A formal assessment of new-developmentalist theory and policy

RESUMO: Desenvolvemos uma estrutura formal que endogeneiza a estrutura produtiva de uma pequena economia periférica aberta como o resultado de um problema de escolhas técnicas. Em seguida, examinamos as principais teses teóricas e prescrições políticas da abordagem neodesenvolvimentista do desenvolvimento econômico. Argumentamos que: a) não apenas o padrão de especialização depende das condições técnicas, mas também da distribuição de renda; b) numa economia sem rendas, o nível do rácio do salário nominaltaxa de câmbio é determinado univocamente quando a taxa de lucros é conhecida e mostra uma relação inversa com a mesma; c) se as rendas diferenciais forem consideradas, o nível da taxa de lucros pode ser estabelecido independentemente do rácio do salário monetário; d) o nível da taxa de câmbio que garante a rentabilidade normal do setor primário não precisa coincidir com a taxa de equilíbrio em conta-corrente; e) a taxa de câmbio efetiva não precisa gravitar em torno de nenhum desses dois níveis anteriores, que devem ser vistos como limiares mínimos da taxa efetiva; f) as consequências distributivas desagradáveis da depreciação da taxa de câmbio podem ser parcialmente evitadas por meio de impostos de exportação que não elevam os custos de produção de commodities primárias. PALAVRAS-CHAVE: conflito distributivo; doença holandesa; política cambial; novo desenvolvimentismo; padrões de especialização.


INTRODUcTION
As is well known, the Neo-Developmentalist Position (henceforth, NDP), which has Professor Bresser-Pereira 3 as its major proponent, views exchange-rate policy as the key variable for sustained economic growth in latin American peripheral economies. Inspired in the seminal contribution by lewis (1954), the NDP envisages a two-sector economy: on the one hand, there is a sector of primary goods (e.g., corn), or inputs of these goods (e.g., oil) -sector c -. Thanks to its "extremely favourable conditions", this sector is able to export its production to the global markets. On the other hand, there is an industrial sector of consumption goods -sector I -. Even though the latter is a potentially more dynamic industry, it is also a relatively backward sector that cannot profitably compete at international prices unless the exchange rate is sufficiently high, at its "industrial equilibrium" level. Without Government intervention, however, the effective exchange rate is argued to gravitate around a lower level, the "current-equilibrium exchange rate", which is the value that is assumed to simultaneously ensure the normal profitability of sector c and the equilibrium of the current account. Therefore, this dynamics sets an "exchange-rate trap" for industry that prevents its consolidation and future expansion, a problem known as "Dutch Disease". The solution, it is finally claimed, consists in introducing a tax on c-production that increases its normal average costs and therefore raises the current-equilibrium exchange rate up to the industrial equilibrium level. Although this measure reduces the real wage in the short run, and hence may face the possible resistance of workers, it would be an unavoidable cost to boost the development of sector I, which would anyway compensate its initial negative consequences in the longer run, through higher job creation and average productivity growth.
The present article attempts to examine the scopes and limits of the NDP, and hence must be interpreted as a constructive criticism towards this approach. To this end, we develop a formal framework that endogenously determines the pattern of specialization of a small-open-peripheral economy as the outcome of a problem of technical choices. The main outcomes of our research can be briefly summarized in the following six propositions: a) not only does the pattern of specialization of the economy depend on technical conditions but also on the distribution of income (the rate of profits); b) in an economy without rents, the level of the money wagenominal exchange rate ratio is univocally determined once the rate of profits is known, and shows an inverse relationship with normal profitability; c) the moment differential rents are considered, an additional degree of freedom in the price system is gained, and hence the level of the rate of profits can be set independently of the money wage-exchange rate ratio; d) the level of the exchange rate that ensures cnormal profitability need not coincide with the current-account equilibrium rate; e) the effective exchange rate need not gravitate around any of these two former levels; which, for given levels of autonomous expenditures, must be rather seen as minimum thresholds of the effective rate; f) it is possible to avoid the unpleasant distributive consequences of exchange-rate depreciation by means of export duties that subtract a portion of the differential rent appropriated by landowners, but, differently from production taxes, do not raise c-production costs.
The rest of the paper is structured as follows: the second section presents the basic analytical framework, while the third section incorporates the problem of differential rent. The fourth section examines the relation among the effective exchange rate, the level that ensures c-normal profitability and the current-account equilibrium rate. The fifth section addresses the problem of distributive conflict in general terms, while the sixth section studies how the issue manifests itself within the New-Developmentalist approach, namely as a by-product of the so-called Dutch-Disease problem, and it examines its possible solutions. The last section resumes the argument and presents the main conclusions of the paper.

ANAlyTIcAl FRAMEWORK 4
We conceive a small open economy with persistent unemployment and two productive sectors: an industrial sector (I) and a primary sector (c). The sectors distinguish themselves by two main features: i) the productive methods employed and ii) the destination of production.
Regarding the methods of production, it is assumed that sector I is produced by labour and an imported capital good, while c requires labour and a fixed factor, typically land. 5 commodity prices can be represented by the following equations: where ! ! ! ! !! ! !! ! !! and ! ! ! ! !!! ! ! !"! ! ! !!! ! !! stand respectively, for the supply prices of commodities c and I, namely the minimum amount of money per unit of output that producers must receive to regularly (under "normal conditions") deliver each commodity on the market. Additionally, w stands for the nominal wage rate, r for the normal rate of profits, ! !
are the unitary labour requirements of sectors c and I, k is the unitary requirement of the imported capital good K, is its exogenously given price in foreign currency and E is the nominal exchange rate.
It is now convenient to introduce a second notion of price, which we shall denominate demand or selling price, and that represents the maximum amount of money that consumers are willing to pay for a commodity. Since the domestic economy takes the international prices of c ( ! ) as given, once the level of the exchange rate is fixed, demand prices are univocally determined. Therefore, demand prices for commodities c ( ! ! ! ! !! ! ! ) and I (!! ! ! !! ! ! ) are, respectively: The four equations have seven unknowns: If we take nominal wages as given: There are still two degrees of freedom left. Furthermore, at the level of abstraction we are working with, we can safely assume free capital mobility across countries. This means that the rate of profits is determined by the international rate (r*): There is still one degree of freedom left. Before we suggest how to eliminate it, let us consider the abovementioned feature ii) of the productive system, i.e., the destinations of production. Note that it is not possible to ascertain which productive sector will be internationally competitive before the relationship between demand and supply prices of each commodity is established. hence, before income distribution is known. Therefore, the pattern of specialization of the economy will be regulated by the following conditions: commodity j is produced and (potentially) exported only if ! !
, sector j will not be viable without protection 7 . Therefore, we can derive for each commodity j an (E/w)-r relation that determines, for each level of r, the minimum E/w ratio that allows sector jj to be internationally competitive. (Alternatively, for each level of r, we can determine the maximum level of money wages in foreign currency, w/E, affordable by each sector. We will return to this second interpretation of the curve below.) For each commodity j=c,I, the level of E that is implied by the E/w ratio is obtained by equalizing the respective supply and demand prices. We thus consider conditions [1] and [3] for sector c and obtain: is what new-developmentalist authors denominate the "current equilibrium" exchange rate (Bresser-Pereira, 2008, pp. 53-55). 8 Analogously, for sector I we consider conditions [2] and [4]: Where the level E I is what ND authors call the "industrial equilibrium" exchange rate (Bresser-Pereira, 2008, p. 53-55) 9 . Figure 1 (left-hand side) represents the shape of these curves: 9 Bresser-Pereira does not offer a formal definition of EI, but he does define the "necessary price" of a generic industrial good as "the price that makes economically profitable to produce other tradables [other than c] using technology in the state-of-the-art". It therefore coincides with ! ! ! in equation [2]. And "The 'industrial' equilibrium exchange rate [is] the one that enables the production of tradables in the country without the need of duties and subsidies […] the exchange rate that, on average, allows companies using state-of-the-art technology to be profitable or competitive" (Bresser-Pereira, 2008, p. 54). This is, in other words, the level of E that arises from condition [9].
The curves have two intersections. The first one is at r = -1, while the second one is at: (We shall interpret below what happens if condition [11] does not hold) These curves can be used to determine the pattern of specialization of the economy. To see this, let us first define the real wage w for a given consumption basket (C C ; C I ) as: Therefore, the real wage w is univocally determined once the level of E/w is known, and shows an inverse relationship with the latter. According to a wellknown result of choice of techniques (see Kurz and Salvadori, 1995, ch. 5), all this means that the economy will fully specialize in the sector that, given the rate of profits r*, can afford the lowest E/w ratio, which is none other than the sector that can pay the highest w. Therefore, if r > 10 Note that when the production of more than two commodities is considered, in general there will be no level of r that will allow the coexistence of all sectors. On this point, see Steedman (1999, p. 272) and Baldone (2001).
The outer envelope of the curve (black line of the right-hand side of Figure 1) illustrates the economically relevant (E/w)-r configuration. Formally, this relationship is represented by the following condition: From conditions [10] and [11] it can be seen that, given the terms of trade, the size of the interval [0; (the range of full specialization in I) will be smaller the higher is i) the import coefficient of the industrial sector (k) and ii) the higher the labor productivity of the primary sector vis-à-vis labour productivity in the industrial sector. On the other hand, given technical conditions, this range will be smaller the higher are the terms of trade in favour of the primary sector. (Notice incidentally that, if condition [11] does not hold, then ] will be small, and hence, without protection the economy will fully specialize in the production of commodity c. In fact, this is the kind of productive structure envisaged by ND authors, and can be easily represented in terms of our model if, given the rate of profits, E (the "market exchange rate", as the ND authors call it), is en dogenously determined by the "lower branch" of condition [13], i.e., at the level E = E C . This is how the last degree of freedom is eliminated. Then, equations We finish this section with the following remark: while the NDP assumes a plausible productive structure for lA countries, the proponents of this approach seem to conceive this configuration as a purely technical aspect of the economies under consideration, while our framework has revealed that this outcome will generally depend on income distribution (except under the special case in which ). In other words, ND authors believe that it is always the case that holds, and hence, since in this case E I < E C , it is convenient for the economy to specialize in the production of industrial goods.

"RIcARDIAN RENTS" IN ThE BASIc FRAMEWORK
At this point of the exposition we should note that the inverse relationship between r and w/E that emerges from the canonical model is modified once we recall that, according to the NDP, good c is produced by employing a fixed factor whose exploitation under "conditions of extraordinary productivity" allows the sector to earn "Ricardian rents" (Bresser-Pereira, 2008, p. 50). If we assume full employment of the fixed factor 11 , these rents will not be eliminated by the action of competition, and they will eventually be appropriated by the owners of the respective natural resources. The magnitude of this unitary rent (p) is determined by: The most relevant implication in terms of the basic model is that it is now possible to fix the exchange rate independently of the value of the rate of profits. Put it differently, it is now possible to set both r (by its international level) and w/E exogenously, being the rent the endogenous distributive variable of the system 12 . Within this modified framework, there is a linear inverse relationship between the magnitude of rent in terms of commodity C (p C ) and the real wage measured in terms of this commodity ( w . This new definition of the real w age will prove useful in the fifth and sixth sections since, unlike the industrial good, commodity c can be seen as a necessary consumption good ("food"), and hence will play a central role when we address the potential distributive limits of devaluation policy.
If we divide equation [14] by Figure 2 illustrates this relationship.
The rent is maximum when w C is zero, and it disappears when the real wage is the highest compatible with the existence of sector c at the level r*, which occurs when E = E C .
Before stressing the second implication, notice that since E I >E C (or equivalently, r* > ). Therefore, its production can only be sold in the domestic market, provided that there exists some kind of extraeconomic measure that forbids imports of competing goods from abroad and/or tariffs that increase the selling price of the foreign commodity 13 . Then, the second implication is that the additional degree of freedom of the price system allows setting the exchange rate at the level E I to allow the international competitiveness of the industrial sector. Formally, this means that E is determined at the level E I in [13], while sector c earns a persistent extraordinary rent per unit of output. We thus have 8 In our basic framework, the level of E that the economy tends to realize, the "market rate" as Bresser-Pereira calls it, is the outcome of a problem of technical choices. In the absence of rents, given the rate of profits and the relevant conditions of production, we have argued that it was plausible to assume that this level would gravitate around E C for latin American countries (see conditions [10] and [11] in the second section). While it could be set at higher levels (including the level E I ) when differential rents were considered (third section). For ND authors, on the contrary, in these economies the action of market forces causes the effective exchange rate to necessarily gravitate around the level E C . The reason is that this is the only level that ensures the equilibrium of the current account: "The 'current' equilibrium exchange rate [is] the one that balances intertemporally a country's current account and that is therefore also the market rate, the rate on which the market shall converge." (Bresser-Pereira, 2008, p. 53).
We should note at this point that the previous claim combines three different statements: The first two are rather straightforward: a) that the minimum level of the exchange rate that ensures viability of sector c, E C , is also the level that ensures current account equilibrium, say E Eq ; b) that neither current account deficits nor surpluses can be sustained in the long run. And finally, c) it is also implicitly argued that the balance of payments can be brought into equilibrium only by adjustments in income distribution (in the w/E ratio). 13 In formal terms, the tariff ( ) To assess the validity of each of these three statements, let us suppose that the behaviour of the current account (cc) can be described by the following condition: Where X C is the level of exports of the primary good, assumed to be autonomous, and M I is the level of imports of commodity I, which in turn depends on the given level of public expenditures (G), and on private consumption (PC), which is a positive function of the real wage and, therefore, of the w/E ratio. For simplicity the equation abstracts from the flow of financial services. Notice moreover that it is implicitly assumed that the economy fully specializes in the production of C, while it imports all its consumption of commodity I.
By setting CC = 0 in [16], we can derive all the possible w/E -G configurations that equilibrate the current account. This is shown in Figure 3.
The curve is negatively sloped since a rise in the real wage increases PC and hence imports. For a given level of exports, equilibrium in the current account needs a decrease in the level of G. To the right (left) of CC = 0, the economy is in a situation of external deficit (surplus) since for a given real wage, the level of public expenditure is too high (low).
We are now in position to assess whether the abovementioned statements a), b) and c) hold or not. To assess claim a), notice that if we fix the level of G at G in Figure 3, there is a corresponding level of E = E Eq that, given money wages, equilibrates the external sector. And there is no reason why E Eq should be equal to E C , since the latter solves equation [8], while the former is the solution of CC(w/E,G) = 0. The coincidence between E C and E Eq will therefore occur only by a fluke. And this denies the first statement.
let us consider statement b), namely whether surpluses or deficits in the current account are sustainable when, given G = G -, the effective rate that emerges from the problem of technical choices, say E = E - (≥ E C ), differs from E Eq . In other words, is it the case that, if E -≠ E Eq , the effective exchange rate will adjust to E Eq ? The answer needs distinguishing between situations of surpluses and deficits. The NDP argues that a foreign exchange surplus cannot last because there will be a tendency to appreciation of the domestic currency. The argument, however, seems to forget the possibility of quantity adjustments in the market of foreign exchange; in other words, through the working of "compensation mechanisms", either by a central-Bank policy of international reserve accumulation, or by private hoarding of foreign currency, as argued by many scholars of the Post Keynesian School (see lavoie 2001;Frenkel, 2004) 14 .
It is clear, on the other hand, that persistent situations of current account deficits are not sustainable since they would imply an explosive dynamics of the country's external debt 15 . Therefore, if the level of G does not change, there will be a tendency towards depreciation, at least up to the level E Eq .
Finally, with respect to statement c), it can be immediately noticed that given an initial situation of current account surplus or deficit, if equilibrium in the external sector is to be re-established, this can occur either by a change in the w/E ratioas is argued by NDP -or in the level of G (or through a combination of both). Therefore, the necessary connection between income distribution and the balance of payments adduced by the NDP is lost once we admit the possibility of adjustments in the autonomous components of effective demand 16 . hence, statement c) is not generally valid.
Several conclusions emerge from the analysis. While it is still true that the level of E that the economy tends to realize emerges from the technical-choice problem addressed in the second and third sections, this statement must be qualified in the following sense: E cannot be persistently lower than E Eq . hence, if: E C > E Eq , then: E -≥ E C , while if: E Eq > E C , then: E -≥ E Eq . All this can be summarized by saying that, to be persistent, the effective exchange rate must satisfy the following condition: In other words, E C and E Eq must be seen as the minimum thresholds of the effective exchange rate E -.
14 While for Frenkel the sustainability of the external surplus depends on the central Bank's decision to fix the exchange rate and the size of the intervention in the foreign exchange market vis-à-vis the practical limits of monetary sterilization, for lavoie, who follows the Post-Keynesian endogenous money doctrine, compensation mechanisms are endogenous in nature. We will not discuss here whether the compensations mechanisms are exogenous (Frenkel) or endogenous (lavoie). Our point is that there seems to be agreement among heterodox scholars that these mechanisms exist and are effective. 15 We are of course assuming that the economy in question does not issue the currency that is internationally accepted as a means of payment.
Second, notice that, if the condition E Eq > E C holds, since by [17] E must be at least equal to E Eq , it follows that there will be "Ricardian rents" even if the current account is in equilibrium, that is, if E -= E Eq . This feature does not arise within the NDP framework because E C is arbitrarily argued to be equal to E Eq , at least in the long run.
Finally, note that the previous two remarks hold for given levels of autonomous expenditures G and X (and terms of trade). conversely, once G or X are allowed to vary, there is an additional degree of freedom to achieve condition [17]. Then, for instance, if the effective rate is such that: E Eq > E -≥ E C , the reestablishment of condition [17] can be achieved by a decrease in G that reduces the corresponding level of E Eq (or by an increase in X, although this is beyond the control of the Government). This can be summarized by re-expressing condition [17] as:

INDUSTRIAl EQUIlIBRIUM EXchANGE RATE AND DISTRIBUTIvE cONFlIcT
We have already shown that Ricardian rents introduce an additional degree of freedom that allow policy-makers to fix the level of the E/w ratio independently of the behaviour of the rate of profits, and therefore devaluation policy can be employed to promote the development of the industrial sector. For a given value of the rate of profits, the required (gross) rate of devaluation is determined by what Bresser-Pereira (2008, p. 55) calls the "severity of the Dutch Disease" 17 , dh, which is given by: For a given rate of profits and terms of trade, dh positively depends on the degree of structural heterogen eity of the economy, namely: a) the relatively low labour productivity of sector I and b) the high import coefficients of this industry. From conditions [10] and [11] (see the second section), both factors suggest that for latin American peripheral economies, dh, and therefore the required rate of devaluation, could be drastic. This implies that the target exchange rate, E I , may not be socially feasible.
To see this it may be useful to recall that there is a sort of hierarchy among commodities: the reason is that good c can be conceived as the only consumption 17 The Dutch Disease problem will be discussed in the following section. good of which a minimum amount, w a , is necessary for workers' subsistence ("food for the cattle"). Given w = w -, it is possible to determine the maximum level of the exchange rate (E a ) that allows workers to buy that quantity c as: The implication is the following: since E I = E C * dh, if structural heterogeneity, measured by dh, is severe enough, it may well happen that E I > E a . As a result, the required rate of devaluation will be socially unviable, it will provoke full wage resistance, and will therefore neutralize the initial effects of the rise of the exchange rate on international competitiveness.
All this forces us to redefine Figure 2 to incorporate the wage limit to devaluation.  Figure 4, at the minimum necessary level of consumption, a, the maximum rent attainable is determined by equation [15] at the level: ρa = 1 -al C (1 + r). Since , we can also illustrate this limit in terms of an E-w c equilateral hyperbola that shows how the level of the real wage changes when the nominal exchange rate varies. Figure 5 shows the possible range of variations of the nominal effective exchange rate: [20] Its lower limit is E C , since below this level commodity c cannot be profitably produced; while its upper bound is given by E a , since above this value workers are not able to consume the quantity a of c. Finally, the figure shows that, since E a > E I , either the industrial sector does not earn the normal rate of profits or, alternatively, workers are not able to consume the minimum amount of the necessary good.
In the following section we will discuss how the problem manifests itself within the New-Developmentalist framework, and its possible solutions.

DUTch DISEASE AND ITS NEUTRAlIZATION
The NDP subordinates distributive conflict to the potential problem of "reprimarization" of the productive structure, a phenomenon known in the literature as "Dutch Disease" (see corden and Neary, 1982). In general terms, the issue can be described as follows: consider an economy with significant industrial development, in which differences in productivity across sectors are small. In other words, within the relevant interval for the rate of profits, the industrial and primary exchange rates are approximately equal: Then, the argument follows, the discovery of a new natural resource allows exploiting c under particularly favourable conditions. This decreases c-supply price: If we further assume that the higher productivity of c allows the sector to increase its exports level, a foreign exchange inflow will take place. If there are no "compensation mechanisms" (see the fourth section), this inflow will induce a ten-dency to the appreciation of the domestic currency towards the level , now considerably lower than E I 19 . And this provokes, in the absence of protection, the disappearance of industry. The Dutch Disease is illustrated in Figure 6.
According to the NDP, the "correct way" (Bresser-Pereira, 2012, p. 66) to solve the problem consists in implementing a tax (θ ) on c-production that increases the supply price of this commodity ( ) (or what Bresser-Pereira denominates its "necessary price"), and therefore its demand price, by appropriately rising the level of E C . The new supply price is: , which means that the new (and higher) level of the exchange rate that satisfies this condition is: The tax allows re-establishing condition [21], and hence the viability of the industrial sector. however, this solution is inflationary and it may therefore face a "major political obstacle", since the real-wage fall in terms of tradable goods may tackle the resistance of the working class (Bresser-Pereira, 2008, p. 59). 18 In the light of our discussion of the third section about the relationship between the level EC and EEq, we will here assume that EC ≥ EEq and hence the effective exchange rate will gravitate around EC. 19 An analogous effect to the discovery of a natural resource may be obtained if initially both sectors are equally efficient ( ) and a rise of takes place, allowing sector c to tolerate a lower exchange rate, . At the initial level of E, there will exist a foreign exchange inflow due to the price effect, which will push the exchange rate towards the level There are several aspects of the Dutch Disease that are worth discussing. First, notice that when the problem manifests itself, the distributive conflict has already been solved, since the tax limits to counteract the pervasive effects of appreciation, and hence increases the level of exchange rate, and therefore reduces the real wage, only up to its initial value , which by hypothesis was socially viable. Note in any case that, at first sight, there seems to be no way to avoid the potential inflationary pressures associated with the neutralization of the disease by means of the tax. At closer inspection, however, this is the result of the assumptions underlying NDP's framework, in which the only way to induce an increase in the level of the effective exchange rate is through an increase in the supply price of c, since the presence of differential rents in "equilibrium" is neglected.
As we have already seen, on the contrary, the moment these rents are sufficiently persistent, monetary authorities can manage the nominal exchange rate directly (without relying on the aid of a tax), even if the rate of profits is exogenously given. Therefore, it is possible to eliminate the competitiveness gap of the industrial sector by raising E up to its initial level . like the tax on production costs, this policy alone is inflationary too. In this case, however, it limits itself to increase the demand price of c, but does not alter its supply price (in other words, the original value of E C does not change), and therefore its inflationary effect on the necessary good c can be avoided. To achieve this goal, it is enough that, at the same time, the Government imposes a tax on exports (t) that fixes the exchange rate net of taxes faced by c at the lower level . This is none other than a compensated devaluation policy, or effective exchange rate differentiation, endorsed by several old latin American Structuralist scholars, the Argentinean economist Marcelo Diamand (1972) among them 20 .
To see how this works, let us recall that E a is the maximum level of the exchange rate tolerable by workers, and further assume that , which means that the gross rate of devaluation required by the industrial sector, dh = This is shown in Figure 7 by means of the E-w C equilateral hyperbola introduced in the fifth section. An export duty of magnitude t shifts the curve upwards (dotted line), and allows a higher real wage for each level of E. Moreover, the effective ex-20 See Dvoskin and Feldman (2015) for a formal reconstruction of Diamand's thought. change rate for the industrial sector is equal to E I , while the one faced by sector c is E a . , since in this case the maximum rate of c-consumption, w C , will remain at its initial level.

cONclUDING REMARKS
The present paper has attempted to reconstruct the New-Developmentalist approach, by means of a formal framework that endogenizes the productive structure of a small-open-peripheral economy, as the outcome of a problem of technical choices. This framework allowed us to inspect the conditions that the exchange rate must satisfy to ensure the viability of the industrial sector while simultaneously respecting the restrictions imposed by the balance of payments and the distributive conflict. Once differential rents are admitted, we have shown that export duties are a more efficient tool to neutralize the Dutch Disease than the imposition of a tax on the primary sector production costs, since the former extracts rents from the owners of natural resources without necessary increasing the domestic price faced by workers. At this juncture, one must stress, however, that distributive conflict may also emerge from the resistance of rentiers who may exert sufficient political power to hinder the policy of exchange rate differentiation.
If in the light of our framework we now return to the more general discussion about the role of the exchange rate as a tool for sustained economic growth, one should warn that setting the exchange rate at the industrial equilibrium level seems to be a necessary yet not a sufficient condition for industrial development. In particular, once the viability of the sector is ensured, the rise of exports will depend on the "extent of the market", i.e., on the evolution of global effective demand. And given this level, it is clear that the industrial exports of a particular economy can increase only to the extent that there is a simultaneous fall in the level of exports of her competitors, and this will imply "exporting unemployment" to her trade partners 21 . Thus, one should expect that in response to devaluation in the domestic economy, competitors devalue their own currencies as well, thereby starting a currency war that ends up in a zero-sum game.
At the global level, on the other hand, we should also expect that if a significant group of nations simultaneously devalue their currencies to promote exports, the growth rate of global output decelerates, due to the negative influence that the exchange rate exerts on the real wage. And this effect will tend to diminish the level of exports of all the economies involved in trade. It could be therefore concluded that exchange rate policy seems to be more effective to counteract negative shocks, such as devaluations pursed by trade partners or the tendency towards reprimarization, than as an effective tool for economic development.