a. Internal trade barriers

Despite past efforts to liberalize their trade regimes, the Common Market for Eastern and Southern Africa (COMESA) ^12/ countries still have relatively high trade barriers. Burundi, for example, has a maximum rate of import duty of 100 percent, Kenya 45 percent, Tanzania 40 percent, and Uganda 30 percent. Although intra-COMESA import duty rates now stand at nearly 40 percent of the regular import duty rates, there is considerable room for gaining efficiency from a liberalization of internal trade and a move toward a CET. The use of nontariff barriers is also widespread in the region as a whole. Import licensing, quantitative restrictions, and import bans are still employed to control imports in order to preserve foreign exchange or provide protection to some domestic producers. Removing these restrictions would produce positive welfare effects for the region as a whole.

b. Intraregional trade and the expected crade pattern

While the share of export trade within the COMESA to total exports is low (5.9 percent), significant trading links already exist nonetheless between some countries and their regional partners. For example, in 1993, 10.5 percent of Burundi’s trade was with the region and 18 percent of Kenya’s exports went to the region. Uganda also bought nearly 19 percent of its imports from within the region, mainly from Kenya (Table 3).

Table 3.

Regional Trade Indicators, 1993.

Source: Direction of Trade Statistics Yearbook, 1987-93, IMF (Washington; 1994).

Table 3.

Regional Trade Indicators, 1993.

	Imports from Eastern and Southern Africa	Exports to Eastern and Southern Africa
	(In percent of total imports)	(In percent of total exports)
Burundi	10.5	10.4
Kenya	2.3	18.0
Tanzania	5.3	10.8
Uganda	18.9	3.0

Source: Direction of Trade Statistics Yearbook, 1987-93, IMF (Washington; 1994).

View Table

Table 3.

Regional Trade Indicators, 1993.

	Imports from Eastern and Southern Africa	Exports to Eastern and Southern Africa
	(In percent of total imports)	(In percent of total exports)
Burundi	10.5	10.4
Kenya	2.3	18.0
Tanzania	5.3	10.8
Uganda	18.9	3.0

Source: Direction of Trade Statistics Yearbook, 1987-93, IMF (Washington; 1994).

These data are evidence that regional integration could play a key role in reinforcing and expanding trade flows in Eastern and Southern Africa, including countries considered here.

In light of the major contribution of the agricultural sector to GDP and the existence of cheap and abundant labor, agricultural products and labor-intensive industries would primarily benefit from the integration-stimulated trade. In the agricultural sector, the CBI will mainly stimulate trade flows in nontradeable crops and livestock given the relatively low share of cash crops in the GDP. In view of the labor-intensive production of agricultural goods in these countries, agricultural exports would have a greater competitive edge and find outlets in the partner countries. Because they may not require the same promotion and marketing costs as exports of manufactured products, it would be easier to realize an expansion in agricultural exports. Thus, as in the case of the European Union (EU) and other regional groupings, trade liberalization could easily result in a greater volume of agricultural exports in each country. Likewise, the agricultural sectors in the countries under review should expand and sustain growth under the CBI as each country specializes in the production of items for which it has a comparative advantage (a lower pre-CBI domestic relative price).

With its relatively developed industrial sector by the region’s standards, Kenya would develop production and exports of more capital-intensive goods. In the manufacturing sector, the CBI would create new trade opportunities and would encourage the existing plants to expand their markets outside their own borders. However, in the countries under consideration, the short-term response of the manufacturing sector to the CBI would be limited by its size, its primarily import-substitution nature, as well as its limited marketing experience. Since most of the existing plants were initially designed to cater to the domestic market and still enjoy substantial protection, the success of their exports strategy would depend on their ability to restructure, allocate resources for marketing, and substantially reduce their costs. A long-term response could be supported by actions in these areas, including improvements in the economic infrastructure (particularly transport and telecommunications) and in the general provision of services. Existing plants would face competition from foreign producers in the partner countries, and would need to have a competitive edge in order to make inroads into the new regional market. New investment encouraged by the CBI incentives could be undertaken, and new export-oriented enterprises would likely emerge.

The trade flows in this sector would be patterned according to the law of comparative advantage and the domestic relative prices before the CBI. In the context of a sound macroeconomic framework, the removal of tariffs and other barriers to trade would lower domestic prices and lead manufacturing enterprises to shift production toward those goods that could be produced more cheaply and exported to regional partners. Overall, the CBI would boost the level of trade among the participating countries, reflecting mainly their rather similar economic structures. However, the uncompetitive cost structures of these countries resulting from previous import-substitution strategies is likely to entail a large cost in terms of economic efficiency as a result of trade diversion, unless external barriers are simultaneously reduced. This condition would be met, because the introduction of the CET is scheduled to be at a lower level of external protection, thus reducing the welfare impact of any trade diversion that may occur (Table 4).

Table 4.

Existing and CBI-proposed Tariff Structures of the Sample Countries, 1993

Sources: Data provided by the authorities.

^1/As of June 1994.

Table 4.

Existing and CBI-proposed Tariff Structures of the Sample Countries, 1993

	In the pre-CBI situation					The CBI-proposed structure of the CET
	Burundi	Kenya		Tanzania	Uganda
	(In percent)
Intermediate goods	12	15		20	10	15
Capital goods	10	15		30	10	10
Consumer goods	43	45	1/	40	30	20

Sources: Data provided by the authorities.

^1/As of June 1994.

View Table

Table 4.

Existing and CBI-proposed Tariff Structures of the Sample Countries, 1993

	In the pre-CBI situation					The CBI-proposed structure of the CET
	Burundi	Kenya		Tanzania	Uganda
	(In percent)
Intermediate goods	12	15		20	10	15
Capital goods	10	15		30	10	10
Consumer goods	43	45	1/	40	30	20

Sources: Data provided by the authorities.

^1/As of June 1994.

c. Terms of trade effects

The present lower tariffs for consumer goods in Tanzania and Uganda would help these two countries reap terms of trade gains vis-à-vis their regional trading partners as internal trade is Liberalized. Internal liberalization would induce a larger shift in their partners’ reciprocal demand, thus leading to a favorable terms of trade movement for Tanzania and Uganda. Implementation of the CBI measures would not affect the terms of trade of any participating country with nonmember countries, as the member countries together are small relative to the world market. Nevertheless, their joint bargaining power vis-à-vis third countries could be improved as collective and coordinated demands carry more weight than individual demands. ^13/

d. Regional market

As indicated by Table 5, the four countries considered are low income countries, with per capita GDP in 1993 ranging from about US$100.0 in Tanzania to US$224.5 in Kenya. The economic size of these countries varies considerably. In terms of economic activity in 1993, measured by the level of GDP, Kenya was the most important with a GDP of US$6,308 million, which was more than six times that of Burundi, more than twice that of Tanzania, and almost twice that of Uganda. The entire bloc of COMESA countries represented a total population of nearly 215 million persons, and a total GDP of approximately US$53.5 billion. In comparison with the pre-integration situation where each country relied first on its own domestic market, the regional market would offer each individual country in the region a significant advantage in terms of market access in which full economies of scale could be realized.

Table 5.

Population and Par Capita GDP, 1993

Source: Economic Trends in Africa, IMF (Washington, 1994).

^1/In millions.

^2/In millions of U.S. dollars.

^3/In U.S. dollars.

Table 5.

Population and Par Capita GDP, 1993

	Population 1/	GDP 2/	GDP Par Capita 3/
Burundi	5.8	1,017.0	175.3
Kenya	28.1	6,308.0	224.5
Tanzania	25.9	2,596.0	100.2
Uganda	18.1	3,266.0	180.4
Sub-total	77.9	13,187.0	—
COMESA 3/	214.7	53,521.0	—

Source: Economic Trends in Africa, IMF (Washington, 1994).

^1/In millions.

^2/In millions of U.S. dollars.

^3/In U.S. dollars.

View Table

Table 5.

Population and Par Capita GDP, 1993

	Population 1/	GDP 2/	GDP Par Capita 3/
Burundi	5.8	1,017.0	175.3
Kenya	28.1	6,308.0	224.5
Tanzania	25.9	2,596.0	100.2
Uganda	18.1	3,266.0	180.4
Sub-total	77.9	13,187.0	—
COMESA 3/	214.7	53,521.0	—

Source: Economic Trends in Africa, IMF (Washington, 1994).

^1/In millions.

^2/In millions of U.S. dollars.

^3/In U.S. dollars.

e. Investment and growth effects of the CBI

Regional integration can increase both the rate and the productivity of investment. First, the emergence of large markets would improve prospects for profits and ultimately could attract higher levels of domestic and foreign direct investment in member countries. The prices of intermediate and capital goods could also fall, as integration measures would be introduced, further providing additional incentives to invest and/or to increase the rate of utilization of existing capacities. The investment environment would be further improved by the removal of payment restrictions and simplification of administrative procedures. Since the decision-making process affecting important policy instruments (trade liberalization, payments, and investment policies) would shift to the regional level, the CBI would contribute to reducing perceived uncertainty, thus boosting the investment rate, while providing institutional gains to participating countries. Second, these institutional effects can be instrumental in increasing the overall quality of policy in the union and in enhancing the quality and the productivity of investment. The investment effects of regional integration could be reinforced by further efforts to harmonize the fiscal, monetary, and exchange rate policies of the participating countries.

The CBI would also stimulate growth in the participating countries, as liberalization measures would contribute to a more efficient intersectoral reallocation of productive resources. In each country, producers would seek to invest in those sectors in which they can secure a competitive edge. Another positive effect on growth would come from the reduction of costs. The dismantling of present trade barriers would lower producers’ costs and free resources that would be subsequently used more efficiently in other activities. When combined with the implementation of a CET that is lower than the pre-integration external tariffs, this would reduce the anti-export bias and give a more outward-oriented stance to economic policy in each of the participating countries. As a result, increasing investment flows could be expected, thus stimulating economic growth. There is clear evidence from cross-sectional studies that suggests a positive relationship between outward-orientation and economic growth (Feder 1983; Belassa 1985; World Bank 1987; Guillaumont 1994). As indicated above, the agricultural sector, which is potentially the main beneficiary of the CBI, will lead growth.

f. Factor price effects

The CBI is expected to influence the behavior of factor prices. Trade effects would be felt by the agricultural sector, which is the most important sector in the participating countries. Opportunities to trade in agricultural products would be enhanced, since crade between partner countries would be conducted on the basis of international market prices that are higher than domestic ones. A rise in the real income in the exporting segments of the agricultural sector in each of the participating countries could be expected. As mentioned before, import-substitution strategies have provided high rates of effective protection to the manufacturing sector in the region as a whole. In the face of increased competition, the inefficient firms would thus be driven out of the market. To survive competition, the only viable option would be to introduce significant adjustment measures to reduce costs and enhance competitiveness. Exporting firms that are competitive relative to other regional producers would stand to expand their exports, and a rise in the real income of the factors they use intensively would be anticipated. The rise in real income would certainly be more pronounced for the unskilled workers, considering the fact that most of the region’s industries are still very light and produce goods that do not require very skilled labor. In Kenya the manufacturing sector is already using skilled workers to produce some capital-intensive goods. These workers would certainly see their real wages rise if Kenya develops its exports of such goods. Overall, according to the factor price equalization theorem, ^14/ the long-run result would be the elimination of differences in income levels among the countries in the region.

g. Transitional adjustment costs

Other important issues pertain to the transitional adjustment costs and the reform of the incentive system. With internal liberalization, the industrial sector in Burundi, Tanzania, and Uganda would each face increased competition from the more developed Kenyan industrial sector. After the CET is introduced, import-competing industries in all the participating countries would face more competition from foreign firms. The move to a CET would entail transitional adjustment costs, which is a concern of any authorities considering whether to embark on a liberalization program. The main argument is that some domestic industries may not survive the increased competitive pressure. Accordingly, the liberalization program should be implemented together with a reform of the overall industrial incentive system in order to provide adjustment assistance to some import-competing domestic producers. However, this rationale is not sustainable on several grounds. First, the adjustment process that drives inefficient firms out of the market is the way to achieve efficiency gains. Second, the tariff reform would eliminate the antiexport bias of the preintegration incentive system, and thus reduce the need for its reform. In addition, preannouncing the tariff reform would provide time to domestic firms to adapt and undertake adjustment measures. Third, it is difficult to identify, without encouraging rent-seeking activities, industries that in return for transitional short-term assistance, would be viable in the long term. The problem is exacerbated by the weak institutional capacity required to manage such adjustment assistance on a transparent basis. In the few cases where adjustment assistance would be warranted, this should be carried out selectively (based on detailed studies at the firm level), within a strictly defined time period, and made available through budgetary channels (preferably in the form of a direct subsidy) such that its authorization (and justification) could be renewed regularly at the beginning of each fiscal year. Also, donors should be closely associated in identifying the recipient industries, and in designing and monitoring the adjustment package.

a. Theoretical issues

Long-term effects of regional integration on government revenue and on the overall fiscal situation are likely to be positive in participating countries, as a combined result of the favorable impact on growth as well as the adoption of well-designed offsetting measures. The latter are intended to eliminate the negative short-term impact stemming from the elimination of internal tariffs and the reduction of external ones.

The elimination of tariffs on intraregional trade will cause per se a loss in international trade-related fiscal revenue. With respect to extraregional trade, it is more difficult to estimate the revenue impact of the move toward a CET matching the lowest tariff in effect. To quantify the impact on revenue, the “Laffer Curve” ^17/ analysis may well be relevant since a tariff could be considered as a tax. Reducing tariff levels could actually boost revenue if the official tariff rate subject to the adjustment is on the downward sloping portion of the curve. Research on Kenya pointed to a positive, nonlinear relation between official rates and effective rates of collection (estimated as the ratio of customs duties collected to the value of imports) with a different slope above and below a given level of the tariff (Pritchett and Sethi, 1993). Above that given level, the slope declines, suggesting that the collection performance at higher rates of customs duties is less than at lower rates. Another interesting result of that research is that the slope of the estimated relationship between effective rates of collection and official rates is less than 1. This clearly indicates that the collection of customs duties does not rise or fall proportionately with tariff increases or reductions. ^18/ In fact, evidence supports this result by showing that the change in the level of the official tariff is just one of several determinants of government revenue growth. Others include procedures for granting and managing exemptions; effective levels of collection; and solid safeguards against misdeclarations, under-declarations, and smuggling. If changes in these areas are introduced along with a tariff elimination on intra-regional trade and a move toward the CET, the result could even be an increase in government revenue. However, these changes can produce their full effects only in the long run. In a short- to medium-term horizon, it would be reasonable to consider that only the tariff changes will be fully implemented, with other aspects of the tax system remaining broadly unchanged. However, one could also expect that reducing tariffs would lower incentives to smuggling and non-compliance, and bring more trade into legal channels, thus generating more taxes and improving fiscal revenue. Overall, the introduction of a CET will cause revenue losses, but in a proportion that is less than unitary.

b. Methodological issues

Regional integration has economy-wide effects through relative price changes and exchange rate adjustments. The economy will move to a new general equilibrium, and the subsequent change in the level of government revenue is in itself a measure of the effect of integration. This measure could be estimated by using either an applied general equilibrium approach or a partial equilibrium approach. The first approach allows the economy-vide effects on government revenue to be captured, and a full model of the economy needs to be specified and applied. However, data requirements make it difficult to extend the approach to each country. Therefore, the alternative is to consider the second approach. While excluding sectoral interactions, it nonetheless focuses on the effects of tariff changes on each particular sector or product category. The partial equilibrium approach, though less comprehensive, is still useful in analyzing the short-term effects, which often are the main concern of governments.

This paper develops a partial equilibrium approach to estimate the impact of the elimination of customs duties on intraregional trade as well as the impact of the move toward the CET on government revenues. The net impact will be obtained by summing the revenue loss incurred while bringing intraregional tariffs down to zero and the loss incurred while moving toward the CET.

c. Estimates of the revenue impact of regional integration

Indirect taxes are an important source of government revenue in the countries under review (Table 6). Customs duties are among the major contributors to total revenue, ranging from 10.2 percent of total revenue in Kenya to 44.1 percent in Uganda (Table 7).

Table 6.

Structure of Central Government Revenue, 1993

Sources: Data provided by the authorities; and IMF staff estimates.

^1/Covering FY 1992/93 (July-June).

Table 6.

Structure of Central Government Revenue, 1993

	Indirect taxes	Income tax	Nontax revenue
	(As a percentage of total revenue)
Burundi	58.7	26.7	14.6
Kenya 1/	54.0	28.7	17.3
Tanzania 1/	61.5	27.7	10.8
Uganda 1/	81.3	14.5	4.2

Sources: Data provided by the authorities; and IMF staff estimates.

^1/Covering FY 1992/93 (July-June).

View Table

Table 6.

Structure of Central Government Revenue, 1993

	Indirect taxes	Income tax	Nontax revenue
	(As a percentage of total revenue)
Burundi	58.7	26.7	14.6
Kenya 1/	54.0	28.7	17.3
Tanzania 1/	61.5	27.7	10.8
Uganda 1/	81.3	14.5	4.2

Sources: Data provided by the authorities; and IMF staff estimates.

^1/Covering FY 1992/93 (July-June).

Table 7.

Importance of Selected Indirect Taxes, 1993

Sources: Data provided by the authorities; and IMF staff estimates.

^1/Covering FY 1992/93 (July-June).

^2/Value-added tax.

Table 7.

Importance of Selected Indirect Taxes, 1993

	Sales tax		Excise tax	Customs duties
	(As a percentage of total revenue)
Burundi	17.2		21.5	19.3
Kenya 1/	31.8	2/	11.2	10.2
Tanzania 1/	19.2		10.9	19.3
Uganda 1/	36.2		6.7	44.1

Sources: Data provided by the authorities; and IMF staff estimates.

^1/Covering FY 1992/93 (July-June).

^2/Value-added tax.

View Table

Table 7.

Importance of Selected Indirect Taxes, 1993

	Sales tax		Excise tax	Customs duties
	(As a percentage of total revenue)
Burundi	17.2		21.5	19.3
Kenya 1/	31.8	2/	11.2	10.2
Tanzania 1/	19.2		10.9	19.3
Uganda 1/	36.2		6.7	44.1

Sources: Data provided by the authorities; and IMF staff estimates.

^1/Covering FY 1992/93 (July-June).

^2/Value-added tax.

(1) Revenue loss from internal liberalization

The revenue lost by each country as a result of internal liberalization is in fact the share of customs duties from intraregional imports with respect to the total amount of customs duties collected. Assuming the same performance in collecting duties on intra-regional imports and on imports from outside the region, this share can be estimated as follows:

Let DC, DCr, DCnr be the total duties collected, duties collected on intraregional imports, and duties collected on imports from nonmember countries, respectively; then

$\begin{array}{cc}\mathrm{DC}=\mathrm{DCr}+\mathrm{DCnr}& \left(1\right)\end{array}$

$\begin{array}{cc}\mathrm{DC}\mathrm{=}\mathrm{\left(}\mathrm{1}\mathrm{-}\mathrm{p}\mathrm{\right)}\mathrm{*}\mathrm{t}\mathrm{*}\mathrm{a}\mathrm{*}\mathrm{M}\mathrm{+}\mathrm{t}\mathrm{*}\mathrm{\left(}\mathrm{1}\mathrm{-}\mathrm{a}\mathrm{\right)}\mathrm{*}\mathrm{M}\mathrm{,}& \mathrm{\left(}\mathrm{2}\mathrm{\right)}\end{array}$

With p = level of the tariff preference enjoyed by intraregional imports over imports from outside the region;

t = effective rate of collection (ratio of customs duties collected to value of imports);

a = share of intra-regional imports in total imports; and

M = total imports.

Let b represent the share of customs duties collected on intra-regional imports in total customs duties collected, then from (2):

$\begin{array}{cccc}\mathrm{b}& \mathrm{=}& \mathrm{\left(}\mathrm{1}\mathrm{-}\mathrm{p}\mathrm{\right)}\mathrm{*}\mathrm{t}\mathrm{*}\mathrm{a}\mathrm{*}\mathrm{M}\mathrm{/}\mathrm{\left[}\mathrm{\left(}\mathrm{1}\mathrm{-}\mathrm{p}\mathrm{\right)}\mathrm{*}\mathrm{t}\mathrm{*}\mathrm{a}\mathrm{*}\mathrm{M}\mathrm{+}\mathrm{t}\mathrm{*}\mathrm{\left(}\mathrm{1}\mathrm{-}\mathrm{a}\mathrm{\right)}\mathrm{*}\mathrm{M}\mathrm{\right]}\hfill & \\ & \mathrm{=}& \mathrm{\left(}\mathrm{1}\mathrm{-}\mathrm{p}\mathrm{\right)}\mathrm{*}\mathrm{a}\mathrm{/}\mathrm{\left[}\mathrm{\left(}\mathrm{1}\mathrm{-}\mathrm{p}\mathrm{\right)}\mathrm{*}\mathrm{a}\mathrm{+}\mathrm{\left(}\mathrm{1}\mathrm{-}\mathrm{a}\mathrm{\right)}\mathrm{\right]}\hfill & \mathrm{\left(}\mathrm{3}\mathrm{\right)}\end{array}$

The revenue loss from internal liberalization is then a function of both the share of intraregional imports in total imports and the level of tariff preferences in effect.

Using figures on intraregional imports from Table 3, and since import duty rates between COMESA members now stand at nearly 40 percent of the regular import duty rates—thus implying a 60 percent preference, ^19/ Table 8 shows the ratio of customs duties collected on intraregional imports as a percent of total customs duties collected (coefficient b in equation 3). Using figures in Table 7, customs duties collected on intraregional imports are further expressed as a percentage of total revenue (Table 9). This is obtained by multiplying data for respective countries in Table 8 by data in Table 7. The outcome is that duties on intraregional imports do not contribute significantly to total revenue, particularly in Kenya and Tanzania. This fraction of total revenue will automatically be lost by national treasuries as they remove internal tariffs.

Table 8.

Ratio of Customs Duties Collected on Intraregional Imports, 1993

Sources: Calculations from Table 3, using equation (3).

Table 8.

Ratio of Customs Duties Collected on Intraregional Imports, 1993

	Percent of total customs duties collected
Burundi	4.5
Kenya	0.9
Tanzania	2.2
Uganda	8.5

Sources: Calculations from Table 3, using equation (3).

View Table

Table 8.

Ratio of Customs Duties Collected on Intraregional Imports, 1993

	Percent of total customs duties collected
Burundi	4.5
Kenya	0.9
Tanzania	2.2
Uganda	8.5

Sources: Calculations from Table 3, using equation (3).

Table 9.

Estimated Shares of Total Revenue from Customs Duties on Intraregional Imports, 1993

Sources: Calculations from Tables 7 and 8.

Table 9.

Estimated Shares of Total Revenue from Customs Duties on Intraregional Imports, 1993

	In percent of total revenue
Burundi	0.9
Kenya	0.1
Tanzania	0.4
Uganda	3.7

Sources: Calculations from Tables 7 and 8.

View Table

Table 9.

Estimated Shares of Total Revenue from Customs Duties on Intraregional Imports, 1993

	In percent of total revenue
Burundi	0.9
Kenya	0.1
Tanzania	0.4
Uganda	3.7

Sources: Calculations from Tables 7 and 8.

(2) Revenue impact of the introduction of the CET

Let M_i = value of imports of products in the pre-CET situation in local currency;

Q_i = quantity of product i imported in the pre-CET situation;

P_i = price of product i in foreign currency;

t_i = external tariff on product i in the pre-CET situation;

CET_i = Common External Tariff on product i in the after-CET situation;

e_i = price elasticity of imports;

PD_i - domestic price of product i;

E = local currency value of a unit of foreign currency; and

d = rate of change of E following integration. In the case of a devaluation, then d > 0;

M = total value of imports in the pre-CET situation;

R = hypothetical revenue in the pre-CET situation;

p_i = rate of change in the domestic price of product i associated with d_i. We expect 0 ≤ p_i ≤ d.

Let a variable with a prime indicate the same variable after introduction of the CET. Then we have:

$\begin{array}{cc}{\mathrm{M}}_{\mathrm{i}}\mathrm{=}{\mathrm{Q}}_{\mathrm{i}}\mathrm{*}{\mathrm{P}}_{\mathrm{i}}\mathrm{*}\mathrm{E}\mathrm{,}& \mathrm{\left(}\mathrm{1}\mathrm{\right)}\end{array}$

$\begin{array}{cc}{\mathrm{Pd}}_{\mathrm{i}}\mathrm{\prime }\mathrm{/}{\mathrm{Pd}}_{\mathrm{i}}\mathrm{=}\mathrm{\left(}\mathrm{1}\mathrm{+}{\mathrm{p}}_{\mathrm{i}}\mathrm{\right)}\mathrm{\left(}\mathrm{1}\mathrm{+}{\mathrm{CET}}_{\mathrm{i}}\mathrm{\right)}\mathrm{/}\mathrm{\left(}\mathrm{1}\mathrm{+}{\mathrm{t}}_{\mathrm{i}}\mathrm{\right)}\mathrm{,}& \mathrm{\left(}\mathrm{2}\mathrm{\right)}\end{array}$

$\begin{array}{cc}{\mathrm{e}}_{\mathrm{i}}\mathrm{=}\mathrm{\left[}\mathrm{\left(}{\mathrm{Q}}_{\mathrm{i}}\mathrm{\prime }\mathrm{-}{\mathrm{Q}}_{\mathrm{i}}\mathrm{\right)}\mathrm{/}{\mathrm{Q}}_{\mathrm{i}}\mathrm{\right]}\mathrm{/}\mathrm{\left[}\mathrm{\left(}{\mathrm{Pd}}_{\mathrm{i}}\mathrm{\prime }\mathrm{-}{\mathrm{Pd}}_{\mathrm{i}}\mathrm{\right)}\mathrm{/}{\mathrm{Pd}}_{\mathrm{i}}\mathrm{\right]}\mathrm{.}& \mathrm{\left(}\mathrm{3}\mathrm{\right)}\end{array}$

From (2) and (3), we have:

$\begin{array}{cc}{\mathrm{Q}}_{\mathrm{i}}\mathrm{\prime }\mathrm{=}{\mathrm{Q}}_{\mathrm{i}}\mathrm{\left[}\mathrm{1}\mathrm{+}{\mathrm{e}}_{\mathrm{i}}\mathrm{\left[}\mathrm{\left(}\mathrm{1}\mathrm{+}{\mathrm{p}}_{\mathrm{i}}\mathrm{\right)}\mathrm{\left(}\mathrm{1}\mathrm{+}{\mathrm{CET}}_{\mathrm{i}}\mathrm{\right)}\mathrm{/}\mathrm{\left(}\mathrm{1}\mathrm{+}{\mathrm{t}}_{\mathrm{i}}\mathrm{\right)}\mathrm{-}\mathrm{1}\mathrm{\right]}\mathrm{\right]}\mathrm{,}& \mathrm{\left(}\mathrm{4}\mathrm{\right)}\end{array}$

$\begin{array}{cc}{\mathrm{M}}_{\mathrm{i}}\mathrm{\prime }\mathrm{=}{\mathrm{Q}}_{\mathrm{i}}\mathrm{\prime }\mathrm{*}{\mathrm{P}}_{\mathrm{i}}\mathrm{*}\mathrm{E}\mathrm{*}\mathrm{\left(}\mathrm{1}\mathrm{+}\mathrm{d}\mathrm{\right)},& \left(5\right)\end{array}$

$\begin{array}{cc}{\mathrm{M}}_{\mathrm{i}}\mathrm{\prime }\mathrm{=}{\mathrm{M}}_{\mathrm{i}}\mathrm{(}\mathrm{1}\mathrm{+}\mathrm{d}\mathrm{)}\mathrm{\left[}\mathrm{1}\mathrm{+}{\mathrm{e}}_{\mathrm{i}}\mathrm{[}\mathrm{(}\mathrm{1}\mathrm{+}{\mathrm{p}}_{\mathrm{i}}\mathrm{)}\mathrm{(}\mathrm{1}\mathrm{+}{\mathrm{CET}}_{\mathrm{i}}\mathrm{)}\mathrm{/}\mathrm{(}\mathrm{1}\mathrm{+}{\mathrm{t}}_{\mathrm{i}}\mathrm{)}\mathrm{-}\mathrm{1}\mathrm{\right]}\mathrm{]}\mathrm{,}& \mathrm{\left(}\mathrm{6}\mathrm{\right)}\end{array}$

$\begin{array}{cc}\mathrm{R}=\underset{i}{\mathrm{\Sigma }}{M}_i{t}_i,& \left(7\right)\end{array}$

$\begin{array}{cc}\mathrm{R}\prime =\underset{i}{\mathrm{\Sigma }}{\mathit{CET}}_i{M}_i\prime ,& \left(8\right)\end{array}$

$\begin{array}{cc}R\prime \mathrm{=}{\underset{i}{\mathit{\Sigma }\mathit{}{\mathit{CET}}_{\mathit{i}}*}}_{}{\mathit{M}}_{\mathit{i}}*\mathit{(}\mathit{1}\mathit{+}\mathit{d}\mathit{)}\mathit{[}1\mathit{+}{\mathit{c}}_{\mathit{i}}\mathit{(}\mathit{(}1\mathit{+}{\mathit{p}}_{\mathit{i}}\mathit{)}\mathit{(}1\mathit{+}{\mathit{CET}}_{\mathit{i}}\mathit{)}\mathit{/}\mathit{(}1\mathit{+}{\mathit{t}}_{\mathit{i}}\mathit{)}\mathit{-}1\mathit{\left)}\mathit{\right)}]\mathrm{.}& \mathrm{(}8\prime \mathrm{)}\end{array}$

Let R_im be the revenue impact; then:

R_im = R’ - R

$\begin{array}{cc}{\mathrm{R}}_{\mathrm{im}}\mathrm{=}{\underset{}{\underset{i}{\mathit{\Sigma }}\mathit{}{\mathit{CET}}_{\mathit{i}}*}}_{}{\mathit{M}}_{\mathit{i}}*\mathit{(}1\mathit{+}\mathrm{d}\mathit{)}-\underset{}{\underset{i}{\mathrm{\Sigma }}\mathrm{}{M_i}_{}{t}_i+\underset{i}{\mathrm{\Sigma }}{e}_i}\left[\mathit{\right(}1\mathit{+}{\mathit{p}}_{\mathit{i}}\mathit{)}\mathit{(}1\mathit{+}{\mathit{t}}_{\mathit{i}}\mathit{)}-1\left)\right]{\mathit{CET}}_{\mathit{i}}(1+d\mathit{)}{M}_i\mathrm{.}& \mathrm{\left(}10\mathrm{\right)}\end{array}$

Dividing R_im by total imports, we obtain the revenue impact as a percentage of imports:

$\begin{array}{c}{\underset{}{\underset{i}{\mathit{\Sigma }}\mathit{}{\mathit{CET}}_{\mathit{i}}(1+d)}}_{}({\mathit{M}}_{\mathit{i}}/M)-\underset{}{\underset{i}{\mathrm{\Sigma }}\mathrm{}({M_i}_{}/M)t_i+\underset{i}{\mathrm{\Sigma }}{e}_i}\left(\right[(1+p_i)\mathit{(}1\mathit{+}{\mathit{CET}}_i\mathit{)}/\mathit{(}1\mathit{+}t\mathit{)}-1\left)\right]\end{array}$

$\begin{array}{cc}+{\mathit{CET}}_{\mathit{i}}\left(M_i/M\right)& \left(10\prime \right)\end{array}$

Under the no devaluation hypothesis, the formula reduces to:

$\begin{array}{cc}{\underset{}{\underset{i}{\mathit{\Sigma }}\mathit{}{\mathit{CET}}_{\mathit{i}}}}_{}({\mathit{M}}_{\mathit{i}}/M)-\underset{}{\underset{i}{\mathrm{\Sigma }}\mathrm{}({M_i}_{}/M)t_i+\underset{i}{\mathrm{\Sigma }}{e}_i}\left[\mathit{\right(}1\mathit{+}{\mathit{CET}}_i\mathit{)}/\mathit{(}1\mathit{+}{\mathit{t}}_{\mathit{i}}\mathit{)}-1\left)\right]{\mathit{CET}}_{\mathit{i}}(M_i/M)\mathrm{.}& \mathrm{\left(}11\mathrm{\right)}\end{array}$

The first two terms of equation (11) are the trade-weighted average tariffs in the context of the CET and before introduction of the CET, respectively. The last term captures the elasticity effect, as the quantity imported will increase and may partly offset the revenue impact of the CET. Trade-weighted average tariffs indicate the hypothetical (average) revenue that would be generated if the full official tariff were applied to the import value of each product category. When applying equation (11), imports have been disaggregated into three categories: intermediate goods, capital goods, and consumer goods (see Table 4 for their tariff structures before and after the CET). Table 10 shows their average shares in total imports over the period 1991-93.

Table 10.

Structure of Imports, Average 1991-93

Sources: Data provided by the authorities; and IMF staff estimates and calculations.

Table 10.

Structure of Imports, Average 1991-93

	Burundi	Kenya	Tanzania	Uganda
	(In percent)
Intermediate goods	35.5	61.3	23.3	24.3
Capital goods	31.4	28.5	50.3	39.9
Consumer goods	33.1	10.1	26.3	35.8

Sources: Data provided by the authorities; and IMF staff estimates and calculations.

View Table

Table 10.

Structure of Imports, Average 1991-93

	Burundi	Kenya	Tanzania	Uganda
	(In percent)
Intermediate goods	35.5	61.3	23.3	24.3
Capital goods	31.4	28.5	50.3	39.9
Consumer goods	33.1	10.1	26.3	35.8

Sources: Data provided by the authorities; and IMF staff estimates and calculations.

Based on present available empirical evidence, we have used two aggregate import elasticities, -0.88 and -1.4, considered to be reasonable estimates of the magnitude of the import response to price changes in African countries. ^20/ The import trade-weighted average tariff using the existing tariff structure and the CBI-proposed tariff schedule are both shown in Table 11 (Columns 1 and 2, respectively). The difference between these two trade-weighted tariffs (Column 3) thus gives the revenue impact of the CET, assuming a zero import elasticity.

Table 11.

Trade-Weighted Average Tariff Rate, 1993

Sources: Calculations from Tables 4 and 10.

^1/As of June 1994.

Table 11.

Trade-Weighted Average Tariff Rate, 1993

	Under existing tariff structure (1)	Under the CBI-proposed CET (2)		Difference (3) = (2) - (1)
Burundi	21.6		15.1	-6.5
Kenya	18.0	1/	14.1	-3.9
Tanzania	30.3		13.8	-16.5
Uganda	17.2		14.8	-2.4

Sources: Calculations from Tables 4 and 10.

^1/As of June 1994.

View Table

Table 11.

Trade-Weighted Average Tariff Rate, 1993

	Under existing tariff structure (1)	Under the CBI-proposed CET (2)		Difference (3) = (2) - (1)
Burundi	21.6		15.1	-6.5
Kenya	18.0	1/	14.1	-3.9
Tanzania	30.3		13.8	-16.5
Uganda	17.2		14.8	-2.4

Sources: Calculations from Tables 4 and 10.

^1/As of June 1994.

The revenue impact of the CET in the cases of high and low import price elasticity are also estimated and expressed as ι percentage of the value of imports (Table 12). Khan (1974) demonstrated that import-price-demand elasticity in developing countries is high and close to unity. Accordingly, and to remain conservative, the -1.4 elasticity case seems the most reasonable estimate, and subsequent results are reported only for this case.

Table 12.

Estimated Revenue Impact of the CET Under Alternative Assumptions of Aggregate Import Price Elasticity

Sources: Calculations from Tables 4 and 10, using equation (11).

Table 12.

Estimated Revenue Impact of the CET Under Alternative Assumptions of Aggregate Import Price Elasticity

	Price elasticity of demand for imports
	(0)	(-0.88)	(-1.4)
	(As a percentage of the value of imports)
Burundi	-6.5	-5.7	-5.2
Kenya	-3.9	-3.5	-3.2
Tanzania	-16.5	-15.0	-14.2
Uganda	-2.4	-2.1	-1.9

Sources: Calculations from Tables 4 and 10, using equation (11).

View Table

Table 12.

Estimated Revenue Impact of the CET Under Alternative Assumptions of Aggregate Import Price Elasticity

	Price elasticity of demand for imports
	(0)	(-0.88)	(-1.4)
	(As a percentage of the value of imports)
Burundi	-6.5	-5.7	-5.2
Kenya	-3.9	-3.5	-3.2
Tanzania	-16.5	-15.0	-14.2
Uganda	-2.4	-2.1	-1.9

Sources: Calculations from Tables 4 and 10, using equation (11).

Using the ratios of imports to GDF and customs duties to GDP (from Table 13), the estimated share of total revenue from customs duties with respect to intraregional imports (Table 9), and an estimated efficiency ratio of import duty collection, ^21/ we calculate in Table 14 the revenue impact of the CET as a percentage of GDP, (Column 2); when added to estimates of the revenue impact of internal liberalization (Column 1), we obtain the total revenue impact as a percentage of GDP (Column 3). The results show that Burundi and Tanzania would experience the most important impact. Therefore, for fiscal policy to be neutral integration-oriented, the authorities in the countries under review must raise additional revenue equivalent to 0.95 percent of GDP in Burundi, 0.77 percent in Tanzania, 0.64 percent in Uganda, and 0.43 percent in Kenya. Otherwise, the revenue forgone will exacerbate the already existing fiscal deficits indicated in Table 2. Table 14, which breaks down the revenue impact, indicates that the implementation of the CET is likely to have the higher effect.

Table 13.

Ratio of Imports, Indirect Taxes, and Total Revenue to GDP, 1993

Sources: Data provided by the authorities; and IMF staff calculations.

^1/Covering FY 1992/93 (July-June)

^2/Value-added tax.

Table 13.

Ratio of Imports, Indirect Taxes, and Total Revenue to GDP, 1993

	Imports	Indirect taxes	Sales tax	Excise tax		Customs duties	Other indirect taxes	Tax revenue	Total revenue
	(In persentage of GDP)
Burundi	18.2	10.1	2.9	3.7		3.3	0.1	14.8	17.3
Kenya 1/	29.0	11.9	7.0	2.7	2/	2.3	0.1	18.3	22.1
Tanzania 1/	53.8	7.7	3.9	1.9		1.5	—	13.9	16.8
Uganda 1/	17.7	5.8	1.7	0.5		3.2	0.4	7.1	7.7

Sources: Data provided by the authorities; and IMF staff calculations.

^1/Covering FY 1992/93 (July-June)

^2/Value-added tax.

View Table

Table 13.

Ratio of Imports, Indirect Taxes, and Total Revenue to GDP, 1993

	Imports	Indirect taxes	Sales tax	Excise tax		Customs duties	Other indirect taxes	Tax revenue	Total revenue
	(In persentage of GDP)
Burundi	18.2	10.1	2.9	3.7		3.3	0.1	14.8	17.3
Kenya 1/	29.0	11.9	7.0	2.7	2/	2.3	0.1	18.3	22.1
Tanzania 1/	53.8	7.7	3.9	1.9		1.5	—	13.9	16.8
Uganda 1/	17.7	5.8	1.7	0.5		3.2	0.4	7.1	7.7

Sources: Data provided by the authorities; and IMF staff calculations.

^1/Covering FY 1992/93 (July-June)

^2/Value-added tax.

Table 14.

Estimated Revenue Impact of Regional Integration-Oriented Tariff Reform

Sources: Calculations from Tables 9, 12, and 13.

Table 14.

Estimated Revenue Impact of Regional Integration-Oriented Tariff Reform

	Revenue impact of removing tariff on internal trade	Revenue impact of introducing the CET	Total revenue impact
	(1)	(2)	(3) = (1) + (2)
	(In percent of GDP)
Burundi	-0.16	-0.79	-0.95
Kenya	-0.02	-0.41	-0.43
Tanzania	-0.07	-0.70	-0.77
Uganda	-0.28	-0.36	-0.64

Sources: Calculations from Tables 9, 12, and 13.

View Table

Table 14.

Estimated Revenue Impact of Regional Integration-Oriented Tariff Reform

	Revenue impact of removing tariff on internal trade	Revenue impact of introducing the CET	Total revenue impact
	(1)	(2)	(3) = (1) + (2)
	(In percent of GDP)
Burundi	-0.16	-0.79	-0.95
Kenya	-0.02	-0.41	-0.43
Tanzania	-0.07	-0.70	-0.77
Uganda	-0.28	-0.36	-0.64

Sources: Calculations from Tables 9, 12, and 13.

a. Theoretical considerations on the optimum tax structure

A tax structure is supposed to be: (i) efficient, by minimizing the distortion in the allocation of resources; (ii) equitable, through its redistributional role; (iii) of minimal cost with respect to resources required when assessing and collecting the tax; (iv) flexible, so as to allow needed adjustments; and (v) able to provide adequate resources to governments. This raises three issues: the respective weights of direct and indirect taxes in the government budget, the distribution of indirect taxation between external trade and domestic consumption-related transactions, and the structure of tax rates.

b. Estimates of some offsetting fiscal measures

(1) Indirect tax rates adjustment to offset the revenue impact

As shown in Table 13, the ratio of tax revenue to GDP in the countries under review is very low, ranging from 7.1 percent in Uganda to 18.3 percent in Kenya. Table 15 provides estimates of additional tax revenue required to offset the revenue losses from regional integration (Table 14) and maintain the fiscal policy stance unchanged in the countries under review. Increases in tax collection amounting to 2.3 percent of current tax revenue in Kenya, 5.5 percent in Tanzania, 6.4 percent in Burundi, and 9.0 percent in Uganda are called for. Furthermore, Table 15 also indicates percentages by which indirect taxes as a whole, sales taxes, and excise taxes would have to be increased to offset the estimated revenue losses.

Table 15.

Estimated Extra Revenue Needed to Achieve a Neutral Fiscal Policy in the Context of an Integration-Oriented Tariff Reform

Sources: Calculations from Tables 13 and 14.

Table 15.

Estimated Extra Revenue Needed to Achieve a Neutral Fiscal Policy in the Context of an Integration-Oriented Tariff Reform

	Extra revenue needed as percent of
	Indirect taxes	Sales taxes	Excise taxes	Tax revenue
Burundi	9.4	32.8	25.7	6.4
Kenya	3.6	6.1	15.9	2.3
Tanzania	10.0	19.7	40.5	5.5
Uganda	11.0	37.6	128.0	9.0

Sources: Calculations from Tables 13 and 14.

View Table

Table 15.

Estimated Extra Revenue Needed to Achieve a Neutral Fiscal Policy in the Context of an Integration-Oriented Tariff Reform

	Extra revenue needed as percent of
	Indirect taxes	Sales taxes	Excise taxes	Tax revenue
Burundi	9.4	32.8	25.7	6.4
Kenya	3.6	6.1	15.9	2.3
Tanzania	10.0	19.7	40.5	5.5
Uganda	11.0	37.6	128.0	9.0

Sources: Calculations from Tables 13 and 14.

The main features of the indirect tax systems of the countries under review are attached as an Appendix. Given theis actual fiscal situation, broad-based and far-reaching revenue-raising measures are needed to meet the required additional fiscal effort. Looking at sales taxes (and/or the VAT) and on the basis of the standard rate of the sales tax in each country, Table 16 indicates the required adjustment of the sales tax rates necessary to offset the revenue impact. ^23/ It is worth noting that only Kenya would be in a position to offset the revenue impact stemming from the regional integration-oriented tariff reform by raising the standard VAT tax rate by only 1.1 percentage points, from 18 percent to 19.1 percent. In Burundi, Tanzania, and Uganda, adjustments of the sales tax rates ranging from 4.9 to 7.5 percentage points would be required.

Table 16.

Estimated Adjustment of Sales Tax Rates

Sources: Calculations from Tables 13 and 14.

Table 16.

Estimated Adjustment of Sales Tax Rates

	Current standard rate	Required adjustment	Adjusted rate
Burundi	15	4.9	19.9
Kenya	18	1.1	19.1
Tanzania	25	4.9	29.9
Uganda	20	7.5	27.5

Sources: Calculations from Tables 13 and 14.

View Table

Table 16.

Estimated Adjustment of Sales Tax Rates

	Current standard rate	Required adjustment	Adjusted rate
Burundi	15	4.9	19.9
Kenya	18	1.1	19.1
Tanzania	25	4.9	29.9
Uganda	20	7.5	27.5

Sources: Calculations from Tables 13 and 14.

(2) Potential revenue gains from removal of import bans and fiscal exemptions

In import-compressed countries—where import bans and other import restrictions are a prime feature of the trade regime—removing quantitative restrictions and replacing them with tariffs could significantly improve government revenue. However, this is not the case for the countries under review, which have been liberalizing their trade regimes; and as a result, restrictions apply only on a few products. Their removal would not produce significant additional revenue. Nevertheless, the alleviation of the fiscal impact discussed above could result not only from the removal of import bans, but also from an improved management of fiscal exemptions and tax smuggling. Table 17 provides estimates of the extent of fiscal exemptions and tax smuggling (including misdeclarations, underdeclarations, and underinvoicing) as a percentage of imports (Column 3), by comparing the trade-weighted average tariff under existing tariff structures (Column 1), with effective rates of collection expressed as the ratios of customs duties to imports (Column 2). It also provides estimates of the extent of fiscal exemptions and tax smuggling as a percentage of GDP in each country (Column 4). Table 17 shows that the estimated revenue losses stemming from the tariff reform (Column 5) would be, with the exceptions of Burundi and Uganda, ^24/ significantly reduced, or even eliminated, by the removal of all fiscal exemptions as well as by a more effective control of tax smuggling. As mentioned before, the reform of the exemptions system and the reduction of the extent of smuggling are medium-term actions. Therefore, not much should be expected from this in terms of possible short-term revenue gains.

Table 17.

Estimated Extent of Fiscal Exemptions and Tax Smuggling, 1993

Sources: Calculations from Tables 11 and 13.

^1/Column 1 of Table 11.

^2/Calculated by dividing Column 5 of Table 13 by Column 1 of Table 13.

^3/Calculated by multiplying Column 3 of Table 17 by Column 1 of Table 13.

^4/Column 3 of Table 14.

Table 17.

Estimated Extent of Fiscal Exemptions and Tax Smuggling, 1993

	Trade-weighted average tariff 1/ (In percent)	Effective tariff rates 2/ (In percent)	Extent of Exemptions		Estimated Revenue Forgone
			In percent of imports)	(As percent of GDP) 3/	(As percent of GDP) 4/
	(1)	(2)	(3)=(1)-(2)	(4)	(5)
Burundi			21.6	18.1	3.5	0.6	0.95
Kenya	18.0	7.9	10.1	2.9	0.43
Tanzania	30.3	2.8	27.5	14.8	0.77
Uganda	17.2	18.1	(0.9)	0.2	0.64

Sources: Calculations from Tables 11 and 13.

^1/Column 1 of Table 11.

^2/Calculated by dividing Column 5 of Table 13 by Column 1 of Table 13.

^3/Calculated by multiplying Column 3 of Table 17 by Column 1 of Table 13.

^4/Column 3 of Table 14.

View Table

Table 17.

Estimated Extent of Fiscal Exemptions and Tax Smuggling, 1993

	Trade-weighted average tariff 1/ (In percent)	Effective tariff rates 2/ (In percent)	Extent of Exemptions		Estimated Revenue Forgone
			In percent of imports)	(As percent of GDP) 3/	(As percent of GDP) 4/
	(1)	(2)	(3)=(1)-(2)	(4)	(5)
Burundi			21.6	18.1	3.5	0.6	0.95
Kenya	18.0	7.9	10.1	2.9	0.43
Tanzania	30.3	2.8	27.5	14.8	0.77
Uganda	17.2	18.1	(0.9)	0.2	0.64

Sources: Calculations from Tables 11 and 13.

^1/Column 1 of Table 11.

^2/Calculated by dividing Column 5 of Table 13 by Column 1 of Table 13.

^3/Calculated by multiplying Column 3 of Table 17 by Column 1 of Table 13.

^4/Column 3 of Table 14.

a. Regional harmonization: Issues and difficulties

The removal of fiscal distortions between CBI countries would significantly contribute to economic integration, as much as the elimination of internal trade barriers and related border controls would. The path toward fiscal harmonization raises two broad issues: fiscal borders and harmonization of the taxable basis, the rate structures, and fiscal exemptions.

With respect to the fiscal borders issue, ^25/ indirect taxes may be levied on the basis of either the destination or the origin principle. According to the destination principle, on which WTO rules are based, all goods bear the taxes of the consuming country, regardless of where they are produced. A levy equal to domestic indirect taxes is imposed on imports, and exports are exempt from internal indirect taxes. Under the origin principle, all goods should be taxed in the country where they are produced. Domestic indirect taxes are then levied on exports and imports enter the country free of internal indirect taxes. Accordingly, the move to an advanced economic integration arrangement requires the adoption of the origin principle between members. In this case, fiscal borders disappear and internal fiscal border controls become unnecessary. However, this option requires a high degree of collaboration (to avoid evasion) and mutual trust between national fiscal administrations. Furthermore, there are budgetary implications in operating the origin principle. As a matter of fact, this system would lead to important budgetary transfers from countries that are net importers from their regional partners to countries that are net exporters, and create the need for a regional clearing system. On both grounds, this option is not a feasible one in the short term for the CBI countries.

The second issue is the harmonization of the taxable basis, the rates structures, and the fiscal exemptions. The domestic indirect tax systems in Burundi, Kenya, Tanzania, and Uganda actually differ in many respects (see Appendix). Kenya has already introduced the VAT as the sales tax instrument; Burundi, Tanzania, and Uganda have yet to do so. ^26/ Also, the taxable basis for the sales tax varies. In Tanzania, the taxable basis for the sales tax is the price inclusive of the import duty and excise tax. In Burundi, a taxe de service and a surtaxe à 1’importation are added to the import duty to make up the taxable price of imports. These two taxes share the protective role played by the import duty, since they do not apply to domestic goods. The rate structure varies widely from one country to another. Kenya has the greatest number of excise tax rates, and Uganda the greatest number of sales tax rates. The range of sales tax rates extends from the highest rate (40 percent) in Kenya to a zero percent rate in Tanzania and Uganda. The move to a uniform rate in these countries will certainly be difficult, considering the budgetary implications in countries with currently higher rates. Harmonizing excise taxes will prove even more difficult because they are applied to only a few products; countries producing them will favor low rates, contrary to consuming countries’ preferences. A more important obstacle in the way of harmonization is the loss of indirect taxes as an autonomous fiscal policy instrument.

b. Approaches to regional harmonization

On the basis of the above findings, the envisaged tariff reform should be supported by changes in the indirect tax systems of the countries under review, including: (i) the introduction of a VAT system in those countries that have not already done so, namely Burundi, Tanzania, and Uganda; (ii) a continued reliance on the levy of indirect taxes under the destination principle, given their present weak administrative capacity; (iii) the shift to a two-rate structure, with a standard rate applying to most goods, and a zero rate applying to a few products (some specific farm inputs) and exported goods, for example; ^27/ and (iv) a reliance on fewer excise duties at higher rates that would apply to luxury goods or on those goods whose consumption the authorities wish to discourage.

These recommended changes would certainly be revenue-enhancing, since domestic indirect taxes apply not only on imports but also on domestically produced goods. A narrow band would be determined, within which countries would have to fix their rates. ^28/ In Burundi, the taxe de service and the surtaxe à 1’importation would have to be removed after the rate structure is changed. The taxable basis would also have to be harmonized with the sales tax levied on a price that includes import duties and excises in all countries. ^29/ In addition, medium-term reforms should be implemented, with the view of effectively eliminating all unnecessary fiscal exemptions, ^30/ together with the establishment of reliable and effective safeguards against tax smuggling.