TARIFF ESCALATION AND AFRICAN COUNTRIES: WHO ARE THE REAL FRIENDS?
By
Alessandro Antimiani,* Michele Di Miao+ & Fransesco Rampa#
Abstract
Tariff escalation is said to be one of the major obstacles for developing countries to enjoy the gains from trade. For this
reason, tariff escalation is a key issue to analyse trade negotiations. Unfortunately measuring tariff escalation is not an easy
task at all. In this paper we provide different methodologies to measure it. Our analysis focuses on the trade relations
between nine African countries and some of the big world trade players: USA, China, India, Japan and EU. First, using the
WITS database, we study tariff escalation using descriptive statistics in order to show a broad picture of the tariff structure
which characterizes the bilateral trade between African Countries and those partners. Then we assess the effective rate of
protection using a CGE model through GTAP. We evaluate the tariff escalation by both the traditional ERP and the OERP
index by Anderson (1998). We compare the results from the two exercises showing differences and complementarities in the
two analysis presented in the paper.
1. INTRODUCTION
Market access remains one of the most important trading issues between developing and developed countries. Recently, the demand from
developing countries in terms of market access mostly focused on the reduction of distortions affecting trade in agriculture, in particular tariff
peaks, tariff dispersion and tariff escalation. Among the issues related to market access tariff escalation ranks among the highest. Tariff escalation,
i.e. higher import duties on processed products than on their input commodities, is one of the objects of controversy between developed and
developing countries. For importing developed countries, tariff escalation implies advantageously low rates of duty on imported inputs in relation
to the protection they receive against imported products competing with those they produce. On the contrary, for developing countries tariff
escalation is an obstacle to the move into higher stages of processing, shifts the economic activity toward primary production and away from
processing (WTO, 1996). Among the reasons for which moving to higher stages of production is beneficial are: 1) escaping the deterioration in the
terms-of-trade for primary commodities and the instability of primary product prices on international markets (Prebisch, 1959) ; 2) enjoying the
higher employment opportunities and higher profits associated with the production of manufactured goods (Yeats, 1984); 3) avoid the excessive
exploitation of natural resources and the damage of the environment caused by over-specialization in primary commodities (Hecht, 1997).
*
Italian Institute for Agricultural Economics.
University of Naples, “Parthenope”.
#
European Centre for Development Policy Management; Corresponding author, ([email protected]).
+
While developing have emphasised the necessity to eliminate tariff escalation and developed countries (US, EU, Canada) stated their willing to do
so, at the moment no solution or agreed solution has been defined. In fact, while if tariff escalation is said to be one of the major limitations to
trade-induced growth in developing countries and the debate on the issue is increasing, still there is no agreement on several aspect of the issue.
One of the reasons is that tariff escalation is not easy to evaluate.
This paper is s contribution to the understanding and measurement of tariff escalation. Our analysis focus on the trade relations between African
countries and some of the big world trade players: USA, China, India, Japan and EU. Three are the main element that differentiates this paper from
previous ones. First our analysis does not focus on specific products but the object of analysis are countries. We consider four LDC (Ethiopia,
Senegal, Tanzania, Uganda) and six non LDC (Botswana, Cameroon, Ghana, Kenya, Nigeria, South Africa) African countries. We evaluate the tariff
escalation they face with respect to the biggest world player. The second element of interest is that in our analysis we explicit take into
consideration the existence of preferential agreements between the country in our sample. This is particularly important when considering EU
(Chevassus-Lozz and Gallezot, 2003) and countries which have a number of different preferential agreement with developing countries. Third, this
is the first paper that presents an analysis of tariff escalation for African countries with respect to India and China, two countries which are
becoming increasingly important in their trade relations.
An interesting and challenging aspect of the analysis of tariff escalation (TE) is related to the difficulties to measure it. Moreover it is almost
impossible to synthesize the information in one number. While this is obviously not a specific problem of tariff escalation, in this case these
difficulties seem quite relevant because of the need to clarify the debate on this intrinsic complex issue. While when dealing with tariff each
aggregation has its own limits, in the case of TE these problems are much lager.
We try to make the information as clear as possible by presenting both descriptive tables with data for each country and proposing some synthetic
– even if quite naïve – aggregate measures. In particular we will use the most simple (and common) methodology to measure tariff escalation: we
use the difference between two subsequent stages of production (raw-intermediate; intermediate-final) as our preferred measure of tariff
escalation. Thus, in the first part of the paper we show a broad picture of the tariff structure which characterizes the bilateral trade between
African Countries and the big partners. We use data from WITS database and the MTN nomenclature in order to evaluate the tariff escalation. In the
second part of the paper we presents the results from a simulation exercise carried on to measure the Output Rate of Effective Protection (OERP).
We used GTAP to evaluate the magnitude of tariff escalation following a new methodology implemented in Antimiani et al. (2003) in order to give a
theoretical based measure of effective protection able to solve some of the problems related to the standard ERP index.
The structure of the paper is the following. In the next section we give a brief account of previous contribution to the tariff escalation literature. In
section 3 we present our dataset, the methodology to measure tariff escalation. Section 4 describes the main results of the descriptive analysis and
discuss which trade partner is the “best friend” of the sample of African countries in terms of tariff escalation. In section 5 we introduce the and we
present the results for a simulation exercise implemented by using GTAP. Section 6 concludes.
2. LITERATURE
Although tariff escalation has been reduced, it still persists in many commodity chains. Tariff escalation in agricultural markets is said
to be a major factor of difficulties for exporting developing countries, hindering export growth and diversification into processed products. Recent
analysis shows that in 12 out of 17 major commodity chains, significant tariff escalation exists, mostly at the first stage of processing (FAO, 2003).
This is the reason fro which tariff escalation is one of the important market access issues to be addressed in the current WTO negotiations on
agriculture. The March 2003 Draft of Modalities for the Further Commitments in the context of the WTO Agreement on Agriculture (AoA), proposes
steeper cuts in the higher tariffs.1
There is a large agreement among both economists and politicians on the importance of moving towards higher values-added production for
developing countries. The basic idea is that what you export matters (Hausmann et al., 2007). Following this idea, developing countries have
recently started to increasingly stress the need to move to value–addition production. To them, tariff escalation appears one of the stumbling
blocks in this direction. It follows that the issue of tariff escalation (TE) becomes crucial in evaluating the growth enhancing potential provided by a
reduction in tariff along the lines proposed in the last DOHA Round meetings. Furthermore, the presence of tariff escalation also significantly
counters the perceived preferences targeted at the LDCs.
Among the issues of concern in the current debate there are the implications of the proposed tiered tariff reduction formula on LDCs and the
quantitative assessment of tariff escalation on the main value added export products. In this paper we instead focus on a more fundamental and
preliminary issue. We aim at give a detailed picture of the tariff escalation faced by a group of Africa countries with respect to developed importing
countries. This task appear to be crucial in order to make the discussion well informed and based on the most accurate description of the situation.
To the best of our knowledge such an analysis is still missing for most of Africa countries and thus the gap must be filled. Our paper is meant to be a
contribution in this direction.
Two are the main methodological issues in measuring tariff escalation. The first concerns the way to construct the categories of products to be
evaluated and the second one is tarification. Concerning the data and the method employed to measure TE, the literature has shown a large variety
of dataset from which compute tariff values. Since recently, one big challenge for TE analysis was the issue of non-ad valorem tariffs. Most studies
in past years have done a considerably effort to implement the most accurate conversion of non-ad valorem tariffs into ad valorem equivalents
(AVEs). Luckily there are now easily available data on ad-valorem transformation that can be directly used to compote tariff escalation.
In the present study we used the TRAINS dataset. Chevassus-Lozz and Gallezot (2003) show that preferences are to be taken into consideration to
correctly evaluate the tariff escalation. Indeed the issue of tarification very important as we are discussing access of developing countries to the
European market: this requires that we take into account all preferential agreements that the EU has signed with its partners. In this paper we
have take this in deep consideration this issue thanks to the information contained in the TRAINS dataset.
The second issue concern the way in which product categories are constructed. Again, there are no official method, nor any real consensus that
would help determine which list of products are used in the manufacturing of a processed product. The studies carried out by the WTO (1996) and
the OECD (1999) used the BEC classification. They considered all products, including agricultural and Agri-food products imported by thirty
countries. In both cases the analysis remain at a highly aggregated level by distinguishing three stages of transformation (raw products, semiprocessed and fully processed products). A selection of individual product pairs was then made for each of the three processing stages in order to
1
The last proposal is the following: where the tariff on a processed product is higher than for its primary form, the proposed tariff reduction for the processed
product would be equivalent to that for its primary form, multiplied by at least a factor of [1.3].
identify nominal tariff escalation between them. Furthermore, talking about LDCs, a limit of the BEC classification is that for agricultural and food
industry there are only four categories. Since often many LDCs are highly specialized in some sectors, by the BEC is hard to figure out the level of
tariff escalation.
A largely used source of categorization is the FAOSTAT classification system. It comprises 226 processed commodities, that is 377 commodity
pairs. Each pair represents a processing relation between one input commodity and one output commodity. For instance, Lindland (1997) uses this
method determining tariff escalation at the level of product “pairs”, distinguish 26 product pairs at an eight-digit level.2 Other works have instead
use the scheme developed by the World Bank. Among these, Amaji et al (1996) measure tariff escalation at primary product level to analyze the
structure of European tariffs on 19 primary commodities exported by Africa. Finally, some studies choose specific pair of product and construct the
relationship between primary products and processed ones (see for instance Sharwa, 2006, Chevassus-Lozz and Gallezot, 2003). For instance,
Chevassus-Lozz and Gallezot (2003) adopt two different methodologies. First they use the BEC classification. Then they use a modified BEC
employing information from experts to determine the technical relationship between the products considered. They cover all agricultural and agrifood products, but theey measure tariff escalation, not for pairs of products but at a more aggregated level, by distinguishing three stages of
transformation per commodity processing chain. They end up with re-classification of the first 24 chapters of the CN8-digit nomenclature (i.e. all
agricultural and agri-food products) by combining two approaches. Similarly Sharma (2006) chooses some specific product line and aggregates
them in primary and processed. In the present study we use the official classification elaborated from the WTO board, namely the MTN.
3. DATA AND METHODOLOGY
3.1 Datase
As for the tarification we employed the TRAINS dataset. As for product classification, we choose to use the MTN (WTO Multilateral Trade
Negotiation) classification elaborate by the WTO. Two are the main reasons for this choice. First, by using an official classification we minimize
arbitrariness in the choice of product categories and product process lines. Second, we believe that adopting an official classification is a first step
in the direction of making possible comparability across studies and allow more easily knowledge accumulation on this topic. The MTN aggregates
Agricultural and Industrial Products into broad categories of interest such as ‘Fish and Fish Products’, ‘Tropical Beverages’, Transport Equipment’,
‘Electric Machinery’, ‘Petroleum’, etc. For the purpose of our analysis, MTN is relevant since it covers 3 stages of processing: raw, semi, and finished
product categories, each of which is subdivided into agriculture and industry3. With respect to other approach MTN is sufficiently disaggregated
(23 chapter plus petroleum or 51 headings plus 2 for petroleum) in order to evaluate trade and/or tariffs for stage of processing but not deriving
from “author’s choices”. Regarding the tariff escalation, having 3 stage of processing allows better analysis of the structure of trade policies.4
2
3
For a few products more than two processing stages are taken into account (durum wheat – durum wheat flour – pasta without eggs for example).
BEC take into account only 2 stage of processing.
For example, there may be cases of tariff escalation between raw and intermediate while descalation between raw and final. Consider the following filiera: 20%
on raw, 40% on intermediate and 10% on final. If the country who apply this scheme is highly specialized in producing and trading intermediate goods it protect firms at
intermediate stage by tariffs and firms at final stage by the national production of input. If we evaluate the escalation, we have 20 percentage points of escalation
4
3.2 Methodology: measuring tariff escalation
The first objective of our paper is to present an evaluation of the tariff escalation for all LDC African and three non-LCD African countries. 5 In the
literature numerous methodologies have been implemented. Most studies quantify TE on the basis of the bound tariffs. Some studies instead
consider applied tariffs to analyse the extent of TE in practice. Other use trade (import) weights to average tariffs, bound or applied, where many
tariff lines define a product. Lastly, some studies have attempted to measure TE also using the concept of effective protection (e.g. Lindland 1997;
Burman and others 2001).6
The analysis presented below is based on nominal applied tariffs. We (as among others in FAO (1997), Sharma (2006)), measure nominal tariff
escalation on the basis of nominal tariff wedge, the difference in tariffs between two subsequent stages of production. Nominal tariff wedge (TW)
for a given period is defined as:
TW = T – t
(1)
where T is tariff on the given stage of production and t is tariff on the previous stage. Three situations can be characterized based on the tariff
wedge (TW):
• Tariff escalation: TW > 0
• Tariff de-escalation: TW < 0
• Tariff parity (neither escalation nor de-escalation): TW = 0.
4 - WHO ARE THE REAL FRIENDS?
4.1 - Tariff values
Since we used the WITS database, we considered all the tariff lines for which there is trade between the importing country and our sample of
African countries. Thus cases in which the country gives preferences but import is zero are not reported. Chevassus-Lozza and Gallezot (2003) we
chose to take into account all preferential agreements signed by all the importing countries we are considering. 7 This implies that we base our
between raw and intermediate and –30 percentage point between intermediate and finale, so it could be the case of descalation by 10 percentage point. Availability of
data on all the stages allows us to detect such cases.
5
Since access of LDC to EU occurs essentially through preferential agreements, some developing countries are worried that their advantage is eroded by the tariff
cuts of MFN duties to be decide in the WTO negotiations. We will not discuss this issue in this paper.
6
Effective protection is usually considered a better indicator of trade protection than is nominal protection is, especially where a processed product is produced
from multiple primary products. But in practice its use is very difficult. Indeed effective protection for large number of products or tariff lines it is very difficult to be
computed as this requires input-output coefficients. More on this below.
7
This is an important element in the analysis. Indeed Chevassus-Lozza and Gallezot (2003) show that taking into account preferential agreements shows that in the
year 2000 the phenomenon of progressivity of duties of the EU vis a vis developing countries is relatively minor except for countries benefiting from the GSP.
analysis on the tarification applied – because we are using TRAINS by WITS - and not on the commitments made by the these countries the WTO.
Thus our tariff are applied tariffs as measured by MFN tariffs in the WITS database.
4.2 - Tariff escalation
We begin describing Table 1 and 2. The Tables reports the tariff escalation that the non LDCs (Table 1) and LDC( Table 2) African countries in our
sample face with respect to the importing countries. Each cell report the TW for the corresponding MTN category.
Two observations concerning the methodology we adopt to construct the tables are in order. First, when there is no preferential tariff for a given
stage, we have consider the MFN tariff. Thus when there is no preferential tariff for the previous stage, we computed the tariff escalation as the
difference between the preferential for the last stage and the MFN of the previous one. Second, in the case of descaltion we report TE to be zero.
As for comparing the TE, we choose not to weight the tariff by import value or to implementing any other weighting scheme. We compare tariff
escalation across importing countries by using the simple average TE for the 25 MTN categories. The reason why we choose the simple average
rather than a more sophisticated one is that of endogeneity. Since a low import (low weigh) may be caused by a high tariff, the weighted measure
of tariff escalation is likely to be biased. Thus we preset simple average of the tariff escalation for our list of categories. This approach implies that
in comparing tariff escalation across African countries, we are making a strong assumption. We are evaluating TE assuming that if TE was the same
across countries these would export the same products in the same amount. Thus what counts in the present analysis is only the simple average:
we do not consider the importance that a specific product may have in the country’s (potential) export vector.
Table 1: Non-LDC countries: Tariff escalation for 25 MTN categories.
Note: For each exporting country, each cell is the TE by importing country for the corresponding MTN category calculated according to equation
(1). Average is the simple average by importing country for all the products. Number of categories gives the number of MTN categories for which
TE is positive. Percentage is the percentage of MTN categories with positive tariff escalation for importing country.
Table 2: LDC countries: Escalation for 25 MTN categories
Note: For each exporting country, each cell is the TE by importing country for the corresponding MTN category calculated according to equation
(1). Average is the simple average by importing country for all the products. Number of categories gives the number of MTN categories for which
TE is positive. Percentage is the percentage of MTN categories with positive tariff escalation for importing country
As it clearly emerges from Table 1 and 2 for each exporting country a large number of categories are under
tariff escalation (from 0% to 80%) and all importing countries impose TE for some category.
We now move to compare of TE across importing countries. Table 3 reports the average TE by importer
country for each African country in our sample. It results that the average TE is lowest in the case of EU and
highest in the case of China which is the trading partner of African countries displaying the highest average
TE, i.e. providing the most significant disincentive to process the goods to be exported. USA Japan and India
is the order of the countries in between the two extremes.
Table 3: Average Tariff Escalation by country of import
Usa
China
Japan
India
EU
Botswana
1.83
4.17
2.91
2.70
0.00
Cameroon
1.32
4.17
3.67
2.70
0.00
Ghana
1.98
4.17
3.36
2.70
0.00
Kenya
2.16
4.17
2.51
2.70
0.00
Nigeria
1.66
4.17
2.74
2.70
0.00
South Africa
1.97
4.17
3.68
2.70
0.67
Average Non LDC
1.82
4.17
3.15
2.70
0.11
Ethiopia
2.09
4.42
3.03
2.70
0.00
Senegal
1.75
4.94
2.61
2.70
0.00
Tanzania
1.73
4.42
3.49
2.70
0.00
Uganda
1.84
4.42
2.96
2.70
0.00
Average LDC
1.85
4.55
3.02
2.70
0.00
AVERAGE
1.83
4.32
3.10
2.70
0.07
Non LDC
LDC
Note: average is the simple average of the TE imposed by the importing
country on the exporting one calculated on the 25 MTN categories.
This comparison is of aggregate nature only, but is in line with the general characteristic of trade policy
regimes offered by the major world trade players to African countries8. The EU preferential tariff structure
is widely considered to be very open vis a vis African countries: LDCs can export through the Everything but
Arms regime or an equivalent treatment if they signed the Economic Partnership Agreements -EPAsrecently negotiated with the EU. This means that they face zero tariffs for all exports a part from arms and
few agricultural commodities (such as rice, sugar and fresh bananas, that are in the process of becoming
duty and quota free as well). Non-LDC countries that signed EPAs face tariffs equivalent to those regulated
by the Cotonou Agreement, with substantial share of duty and quota free access and other tariffs considered
generally very low, and in coming years will be all reduced to duty free quota free; African non-LDCs that
did not sign EPAs export under the EU's GSP. The USA and Japan also offer preferential trade schemes to
African countries, though with more limited product coverage compared to EU's regimes. Our result that the
US displays the second best average TE could be explained by the fact that in addition to its GSP Scheme the
USA import goods from African countries under the AGOA regime, offering duty and quota free access on
several products. China's tariff preferences are granted to only few African countries while India has no
preferential scheme for Africa.
In addition to comparing the overall average TE imposed by the five trading partners on the sample African
countries it is also interesting to look at the TE profile of each of the five major players. Only the EU and
India display homogenous TE across African countries, with the exception for South Africa in the case of the
EU since their trading relations are regulated by a bilateral FTA (called TDCA). The other major trading
partners show a varying degree of TE. It is important to note that LDCs receive no better treatment by the
five partners in terms of TE compared to non-LDC, despite the fact that some of them do offer a better
preferential scheme in general to LDCs (such as EBA) and that WTO (as well as UN and World Bank)
strongly encourage OECD countries to include special and differential treatment to LDCs in all their trade
policies and agreements. This is the case for India and the EU, where at least for the EU there is no TE at all
and for India LDC are treated no worse than non-LDCs. It is not the case for rich countries such as USA and
Japan where in many cases LDCs are worse off than non-LDCs: a country like Botswana (with a much wider
manufacturing sector and recognised export competitiveness than LDCs) is imposed a much lower TE than
LDCs like Ethiopia in the first case and Tanzania in the second, while it is these countries that need no
disincentive to move up the value chain through international trade regimes. It is the case for China: our
results show that consistently the group of LDCs face a higher escalation than non-LDCs, and the TE profile
seems exactly dual with respect to these two groups with two uniform TE (with the particularly high TE for
Senegal receiving even worse treatment than other LDCs).
The TE profile of China merits a special consideration given the important role this country has increasingly
acquired in the African continent, as major trading partners and investor. Given the size of the Chinese
economy, and the potential for future growth of imports from Africa , with spillovers on income generation
in African countries, the Chinese Government should consider liberalizing its import regime vis a vis African
countries, starting with the LDCs. This would be beneficial and in line with the decision to become a major
cooperating partner of Africa, as value addition in African exports going to China would entail also increased
interaction in terms of investment with African and Chinese enterprises working together with benefits in
terms of capital, technology and knowledge transfer as well as much needed financial resources for the
African processing sector. The apparently worse treatment by China of LDCs compared to their non-LDCs
neighbours should also be urgently addressed.
China, and India that has not yet adopted any preferential scheme for African exports should do more in
terms of eliminating their TE for Africa. However, the fact that USA and Japan display significant TE is in a
way more serious than for countries like China and India that are still considered developing countries
included by the WTO and may have non-trade concerns such as food security and poverty reduction in their
8
Again, our analysis does not take into account non-tariff barriers to trade: whether goods from African countries can
actually be exported under the preferential tariff regime we describe depends in reality on the capability of the exporter to
satisfy rules of origin and SPS or technical standards. Our measures of TE do not compare the impact of such non-tariff
barriers on the actual incentives offered by the five trading partners to African countries.
trade policy design. The so-called QUAD countries (EU J US CAN) are the largest traders in the world and the
major importers of African exports and are under pressure by public opinion and international
organisations to relax their trade polices to support African plans to move out of merely exporting primary
goods towards manufacturing with increasing value addition. QUAD countries do offer GSPs and have
pledged to even improve their treatment to Africa but our result show that, with the exception of EU, much
more needs to be done in terms of elimination of TE.
The above findings and observations could lead to a direction for policy recommendations and reform. The
large trading partners, in the spirit of improved cooperation with poorer countries, should converge
towards the best possible preferential regime for African countries, EU’s. At least OECD countries should
make these improvements a priority and as they have maintained and stated many times that they want to
support export-led poverty reduction strategies to African countries also through trade policy. In particular
they should offer LDCs more favourable treatment also in terms of TE to give an incentive to the African
economic actors to move up the value chain. This is included in the Special and differential treatment
principle of WTO treaties and should be reflected in all trade polices including TE.
5. AN ALTERNATIVE: DETERMINING THE UNIFORM “REAL” TARIFF USING THE OERP INDEX
In the literature, the issue of tariff escalation is sometimes strictly related to the one of effective protection.
The two concepts are related even if quite different. It is easy to show that an effective protection of the
production process has not to be the result of the presence of tariff escalation. But, conversely, a situation of
tariff escalation definitely reveals an effective protection of the activity of transformation. The two concepts
are also similar in that as there is not a unique or commonly agreed upon method to calculate a synthetic
measure to describe TE, the same difficulties are present in the case of ERP. While the ERP has some
advantages with respect to measuring TE as the difference between tariff in the different stages of
production, it also suffer severe methodological limitations. Iin order to give the broadest view on the
analysis of the issue of tariff escalation, we now move also to measure the effective rate of protection for our
sample of importing countries. But we do not go for the standard ERP and we take an alternative approach.
Before that we briefly describe the standard ERP measure and its limitations.
5.1 Some problems related to the concept of ERP
The effective rate of protection can be defined as the proportional increase in the price of a sector’s gross
output relative to free trade. Since the total value of gross output priced at value added per unit equals the
total value of net output valued at equilibrium prices, an appropriate price for gross output is the valueadded per unit. Accordingly, the effective rate of protection (ERP) of industry j (Ej) measures the increase in
industry’s value added per unit of output under protection (Vj’) as a percentage of the free trade value added
per unit (Vj):
Vj' Vj
Ej 
.
Vj
(2)
Assuming that one unit of output j necessitates the use of aij quantity of inputs i, we can write:
*
*
V
p

a
p

j
ij
j
i
i
*
1tj
1
V
p
a
p

ti
ij
j
i
'
j
*
i
*
*
If cij aijpi pj is the cost share of input i in output j, after simplification we get:
(3)
tj 
c
ti
'

ij
V
V
j
j
i
E

.
j
V
1

c

j
ij
(4)
i
The traditional definition of the ERP is based on restrictive assumptions (fixed coefficient and/or
separability) regarding the production functions (Anderson and Naya, 1969). If the assumption of fixed
physical input coefficients does not hold, free trade input-output coefficients must be inferred from the
observed distorted coefficients (Bureau and Kalaitzandonakes, 1995). The fundamental theoretical critique
moved to the effective protection concept, though, stems largely from concerns about drawing general
equilibrium inferences from a partial equilibrium measure (Ethier, 1971, 1977; Bhagwati and Srinivasan,
1973; Davis, 1998). The development of the concept of effective protection, as a matter of fact, may be seen
as an attempt to define the index as a pure production concept – expressed in terms of nominal prices and
input coefficient – making enough assumptions so that demand might be ignored: "Effective protection is
the ranch house of trade policy construction – ugly but apparently too useful to disappear" (Anderson,
1998). Also in terms of the possibility for the ERP to be a good predictors of gross outputs change, effective
protection is a partial equilibrium index, since in reality the prices of primary (non-produced) factors are
endogenous, and the prices of (internationally) non-traded goods may change as well. As a consequence,
even if the fixed coefficient assumption is met, ranking effective rates may not allow ranking percentage
output changes: a non-prohibitive import tariff or export tax in partial equilibrium might become
prohibitive in general equilibrium (Anderson, 1970).
5.2 Our analysis
To measure the effective protection granted by a country’s trade policy regime, one needs to overcome
some important hurdles. As we said, among these one of the main issue is to have a single “measure” to
compare trade policy in term of effective protection. In order to solve this problem using theoretically
sound aggregation procedures, it is necessary to specify the type of information we want to maintain: the
single “measure” has to be equivalent to the original multiple data with respect to the dimension we are
interested in. According to Anderson and Neary (1996), a general definition of a policy index is as follows:
depending on a pre-determined reference concept, any aggregate measure is a function mapping from a
vector of independent variables – defined according to the policy coverage – into a scalar aggregate. The
greatest advantage of this approach is that it is theoretically consistent, since the equivalence is determined
according to a fundamental economic structure. Secondly, it provides unequivocal interpretation of the
results, since the definition and properties of these “equivalence-based” indicators are predetermined. To
cope with some of the problems related with the standard definition of ERP, Anderson (1998) suggested an
interesting new definition of an ERP index: the distributional effective rate of protection, which is the
uniform tariff which is equivalent to the actual differentiated tariff structure in its effects on the rents to
residual claimants in a given sector. The same approach can be used in order to define an index which is
able to measure the impact of protection on the ability of sectors to compete with other industries in factor
markets: the Output Effective Rate of Protection (OERP), which is the uniform tariff on all distorted sectors
which produces the same level of output, sector by sector, as does the initial differentiated tariff structure
(Anderson, 1998).
In this section we apply the Output Effective Rate of Protection (OERP), that is an index which, focusing on
gross (rather than net) output, is able to take into account the role of the protection provided to the
intermediate inputs. The OERP ej of sector j in general equilibrium is defined as the uniform tariff which
exert on the output of j an effect which is equivalent to the initial tariff structure. That is
  
  
e e e
00 0
e
*
e
Y
,
w
p
,
v

Y
,
w
j:
jp
j
j
j p
, with pj pj1ej .
(5)
where Yj is j supply function, and w is the vector of competitive factor prices (w is function of the price
vector p and of the fixed factor supply v). The previous definition is based on the "small country"
assumption. If we want to allow for endogenous world prices, we need to define the vector p as a function of
the tariff vector (t). Equation (11) becomes:
 





 


w w
*
ew
*
00
*
00
*








e
:
Y
1

e
p
t
,
w
1

e
p
t
,
v

Y
1

t
p
t
,
w
1

t
p
t
,
v(6)
j
j
j
j
j
j j
j j
j j
w
where ( e j ) is the OERP uniform tariff with endogenous world prices.
Output variations across sectors reflect both the structure of protection (which the “standard” effective
protection index tries to measure) and differences in the production structure of the economy. The two
questions, ‘how much protection is given’ and ‘how much does supply change as a result’ are distinct, and
the OERP gives a precise answer to the latter.
Nowadays, the development of computable general equilibrium models implies that the ERP can be
computed as a general equilibrium index summarising all the model information (Stevens, 1996). In this
section, we assess the effective (rate of) protection using a CGE model through GTAP. The main difference
with the standard ERP is that our measure is derived in the context of general equilibrium. Simulations have
been carried out using the Global Trade Analysis Project (GTAP) model and its database, based on the latest
release of version 7, providing a baseline with reference to the year 2004. The GTAP model is a computable
general equilibrium (CGE) model based on Input Output tables (IO) representing 57 activities and 113
countries. Since we are interested mainly in understanding the level of effective protection we use the 2001
database of GTAP, which includes a new input-outputs table for Africa, updating the tariffs to 2004. On this
baseline we evaluate the tariff escalation by both the traditional ERP and the OERP index by Anderson. In
our version the database was aggregated in order to include 14 regions/countries and 19 sectors (Table 4),
taking all the raw agricultural products disaggregated and having a single sector for the food, for which we
want to evaluate the level of effective protection.
Table 4: GTAP data accounts
Commodities and Activities
Paddy rice
Wheat
Other cereals
Vegetables and fruit
Oil seeds
Sugar cane and beet
Plant based fibers
Other crops
Cattle, sheep, goats and horses
Other live animals
Raw milk
Wool and silk
Forestry
Fishing
Minerals
Food sector
Wearing sector
Manufacture
Services
Factors
Land
Labour skilled
Labour unskilled
Capital
Natural resources
Regions
China
India
Japan
USA
EU_25
Uganda
Tanzania
South Africa
Senegal
Nigeria
Ethiopia
Botswana
Sub-Saharan countries
Rest of the World
Trade policy at the tariff line level implies a level of detail by far higher than any existing model can allow
for: the EU tariff schedule, for example, includes more than 10,000 tariff lines. Therefore, in order to reach
the consistency between the information on trade distortions and the model aggregation it is necessary to
compute some kinds of average tariffs. However, the quality of trade distortion data included has
considerably improved compared to those in the previous release, due to the use of the MacMap-HS6: a
database including HS-6 level details, providing consistent and exhaustive ad valorem equivalents (AVEs) of
applied border protection across the world. This improvement allows considering applied/preferential
tariffs rather than bound ones, and includes the AVEs of some NTBs (Bouët et al., 2005). A positive OERP
means that a positive uniform rate would be needed in order to maintain the output of the given sector, i.e
the food sector should be “protected” in a bilateral free trade scenario. A negative OERP means that the
sector would be fairly competitive even in the absence of any protection.
In order to complete the analysis we also evaluate the OERP for the wearing and manufacturing sector9.
Table 5 reports the results of our analysis of the OERP index. The data show that – using OERP as our
measure of effective tariff protection - USA is the best friend of the our sample of African countries, followed
by EU_25 and China. India and Japan show the highest effective rate of protection with respect to the African
countries considered. The reason is that India has no preferential scheme while Japan shows in some cases
the second highest average tariff escalation after China. While China show almost the highest tariff
escalation in all the cases, the third position is not surprising. Indeed China became a WTO member in 2001,
finishing his access by 2004, so with a generalized MFN low tariffs. In other words, the general equilibrium
analysis show that China do not have a significant effective protection probably due to a relative “world
wide” high efficiency.
Difference in efficiency explains also some results like the value for USA with respect to Tanzania (41%)
and the one of EU_25 with respect to Ethiopia. These results are not quite surprising if we take into account
the difference in the efficiency in the food industry technology. As mentioned above, for USA and EU the
food sector would be fairly competitive even in the absence of any protection, so that an import subsidy
9
Due to a computable time limit (many sectors and countries considerably expand time running of the model) and to
the desegregation between raw and final of manufacturing products in the GTAP database (there is not a clear division
between input and final goods in the manufacturing sectors) we plan to perform such an analysis in a second step.
would be necessary in order to maintain output unchanged. On contrary, the effective protection for USA
and EU with respect to South Africa is high, since the degree of (or the capability) competition of South
Africa is higher compared with the other countries. To better understand this point it is worth to note
results for USA with respect to Uganda and Tanzania. While the USA trade policies toward these countries is
almost equal, the results in terms of OERP is strongly different, due to the different value added structure of
the two African countries.
Considering (the average for the) food sector the best friend of our sample of countries US followed by
China and EU while India and Japan are last. It is not surprising for India and for Japan. For India both
because of the tariff scheme which has not preferential regimes and probably because the efficiency in food
sector is still growing but is not yet so high. For Japan it is quite know that this countries has one of the
highest tariffs scheme and about preferential regimes is more focused on Asia partner than African. EU and
US have a similar level and it something like we expected due to a strong preferential regime and high
efficiency in food sector compare to African countries. On other hand is not intuitive the position of China
but it could be explained by the low MFN tariffs world wide and the high capability to compete on world
trade by state trading. Furthermore, it must be stressed that compared to USA and EU the supply of raw
input for food sector is more inland for china than USA and EU. Looking at the results for wearing and
manufacturing sectors they more intuitive. For manufacturing sectors the general level is quite low since
tariffs are generally low. It is interesting to stress the values for India which emphasized the “new
industrialized country” role of India. Even if tariffs on manufacture are not higher than other countries
probably the level of valued added in this sector is still low and then the effective protection is relative high
Table 5 - Output Effective Rate of Protection (OERP) (2004) – All sectors
Note; Average is simple average by importing country. * For South Africa is not possible to find an uniform tariffs equivalent to the actual tariffs schedule
imposed by India.. Source: own calculation on GTAP.
Finally, it is interesting to look at same results for the wearing sector for EU25, where with
respect to some countries like Ethiopia, Tanzania and South Africa, the level of OERP is high,
probably because the difference in tariff between raw materials and final goods and because
wearing is a high labour intensive sector and then developing countries could be
competitive. This last results also explains the general high level of OERP fro wearing
compared with food sector and manufacture.
1. CONCLUSIONS AND FURTHER RESEARCH
Developing countries are denouncing the tariff escalation practised by developed countries
as an obstacle to the development of their processing industry. This situation is particularly
important for agricultural and food products. Our analysis of tariff escalation shows that the
EU markets are the most open to developing countries thanks to the many preferential
agreements that the EU has signed with African countries. Using an alternative methodology
– the OERP index – we instead found that – for a smaller sample of African exporting
countries - the best friend is USA, followed by EU and China. Even if interesting, these
results need several refinements and further analysis to include more countries and a more
detailed sectoral analysis. This is left for future research.
Some policy recommendations drawing on our results on today's tariff escalation profiles
are highlighted in the paper. However new waves of preferential trade agreements – in
particular related to regional integration and North-South FTAs- are occupying the
negotiating, political and academic fora. Clearly, any new proposal of reduction of duties
leads to a reduction of the tariff escalation. However, as this paper demonstrated, these
liberalization proposals have to be evaluated also in the context of existing preferential
agreements; nominal improvements in the tariff structures faced by African exporters, as in
the cases of EU preferential schemes for Africa, do not seem to have led to significant
improvements in the incentives provided for processing of goods to be exported. Identifying
the real friends is not a clear-cut exercise, but certainly a much needed one. Consequently,
developing countries should not be pressurized to quickly negotiate and implement trade
reforms, but left sufficient time to analyse in depth new liberalization proposals, since the
devil is in the details, and the economic impact of changes in the tariff regimes of major
trading partners, including the escalation profile, may not be unequivocal.
In future research, the implications of today's tariff structures and of their possible future
modifications on investment patterns should also be considered. It will be particularly
relevant to investigate the possible relationships between the FDI flows from USA, China,
India, Japan and EU towards African countries and the sectoral patterns of tariff escalation
they face.
References
Anderson J., Naya S. (1969), Substitution and two concept of effective rate of protection, The
American Economic Review, volume 59, issue 4.
Anderson J., (1998), Effective protection redux, Journal of international economics, 44, pp.
21-44.
Antimiani A., Conforti P., Salvatici L. (2003), The effective rate of protection of European
agrifood sector, paper presented at “Agricultural policy reform and the WTO: where are we
heading?”, Capri, Italy, 23-26 June 2003.
Bhagwati J. N., Srinivasan T. N., (1973), The general equilibrium theory of effective protection
and resource allocation, Journal of International Economics, n. 3, pp. 259-282
Bouët A., Decreux Y., Fontagné L., Jean S., Laborde D., (2005), Tariff Data, Documentation
GTAP Database version 6, draft.
Bureau J.-C., Kalaitzandonakes, N., (1995), Measuring effective protection as a superlative
index number, American Journal of Agricultural Economics, Vol°77, May.
Burman, C., Johansson, K., Karlsson, A., Loxbo, H., Norell, B., Mattson, Y., & Wilhelmsson, M.
(2001). Tariff Escalation for Agriculture and Fishery Products: Swedish Board of Agriculture