1. Introduction

1In a seminal paper, Bhagwati [1958] using a two-good, two-country HOS framework showed that economic expansion (due to an increase in the volume of production factors, or to a technological shock) “may harm the growing country itself”. This paper has given birth to a huge literature on the effects of income transfers (Bhagwati, Brecher, Hatta [1983]; Majundar and Mitra [1985]; Yano [1983]) and technological progress (Hatta [1984]; Mantel [1984]) on developing countries’ welfare. Hatta [1984] has demonstrated the analytical conditions for non-immiserizing growth in an n-commodity, two-country HOS framework, following the progress of the developing country’s technology. The most important conditions are: all commodity pairs are substitutes in consumption in the developing country, there are no complements in commodity pairs in production of the developing country, and in the developed country the import demand functions satisfy gross substitutability! [1]

2The effect of the relaxation of some of these restrictive conditions on the developing countries’ welfare has become a wide field of research. Most of these contributions use the Ricardian trade model that takes into account the international differences in technology, rather than the Heckscher-Ohlin trade model. Using a Ricardian setting, Samuelson [2004] considers that, under free trade, technological improvement in the industry where the developing country has a comparative advantage may harm its welfare. He states that: “Self-immiseration is a well-known phenomenon in the economic literature, and it does crop up in the debate over globalization”. [2]

3Moreover, in a world where technology freely flows between nations, economic expansion can originate from developed countries which transfer their superior technology to developing countries (Cheng, Qiu, and Tan [2005]; Jones and Ruffin [2008]; Ruffin and Jones [2007]).Beladi, Jones and Marjit [1997] in a two-country Ricardian model consider that a developed country transfers the technology of an industry where it has a comparative disadvantage into the developing country’s exporting industry. Using a diagramatic analysis, they show that immiserizing growth may hurt the developing country’s welfare. Ruffin and Jones [2007, p. 212] also study geometrically this pattern of technology transfer, and conclude that: “Sufficiently low elasticity, coupled with a large value for the developing country import propensity, could result in a drop of its real income”. The case, where a developed country transfers the technology of the industry where it has a comparative advantage into a less developed country, has been formalized and analyzed in depth by Jones and Ruffin [2008] and Ruffin and Jones [2007]. The two authors disclose a “technology transfer paradox”. They show, in a two-country Ricardian model, that even in the situation where the specialization of the developed country is reversed, its welfare may increase following the transfer of technology (TT) to the developing country.

4In the model we build hereafter, following Beladi, Jones and Marjit [1997], Cheng, Qiu, and Tan [2005], Jones and Ruffin [2008], Ruffin and Jones [2007], technology is the know-how, blueprints or management skills which can be transferred by developed countries’ multinational firms (MNFs) to developing countries. This corresponds to the view that capital accumulation is no more the main factor of economic development, but “that in recent decades more attention is paid by economists to factors such as education and technological change” [Weil, 2009]. This statement is confirmed by Cheng, Qiu, and Tan [2005] when they study the MNFs behavior: “Capital movement, from the home countries of MNFs to the host countries, seems to have become the least important ingredient of FDI. In contrast, technology and managerial talent have become the key ingredient” [3]. From an empirical point of view, it is noteworthy that The World Investment Report (UNCTAD), each year, contains a special section on new “non-equity forms” of transnational corporate activity. These include subcontracting, management contract, franchising, licensing and product-sharing [4].

5Moreover, in this contribution, we consider a two-country, two-good model (extended in the last section to n goods) in which technology is transferred for free from the industry where the developped country has a comparative disadvantage to the developing country’s exporting industry (following Beladi, Jones and Marjit [1997]; Cheng, Qiu, and Tan [2005]; Ruffin and Jones [2007], this last paper even envisages that technology may be stolen by the developing country). There are two basic ideas behind this hypothesis of free transfer of technology.

6First, in our model, TT does not concern cutting edge technology, but technology which is standard in the developed country’s comparatively disadvantaged industry. This technology can be regarded as a public good, since we hypothesize that it has been used extensively, and for a long time, by the developed country’s FMNs. At the opposite, for FMNs, high technology is a fundamental factor of their comparative advantage in their competition with the developing country’s firms. There are strong arguments why these FMNs do not transfer their advanced technology. They fear not to be able to protect this technology from imitation, or even robbery in the foreign countries. Japanese economists (Kojima [1978, 2000]; Ozawa [1992, 2005, 2009] have formalized a so-called model of “flying geese” in which the leading country transfers its standard technology, in the industries where it has a comparative disadvantage, to the countries which have a comparative advantage in the same industries. The leader keeps its supremacy in high technology industries.

7Second, for non equity forms of corporate activities, it is the interest of the developed country’s mnfs to enter into technological cooperation with foreign firms (Hoekman, Maskus and Saggi [2005]; Pack and Saggi [2001]; Saggi [2002]) and to support the efforts of their suppliers to increase the efficiency of their production process. By this way, firms which transfer their technology benefit from the cost reduction of the goods they import. This reduction may result in the fall of the terms of trade of the developing country, following the TT. There is much descriptive and empirical research which puts forward this deterioration of the terms of trade [5]. Sarkar [2004, p. 167-168] has estimated a log-linear trend equation which shows the terms of trade deterioration for a wide range of developing and emerging countries, in Asia and south America, in the 1980s and 1990s. The same evolution is observed for the 2000s (World Bank [2009], p. 335). Li, Huang and Li [2007] confirm that, in China, the terms of trade of commodities dropped by 17% between 1995 and 2004. In the 1990s, processing trade which resulted from different sub-contracting and licensing agreements between Western MNFs and Chinese firms were found in clothing, shoes, toys, and furniture industries. At the beginning of the 2000s, consumer electronics became dominant, but this shift has not reversed the deterioration of this country’s terms of trade.

8However, the deterioration of the terms of trade does not necessarily entail immiserizing growth (Bhattacharya & Raychaudhuri [2004]). Sawada [2009] has investigated the empirical reality of this negative evolution. He evaluates a representative consumer’s welfare, based on revealed preference theory, for a wide range of developed as well as developing countries. He singles out mostly sub-Saharan, but also Central and South American developing countries which during the period under survey (1975-1985) experienced immiserizing growth, that is the growth of their GDP per capita coupled with a decline of their welfare. This study also shows that many developing countries (including most Asian countries) increased both their welfare and their GDP per capita during this period. To our opinion, these diverging evolutions justify that we analyse in depth the factors which determine the positive or, at the opposite, the negative impact of international technology and trade flows on developing countries’ welfare.

9Concerning our model fundamental hypotheses, it must be noted that some theoreticians (Markusen [1984], Helpman [1984]) consider that FMNs originate in imperfectly competitive markets, which result from increasing returns to scale, or product differentiation. However, our contribution differs in important ways from these models. First, in our contribution, trade and MNFs arise from technological differences between a developed and a developing country. Second, we adopt the view that, between countries with different level of economic development and technological advancement, inter-industry trade prevails over intra-industry trade. Third, there is free competition on the products markets with neither transportation costs, nor trade barriers. We think that this hypothesis holds for industries where developing countries are specialized, and benefit from TT: textile, clothing industry and consumer electronics.

10Moreover it is clear that, when a developed country transfers its more efficient technology to a less developed country, the world production increases. The main objective of our contribution is to study how this increase is shared between both countries, in other words who gains and who loses from this transfer (Ruffin and Jones [2007]). Thus, we analyze the impact of TT on the developing country’s welfare, and the possibility of immiserizing growth for this country. As a consequence, we leave aside employment adjustment on the labor market, and hypothesize that it is in equilibrium [6].

11Finally, in our contribution we build a model which formalizes and generalizes the situation which has been represented and analyzed geometrically by Beladi, Jones and Marjit [1997], and Ruffin and Jones [2007]. First, using a Ricardian setting, we take into account the fundamental factors of the free trade equilibrium. Then, we show that TT is a shock which may result in a change of the specialization pattern of the developing country. Our contribution is the first to show how this change is a buffer which may protect it against the adverse evolution of the terms of trade. We disclose the analytical conditions of such a protection.

12The organization of the rest of this paper is the following. In section 2, we present our Ricardian model which includes the consumer preferences’ structure, the efficiency of the technology which is transferred, and the relative size of both countries. In section 3, we study the developing country’s welfare after TT, taking into account the possible change of its specialization pattern. We establish the detailed conditions of its welfare decrease. In section 4, we generalize our model in a world with n goods, and envisage different other extensions. In section 5, we conclude.

2. The model: introducing technology transfer in a Ricardian setting

13We distinguish a developed country (noted by *) and a developing country, two goods 1 and 2. Labor is the only factor of production in each country. Labor supply in the developed country and the developing country is fixed at L * and L, respectively (this hypothesis is relaxed in our section 3). Labor is fully employed in each country, and is internationally immobile. Goods are freely traded in the absence of any transportation costs. The developing country’s production technology for good 1 and good 2 is described by two unit labor requirements a₁ and a₂, respectively. Similarly, the developed country’s unit labor requirements are given by and .

14Given our interest in analyzing TT from an advanced country to a less developed country, we shall focus on the case in which the developed country has an absolute advantage in the production technology of both goods  [7]. The underlying conception is that the effects of R & D investment in the developed country unevenly spread over its industries, nevertheless it results in an absolute advantage for this country over all the developing country’s industries [8].

15We suppose that the developing country has a comparative advantage in the production of good 1, and the developed country in the production of good 2. The prices of each good is respectively p₁ and p₂. We consider good 1 as numeraire. So the terms of trade are: p = p₂/p₁.

16In order to investigate the consequences of free trade and TT on both countries’ welfare, we use a CES utility function which represents the consumers’ preferences. The condition of equilibrium of the trade balance, gives the terms of trade between the two goods in the case of full specialization of each country (appendix A).

17where ? is the parameter of the CES utility function, and:

18 is the elasticity of substitution between the two goods.

19If the developing country is relatively large compared with the developed country , for a given level of the labor coefficients, p increases with ?, that is when the elasticity of substitution between the two goods diminishes. In that case the developing country production’s is higher than the developed country production’s. If the substitution between these two goods is low (p> 0, ? < 1 ), good 1 does not compete on the world market with good 2, and its relative price falls. Conversely, its terms of trade increase with ? if it is relatively small compared to the developed country, that is to say if: . So, if it is completely specialized in the production of one good, the issue of the developing country immiserizing growth is linked with its large size (see figure 1 here after).

20Since technology is captured by the set of labor coefficients per unit of output of each commodity, following the TT, the developing country’s labor coefficient is reduced ( replaces a₁ ).

21Taking into account (1), we obtain:

22In the rest of our paper, a prime ( ' ) refers to the variables after TT.

23These relative terms are an exponential function of ?; they increase when the elasticity of substitution between the two goods decreases. Indeed, in that case, good 1, whose production increases, does not compete with good 2 on the international market. Thus, the deterioration of the terms of trade for the developing country increases with the efficiency of the technology which is transferred, and when the elasticity of substitution between the two goods diminishes.

24Each country’s welfare is modeled by the consumers’ utility function (Samuelson [2004], p. 138-140). Thus to study the evolution of the welfare of each country, we must formalize U as a function of y and p. Taking into account, under free trade, the welfare before (B *) and after (B *' ) the TT, we obtain for the developed country, under full as well as under partial specialization of this country (demonstration in appendix A):

25Since p' >p and ? > - 1 ==> B * /B * > 1

26By comparison with free trade without TT, the developed country always gains from the TT. Indeed, this result is straightforward since the developed country has the same production as before the transfer, and the terms of trade of good 2 improve unambiguously.

27The situation of the developing country is more complex to study. Indeed, compared with free trade before the TT, its production of good 1 increases with labor productivity, but its terms of trade deteriorate. Equation (4) applies for full, as well as for partial specialization, of the developing country. For ?> 0, the first factor is larger than 1, and the second smaller than 1.

28Our objective is to study the evolution of the different effects on the developing country’s welfare: the increase of the production due to TT (first factor in equation 4), the deterioration of the terms of trade for the developing country (second factor), when the elasticity of substitution varies. But we must also take into account that, following the TT, this deterioration may be limited by the possible partial specialization of the developing country in the production of commodity 2.

3. Changes in the specialization may benefit the developing country

29If the developing country’s welfare is formalized by a Cobb Douglas function ( ? = 1 ), it always gains from the TT regardless of its relative size and its specialization pattern. When the elasticity of substitution is higher than 1, and when ? decreases (between 0 and -1), the developing country’s welfare (equation 4) increases compared with free trade on this interval of variation of ?. For example, when it tends to -1 (the elasticity of substitution tends to infinity), the terms of trade do not change compared with free trade (equation 2), and the developing country reaps all the benefits of the transfer.

30From now on, we study the evolution of the developing country’s welfare when the elasticity of substitution between the two commodities is low (with ? varying between 1 to 0, that is to say with ? varying between 0 and infinity), which is supposed to correspond to immiserizing growth [9].

31If we start the analysis considering that the elasticity of substitution between the two goods is very low ( ? tends to 0), a paradoxical result emerges. Our model shows that, with the deterioration of its terms of trade, the developing country shifts to partial specialization (the production of commodity 2). Thus, it can be proved that when the elasticity of substitution between the two goods is very low, the developing country’s welfare never decreases whatever its size (see table 1 hereafter and appendix B).

32Does this mean that the developing country can never lose from the TT? The answer is no, since when the elasticity of substitution varies between 1 and 0, there are certain conditions where the shift to partial specialization does not prevent the fall of the developing country’s welfare.

33We call the critical point [10], the point where the developing country shifts from complete specialization to partial specialization, which also corresponds to its minimum welfare (compared with free trade before TT) and to the “critical relative size” (L/L* )_C of the two countries. For a given value of ?> 0, that is to say an elasticity of substitution inferior to 1, the developing country’s relative welfare is a decreasing function of its relative size if it is completely specialized in the production of good 1 [11]. However, when L/L * increases above the critical size, the developing country shift to partial specialization and the terms of trade no longer deteriorate for this country after TT. Before TT, it continues to deteriorate, since under free trade, the shift to partial specialization takes place for a superior relative size of the developing country. So, for a given value of the elasticity of substitution (inferior to 1), there is one and only one critical point for which B ' /B is minimal. This can be illustrated by our figure 1. We compare the relative developing country welfare before and after the TT (B ' /B ) with an elasticity of substitution between both goods of 0.5. The critical point is in F. Before this point the developing country is completely specialized in the production of good 1. After this point if it is larger than the critical size (L/L * > 12.6 ), its shift to partial specialization, its terms of trade no longer deteriorate, and it takes advantage of the transfer of the more efficient technology.

34The above definition of the critical point yields the following expression (5): this is the point where the price which corresponds to complete specialization of the developing country after TT equals the price corresponding to partial specialization of this country in the same situation:

35So the critical relative size is:

36For a given ?>0, (L/L* )_C is an increasing function of that is to say of the ratio of the relative efficiency of the technology in industry 2. If this ratio is high (developing country uses an inefficient technology in industry 2 compared to the developed country), it will shift to partial specialization only if it is big enough relative to the developed country (equation 6). That is to say, it may lose from the TT if it is inferior to the critical size (see hereafter figure 1, the case where ? =0.5, and L/L * < 12.6 and

37The critical point, and hence the developing country’s welfare, depends fundamentally of the relative efficiency of its technology in the industry where it has a comparative disadvantage. So, it is interesting to single out, for what value of this parameter, this welfare may decrease.

38Taking into account (4), (5) and (6), the developing country’s welfare decreases if:

39If (7) is satisfied, the technology in the industry where the developing country has a comparative disadvantage is backward. When it shifts to partial specialization, it is already in a situation of an adverse evolution of its welfare (B’ /B < 1 ).

40To discuss more empirically of the possibility of immiserizing specialization, we introduce numerical examples hereafter. For that purpose, we use some of the parameters drawn from Samuelson’s famous article on globalization (2004). Using a Ricardian framework, he studies how the USA and Chinese trade and welfare are affected by a technological shock on the Chinese economy. For his simulation, he uses the following values: a₁ = 5  [12] which are supposed to approximate the economic parameters which were observed for each country, at the end of the 1990s. It is noteworthy that the relative efficiency of the USA is much higher in the industry where it has a comparative advantage than in the other sector. In the industry where the developed country has a comparative advantage, the efficiency ratio between both economies is 1 to 40!

41In table 1, we used equations (6) and A10 (in appendix) to compute the relative critic size of both countries and the developing country’s welfare, before and after TT, for different relative efficiency of the technology where it has a comparative disadvantage. For we have taken Samuelson’s data about the developing country’s technology which is considered as very backward . In a second simulation, the gap between both countries is only 1 to 10 . Moreover simulations are computed for a wide range of the elasticity of substitution varying from 0 to 1.

42We remind that the critical relative size is the point where the developing country shifts to partial specialization, which corresponds to the minimum value of B ' /B. Table 1 confirms that the developing country always gains from the transfer (B ' /B > 1 ) when ? is close to 1 or 0. In the first case, the effect of the increase of the production of good 1 exceeds the effect of the fall of the terms of trade. In the opposite situation, the terms of trade falls following the TT, the developing country shifts to partial specialization for a relatively small critical size, which protects it against the adverse evolution of the terms of trade.

Table 1. Relative critic size (L/L* )_C and minimum welfare of the developing country after and before TT (B ' /B ) for different values of the elasticity of substitution between goods ( ? ) and relative efficiency in the industry where it has a comparative disadvantage

Table 1. Relative critic size (L/L* )_C and minimum welfare of the developing country after and before TT (B ' /B ) for different values of the elasticity of substitution between goods ( ? ) and relative efficiency in the industry where it has a comparative disadvantage

43For intermediate values of ?, a very backward technology is an handicap for the developing country since it prevents it from shifting to partial specialization which hurts its welfare. If its technology is moderately backward, this shift is a buffer which protects it against the adverse evolution of the terms of trade (if ? = 0.5) or it limits its lost (if ? = 0.09).

44So it is remarkable that immizerizing growth following the TT is limited to a situation where the elasticity of substitution between the two goods is small, but not too small, and where the technology used by the developing country in its comparatively disadvantage industry is very inefficient compared with the developed country.

45However, if we consider that the set of labor coefficients in table 1 concerns the USA and China, we find that the latter may lose in welfare following the TT from the former [13]. Indeed, the ratio of the labor employed in manufacturing industries as well as the ratio of population between China on the one hand, and the United States on the other hand, was approximately 4.5 at the beginning of the 2000s, which corresponds roughly to the critical size, for a very low elasticity of substitution between both goods ( ? = 0.09, table 1). In fact, several factors may explain that China has gained in welfare from the TT. First, Chinese trade concerns other developed countries (Europe and Japan) which reduce the ratio L/L*. Second, China has also imported technology into the industry where it has a comparative disadvantage (decreasing the ratio , thus shifting to partial specialization in this industry, and escaping for a large fall of its terms of trade (see next section, the n commodity model on this point).

46But a last question arises: even if the developing country loses at the critical point, if it is larger than the critical size, may it gain from the TT (always compared with no TT)? We are going to show that the answer to this question is yes, studying in details the case where, ? = 0.5 and (see table 1 and figure 1).

47In Figure 1, we have represented the evolution of the developing country’s welfare before and after TT relaxing the hypothesis that L/L * is fixed (and using equations 4, 5, 6, 7, the demonstration is given in appendix B). On the segment AC, the developed country is big compared to the developing country. This is the most favorable situation for the latter since the former is partially specialized before and after the TT, thus the terms of trade do not change. On the segment CD ( 2 < L/L * < 5 ), the developed country is partially specialized under free trade before the TT, but both countries are completely specialized after the TT. The developing country still gains from the TT compared with free trade before the transfer. Between D and F, both countries are completely specialized before and after the TT. When the relative size of the developing country increases, its terms of trade deteriorate (see appendix B). For L/L * = 6.3 (point E), its welfare does not change after the TT (B/B ' = 1 ). Then it declines between E and F. F is the critical point, where L/L * = 12.6, and its welfare is minimum (B ’ /B = 0.74 ).

Figure 1. Evolution of the developing country’s welfare ( B ' /B ) after and before the TT, when its relative size compared to the developed country ( L/L * ) varies (and with a elasticity of substitution between goods of 0.5).

48Indeed for a larger L/L*, the developing country is partially specialized after the TT, and begins to produce commodity 2. Thus it allocates a part of its labor from industry 1 to industry 2, and the terms of trade do not deteriorate any longer. Its welfare improves compared with free trade before TT, since in the latter case this country is still fully specialized in the production of good 1 and its terms of trade continue to deteriorate (segment FH). Then (on segment HI with L/L * > 20), when the developing country is partially specialized before and after the TT, its welfare gain is constant (B ' /B = 1.29 ).

49In this example, with a realistically inelastic consumer utility function ( ? = 0.5 ), if the ratio of the population of the developing country divided by the developed country’s is smaller than 6.3, or bigger than 16.3, the developing country’s welfare increases instead of the very poor efficiency of industry 2.

50Finally, compared with the literature on technological change in the industry where the developing country has a comparative advantage (Beladi, Jones, Marjit [1997]; Jones and Ruffin [2008]; Ruffin and Jones [2007]; Samuelson [2004]), we have disclosed the precise conditions under which the developing country reallocates a part of its labor force, from the industry which benefits from this transfer, to its comparatively disadvantage industry. By this way, the adverse evolution of the terms of trade is limited. As a consequence the possibility for the developing country to shift its specialization towards the industry where it has a comparative disadvantage is crucial. From a general point of view, the developing country risks to lose from the TT if three conditions are simultaneously met (see equations 5, 6, 7).

1. The elasticity of substitution between the two goods is low, but not too low otherwise the developing country shifts to partial specialization and its welfare never decreases (see Table 1).
2. The relative size of the developing country compared to the developed country must not be too big (otherwise the developing country shifts to partial specialization, thus reducing the fall of the terms of trade, equation 6 and figure 1).
3. The technology which is used by the developing country in its comparatively disadvantage industry is very backward compared with the developed country (equation 7, table 1, and figure 1).

4. Extensions of the model

51In a two-good model, at least one country produces one good. So the change in the production specialization of both countries, which plays an important role in our analysis is very simple. Here after, we extent briefly our analysis to a model with n goods (n finite), focusing on the effect on the developing country’s welfare of the change in its specialization pattern following the TT [14]

52The set of comparative advantages can be written as:

53The developing country has the highest comparative advantage in the production of commodity 1, and the developed country in the production of commodity n. We also suppose that the developed country has an absolute advantage in the production of the n commodities.

54Following our previous results (equation 4), the impact of the TT on the developing country’s welfare depends on two opposite factors.

1. The increase in the production in the industries whose efficiency is improved by the TT.
2. The fall in prices of these industries.

55To analyse this impact we first consider that before TT, the developing country is completely specialized in the production of the k first goods (and therefore the developed country in the production of commodities k + 1 to n), and that the TT concerns all the k commodities produced by the developing country. If there is no change in the specialization, of both countries after TT, it means that all the relative prices of the k first goods will drop. This is an adverse evolution which may occur if the substitution between the goods which are produced by both countries is low (Jones [1979], p. 284-285). However the above analysis has shown that if the developing country is big enough, or (and) its technology to produce commodity k +1 is efficient enough, it will shift to the production of this commodity. The point where the developing country shifts to the production of commodity k +1 is the critical point in our new n commodity framework.

56Let’s now turn to the case where the developing country produces commodity k + 1. Then, the developed country may still be partially specialized in the production of k +1, or give it up. In both cases, the developing country transfers a part of its labor force to industry k + 1, and the developed country withdraws a part of its labor force from this industry, to industries k + 2... n. These two opposite movements of factors, and hence of the supply of goods will result in a moderation of the fall of the developing country’s prices, and in a decrease in the developed country’s prices. Like in the two-good model, the transfer of labor force by the developing country to new industries is a buffer which reduces the adverse evolution of prices.

57Let’s study a particular situation where the developing country is partially specialized in commodity k before TT, and jumps to partial specialization in the production of k + 1 after the TT. Before the TT, we have

58and after the TT:

59and

60with

61Where w, w*, w', w'* are respectively the nominal wage in the developing and in the developed country, before and after the TT. Thus, in this particular case, the developing country’s relative wage inevitably decreases, but this decrease is small if a_{k +1} is close to , a condition that we have also obtained here above in the two-good model (equations (6) and (7), see also Jones and Ruffin [2008], p. 325-326).

62We consider now another case where the TT concerns only industry 1. This hypothesis is favourable to the thesis of immiserizing growth [15]. Again, the important point is to find out if the developing country will shift from production k to production k + 1 (possibly k + 2...) following the TT. This will depend on the importance of the reallocation of workers from industry 1, to industry 2, ... k +1, and hence of the possibility of substitution between commodity 1 and the goods which are produced by the developing country on the one hand, and also of the possibility of substitution between these goods, and those which are produced by the developed country.

63If the goods which are produced by the developing country have a low elasticity of substitution with those which are produced by the developed country (but not “too low”, see the two-good model), and the comparative advantage of the developed country for the industry k +1 is much higher than for industry k, the developing country may lose from the TT. Indeed, in that case the developing country will not shift to the production of good k + 1. We find here again the fundamental argument concerning the relative efficiency of the technology of the developing country in the neighbourhood of the critical point.

64Finally, it is noteworthy that, in the case of a continuum Ricardian trade model (Dornbusch, Fisher, and Samuelson [1977]), there cannot be any critical point, so the developed country can only transfer a subset of its comparative advantaged industries to the developing country. Thus, there is no possibility that trading partners expand, or contract, the production of one commodity. So this model is less appropriate than ours to take into account large technological shocks which may change their specialization patterns, and affect their welfare (Jones and Ruffin [2008]).

65Other extensions of our basic model can be envisaged. Here after we briefly outline the most important of them.

66First, we have hypothesized that technology is transferred for free from the developed to the developing country. However, the developed country’s MNFs may patent their inventions and sell it to the developing country’s firms. In that case, to counterbalance the MNFs’ R&D investments, these firms commonly pay royalties which are proportional to the production of the industry where the technology has been transferred (see Pugel [1988]; Brecher [1982]; Kopitz [1979]). The developing country’s welfare is then negatively affected, and it can lose even if the elasticity of substitution between goods which are produced by both countries is relatively high (equal to 1 with a Cobb Douglas utility function) [16].

67Second, the change in the developing country’s specialization is the main mechanism which protects it against the fall of the terms of trade. Introducing economic and social costs which are linked with this change into our model could affect negatively the developing country’s welfare.

68Third, in our model, labor is homogenous. However, it could be interesting to distinguish between different labor skills and to consider that skilled labor is complementary with modern imported technology. In our model, the fall in the labor coefficient of the industry (industries) where the modern technology is transferred implies that the developing country’s labor has the necessary skill to use this technology. Introducing the investments in education and training which corresponds to this hypothesis could be a useful extension of the model.

5. Conclusion

69The study of the model with n goods confirms that immiserizing growth linked to the TT from a developed country is a phenomenon which concerns the least developed countries. These countries produce manufacturing goods which are not substitutes with those which are produced by developed countries. In addition, they have very inefficient technology in the industry where the developed countries have a comparative advantage. This factor is fundamental, because we have shown that any progress in the ladder of comparative advantages cushions the adverse evolution of the terms of trade for these countries. These results have two important consequences

70First, for developing countries, it is crucial to upgrade their specialization pattern. Such an evolution is the condition to take advantage of the globalization of trade and of the technology which may be transferred by developed countries. However, the poor quality of infrastructure, or the low qualification of the manpower, may be obstacles to this change of specialization. If the economic and social costs to create new emerging activities, and move internal resources to these activities, are very high, the spectre of immiserizing growth cannot be ruled out.

71Second, the “emerging” countries (say China, India), which could be regarded decades ago as underdeveloped, have gradually climbed up the ladder of comparative advantages. To do so, they have reallocated their manpower, from their traditional industries where they had their highest comparative advantages, to produce more and more commodities of the industries of the middle of the ladder. The TT into their “traditional” industries (textile, clothing) from developed countries was in the interest of the latter (because they benefited from the evolution of the terms of trade), and also, if we consider our model, of the former. An important general result of this paper is that, even if such transfers are concentrated in the traditional industries of these countries, they save (human) resources which can be employed in new emerging industries (say consumer electronics), and by this way, help them to change their specialization pattern, thus protecting them against immiserizing growth.