## 1. The VAT and Net Exports

## Source details

**Canonical URL:** [1. The VAT and Net Exports](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0647.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0647.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0647.pdf.json)

---

### I. Research question and context
- Using panel data for 27 OECD member countries over 1967–2003 (573 observations), the paper asks: do countries that rely more on value added taxes (VAT), and/or less on corporate taxation, tend to have higher (or lower) net exports?
- Motivation:
  - Decline of explicit taxes on international trade raises interest in how domestic taxes affect international trade.
  - Prior findings: Desai and Hines (2005) report a negative relation between VAT share and export/trade intensity; Slemrod (2004) finds a positive association between corporate tax revenues/GDP and trade intensity.
- Empirical focus: net exports (trade balance) rather than export intensity, because corporate tax effects can operate via the capital account and require dynamic (intertemporal) analysis.

### II. Theoretical framework and main predictions
- Model setup (small open, two-period economy; single produced good; representative consumer; all nominal prices and exchange rate normalized to unity).
- Four tax types defined exactly as in the source:
  - Destination-based ad valorem consumption tax (VAT) at rate TV (tax-exclusive).
  - Origin-based ad valorem consumption tax at rate TO (tax-inclusive).
  - Source-based tax on return to home investment at rate TS.
  - Residence-based tax on all home resident savings at rate TR (tax-inclusive).
- Key theoretical identities (as presented in the source):
  - Present-value net exports identity: E1 + E2/(1 + r) = 0.
  - Arbitrage condition: r = (1 − TS) TR.
- Main analytical results:
  - Destination-based and origin-based consumption taxes (TV and TO) cancel from the necessary conditions and have no effect on the real equilibrium and on the level of exports in any period under the model assumptions (irrelevance result), subject to qualifications.
  - Conditions and qualifications to irrelevance:
    - Rates must be constant over time; anticipated changes affect timing of consumption and net exports.
    - Single-good assumption masks heterogeneous effective tax rates across commodities (e.g., nontradables exempted or lightly taxed) which can reduce tradable sector size.
    - Imperfect refunding of VAT on inputs used by exporters can make VAT act partly as an export tax.
  - Corporate taxation:
    - Source-based tax TS raises the domestic marginal condition: F'(K) = r/(1 − TS) (as in the source).
    - Source-based corporate tax reduces domestic investment and raises first-period net exports (capital outflow), with later income inflows reducing second-period net exports; present-value constraint can imply eventual convergence to zero net-export effect.
    - Residence-based tax TR leaves domestic investment unchanged in the model and produces opposite short-run signs compared with a source tax.
  - Practical considerations:
    - Many corporate tax systems have de facto source-based features (e.g., deferral of repatriated dividends, limited controlled foreign corporation rules).
    - Taxation of rents depends on ownership and mobility; effects differ if rents are owned domestically or by nonresidents.
    - Model limitations: overlapping generations frameworks alter present-value constraints; non–lump-sum use of tax revenue (public spending) can affect export performance; sluggish real investment adjustment can delay effects.

### III. Data and descriptive statistics
- Sample details:
  - Unbalanced panel of 27 OECD countries with a VAT, covering 1967–2003, each country observed from VAT introduction onward.
  - Luxembourg and Mexico excluded for lack of tax revenue data; United States excluded because it does not have a VAT.
  - Total observations: 573.
- Key descriptive statistics (Table 5 preserved exactly):
  - Net exports as a fraction of GDP: Number of Observations 573; Mean 0.005; Mean Standard Deviation 0.047.
  - Exports as a fraction of GDP: 573; Mean 0.355; Mean Standard Deviation 0.158.
  - Log GDP per capita: 573; Mean 10.767; Mean Standard Deviation 2.436.
  - Total tax revenue, as a fraction of GDP: 573; Mean 0.372; Mean Standard Deviation 0.077.
  - VAT revenue, as a fraction of GDP: 573; Mean 0.066; Mean Standard Deviation 0.020.
  - VAT revenue, as a fraction of total consumption: 573; Mean 0.082; Mean Standard Deviation 0.036.
  - Standard VAT rate: 461; Mean 17.500; Mean Standard Deviation 5.373.
  - Corporate tax revenue, as a fraction of GDP: 573; Mean 0.026; Mean Standard Deviation 0.013.
  - Statutory corporate tax rate: 334; Mean 0.389; Mean Standard Deviation 0.120.

### IV. Empirical specification and identification
- Baseline dynamic specification (allowing up to two lags):
  - NX_it = sum_{k=0 to 2} β_k VAT_{i,t−k} + sum_{k=0 to 2} γ_k CIT_{i,t−k} + sum_{k=0 to 2} φ_k TAX_{i,t−k} + θ' X_it + α_i + μ_t + ε_it
  - Definitions preserved: NX = net exports of goods and services (relative to GDP); CIT, VAT = corporate tax and VAT variables; TAX = sum of all tax revenues (relative to GDP); X = conditioning variables (including GDP per capita and geographic controls); α_i = country fixed effects; μ_t = time effects.
- Long-run effects (as given in the source, equation form preserved):
  - ∆_CIT = (γ_0 + γ_1 + γ_2) / (1 − φ)
  - ∆_VAT = (β_0 + β_1 + β_2) / (1 − φ)
- Estimation issues and approach:
  - VAT and CIT variables likely endogenous (share denominators, simultaneity).
  - Authors use IV/GMM (lagged regressors as instruments) and report specifications with statutory rates (smaller sample) as robustness.
  - Dynamic estimates obtained via first-differenced GMM (Arellano-Bond style), including time dummies; diagnostics reported include Sargan (Hansen’s J), m1 and m2 tests for serial correlation; heteroskedasticity-robust standard errors used.

### V. Static empirical results (Table 1 and Table 2, key estimates preserved exactly)
- Table 1 — The VAT and Net Exports (selected coefficients):
  - Column 1 (Year effects only): VAT reliance (VAT/GDP)t = 0.070 (0.078). R-squared 0.09. Observations 573.
  - Column 2 (Year effects, GDP and geographic controls): VAT reliance (VAT/GDP)t = 0.106 (0.092). R-squared 0.11. Observations 573.
  - Column 3 (Year, country effects, GDP and geographic controls): VAT reliance (VAT/GDP)t = -0.575** (0.245). R-squared 0.62. Observations 573.
  - Column 4 (adds TAX/GDP): VAT reliance (VAT/GDP)t = -0.544** (0.245); (TAX/GDP)t = -0.194*** (0.067). R-squared 0.63. Observations 573.
- Table 2 — Corporate Taxes and Net Exports (selected coefficients):
  - Column 1: (CIT/GDP)t = 1.182*** (0.183). R-squared 0.17. Observations 573.
  - Column 2: (CIT/GDP)t = 1.219*** (0.180). R-squared 0.20. Observations 573.
  - Column 3: (CIT/GDP)t = 1.291*** (0.167). R-squared 0.66. Observations 573.
  - Column 4: (CIT/GDP)t = 1.479*** (0.169). R-squared 0.68. Observations 573.
  - Column 5 (CIT and VAT together): (CIT/GDP)t = 1.441*** (0.172); (VAT/GDP)t = -0.300 (0.215). R-squared 0.68. Observations 573.
  - Column 6 (instrumenting current tax variables with their first lags): (CIT/GDP)t = 1.270*** (0.244); (VAT/GDP)t = -0.182 (0.327). R-squared 0.68. Observations 546.
- Static interpretation:
  - VAT reliance insignificant in simple specifications but becomes significantly negative with country fixed effects.
  - CIT/GDP consistently positive and highly significant across specifications; when both taxes included, VAT becomes insignificant, suggesting VAT may proxy for omitted corporate tax effects in some specifications.
  - Authors note: "A one percentage point increase in corporate tax revenue relative to GDP, compensated by increases in other taxes, increases net exports by rather more than one percentage point." (exact phrasing preserved from the source).

### VI. Dynamic empirical results (GMM first-differenced estimates; main findings preserved exactly)
- Dynamic specification estimated by GMM after first-differencing; includes lagged dependent variable and first differences of VAT, CIT and TAX up to two lags; time dummies included.
- Representative dynamic coefficient patterns (selected entries preserved exactly):
  - NX t-1: 0.732*** (0.090); 0.404*** (0.066); 0.715*** (0.077); 0.661*** (0.103); 0.781*** (0.045); 0.772*** (0.047).
  - (CIT variable) t: 0.831** (0.397); 1.307** (0.526); 1.000** (0.507); 1.006* (0.538); 0.057** (0.029); 0.059** (0.029).
  - (CIT variable) t-1: -0.838** (0.410); -0.757* (0.457); -0.935* (0.489); -0.955** (0.453); -0.066** (0.032); -0.061* (0.033).
  - (VAT variable) t: -1.371*** (0.521); 0.001 (0.001); 0.008 (0.335); -0.049 (0.325).
  - (VAT variable) t-1: 0.974** (0.427); -0.003** (0.001); 0.105 (0.290); 0.029 (0.254).
  - (TAX/GDP) t: -0.197** (0.100); -0.499*** (0.188); -0.335*** (0.147); -0.376** (0.163); -0.261*** (0.086); -0.282*** (0.108).
  - (TAX/GDP) t-1: 0.256** (0.110); 0.173 (0.126); 0.352** (0.124); 0.347*** (0.116); 0.204*** (0.061); 0.209*** (0.074).
- Diagnostics (selected p-values preserved exactly):
  - First-order serial correlation (m1): 0.006, 0.022, 0.007, 0.011, 0.004, 0.003 (across Table 3 columns).
  - Second-order serial correlation (m2): 0.490, 0.167, 0.402, 0.425, 0.201, 0.196 (Table 3 columns).
  - Sargan: 1.000 across reported specifications.
- Dynamic interpretation and exact numeric implications:
  - Short-run effect of corporate tax increases: current-year coefficient example (Table 3, column 1): (CIT variable) t = 0.831**; (CIT variable) t-1 = -0.838** — consistent with a short-run increase in net exports followed by a reversal.
  - Point estimate implication stated in the source: "a one point increase in reliance on the corporate tax is associated with an increase in net exports of 0.83 percent of GDP in the first year." (exact wording preserved).
  - Medium- to long-run effect: the short-run increase turns into a persistent reduction after one period and then converges to zero in the long run. After 10 years, net exports are a little under 0.02 percentage points lower than prior to the tax increase. The null that 0 = ∆CIT cannot be rejected; the p-value on this hypothesis test is 0.91 (exact numbers preserved).
  - VAT dynamics: an increase in VAT reliance reduces net exports sharply in the short run but the effect is essentially reversed in the second year; the null of no long-run VAT effect cannot be rejected (p-value 0.74). Using alternative VAT measures (standard VAT rate; VAT revenue as a fraction of total consumption) the short-run effects vanish.

### VII. Robustness and alternative specifications (selected exact entries)
- Robustness table (Table 8; selected entries preserved exactly):
  - LDV t-1: 0.862*** (0.064); 0.915*** (0.036); 0.729*** (0.092).
  - (CIT variable) t: 0.510** (0.223); 0.067** (0.028); 0.278* (0.159).
  - (CIT variable) t-1: -0.602** (0.299); -0.075*** (0.016); -0.268* (0.150).
  - (TAX/GDP) t: -0.276*** (0.078); -0.248** (0.090); -0.397** (0.156).
  - Observations in robustness specifications: 492; 298; 492.
- Diagnostics in robustness specifications (selected p-values preserved exactly):
  - First-order s.c.: 0.003, 0.004, 0.007.
  - Second-order s.c.: 0.203, 0.207, 0.225.
  - Sargan: 1.000.

### VIII. Interpretation, caveats, and main conclusions (exact interpretations preserved)
- VAT:
  - Evidence supports VAT’s inherent trade neutrality: short-run deterioration in net exports associated with higher VAT reliance likely reflects confounding consumption shocks rather than a causal VAT trade effect.
  - Alternative VAT measures (standard rate; VAT revenue/consumption) eliminate the short-run effect.
- Corporate tax:
  - Empirical behavior consistent with an essentially source-based tax in practice: short-run increase in net exports (capital outflow), followed by income inflows reducing net exports, with eventual convergence to zero long-run effect.
  - Results robust to using statutory corporate tax rate and to instrumenting tax variables; marginal effective tax rate (METR) tried and proved insignificant (not shown).
- Other domestic tax measures:
  - Total tax revenue/GDP typically shows an initial increase associated with reduced net exports, followed by reversal and no long-run effect; similar patterns when replacing tax ratio by general government expenditure/GDP (not shown).
- Summary conclusions (section V, preserved exactly):
  - Two consistent patterns over the last 35 years in OECD experience:
    - VAT appears to have no impact on the trade balance in either the short or the long run; observed dynamics can be explained by unrelated consumption shocks, supporting VAT’s inherent trade neutrality.
    - Corporate tax changes have powerful dynamic effects consistent with source-based taxation in practice: short-run increase in net exports due to capital outflow, subsequent reduction from income inflows, and eventual convergence to zero long-run effect.
  - Overall implication: some aspects of domestic tax policy have strong, complex, and temporally varying effects on trade performance, though long-run net export effects from VAT and corporate tax changes converge to zero in these specifications.

*Source: _wp0647 — IMF working paper excerpt provided in the content unit*

### 1. The VAT and Net Exports .............................................................................................

### 1. The VAT and Net Exports

### Major Sections in the Content Unit
- 1. The VAT and Net Exports ...................................................................................................16
- 2. Corporate Taxes and Net Exports ........................................................................................17
- 3. Dynamic Specifications .......................................................................................................19

### Figures Included
- Figure 1. Trade Balance Against VAT/GDP.......................................................................................12
- Figure 2. Trade Balance Against CIT/GDP ........................................................................................12
- Figure 3. Trade Balance and VAT Reliance, Unweighted Yearly Average .......................................13
- Figure 4. Trade Balance and Corporate Tax Reliance, Unweighted Yearly Average ........................13

### Appendix and Tables
- Appendix: Data ..........................................................................................................................................23
- Appendix Tables:
  - 4. Country (Year) Coverage of Sample ...................................................................................23
  - 5. Descriptive Statistics............................................................................................................24
  - 6. Distribution of Countries Over Years ..................................................................................25
  - 7. Balance of Panel ..................................................................................................................25
  - 8. Robustness ...........................................................................................................................26

*Source: _wp0647 - 1. The VAT and Net Exports*

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

### _wp0647 - References

### I. Introduction
- Context:
  - Decline of explicit taxes on international trade has increased attention on how domestic taxes affect international trade.
  - Historical debates: EU internal customs elimination and differing tax structures (Sinn, 1990); U.S. debates over VAT remission on exports and corporate tax border adjustments (Hartman, 2004) and WTO disputes (Foreign Sales Corporation, DISC, EITI).
- Research question:
  - Using panel data for 27 OECD member countries over 1967–2003, the paper asks: do countries that rely more on value added taxes, and/or less on corporate taxation, tend to have higher (or lower) net exports?
  - Focus is on net exports rather than export or trade intensity because corporate tax effects may operate via the capital account and require dynamic (intertemporal) analysis.
- Relation to prior work:
  - Desai and Hines (2005): for high-income countries, a VAT dummy has no effect on export or trade intensity with fixed effects; VAT share in total tax revenue in 2000 is significantly and negatively related to export and trade intensity.
  - Slemrod (2004): finds a significant positive association between corporate tax revenues relative to GDP and trade intensity (about 100 countries observed in four years).
- Plan:
  - Section II: theoretical framework
  - Section III: sample properties
  - Section IV: empirical strategy and results
  - Section V: conclusion
  - Appendix: dataset, variables, robustness

### II. Tax Structure and Net Exports: Analytics
- Framework:
  - Small, open, two-period economy with a representative consumer, single produced good (used for consumption or investment), all nominal prices and exchange rate normalized to unity.
  - Period 1 endowment Y, borrowing b from rest of world, domestic investment K, lending abroad B.
  - Period 2 production from investment is F(K) (strictly concave), gross domestic interest R, foreign gross interest r.
  - Four tax types introduced:
    - Destination-based ad valorem consumption tax (VAT) at rate TV (tax-exclusive).
    - Origin-based ad valorem consumption tax at rate TO (tax-inclusive).
    - Source-based tax on return to home investment at rate TS.
    - Residence-based tax on all home resident savings at rate TR (tax-inclusive).
  - Government returns period-i tax revenue to consumer as lump-sum Ti.
- Key equilibrium and identities (as presented in text):
  - Period 1 budget (equation (1)): 101 ))(1()1(TBbKYTCT V +−+−−=+  where T1 = (1−TV) + (1−TO) presented in the source.
  - Period 2 budget (equation (2)): 202 ))1(1()1())()1)(1()(1()1(TBTrbRKFTTKTCT RRSV +−+++−−−+−=+  with T2 defined in text.
  - Arbitrage condition (equation (3)): r = (1 − TS) TR
  - Present-value consumption identity (equation (4)): C1 + C2/(1 + r) ≡ Y* + [r/(1+r)](K + F(K))
  - Net exports definitions (equations (5) and (6)):
    - Period 1: E1 ≡ Y − C1 = −B + b (text expresses net exports as b = Y − C1)
    - Period 2: E2 ≡ F(K) − C2 + (1 − TS)(K) + (1 − TS)B + ... (as given)
  - Present-value of net exports (equation (7)): E1 + E2/(1 + r) = 0

A. Simple model notes
- Consumer chooses K and B to maximize intertemporal utility U(C1, C2) subject to the tax-augmented budgets and r taken as given (small country).
- First-order conditions (equations (9) and (10)) characterize optimal B and K with derivatives of U and F.

B. Trade and indirect taxation
- Main theoretical result: destination-based and origin-based consumption taxes (TV and TO) cancel from the necessary conditions and have no effect on the real equilibrium and on the level of exports in any period under the model assumptions (irrelevance result).
- Conditions and qualifications for irrelevance:
  - Rates must be constant over time. A fully anticipated increase in VAT is akin to an increase in residence-based tax and affects timing of consumption and net exports.
  - Single consumption good assumption masks heterogeneity of effective tax rates across commodities (e.g., nontradables often exempted or lightly taxed), which can reduce the tradable sector and export intensity (Feldstein and Krugman, 1990).
  - Imperfect refunding of VAT on inputs used by exporters (administrative difficulties) can make VAT act partly as an export tax, reducing tradable sector size and export intensity (Desai and Hines, 2005).
  - Origin-based consumption tax is distinct from source-based corporate tax because origin tax applies to final consumption goods only, while source-based corporate tax applies to all output and distorts production decisions.

C. Trade and corporate taxation
- From first-order conditions, domestic marginal condition implies (equation (11)): F'(K) = r/(1 − TS) (as presented in the source).
- Effects of source-based tax TS (with TR = 0):
  - Present value of consumption falls as TS increases (equation (13) and discussion).
  - Both C1 and C2 fall; fall in C1 implies first period net exports rise (greater trade surplus or smaller deficit) and second period net exports fall (present-value constraint).
  - Intuition: source-based corporate tax reduces domestic investment, increasing capital exports in period 1 and inducing a trade surplus; later income from abroad raises second-period resources, reversing net exports.
- Effects of residence-based tax TR (with TS = 0):
  - Domestic investment K unchanged; present value of lifetime consumption Y* unchanged.
  - C1 increases (given concave utility assumption), so net exports fall in first period and rise in second—opposite sign pattern to source tax.
  - Intuition: residence tax lowers net return to saving for residents, reducing savings and investment abroad in period 1, lowering the trade surplus; reverses in period 2.
- Practical considerations:
  - Many corporate tax systems have de facto source-based features (e.g., deferral of repatriated dividends, controlled foreign corporation rules limited in scope).
  - Corporate taxes often affect marginal and intra-marginal returns; taxation of rents (fixed factor returns) has different effects depending on ownership and mobility:
    - If rents are owned domestically and immobile, taxing them has no real-investment effect; if owned by nonresidents, taxing rents can increase present-value consumption and affect net exports.
    - If fixed factor is internationally mobile or average effective tax rates influence location decisions, corporate tax increases can push investment abroad, similar to source-tax effects.
  - Model limitations:
    - Overlapping generations frameworks invalidate the present-value net-export zero condition but preserve intertemporal optimization logic.
    - Short-run vs long-run interpretation: first period captures brief short-run effects; sluggish real investment adjustment can delay responses to source taxes.
    - Assumption that tax revenue is returned lump-sum is strong; if not, changes in public expenditure paths can affect export performance (e.g., government spending on non-tradables).

### III. The Data: A First Look
- Sample:
  - Unbalanced panel of 27 OECD countries.
  - Estimation period: 1967 to 2003.
  - Total number of observations: 573.
- Preliminary correlations and visual evidence:
  - Scatter plots (Figures 1 and 2) show:
    - Correlation coefficient between trade balance (net exports/GDP) and VAT reliance: positive but marginally significant at 0.07.
    - Correlation coefficient between trade balance and corporate tax reliance: positive and highly significant at 0.35.
  - Time-series unweighted annual averages (Figures 3 and 4) suggest a positive association between net exports and both VAT and corporate tax reliance, though correlations are not significant.
- Caveat:
  - Simple correlations do not control for other factors, potential biases, or dynamics—necessitating a closer econometric analysis.

### IV. Empirical Analysis
A. Econometric specification and issues
- Estimated basic equation (for country i at time t), allowing up to two lags:
  - NX_it = sum_{k=0 to 2} β_k VAT_{i,t−k} + sum_{k=0 to 2} γ_k CIT_{i,t−k} + sum_{k=0 to 2} φ_k TAX_{i,t−k} + θ' X_it + α_i + μ_t + ε_it
  - Where:
    - NX: net exports of goods and services (relative to GDP)
    - CIT, VAT: corporate tax and VAT variables (several variants used)
    - TAX: sum of all tax revenues (relative to GDP)
    - X: conditioning variables (including GDP per capita and geographic controls described in Appendix)
    - α_i: country fixed effects; μ_t: time effects; ε_it: idiosyncratic error
- Interpretation:
  - Short-run impacts: coefficients on current VAT and CIT (β_0, γ_0).
  - Long-run effects (with up to two lags) defined as:
    - ∆_CIT = γ_0 + γ_1 + γ_2 − φ_1 γ_0 − ... (presentation summarized in source text as equations (15)): specifically
      - ∆_CIT = (γ_0 + γ_1 + γ_2) / (1 − φ)
      - ∆_VAT = (β_0 + β_1 + β_2) / (1 − φ)
      - (equations preserved as in source: φ γ γ γ − + + =∆ 1 210 CIT and φ β β β − + + =∆ 1 210 VAT)
  - Theoretical expectations:
    - If corporate tax approximates source-based tax: short-run increase in net exports (γ_0 > 0), possible subsequent reduction and sign reversal in γ_1 and/or γ_2, with an ultimate permanent reduction in net exports; possible hypothesis of no long-run effect (∆_CIT = 0) if present-value net exports sum to zero.
    - If corporate tax approximates residence-based tax: opposite sign pattern.
    - For an idealized single-rate VAT, theory predicts no effect in short or long run (β_k = 0 ∀k), though real-world deviations weaken this.
- Endogeneity and estimation approach:
  - VAT and CIT variables likely endogenous (share denominators, simultaneity).
  - Authors also report results using basic statutory rates (conceptually imperfect, smaller sample).
  - Preferred strategy: IV/GMM (generalized method of moments) using suitably lagged regressors as instruments to control for country-specific effects and lagged endogenous variables.
  - Identification requires absence of higher order serial correlation in residuals and valid instruments (correlated with endogenous regressors, uncorrelated with errors).
- In all specifications:
  - Year dummies included to control for common time-specific effects.
  - Standard errors are heteroskedasticity-robust.

B. Results (overview up to text cutoff)
- The paper reports a range of empirical equations estimated along the described lines.
- Static results introduction appears next in the source (text cuts off before static estimates are presented).

*Source: _wp0647 - References (excerpt provided) — IMF working paper content as supplied.*

### introduction, we begin by presenting results using a static specification, constraining the

### _wp0647 - introduction, we begin by presenting results using a static specification, constraining the

### Static specification and approach
- The static specification constrains the coefficients on all the lagged variables in equation (14) to be zero, relating the trade balance to only the current values of the corporate tax and VAT reliance variables.
- Estimation results are summarized in Tables 1 and 2 using specifications that progressively add controls: year effects; GDP controls and geographic controls; and country-specific fixed effects.

### Table 1 — The VAT and Net Exports
- Column 1 (Year effects only): VAT reliance (VAT/GDP)t coefficient reported as 0.070; standard error (0.078). R-squared 0.09. Observations 573.
- Column 2 (Year effects, GDP controls and geographic controls): VAT reliance (VAT/GDP)t coefficient reported as 0.106; standard error (0.092). R-squared 0.11. Observations 573.
- Column 3 (Year effects, country effects, GDP controls and geographic controls): VAT reliance (VAT/GDP)t coefficient reported as -0.575**; standard error (0.245). R-squared 0.62. Observations 573.
- Column 4 adds overall tax revenue (TAX/GDP)t as a regressor:
  - VAT reliance (VAT/GDP)t coefficient reported as -0.544**; standard error (0.245).
  - (TAX/GDP)t coefficient reported as -0.194***; standard error (0.067).
  - R-squared 0.63. Observations 573.
- Notes on Table 1:
  - Robust standard errors in parentheses.
  - * means significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.
  - Geographic controls include area of the country and dummies for landlocked and island economies.

- Interpretation provided in text:
  - VAT reliance is insignificant in simple specifications (Cols 1–2) but becomes significantly negative with country-specific fixed effects (Col 3).
  - When overall tax revenue is included (Col 4), TAX/GDP enters significantly negative while VAT reliance remains significantly negative and of similar magnitude to Col 3.
  - The VAT result may reflect omitted-variable bias in simpler specifications, and the VAT pattern resembles results in Desai and Hines (2005) for gross exports.

### Table 2 — Corporate Taxes and Net Exports
- Columns 1–4 replicate the Table 1 specifications but replace VAT with corporate tax reliance (CIT/GDP)t.
- Coefficients on (CIT/GDP)t:
  - Column 1: 1.182***; standard error (0.183). R-squared 0.17. Observations 573.
  - Column 2: 1.219***; standard error (0.180). R-squared 0.20. Observations 573.
  - Column 3: 1.291***; standard error (0.167). R-squared 0.66. Observations 573.
  - Column 4: 1.479***; standard error (0.169). R-squared 0.68. Observations 573.
- Column 5 (including VAT/GDP alongside CIT/GDP):
  - (CIT/GDP)t coefficient: 1.441***; standard error (0.172).
  - (VAT/GDP)t coefficient: -0.300; standard error (0.215) — reported as insignificant in the presence of corporate tax reliance.
  - R-squared 0.68. Observations 573.
- Column 6 (instrumenting current tax variables with their first lags):
  - (CIT/GDP)t coefficient: 1.270***; standard error (0.244).
  - (VAT/GDP)t coefficient: -0.182; standard error (0.327).
  - (TAX/GDP)t coefficient(s) reported in other columns as negative: -0.317***, -0.309***, -0.233*** (standard errors (0.064), (0.064), (0.077) respectively).
  - R-squared 0.68. Observations 546.
- Notes on Table 2:
  - Robust standard errors in parentheses.
  - * means significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.
  - Column 6 uses first lags of VAT, corporate tax and total tax variables to instrument their current values.
- Interpretation provided in text:
  - There is a significant, robust, large, and positive association between corporate taxes and export performance across specifications.
  - Broad statement: "A one percentage point increase in corporate tax revenue relative to GDP, compensated by increases in other taxes, increases net exports by rather more than one percentage point."
  - When both taxes are included, VAT reliance is insignificant, suggesting VAT may have been proxying for omitted corporate tax effects in Table 1.
  - Instrumenting tax variables using first lags (Col 6) does not change the conclusion: export performance is unrelated to reliance on VAT, but positively related to reliance on corporate taxes.

### Dynamic results (GMM, first-differenced specification)
- Main interest is the dynamic specification (14), estimated by GMM using lagged values of net exports and regressors as instruments, after first-differencing to eliminate country fixed effects and including time dummies.
- The estimated model relates first-differenced trade balance in period t to:
  - its own lag;
  - first-differenced measures of VAT and CIT in periods t, (t -1) and (t -2);
  - differences in controls (time dummies and log GDP per capita included; time-invariant geographic controls drop out).
- Diagnostics reported include:
  - p-value of the Sargan Statistic (Hansen’s J statistic) for overidentifying restrictions.
  - p-values of the m1 test for first-order serial correlation (expected to be present).
  - p-values of the m2 test (Arellano and Bond) for second-order serial correlation (desired to be rejected).
- Column 1 (dynamic specification using VAT and corporate tax reliance):
  - The coefficient on the lagged dependent variable is significantly positive; diagnostics satisfactory.
  - Corporate tax dynamics:
    - Coefficient on current value of corporate tax reliance is significantly positive.
    - Coefficients on the first lag and second lag are negative (second lag individually insignificant).
    - The pattern of sign reversal is consistent with theoretical predictions for a source-based corporate tax.
    - Point estimate implication: "a one point increase in reliance on the corporate tax is associated with an increase in net exports of 0.83 percent of GDP in the first year."
    - In the following year, net exports decline by 0.84 percent of GDP (text breaks off after this figure).

*Italic source: Excerpt from PDF chapter _wp0647 - introduction, we begin by presenting results using a static specification, constraining the*

### 0.23 points lower than prior to the

### _wp0647 - 0.23 points lower than prior to the

### Main empirical findings on tax changes and net exports
- Short-run effect of corporate tax increases:
  - Increased corporate taxation is associated with increased net exports in the short run (consistent with capital flowing abroad).
  - Example coefficient evidence (Table 3, column 1): (CIT variable) t = 0.831**; (CIT variable) t-1 = -0.838**.
- Medium- to long-run effect of corporate tax increases:
  - The short-run increase in net exports turns into a persistent reduction after one period and then converges to zero in the long run.
  - After 10 years, net exports are a little under 0.02 percentage points lower than prior to the tax increase.
  - The null that 0 = ∆CIT cannot be rejected; the p-value on this hypothesis test is 0.91.
- VAT reliance effects:
  - An increase in VAT reliance reduces net exports sharply in the short run, but the effect is essentially reversed in the second year.
  - The null hypothesis of no long-run VAT effect cannot be rejected (p-value in this case 0.74).
  - Using alternative VAT measures (standard VAT rate; VAT revenue as a fraction of total consumption) the short-run effects found for VAT reliance vanish—no convincing evidence of short- or long-run VAT effect when using these proxies.
- Interpretation:
  - Results are consistent with VAT being inherently trade neutral.
  - Corporate tax behavior is consistent with an essentially source-based tax effect in practice: short-run capital outflows raise net exports initially, followed by income inflows that reduce net exports, with eventual convergence to zero.

### Key regression and robustness results (selected values preserved exactly as reported)
- Dynamic coefficient estimates (from Table 3; selected entries and standard errors in parentheses):
  - NX t-1: 0.732*** (0.090); 0.404*** (0.066); 0.715*** (0.077); 0.661*** (0.103); 0.781*** (0.045); 0.772*** (0.047).
  - (CIT variable) t: 0.831** (0.397); 1.307** (0.526); 1.000** (0.507); 1.006* (0.538); 0.057** (0.029); 0.059** (0.029).
  - (CIT variable) t-1: -0.838** (0.410); -0.757* (0.457); -0.935* (0.489); -0.955** (0.453); -0.066** (0.032); -0.061* (0.033).
  - (VAT variable) t: -1.371*** (0.521); 0.001 (0.001); 0.008 (0.335); -0.049 (0.325).
  - (VAT variable) t-1: 0.974** (0.427); -0.003** (0.001); 0.105 (0.290); 0.029 (0.254).
  - (TAX/GDP) t: -0.197** (0.100); -0.499*** (0.188); -0.335*** (0.147); -0.376** (0.163); -0.261*** (0.086); -0.282*** (0.108).
  - (TAX/GDP) t-1: 0.256** (0.110); 0.173 (0.126); 0.352** (0.124); 0.347*** (0.116); 0.204*** (0.061); 0.209*** (0.074).
- Robustness table (Table 8; selected entries):
  - LDV t-1: 0.862*** (0.064); 0.915*** (0.036); 0.729*** (0.092).
  - (CIT variable) t: 0.510** (0.223); 0.067** (0.028); 0.278* (0.159).
  - (CIT variable) t-1: -0.602** (0.299); -0.075*** (0.016); -0.268* (0.150).
  - (TAX/GDP) t: -0.276*** (0.078); -0.248** (0.090); -0.397** (0.156).
  - Observations in robustness specifications: 492; 298; 492.
- Diagnostic test p-values (Table 3 and Table 8; examples):
  - First-order s.c.: 0.006, 0.022, 0.007, 0.011, 0.004, 0.003 (across Table 3 columns) and 0.003, 0.004, 0.007 (Table 8 columns).
  - Second-order s.c.: 0.490, 0.167, 0.402, 0.425, 0.201, 0.196 (Table 3 columns); 0.203, 0.207, 0.225 (Table 8 columns).
  - Sargan: 1.000 across reported specifications.

### Data, sample, and variable definitions
- Sample:
  - Unbalanced panel of 27 current OECD member countries with a VAT, covering the period 1967–2003, in each case covering the period after VAT introduction in each country.
  - Luxembourg and Mexico excluded due to lack of tax revenue data; United States excluded because it does not have a VAT.
  - Country-year coverage examples: Australia (2000–2002); Austria (1973–2003); Japan (1989–2003); United Kingdom (1973–2003); others as listed in the sample.
- Key data sources:
  - GDP and exports: World Economic Outlook database.
  - Tax revenue: OECD Revenue Statistics Database.
  - Final consumption expenditure: World Development Indicators database.
  - Statutory corporate tax rates: Devereux, Griffith and Klemm (2002) for 16 OECD countries during 1983–2001.
  - Statutory VAT rates: Tax Policy Division at the International Monetary Fund.
- Descriptive statistics (Table 5; preserved exactly):
  - Net exports as a fraction of GDP: Number of Observations 573; Mean 0.005; Mean Standard Deviation 0.047.
  - Exports as a fraction of GDP: 573; Mean 0.355; Mean Standard Deviation 0.158.
  - Log GDP per capita: 573; Mean 10.767; Mean Standard Deviation 2.436.
  - Total tax revenue, as a fraction of GDP: 573; Mean 0.372; Mean Standard Deviation 0.077.
  - VAT revenue, as a fraction of GDP: 573; Mean 0.066; Mean Standard Deviation 0.020.
  - VAT revenue, as a fraction of total consumption: 573; Mean 0.082; Mean Standard Deviation 0.036.
  - Standard VAT rate: 461; Mean 17.500; Mean Standard Deviation 5.373.
  - Corporate tax revenue, as a fraction of GDP: 573; Mean 0.026; Mean Standard Deviation 0.013.
  - Statutory corporate tax rate: 334; Mean 0.389; Mean Standard Deviation 0.120.

### Interpretation, caveats, and robustness checks
- VAT:
  - Short-run deterioration in net exports associated with higher VAT reliance likely reflects confounding shocks to consumption rather than a causal VAT trade effect.
  - Alternative VAT measures (standard rate; effective VAT as VAT revenue/consumption) eliminate the short-run effect.
- Corporate tax:
  - Results robust to using statutory corporate tax rate (columns 5 and 6) and to alternative specifications; short-run positive relation with net exports, reversing after one period and converging to zero.
  - Marginal effective tax rate (METR) was tried and proved insignificant (not shown).
  - Potential endogeneity concerns (e.g., profit cycles affecting corporate tax revenue) are addressed by replacing CIT reliance with statutory rate and by instrumenting; qualitative results persist.
- Other domestic tax measures:
  - Total tax revenue as a proportion of GDP typically shows an initial increase leading to a reduction in net exports, followed by reversal and no long-run effect.
  - Similar patterns obtained replacing tax ratio by general government expenditure relative to GDP (not shown).

### Summary conclusions (section V)
- Two consistent patterns over the last 35 years in OECD experience:
  - VAT appears to have no impact on the trade balance in either the short or the long run; observed dynamics can be explained by unrelated consumption shocks, supporting VAT’s inherent trade neutrality.
  - Corporate tax changes have powerful dynamic effects consistent with source-based taxation in practice: short-run increase in net exports due to capital outflow, subsequent reduction from income inflows, and eventual convergence to zero long-run effect.
- Overall implication:
  - Some aspects of domestic tax policy have strong, complex, and temporally varying effects on trade performance, though long-run net export effects from VAT and corporate tax changes converge to zero in these specifications.

*Source: _wp0647 - 0.23 points lower than prior to the (IMF working paper PDF content as provided).*

### References

### _wp0647 - References

### References list
- Arellano, Manual, and S. Bond, 1991, “Some Tests of Specification for Panel Data: Monte Carlo Evidence and An Application to Employment Equations,” Review of Economic Studies Vol. 1, 58, pp. 277–97.  
- de Meza, David, Ben Lockwood, and Gareth Myles, 1994, “When are Destination and Origin Regimes Equivalent?” International Tax and Public Finance, Vol. 1, pp. 5–24.  
- de Mooij, Ruud and Sjef Ederveen, 2003, “Taxation and Foreign Direct Investment: A Synthesis of Empirical Research,” International Tax and Public Finance, Vol. 10, pp. 673–93.  
- Desai, Mihir A., and James R. Hines Jr., 2005, “Value Added Taxes and International Trade: The Evidence” (unpublished; Ann Arbor: University of Michigan).  
- Devereux, Michael and Rachel Griffith, 2003, “Evaluating Tax Policy for Location Decisions,” International Tax and Public Finance, Vol. 10, pp. 107–26.  
- Devereux, Michael, Rachel Griffith, and Alexander Klemm, 2002, “Corporate Income Tax Reforms and Tax Competition,” Economic Policy, pp. 451–95. [Data available at http://www.ifs.org.uk/publications.php?publication_id=3210]  
- Ebrill, Liam, Michael Keen, Jean-Paul Bodin, and Victoria Summers, 2001, The Modern VAT (Washington: International Monetary Fund).  
- Feldstein, Martin S., and Paul Krugman, 1990, “International Trade Effects of Value Added Taxation,” ed. by Assaf Razin and Joel Slemrod, Taxation in the Global Economy, (Chicago: Chicago University Press), pp. 263–82.  
- Genser, Bernd, Andreas Haufler, and Peter Birch Sørensen, 1995, “Indirect Taxation in an Integrated Europe: Is There a Way of Avoiding Trade Distortions Without Sacrificing National Tax Autonomy?” Journal of Economic Integration, Vol. 10, pp. 178–205.  
- Hartman, David A., 2004, “The Case for Border-Adjusted Taxation in the United States,” Tax Notes International, September 27.  
- Hines, James R. Jr., 1999, “Lessons from Behavioral Responses to International Taxation,” National Tax Journal, Vol. 52, pp. 305–22.  
- Keen, Michael, and Ben Lockwood, 2005, “Causes and Consequences of the Value-Added Tax,” forthcoming.  
- Rodrik, D., 1998, “Why Do More Open Economies Have Bigger Governments?” Journal of Political Economy, Vol. 106, pp. 997–1032.  
- Rose, Andrew, 2002, “Do WTO Members Have a More Liberal Trade Policy?” NBER Working Paper No.9347 (Cambridge, Massachusetts: National Bureau of Economic Research) [Data available at http://faculty.haas.berkeley.edu/arose/RecRes.htm].  
- Sinn, Hans-Werner, 1985, “Why Taxes Matter: Reagan’s Accelerated Cost Recovery System and the U.S. Trade Deficit,” Economic Policy, pp. 239–50.  
- ―——1990, “Can Direct and Indirect Taxes Be Added for International Comparisons of Competitiveness?” Reforming Capital Income Taxation, ed. by Horst Siebert (Tubingen: J.C.B. Mohr), pp. 1–19.  
- Slemrod, Joel, 2004, “Are Corporate Tax Rates, or Countries, Converging?” Journal of Public Economics, Vol. 88, pp. 1169–86.  
- Viard, Alan D., 2004, “Border Adjustments Won’t Promote U.S. Competitiveness,” Tax Notes International, October 11.

*Source: _wp0647 - References*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2006/_wp0647.pdf_