Outgrowing Resource Dependence
Theory and Some Recent Developments
by
Will Martin
Development Research Group
World Bank
Abstract
Many policy makers are concerned about dependence on resource exports. This paper
examines four changes that reduce this dependence: (i) accumulation of capital and skills;
(ii) changes in protection policy, particularly reductions in the burden of protection on
exporters; (iii) differential rates of technical change; and (iv) declines in transport costs.
Developing countries as a group have made enormous progress in diversifying their
exports away from resources in recent decades, a development that appears to have been
aided by accumulation of capital and skills and by dramatic reductions in the cost of
protection to exporters, but slowed down by technological advances that favored
agriculture.
World Bank Policy Research Working Paper 3482, January 2005
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the
exchange of ideas about development issues. An objective of the series is to get the findings out quickly,
even if the presentations are less than fully polished. The papers carry the names of the authors and should
be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely
those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors,
or the countries they represent. Policy Research Working Papers are available online at
http://econ.worldbank.org.
Outgrowing Resource Dependence:
Theory and Some Recent Developments
Non-Technical Summary
Policy makers in many developing countries are concerned about continuing dependence
on exports of resource-based products. Concerns about such dependence arise from a
number of factors including concerns about: adverse trends in the terms of trade for
commodities; the perceived volatility in their prices; the possibility of lower rates of
productivity growth in resource production; and incentives for rent-seeking.
Frequently, this concern manifests itself in policy approaches that are either ineffective in
reducing resource dependence, or unhelpful to economic development. Approaches that
are likely to be ineffective include exhortations to the private sector to diversify their
export mix. Proposals that are likely to be unhelpful include the provision of protection to
import-competing industries.
The first part of the paper identifies policies that can both contribute to economic
development and to reduction in resource dependence. These include: (i) accumulation of
capital and skills; (ii) changes in protection policy, particularly reductions in the burden
of protection on exporters; (iii) differential rates of technical change; and (iv) declines in
transport costs. Then, the paper examines recent changes in developing country policies,
and their environment, and export outcomes.
There has been a dramatic change in the composition of developing country exports since
the mid-1960s. The share of manufactures in has risen from around 15 percent to over 80
percent, with a corresponding reduction in the importance of agricultural and mineral
resource-based products. This is shown to reflect changes in the volumes of exports,
rather than changes in relative prices.
Possible contributors to this development are identified as higher rates of accumulation of
physical capital in many developing countries than in the industrial countries; and higher
rates of growth in education per worker in all developing regions. Another fundamental
change that appears likely to have contributed to this transformation is the dramatic
reduction in developing-country trade barriers. Tariffs, non-tariff barriers, and exchange
rate overvaluation have all been reduced dramatically. This has greatly reduced the
burden imposed by protection on export-oriented manufactures. The role of technological
change is less clear. The evidence suggests that, if anything, technical change has been
more rapid in agriculture than in manufactures, a factor that would be expected to retard
reductions in dependence on agricultural exports. Clearly, however, reductions in
transport costs have contributed to the development of global production sharing models
that have increased developing countries’ ability to participate in manufactures trade.
Outgrowing Resource Dependence: Theory and Some Recent
Developments
Countries vary greatly in the share of their exports derived from resource-based activities.
In those countries that obtain a large share of their export revenues from resource-based
activities, a goal of reducing resource dependence is frequently a major influence on
policy. The importance placed on this goal is particularly marked in resource-dependent
developing countries, but has also emerged in high-income countries such as the
Netherlands and Australia in the form of concerns about de-industrialization during
periods of growth in resource-based industries (Gregory 1976; Snape 1977).
There are many reasons why policy makers may wish to reduce the share of a
country’s export revenues obtained from commodities produced using resource-intensive
procedures. These include: (i) the concerns about potentially adverse trends in the terms
of trade for commodities raised by Prebisch, (ii) concerns about the perceived instability
of returns from commodities and possible resulting problems of unemployment and
output loss (Cashin and McDermott 2002), (iii) perceptions that the rate of technological
change in resource-dependent activities may be lower than in manufactures or services,
and (iv) concerns that resource-intensive production may promote rent-seeking activities,
lower growth rates, and increase the risk of civil war (Sachs and Warner 1995, Collier
2000).
Clearly, given the potential stakes involved with decisions about changing
resource dependence, and the fundamental nature of many of the policies advocated for
achieving this objective, there is a great need for carefully formulated policies if this
objective is to be achieved. Unfortunately, much of the policy debate surrounding these
objectives takes place at a sufficiently high level of abstraction that it does not provide
much guidance. Consequently, many of the policies adopted to this end seem ad hoc and
potentially counter-productive. A very common response, for example, is a relatively
arbitrary set of protectionist measures designed, perhaps, to promote activity and learning
in manufacturing sectors. But, as we shall see, protectionist policies may have quite
2
contrary effects. In fact, it seems likely that liberalization is key to increasing
diversification and not, as many have feared (eg Parris 2003), to continuing dependence.
The choice of policy options for dealing with this problem also needs to be based
on good diagnostics, and to take a broad view of the policy options. It is possible, for
instance, that a country relying on a set of different commodities may find that the
variance of returns from the resulting portfolio is not excessive—or that shifting from
commodities to manufactures would not reduce the variance of returns (see Martin 1988
for example). Further, if excessive instability of export returns is identified as a problem,
then the most effective solution may lie in portfolio management approaches that allow
reductions in the volatility of consumption without attempting to reduce the volatility of
annual earnings. Such a solution is consistent with the general principle in economic
policy of targeting the policy solution as closely as possible to the problem at hand.
Policies that attempt to deal with the risks associated with commodity dependence
by diversifying the structure of output should not generally be undertaken unless analysis
indicates that: (i) there are market failures that are reducing the extent to which the
production structure should shift away from commodities, and that (ii) policy options are
available that will diversify output and improve overall economic performance. While
these criteria might appear daunting, there are many cases where they will be fulfilled.
Potential causes of resource dependence in the structure of output and exports
include: (i) unusually large endowments of natural resources; (ii) limited supplies of
factors such as capital and human capital that are used more intensively in manufactures
and services than in resource-based industries; (iii) low productivity in manufactures and
services; (iv) trade and pricing policies that discriminate against export-oriented
manufactures and services; and (v) high transport and communication costs. Since
countries would not generally wish to reduce their endowments of natural resources1, the
policy solutions to what is regarded as an “excessive” level of dependence on natural
resources are likely to lie in the four areas (ii) to (v).
1 Although they may wish to consider the timing of exploitation of non-renewable resources.
3
These four influences on resource dependence are clearly strongly related to the
basic determinants of structural change identified in the classic Chenery, Robinson and
Syrquin (1986) study of industrialization and structural change. One other influence on
the structure of output, and of exports, identified by Chenery, Robinson and Syrquin is
non-homotheticity of consumer demand, although this is difficult to use this for policy
purposes. Low income elasticities of demand may, in fact, cause a country undergoing
unbiased growth to become more reliant on exports of commodities.
A wide range of policies designed to promote the development of favored sectors
have been discussed under the rubric of industrial policy (see Pack 2000 and Stiglitz
1996). Industrial policies have included many specific policies, such as provision of
infrastructure, support for education; export promotion activities; technology promotion
programs; duty exemption and drawback arrangements for exporters; and preferential
allocation of credit to exporting industries. All of these policies can be seen ultimately as
affecting the level and structure of output through one of the four channels considered in
this paper.
The process of developing growth models that go beyond balanced growth is only
now getting under way (see, for example, Kongsamut, Rebelo and Danyang Xie 2001).
Specifying model features in a way that will allow them to be useful in analyzing the
profound structural changes associated with reducing resource dependence seems likely
to require more sources of structural change than are included in most current growth
models.
As noted in World Bank (2003), and in Martin (2003), there have been dramatic
changes in the participation of developing countries in world trade. The share of
manufactures in total merchandise exports has increased dramatically, at the expense of
the traditional stalwarts—agricultural products and minerals. This change has been
associated with dramatic shifts in policy toward trade openness, and with increases in
factor endowments, that raise the available capital and skills per worker.
4
In this paper, a simple general equilibrium framework sufficiently general to
incorporate the structural changes associated with reductions in resource dependence is
specified. It is then used as an organizing framework to examine some data on changes in
export patterns of developing countries and indicators of the influences on resource
dependence identified in the conceptual framework. This analysis is then followed by
consideration of policies that might be used to reduce resource dependence.
A Framework
For this paper, we need a formulation sufficiently general that it can encompass
changes in factor endowments, changes in technology, and changes in price policies. The
dual approach popularized by Dixit and Norman (1980) provides this flexibility. The
production side of the economy can be represented using a restricted profit function
specifying the value of net output in the economy as a function of the domestic prices of
outputs and intermediate inputs:
(1) π = π( p , v) = maxx {p.x (x, v) feasible}
where π is the value-added accruing to the vector of quasi-fixed factors ,v, in the
economy given the vector of domestic prices, p, for gross outputs of the vector of
produced goods, x. The vector v includes economy-wide stocks of mobile factors, any
sector-specific factor inputs, and public goods such as infrastructure, that may not be
readily allocable to particular sectors.
As Dixit and Norman (1980) note, the specification in equation (1) represents all
of the properties of the production technology. It is extremely general, being able to
represent many different types of technology depending upon the particular functional
form used to specify the GDP function. These specifications may include the familiar 2*2
Heckscher-Ohlin model with two factors and two outputs, and no intermediate inputs,
through a range of specifications of much greater generality. It may also include
specifications such as the Leamer (1987) model in which there are more goods than
factors, and small, open economies move between different cones of diversification in
5
which the set of commodities produced change. The specification is also sufficiently
general to include forward and backward linkages induced by input-output linkages and
transport costs.
Over the range where the profit function is differentiable, its derivatives with
respect to the prices of output yield a vector of net output supplies:
(2) πp = πp(p,v)
Depending upon the specification of the profit function, it may be possible to
identify the gross outputs of each good, and the quantities of these goods used as
intermediate inputs in production. For some purposes, such as estimating the incentives
created by a protection structure, it is very important to be able to identify the net outputs.
The derivative of the profit function with respect to the factor endowments gives
the vector of factor prices.
πv(p, v)
One additional important expression is the matrix of Rybczynski derivatives.
Differentiating the vector of price derivatives, πp , by the vector of resource endowments
(or, equivalently by Young’s theorem, differentiating the vector of factor prices by the
price vector) yields a matrix, πpv , of changes in the net output vector resulting from
changes in factor endowments. This matrix is clearly critical for our analysis, but its exact
structure depends heavily upon the particular situation.
In the simple, two factor, two output model used in textbook treatments, the
Rybczynski responses take a very clearly-defined form in any economy that is producing
both outputs. As the supply of one factor increases, the output of the sector in which that
factor is used intensively increases. The output of the other good declines, despite the
increase in the total resources available to the economy. Importantly, factor prices do not
change. The required change factor use is achieved by changing the mix of outputs, rather
6
than by changing factor prices. As long as the number of factors and the number of
outputs remains the same, this mechanism can be generalized to economies in which
there are multiple factors and multiple outputs. The concept of relative factor intensity
can be generalized to indicate the increase in the cost of producing a good when the price
of a factor increases (Dixit and Norman 1980, p57).
The most difficult case to analyze is the realistic situation in which there are more
goods than factors. Leamer (1987) and Leamer, Maul, Rodriguez and Schott (1999)
provide an extremely useful analytical framework for analyzing this problem where there
are three factors and many goods. In simple cases2, countries with three factors will
specialize in the production of three goods. Over some range, the features of the
Rybczynski theorem will hold and changes in factor endowments will result in changes in
the mix of output without changes in factor prices. However, changes beyond that point
will result in shifts into a new cone of diversification, with a change in the mix of output
and a fall in the return to the factor whose relative supply is being augmented. As Leamer
(1987, p967) points out the location of these cones of diversification depends upon
commodity prices, and hence is not merely a function of technology.
In the case of resource-poor economies, Leamer et al show that the adjustment
path associated with accumulation of human and physical capital is likely to be relatively
smooth, with increases in the supply of capital raising the demand for raw labor as the
economy moves through different cones of diversification. For resource-abundant
economies, however, the path may involve reductions in unskilled labor as the economy
moves from, say, peasant farming to resource-based systems involving greater use of
capital. This move may be associated with reductions in the returns to unskilled labor that
increase income inequality.
For some problems, such as situations where some goods are nontraded, we need
to consider the consumption side of the economy as well as the production side. The
2 In the absence, for instance, of nontraded goods.
7
consumption side of the economy can be represented similarly using an expenditure
function:
(3) e(p, u)
where e represents the expenditure required to achieve a specified level of utility, u, and
represents all of the economically relevant features of consumer preferences. Assuming
differentiability of the expenditure function, the vector of consumer demands can be
obtained as:
(4) ep(p, u)
An important feature of real-world consumer preferences is their non-
homotheticity, with commodities like basic food having small or negative responses to
income increases, while luxury goods have large positive income effects. The vector of
Marshallian income effects can be derived from (4) as:
cY = (epu/eu)
where eu is the marginal impact of a change in utility on expenditure, and epu is the
marginal impact of a change in utility on the consumption of each good.
The vector of net imports of commodities is given by m, which is the difference
between the vector of consumption and the vector of net outputs:
m = ep - rp
World prices of traded goods are determined by the market clearing condition that
the sum of the net trade vectors for all regions must equal zero. Where some goods are
non-traded, the relevant sub-vector of m is exogenously equal to zero and equilibrium in
the market for these goods is achieved by adjustments in the prices of these goods.
Similarly, where trade in some goods is determined by binding quotas, the relevant sub-
vector of m is set exogenously at the quota level and equilibrium is achieved by
endogenous determination of these prices.
Trade policy distortions are represented very simply as creating a difference
between the vector of domestic prices, p, and world prices, pw for the small representative
8
economy. It is frequently useful to define a net expenditure function z = (e –r ). The
derivative of this function with respect to prices, zp = (ep – rp ) is also equal to the vector
of net imports. This function also provide