Domestic Competition and Export Performance of Manufacturing Firms in Cote d'Ivoire

Revised Version
July 2000


Domestic Competition and Export Performance
of Manufacturing Firms in Côte d’Ivoire

by

Jean-Paul Azama,b,c,
Marie-Françoise Calmettea,b,
Catherine Loustalana,
and Christine Maurela,b.

a: ARQADE, University of Toulouse,
b: IDEI
c: Institut Universitaire de France.


WPS/2001-1






Abstract: Because of transportation costs, African manufacturing firms benefit from some
market power on their domestic market, where they can charge a higher price than the export
price, net of transportation cost. We present a simple theoretical model of an exporting firm
that discriminates between the export and the domestic markets, where firms engage in
Cournot competition. It is then shown that the impact of increased competition on export
performance by the firms is ambiguous, and may be negative for a non trivial range of
parameter values. Using survey data on Ivoirian firms, our empirical analysis gives some
support to this prediction, showing that the probability of a firm exporting decreases with
increased competition.

Acknowledgements: We wish to thank Jean-Jacques Laffont and Patrick Rey for useful
discussions, as well as participants at the ARQADE-RPED lunch workshop in Toulouse and
at the CSAE Conference 2000 in Oxford, and in particular Simon Appleton and Arne Bigsten,
for useful comments, without implicating. Remco Oostendorp also offered useful comments
on a summary version of this paper at the ABCDE-Europe conference in Paris, which are
gratefully acknowledged, with the same caveat.

---

1
1. Introduction
The need for diversifying the exports of African economies is well known, as most of
these countries rely on exporting a small number of primary commodities on highly volatile
markets. A recent series of case studies has shown how the resulting trade shocks are liable to
destabilise these economies, especially when the government responds by inappropriate
policies (Collier, Gunning et al., 1999) ; this occurs quite frequently. The whole theory of the
‘Dutch Disease’, and the associated de-industrialisation, was motivated by this type of issues.
In particular, Van Wijnbergen (1984) suggested that the commodity shocks have a lasting
detrimental dynamic effect on industrialisation, as the Dutch Disease retards the process of
learning by doing in the industrial sector. However, commodity booms often entail a massive
inflow of foreign currency, and there are examples of economies, like Mauritius, that have
used the funds accumulated during such trade booms, concerning sugar in this case, for
financing a diversifying industrialisation of their countries, with very positive effects on
growth. Consequently, the impact of these shocks tends to work either way, and does not
come out significantly in single-equations cross-country growth regressions (Deaton and
Miller, 1996), probably because it is conditioned by the response of the government, and
hence by the political economic circumstances of the country affected. Nevertheless, the case
for export diversification in developing countries in general seems to be supported by most
observers. The jury is still out for determining whether manufacturing industries will provide
the right solution for diversifying African economies, or whether services or other industries
will be more appropriate.
The aim of the present paper is to contribute some empirical elements for evaluating
this issue, by analysing the determinants of the export performances of manufacturing firms in
one of the most industrialised economies of this continent, namely Côte d’Ivoire. We focus on
the links between domestic competition and export performance. A widespread argument in
the ‘export-led-growth’ literature is that export-orientation, especially in the non-traditional
sectors, is the key to fast growth and industrialisation for developing countries. In many
popular expositions of this doctrine, exporting firms are regarded as necessarily competitive,
and hence, progressive. However, the relationships between export performance and
competition are not as straightforward as generally thought. This is discussed in the next
section, using a simple model of the type used in the literature on ‘dumping’ (e.g. Deardorff,
1990). In this model, firms are able to discriminate between the domestic market and the
export market, charging a different price on each segment. However, the model used does not
provide any argument for the export-destination countries to take any anti-dumping actions
against Côte d’Ivoire. In particular, because the exporting firm charges a domestic price that is
higher than the net export price, net of transportation costs, the model does not make any
predictions about the relationship between the domestic selling price and the foreign selling
price, transportation costs included.

The handicaps faced by industrial firms in Africa are well known, even in largely open
economies like Côte d’Ivoire, where the discretion of the government for soaking up the
profits of incumbent firms is credibly kept under control, in an orderly polity (Azam and
Morrisson, 1994). Labour in the formal sector is relatively expensive, given its productivity;
transportation costs are high, and contribute to make imported goods dearer than in other
continents (Yeats, 1990) ; etc. This is true not only of landlocked countries, but also of coastal
ones, as the cost of shipping goods to or from African countries are higher than for other
continents, probably because of the small size of the African markets, that precludes the full
exploitation of economies of scale in shipping, and of the correlative market power of the
incumbent shipping firms, that allows to sustain a profitable traffic below the optimal scale.
The other side of this coin is that firms located in African countries benefit from some natural

---

2
protection by these high transportation costs, in addition to some possible barriers erected by
the domestic government for the sake of protecting some domestic firms. Market power is
temporarily good when it enables the benefiting firm to make the most of learning-by-doing
effects, as in the standard ‘infant-industry’ argument, but the resulting excessive comfort of a
monopolistic position may also inhibit any dynamic behaviour of the firm in the fields of
innovation and commercial aggressiveness.

The next section discusses the theoretical model, showing that the relation between a
firm’s export performance and domestic competition is ambiguous, with a range of parameter
values for which an increase in domestic competition reduces the firm’s quantity exported.
Although the analytic condition for this result to prevail is relatively complex, involving the
change in the elasticity of demand, we show with a very simple example that this phenomenon
is not a curiosum, and can arise with a fairly standard specification of the model. In this
theoretical framework, the number of incumbent firms is regarded as exogenous, and the
impact of increased competition is analysed as the result of the entry of a new competitor.
Hence, implicitly, we rule out free entry in the market, and postpone the analysis of
endogenous entry until further research. Because of the discretion left to the government by
the investment code, in Côte d’Ivoire, incumbent firms often benefit from secret tax
exemptions, and other privileges that do not help creating a level playing field. Therefore,
treating the number of firms as exogenous, or policy-determined, is undoubtedly a better first
approximation than assuming that free entry prevails. Some thoughts about the potential
impact of exports as an entry deterrent are offered in the concluding section.
Section 3 presents the data used, coming from the RPED (Regional Program for
Enterprise Development) survey performed in Côte d’Ivoire under the direction of one of the
authors in two rounds, in 1995 and 1996, and some descriptive statistics. The latter show in
particular that the most important exporting industries of the manufacturing sector, in terms of
their share of the value of total manufacturing exports, are comprised of firms that sell most of
their output in the domestic market, and export the remainder to a diversified set of countries.
Section 4 presents an empirical test of the predictions of the model, showing that the
probability of a firm exporting is negatively related to the number of competitors in the
domestic market, when the latter is small, and becomes positively related for a larger number
of them. However, the impact of the number of competitors on the export performance of the
firms, conditional upon their actually exporting, is not significant. Section 5 concludes, and
suggests extensions of this analysis.

2. The Model

In order to bring out clearly the effect of increased competition on export performance,
we restrict the analysis to the case where n identical firms are competing à la Cournot on the
domestic market, while they are facing a perfectly elastic demand for their product on the
international market. Transportation costs, and other possible tariff or non tariff barriers, play
a central part in making such a setting possible, in particular by making the re-import of
exported goods very costly. Otherwise, arbitrage would prevent the domestic firms from
exerting any market power on the domestic market. So, if p* is the price of the exportable
good on the foreign market, µ the unit cost of importing the good, and ξ the unit cost of
exporting it, there exists a price band:


p ∈[ p *−ξ , p*+ µ]

within which the price of the good is determined by the domestic market conditions, rather
than by the foreign market. In Africa, this price band is probably quite wide, because of the

---

3
transportation costs and the tariffs and non tariff barriers which are common in this continent.
Denote px = p *− ξ the export price that the firm gets, net of exportation costs, and x the
quantity exported.
Let
q represent the level of domestic sales by the individual firm, and Q = n q the
industry’s total domestic sale in a symmetric equilibrium. Assume that the domestic demand
for the good may be represented by the following inverse demand curve:


p = p(Q), p' (Q) < ,
0






(1)

and define the demand elasticity ε such that:

− 1 Q p' (Q)

=
>0.







(2)
ε
p(Q)


Now, under the Cournot-Nash conjecture, the firm will regard the domestic sale level
_
of the other firms q as given, and will choose the quantity produced z, the domestic sale q,
and the level of export x, that maximise:

_
max
x
q, z p(q + (n − )
1 q) q + p .[z − q]− c(z), (3)
s.t.
x = z - q ≥ 0,

where c(z) is the cost of producing z, such that c’(z) > 0.

Examination of (3) shows clearly that, for an exporting firm, the marginal price of
selling the output z is px , while the marginal cost of selling q on the domestic market is also
px, its opportunity cost. Hence, in an interior symmetric equilibrium, with identical positive
exports for all the firms in the industry, the first-order conditions will read:


c’(z) = px,








(4)
and

1 

p n q  −
 = px
(
). 1
.






(5)

n ε 


Condition (4) simply states that the marginal cost of producing the good should be
equated to its marginal price, i.e. the export price, net of transportation costs. This condition
would hold also were the industry perfectly competitive. It determines the output level
irrespective of the domestic demand conditions. Condition (5) shows that the level of
domestic sales should be chosen so as to equate the marginal revenue on this segment of the
market to the opportunity cost of selling the good locally rather than abroad. Because of the
symmetric equilibrium assumption, the perceived elasticity of the demand facing the firm is
increasing in the number of competitors. The larger the latter, the more elastic is the demand
curve facing each firm, and the lower is the incentive for it to restrict sales on the domestic
market with a view to increase the selling price. In the limit, if n → ∞ , the firm has no market
power on the domestic market, and is unable to discriminate between the two segments.
However, in African countries, we expect the number of domestic competitors to be relatively
small, in manufacturing industries.

---

4

In a corner symmetric equilibrium, with no exports, the first-order conditions (4) and
(5) become:


1 

p n q 1−
 = c q ≥ p x
(
).
' ( )
.





(6)

n ε 

Hence, the firms that specialise in selling on the domestic market should have a larger
output level, ceteris paribus, and thus a higher marginal cost of production. From the
viewpoint of a simple analysis of their production cost, they would thus look less competitive.

Figure 1 represents the interior equilibrium, for a given (small) value of n. The level of
output is determined where the marginal cost of production is equal to the net export price,
irrespective of domestic demand conditions, while the level of domestic sale is found where
the domestic marginal revenue R’, i.e. the left-hand side of (5) and (6), is also equal to the net
export price. In this equilibrium, the firm discriminates between the foreign market and the
domestic market, where it charges a price p(nq) > px.



p
p*+µ




p(nq)

c’(z)


px

p(nq)
R’


q, z
q
z
x








Figure 1: Output and Domestic Sale Levels

This theoretical framework is akin to the classic analysis of the dumping problem (see
e.g. Deardorff, 1990). However, this case would provide no justification for the adoption of an
anti-dumping action by the destination market, as the exported goods are sold there at the
market price, which is equal to the marginal cost of delivering the good there, transportation
cost included, with no cross subsidy by the domestic market. This model does not rule out the
case where the selling price at destination is higher than the domestic price.

With the empirical analysis of the next two sections in view, we are interested in the
comparative statics of the quantity exported by the firm, which may be written as:


x p x Φ n Ψ = z p x Φ − q n p x
(
, , , )
(
, )
( ,
, Ψ) ,



(7)

where Φ and Ψ represent respectively the vectors of parameters affecting the marginal cost
of production and the demand curve. Most effects are pretty obvious, and can be derived
intuitively: any reduction in the marginal cost of production will increase exports, ceteris
paribus, as will an increase in the net export price, while an increase in domestic demand, not
affecting the number of firms or the elasticity of the demand curve, will reduce exports. The
only ambiguous impact is that of an increase in the number of competitors in the domestic

---

5
market. There are two conflicting effects: an increase in the number of competitors will
reduce the level of domestic demand addressed to each individual firm, as its market share
falls, which tends to push it to export more, but, under the Cournot-Nash conjecture, this will
also entail an increase in the perceived elasticity of the demand curve facing each firm. Then,
the payoff to restricting the sales on the domestic market with a view to get a higher price
falls, providing a countervailing incentive to sell more on the domestic market. Therefore, the
net impact will depend on the relative strength of these two opposing effects. The outcome is
clearly driven by the change in the elasticity of demand facing the firm as the intensity of
competition increases. This is formalised by the following proposition, taking due account of
the integer constraint that bears on the number of firms. Denote ε+ and q+ respectively the
elasticity of demand and the level of output, when the number of competitors is n + 1,
and define ∆ ε = ε+ - ε and ∆ q likewise. Then we can prove the following proposition:

Proposition 1: The impact of an increased number of competitors in the industry will increase
the firm’s exports unless:


∆ ε > ε .(n ε+ - 1) > - ε+ ε.





(8)

or unless these two inequalities are simultaneously reversed.

Proof: Comparing the first-order condition (5) when there are n and n+1 competitors, we find:


1

1
+



p[(n + )
1 q ].1−
1
.
+  = p(nq).
−

(n + )
1 ε 


nε 

Now, take a Taylor expansion of p[(n+1)q] about nq, limited to the first order, to find:


p[(n+1) q+] = p(nq) + p’(nq).[(n+1)q+ - nq].


Then, substituting and rearranging yields:

∆q
1− ε n

=
.
q
(n + 1)[(n +
+
1) ε − 1]


Then proposition 1 follows as the firm’s exports increase when q falls.


As condition (8) is not particularly illuminating, from an intuitive point of view,
except insofar as it draws attention to the change in the elasticity of demand, it is worth
working out a simple example in order to show that a negative impact of increased domestic
competition on exports can arise without making any heroic assumptions.

Example: Assume that the inverse demand curve is given by:


p = Q−1/2 ,

and that:


px = 1.

---

6


Then, output is fixed, and export will fall if domestic sales increase. Table 1 describes
the outcome for three levels of domestic competition.

Table 1: Domestic Sales and Price
(example where p = Q−1/2 and px = 1)


q
p
n = 1
0.25
2
n = 2
0.28
1.333
n = 3
0.231
1.2


This example thus shows that, in this case, the quantity exported by each firm would
fall were a new entrant to turn a monopoly into a duopoly. After that point, any new entrant
would lead all the firms to increase their exports. Obviously, this simple example is just that,
and thus only provides simple predictions. However, proposition 1 shows that a wider array of
possible cases can be constructed by allowing a variable elasticity of demand in the domestic
market.

Notice that a similar argument would apply to analyse how an increase in the number
of competitors would turn firms from exclusive dealings in the domestic market to exporting.
Then, for some values of the parameters of the model, it is possible that an exporting industry
will stop exporting at some point when the number of competitors increases, and then will
become exporter again at a higher level of competition. This happens very simply in the
simple example analysed above, if one completes the model by assuming that the marginal
cost of production is c’(z) = 3.9 z. Then this industry exports when there is a monopoly, stops
exporting when a duopoly is established, and becomes again an exporter when three or more
firms are active.

We now turn our attention to the case of manufacturing firms in Côte d’Ivoire, in order
to confront the predictions of this theoretical framework to the actual experience of these
firms. We first describe some salient features of our sample.

3. Description of the Data

A panel of about 230 manufacturing firms has been randomly selected at the beginning
1995 and 1996, with replacement of the exiting ones (about 10 %), for collecting the data
relative to the preceding years. The firms have been picked at random in four industries:
textile and clothing, wood, metal, and food processing and agro-industry. Formal sector firms
are defined as firms which do pay national taxes, or at least are supposed to do so in case of
positive profits, while informal sector firms are those which only pay communal taxes (« la
patente »). We have created a sub-group of semi-formal firms, that are informal according to
the previous criterion, but are organised in an association which lobbies the government and
manages to get various advantages from it. The sample is stratified insofar as about 75 % of
the formal and semi-formal sector firms from the cities of Abidjan, Bouaké and San Pedro, in
four manufacturing industries (wood, metal, textile, and agro-industry), have been drawn in
the sample, while the sampling rate for the informal sector is much smaller. However, some
experimentation with weighted estimation has shown that the problem can safely be
neglected. Less than a fifth of our sample firms are from the informal sector. A close scrutiny
of the data has led to the identification of seven markets, to which firms have been allocated,
with a view to better capture the extent of competition, according to the substitutability or

---

7
otherwise of their products: clothing industry, furniture, metalwork, plywood, primary
consumption goods, secondary consumption goods (i.e. more elaborated products), other
(mainly bread).

Firms have been asked: « what fraction of your production do you export ? » Out of the
initial sample of 234 firms in 1994 (230 in 1995), we have:
• ten firms which failed to answer in 1994 (four in 1995),
• 139 firms (59.4 % of the sample in 1994, 60.4 % in 1995) that declare exporting nothing ;
the majority of them are in the informal or semi-formal sector. In 1994, 83 % of the
informal and 88.5 % of the semi-formal sector firms are not concerned by exports. This
percentage falls to 46.5 % in the formal sector. Of course, firms only answer about the
exports that they process themselves, and are unable to tell if their products are exported by
informal traders to the neighbouring countries.
• 85 firms (81 of them in the formal sector) that answer that they actually export a positive
share of their production in 1994 (87 in 1995). Chart 1 shows the number of the exporting
firms by market for 1994 and 1995.

30
25
20
15
1994
1995
10
5
0
clothing furniture metalwork other
plywood primary secondary
industry
product
product


Chart 1 Number of Exporting Firms per Market

As the samples have very similar properties for 1994 and 1995, our descriptive
analysis of export performance is based on the 1994 data.

Table 2 clearly shows that a large fraction of the exporting firms (37.6 %) are export-
specialist, as 32 out of the 85 exporting firms export more than 75 % of their production,
including 15 firms that export 100 %. Chart 2 brings out more strikingly that firms tend to
polarise at the two ends of the distribution, exporting either very little, or quite a lot of their
production.

A breakdown by market shows that export orientation is very contrasted by type of
product (table 3). For example, 60 % of the firms exporting more than 75 % of their
production are in the plywood industry, while 40 % of the firms exporting less that 25 % of

---

8
their production are in the metalwork industry. Firms do not export anything in the bakery
industry, and almost nothing in the furniture industry, with only one firm being highly
specialised in producing furniture for export.

Table 2: Distribution of Exported Share Among Exporting Firms

Fraction Exported
Number
Percentage
Between 0 and 25 %
23
27.5
Between 25 and 50 %
16
18.8
Between 50 and 75 %
14
16.5
Between 75 and 100 %
32
37.6
Source: Computed from the RPED survey 1994.




35
30
25
y
nc 20
ue
15
f
r
e
q
10
5
0
0 - 25
25 - 50
50 - 75
75 - 100


Chart 2: Distribution of Firms per Share of Exported Production

Table 3: Export Shares per Firm and per Market

bread
clothing
furniture
metalwork
plywood
primary
secondary
total
0
% 16 35 22 21 12 6 5 117
0
to
25
%
0 1 0 8 1 2 5 17
25
to
50
%
0 3 0 1 3 2 1 10
50 to 75 % 0
1
0
2
5
1
3
12
75
to
100
%
0 3 1 1 18 3 3 29
total 16 43 23 33 39 14 17 185
Source: Computed from the RPED survey 1994.

---

9
100
80
60
40
20
0
0
20
40
60
80
100



Chart 3: Distribution of Exported Value (1994)


100
80
60
clothing indus
metalwork
40
plywood
primary prod
20
secondary prod
0
0
20
40
60
80
100


Chart 4: Distribution of Exported Value
(per market in 1994)

---