Previous - Next
This is the old United Nations University website. Visit the new site at http://unu.edu
Theories of structural adjustment
The nature of theories is that they are abstract, for otherwise they would not reveal the general properties of the systems they describe. Yet, it can certainly be argued that the theories hitherto described are too abstract, for they fail to include two other crucial issues, the allocation of scarce resources within the productive sector and the destination of its outputs. These two issues are interrelated, and so can best be considered - logically and coherently - in the context of an expanded mathematical model of economic growth. The model will necessarily be one that expands upon our earlier models but, also necessarily because of its increased complexity, one that is not so clear in its policy implications, i.e. one that does not yield, as a logical deduction, an optimal policy. For the model that we will describe, the optimal policy cannot be derived, only guessed at.
Of all the theories available, this model appears to be the most relevant to our research and the most suggestive for policy, not only domestic (that is within a developing country) but also foreign (that is within the international agencies and the overseas donors). If our prescriptions are to be international, our theory must be international too, so it is within the international elements of the theory that we will commence our description. The theory is recent, having been formulated only recently by Ka-yiu Michael Fung and Jota Ishikawa (Fung and Ishikawa, 1992). In their paper the authors start by assuming that the prices which govern in the developing economy are those established in the world at large. The developing country under examination can trade at those prices, but must also accept them for its own internal transactions. It is, in the language of trade theorists, an open economy. The country is also assumed to have such a meagre effect in the world's demand for and supply of the traceable commodities that it alone has no effect on their prices. It is a small open economy.
For us, the traceable commodities must be defined with great care. In Fung and Ishikawa's model they are two in number and labelled X and Y: both X and Y are final goods, which can either be traded abroad, generating foreign exchange, or consumed at home, yielding satisfaction (see Figure 1.1). Taking the world market price of good X as unity, the world sets a price of P* on good Y. Since the goods within the developing country are valued at world market prices, their domestic prices are also 1 (unity) and P*. These prices are assumed to remain constant throughout all time.
Since the relative prices of the two goods are exogenous and fixed, something else must distinguish the two. The distinguishing characteristics are the resources that they need for their production and the impact upon them of advances in science and technology. Briefly, good Y is produced under constant returns to scale, using labour and a second factor unique to itself, labelled R. For the moment, we will say no more about R. beyond that it is necessary in the production of good Y. is in fixed supply, and, consequently, yields to its possessors an income, or rent, hence the letter R.
Good X, the other traded good, is produced under increasing returns (to be explained later) using a single factor, intermediate goods. Intermediate goods, S. are an exception to the category of goods consisting of X and Y: they are assumed not to be traceable. In other words, goods X and Y can enter into international trade but goods S. the intermediate goods, cannot. They have a price in the domestic economy alone, and that price is determined solely in the domestic economy.
We cannot bring out the important distinction between goods X and Y without going further into the origin of the intermediate goods S. This can most clearly be done by counting up the sectors in Fung and Ishikawa's model. So far we have identified three, those responsible for the production of goods X, Y and S. But there is a fourth and final sector, called by the authors the R&D sector, but by us the sector that generates advances in science and technology.
In the sector that generates advances in science and technology, there is only one factor, labour. In this assumption, Fung and Ishikawa's model is similar to Phelps' simpler model, and to those of Uzawa and Enos; but not to Phelps' more complicated model. Working with the accumulated science and technology, labour engaged in this sector is able to advance it, the rate of advance being proportional to the number of workers allocated, and to its existing level. Thus the higher is the current level of science and technology, the faster it will advance; and the more workers that are employed in advancing it, the faster it will advance. The two elements (the state of science and technology and the number of scientists, technologists and allied workers) reinforce each other. Each alone is capable of promoting science and technology, but together they have a greater capability. This greater capability is acknowledged in the expression 'increasing returns to scale'.
Increasing returns to scale are the characteristic of production of tradeable good X, but accrue not directly but indirectly through the production of the intermediate good S. Recall that S. and S alone, is necessary for the production of good X; yet labour and the accumulated amount of science and technology is necessary for the advance of science and technology, which is incorporated in good S. The way in which Fung and Ishikawa account for the effect of advances in science and technology in S is now relatively common in economic theorizing and owes its formulation to Romer and Lucas (Romer 1989 and 1990; Lucas, 1988). The properties of the specific function chosen are discussed in Grossman and Helpman, 1989. Essentially, Fung and Ishikawa assume that the further has science and technology advanced, the greater is the number of different intermediate goods available to producers of X; and the greater the number of intermediate goods available, the lower are their prices. Thus, the further advanced is science and technology, the cheaper is the production of X. This section of the model itself is more complicated, involving many producers of differentiated intermediate goods in a monopolistically competitive industry.
In résumé, the two traceable goods, Y and X, are produced under different conditions: Y is produced employing a technology that is static, whereas X is produced under an advancing technology. The more labour that is allocated to securing advances, the faster are those advances; also, the further has the technology advanced already, the faster are those advances. To concentrate today on advancing science and technology is to lower the costs of producing X today; moreover, to have concentrated on advancing science and technology in the past is to have lowered the cost of producing X today. Concentrating on advancing science and technology provides two dividends, one immediate and the other from past accumulations.
Perhaps it is now obvious why it is so important for us to define our goods X and Y carefully. Depending upon what real commodities we assign to the category X and what other commodities to category Y. we will be determining deductions from the model and influencing the recommendations for policy. One's intuition can be relied upon already; it suggests that whatever commodities receive the attention of grantors of money for R&D will have greater prospects for the future than those which receive no attention. But there is a danger of circular reasoning: since, according to the model, the commodity towards whose production R&D is directed will have its costs of production lowered, so it will be worth while directing R&D towards that commodity. That commodity automatically falls in category X. The allocation is self-fulfilling. But there may well be another commodity whose cost of production would fall even further if the same amount of R&D were devoted to advancing its technology. Yet, if that potentially more attractive commodity does not receive the attention of the potentially less attractive commodity, it would fall in category Y.
So, we must assign commodities to categories X and Y not on the basis of whether or not they currently have R&D directed towards them but on the basis of whether or not they have great potential, potential that could become manifest were R&D to be directed towards them. It is potential to respond to R&D that determines in which category actual commodities belong. This is not a theoretical issue but a practical one; this is not an issue that we can dispose of now, but one that will occupy us throughout our study.
Even before our theory has been fully formulated it has already illuminated one issue that we must address - the issue of the direction of R&D. In our language, the issue is in which areas and on behalf of which commodities are advances in science and technology to be sought? It is not enough for us to identify those areas in which advances in science and technology are currently being sought; we must ask if these are the best areas. Are there better areas to be exploited? Why are these areas not being opened? Why is R&D not being undertaken there?
Other collateral questions immediately come to mind: on what basis and through what mode of analysis is the potential of individual commodities determined? Who makes the choice? With what objective in mind? These are all questions that we must ask, questions that are of consequence not only for the narrow purpose of ensuring as good a fit between our model and reality in Sub-Saharan Africa but also for the much graver purpose of ensuring that those resources necessary to advance science and technology there, so scarce as they be, be put to their best use.
Having interrupted our description of Fung and Ishikawa's model to identify our first set of questions, we shall now proceed to determine its solution and its policy implications. Although we are avoiding mathematics in this chapter, we will find it useful to appeal to five further diagrams, in the hope that they will make the analysis clearer. The first is Figure 2.1 which purports to illustrate the state of the economy before and after substantial resources are allocated to advancing science and technology. The output of good Y (the good whose potential is low, i.e. whose response to future R&D would be less beneficial from the point of view of the developing economy) is measured on the vertical axis; that of good X (the good with potential) on the horizontal. The curve, concave to the origin, joining the two axes, is the locus of the maximum amounts of the two goods, X and Y in combination, that can be produced currently, assuming that all the resources in the economy are fully employed, that they can be allocated to the production of X and Y in any proportion, and that they can be shifted costlessly from the production of less of one good to the production of more of the other. In other words, any combination of goods X and Y indicated by the coordinates of a point on the concave curve labelled (X,Y) c, the 'c' for current, is assumed to be feasible.
How, in theory, will the developing country allocate its scarce resources? Assuming that internal prices of the two commodities match world market prices, and assuming also that the objective of the country is to maximize the total value of output (equal, in macro-economic terms, to maximizing GDP), it should allocate them so as to produce the amounts Y c and X c, as indicated by the intersections on the vertical and horizontal axes respectively. Graphically, coordinates of the amounts Y c and X c are located where the concave production possibility curve and the dashed relative price line (P* is the slope of the dashed line) are tangent.
Figure 2.1 Production of goods X and Y by the economy, before and after advances in science and technology
This exercise established the equilibrium for the economy before there is any shift of resources into advancing science and technology. In the absence of the allocation of resources to advancing science and technology, this equilibrium will persist through time: the economy described by the model will be static; i.e. it will not grow in per capita terms. For ease of display, it is also assumed that population is static; relaxing this assumption does not change the results in per capita terms.
An economy that grows is presumably better than one that does not. In Figure 2.1 we also depict the optimal allocation of resources some time in the distant future, based on the assumptions that the relative prices of X and Y do not change, that the scarce resources have been devoted to advancing science and technology in the production of the good X where the potential lies, and that the country's objective is still that of maximizing GDP. The optimal allocation is determined in exactly that same manner as for the 'current' case; the only change graphically is that the production possibility curve has moved outwards to the right. The new curve, (X,Y)f where the superscript 'f' indicates future, illustrates the greater productivity of the country's resources when assigned to the production of good X, but a constant productivity through time when assigned to the production of good Y.
In the future, the country will maximize its GDP by producing Yf of good Y and Xf of good X. Total GDP will be higher in the future than the present, by the amount by which (1)(Xf) + (P*)(Yf) exceeds (1)(Xc) + (P*)(Yc). The production of good Y will have fallen, and that of good X will have risen more than proportionately. Resources will have been shifted from the production of Y to that of X, although in lesser proportion than the shift in output, because of the growth of productivity of the country's resources in the production of the good X towards which it was worth directing R&D expenditures.
Although Figure 2.1 displays the before and after, it does not give any indication of the process by which growth was obtained. This is the function of Figures 2.2-2.3. There, the initiation of R&D is depicted; the changes have come about as a result of the reallocation of scarce resources away from the production of good Y. Logically, given the properties of the model, its output will fall from Yc to a new level Y1. The resources reallocated will have two competing uses, to increase the immediate output of X, and to advance science and technology so as to increase the productivity of these resources that are devoted to producing X. How should the newly available resources, withdrawn from the production of good Y. be split between the two competing uses? The optimal growth theories already laid out in the preceding section of this chapter (those attributable to Phelps, Uzawa and Enos) deduce that the bulk of the newly available resources should be allocated to advancing science and technology.
So, previous theory suggests that it is optimal to shift most of the newly available resources into R&D; but the models that yield this theorem are simpler than Fung and Ishikawa's, lacking both the sector which does not benefit from advances in science and technology and the expression which provides increasing returns through the accumulation of scientific and technological knowledge. Therefore previous theory is silent about whether the output of good X will rise or fall. Fung and Ishikawa recognize that there are the two possibilities. These are reflected in Figures 2.2 and 2.3 by the alternative production possibility curves (XY)1a and (XY)1b; the implication of the former is that shifting the resources previously utilized in the production of good Y to advancing science and technology will lead to less X being produced too; the implication of the latter that shifting the resources will lead to producing more X. To Fung and Ishikawa, as to ourselves, which of the two alternatives will be forthcoming is not obvious, but they have been able to specify, quite ingeniously, the conditions that will determine the answer. The determining condition is expressed by two parameters (pare = determined outside; meter = the model) which they label d and b , plus one initial condition and one control variable. The initial condition, which we label Nc, reflects the level of science and technology at the time the shift occurs; the control variable is labelled LN (as in figure 1.1), and measures the amount of labour allocated to advancing science and technology.
Figures 2.2 and 2.3 Production possibilities for the economy with different elasticities of substitution for intermediate goods
The first of the two parameters, d , easy to interpret: it represents the efficiency with which the labour employed in the research sector advances science and technology. The 'production function' for advances in science and technology contains the efficiency parameter, d , together with the other two factors determining the advance, namely the level of science and technology already reached and the number of workers assigned to the sector.
The other parameter, b , is difficult to interpret, although no less important in determining the outcome. It is contained in the 'production function' for the output of the good, X, whose potential is being realized through advances in science and technology. This equation states that the output of good X is directly dependent upon the number of intermediate goods available, which themselves are dependent upon the current state of the arts. The production function exhibits constant returns to scale at any instant; the parameter b represents the elasticity of substitution between the various intermediate goods. Since b is a constant, the production function can be recognized as one with a constant elasticity of substitution (abbreviated, commonly, as a CES function).
What does the elasticity of substitution signify? Its significance can perhaps be understood most easily if we consider the two polar cases, those of no substitutability and of perfect substitutability. Imagine the case of no substitutability; in this case there is no possibility of substituting in production one intermediate good for another (say, one lathe for another lathe). Intermediate goods must be combined in fixed proportions, and a lack of one sort, necessary for the production of X, cannot be compensated for by a surplus of another sort. The case of perfect substitutability is the antithesis of no substitutability: all lathes are perfectly substitutable; and none will be in surplus. In the case of perfect substitutability, the parameter b takes on the value of zero; in the case of perfect substitutability, the value of infinity. A realistic value would be somewhere in between - greater than zero and less than infinity. What value b does take on is an empirical matter, but presumably subject to influence through policy, an issue to which we will return shortly.
Let us now illustrate the effects of various values of the two crucial parameters d and b on the total output and income of the country. For this purpose we will utilize Figures 2.2-2.3. In these, are displayed the alternative outcomes in cases of relatively low (less than one-half) and relatively high (more than one half) for b , the elasticity of substitution. Note that in the first case, illustrated in Figure 2.2, the production possibility curve (X,Y)1a bends back upon itself at lower outputs of good Y than the previous equilibrium output (X,Y)c (refer to Figure 2.1, for the determination of the pre-research equilibrium). The reason that the production possibility curve bends backwards is that the productivity of intermediate goods is low because of their scarcity and lack of substitutability. Points lower down on the production possibility curve (X,Y)1a indicate greater allocation of labour to advancing science and technology, and, consequently, a lesser allocation to production of good X. This reflects the trade-off between lower production today, with more resources assigned to R&D, and higher production in the future, as a result of the advances in science and technology which today's research workers achieve.
Figure 2.4 Trajectory with low b high d
Figure 2.5 Trajectory with high b , low d
Figures 2.4 and 2.5 Alternative trajectories with different values for the parameters measuring the elasticity of substitution (b ) and the efficiency of inputs (d )
But, if the value of b exceeds one-half, it is possible to allocate some resources to advancing science and technology and produce more of good X simultaneously (although, of course at the expense of producing less of good Y). This more favourable outcome is illustrated in Figure 2.3, in which the production possibility curve does not bend back upon itself for outputs of Y less than Xc,Yc.
We have therefore seen the difference between lower and higher values of the parameter b , the difference being that, with lower values of b committing resources to advancing science and technology reduces, in the short run, the output of both final goods. With higher values of b committing resources to advancing science and technology reduces only the output of good Y.
How can we illustrate the effect of changing values of the other crucial parameter, d , the efficiency with which resources are employed in the research sector? To illustrate varying efficiencies, we have drawn Figures 2.4 and 2.5. These two consider a longer period than the previous figure, and depict outcomes in a succession of years. These yearly outcomes are shown as points along two trajectories; each trajectory is the locus of yearly outcomes of combinations of goods X and Y produced by the economy, with different values for the two parameters d and b . Taking the upper trajectory, the outcome for the year before resources are committed to advancing science and technology is the familiar Xc, Yc. Given a low value for the elasticity of substitution b , the outcome for the next year will be represented by the point X1a, Y1. Assuming that the resources employed in advancing science and technology are very productive (i.e. assigning a high value to the efficiency parameter d ) progress along the trajectory (i.e. progress from point (X,Y)1a to point (X,Y)2a, to point (X,Y)3a and so on) will be rapid.
The alternative trajectory represents opposite values for the crucial parameters d and b . As in Figure 2.3, it is assumed that the value of b is high, and so the first year's outcome is indicated by the point (X,Y)1b. But this trajectory is drawn on the assumption that the value of the second parameter, d , is low, i.e. that resources committed to advancing science and technology are employed less effectively. As a consequence, yearly progress along the trajectory is slower, successive outcomes (points (X,Y)2b, (X,Y)3b, and so on) being more closely spaced that equivalent points (X,Y)1a, (X,Y)2a, (X,Y)3a.
The best combination for the developing country is high values for both parameters. In such a desirable case, the initial commitment of resources to advancing science and technology would not involve any immediate sacrifice of good X, although there would be a sacrifice of good Y through the reallocation of resources out of the good with no potential to respond to advances in science and technology. Subsequently, progress along the economy's trajectory would be rapid, so that the country would in a relatively short period of time reach, and subsequently surpass, the situation illustrated in Figure 2.1 as the production possibility curve (X,Y)f.
Summarizing the deductions we have made from Fung and Ishikawa's model, we can now recognize the significance of the two parameters b and d , the elasticity of substitution of intermediate goods in the production of good X and the efficiency with which resources are employed in research, respectively. Added to the other two important variables, LN and Nc, the number of workers committed to advancing science and technology and the initial level reached by the country, we can see how, within the context of the model, the country's productive future is determined.
Since these four elements, the parameters b and d and the variables LN and Nc determine the course of the future in principle, we, who are using economic principles as our guide, should focus upon the actual phenomena they represent. We should attempt to find out the extent to which intermediate goods can be substituted for each other, the extent to which resources allocated to advancing science and technology are employed effectively, the extent to which substantial numbers of workers are actually committed to advancing science and technology, and the level of science and technology already achieved. Moreover, to fulfil the objective of our study, we wish to discover how these phenomena are being affected by Structural Adjustment Programmes, and whether or not their values are being improved. Do the Structural Adjustment Programmes lead to an increase in the value of b , so important in the initial years? Do they lead to an increase in the value of d , so important along the country's trajectory? Do they lead to an increase in the resources committed to advancing R&D, so important throughout? Do they lead to the country's ability to exploit, to the fullest extent, its current stock of scientific and technological knowledge? These are the questions that our theory suggests are of such importance as to demand answers.
The theories presented in this chapter have been of some use. They, together with some statistical analysis, have provided justification for concentrating our inquiry on advancing science and technology in the developing countries. They have suggested that the optimal policy for these developing countries is to commit, immediately and in substantial volume, additional resources to promoting the advance of science and technology. They have alerted us to the importance of certain phenomena, upon which we should focus our attention. Finally, they will provide us, in Parts III and IV of this study, with a means of separating the effects of Structural Adjustment Programmes from all the other events that impinge upon the developing countries in our sample; i.e. they will enable us to construct counterfactual cases, so as to guess at what would have transpired in the absence of the Structural Adjustment Programmes.
Before moving on to the country studies, let us restate the list of phenomena to which theory has alerted us. First, when defining the terms of out theory we saw that a clear distinction had to be made between those goods which the developing country could, through advances in science and technology, produce increasingly profitability, and those, other, goods which it could not. We called these goods with a 'potential' and goods without, respectively, and admitted that it will be important in principle for our study, and in practice for the developing country and for those who advise it correctly to identify the goods. This is a real issue: in which direction should science and technology be advanced? Which developments should be encouraged? Given that there are never enough resources to pursue all lines of research, which developments should not?
Having distinguished in principle between goods with potential and goods without, we then learned that theory suggests we focus particularly upon four phenomena as having a profound influence upon the future. These are the country's ability to substitute 'intermediate' goods one for another in the production of the good in which the country has potential, the efficiency with which R&D is carried out, the number of individuals committed to advancing science and technology, and the extent to which the already existing body of science and technology is drawn upon. Theory implies that these are vital matters in their own right, and that the effects upon them of the adoption of Structural Adjustment Programmes will be of great importance for the country's progress. Some empirical evidence concerning these phenomena, and others of interest, will be reported in the next four chapters, one each on Ghana, Kenya, Tanzania and Uganda.