressive tax planning.
  - May lead to economic double taxation when adopted unilaterally.
  - Can generate efficiency costs from strategic responses to qualify.
  - Meaningful deviations from OECD TP Guidelines (e.g., disallowing taxpayers’ right to rebut profit margins) risk international scrutiny (Brazil example).
- Mitigation examples:
  - Mexico negotiated agreements between jurisdictions to avoid double taxation via MAP; India pursued similar approaches.
  - Small developing countries can implement well-researched sectoral benchmarks successfully (Dominican Republic tourism example).

### Conceptual Model (Presumptive Taxation and Misallocation)
- Presumptive tax base represented as E[πi | Xi] = πi + εi, where εi is uncorrelated with presumed base.  
- Simplification modeled as reduction in information in Xi, leading to increased residual variation.  
- MNE production: Si = Ai Ki^α with profits πi = Si − r Ki.  
- Net-of-tax profit under presumptive taxation:
  - πi − E[πi | Xi] τi = Si (1 − δi τi) − r Ki (1 − τi), where δi = E[Ai | X]/Ai.  
- Presumptive taxation implies a firm-specific tax wedge: δi−1 = τi/ (τi δi) − τ (text defines wedge as 훿훿
푖푖
−1 = 휏휏
푖푖
훿훿
푖푖
−τ휏, described in narrative).  
- First-order condition for optimal capital allocation:
  - ∂Si/∂Ki = r (1−τi) / (1−τi δi) for i = 1,...,N.  
- Key theoretical insights:
  - Presumptive taxation leads to misallocation of capital across subsidiaries and reduces aggregate factor productivity relative to profit-based taxation.
  - Reduction in allocative efficiency is a function of variation in δi.
  - Efficiency losses increase with returns-to-scale parameter α (or with elasticity of substitution σ) and are quadratic in the tax rate τ.

### Empirical Strategy and Data
- Approach: treat benchmarking/transfer pricing practice as a form of presumptive taxation and quantify efficiency effects by measuring variation in unpredictable profit margins.  
- Two information scenarios for predictive regressions:
  - Full information: Xi includes fixed assets, total assets, cost of goods sold, number of employees, GDP, GDP per capita, GDP growth, and inflation.
  - Limited information: Xi includes fixed assets only (used for manufacturing subsector analysis).
- Sample:
  - ORBIS-based sample of independent companies with positive profits, sales, fixed assets; drop entities with intangible assets > 10% of fixed assets and outliers relative to MNE size.
  - Baseline sample: about 343,000 entities observed 2011–2019.
  - Manufacturing subset: 66,000 independent entities in 22 NACE 2-digit sectors across 21 countries.
- Predictive regressions (summary from Table 2):
  - Average observations: 32,968 (All industries, full info); 17,454 (Manufacturing, full info); 17,454 (Manufacturing, limited info).
  - Average Adj. R2: 0.748 (All industries, full info); 0.773 (Manufacturing, full info); 0.63 (Manufacturing, limited info).
  - Observations: predicted profit slightly above recorded profit on average (predicted ≈ 500,000 USD; recorded ≈ 460,000 USD).

### Quantitative Findings on Residual Variation and Efficiency Losses
- Firm-specific tax wedges (distribution from Table 3):
  - Baseline sample, full information: Min. −0.468; Median 0.036; Mean 0.012; Max. 0.377; Variance 0.035.
  - Manufacturing sample, full information: Min. −0.431; Median 0.039; Mean 0.009; Max. 0.291; Variance 0.028.
  - Manufacturing sample, limited information: Min. −0.380; Median 0.090; Mean 0.202; Max. 1.970; Variance 0.35.
- Approximate implications (illustrative calculation in text):
  - With a tax rate of 25 percent and elasticity of substitution σ = 3, approximate efficiency losses could amount to 0.3 percent (0.003 = 3*025^2/2*0.035) under the full information scenario in the baseline sample.
- Sectoral variation:
  - Variation in unpredictable profit margins differs considerably across sectors; highest variability among head office activities and lowest among postal and courier services (Figure 1 reference).
- Drivers of residual variation (Table 4 summary, N = 3,950 observations):
  - Log variation in residual profit margins negatively correlated with log(GdpPc) (GDP per capita) across models; coefficients: Model 1 −0.330***, Model 2 −0.493***, Model 3 −0.457***, Model 4 −0.294***.  
  - log(Predicted profit margin) positively associated with residual variation: coefficient ~1.490*** (Model 2), 1.363*** (Model 3), 1.285*** (Model 4).  
  - log(Mark-up) positive and significant in Model 3: 1.138*** (Model 3) and 0.989*** (Model 4).  
  - Adjusted R2 increases from 0.055 (Model 1) to 0.452 (Model 4 with industry FE), indicating substantial explanatory power from predicted margins, mark-ups, and industry fixed effects.
- Direct estimates of efficiency losses from benchmarking (Figure 2 summary):
  - Under full information scenario (benchmarking akin to ALP practice), estimated reduction in total factor productivity ranges between 0.3 percent and 2.3 percent across sectors (assuming σ = 3 and average tax rate τ = 0.23).
  - Losses smallest for trading activities; larger losses observed in legal services, R&D, management and consulting—sectors where benchmarking is less common and simplification is not advisable.
- Incremental costs of simplification (manufacturing subsectors, Figure 3 summary):
  - Using fixed assets alone (limited information) raises efficiency losses relative to full information benchmarking in many manufacturing subsectors.
  - The difference between full- and limited-information scenarios approximates incremental efficiency costs of fixed margin simplification; some subsectors (e.g., apparel) show notably larger costs when fixed assets are a poorer profit proxy.
  - Allowing mark-ups to vary across subsectors increases estimated efficiency losses, suggesting σ = 3 may overestimate market power for manufacturers in the sample.

### Policy Implications and Recommendations
- Potential gains from simplification:
  - Simplification (safe harbors, fixed margins, deemed pricing) can substantially reduce compliance and administrative costs, especially where benchmarking is costly or data-poor.
  - Returns to simplifying the ALP are likely larger in developing countries with limited public information and high compliance burdens.
- Trade-offs and safeguards:
  - Policymakers should weigh compliance cost savings against efficiency losses from misallocation of capital and risks of tax abuse and double taxation.
  - Optional (opt-in) or opt-out safe harbors with reversing burdens of proof (taxpayer shows rule inappropriate) may help balance risks and uptake (experience in India, Dominican Republic).
  - Coordination with major trading partners (e.g., treaty partners) and use of MAP agreements can mitigate double taxation and treaty conflict (Mexico and India examples).
  - Sectoral and country heterogeneity matters: standardized fixed margins across diverse markets may be problematic given differences in predicted profit variation and market power; sector- and region-specific benchmarks or administrative-data-based regional safe harbors may be preferable.
- Design considerations:
  - Focus simplified regimes on sectors where observable inputs predict profitability well (low residual variation) and where activities are routine (limited intangible contributions).
  - Ensure scope, margin levels, and opt-in/opt-out rules reflect expected benefits (deemed profits, reduced documentation, lower litigation and double taxation risk) to encourage taxpayer participation.
  - Monitor for strategic responses and potential tax planning; consider transitional or negotiated agreements to address international allocation issues.

*Source: IMF Working Paper text excerpt (Keeping It Simple–Efficiency Costs of Fixed Margin Regimes in Transfer Pricing).*

### Annex 2 for computation of mark-ups at the firm level.

### Annex 2 for computation of mark-ups at the firm level

### Conclusion: trade-offs from transfer-pricing simplification
- Assessing desirability of tax simplification options requires contrasting associated costs with potential gains.
- Analysis provides insight into the risk of creating distortions when presuming returns for multinational subsidiaries.
- There are large asymmetries across sectors in the reliability of different profit predictions.
- Where efficiency losses from predicting profit are comparatively large, benchmarking to determine arm’s length pricing is less common practice and simplification to avoid case-by-case assessments is not advisable.
- Efficiency costs are moderate in several sectors; this supports exploring widening simplified approaches for those sectors but does not suffice alone to determine application.
- Simplified approaches can increase welfare if reductions in compliance and administrative costs and other benefits outweigh resource misallocation.
- Country experiences suggest benefits likely greater for developing economies, including increased taxpayer certainty, lower compliance and administrative cost, and reduced scope for tax avoidance.
- Given heterogeneity across sectors, an incremental approach at country or regional level—initially targeting very specific activities—may be the most promising route for policy reform.

### Annex I — Descriptive statistics: profits and predicted profits
- Table A1: Descriptive Statistics: Profits and Predicted Profits – Full Sample (values in thousands of USD)
  - Baseline sample, N=1846243
    - Profit: Min. 0, Median 58, Mean 461, Max. 5183, Variance 1086286
    - 휋휋
푓푓푓푓푓푓푓푓: Min. 1, Median 96, Mean 499, Max. 4911, Variance 1011934
  - Manufacturing sample, N=384009
    - Profit: Min. 1, Median 125, Mean 836, Max. 8588, Variance 3086555
    - 휋휋
푓푓푓푓푓푓푓푓: Min. 2, Median 208, Mean 889, Max. 7843, Variance 2667236
    - 휋휋
푓푓푖푖푙푙 푖푖푖푖푙푙푙푙: Min. 3, Median 297, Mean 890, Max. 6420, Variance 1946092
- Notes: table provides distribution of profits and predicted profits in two subsamples, in thousands of USD.

### Annex II — Estimation of firm-specific mark-ups
- Firm-specific mark-ups computed following De Loecker and others (2020).
- Formula provided in the source:
  - 휇휇
푖푖푖푖
 as
  - 휇휇
푖푖푖푖푖푖
=휃휃
푖푖
퐿퐿
푆푆
푖푖
푖푖푖푖푖푖
푤푤퐿퐿
푖푖푖푖푖푖
  - (A.1)
- Definitions and implementation details from the source:
  - 휃휃
푖푖
퐿퐿 is the elasticity of output with respect to intermediate inputs in sector s.
  - 푆푆
푖푖 are sales.
  - 푤푤퐿퐿 are intermediate inputs.
  - Use accounting variable "cost of goods sold" as the measure for intermediate inputs at the firm level.
  - Industry-specific elasticities 휃휃
푖푖
퐿퐿 are obtained at the 2-digit NACE level using panel regression methods by regressing the logarithm of turnover on firm-specific variables, including cost of goods sold, and firm- and time-specific fixed effects.

### Key methodological and policy implications
- Simplification via fixed margin regimes entails efficiency costs that vary substantially across sectors.
- Empirical computation of firm-specific mark-ups relies on sectoral elasticities and firm-level cost of goods sold; accuracy of these inputs matters for assessing misallocation risk.
- Policymakers should weigh direct efficiency losses against gains from lower compliance and administrative costs, with particular attention to developing-economy contexts.
- An incremental, activity-targeted approach is recommended given sectoral heterogeneity.

_Keeping It Simple–Efficiency Costs of Fixed Margin Regimes in Transfer Pricing, Annex 2 and Annexes I–II (from Working Paper No. WP/2022/193)_

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022193-print-pdf.pdf_