## wpiea2019096

## Source details

**Canonical URL:** [wpiea2019096](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019096.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wpiea2019096.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019096.pdf.json)

---

### Overview and research question
- Research question: Does the growth effect of a tax depend on whether it is raised through a rate increase or through a base broadening, with focus on the value-added tax (VAT)?
- Key empirical claim: Increasing VAT revenue through base broadening (higher C-efficiency) is more growth friendly in OECD countries than increasing VAT revenue via the VAT standard rate.

### Decomposition of the VAT (formal framework)
- VAT revenue share decomposition: V/T = τ e c, where:
  - τ is the VAT standard rate.
  - e (= V/(τ C)) is C-efficiency.
  - c (= C/T) is the ratio of final consumption to total tax revenue.
- Interpretation:
  - C-efficiency measures departures from a perfectly enforced uniform VAT; values lower than one reflect reduced rates, exemptions, and/or compliance limitations.
  - For advanced economies, trends in C-efficiency are argued to be primarily explained by policy choices (rate differentiations and exemptions) rather than compliance changes.

### Theoretical mechanisms and hypotheses
- Hypothesis: Raising VAT revenue through increases in C-efficiency (base broadening — removing reduced rates and exemptions) is more conducive to long-run growth than raising VAT revenue through increases in the standard VAT rate.
- Mechanisms:
  - Exemptions create production taxation because VAT on inputs is not refunded when sales are exempted, distorting firms’ input choices and potentially lowering long-run growth.
  - Eliminating rate differentiations reduces administration costs and decreases consumer spending distortions.
  - Raising VAT via the standard rate forgoes potential efficiency gains from base broadening.

### Data and sample
- Sample: 30 OECD countries during the 1970-2016 period; annual data; unbalanced panel; only countries with at least 20 annual successive observations included.
- Countries: Austria, Belgium, Chile, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy, Japan, Korea, Latvia, Luxembourg, Netherlands, New Zealand, Norway, Poland, Portugal, Slovak Republic, Spain, Sweden, Switzerland, Turkey, United Kingdom.
- VAT adoption context: "As of November 1, 2018, 168 countries and territories worldwide have adopted the VAT, including all the OECD countries with the only exception of the US."
- Average statistic reported: "On average, 20 percent of total taxation is collected through the VAT in the OECD."
- Crisis period indicator: crisis_i,t = 1 for years after 2008 (inclusive), 0 otherwise; Global Financial Crisis dated 2007-09 is referenced.

### Descriptive statistics (selected exact values from Table 1)
- Number of observations: 944 from 30 countries (for variables in Table 1).
- Growth rate of real GDP pc: Mean 0.0226; Standard deviation 0.0314; Minimum -0.145; Maximum 0.251.
- Total taxes/GDP: Mean 0.342; Standard deviation 0.0742; Minimum 0.108; Maximum 0.494.
- Consumption taxes/Total taxes: Mean 0.336; Standard deviation 0.0872; Minimum 0.140; Maximum 0.686.
- Personal income taxes/Total taxes: Mean 0.250; Standard deviation 0.104; Minimum 0.0496; Maximum 0.574.
- Corporate income taxes/Total taxes: Mean 0.0864; Standard deviation 0.0459; Minimum 0.0167; Maximum 0.332.
- VAT/GDP: Mean 0.0667; Standard deviation 0.0176; Minimum 0.0128; Maximum 0.110.
- VAT/Total taxes: Mean 0.200; Standard deviation 0.0581; Minimum 0.0460; Maximum 0.451.
- Standard rate: Mean 0.182; Standard deviation 0.0489; Minimum 0.0300; Maximum 0.270.
- C-efficiency: Mean 0.563; Standard deviation 0.131; Minimum 0.323; Maximum 1.244.
- Consumption ratio: Mean 2.106; Standard deviation 0.687; Minimum 1.060; Maximum 6.468.
- Investment rate: Mean 0.238; Standard deviation 0.0456; Minimum 0.0982; Maximum 0.415.
- Employment growth: Mean 0.00895; Standard deviation 0.0212; Minimum -0.136; Maximum 0.106.
- Additional notes:
  - Average annual growth rate of real GDP per capita is 2.3 percent (reported as 0.0226).
  - Total taxes as share of GDP is 34.2 percent (0.342).
  - Share of VAT within total taxes is 20.0 percent (0.200).
  - C-efficiency highest on average in New Zealand at 97.5 percent (sample period).
  - Maximum C-efficiency in sample: Luxembourg 124.4 percent.
  - Consumption ratio averaged at 211 percent (reported as 2.106).

### Econometric methodology
- Primary estimator: Pooled Mean Group (PMG) of Pesaran et al. (1999); Hausman test predominantly supports PMG (p-value > 0.05 in all cases).
- Alternative estimator: Mean Group (MG) of Pesaran and Smith (1995) used for heterogeneity checks.
- Baseline model: ARDL(1) re-parameterized in error-correction form; dependent variable g_i,t is growth rate of annual real GDP per capita; controls include investment rates and employment growth; includes crisis dummy and lagged dependent variable.
- Government budget constraint (GBC) framework: tax variable changes specified so changes in one component are offset by changes in others.

### Main empirical findings — long-run PMG estimates (exact reported coefficients)
- A. Growth effects of VAT revenue as a whole (Table 2; financing taxes = Rest / Total income / Personal income / Corporate income)
  - Common findings:
    - VAT/Total taxes positive and significant across specifications when VAT rise is offset by falls in income taxes.
    - Total taxes/GDP negative and significant throughout.
    - Investment rate and employment growth positive and significant.
    - EC coefficient (φ) strongly negative and significant.
    - Crisis dummy negative and significant.
  - Representative coefficients (Column (2), financing taxes = Total income; Countries: 30; Observations: 981):
    - Total taxes/GDP: -0.0872*** (t = -3.98)
    - VAT/Total taxes: 0.0848*** (t = 3.18)
    - Investment rate: 0.0480** (t = 2.46)
    - Employment growth: 0.465*** (t = 12.43)
    - EC coefficient (φ): -0.964*** (t = -20.09)
    - Crisis: -0.0168*** (t = -7.77)
  - Interpretation: coefficient of 0.085 indicates that a one percentage point increase in the share of the VAT in total taxes, offset by a fall in the share of total income taxes of the same size, is associated with a 0.085 percentage points increase in annual growth in the long run.
  - Larger VAT coefficient when offset by corporate income taxes (0.107) than by personal income taxes (0.0731).

- B. Growth effects of VAT components (Table 3; focus on C-efficiency vs standard rate)
  - Column (2) (omit standard rate; include ln(C-efficiency) and ln(consumption ratio); Countries: 30; Observations: 944):
    - Total taxes/GDP: -0.0558 (t = -1.40)
    - VAT/Total taxes: -0.0166 (t = -0.53) — not significant
    - C-efficiency (log): 0.0347*** (t = 4.86)
    - Consumption ratio (log): 0.0209* (t = 1.79)
    - EC coefficient (φ): -0.911*** (t = -17.57)
    - Crisis: -0.0133*** (t = -6.08)
    - Interpretation: when the VAT is raised only through a rise in the standard rate (omitted), a rise in the VAT financed by a fall in income taxes does not promote growth (VAT coefficient not significant).
  - Column (3) (omit C-efficiency; include standard rate and consumption ratio):
    - Total taxes/GDP: -0.0447 (t = -1.12)
    - VAT/Total taxes: 0.187*** (t = 5.70)
    - Standard rate (log): -0.0277*** (t = -4.87)
    - EC coefficient (φ): -0.909*** (t = -17.34)
    - Crisis: -0.0138*** (t = -6.25)
    - Interpretation: if VAT is raised only through a rise in C-efficiency (omitted here), the VAT share coefficient is positive and highly significant — raising VAT via C-efficiency is associated with higher long-run growth.
  - Column (4) (alternate specification):
    - Total taxes/GDP: -0.0573* (t = -1.66)
    - VAT/Total taxes: 0.102* (t = 1.68)
    - Standard rate (log): -0.0175** (t = -2.01)
    - C-efficiency (log): 0.0153 (t = 1.53)
    - EC coefficient (φ): -0.924*** (t = -17.41)
    - Crisis: -0.0136*** (t = -6.03)
  - Key qualitative result: A VAT increase driven by higher C-efficiency is associated with positive long-run growth effects; a VAT increase driven by a higher standard rate is not growth-promoting and the log standard rate often has a negative and significant coefficient (e.g., -0.0277***).

### Channels, magnitudes, and additional numeric findings
- C-efficiency magnitude example: coefficient of 0.0347*** on C-efficiency (log) in Table 3 Column (2) — interpreted as a one percent increase in C-efficiency, offset by a fall in the standard rate, promotes growth by 0.035 percentage points (text reports "The coefficient of 0.035 means that a one percent increase in C-efficiency, offset by a fall in the standard rate, promotes growth by 0.035 percentage points.").
- Table 4 (allowing for different channels; Countries: 30; Observations: 944) selected coefficients:
  - C-efficiency: 0.0461*** (Col 1); 0.0184* (Col 2); 0.0495*** (Col 3); 0.0332** (Col 4).
  - Standard rate: -0.0338*** (Col 1); -0.0289*** (Col 2); -0.0378*** (Col 3); -0.0201 (Col 4).
  - VAT/Total taxes: values vary across columns including 0.227*** (Col 2), 0.126* (Col 3), -0.0406 (Col 1).
  - EC coefficient (φ): ranges from -0.930*** to -0.792*** across columns.
  - Crisis dummy: around -0.0159*** to -0.0130*** across columns.
- Channels of transmission:
  - When investment rate is omitted, coefficients on VAT and C-efficiency increase, consistent with VAT design affecting growth via capital accumulation as well as TFP.
- Cyclicality:
  - VAT/Total taxes correlation with output gap not statistically significant (VAT: coef = .022; (robust) se = .0365; t = 0.60).
  - Income taxes/Total taxes positively correlated with output gap (Income taxes: coef = .219; (robust) se = .043; t = 5.08).

### Robustness checks and sensitivity analysis (exact reported outcomes)
- Robustness checks include:
  1. Excluding the Global Financial Crisis period and aftermath (2008-16).
  2. Including lagged real GDP per capita to address convergence.
  3. Adopting different short-run dynamics (e.g., ARDL with 2 lags).
  4. Addressing reverse causality by omitting contemporaneous relations.
- Selected robustness table results:
  - Table 5 (omitting 2008-16): VAT/Total taxes coefficients include 0.0825** (omitting 2008-16, Col 1); C-efficiency 0.0275*** (omitting 2008-16, Col 5).
  - Table 5 (adding lagged GDP): VAT/Total taxes 0.130*** (Col 1); C-efficiency 0.0213*** (Col 5); lagged real GDP per capita coefficients -0.0230*** to -0.0268*** where included.
  - Table 6 (ARDL with 2 lags): VAT/Total taxes include 0.114*** (Col 1), 0.134*** (Col 2), 0.135*** (Col 3), 0.0822* (Col 4), 0.257*** (Col 5); C-efficiency 0.0348*** (upper sub-table Col 5).
  - Table 6 (omitting contemporaneous relation): VAT/Total taxes 0.0654** (Col 1), 0.272*** (Col 6); C-efficiency 0.0238*** (Col 5).
- Robustness summary:
  - The result that VAT reallocations are generally more growth promoting than total income taxes persists.
  - Increasing VAT via C-efficiency, offset by a fall in income taxes, remains growth promoting across robustness checks.
  - A VAT increase through the standard rate or consumption ratio does not robustly foster growth.

### Interpretation, policy implications, and caveats
- Interpretation and policy-relevant conclusions:
  - Revenue-neutral tax reform that shifts the tax mix toward VAT and away from income taxes can be growth-promoting in the long run, but the effect depends critically on how VAT revenue is increased.
  - Raising VAT revenue via improvements in C-efficiency (broadening the base and improving collection) tends to be growth-enhancing.
  - Raising VAT revenue via increases in the statutory standard rate tends to be neutral or growth-reducing.
  - Corporate income taxes appear more distortionary than personal income taxes in the context of VAT-financed tax shifts (larger VAT coefficient when offset by corporate income taxes).
  - Higher overall tax levels (Total taxes/GDP) are associated with lower long-run growth ceteris paribus.
- Caveats:
  - Decomposition does not capture behavioral responses of taxpayers to policy changes (e.g., compliance changes induced by rate changes).
  - C-efficiency increases can sometimes reflect policy changes that degrade VAT quality (e.g., denying VAT refunds to exporters), which may not yield efficiency gains.
  - Results focus on long-run growth effects; distributional effects are not analyzed here.

### Data sources, tests, and documentation notes
- Data sources:
  - Real GDP per capita growth: Penn World Tables (PWT 9.0) and IMF WEO for 2015 and 2016.
  - Tax revenue data, including VAT at general government level: OECD Revenue Statistics Database.
  - VAT standard rates: assembled using OECD, European Commission, IMF’s Tax Policy Reform Database, CIAT (for Chile), and official country documents.
  - Final consumption and controls (investment rate, employment growth): WEO.
- Panel unit root tests:
  - Maddala and Wu (1999) and Im et al. (2003) reject the null of non-stationarity at the 1% level for all variables except VAT standard rate which cannot be tested by Im et al. (2003).
  - Residuals from estimated models are stationary, supporting long-run relations.

*Source: IMF Working Paper (wpiea2019096) — content excerpt covering model, data, Table 1 descriptive statistics, Tables 2–4 long-run PMG estimation results, robustness checks, and related discussion.*

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

### wpiea2019096 - References .............................................................................................................

### Overview and research question
- Research question: Does the growth effect of a tax depend on whether it is raised through a rate increase or through a base broadening, with focus on the value-added tax (VAT)?
- Key empirical claim: Increasing VAT revenue through base broadening (higher C-efficiency) is more growth friendly in OECD countries than increasing VAT revenue via the VAT standard rate.

### Decomposition of the VAT (formal framework)
- VAT revenue as a share of total tax revenue, V/T, is decomposed as V/T = τ e c, where:
  - τ is the VAT standard rate (the rate applied to the largest portion of taxed consumption).
  - e (= V/(τ C)) is C-efficiency, the ratio of VAT revenue to the product of the VAT standard rate and final consumption C (excluding VAT revenue collection).
  - c (= C/T) is the ratio of final consumption to total tax revenue.
- Interpretation of C-efficiency:
  - C-efficiency measures departures from a perfectly enforced uniform VAT; it takes a value lower than one to the extent that reduced rates and exemptions apply and/or compliance is limited.
  - For advanced economies, trends in C-efficiency are argued to be primarily explained by policy choices (rate differentiations and exemptions) rather than compliance changes.

### Theoretical mechanisms and hypotheses
- Hypothesis: Raising VAT revenue through increases in C-efficiency (base broadening — removing reduced rates and exemptions) is more conducive to long-run growth than raising VAT revenue through increases in the standard VAT rate.
- Mechanisms:
  - Exemptions create production taxation because VAT on inputs is not refunded when sales are exempted, distorting firms’ input choices and potentially lowering long-run growth.
  - Eliminating rate differentiations reduces administration costs and decreases consumer spending distortions.
  - Raising VAT via the standard rate forgoes these potential efficiency gains from base broadening.

### Data and sample
- Novel dataset: 30 OECD countries during the 1970-2016 period.
- VAT adoption context: "As of November 1, 2018, 168 countries and territories worldwide have adopted the VAT, including all the OECD countries with the only exception of the US."
- Average statistic reported: "On average, 20 percent of total taxation is collected through the VAT in the OECD."
- Crisis period indicator: dummy variable crisis_i,t equals one for years after 2008 (inclusive) and zero otherwise; Global Financial Crisis dated 2007-09 is referenced.

### Econometric methodology
- Primary estimator: Pooled Mean Group (PMG) method of Pesaran et al. (1999), which:
  - Estimates long-run relationships constrained to be equal across countries, while allowing heterogeneous short-run dynamics.
  - Is preferred when the Hausman test cannot reject equality of long-run coefficients (authors report the Hausman test predominantly supports PMG in their estimations).
- Alternative estimator considered: Mean Group (MG) method of Pesaran and Smith (1995) for heterogeneity checks.
- Baseline model:
  - Uses an ARDL(1) structure re-parameterized in error-correction form.
  - Dependent variable g_i,t is the growth rate of annual real GDP per capita.
  - Vector f_i,t contains tax variables; z_i,k,t contains controls (investment rates and employment growth among them).
  - Includes lagged dependent variable, crisis dummy, and an error term.
- Government budget constraint (GBC) framework:
  - Tax variables are specified so that changes in one tax component are offset by changes in others (e.g., a rise in VAT offset by a fall in income taxes for a given total tax revenue).

### Main empirical findings
- Two complementary exercises:
  1. For a given level of total tax revenue, a rise in VAT offset by a fall in income taxes increases growth if the VAT rise comes through higher C-efficiency, but not if it comes through a higher VAT standard rate.
  2. For a given level of VAT revenue, a rise in C-efficiency financed by a fall in the standard rate promotes growth — implying C-efficiency increases are more growth promoting than rate increases.
- Robustness checks reported to largely support the results (see Robustness subsection).

### Robustness checks and sensitivity analysis
- Robustness checks include:
  1. Excluding the Global Financial Crisis (GFC) period and aftermath (2008-16) to account for systematically different post-GFC growth patterns.
  2. Including lagged real GDP per capita as an independent variable to address possible convergence effects.
  3. Adopting different short-run dynamics (different lag structures in the regression).
  4. Considering possible endogeneity biases due to reverse causality (growth affecting tax variables).
- Additional robustness elements mentioned elsewhere in the document:
  - Use of ARDL with 2 lags and omission of contemporaneous relation (Table 6).
  - Panel unit root tests (Table 7).
  - Consideration of omitting the post-crisis period and adding initial GDP per capita (Table 5).
  - Investigation of allowing for different channels of transmission (Table 4).

### Relation to literature
- Links to literature finding consumption taxes are less harmful for growth than income taxes in revenue-neutral reallocations (Arnold et al., 2011; Acosta-Ormaechea et al., 2018).
- Contrasts with EC (2011), which finds negligible long-run effects of VAT components on consumption or GDP growth in 40 countries; current study emphasizes policy-relevant identification via the GBC.
- C-efficiency dynamics and determinants discussed in prior work:
  - Keen (2013) and Ueda (2017) decompose C-efficiency into policy and compliance gaps and find policy gaps drive C-efficiency trends in advanced economies.
  - Sancak et al. (2010) find C-efficiency is pro-cyclical in a broad sample.
  - Tanzi and Davoodi (2000), Aizenman and Jinjarak (2009), De Mello (2009) identify structural correlates of C-efficiency (corruption, political regime durability, tax administration efficiency).

### Conclusion and policy implication
- Policy implication explicitly stated: In OECD countries, increasing VAT revenue through base broadening measures that raise C-efficiency (fewer reduced rates and exemptions) is more growth friendly than increasing revenue via higher VAT standard rates.
- The paper emphasizes the importance of tax collection design (not just tax composition) for long-run growth outcomes.

*Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019096.pdf*

### 2.78 percent, while over 2010-2016 is 1.56 percent. One theoretical explanation for the apparent structural

### wpiea2019096 - 2.78 percent, while over 2010-2016 is 1.56 percent. One theoretical explanation for the apparent structural

### Theoretical framework and econometric setup
- Model specification:
  - Short-run and long-run decomposition via error-correction representation (Eq. (4)):
    - ∆g_{c,t} = φ_c [ g_{c,t−1} − θ_c s_{VAT,c,t−1} − ... ] + long-run and short-run terms + ζ_c crisis_{c,t} + δ^0_c + δ^1_c + ε_{c,t}.
    - φ_c = −(1 − λ_c) represents the error-correction speed of adjustment.
  - Long-run coefficient of interest on VAT share: θ_{i,VS} − θ_{i,IS} (or variants) measures growth effect of revenue-neutral increase in VAT offset by income taxes.
- Decomposition of VAT revenue share:
  - s_{c,t} (= V_{c,t}/T_{c,t}) = τ_{c,t} e_{c,t} c_{c,t}, where:
    - τ_{c,t} is the VAT standard rate,
    - e_{c,t} = V_{c,t}/(τ_{c,t} C_{c,t}) is C-efficiency,
    - c_{c,t} = C_{c,t}/T_{c,t} is the ratio of final consumption to total tax revenue.
- Linear approximation used:
  - s_{c,t} ≅ a + b ln(s)_{c,t} = a + b( ln(τ)_{c,t} + ln(c)_{c,t} + ln(e)_{c,t} ). (Eq. (6))
  - Omitting one VAT component (e.g., ln(τ)) allows identification of effects of the remaining components.
- Alternative long-run coefficients:
  - When the VAT increase is driven only by the standard rate (omitted), long-run coefficient on VAT share becomes θ_{i,VS} + θ_{i,T}/b − θ_{i,IS} (Eq. (8)).
  - Long-run effect of a rise in C-efficiency (offset by a fall in the standard rate) captured by coefficient on ln(e): θ_{i,E} − θ_{i,T}.

### Dataset
- Coverage:
  - 30 OECD countries during the 1970-2016 period.
  - Annual data; only countries with at least 20 annual successive observations included.
  - The dataset is an unbalanced panel.
- Countries included: Austria, Belgium, Chile, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Israel, Italy, Japan, Korea, Latvia, Luxembourg, Netherlands, New Zealand, Norway, Poland, Portugal, Slovak Republic, Spain, Sweden, Switzerland, Turkey, and United Kingdom.
  - Australia and Canada not included; US does not have a VAT.
- Tax revenue definitions:
  - Total taxes = consumption taxes (including VAT) + personal income taxes + corporate income taxes + property taxes + social security contributions.
- Treatment of VAT standard rate changes:
  - Adjusted by month of change: adjusted rate for year t = ((x-1)/12)*rate_t + ((12-x+1)/12)*rate_{t+1} when change month = x and x ≠ 1. Example: Japan 1997 change from 3 to 5 percent on 1 April (x=4) → adjusted 1997 rate = 4.5 percent.
- Data on final consumption excludes VAT revenue.

### Descriptive statistics (Table 1) — all numeric values preserved exactly
- Number of observations: 944 from 30 countries (for variables in Table 1).
- Growth rate of real GDP pc:
  - Mean 0.0226; Standard deviation 0.0314; Minimum -0.145; Maximum 0.251.
- Total taxes/GDP:
  - Mean 0.342; Standard deviation 0.0742; Minimum 0.108; Maximum 0.494.
- Consumption taxes/Total taxes:
  - Mean 0.336; Standard deviation 0.0872; Minimum 0.140; Maximum 0.686.
- Personal income taxes/Total taxes:
  - Mean 0.250; Standard deviation 0.104; Minimum 0.0496; Maximum 0.574.
- Corporate income taxes/Total taxes:
  - Mean 0.0864; Standard deviation 0.0459; Minimum 0.0167; Maximum 0.332.
- Property taxes/Total taxes:
  - Mean 0.0526; Standard deviation 0.0317; Minimum 0.00699; Maximum 0.221.
- Social security contribution/Total taxes:
  - Mean 0.275; Standard deviation 0.122; Minimum 0; Maximum 0.496.
- VAT/GDP:
  - Mean 0.0667; Standard deviation 0.0176; Minimum 0.0128; Maximum 0.110.
- VAT/Total taxes:
  - Mean 0.200; Standard deviation 0.0581; Minimum 0.0460; Maximum 0.451.
- Standard rate:
  - Mean 0.182; Standard deviation 0.0489; Minimum 0.0300; Maximum 0.270.
- C-efficiency:
  - Mean 0.563; Standard deviation 0.131; Minimum 0.323; Maximum 1.244.
- Consumption ratio:
  - Mean 2.106; Standard deviation 0.687; Minimum 1.060; Maximum 6.468.
- Investment rate:
  - Mean 0.238; Standard deviation 0.0456; Minimum 0.0982; Maximum 0.415.
- Employment growth:
  - Mean 0.00895; Standard deviation 0.0212; Minimum -0.136; Maximum 0.106.
- Additional dataset notes:
  - Average annual growth rate of real GDP per capita is 2.3 percent (reported as 0.0226 in Table 1).
  - Total taxes as share of GDP is 34.2 percent (0.342).
  - Share of VAT within total taxes is 20.0 percent (0.200).
  - Aggregate consumption taxes share is 33.6 percent (0.336).
  - Combined share of personal and corporate income taxes equals 34.0 percent of total taxes.
  - C-efficiency highest on average in New Zealand at 97.5 percent (sample period).
  - Maximum C-efficiency in sample: Luxembourg 124.4 percent.
  - Consumption ratio averaged at 211 percent (reported as 2.106 in Table 1).

### Time series properties
- Panel unit root tests:
  - Maddala and Wu (1999) test: rejects null of non-stationarity at 1% level for all variables.
  - Im et al. (2003) test: rejects null of non-stationarity at 1% level for all variables except log of VAT standard rate (cannot be tested via this method).
  - Residuals from estimated models (Table 3 illustrations) are stationary per these tests, supporting presence of long-run relations.

### Main empirical results — long-run growth effects (PMG estimates)
- Estimation approach:
  - PMG estimator used; Hausman test supports PMG (p-value > 0.05 in all cases).
  - Crisis dummy = 1 for 2008 to 2016 (inclusive), 0 otherwise.
  - Results robust to restricting sample to countries with at least 30 successive observations (not shown).
- A. Growth effects of VAT revenue as a whole (Table 2)
  - Column (1) financing taxes = Rest (all taxes apart from VAT):
    - Total taxes/GDP: -0.0858*** (t = -3.95)
    - VAT/Total taxes: 0.0768*** (t = 2.84)
    - Investment rate: 0.0508** (t = 2.54)
    - Employment growth: 0.444*** (t = 11.38)
    - EC coefficient (φ): -0.944*** (t = -21.42)
    - Crisis: -0.0172*** (t = -8.47)
    - Countries: 30; Observations: 981.
  - Column (2) financing taxes = Total income:
    - Total taxes/GDP: -0.0872*** (t = -3.98)
    - VAT/Total taxes: 0.0848*** (t = 3.18)
    - Investment rate: 0.0480** (t = 2.46)
    - Employment growth: 0.465*** (t = 12.43)
    - EC coefficient (φ): -0.964*** (t = -20.09)
    - Crisis: -0.0168*** (t = -7.77)
    - Countries: 30; Observations: 981.
    - Interpretation: coefficient of 0.085 indicates that a one percentage point increase in the share of the VAT in total taxes, offset by a fall in the share of total income taxes of the same size, is associated with a 0.085 percentage points increase in annual growth in the long run.
  - Column (3) financing taxes = Personal income:
    - Total taxes/GDP: -0.0848*** (t = -3.76)
    - VAT/Total taxes: 0.0731*** (t = 2.68)
    - Investment rate: 0.0388* (t = 1.90)
    - Employment growth: 0.463*** (t = 11.92)
    - EC coefficient (φ): -0.947*** (t = -21.51)
    - Crisis: -0.0171*** (t = -8.42)
    - Countries: 30; Observations: 981.
  - Column (4) financing taxes = Corporate income:
    - Total taxes/GDP: -0.0773*** (t = -3.57)
    - VAT/Total taxes: 0.107*** (t = 2.98)
    - Investment rate: 0.0594*** (t = 2.98)
    - Employment growth: 0.410*** (t = 10.14)
    - EC coefficient (φ): -0.940*** (t = -21.34)
    - Crisis: -0.0158*** (t = -7.79)
    - Countries: 30; Observations: 981.
  - Key findings from Table 2:
    - VAT is growth-promoting relative to all other remaining taxes combined.
    - VAT increase offset by a fall in income taxes (total, personal, or corporate) yields a positive and significant long-run growth effect.
    - Coefficient larger when offset by corporate income taxes (0.107) than by personal income taxes (0.0731), suggesting corporate income taxes may be more distortionary.
    - Total taxes/GDP coefficient negative throughout, implying increases in total taxes are growth-reducing in the long run (interpretation caveat: financing not specified).
    - Investment rate and employment growth long-run coefficients are significantly positive.
    - Crisis dummy is significantly negative across specifications.
- B. Growth effects of VAT components (Table 3)
  - Focus: VAT increase financed by a fall in total income taxes.
  - Column (1) (replication, fewer observations):
    - Total taxes/GDP: -0.0982*** (t = -3.91)
    - VAT/Total taxes: 0.0670** (t = 2.36)
    - Investment rate: 0.0634*** (t = 3.24)
    - Employment growth: 0.457*** (t = 11.73)
    - EC coefficient (φ): -0.958*** (t = -19.77)
    - Crisis: -0.0159*** (t = -7.64)
    - Countries: 30; Observations: 944.
  - Column (2) (omit standard rate; include ln(C-efficiency) and ln(consumption ratio)):
    - Total taxes/GDP: -0.0558 (t = -1.40)
    - VAT/Total taxes: -0.0166 (t = -0.53) — not significant
    - C-efficiency (log): 0.0347*** (t = 4.86)
    - Consumption ratio (log): 0.0209* (t = 1.79)
    - Investment rate: 0.0436** (t = 2.04)
    - Employment growth: 0.394*** (t = 10.23)
    - EC coefficient (φ): -0.911*** (t = -17.57)
    - Crisis: -0.0133*** (t = -6.08)
    - Interpretation: when the VAT is raised only through a rise in the standard rate (omitted), a rise in the VAT financed by a fall in income taxes does not promote growth (VAT coefficient not significant).
  - Column (3) (omit C-efficiency; include standard rate and consumption ratio):
    - Total taxes/GDP: -0.0447 (t = -1.12)
    - VAT/Total taxes: 0.187*** (t = 5.70)
    - Standard rate (log): -0.0277*** (t = -4.87)
    - Consumption ratio (log): -0.00823 (t = -0.72)
    - Investment rate: 0.0388* (t = 1.79)
    - Employment growth: 0.389*** (t = 10.16)
    - EC coefficient (φ): -0.909*** (t = -17.34)
    - Crisis: -0.0138*** (t = -6.25)
    - Interpretation: if VAT is raised only through a rise in C-efficiency (omitted here), the VAT share coefficient is positive and highly significant — raising VAT via C-efficiency is associated with higher long-run growth. Quantitative implication (not fully shown in excerpt): a one percentage point increase in the share of the VAT through C-efficiency increases annual GDP per capita growth (magnitude discussed in text; table coefficient on VAT/Total taxes = 0.187 when C-efficiency omitted).
  - Column (4) (alternate specification):
    - Total taxes/GDP: -0.0573* (t = -1.66)
    - VAT/Total taxes: 0.102* (t = 1.68)
    - Standard rate (log): -0.0175** (t = -2.01)
    - C-efficiency (log): 0.0153 (t = 1.53)
    - Investment rate: 0.0326 (t = 1.53)
    - Employment growth: 0.409*** (t = 10.97)
    - EC coefficient (φ): -0.924*** (t = -17.41)
    - Crisis: -0.0136*** (t = -6.03)
  - Key findings from Table 3:
    - The growth impact of a revenue-neutral shift toward VAT depends on how VAT revenue is raised.
    - A VAT increase driven by higher C-efficiency is associated with positive long-run growth effects (VAT coefficient positive and significant when C-efficiency is the driving component).
    - A VAT increase driven by a higher standard rate is not growth-promoting; the log standard rate has a negative and significant coefficient where included (e.g., -0.0277*** in Column (3)).
    - The coefficient on VAT/Total taxes is not significant when standard rate is omitted from the regressions (Column (2)), indicating the omitted standard rate matters for interpretation.
    - The long-run coefficient on the log of C-efficiency (θ_{i,E} − θ_{i,T} in reparameterization) captures the growth effect of a rise in C-efficiency offset by a fall in the standard rate, enabling direct comparison between raising VAT via C-efficiency versus via standard rate.

### Interpretation and policy-relevant conclusions (from reported results)
- Revenue-neutral tax reform that shifts the tax mix toward VAT and away from income taxes can be growth-promoting in the long run, but the effect depends critically on how VAT revenue is increased:
  - Raising VAT revenue via improvements in C-efficiency (broadening the base and improving collection) tends to be growth-enhancing.
  - Raising VAT revenue via increases in the statutory standard rate tends to be neutral or growth-reducing (standard rate coefficients negative when included).
- Corporate income taxes appear more distortionary than personal income taxes in the context of VAT-financed tax shifts (larger VAT coefficient when offset by corporate income tax cuts).
- Higher overall tax levels (Total taxes/GDP) are associated with lower long-run growth ceteris paribus.
- Investment rate and employment growth are robustly positively associated with long-run GDP per capita growth.
- The Global Financial Crisis period (2008–2016) is associated with a significant negative effect on growth across specifications.

*Italic source: IMF Working Paper (wpiea2019096) — content excerpt covering model, data, Table 1 descriptive statistics, Tables 2–3 long-run PMG estimation results, and related discussion.*

### 0.19 percentage points, 2.4 times larger than the effect when the form in which the VAT is

### wpiea2019096 - 0.19 percentage points, 2.4 times larger than the effect when the form in which the VAT is

### Key findings on VAT design and growth
- An increase in VAT revenue financed by a fall in income taxes promotes growth only when the VAT rise occurs through a rise in C-efficiency, but not when it occurs through a rise in the standard rate.
- A rise in C-efficiency, offset by a fall in the standard rate, also fosters growth, indicating C-efficiency is significantly more growth friendly than the standard rate.
- The coefficient of 0.035 means that a one percent increase in C-efficiency, offset by a fall in the standard rate, promotes growth by 0.035 percentage points.
- Table 4 (allowing for different channels) reports:
  - Total taxes/GDP: -0.0527 (Column 1) and up to -0.159*** (Column 4) with t-statistics shown in parentheses.
  - VAT/Total taxes: -0.0406 (Col 1), 0.227*** (Col 2), 0.126* (Col 3), -0.0918** (Col 4), 0.217*** (Col 5), 0.0274 (Col 6).
  - C-efficiency: 0.0461*** (Col 1), 0.0184* (Col 2), 0.0495*** (Col 3), 0.0332** (Col 4).
  - Standard rate: -0.0338*** (Col 1), -0.0289*** (Col 2), -0.0378*** (Col 3), -0.0201 (Col 4).
  - Consumption ratio: 0.0249* (Col 1), -0.00916 (Col 2), 0.000461 (Col 3), -0.0324** (Col 4).
  - Employment growth: 0.507*** (Col 1), 0.502*** (Col 2), 0.531*** (Col 3).
  - EC coefficient (휙): ranges from -0.930*** to -0.792*** across columns.
  - Crisis dummy: coefficients around -0.0159*** to -0.0130*** across columns.
- Observations and sample: Countries 30; Observations 944 in Table 4.

### Interpretations and mechanisms
- Exemptions narrow the VAT base and lower C-efficiency by breaking the chain of output tax and input credit, creating production taxation and distorting competition.
- Exemptions may incentivize self-supply, hamper contracting out, and reduce operational efficiency.
- Exemptions in healthcare and education raise costs for firms conducting research with hospitals and universities because exempted sectors cannot take credit for VAT on inputs; this may discourage research and lower total factor productivity (TFP).
- Unrecoverable VAT liabilities from intermediate inputs of exempted sectors are quantitatively nontrivial:
  - In 2011 in EU-27, unrecoverable VAT liabilities from intermediate inputs purchased by sectors producing exempted supplies accounted for 19% of all VAT liabilities.
  - Unrecoverable VAT on gross fixed capital formation expenditure of exempted sectors accounts for an additional 17% of total VAT liabilities.
- Rate differentiation (reduced rates and exemptions) increases administration costs and can distort consumer choices through relative price effects.
  - As an illustration, EC (2012) calculated an “effective” VAT rate in 2011 in EU-27 of 9 percent on average versus an average of standard rates of 20.7 percent.

### Channels of transmission (investment and TFP)
- When investment ratio is omitted (allowing VAT to affect growth through capital accumulation), coefficients on VAT and C-efficiency increase relative to specifications that control for investment, suggesting C-efficiency also promotes growth via the investment channel.
  - Example: VAT share coefficient in Column 2 of Table 4 is 0.23 versus 0.19 in Table 3; C-efficiency coefficient in Column 1 of Table 4 is 0.046 versus 0.035 in Table 3.
- Even when employment growth and investment are omitted, key results on C-efficiency remain essentially unchanged, reinforcing robustness of the C-efficiency effect across channels.

### Robustness tests and summarized results
- Robustness tests include: excluding 2008-2016 period; controlling for lagged GDP per capita; using ARDL model with 2 lags; omitting contemporaneous relation between fiscal variables and growth (to address reverse causality).
- Main robustness outcomes:
  - The result that VAT reallocations are generally more growth promoting than total income taxes persists across tests.
  - Increasing VAT via C-efficiency, offset by a fall in (total) income taxes, remains growth promoting in robustness checks.
  - A VAT increase through the standard rate or consumption ratio does not robustly foster growth across specifications; effects when present are smaller and less consistent.
  - The VAT-revenue-neutral increase in C-efficiency, offset by a fall in the standard rate, is confirmed to promote growth in multiple robustness checks.
- Specific robustness table highlights:
  - Table 5 (omitting 2008-16 and adding lagged GDP): VAT/Total taxes coefficients include 0.0825** (omitting 2008-16, Col 1) and 0.130*** (adding lagged GDP, Col 1). C-efficiency shows 0.0275*** (omitting 2008-16, Col 5) and 0.0213*** (adding lagged GDP, Col 5). Lagged real GDP per capita coefficients are -0.0230*** to -0.0268*** across columns where included, indicating convergence effects.
  - Table 6 (ARDL with 2 lags and omitting contemporaneous relation): Using ARDL 2 lags, VAT/Total taxes coefficients include 0.114*** (Col 1), 0.134*** (Col 2), 0.135*** (Col 3), 0.0822* (Col 4), 0.257*** (Col 5). C-efficiency 0.0348*** (upper sub-table Col 5). Omitting contemporaneous relation (1 lag), VAT/Total taxes coefficients include 0.0654** (Col 1) and 0.272*** (Col 6); C-efficiency 0.0238*** (Col 5).

### Policy implications and caveats
- Policy implication: Base-broadening measures that raise C-efficiency (e.g., removing reduced rates and exemptions) are key long-run drivers of growth-friendly VAT reform in advanced economies.
- Raising VAT via higher standard rates forgoes efficiency gains associated with base broadening and exemptions removal.
- Caveats:
  - The VAT decomposition method used does not capture behavioral responses of taxpayers to policy changes (e.g., changes in compliance induced by changing the standard rate).
  - C-efficiency interpretation has conceptual issues: increases in C-efficiency can result from policy changes that degrade VAT quality (e.g., introducing exemptions for intermediate goods or denying VAT refunds to exporters), which may not yield efficiency gains.
  - Results focus on long-run growth effects; distributional effects (income inequality) from VAT design are not analyzed here and are suggested as a useful extension.

### Data sources, tests, and sample details
- Data sources:
  - Real GDP per capita growth: Penn World Tables (PWT 9.0) and IMF WEO for 2015 and 2016.
  - Tax revenue data, including VAT at general government level: OECD Revenue Statistics Database.
  - VAT standard rates dataset assembled using OECD, European Commission, IMF’s Tax Policy Reform Database, CIAT (for Chile), and official country documents.
  - Final consumption and control variables (investment rate, employment growth): WEO.
- Panel unit root tests (Table 7):
  - Maddala and Wu (1999) and Im et al. (2003) reject the null of non-stationarity at the 1% level for all variables except VAT standard rate which cannot be tested by Im et al. (2003).
  - Examples of test statistics: GDP p.c. growth rate χ(60): 306.60 p-value: 0.00; t-bar: -14.44 p-value: 0.00.
- Cyclicality of tax compositions:
  - VAT/Total taxes: correlation with output gap not statistically significant (VAT: coef = .022; (robust) se = .0365; t = 0.60).
  - Income taxes/Total taxes: positively correlated with output gap (Income taxes: coef = .219; (robust) se = .043; t = 5.08).
- Sample sizes and observations cited:
  - Table 4: 30 countries, 944 observations.
  - Table 5 omitting 2008-16: 18 countries/543 observations (upper sub-table); adding lagged GDP: 30/981 or 30/944 depending on column.
  - Table 6 ARDL 2 lags: 14/573 or 13/522 depending on column; omitting contemporaneous relation: 30/981 or 30/944 depending on column.

*Source: wpiea2019096*

### REFERENCES

### REFERENCES

### VAT and Consumption Taxes
- Acosta-Ormaechea, S. S. Sola and J. Yoo, 2018, “Tax Composition and Growth: A Broad Cross-country Perspective”, German Economic Review, https://doi.org/10.1111/geer.12156.  
- Aizenman, J., and Y. Jinjarak, 2008, “The Collection Efficiency of the Value Added Tax: Theory and International Evidence,” Journal of International Trade & Economic Development, Vol. 17, No. 3, pp. 391-410.  
- Alm J., and A. El-Ganainy, 2013, “Value-added taxation and consumption,” International Tax and Public Finance, 20:1, pp. 105-128.  
- Crawford, I., Keen, M., and Smith, S., 2010, “Value Added Taxes and Excises”, pp. 275-362 in James Mirrlees and others (eds), Dimensions of Tax Design: The Mirrlees Review, Oxford: Oxford University Press for Institute for Fiscal Studies.  
- Cnossen, S., 2010, “Value Added Taxes and Excises: Commentary”, pp. 370-386 in James Mirrlees and others (eds), Dimensions of Tax Design: The Mirrlees Review, Oxford: Oxford University Press for Institute for Fiscal Studies.  
- De Mello, Luiz, 2009, “Avoiding the Value-added Tax: Theory and Cross-country Evidence,” Public Finance Review, Vol. 37, No. 1, pp. 27-46.  
- Ebrill, L., M., Keen, J-P Bodin, and V. Summers, 2001, The Modern VAT. International Monetary Fund, Washington DC.  
- Keen, M., 2013, “The Anatomy of the VAT.” National Tax Journal ̧ Vol. 66, pp. 423-446.  
- OECD, 2008, “Consumption Tax Trends 2008, VAT/GST and Excise Rates, Trends and Administration Issues”, OECD Publishing, Paris, https://doi.org/10.1787/ctt-2008-en.  
- OECD, 2018, “Consumption Tax Trends 2018, Trends and Policy Issues”, OECD Publishing, Paris, https://doi.org/10.1787/ctt-2018-en.  
- Ueda, J., 2017, “The evolution of potential VAT revenues and C-efficiency in advanced economies” IMF Working Paper 17/158.  
- European Commission (EC), 2011, “A retrospective evaluation of elements of the EU VAT system,” Unpublished. http://ec.europa.eu/taxation_customs/resources/documents/common/publications/studies/report_evaluation_vat.pdf  
- European Commission (EC), 2012, “A study on the economic effects of the current VAT rates structure,” Unpublished. https://ec.europa.eu/taxation_customs/sites/taxation/files/docs/body/vat_rates_structure_final_report.pdf

### Tax Policy, Growth, and Public Finance
- Amaglobeli, D., V. Crispolti, E. Dabla-Norris, P. Karnane, and F. Misch, 2018, “Tax Policy Measures in Advanced and Emerging Economies: A Novel Database” IMF Working Paper 18/110.  
- Arnold, J.M., Brys, B, Heady C., Johansson, Å., Schwellnus, C. and Vartia L., 2011, “Tax Policy for Economic Recovery and Growth.” Economic Journal, 121, pp. F59-F80.  
- Bleaney, M., Gemmell, N., and Kneller, R., 2001, “Testing the endogenous growth model: Public expenditure, taxation, and growth over the long run.” Canadian Journal of Economics, 34, 36-57.  
- Dabla-Norris, E., and F., Lima, 2018, “Macroeconomic Effects of Tax Changes: Evidence from Fiscal Consolidations,” IMF Working Paper 18/220.  
- Elgin, C., and S. Birnici, 2016, “Growth and Informality: A Comprehensive Panel Data Analysis,” Journal of Applied Economics, XIX:2, pp. 271-292.  
- Gemmell, N., Kneller, R., and Sanz, I., 2011, “The timing and persistence of fiscal policy impacts on growth: Evidence from OECD countries.” Economic Journal, 121, F33-F58.  
- Gemmell, N., Kneller, R., and Sanz, I., 2014, “The growth effects of tax rates in the OECD.” Canadian Journal of Economics, 47, 1217-1255.  
- Gemmell, N., Kneller, R., and Sanz, I., 2016, “Does the composition of government expenditure matter for long-run growth levels?” Oxford Bulletin of Economics and Statistics, 78, 522-547.  
- Kneller, R., Bleaney, M., and Gemmell, N., 1999, “Fiscal policy and growth: evidence from OECD countries.” Journal of Public Economics, 74, 171-190.  
- Kneller, R., and Misch, F., 2011, “What Does Ex-post Evidence Tell us About the Output Effects of Future Tax Reforms?” ZEW Discussion Papers, No. 11-029.  
- Morozumi, A. and Veiga, F.J., 2016, “Public spending and growth: The role of government accountability” European Economic Review 89, pp. 148-171.  
- Mirrlees, J., Adam, S., Besley, T., Blundell, R., Bond, S., Chote, R., Gammie, M., Johnson, P., Myles, G. and Poterba, J., 2011, Tax by Design: The Mirrlees Review, Oxford: Oxford University Press for Institute for Fiscal Studies.  
- Summers, L.H., 2014, “U.S. Economic Prospects: Secular Stagnation, Hysteresis, and the Zero Lower Bound” Business Economics, 49:2.  
- Xing, J., 2012, “Tax structure and growth: How robust is the empirical evidence?” Economics Letters, 117:1, pp. 379-382.  
- IMF, 2011, “Revenue mobilization in developing countries,” Policy Paper Series, International Monetary Fund, Washington, D.C.

### Informality, Underground Economy, and Corruption
- Elgin, C., and S. Birnici, 2016, “Growth and Informality: A Comprehensive Panel Data Analysis,” Journal of Applied Economics, XIX:2, pp. 271-292.  
- Schneider, F, 2005, “Shadow economies around the world: what do we really know?” European Journal of Political Economy, 21, pp. 598-642.  
- Tanzi, V, 1999, “Uses and Abuses of Estimates of the Underground Economy,” Economic Journal, 109, pp. F338-F347.  
- Tanzi, V., and H.R. Davoodi, 2000, “Corruption, Growth, and Public Finances,” IMF Working Paper 00/182.  

### Empirical Methods, Econometrics, and Data Sources
- Feenstra, R.C., R. Inklaar, and M.P. Timmer, 2015, “The Next Generation of the Penn World Table” American Economic Review, 105:10, pp. 3150-3182.  
- Im, K. S., Pesaran, M. H., and Shin, Y., 2003, “Testing for unit roots in heterogeneous panels”, Journal of Econometrics, 115:1, pp. 53-74.  
- Maddala, G. S., and Wu, S., 1999, “A Comparative Study of Unit Root Tests with Panel Data and a New Simple Test”, Oxford Bulletin of Economics and Statistics, 61:0, pp. 631-52.  
- Pesaran, M.H., Y. Shin, and R.P. Smith, 1999, “Pooled mean group estimation of dynamic heterogeneous panels” Journal of the American Statistical Association 94:446, pp. 621-634.  
- Pesaran, M.H., and R.P. Smith, 1995, “Estimating long-run relationships from dynamic heterogeneous panels” Journal of Econometrics 68, pp. 79-113.  
- Riera-Crichton, D., C. Vegh and G. Vuletin, 2016, “Tax multipliers: Pitfalls in measurement and identification,” Journal of Monetary Economics, 79, pp. 30-48.  
- Sancak, C., J. Xing and R. Velloso, 2010, “Tax Revenue Response to the Business Cycles,” IMF Working Paper 10/71.

### Other Relevant Contributions
- Summers, L.H., 2014, “U.S. Economic Prospects: Secular Stagnation, Hysteresis, and the Zero Lower Bound” Business Economics, 49:2.  
- Declarations and commentary pieces embedded in edited volumes and policy series cited above.

*Content extracted from the PDF file "wpiea2019096 - REFERENCES".*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019096.pdf_