## wpiea2025001-print-pdf

## Source details

**Canonical URL:** [wpiea2025001-print-pdf](https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025001-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2025/english/wpiea2025001-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2025/english/wpiea2025001-print-pdf.pdf.json)

---

### Methodology
- Estimation approaches:
  - Production Function Approach (PFA) following De Loecker & Warzynski (2012): markups recovered as μ_it = P_it / MC_it = ε_it · [S_it / R_it]^{-1}, where ε_it is the output elasticity of the variable input s, S_it is expenditure on input s, and R_it is sales/turnover.
  - Cost-Share Approach (CSA) assuming constant returns to scale: ε_it^s = E[ω_s · s_it] / E[∑_j ω_j · j_itj], yielding μ_it = R_it / E[∑ ω_j · j_itj] (equation 4).
- Identification and implementation details:
  - PFA auxiliary regression for revenue elasticity η_st: y_it = η_st^v v_ist + η_st^x x_ist + η_st^k k_ist + ω_ist + u_ist (equation 2).
  - ω_it modeled following Olley & Pakes (1996) as ω_it = φ_st(i_it, k_it, z_it) + u_ist (equation 3).
  - Baseline assumes η_st constant across sample (η_st = η_s) and estimates a unique η_s for each sector.
  - CSA relaxes need to identify output elasticity from quantity data but imposes constant returns to scale.
- Robustness and limitations noted:
  - Sensitivity to input-bundle choice (Raval 2023); revenue vs quantity data may cause downward bias (Klette & Griliches 1996); concerns on omitted price bias (Bond et al. 2021).
  - Sample-size constraints addressed by aggregating sectors, assuming time-constant production functions, and expanding to non-listed firms from BvD Orbis (with caution).

### Firm-Level and Macroeconomic Data
- Primary sample coverage:
  - Publicly listed firms from Compustat Global headquartered in one of 13 ME economies between 2000 and 2022; earlier references use 2004–2022 for some estimates.
  - Expanded sample includes non-listed firms from BvD Orbis for robustness.
- Data processing:
  - Deflated using country-level deflators from IMF WEO and converted to USD.
  - Observations with negative sales or cost of goods sold dropped; sales-to-cogs ratio winsorized at the 1 percent double-sided.
- Sample size after cleaning:
  - Approximately 1300 firms and 20,700 firm-year observations.
- Benchmarking and macro variables:
  - Constructed comparable US sample from Compustat with identical restrictions.
  - GDP, production by industry, inflation, and VAT rates sourced from IMF WEO.

### Main Results — Market Power in the Middle East
- Aggregate and cross-region findings:
  - Corporate market power among listed firms in the Middle East is higher than in the US.
  - GCC market power has not exhibited a secular upward trend over the last twenty years; overall ME markups are on a downward trend (except for the COVID-19 period).
  - A significant spike in market power in 2016 after the listing of Saudi Arabia’s ARAMCO.
  - Excluding ARAMCO, GCC markups were stable around 1.45 between 2010 and 2020; average markup in the GCC was close to 1.6 around 2010.
- Robustness across estimators:
  - PFA and CSA reveal similar dynamics for the ME region, with markups decreasing over time.
- Firm-level heterogeneity and superstar phenomenon:
  - Larger firms with higher sales exhibit higher market power across sectors (oil, mining & utilities and other sectors).
  - Oil, mining, and utilities sector has higher market power than other sectors across the sales distribution.
  - Resource-rich ME countries show higher average markups; Morocco is the only non-oil country with comparable markup levels.
- Country-level heterogeneity within GCC:
  - Average markups broadly stable overall; weak upward trends in Bahrain, Kuwait, and Oman in recent years; downward trends in Qatar, Saudi Arabia, and the UAE.
  - Possible drivers: structural reforms (e.g., Qatar competition law of 2006; Saudi competition law of 2019), changes in price elasticity, licensing/regulatory practices.
- Industry-level heterogeneity:
  - Construction sector among listed firms shows the lowest average markups (caveat: missing large construction firms may bias downward).
  - Oil, mining, and utilities have the highest average markups. Market services display higher markups than many other industries.

### Macroeconomic Implications and Inflation
- Inflation passthrough and "greedflation":
  - Across the GCC, firms absorbed some price rises by lowering markups following inflation surprises; the markup response to inflation surprises is negative (significance varies by country).
  - Authors state they “do not find evidence of ‘greedflation’ in the Middle East.”
  - Alignment with evidence reported for advanced economies in Chapter 1, October 2023 World Economic Outlook.
- Impulse response to inflation:
  - After a 1 percent inflation shock, firms in the GCC on average reduce their markups by 0.05 units relative to an average of 1.3 after two years of the shock.
  - Relationship negative for all GCC countries (significance varies); non-GCC ME firms do not adjust markups and tend to pass price rises to customers.

### Decomposition of Markup Evolution (Annex I)
- Decomposition components:
  - evolution due to net firm entry
  - evolution due to within-firm reduction in markups
  - evolution due to reallocation of economic activity
- Scenario outcomes (dashed lines start from 2004):
  - Dashed red line (only within-firm markups allowed to change): in both markup measures, red dashed line drives the reduction in markups.
  - Dashed grey line (relocation only): stagnates in PFA markups; rises in CSA markups. Implication: reallocation did not play a part in PFA markup evolution; in CSA, reallocation to firms with higher markups increased markups.
  - Net entry (black dashed line): broadly constant for PFA; fell similarly to within for CSA.
- Summary finding:
  - Declining markups are partly determined by within-firm changes in markups rather than entry alone; net firm entry plays an extremely limited role for one markup measure.

### VAT Reform and Market Power (GCC 2018–2022)
- Institutional background:
  - GCC unified VAT framework agreement year: 2017; mandates standard VAT rate at 5 percent; requires local laws to implement the treaty.
  - Implementation status through 2023: four out of six countries implemented the tax; Saudi Arabia and the United Arab Emirates implemented in 2018; Kuwait and Qatar yet to implement. Saudi Arabia hiked rates in 2020. Bahrain hiked rates in 2022.
  - Staggered implementation enables multiple staggered difference-in-difference strategies.
- Theoretical prediction:
  - Firm markup expression with consumption tax τ: μit := Pit / C′(Q(P(1+τ))) = [1 + 1/((1+τ)γ)]−1 where γ := (Qit / (P·Q)) p′.
  - Effect on logged firm markups: ∂ ln(μit)/∂τ = γ·[γ(1+τ)]−2 / {1 + [γ(1+τ)]−1}. Since γ < 0, numerator negative; for sufficiently elastic demand (γ·(1+τ) < −1) the tax reduces the wedge between output prices and marginal costs, implying VAT reduces markups.
- Empirical identification:
  - Staggered adoption diff-in-diff with parallel trends assumption (equation (8)); robustness: pre-trends tests and conditional parallel trends on matched sample.
  - Intermediate-final goods diff-in-diff: within-country comparison using intermediate-sector firms as controls (equation (9)).
- Empirical findings:
  - Staggered Diff-in-Diff: a 1 percentage point increase in the VAT rate is associated with a 5 percent reduction in markups. Effect is dynamic and not immediate.
  - No evidence of pre-reform common trends violation.
  - Limitation: cannot trace effects beyond 3 years after reform due to limited firm accounts after 2021.
- Robustness table (Annex IV) highlights:
  - Headline average impact (Column (1) of Annex IV table): an additional 1% point of VAT is an average reduction of 1% in markups estimated using PFA in the GCC sample.
  - Regression VAT coefficients and standard errors:
    - (1) VAT -0.0117** (0.00461)
    - (2) VAT -0.00798** (0.00352)
    - (3) VAT -0.00667*** (0.00247)
    - (4) VAT -0.00159** (0.000738)
    - (5) VAT -0.00148** (0.000734)
    - (6) VAT -0.00728** (0.00315)
  - Observations by column:
    - (1) 3,691
    - (2) 3,691
    - (3) 3,606
    - (4) 3,691
    - (5) 3,691
    - (6) 3,691
  - R-squared by column:
    - (1) 0.809
    - (2) 0.870
    - (3) 0.909
    - (4) 0.983
    - (5) 0.984
    - (6) 0.818
  - Additional settings: Firm FE = Y for all columns; Year Trend = N for (1)-(5), Y for (6); Controls = N for (1)-(3),(6), Y for (4)-(5); Weights vary (Sales or COGS); GCC Sample = Y for all columns; Oil indicator varies.

### Policy Channels and Recommendations
- Reforms associated with falling markups:
  - First-generation trade and financial market reforms and improvements in property rights contributed to falling markups.
- VAT as a competition policy instrument:
  - VAT reforms introduced/adjusted by some GCC countries led to reductions in market power—an additional benefit beyond increasing fiscal space.
  - VAT policy could act as a backstop to antitrust where antitrust enforcement capacity is limited.
- Broader policy levers highlighted (presented as potential channels):
  - Trade liberalization, regulatory reform and simplification, FDI promotion, e-commerce development, strengthening antitrust regulations and enforcement, improving access to finance for SMEs.

### Sample Description and Annex Statistics (selected)
- A.II.a. Summary statistics for sample between 2004 and 2022 (Sales, COGS, Pretax Income in ’000s of 2015 USD; reported PFA and CSA markups):
  - ARE: Count 1018; Sales 897804; COGS 518217; Pretax Income 149467; PFA Markups 1.51; CSA Markups 1.15
  - BHR: Count 278; Sales 271782; COGS 173070; Pretax Income 39566; PFA Markups 1.42; CSA Markups 1.09
  - EGY: Count 2064; Sales 302459; COGS 210498; Pretax Income 45091; PFA Markups 1.38; CSA Markups 1.13
  - JOR: Count 1693; Sales 129969; COGS 103593; Pretax Income 10066; PFA Markups 1.12; CSA Markups 1.06
  - KAZ: Count 380; Sales 602456; COGS 259910; Pretax Income 177650; PFA Markups 2.25; CSA Markups 1.66
  - KWT: Count 1466; Sales 284078; COGS 181791; Pretax Income 30654; PFA Markups 1.62; CSA Markups 1.14
  - LBN: Count 46; Sales 135524; COGS 90256; Pretax Income 15971; PFA Markups 1.33; CSA Markups 1.03
  - MAR: Count 980; Sales 447442; COGS 261570; Pretax Income 69099; PFA Markups 1.77; CSA Markups 1.26
  - OMN: Count 1372; Sales 152126; COGS 104384; Pretax Income 12022; PFA Markups 1.39; CSA Markups 1.07
  - PAK: Count 5587; Sales 231704; COGS 186559; Pretax Income 21435; PFA Markups 1.15; CSA Markups 1.12
  - QAT: Count 348; Sales 895081; COGS 530403; Pretax Income 177973; PFA Markups 1.58; CSA Markups 1.14
  - SAU: Count 2377; Sales 1657932; COGS 884909; Pretax Income 577603; PFA Markups 1.61; CSA Markups 1.32
  - TUN: Count 730; Sales 110941; COGS 80482; Pretax Income 5347; PFA Markups 1.16; CSA Markups 1.12
- Annex VIII selected findings:
  - Correlation between unweighted markups & price is 0.23; weighted markups & price is -0.04
  - KZ-index of equity dependence is negatively correlated with PFA Markups.

### Conclusion — Key Takeaways
- Relative levels and trends:
  - Corporate markups in the Middle East are higher compared to the US (and Europe).
  - There has been a downward trend in markups over the years, except during COVID-19.
  - GCC markups have been converging toward US and rest-of-ME levels since COVID-19 onset.
- Market structure:
  - Presence of superstar firms: larger firms with higher sales have higher market power.
  - Oil, mining, and utilities sector exhibits particularly high market power; marketable services sectors also relatively high.
- Macroeconomic implication:
  - No evidence of "greedflation" in the ME region; GCC firms on average absorb some inflationary pressure.
- Policy implication on VAT:
  - VAT reforms (introduction and rate changes in some GCC states from 2018 to 2022) led to reductions in market power while increasing fiscal space; VAT can complement competition policy.

*Source: IMF Working Paper — Market Power in the Middle East (content unit: wpiea2025001-print-pdf).*

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

### References

### Annex I. Sample Representativeness
- A.I.a. Earnings relative to GDP — page 27
- A.I.b. Sales relative to GDP — page 28

### Annex II. Sample Description
- A.II.a. Table 1: Summary Statistics for Sample between 2004 and 2022 — page 29
- A.II.b. Sample Composition: by Firm Location — page 29
- A.II.c. Sample Composition: by Country of Incorporation — page 30
- A.II.d. Sample Composition: by Industry — page 30
- A.II.e. PFA: excluding XSGA — page 31

### Annex III. Revenue-Variable Costs Ratio (CSA)
- Coverage of Revenue-Variable Costs Ratio (CSA) — page 32

### Annex IV. Value Added Tax: Robustness Table
- Robustness table related to Value Added Tax — page 33

### Annex V. Country Specific Results
- A.VI.a. Saudi Arabia — page 34
- A.VI.b. Egypt — page 35

### Annex VI. Breakdown of Markups Dynamics
- Breakdown of Markups Dynamics — page 36

### Annex VII. Correlation between Markup Measures and HH-index Measure of Concentration
- Correlation between Markup Measures and HH-index Measure of Concentration — page 37

*Source: wpiea2025001-print-pdf - References (pages indicated).*

### Annex VIII. Additional Results .........................................................................................

### Annex VIII. Additional Results

### III. Methodology
- Estimation approaches used:
  - Production Function Approach (PFA) following De Loecker & Warzynski (2012): markups recovered as μ_it = P_it / MC_it = ε_it · [S_it / R_it]^{-1}, where ε_it is the output elasticity of the variable input s, S_it is expenditure on input s, and R_it is sales/turnover.
  - Cost-Share Approach (CSA) assuming constant returns to scale: ε_it^s = E[ω_s · s_it] / E[∑_j ω_j · j_itj], yielding μ_it = R_it / E[∑ ω_j · j_itj] (equation 4).
- Key identification and implementation details:
  - PFA auxiliary regression for revenue elasticity η_st: y_it = η_st^v v_ist + η_st^x x_ist + η_st^k k_ist + ω_ist + u_ist (equation 2). ω_it modeled following Olley & Pakes (1996) as ω_it = φ_st(i_it, k_it, z_it) + u_ist (equation 3).
  - Baseline estimates assume η_st constant across the sample (η_st = η_s) and estimate a unique η_s for each sector to increase effective sample size.
  - CSA relaxes need to identify output elasticity from quantity data but imposes constant returns to scale.
- Acknowledged challenges and robustness steps:
  - Sample size limitations addressed by aggregating to broader sectoral classifications, assuming time-constant production functions, and expanding to non-listed firms from BvD Orbis (with caution due to added heterogeneity).
  - Methodological critiques noted: sensitivity to choice of input bundle (Raval 2023), revenue vs quantity data leading to potential downward bias (Klette & Griliches 1996), and concerns from Bond et al. (2021) on omitted price bias. Annex II.e and Annex III present robustness and alternative-specification results.

### IV. Firm-Level and Macroeconomic Data
- Data coverage:
  - Primary sample: comprehensive consolidated accounts of publicly listed firms from Compustat Global for firms headquartered in one of 13 ME economies (including all six Gulf states) between 2000 and 2022.
  - Earlier text references markup estimation “between 2004 and 2022” for the Middle East; baseline descriptions use the 2000–2022 Compustat coverage.
  - Expanded sample includes non-listed firms from BvD Orbis for robustness.
- Data processing:
  - Deflated firm accounts using country-level deflators from the IMF World Economic Outlook (WEO) and converted to USD for cross-country comparability.
  - Observations with negative sales or cost of goods sold dropped; sales-to-cogs ratio winsorized at the 1 percent double-sided.
- Sample size after cleaning:
  - Approximately 1300 firms and 20,700 firm-year observations.
- Benchmarking:
  - Constructed a comparable US-incorporated publicly listed firms sample from Compustat with identical sample restrictions.
- Macroeconomic variables:
  - GDP, production by industry, inflation, and VAT rates sourced from IMF WEO.

### V. Main Results — Market Power in the Middle East
- Aggregate and cross-region comparisons:
  - Corporate market power among listed firms in the Middle East is higher than in the US.
  - Unlike the US, GCC market power has not exhibited a secular upward trend over the last twenty years; overall ME markups are on a downward trend (except for the COVID-19 period).
  - A significant spike in market power in 2016 is observed after the listing of Saudi Arabia’s ARAMCO.
  - Excluding ARAMCO, GCC markups were stable around 1.45 between 2010 and 2020; average markup in the GCC was close to 1.6 around 2010.
- Robustness across estimators:
  - PFA and CSA reveal similar dynamics for the ME region, with markups decreasing over time (see Figure 3; alternative CSA imposes constant returns to scale and obviates quantity data).
- Firm-level heterogeneity and “superstar” phenomenon:
  - Listed firms exhibit a “superstar” phenomenon: firms with higher sales have higher market power across sectors (oil, mining & utilities and other sectors).
  - Oil, mining, and utilities sector has higher market power than other sectors across the sales distribution.
  - Resource-rich ME countries show higher average markups; Morocco is the only non-oil country in the sample with comparable markup levels.
- Country-level heterogeneity within GCC:
  - Average markups did not change much overall but showed weak upward trends in Bahrain, Kuwait, and Oman in recent years, and downward trends in Qatar, Saudi Arabia, and the UAE.
  - Potential explanations cited include structural reforms (e.g., Qatar’s competition law of 2006; Saudi’s competition law of 2019), changes in price elasticity of demand, and licensing/regulatory practices.
- Industry-level heterogeneity:
  - Construction sector among listed firms shows the lowest average markups (sample caveat: absence of some large construction firms may bias this average downward).
  - Oil, mining, and utilities have the highest average markups. Market services display higher markups than many other industries.
- International exposure and who benefits from falling markups:
  - Firms with zero foreign currency exposure generally have lower markups than firms with FX exposure.
  - CSA results: both domestically oriented and externally oriented firms show similar reducing trends — implying both domestic and foreign consumers benefited from falling markups.
  - PFA results: firms with no FX exposure fluctuate around a flat level (domestic consumers did not clearly benefit), while firms with FX exposure show a downward trend (implying rising global competitiveness for ME firms).
  - Result heterogeneity noted by country (e.g., Saudi firms’ markups similar across FX exposure groups).

### V.a. Macroeconomic Implications and Inflation
- Inflation passthrough and “greedflation”:
  - Across the GCC, evidence suggests firms absorbed some price rises by lowering markups following inflation surprises; the markup response to inflation surprises is negative (significance varies by country).
  - Overall, the authors state they “do not find evidence of ‘greedflation’ in the Middle East.”
  - These ME findings align with evidence reported for advanced economies in Chapter 1, October 2023 World Economic Outlook.

### V.b. Decomposition and Dynamics
- Drivers of falling markups:
  - Within-firm markups are falling and drive a significant portion of the decrease in aggregated weighted markups for both PFA and CSA measures.
  - Net entry of more productive firms may contribute to falling markups for one of the markup measures; for the other measure, net entry plays a limited role.
- Implication: falling within-firm markups indicate changes are not solely driven by sample composition but reflect firm-level dynamics.

### V.c. Policy Channels and VAT Evidence
- Reforms associated with falling markups:
  - First-generation trade and financial market reforms and improvements in property rights in the ME region contributed to the falling markups trend.
- VAT as a competition policy instrument:
  - Using cross-country variation in VAT adoption in GCC (2018–2022), the authors find that VAT reforms introduced and adjusted by some GCC countries led to a reduction in market power — an additional benefit beyond increasing fiscal space.
  - The paper suggests VAT policy could act as a backstop to antitrust in promoting competition where antitrust enforcement capacity is limited.
- Broader policy levers highlighted (noted as potential channels rather than prescriptive rankings):
  - Trade liberalization, regulatory reform and simplification, FDI promotion, e-commerce development, strengthening antitrust regulations and enforcement, and improving access to finance for SMEs.

### Robustness and Limitations (as discussed)
- Robustness checks:
  - Alternative markup estimation via CSA presented (Annex III) to mitigate quantity-data criticism (Bond et al., 2021).
  - Expanded sample to BvD Orbis and alternative specifications reported in Annex II and Annex III.
- Limitations and cautions:
  - Small listed-firm sample in ME motivates aggregation and time-constancy assumptions.
  - Expansion to non-listed firms increases heterogeneity and requires careful interpretation.
  - Revenue-based elasticities may differ from output elasticities (potential downward bias); concerns and sensitivity analyses addressed in Annex II.e and elsewhere.

*Source: IMF Working Paper — Market Power in the Middle East; Annex VIII. Additional Results.*

### Annex I reports the results of a decomposition of markup evolution into 3 terms: evolution due to net firm entry,

### wpiea2025001-print-pdf - Annex I reports the results of a decomposition of markup evolution into 3 terms: evolution due to net firm entry,

### Annex I — Decomposition of markup evolution
- Decomposition components:
  - evolution due to net firm entry
  - evolution due to within firm reduction in markups
  - evolution due to reallocation of economic activity
- Presentation:
  - Green line: headline weighted markup estimators.
  - Dashed lines: alternative scenarios starting from 2004.
    - Dashed red line: scenario where only within-firm markups are allowed to change. In both markup measures provided, the red dashed line drives the reduction in markups.
    - Dashed grey line: scenario where relocation of economic activity is the only term allowed to evolve. The relocation term:
      - stagnates in the production function approach markups
      - rises in the cost-share markups
      - implication: for the production function measure, reallocation did not play a part in markup evolution; for the cost-share measure, reallocation to firms with higher markups led to an increase in markups.
    - Net entry (black dashed line): remained broadly constant for the production function measure and fell similarly to within for the other measure.
- Summary finding:
  - For both markup estimates, the declining evolution is partly determined by within-firm changes in markups as opposed to entry on its own.
  - For one markup measure, net firm entry plays an extremely limited role in the change of markups.

### V.b Markups in ME and macroeconomic implications
- Markups and inflation in ME:
  - Finding: a rise in global market power did not contribute to the GCC inflation surge of 2022.
  - Observations 2021–2022:
    - Corporate profits and dividend payouts in the GCC increased robustly over 2021-2022 (Figure 9).
    - Wages rose relatively slowly compared to prices (IMF ME&CA REO October 2023).
  - Firm-level markup evidence:
    - Little or no change in firm markups across various sectors in ME countries (Figure 7).
    - Overall markups have exhibited a downward trend (except for COVID years).
  - Impulse response to inflation:
    - After a 1 percent inflation shock, firms in the GCC on average reduce their markups by 0.05 units relative to an average of 1.3 after two years of the shock.
    - Relationship: negative for all GCC countries (significance varies by country).
    - Interpretation: GCC firms absorb some inflationary pressures and pass less of price changes onto consumers.
    - Non-GCC ME: firms do not adjust markups in response to inflationary shocks and tend to pass price rises to customers.

### V.c VAT reform and market power — institutional background
- GCC unified VAT framework:
  - Agreement year: 2017.
  - Framework key articles:
    - mandates signatories adopt a standard rate for VAT at 5 percent.
    - requires introduction of local laws to implement the treaty.
  - Implementation status (as presented):
    - Six years after signing (i.e., through 2023), only four out of six countries implemented the tax.
    - Saudi Arabia and the United Arab Emirates were the first to implement in 2018.
    - Kuwait and Qatar are yet to implement the tax framework of 2017.
    - Saudi Arabia hiked rates in 2020.
    - Bahrain hiked rates in 2022.
  - Research leverage: staggered implementation allows multiple staggered difference-in-difference identification strategies to study VAT impact on firm-level market power.

### V.c Identification strategies and theoretical prediction
- Theoretical model:
  - Firm profit maximization with final good consumption tax τ and inverse demand Q(P) yields markup expression (as in text):
    - μit := Pit / C′(Q(P(1+τ))) = [1 + 1/((1+τ)γ)]−1 where γ := (Qit / (P·Q)) p′.
    - Effect on logged firm markups: ∂ ln(μit)/∂τ = γ·[γ(1+τ)]−2 / {1 + [γ(1+τ)]−1}.
    - Since γ < 0, numerator negative; for sufficiently elastic demand (γ·(1+τ) < −1) the tax reduces the wedge between output prices and marginal costs, implying VAT reduces markups.
- Empirical identification strategies:
  - Staggered adoption diff-in-diff:
    - Parallel trends assumption stated in equation (8).
    - Robustness tests: pre-trends tests and conditional parallel trends on a matched sample.
  - Intermediate-final goods diff-in-diff:
    - Within-country comparison: intermediate-sector firms (do not pay VAT and claim refunds) used as control for final-sector firms that pay VAT.
    - Identifying assumption stated in equation (9).

### V.c.3 Empirical findings on VAT impact
- Staggered Diff-in-Diff results:
  - A 1 percentage point increase in the VAT rate is associated with a 5 percent reduction in markups.
  - The effect is dynamic and not immediate (adjustment occurs over time; reported in Figure 14).
  - No evidence of pre-reform common trends violation (per equation 8).
  - Limitation: cannot trace effects beyond 3 years after reform due to limited firm accounts after 2021.
- Annex IV / robustness:
  - Headline average impact (Column (1) of table in Annex IV): an additional 1% point of VAT is an average reduction of 1% in markups estimated using the production function approach in the GCC sample.
  - Result robustness: holds across different specifications, weighting methods, and alternative markup measure.

### VI. Conclusion — key takeaways
- Relative levels and trends:
  - Corporate markups in the Middle East are higher compared to the US (and Europe), for both GCC and non-GCC countries.
  - There has been a downward trend in markups over the years, except during COVID-19.
  - GCC markups have been converging toward US and rest-of-ME levels since COVID-19 onset.
- Market structure observations:
  - Presence of superstar firms: larger firms with higher sales tend to have higher market power across sectors.
  - Oil, mining, and utilities sector exhibits particularly high market power.
  - Marketable services sectors and oil/mining/utilities generally have higher markups.
- Macroeconomic implications:
  - No evidence of "greedflation" in the ME region; relationship between market power and inflationary pressures is not significant or is negative for the GCC.
- Policy implication on VAT:
  - VAT reforms (introduction in some Gulf states from 2018 to 2022) led to reductions in market power while increasing fiscal space.
  - VAT can serve as an additional deterrent to corporate market power alongside traditional competition policy.

*Source: IMF Working Paper — Market Power in the Middle East (content unit: wpiea2025001-print-pdf).*

### Annex I. Sample Representativeness

### Annex I. Sample Representativeness

### A.I.a. Earnings relative to GDP
- Aggregated country-level average Earnings relative to GDP.
- Map of average Earnings relative to GDP in 2020.

### A.I.b. Sales relative to GDP
- Aggregated country-level average Sales relative to GDP.
- Map of average Sales relative to GDP in 2020.

---

### Annex II. Sample Description

### A.II.a. Table 1: Summary Statistics for Sample between 2004 and 2022
- Count of Firm-year and summary statistics (Sales, COGS, Pretax Income, PFA Markups, CSA Markups) reported for listed firms headquartered in the ME. Sales, COGS, and Pretax income are reported in ’000s of 2015 USD.
- Country-level entries:
  - ARE: Count 1018; Sales 897804; COGS 518217; Pretax Income 149467; PFA Markups 1.51; CSA Markups 1.15
  - BHR: Count 278; Sales 271782; COGS 173070; Pretax Income 39566; PFA Markups 1.42; CSA Markups 1.09
  - EGY: Count 2064; Sales 302459; COGS 210498; Pretax Income 45091; PFA Markups 1.38; CSA Markups 1.13
  - JOR: Count 1693; Sales 129969; COGS 103593; Pretax Income 10066; PFA Markups 1.12; CSA Markups 1.06
  - KAZ: Count 380; Sales 602456; COGS 259910; Pretax Income 177650; PFA Markups 2.25; CSA Markups 1.66
  - KWT: Count 1466; Sales 284078; COGS 181791; Pretax Income 30654; PFA Markups 1.62; CSA Markups 1.14
  - LBN: Count 46; Sales 135524; COGS 90256; Pretax Income 15971; PFA Markups 1.33; CSA Markups 1.03
  - MAR: Count 980; Sales 447442; COGS 261570; Pretax Income 69099; PFA Markups 1.77; CSA Markups 1.26
  - OMN: Count 1372; Sales 152126; COGS 104384; Pretax Income 12022; PFA Markups 1.39; CSA Markups 1.07
  - PAK: Count 5587; Sales 231704; COGS 186559; Pretax Income 21435; PFA Markups 1.15; CSA Markups 1.12
  - QAT: Count 348; Sales 895081; COGS 530403; Pretax Income 177973; PFA Markups 1.58; CSA Markups 1.14
  - SAU: Count 2377; Sales 1657932; COGS 884909; Pretax Income 577603; PFA Markups 1.61; CSA Markups 1.32
  - TUN: Count 730; Sales 110941; COGS 80482; Pretax Income 5347; PFA Markups 1.16; CSA Markups 1.12
- Note: PFA markup estimation is explained in the text. Share estimator of markups is reported in Annex E.

### A.II.b. Sample Composition: by Firm Location
- Fraction of the Sample by Location of Firm’s Headquarter. (Maps/figures referenced in source.)

### A.II.c. Sample Composition: by Country of Incorporation
- Fraction of the Sample by Firm’s Country of Incorporation. (Maps/figures referenced in source.)

### A.II.d. Sample Composition: by Industry
- Fraction of the Sample by Firm’s Industry. (Maps/figures referenced in source.)

### A.II.e. PFA: excluding XSGA
- Unweighted markups using production function approach excluding XSGA.
- COGS-weighted markups using production function approach excluding XSGA.

---

### Annex III. Revenue-Variable Costs Ratio (CSA)
- Markups based on CSA: Excluding ARAMCO; Including ARAMCO.
- Markups based on CSA for (i) GCC country and (ii) Industries in GCC.
- Markups based on CSA for (i) Superstar Phenomena and (ii) Deciles of Revenue.

---

### Annex IV. Value Added Tax: Robustness Table
- Regression table reports effect of a 1% point increase in value-added tax on market power (sample: GCC headquartered firms). Clustered standard errors in parentheses. PFA markup estimation explained in text. Share estimator markups explained in Annex E. Significance: *** p<0.01, ** p<0.05, * p<0.1.
- Column headers (DV): (1) log(μ_PFA), (2) log(μ_PFA), (3) log(μ_CS), (4) log(μ_PFA), (5) log(μ_PFA), (6) log(μ_PFA)
- VAT coefficients and standard errors:
  - (1) VAT -0.0117** (0.00461)
  - (2) VAT -0.00798** (0.00352)
  - (3) VAT -0.00667*** (0.00247)
  - (4) VAT -0.00159** (0.000738)
  - (5) VAT -0.00148** (0.000734)
  - (6) VAT -0.00728** (0.00315)
- Constant terms and standard errors:
  - (1) Constant 0.503*** (0.0280)
  - (2) Constant 0.382*** (0.0206)
  - (3) Constant 0.263*** (0.0150)
  - (4) Constant -0.176*** (0.0636)
  - (5) Constant -0.134** (0.0642)
  - (6) Constant 18.11 (11.62)
- Observations:
  - (1) 3,691
  - (2) 3,691
  - (3) 3,606
  - (4) 3,691
  - (5) 3,691
  - (6) 3,691
- R-squared:
  - (1) 0.809
  - (2) 0.870
  - (3) 0.909
  - (4) 0.983
  - (5) 0.984
  - (6) 0.818
- Additional model settings:
  - Firm FE: Y for all columns
  - Year Trend: N for (1)-(5); Y for (6)
  - Controls: N for (1)-(3),(6); Y for (4)-(5)
  - Weight: (1) Sales; (2) COGS; (3) Sales; (4) Sales; (5) Sales; (6) Sales
  - GCC Sample: Y for all columns
  - Oil: N for (1)-(4); Y for (5)-(6)

---

### Annex V. Country Specific Results

### A.VI.a. Saudi Arabia
- Figure 17: Markups using PFA in Saudi Arabia.
- Figure 18: Markups based on CSA in Saudi Arabia.

### A.VI.b. Egypt
- Figure 19: Markups using PFA in Egypt.
- Figure 20: Markups based on CSA in Egypt.

---

### Annex VI. Breakdown of Markups Dynamics
- PFA markups excluding ARAMCO.
- CSA-markups excluding ARAMCO.

---

### Annex VII. Correlation between Markup Measures and HH-index Measure of Concentration
- PFA markups excluding ARAMCO.
- CSA-markups excluding ARAMCO.

---

### Annex VIII. Additional Results
- HH-index based on 4-digit industry classification excluding ARAMCO.
- HH-index based on 4-digit industry classification including ARAMCO.
- Correlation between unweighted markups & price is 0.23; weighted markups & price is -0.04
- (i) Gulf Cooperation Council (ii) Rest of Middle East
- KZ-index of equity dependence is negatively correlated with PFA Markups.

*Market Power in the Middle East — Working Paper No. WP/2025/001*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2025/english/wpiea2025001-print-pdf.pdf_