“Trade Taxes Are Better ?!? Short Answer: No” by Can Erbil Brandeis University and EcoMod

1. “Trade Taxes Are Better ?!? Short Answer: No” by Can Erbil Brandeis University and EcoMod July 2003

2. I. Motivation • Trade Liberalization: a powerful policy tool to improve economic performance and raise standards of living. • Doha Round • Barriers to Trade: still wide spread. • trade restrictions as optimal policy • interest group politics • revenue-raising aspect

3. Objective: • To study the welfare implications of trade reforms, and • compensating taxation, • taking into account the importance of the revenue-raising aspect of trade barriers. Direct cost: loss of tax revenue from trade, which must be made up from increases in other distortionary taxes or cuts in government spending. Revenue loss can be substantial for developing countries. Welfare implications of trade liberalization become an important issue. The net effect can be a welfare loss.

4. Welfare Measurement: • Builds on Anderson’s (1999) study • Investigating in more depth the Marginal Cost of Funds --- MCF --- calculations for distortionary taxation. • MCF for any tax increase is given by the ratio of the incremental compensation, required to maintain real income to the incremental tax revenue. • A revenue neutral shift from the tax with high MCF to the tax with low MCF is welfare improving.

5. Major Questions: • What is the MCF for trade taxes versus output taxes? • Are revenue neutral trade reforms welfare improving? • Are there “cheaper” distortionary taxes that we can replace the trade taxes with? Answers: • The MCF of trade taxes versus MCF of output taxes for 34 countries. • For 27 of them trade liberalization compensated with an output tax policy is welfare improving. • Output taxes are “cheaper” than trade taxes for 27 countries.

6. Background On coordinating tariff reductions with domestic tax reforms: • Rajaram (1994) • Abed (1998) • Anderson (1999) On using MCF as the welfare measurement tool of tax reforms • Anderson (1999) • Devarajan, Thierfelder and Suthiwart-Narueput (2000) • Rutherford (2001)

7. Contributions: • Utilization of (compensated) MCF as the measurement tool for welfare impacts of tax reforms • The large number of countries included in the empirical analysis • The explicit usage of intermediate imported inputs in the model • Econometric analysis to shed light on the determinants of the MCF

8. II. Marginal Cost of Funds – Basic Intuition MCF for any tax increase gives the relationship between: ·the marginal compensation required to maintain real incomeand ·the marginal tax revenue raised by this tax increase. mdp is the marginal compensation [m + (p-p*)mp]dp is the marginal tax revenue change. In other words, it is the marginal cost of raising another dollar of tax revenue: Intuitively, it takes morethan onedollar of compensation to maintain real income, when $1 of added tariff revenue is raised.

9. III. The Model A simple computable general equilibrium (CGE) model simulates the working of a small open market economy (SOE). A. Production Block • Perfect Competition • Constant Returns to Scale • Three sectors: agriculture, manufacturing and service • Two primary factors of production: Labor and Capital • Domestic and imported intermediates

10. A multi-level nesting of all the factors of production constructs the sectoral output:

11. B. Supply Behavior • Each sector decides how much of the sectoral output will be produced as exports or domestic products for domestic consumption, which are imperfect substitutes in supply:

12. C. Demand Behavior • The Armington specification between imports and domestic goods indicates imperfect substitution in demand:

13. IV. Empirical Work ELASTICITIES Erbil Literature and GTAP i , elasticity of substitution between capital and labor (production - CES) 1.2 0.9-1.8 i , elasticity of substitution between domestic and imported inputs (production - CES) 2 0.4-3.6 i, elasticity of substitution between value added and composite intermediate (production - CES) 0.1 0-1.7 ELAi, elasticity of substitution between exports and domestically consumed domestic products (supply – CET) 3 2.9-3.9 EPSIi, elasticity of substitution between imports and domestic products (demand – CES) 2 0.4-3.6 A) Data Sources • GTAP (Global Trade Analysis Project) database • Versions: GTAP 4 (1995) and GTAP 5 (1997) B) Elasticities

14. V. Results of the Empirical Analysis A) Range of MCF Estimates • MCF estimates reported by previous studies range from 0.67 to 4.51. – (see Table 5.1 in the paper). • The empirical results for 34 countries fall in this interval and have a narrower range: [0.915-1.558].

15. B) Multi-Country MCF Estimates

16. C) Interpretation of Results • For Turkey, MCF of tariffs is 1.27, and MCF of output taxes is 1.041. • To raise $1, while keeping the government budget constraint constant,$1.27 must be spent when using tariffs as the policy tool, and only $1.041 when using the output taxes. • Output taxes are the “cheaper” distortion, and therefore replacing trade taxes with output taxes is welfare improving. • Net benefit of switching from trade to output taxes is MCF(tariffs)-MCF(output taxes). • For Turkey, the net benefit would be $1.27 - $1.041 = $0.229. • For every $1 raised by output taxes instead of tariffs, Turkey would gain $0.229. This result suggests that there are still welfare gains to be captured from trade liberalization for Turkey.

17. For the majority of countries (27 out of 34), MCF(output taxes) <MCF(tariffs). • For seven countries, trade taxes are the “cheaper” distortion. • For these countries, replacing trade taxes with output taxes is not beneficial. A trade liberalization package financed with output taxes would be welfare decreasing. • For the remaining 27 countries, output taxes are “cheaper” than trade taxes, and trade liberalization would be the correct policy recommendation. • Korea, China, Poland and thePhilippines have the highest potential welfare gains: $0.354, $0.288, $0.251 and $0.24, respectively, for each dollar raised, if they switched from tariffs to output taxes as the revenue collecting policy instrument.

18. Imported intermediate inputs • “Simple Model” (without imported intermediate inputs in production): trade taxes the “cheaper” distortion for the majority of countries. • “Advanced Model” (with imported intermediate inputs in production): trade taxes the “more expensive” distortion for the majority of countries.Reversal of Results • For most of the countries, trade taxes were the “cheaper” distortion in the “simple model”, and became the “more expensive” distortion in the “advance model”. • In the “Advanced Model”, trade taxes impose distortion on both the consumption and production side.

19. D) Sensitivity Analysis • MCF figures are not very sensitive to changes in the elasticity of substitution between imported goods and domestic products (Armington elasticity). • Empirically determined elasticity of the MCF with respect to the elasticity of substitution: for MCF(tariffs) the elasticity is within the range of 0.0122 and 0.0408, and for MCF(output taxes), it is even smaller, between 0.000004 and 0.0186.

20. VI. Conclusion • Are trade taxes better ? Short answer: No. • For 27 out of 34 countries investigated in this study, trade taxes are the “more expensive” distortion. • The relevant policy recommendation that emerges from these results is that there are welfare gains to be exploited by trade liberalization in developing countries. There is still a strong potential for many countries to liberalize their trade. • The fact that 24 of these 27 countries are developing countries makes this recommendation even more encouraging.

21. VII. Extensions A. Distortionary Income Taxation B. Other Types of Taxation C. Other Distortions (Non-tariff Barriers) D. Dynamic Model E. Sensitivity Analysis in a Broader Sense

22. END OF PRESENTATION

23. E) Cross-Country Econometric Analysis • identify the determinants of MCF • Regression Analysis for MCF of Tariffs • Elasticity of MCFTM with respect to a change in the tariff rate is 0.05. The coefficient is highly statistically significant (more than 95% confidence level). • Elasticity of MCFTM with respect to theother distortion, output taxes, is also positive, but very small, 0.002, and not statistically significant. • An increase in the share of imports in GDP, share of imports in consumption and share of intermediate imports over total imports, have small but positive effects on MCFTM.

24. Regression Analysis for MCF of Output Taxes (MCFTY) • Using interactive slope dummy variables for GDP size: high output taxesincrease the magnitude of MCFTY significantly for large countries, while they decrease its magnitude for small countries. • An increase in the tariff rate and in the share of domestic goods in total consumption affect MCFTY positively, an increase in the share of imports in GDP, negatively. All three are intuitive.

25. What did we learn from the regressions? • The conventional wisdom that higher distortions imply higher MCF estimates is confirmed. • High share of imports in production and consumption makes MCF of tariffs more costly. This finding supports trade liberalization recommendations for developing countries with a high demand for imported intermediate inputs. • Effects of distortions on MCF estimates seem to change with the size of the economy (determined by GDP). However, a larger sample with enough degrees of freedom in each size category is necessary to investigate this relationship.

26. Compensated versus Money Metric MCF • shadow price of foreign exchange, also called the fiscal multiplier: • price of an additional unit of money metricutility in terms of external compensation. • adjusts for the fact that changes in welfare have income effects, and these income effects induce changes in tariff revenues. • Compensated MCF is money metric MCF (MMCF) divided by fiscal multiplier: • The compensated version of MCF gives the willingness to pay for a dollar of external transfer to finance a tax reduction, and allows for international and interregional comparisons.

27. Calculating Compensated MCFs • First step: model is perturbed with a transfer of a small external exogenous amount into the government budget. This amount is offset by an endogenous proportionate change in the taxes. The simulation calculates the change in money metric utility. This is the uncompensated MCF (MMCF). • Second step: the same small external exogenous amount is injected in the government budget. It is offset by a lump sum transfer, from the government to the private sector. The change in money metric utility gives us the fiscal multiplier. • Third step: step1 is divided by step 2 to obtain the compensated version of MCF:

29. The Case of MCF<1 • Intuitively, it takes morethan onedollar of compensation to maintain real income, when $1 of added tariff revenue is raised. Hence, MCF>1. • However, three countries, Canada, USA and Chile have MCF estimates less than 1: * MCF of output taxes for Canada is 0.915 * MCF of tariffs for USA and Chile is 0.995 • Devarajan, Thierfelder and Suthiwart-Narueput (2000) argue that “in countries with large distortions, a tax whose substitution effect lowers that large distortion could have a very low or negative welfare cost. A policy which increases the lowest tax rate, in the absence of other distortions, will reduce the marginal cost of funds as the tax structure becomes more uniform”. • In the case of Canada, the presence of high agricultural subsidies would confirm their argument, and the intuition behind our finding of a MCF <1.

30. A dummy variable “large”, indicating countries with large GDPs, is positively related to MCFTM. It implies that tariffs would be less desirable as a policy tool to raise revenue for larger countries. Looking at the data, we observe that this is correct for large developing countries like China and Korea, but not for large developed countries like Canada or Great Britain.