Palma,G.2004DeindustrialisationDutchDisease

contact:e-mail: [email protected] / [email protected]

tel: +27 12 302 2740

innovative employment strategies

employment growth & development initiative

FOUR SOURCES OF ‘DE-INDUSTRIALISATIONAND A NEW CONCEPT OF THE ‘DUTCH

DISEASE’

by Gabriel Palma

University of Cambridge

presented at

HSRC EGDI Roundtable

The Changing Character of Industrial Development:What Implications for Growth, Employment and

Income Distribution?

21 May 2007

HSRC EGDI ROUNDTABLE —

The Changing Character of Industrial Development: What Implications for Growth, Employment and Income Distribution?

2

FOUR SOURCES OF ‘DE-INDUSTRIALISATION AND A NEW CONCEPT OF THE ‘DUTCH

DISEASE’1 2

Gabriel Palma

Faculty of Economics and Politics, University of Cambridge

November 2004

1 This paper builds on Bob Rowthorn’s influential work on this subject. I am extremely grateful to him for sharing his data and for many lengthy discussions on the subject. I would also like to thank Amit Bhaduri, Daniel Hahn, Geoff Harcourt, Andrew Glyn, Richard Kozul-Wright, Hashem Pesaran, Carlota Perez, Bob Sutcliffe, Guy Standing, Ben Turok and especially Stephanie Blankenburg, Jose Antonio Ocampo and Fiona Tregenna; and participants at seminars in Bangkok, Bilbao, Cambridge, Cape Town, Chicago, Geneva, Kuala Lumpur, Mexico City and Santiago for their helpful comments on a previous draft. Lastly, I am very grateful to John Wells, with whom I had frequent discussions on the subject before his sudden death. The usual caveats apply. 2 Paper published in J A Ocampo, Beyond Reforms: structural dynamics and macroeconomic vulnerability, Stanford University Press and World Bank, 2005.

Four Sources of ‘De-industrialisation’ and a New Concept of the ‘Dutch Disease’

3

Contents

Tables……………………………………………………………………………….3 Figures………………………………………………………………………………3

1. Introduction……………………………………………………….………....5 2. The four sources of ‘de-industrialisation’ ...............................................................10

2.1 The first source of ‘de-industrialisation’: an ‘inverted-U’ relationship between manufacturing employment and income per capita………………………………………………………………….………10 2.2 The second source of ‘de-industrialisation’: a declining relationship between income per capita and manufacturing employment…………………..11 2.3 The third source of ‘de-industrialisation’: a declining income per capita corresponding to the turning-point of the regression…………………………13 2.4 The fourth source of ‘de-industrialisation’: the ‘Dutch Disease’………....15

3. Trying to swim against the de-industriali-sation tide ............................................31 3.1 Finland and the diversity of Nordic countries…………………………...31 3.2 East- and Southeast-Asian industrialisation……………………………...34 3.3 Mexico and Central America's ‘maquila’ (labour-intensive assembly-end part of the value chain) industrialisation………………………………………36 3.4 The Andean Pact countries in South America…………………………..40 3.5 China, India and Turkey…………………...……………………………42 3.6 South Africa: still searching for an industrialisation path? …………...….43

4. ‘De-industrialisation’ – does it matter ?...................................................................44 5. Conclusions .................................................................................................................48

Tables

Table 1 – Employment in manufacturing (% of total) ........................................................9

Figures Figure 1 – European Union: manufacturing and services, 1960-2000..............................6 Figure 2 – Rowthorn’s regression: manufacturing employment and income p.c., 1990

...........................................................................................................................................10



4

Figure 3 – Second source of ‘de-industrialisation’: a declining relationship, 1960-98..11 Figure 4 – Third source of ‘de-industrialisation’: a changing turning-point in

regression, 1960-1998 ....................................................................................................14 Figure 5 – Turning-point in the five regressions and United States income p.c.,

1960-98.............................................................................................................................15 Figure 6 – Fourth source of ‘de-industrialisation’: cases of ‘overshooting?’..................16 Figure 7 – The ‘primary commodity’ and ‘export-services’ effects, 1998 ......................18 Figure 8 – Changes in manufacturing employment and income p.c., 1960 and 1998619 Figure 9 – The Netherlands: unravelling the ‘Dutch Disease’? 1960-998 ......................20 Figure 10 – The United Kingdom: catching the ‘Dutch Disease’? 1960-98 ..................22 Figure 11 – The Netherlands and five countries of the European Union, 1960-98.....23 Figure 12 – Netherlands and ‘UCAN’: two processes of de-industrialisation but only

one ‘Dutch Disease’ .......................................................................................................24 Figure 13 – Greece, Cyprus and Malta: a ‘tourism-based Dutch Disease’? ...................25 Figure 14 – Switzerland, Luxembourg & Hong Kong: a ‘financial-services-based

Dutch Disease’? ..............................................................................................................26 Figure 15 – Argentina, Brazil and Chile: Catching the ‘Dutch Disease’? 1960-98........27 Figure 16 – Brazil and the Southern Cone Countries of Latin America, 1998..............29 Figure 17 – A. Nordic Countries: three different industrialisation paths, 1960-98 ......31 Figure 18 –B. Finland & Chile: an ‘anti’-Dutch disease and a Dutch disease

industrialisation? .............................................................................................................32 Figure 19 – C. Finland: changing vertical integration in timber-based exports.............33 Figure 20 – NICs 2: Manufacturing employment and income p.c., 1950-98 ...............35 Figure 21 – NICs 1: Manufacturing Employment and Income p.c., 1970-98..............36 Figure 22 – Mexico and Central America: the 'maquila' effect, 1970-98 ........................37 Figure 23 – Mexico: structure of ‘maquila’ output (%), 1990-1999.................................39 Figure 24 – Mexico: GDP and exports, 1950-2000 ...........................................................40 Figure 25 – The Andean Diversity in South America, 1998.............................................41 Figure 26 – China, India & Turkey: manufacturing employment and Income p.c.,

1950-98.............................................................................................................................42 Figure 27 – South Africa: manufacturing employment and income p.c., 1950-98 .......43


5

1. Introduction

Using Kaldor’s well-known concept, one of the most notable ‘stylised facts’ of the post-war period is the rapid decline in manufacturing employment in most industrialised countries and in many medium- and high-income developing countries (DCs). Although it is well known that over the long-term course of economic development, the structure of employment changes substantially, relative changes in employment on the scale and at the speed occurring during this period constitute a phenomenon without precedent.

In essence, during the long-term course of economic development, changes in the structure of employment are set in motion by an increase in the productivity of the agricultural sector.3 This increase in productivity reduces the labour requirements of agriculture and, at the same time, raises both the demand for agricultural intermediate and capital inputs, and the demand for consumer goods by those benefiting from the increase in productivity in agriculture. As a result, two processes are set in motion: one in which labour begins to be released from agriculture; the other, a process in which labour is progressively absorbed into other sectors of the economy – initially by those activities whose products benefited from higher demand from agriculture and later by the more general dynamics of economic growth. During this new phase, generally called the “industrialisation” phase, labour is absorbed mainly by manufacturing and services. During the next phase, alongside a continuing contraction of employment in agriculture and an expansion of employment in services, comes a tendency for the share of manufacturing employment in total employment to stabilize. Finally, a new phase emerges in which employment in manufacturing begins to fall (first in relative terms and then, at least in some countries, in absolute terms); in the meantime, services continue to be the main source of labour absorption.4 This latter phase is commonly referred to as the “de-industrialisation” phase.5

Most industrialised countries reached this phase of de-industrialisation around the end of the 1960s and the beginning of the 1970s, while some high-income DCs (such as the rapidly industrialising countries of East Asia) began this phase in the 1980s. However, at about the same time (and for a number of different reasons which will be discussed in more detail below), some Latin American countries also began to ‘de-industrialise’ rapidly, despite the fact that their level of income per capita was far lower than that of other countries which had either began to ‘de-industrialise’ earlier, or that were beginning to ‘deindustrialise’ at the same time. Among industrialised

3 In some historical experiences, like that of Great Britain, this process starts with what is usually called the “agrarian revolution”. 4 Often, more labour is released by on sector than is able to be absorbed by others, leading to problems such as the growth of unemployment, informality and so on. 5 Throughout this paper, “de-industrialisation” will be analysed solely from the point of view of manufacturing employment.



6

countries, one group of countries where the size and speed of change were most remarkable was that of the European Union (EU); here, manufacturing employment was reduced by almost a third in the last three decades of the twentieth century.6

Figure 1 – European Union: manufacturing and services, 1960-2000

mf = manufacturing and ser = services Source: Adapted from Rowthorn (1997)

Figure 1 shows how, in the years between 1960 and the early 1970s, both output and productivity in the European Union’s manufacturing sector were growing at similarly high speeds (with annual average rates of 5.9% and 5.3% in 1960-1973, respectively).

6 In the United Kingdom, the fall was even more dramatic, dropping by half in the same period (see Figure 10 below).


7

In the post-1973 period both these rates fell drastically; however, the production rate dropped far more rapidly, ending up at just half that of productivity (1.4% and 2.8%, respectively). Therefore, the origins of the rapid fall in manufacturing employment in the EU can be explained easily – at least in arithmetic terms. As the growth in employment is equal to the growth in output less that of productivity, the result of the above asymmetry is a decline in employment. From this (arithmetic) standpoint, what happened in the European Union after 1973 should be considered as basically a case of “output-led” de-industrialisation – i.e., the result of a rapid slowdown of the rate of growth of output – rather than a case of “productivity-led” (or a ‘new-technological-paradigm-led’) de-industrialisation.7

Figure 1 also shows that in the services sector the reverse phenomenon took place: although in this sector the rate of growth of both output and productivity also fell drastically from the early 1970s, growth in productivity fell much faster than that of production – with productivity growing at less than half the rate of growth of output (1.1% and 2.6%, respectively). This reversed asymmetry resulted in a rapid increase in employment. Consequently, after the early 1970s, both sectors – manufacturing and services – demonstrated significant contrasts in their capacity to absorb labour.

As a result, between 1973 and 2000 the EU was able to raise their production of manufactures by about 50%, while cutting employment by almost 30%. At the same time, in order to continue raising their output of services (even though this was done at a slow pace – and certainly at a much slower pace than in the previous period), the EU had to increase employment in this sector by more than 50%. The literature on the subject has developed various hypotheses to explain the fall in manufacturing employment in industrialised countries since the late 1960s. The four better-known hypotheses are the following:

i. The fall is no more than a ‘statistical illusion’ (caused mainly by the re-allocation of labour from manufacturing to services following a rapid increase in the number of activities being contracted out by manufacturing firms to specialist service producers – such as transport, cleaning, design, security, catering, recruitment and data processing);

ii. The fall is the result of a significant reduction in the income elasticity of demand for manufactures;

iii. The fall is the consequence of a rapid growth in productivity in (at least some sectors of) manufacturing, a product of the propagation of the new technological

7 Of course, in order to understand this process not just simply in arithmetic terms but in its proper macro- and micro-framework one would have to go more deeply into the causal relationships between the trends in output, employment, and productivity growth. Furthermore, at the risk of stating the obvious, surely the issue is not whether the de-industrialisation was due to rising productivity growth or to falling output growth; but more critically to focus on the actual relationship between productivity growth and output growth. Unfortunately, the study of this relationship falls outside the scope of this paper.



8

paradigm of micro-electronics – it would have been a case of the new technology tending to produce ‘jobless growth’; and

iv. The fall is the result of a new international division of labour (including ‘outsourcing’), which is detrimental to manufacturing employment in industrialised countries, especially in regards to its non-skilled labour.8

The main aim of this paper is to study the trajectory of manufacturing employment in the post-war period – in particular, the ‘inverted-U’ phenomenon of the process of economic development in which an increase in per capita income is accompanied by the percentage of employment in manufacturing first rising, then stabilising, and finally falling. For this, I shall use a sample of 105 countries over the period 1970-1998 – the same 105 countries throughout the period – and (due to lack of data) a sample of only 81 in 1960 (see endnote 1); the source of data were the ILO Databank for manufacturing employment and the ‘Pen Tables’ for income per capita.9 Table 1 summarises the employment data by region.

8 For a detailed analysis and critical discussion of (and bibliographical references to) these hypotheses, see Rowthorn (1997) and (1999), and Rowthorn & Ramaswamy (1999). 9 In addition, ECLAC’s Statistical Division provided data for Latin America’s manufacturing employment during the 1960s.


9

Table 1 – Employment in manufacturing (% of total)

1960 1970 1980 1990 1998

Sub-Saharan Africa 4.4 4.8 6.2 5.5 5.5 Latin America 15.4 16.3 16.5 16.8 11.8 Southern Cone and Brazil 17.4 17.2 16.2 16.6 11.8

West Asia and North Africa 7.9 10.7 12.9 15.1 15.3 South Asia 8.7 9.2 10.7 13.0 13.9 East Asia (w/o China and Japan) 10.0 10.4 15.8 16.6 14.9 NICs 1 10.5 12.9 18.5 21.0 16.1 China 10.9 11.5 10.3 13.5 12.3 Third World 10.2 10.8 11.5 13.6 12.5 First World 26.5 26.8 24.1 20.1 17.3

Source: Calculations made using the ILO Databank. Regional averages are weighted by economically active population.

Countries included in Thi rd World: Sub-Saharan Af ric a: Benin, Botswana, Burkina Faso, Cameroon, Central African Rep., Chad, Congo Dem. Rep., Congo Rep., Côte d'Ivoire, Gabon, Ghana, Kenya, Lesotho, Malawi, Mali, Mauritania, Mauritius, Niger, Nigeria, Rwanda, Senegal, South Africa, Togo, Zambia and Zimbabwe. La tin Americ a: Argentina, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Jamaica, Mexico, Nicaragua, Panama, Paraguay, Peru and Uruguay (of which, Southern Cone: Argentina, Chile and Uruguay). West As ia & Nor th Af ric a: Algeria, Egypt, Morocco, Oman, Saudi Arabia, Tunisia, and Turkey. South Asia: Bangladesh, India, Pakistan and Sri Lanka. East As ia: Hong Kong, Indonesia, Malaysia, Philippines, Singapore, South Korea, Thailand and Taiwan (of which, NICs 1: Hong Kong, South Korea, Singapore and Taiwan). Coun tr ies inc luded in Fi rs t World: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Greece, Italy, Japan, Luxembourg, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, United Kingdom and United States.

This paper illustrates that these data provide significant confirmation of this ‘inverted-U’-type of trajectory of manufacturing employment with respect to income per capita. However, it will also show that this relationship has characteristics that are far more complex than has so far been recognised.



10

2. The four sources of ‘de-industrialisation’

2.1 The first source of ‘de-industrialisation’: an ‘inverted-U’ relationship between manufacturing employment and income per capita

The point of departure for this work is the ‘inverted-U’ developed by Rowthorn (1994) who, having discussed and criticised in detail the above-mentioned hypotheses (in particular the second and third ones), defined de-industrialisation as the drop in manufacturing employment that takes place from when countries reach a certain level of per capita income – this level in his cross-section regression for 1990 (built from a sample of 70 countries) being approximately US$12,000 in 1991 international dollars (see Figure 2).

Figure 2 – Rowthorn’s regression: manufacturing employment and income p.c., 1990

Although the analysis of our sample confirms Rowthorn’s hypothesis, there are also precedents for arguing that the process of de-industrialisation is a rather more complex phenomenon. In particular, this study will establish that, in addition to the process identified by Rowthorn – from now on called the ‘first source’ of de-industrialisation – there are three further processes at work. De-industrialisation, therefore, is not simply the result of a single process (the existence of a stable ‘inverted-U’ relationship between manufacturing employment and income per capita), but a consequence of the interaction of four distinct phenomena.


11

2.2 The second source of ‘de-industrialisation’: a declining relationship between income per capita and manufacturing employment

The first phenomenon to note is that Rowthorn’s ‘inverted-U’ relationship is not stable over time, but shows a continuous fall for medium- and high-income countries (see Figure 3).

Figure 3 – Second source of ‘de-industrialisation’: a declining relationship, 1960-98

1960 = cross-section regression for 1960, using a sample of 81 countries; 1970, 1980, 1990 and 1998 = cross-section regressions for respective dates, using a sample of the same 105 countries. For the specification of the regressions, see endnote 2. The regression for 1998 corresponds to employment for that year, but income per capita to 1992 – the last year for which the ‘Pen Tables’ so far offers information. Unfortunately, there were too few countries with reliable employment information for 1950 to construct a comparable regression. In the five regressions, all parameters are significant at the 1% level and the adjusted R2 are between 57% and 72% (see endnote 2). All regressions also pass the tests for homoscedasticity and for the normality of residuals at the 5% level of significance.

Sources: ILO Databank for manufacturing employment; and Summers and Heston ‘Pen Tables’ for income per capita. Unless otherwise stated, these are the sources for all graphs in this paper.



12

This continuous decline over time in the relationship between manufacturing employment and income per capita is the second source of de-industrialisation identified in this work. In essence, for middle- and high-income countries, whether or not they had reached the turning point of the regression, there was a declining level of manufacturing employment associated with each level of income per capita. The reasons for this steady decline, in particular the big drop during the 1980s for industrialised countries, still need to be properly understood. Evidence so far indicates that it is the result of a combination of factors, including, at least in part, three of the hypotheses mentioned above: the ‘statistical illusion’ hypothesis; (in an indirect way) the propagation of the new technological paradigm (micro-electronics)10; and the increasingly important process of breaking-down of the ‘value chain’ by multi-product Transnational Corporations (TNCs), leading to the relocation to DCs of the labour-intensive assembly-end part of the value chain. However, at least of equal (possibly of more) importance are the consequences of the new politics and economics of the 1980s – especially the sharp slowdown of economic growth that followed those policies – and the massive institutional and financial transformations that characterised the world economy in this period.

Although a detailed analysis of the role that each of these factors played in de-industrialisation during this period is outside the scope of this paper, it is important at least to emphasise that the existing literature does not pay sufficient attention to the latter phenomenon: the role that the 1980s’ switch in ‘policy regime’ in OECD countries had on manufacturing employment – from post-war Keynesianism (broadly speaking) to monetarism. As is clear from Figure 3, the combined effect that ‘neo-liberal’ politics and (demand-constraining) monetarist economics had on manufacturing employment was devastating, in particular the stagflation that followed the (barking-up-the-wrong-tree) monetarist response to the second oil shock.

Chicago’s (repackaged) monetarist policies ‘came of age’ (again) in their new incarnation with the Swedish Central Bank giving their economic prize in honour of Alfred Nobel to Milton Friedman in the mid-1970s. Soon afterwards, their big break came with the failure of the Carter administration’s attempt at stimulate aggregate demand in the US between 1977-1978; these expansionary policies resulted in the US trade deficits reaching one-quarter of exports just at a time when the rest of the OECD was no longer willing to absorb ‘excess’ dollars. The resulting dollar weakness, in tandem with rising inflation, set the stage for the FED radical monetarist era following the appointment of Paul Volker and his trebling of interest rates between 1979 and 1981. In turn, the election of Mrs. Thatcher in 1979 was the icing on the cake for the new neo-liberal-cum-monetarist policies. Her election also marked the arrival of the new policies in Europe, and the emergence of the ‘Chicago-Boys’ in Pinochet’s Chile their (not very democratic) appearance in the Third World.

In the EU there was an immediate collapse of growth rates: during Mrs. Thatcher’s first two years in office (1979-81), for example, the growth-rate of output in the UK fell from 3.7% to -2.2% (in fact, output declined by no less than 17% in just 6

10 See especially Perez (2002).


13

quarters); in Germany it dropped from 4.2% (1979) to –0.6% (1982); in France from 3.2% (1979) to 0.8% (1983); and in Italy from 6% (1979) to 0.3% (1983). This sharp slowdown had a major impact on unemployment – in the UK, it grew from 5% (1979) to 12.4% (1982); in Germany from 3.2% (1979) to 8% (1983); and in France from 5.9% (1979) to 10.2% (1985).

Manufacturing was particularly badly hit by these events. In the UK, for example, net manufacturing investment became highly negative in 1981 and 1982. The combination of these factors – massive reductions in private consumption (this fell from an annualised growth rate of 8.5% in the second quarter of 1979 to –3.7% in the same quarter of the following year), high interest rates, low levels of investment and a sharp revaluation of the pound – meant that manufacturing employment fell by 20% in the first three years of Mrs. Thatcher’s reign alone (without experiencing any major recovery thereafter). Therefore, there is little doubt that the remarkable decline in the relationship between income per capita and manufacturing employment in industrialised countries during the 1980s (clearly evident in Figure 3 above) had as much to do with ‘policy’ as with other factors, such as the need for industrial restructuring, technological change, mere accounting issues, or the trend towards “financialisation”.11

2.3 The third source of ‘de-industrialisation’: a declining income per capita corresponding to the turning-point of the regression

A declining income per capita corresponding to the turning-point of the regression The third source of de-industrialisation identified in this work concerns the huge drop in the turning-point of the regressions that relate manufacturing employment to per capita income since 1980. As Figure 4 shows, since the beginning of the 1980s, there has been a dramatic reduction in the level of per capita income from which the downturn in manufacturing employment began: from $ 20,645 in 1980, to just $ 9,805 in 1990 (and $8,691 in 1998; all figures in 1985 international US dollars).

11 “Financialisation”, inter alia, includes the rise in size and dominance of the financial sector relative to the non-financial sector, as well as the diversification towards financial activities in non-financial corporations.



14

Figure 4 – Third source of ‘de-industrialisation’: a changing turning-point in regression, 1960-1998

60, 70, 80, 90 and 98 = regressions for 1960, 1970, 1980, 1990 and 1998. This notation will continue in all graphs below.

This rapid lowering of the turning-point of the regressions since 1980 is crucial to understanding one of the sources of the process that leads to de-industrialisation. Until that date (1980), no country – not even the United States, the country with the highest per capita income in the sample – had reached a level of income per capita anywhere near to that at which the curves begin to fall (see Figure 5). In 1990, by contrast, there are more than 30 countries whose per capita incomes were above that critical point in the curve.12

12 As Rowthorn (1994) used data for 1990 for his regression (see Figure 2), this particular effect was already clearly in operation.


15

Figure 5 – Turning-point in the five regressions and United States income p.c., 1960-98

TP = level of income per capita at which the turning-point in each regression is found; and US = United States' income per capita.

However, as is well known, the drop in manufacturing employment in industrialised countries began in the late 1960s – i.e., well before any industrialised country was anywhere near the turning-point in the curve. This would suggest that the original impulse for de-industrialisation was not the fact that some countries had already reached the level at which the curve begins to slope downwards but the constant fall over time of the actual inverted-U relationship for middle and high-income countries (shown in Figure 3). It would not be until the 1980s that the de-industrialisation phenomenon would include the latter additional element as well (leading to an acceleration of this process). Working from a different perspective, this lowering of the turning-point of the regressions had been predicted by Rowthorn and Wells (1987; pp. 329-332); according to them, as productivity catch-up is fastest in manufacturing, in developing countries de-industrialisation was probably going to start at a lower level of income per capita than in early industrialisers. Nevertheless, nobody could have predicted that the drop in the turning point of the regression could be of such a magnitude.

2.4 The fourth source of ‘de-industrialisation’: the ‘Dutch Disease’

Finally, in addition to those three sources of de-industrialisation already mentioned, in several countries there is a further one: the so-called ‘Dutch Disease’-effect. Some countries, such as the Netherlands and a group of Latin American countries, had a fall



16

in their manufacturing employment that was clearly greater than would have been expected following the three sources of de-industrialisation discussed above (see Figure 6).

Figure 6 – Fourth source of ‘de-industrialisation’: cases of ‘overshooting?’

Neth = The Netherlands; and LA 4 = average of Brazil, Argentina, Chile and Uruguay.

However, the analysis below will show that rather than simple cases of ‘overshooting’, the ‘Dutch Disease’ is a specific additional degree of de-industrialisation characteristic of some countries that have followed at least one of three brands of transformations.

The origin of this ‘disease’ is found in the fact that the relationship between manufacturing employment and per capita income tends to be different in countries that are following an ‘industrialisation agenda’ that seeks to generate a trade surplus in manufacturing and in countries rich in natural resources able to generate a trade surplus in primary commodities that can finance their trade deficits in manufacturing. However, in reality, as will be shown below, what is called here the ‘primary commodity effect’ is a more general phenomenon that also applies to countries that are able to generate a trade surplus in services, especially tourism and finance.

It is important to stress from the beginning that the first category of countries (the ‘manufacturing’ one) includes some that are there out of necessity and others out of growth policy. That is, some countries are there because they have no option but to aim at a trade surplus in manufacturing (in order to be able to pay for their trade deficits in primary commodities and services); others are there because, even though they are able to generate a trade surplus in primary commodities or services, they still


17

try to implement an ‘industrialization agenda’ that aims to generate a trade surplus in manufacturing.

Following the Akaike regression-specification criterion, in this paper the two types of countries (‘manufacturing’ and ‘primary commodity’) are differentiated by an intercept dummy in the above regressions. Countries are classified according to their position at the end of the period – and once classified, they stay in the same group in all regressions (to avoid circular-type arguments). Furthermore, as above, the cross-section regressions for 1970, 1980, 1990 and 1998 are based on the same sample of 105 countries (due to a lack of data, however, the regression for 1960 is based on a sample of only 81 countries; all of them are later included in the larger sample). The intercept dummy shows that the relationship between manufacturing employment and income per capita in the two groups of countries is located at different levels in all five regressions (1960, 1970, 1980, 1990 and 1998); this dummy is significant at the 1% level in all regressions (1.4% in 1998).13 Figure 7 shows this phenomenon for 1998 (superimposing a group of ‘manufacturing’ and primary commodity and export service countries to their respective regressions.)

13 In the five regressions, all parameters are significant at the 2% level of significance (or less), and the adjusted R2 are between 67% and 77% (see endnote 2). All regressions also pass the tests for homoscedasticity and for the normality of residuals at the 5% level of significance.



18

Figure 7 – The ‘primary commodity’ and ‘export-services’ effects, 1998

98 mf = regression corresponding to countries that aim at generating a trade surplus in manufacturing; 98 pc = regression of countries that have a trade surplus in primary commodities or services (like tourism or finance). The same notation will be used in all graphs below. See endnote 1 for the classification of countries in the sample. As mentioned above, the difference between both types of regressions is an intercept-dummy.

Countries shown around the ‘98 mf ’ regression are: Pk = Pakistan; Cn = China; Eg = Egypt; SL = Sri Lanka; Tun = Tunisia; Ko r = Korea; Si = Singapore; EU (5) = five continental countries of the EU (Germany, France, Italy, Belgium and Austria); and Jp = Japan. Countries shown around the ‘98 pc ’ regression are: Ni = Nigeria; Zi = Zimbabwe; PG = Papua New Guinea; Bo = Botswana; Ja = Jamaica; PA = Panama; Br = Brazil; Cl = Chile; Ar = Argentina; U = Uruguay; Ve = Venezuela; Ma = Malta; Gr = Greece; Cy = Cyprus; UK = United Kingdom; Ne = Netherlands; Aus = Australia; No = Norway; and Ca = Canada.

Obviously, the main reasons for the different degrees of industrialization of these two groups of countries are their differences in resource endowment and in growth policy; these ends up being reflected in their patterns of international trade and in their internal economic – and employment – structures. However, as will be shown below, although this primary-commodity (or export-services) effect is a necessary condition for developing Dutch Disease, it is in no way a sufficient one.

Figure 8 shows a remarkable similarity in the decline of the relationship between manufacturing employment and per capita income in both categories of country between 1960 and 1998.


19

Figure 8 – Changes in manufacturing employment and income p.c., 1960 and 19986

One aspect of these relationships must be emphasized: although the “primary-commodity” group of countries do tend to reach a lower level of industrialization at any given point in time,14 the “pc” effect does not lead per se to a greater relative level of de-industrialization in time. In fact, the above graph illustrates that both categories of countries experience a similar drop in relative terms. Taking the highest point of the curves for each group of countries, it is noticeable that in these four decades both categories of countries lost about the same share of manufacturing employment – from 39% to 21% in “manufacturing” countries and from 29% to 16% in the others (i.e., in the former the 1998 inflection point of the regression is located at a share of manufacturing employment that is 45% below that of 1960, while in the latter the respective point is 46% below). In turn, from the point of view of the horizontal axis (income per capita), the turning-points of both curves also fell to half their former levels: from about US$ 18,000 in 1960 to US$ 9,000 in 1998 (both figures in 1985 international dollars).15

After this (of necessity, lengthy) introduction on the differences between these two types of country, it is now possible properly to explain my concept of ‘Dutch Disease’. From the point of view of the methodology developed in this work, what is

14 From now on, the whole group of primary commodity and export services countries will be referred to simply as the “primary commodity” group (“pc”). 15 The drop in the level of income per capita at which the inflection points are located is not surprising, as the difference in the regressions is just the dummy in the intercept. As indicated before, in 1960 no country had reached the income per capita associated with the turning-point of both regressions.



20

most interesting about this ‘disease’ is that in the light of our scrutiny, it acquires a quite different connotation to that with which it has been associated hitherto. This allows us to develop a new, more specific way of looking at it. There is a group of countries – both industrialised and developing (although the latter only when they have at least a middle-income level) – that exhibits a specific additional de-industrialisation phenomenon (additional to the three de-industrialisation forces already discussed, that is), which is associated either with a sudden surge in the exporting of primary commodities or services (particularly in countries which had not previously developed these sectors), or – as in the Southern Cone of Latin America – with a sudden shift in economic policy. The Netherlands rightly gives its name to this phenomenon (see Figure 9).16

Figure 9 – The Netherlands: unravelling the ‘Dutch Disease’? 1960-998

From this perspective, the Dutch Disease is a process in which the discovery of a natural resource (natural gas in the case of the Netherlands) causes a country to switch from one group of reference to the other, i.e., from the group of countries that aim at generating a trade surplus in manufacturing (type-”mf” regressions) to the group that is able to generate a trade surplus in primary commodities (type-”pc” regressions). When this occurs, as figure 8.A shows for the case of the Netherlands, the country experiencing this “disease” goes through two different paths of de-industrialization: the first, which is common to those countries in its original group (from “60 mf” to

16 On this point, I disagree with my friend, the Dutch economist Willem Buiter, who always remarks that the so-called Dutch Disease is a ‘slander’ on the Dutch economy! The literature on the general macro-mechanisms at work in the 'Dutch Disease' is extensive; see, for example, Rowthorn and Wells (1987), Ros (2000), and Pieper (2000).


21

“98 mf”), consists of the three processes discussed above; and, in addition to this “common” one, a second surge of de-industrialization resulting from the change in the reference group – (from “98 mf” to “98 pc”). In this context, the Dutch Disease should only be regarded as the “excess” degree of de-industrialization associated with the latter movement; i.e., only with the difference between employment in manufacturing falling to “98 mf” and falling to “98 pc” – in the case of the Netherlands, then, between having fallen from 30.5% in 1960 to 19.8% in 1998 (in this hypothetical no-Dutch Disease scenario this is the predicted level of manufacturing employment in the “mf” regression given the actual 1998 level of income per capita), and from 30.5% to 14.8%, respectively (actual Dutch Disease situation).17

When thus perceived, it becomes clear that the ‘Dutch Disease’ was not a phenomenon limited to Holland, but also occurred in other industrialised countries such as the United Kingdom, where there was both a significant discovery of natural resources (North Sea oil), and an increase in the trade surplus of financial-services exports (see Figure 10).

17 An alterative form of measuring the Dutch Disease along these lines would be to have a growth model for The Netherlands that somehow included the effect of the discovery of natural resources as an explanatory variable, and predict from this regression what would have been the expected level of income per capita in 1998 had this country not discover natural gas. In this case, the Dutch Disease effect could be measured as the difference in manufacturing employment in 1998 between its actual level and its predicted level in the “mf” regression given the hypothetical income per capita of a non-Dutch Disease scenario.



22

Figure 10 – The United Kingdom: catching the ‘Dutch Disease’? 1960-98

Ne = The Netherlands; and UK = United Kingdom.

In the UK, the improvement in the trade balance in oil between 1979 and 1984 (from -£ 2.2 billion to £ 6.6 billion) in fact mirrored the decline in the trade surplus in manufactures (from £ 3.6 billion to -£ 6.3 billion, respectively). Hardly surprisingly, the UK’s economic and employment structure switched from one ‘category’ of country (and one ‘industrialisation agenda’) to the other.

A comparison between the degree of de-industrialisation of Holland and other continental EU countries can also help to explain the special nature of the ‘Dutch Disease’.


23

Figure 11 – The Netherlands and five countries of the European Union, 1960-98

Ne = The Netherlands; EU 5 = Germany, France, Italy, Belgium and Austria.

Figure 11 shows how, while the share of manufacturing employment in the five EU countries falls according to the ‘aiming-at-a-trade-surplus-in-manufacturing’-type regression (from ‘60 mf’ to ‘98 mf’), Holland’s suffers an added fall (from ‘98 mf’ to ‘98 pc’; as already mentioned, equivalent to five percentage points in this case).

As is clear from the graph, the major fall in manufacturing employment in Holland took place in the first half of this period (following the discovery of natural gas), while in the other countries manufacturing employment began to fall significantly only after 1980. As mentioned above, the latter drop was greatly influenced by a shift in the political and economic ‘policy regime’. The rapid decline in domestic demand (that resulted from high interest rates, the shift in income distribution towards capital, the breaking up of the trade union power, etc.) meant that productivity growth had much higher negative effects on manufacturing employment than it would have had otherwise. In the case of Germany, for example, although its machine-tool industry did not suffer significantly (mainly due to the fact that the relevant markets were characterised by a combination of high income with low price elasticity of demand), other segments of its manufacturing industry – such as steel – were badly hit by having to face rapid technological change in the context of demand-constrained domestic markets.18

18 See especially Solow (1997).



24

At the same time, Figures 10 and 11 indicates the remarkable difference between the UK and ‘EU 5’ (Germany, France, Italy, Belgium and Austria) during the 1980s – in the former, the fall in the share of manufacturing employment amounted to 9.2 percentage points, in the latter to 3.3 points. In the UK, neo-liberal politics and (demand-constraining) monetarist economics not only affected manufacturing via their effect on domestic demand (as in ‘EU 5’), but the discovery of oil, the strengthening of the financial-services export sector (greatly supported and favoured by Mrs. Thatcher) and the higher degree of fundamentalism of the UK government’s neo-liberal-cum-monetarist policies all meant that manufacturing in the UK had the added problem of the loss of export markets (mainly but certainly not exclusively) via the revaluation of the pound. As mentioned above, between 1979 (the year of the election of Mrs. Thatcher and the beginning of oil exports) and 1984 the UK’s trade balance in manufactures switch from a surplus of £ 3.6 billion to a deficit of £ 6.3 billion. In ‘EU 5’ (and, in particular, in Germany), instead, success in export markets was crucial in mitigating their processes of de-industrialisation.

A further comparison between Holland and four other industrialised countries (in this case, countries that have been major primary commodity exporters throughout the period – United States, Canada, Australia and New Zealand) will also help to show the specific nature of the ‘Dutch Disease’ (as understood in this paper).

Figure 12 – Netherlands and ‘UCAN’: two processes of de-industrialisation but only one ‘Dutch Disease’

Ne = The Netherlands; and UCAN = United States, Canada, Australia and New Zealand.

Figure 12 shows that, although these four ‘UCAN’ countries also ended up in the ‘primary commodity’ relationship in 1998 (‘98 pc’), they did not suffer from the ‘Dutch Disease’ – as understood in this study – (as Holland did) because they had also set out in that type of relationship back in 1960 (‘60 pc’). Consequently, although


25

both the “EU-5” and the “UCAN” countries did experience a large (and similar, in relatively terms) drop in the share of total employment accounted for by manufacturing employment during this period (9.2 and 10.5 percentage points, respectively), they each began and ended this four-decade period in the same reference group that they had started with; The Netherlands, instead, switched from one to the other as a result of the Dutch Disease (with an overall manufacturing employment loss of no less than 15.1 percentage points).

Moreover, as already indicated, the phenomenon of the ‘Dutch Disease’ was not limited to those industrialised countries that discovered natural resources, but also occurred in countries that developed important service-exporting sectors, such as tourism (e.g., Greece, Cyprus and Malta; see Figure 13) and financial services (e.g., Switzerland, Luxembourg and Hong Kong; see Figure 14).

Figure 13 – Greece, Cyprus and Malta: a ‘tourism-based Dutch Disease’?

G = Greece; M = Malta; and C = Cyprus (for the clarity of the graph, the trajectory of Cyprus is only shown from 1980; at that point in time, this country had a level of manufacturing employment corresponding to the ‘80 mf ’ regression).



26

Figure 14 – Switzerland, Luxembourg & Hong Kong: a ‘financial-services-based Dutch Disease’?

HK = Hong Kong; L = Luxembourg; and S = Switzerland.

Countries in Figures 13 and 14 also began this period in a ‘manufacturing’ regression (or even to the left of it, e.g. Malta and Hong Kong) but they all ended it in (or very close to) the ‘98 pc’ regression (rather than in their corresponding ‘98 mf’ one).

Finally, this “disease” also spread to some Latin American countries; but the key issue in this case is that it was not brought about by the discovery of natural resources or the development of a service-exporting sector but mainly because of a drastic switch in their economic policy regime. Basically, it was the result of a drastic process of trade and financial liberalisation in the context of a radical process of institutional change, leading to a sharp reversal of their previous (Stateled) import-substituting industrialisation agenda (ISI). Despite their well-known abundance of natural resources, ISI had brought many countries in the region to a level of industrialisation characteristic of the ‘mf’ group (see Figure 15). This shift in ‘policy regime’, although in many ways similar to that of most industrialised countries during the 1980s, hit their level of manufacturing employment more drastically because it brought their process of industrialisation down from their policy-induced ‘mf’ heights to a ‘Ricardian’ resource-rich ‘pc’ level. Brazil and the three Southern Cone countries (Argentina, Chile and Uruguay) were the Latin American countries that experienced the highest levels of de-industrialisation following their economic reforms, while also being among the countries of the region that had previously industrialised the most and that had implemented such reforms most rapidly and drastically (see Figure 15).


27

Figure 15 – Argentina, Brazil and Chile: Catching the ‘Dutch Disease’? 1960-98

Br = Brazil; Cl = Chile; Ar = Argentina; and Ne = The Netherlands. For the clarity of the graph, Uruguay is not included (having a similar trajectory to that of the other three Latin American counties, located between that of Chile and Argentina).

These four Latin American countries began this period – as did Holland – with a level of manufacturing employment typical of countries aiming at a trade surplus in manufacturing (‘60 mf’), although obviously for different reasons. For Holland, it was a matter of the most likely ‘Ricardian’ position given its resource endowments and level of income per capita. For the four Latin American countries, in contrast, it was the deliberate result of a ‘structuralist’ industrialisation agenda – that is, intensive State-led ISI, to which they had moved hoping to improve their chances of ‘catching up’ with industrialised countries by promoting specialisation in products with higher productivity growth potentials that were more likely to move them up the technology ‘ladder’.19 Moreover, both Holland and these four Latin American countries reached 1998 with a level of manufacturing employment corresponding to the other category of countries (‘98 pc’). Again the reasons vary. In Holland, it was because of the effects of the discovery of natural resources in a ‘mature’ manufacturing economy, while in the four Latin American countries it was generated by a sharp reversal of the ISI policies. The end of industrial and trade policies, changes in relative prices, in real

19 Although Latin America’s ‘structuralist’ industrialisation agenda did not succeed (as the one in East Asia did) in creating a trade surplus in manufacturing, ISI in Latin America at least did lead to a significant reduction in the trade deficit in manufactures (as well as in the trade surplus in primary commodities); see Palma (2003).



28

exchange rates, in the institutional framework of the economies, in the structure of property rights and in market incentives in general led them back to their ‘natural Ricardian position’; i.e., one more in accordance with their traditional resource endowment.20

From this point of view, the major difference between Latin American and continental Europe in terms of the consequences of neo-liberal-politics-cummonetarist-economics is that in continental Europe the crucial transformation took place in industrial relations, the welfare State, public corporations, etc., while in Latin American, since they were hit by the new policies at a much lower level of income per capita, neo-liberal-politics-cum-monetarist-economics also obstructed their transition towards a more mature – i.e., self-sustaining (in a Kaldorian sense) – form of industrialisation.21

In terms of their future prospects for manufacturing employment, the situation in the four Latin American countries that have already experienced the ‘Dutch Disease’ seems to be fairly straightforward, as Figure 16 shows.

20 It should be emphasized that in these countries it was policy rather than a surge in primary-commodity exports that brought about their Dutch Disease. In fact, in them these exports only surged after their respective post-neo-liberal-economic-reform financial crises (and subsequent devaluations); therefore, by 1998 only in Chile there had been a rapid expansion of unprocessed primary commodities exports (adding in this case an additional impetus to its de-industrialisation.) 21 See, in particular, Kaldor (1967).


29

Figure 16 – Brazil and the Southern Cone Countries of Latin America, 1998

B = Brazil; A = Argentina; Cl = Chile; and U = Uruguay.

With the drastic switch of economic policies, these four countries have not only returned to their expected ‘Ricardian’ positions, but their middle-income levels of income per capita have already got them almost as far as the levelled-out part of the curve (Argentina, Chile and Uruguay in particular). Furthermore, it is also important to remember that this curve has been going down continuously since the 1960s and there is no reason to suspect that this downward tendency has run its full course – the regressions and their turning-points could easily keep getting lower. Therefore – and with the necessary strong caveats regarding the use of this type of cross-section regression for the purpose of prediction, in particular due to the ‘homogeneity conditions’ required for this22 –, in the absence of any new and imaginative industrial and trade policies (that would rescue their pro-industrialisation agenda), the framework developed here suggests that, at their absolute best, these countries would be able to maintain their present relative levels of manufacturing employment. More likely, the decline will continue, even if their income per capita increases rapidly – a prospect that at the time of writing seems (at best) rather unlikely, at least in the immediate future.23

22 See, for example, Pesaran et all (2000).

23 The signature of the new bilateral ‘free trade’ treaty with the US, such as that recently signed by Chile, is unlikely to help the industrialisation task. In fact, as is well-known, other than for agricultural products, the treaty between Chile and the US had



30

In sum, in this paper the ‘Dutch Disease’ is perceived to be not just a simple ‘overshooting’ of de-industrialisation but a specific type of ‘excess’ – which is associated with the movement from a process of de-industrialisation typical of countries following an ‘industrialisation agenda’ aiming at generating a trade surplus in manufacturing, to a process of de-industrialisation typical of countries generating a trade surplus in primary commodities or services. In general, the shift between the two types of de-industrialisation has taken place for one of three different reasons: the first is due to the discovery of natural resources (e.g., Holland); the second to the development of export-service activities, mainly tourism and finance (e.g. Greece in the former, and Hong Kong in the latter); and finally, the third is due to changes in economic policy, which brought countries that were above their ‘natural’ Ricardian position back to their traditional (static) comparative advantage position (e.g., Chile, Brazil and Argentina).

Therefore, in analytical terms, (policy-induced) Dutch Disease in Latin America should be understood more as a case of “downward” de-industrialisation than in the other Dutch Disease countries discussed above, where it was the result of the emergence of other productive activities. In turn, all Dutch Disease cases should be distinguished from more “normal” processes of de-industrialisation, as those that have taken place in many industrialised countries after they have reached the income per capita associated with the inflection point of the manufacturing employment-income per capita relationship. The latter should be understood more as processes of “upward” de-industrialisation -- i.e., mature economies switching employment from manufacturing into other activities (mainly services) in their normal process of economic development.

Finally, one should also distinguish all the above types of de-industrialisation with that found in the late 1980s and 1990s in many Sub-Saharan African economies, including South Africa, and in some countries of the former Soviet Union and Eastern Europe, which have been associated with a fall in income-per capita. As all these countries had income per capita below the turning point of the curve, a decline in income per capita was associated with a reduction in manufacturing employment along the same relationship among these two variables: a case of “reverse” de-industrialisation.

Then, in all we have at least four different types of de-industrialisation: “upward” de-industrialisation (e.g., continental Europe, Japan and traditional primary-commodity exporting industrialised countries); “normal” Dutch Disease (e.g., Holland and countries with newly-developed service-export activities); “downward” Dutch Disease (e.g., Latin America); and “reverse” de-industrialisation (e.g., Sub-Saharan Africa and countries of the former Soviet Union).

little to do with trade issues and concentrated in restricting Chile’s capacity to implement trade and industrial policies and controls in the capital account of the balance of payments; on this issue, see especially Bhagwati (2003) and Stiglitz (2003).


31

3. Trying to swim against the de-industriali-sation tide

3.1 Finland and the diversity of Nordic countries

Figure 17 shows the diversity of the process of de-industrialisation in three Nordic countries.

Figure 17 – A. Nordic Countries: three different industrialisation paths, 1960-98

Fi = Finland; Sw = Sweden; and No = Norway.

The most interesting case is that of Finland, which follows an industrialisation path that runs opposite to the ‘Dutch Disease’. It is a country rich in natural resources and in 1960 held a position that corresponded with that comparative advantage given its income per capita; i.e., a share of manufacturing employment of 21.6% of total employment – 7.3 percentage points below what would have been its expected position had it then been already in the ‘manufacturing-trade-surplus’ category of countries. However, through a greater processing of the primary commodities that it exports, and the development of sectors such as the mobile phone industry (the paradigmatic case of Nokia), it managed to move in the opposite direction to the ‘Dutch Disease’ – starting in 1960 in the ‘60 pc’ regression but finishing in 1998 spot on the ‘98 mf’ regression (with a 20% share of manufacturing employment, well above the 15.4% conditional expectation had it continued in the ‘pc’ group of countries).



32

Norway is another interesting case. In 1960, it was situated in between the‘60 mf’ and ‘60 pc’ regressions. In fact, its position in 1960 corresponds exactly to its conditional expectation in the original ‘average’ 1960 regression.24 The main reason for this was that at the time Norway had a mixed composition of exports, combining manufactures with primary commodities and services (the latter coming from its strong maritime fleet). However, the discovery of oil and further development of primary commodity exports (such as those from its flourishing fishery industry) brought Norway fully into the ‘primary-commodity-exports’ family of countries (‘98 pc’). In this context, although Norway has not suffered from a fully-fledged ‘Dutch Disease’, it represents what could be called a ‘mild’ case! Sweden, meanwhile, follows a systematic industrialisation path corresponding to the ‘manufacturing-track’ group of countries (such as the five EU countries above), from ‘60 mf’ to ‘98 mf’.

Returning to the case of Finland, when this country is compared with Chile, for example, the contrasting paths of the ‘anti-Dutch Disease’ industrialisation and the ‘Dutch Disease’ become even clearer (see Figure 18).

Figure 18 –B. Finland & Chile: an ‘anti’-Dutch disease and a Dutch disease industrialisation?

Fi = Finland; and Cl = Chile.

While Finland was able to redirect the course of its industrialisation towards a path characteristic of a country aiming at been able to generate a trade surplus in manufacturing (going from ‘60 pc’ to ‘98 mf’), Chile did the opposite by moving from

24 The ‘average’ regression was the one that did not differentiate between ‘primary commodity’ and ‘manufacturing’ countries; see Figure 3.


33

‘60 mf’ to 98 pc’. After Chile had spent many years following strong ISI-type industrial and commercial policies, which led to an industrialisation more typical of a country not rich in primary commodities, in the 1970s it abandoned its ‘pro-industrialisation agenda’ and embarked on a drastic process of trade and financial liberalisation and economic reform. These led it back to its more ‘natural Ricardian position’, a comparative advantage (which despite Chile’s rapid economic growth for about half of the period since the beginning of its economic reforms) that not only resulted in the reduction of the relative size of this country’s manufacturing sector in general, but even ended up drastically reducing the overall share of manufacturing value added in its main primary commodity export (copper) -- with the share of refined copper being drastically reduced in favour of the more primitive copper ‘concentrates’: the share of the latter in total exports of copper increased from 3% at the time of trade and financial liberalisation in the early 1970s, to 17% in 1990 and about 40% in 2002.25 Figure 19 shows how Finland instead moved in the opposite direction – one characterised by an increasingly higher degree of processing of the primary commodities it exported – thus enabling the country to manage this ‘anti-Dutch Disease’ industrialisation.

Figure 19 – C. Finland: changing vertical integration in timber-based exports

w chip = ‘wood-chip’; w mf = wood manufactures; mach p = machinery to produce ‘wood-chip’, pulp and paper. Source: Tradecan (2002)26

25 On this subject, see especially R. Tomić (1985), Caputo (1996), C. Tomić (1999), Lagos (2000), Lavanderos (2001), and The Economist (2001). 26 See also Palma (1996), and Mandeng (1998).



34

Finland and Sweden (and, as figure 20 below shows, also Malaysia, and to a lesser extent other Southeast Asian countries rich in natural resources, such as Thailand, Indonesia and the Philippines) therefore prove that from the perspective of manufacturing employment, there is no such thing as the so-called ‘curse of natural resources’.27

It is blatantly clear that those countries that export primary commodities (and services) have sufficient degrees of freedom not only to allow them to develop policies for avoiding the ‘Dutch Disease’, but also to end up having a manufacturing industry that, in terms of relative size, is more typical of countries that aim at a trade surplus in manufacturing.28

But, as the Latin American experience in particular shows, it would seem that as globalisation progresses, there are fewer and fewer countries wishing to take advantage of such degrees of freedom.29

3.2 East- and Southeast-Asian industrialisation

Starting with the second-tier Newly Industrialising Countries (NICs) of Southeast Asia, the most important feature common to all of them is that they follow a ‘mixed’ path towards industrialisation; this ‘mixed’ path keeps them near or above the original ‘average’ regressions (see Figure 3 above) -- those that did not differentiate between primary and non-primary commodity countries.

27 A concept which is very popular these days in some of the most simplistic ‘newinstitutionalist’ literature trying to explain the development failure of resource-rich countries; for a critical review of this literature, see Di John (2003). 28 For policies for avoiding the ‘Dutch Disease’, see especially Pesaran (1984); for an analysis of the specific experience of a LDC (Chile) that successfully avoided the ‘Dutch Disease’ in the late XIX Century (despite of a sudden more-than-doubling of its level of exports of primary commodities), see Palma (2000). For a comparative analysis of trade and industrial policies in East Asia and Latin America, see Palma (2004a and b). 29 As Stiglitz keeps reminding us, the ‘Washington Consensus’-type of thinking is not renowned for its ‘flexibility’ nor its imagination (see, for example, Stiglitz, 2002).


35

Figure 20 – NICs 2: Manufacturing employment and income p.c., 1950-98

98 ‘a v’ = cross section regression for the year 1998 for the whole sample. NICs 2 = Second-tier NICs; I = Indonesia; P = Philippines; T = Thailand; and M = Malaysia.

Figure 20 shows that Malaysia is another country that clearly swam against the deindustrialising tide, sharply raising the proportion of employment located in its manufacturing industry; i.e., despite being a country rich in natural resources, it moved from a position corresponding to the ‘average’ regression to one corresponding to the ‘manufacturing’ regression – in fact, by 1998, its level of manufacturing employment was even higher than that of the conditional expectation in ‘98 mf’ for a country with its per capita income. As for the other ‘NICs 2’ countries, they clearly follow the intermediate, or ‘average’ course (between ‘98 mf’ and ‘98 pc’) characterised by the original regressions (of Figure 3). In other words, despite being rich in natural resources, these three countries (Thailand, Indonesia and the Philippines) have a 1998 level of manufacturing employment higher than anticipated by a traditional ‘Ricardian’ integration into the international division of labour. However, unlike Malaysia (and Finland, and prereform Latin America), the industrialisation agenda of these countries does not seem to aim at a level of manufacturing employment similar to that of the “manufacturing” ones, but only at an intermediate level (somewhere in between the two types of countries).

In contrast, the ‘NICs 1’ countries did industrialise (and very successfully) in a way that corresponds to their condition of being ‘poor’ in primary commodities; the exception, of course, as mentioned above in Figure 14, is Hong Kong which, having developed a powerful financial service economy in the 1980s, suffered probably the worst case of ‘Dutch Disease’ of them all – its manufacturing employment collapsing



36

from 42.1% of total employment in 1980 to just 12.2% in 1998 – which led to its ending the period even lower than the ‘98 pc’ regression.

Figure 21 – NICs 1: Manufacturing Employment and Income p.c., 1970-98

NICs 1 = First-tier NICs; K = Korea; T = Taiwan; S = Singapore; and HK = Hong Kong.

An important issue that emerges from Figure 21 is that the characteristic component of the industrialisation of Korea, Singapore and Taiwan is not their level of industrialisation -- as measured by their manufacturing employment (although Taiwan in 1998 did in fact have approximately five percentage-points of manufacturing employment above the value suggested by regression ‘98 mf' for a country with its per capita income). What is really unique about their process of industrialisation is the vertiginous rise in per capita income associated with that industrialisation. In other words, the real ‘miracle’ in these countries is not found in their levels of industrialisation but in the ‘income-multipliers’ and ‘export linkages’ they were able to develop with this industrialisation. This contrasts sharply with developments in ‘maquila’ Central America and Mexico.

3.3 Mexico and Central America's ‘maquila’ (labour-intensive assembly-end part of the value chain) industrialisation

One noticeable characteristic of Latin American industrialisation in the post-war period was the homogeneity of their industrialisation agendas during the period of ISI (or State-led industrialisation). However, one noticeable characteristic now (after their


37

economic reforms) is the diversity of their respective agendas. As discussed above, their initial degree of similarity was characterised by the fact that most of them were clearly more industrialised than what countries rich in primary commodities would be expected to be. However, what were relatively similar economic reforms and processes of trade and financial liberalisation in fact led to a marked diversity in the industrialisation agenda of the region. In general, three distinct patterns of industrialisation were developed: (i) the 'Dutch Disease' path followed by Brazil and the Southern Cone countries with higher per capita income and more advanced industrialisation (Argentina, Chile and Uruguay), as illustrated in Figure 15 above; (ii) that of Mexico and the so-called ‘maquila’ Central America (Figure 22); and (iii) the Andean countries, presenting little variation with respect to their earlier patterns -- as if, by 1998, they had still not decided which path to take (Figure 25).

As can be seen in Figure 22, Mexico and ‘maquila’ Central America present a very different picture than Brazil and the Southern Cone countries do in terms of their manufacturing employment, and a remarkably different one from that of East Asian countries in terms of the ‘income-multipliers’ and ‘export-linkages’ associated with their export-led industrialisation.

Figure 22 – Mexico and Central America: the 'maquila' effect, 1970-98

H = Honduras; S = El Salvador; D = Dominican Republic; CR = Costa Rica; and Me = Mexico.

No other country in our sample shows such an steep growth in manufacturing employment as Honduras, El Salvador and the Dominican Republic; in turn, no other country shows such poor manufacturing export-income multipliers. In these countries, ‘maquila’ industrialisation for exports increased manufacturing employment very dramatically, but this is in no way associated with the same income growth as in



38

East and Southeast Asia – in fact, particularly in Honduras and El Salvador, income per capita hardly grew at all during these three decades!30 At the same time, the graph shows the contrasting patterns of growth in Mexico between the (ISI) 1970s and the (export-led growth, with a substantial ‘maquila’ component) period afterwards. In the 1970s (as in the 1950s and 1960s), for all its faults, ISI did deliver one of the fastest rates of growth of income in the Third World – although in the 1970s, at the tail-end of ISI, growth was first helped by substantial foreign borrowing and then by the discovery of large reserves of oil. Furthermore, as the graph also shows, the oil riches do seem to have some negative effect on manufacturing employment. In the period after ISI, however, trade and financial liberalisation (and NAFTA) not only led Mexico into a remarkably rapid growth of manufactured exports, but also brought manufacturing employment back to the ‘average’ regression (following a pattern of manufacturing employment similar to that of Thailand, the Philippines and Indonesia). But, because manufacturing exports had a substantial ‘maquila’ component – and the characteristics of the ‘maquila’ industry, see Figure 23 -- the ‘income-multipliers’ and the ‘export-linkages’ that this export model was able to develop were particularly disappointing.31

30 For a detailed discussion of this issue, see Palma (2002). 31 In a study of the television industry in Tijuana, for example, Carrillo (2002) clearly shows both sides of Mexico’s manufacturing export-led ‘success’. On the one hand, in 2001 Mexico produced about 30 million TV sets, 90% of which were exported to the US (representing 78% of all TV imports into the US). On the other, Carrillo’s calculations show that in value terms, 98% of inputs for the Mexican TV industry are either direct imports or ‘indirect’ ones (i.e., inputs that are supplied by foreign firms operating in Mexico, which themselves import practically all of their inputs). In fact, Mexican companies only supply the remaining 2% of inputs (mostly cardboard boxes and plastic sheets needed for packaging and the printing of manuals).


39

Figure 23 – Mexico: structure of ‘maquila’ output (%), 1990-1999

dom inp = domestic inputs; w = remuneration to blue-collar workers; s = remuneration to white-collar ones; o = operating surplus. Source: Palma (2002, using data from Mexico’s official statistical office, INEGI).

As a result, the ‘export-GDP growth’ elasticity of Mexico during its manufacturing export-led period (since the early 1980s; see Figure 24) is worlds away from those of the above-mentioned countries of Southeast Asia – let alone those of East Asia.32

32 In 2001, for example, Mexico exported the same amount of manufactured exports as Korea (about US$ 150 billion); however, its overall manufacturing sector only generated less than half the level of value added, and absorbed more than twice as much imports as Korea's. See Trade and Development Report (2002) and Palma (2002).



40

Figure 24 – Mexico: GDP and exports, 1950-2000

[X] = exports; and [GDP] = Gross Domestic Product. Exports do not include oil. The original data were first transformed into a 3-year moving average. Source: Palma (2002).

Furthermore, the future perspectives for Mexico and Central America are more complex than is usually acknowledged because the outlook for the whole Latin American ‘maquila’ industry is generally uncertain. Not only these countries have completely failed to ‘anchor’ these activities in the region via increasing the domestic content of the products, but also they are finding it particularly difficult to cope with the growing competition from other DCs. In fact, it is hard to tell whether the current already-rapid process of relocation of Mexican ‘maquila’ into China (where they precisely cease to be ‘maquila’ activities, as China produces a large proportion of inputs as well) will soon become an stampede;33 and the sign of things to come for the rest of Central America as well.34

3.4 The Andean Pact countries in South America

As for the Andean countries, by 1998 Colombia, Ecuador, Peru and Bolivia had not followed either a ‘maquila’ route or a ‘Dutch Disease’ one; see Figure 25.

33 See, for example, ECLAC (2002) and Palma (2002); see also La Jornada, 17th July, 2002. 34 See Mortimore (2000).


41

Figure 25 – The Andean Diversity in South America, 1998

B = Bolivia; P = Peru; E = Ecuador; C = Colombia; and V = Venezuela.

It can generally be said that, until the end of the 1990s, economic reforms have not had any substantial effect on their levels of manufacturing employment. Perhaps a crucial factor for this has been the capacity of these countries to continue using their regional integration (the Andean pact) as a mechanism to offset other de-industrialising forces, at least in part. But it is hard to say at the moment whether the relatively high degree of manufacturing employment in these countries is essentially a reflection of a process of economic reform that still has some way to go, or whether these countries are succeeding in finding an industrialisation path which is different from that of Brazil and the Southern Cone countries.

The case of Venezuela, instead, is perhaps an example of a country that exports primary commodities ‘too easily’. Given Venezuela’s peculiar political and economic institutional setting, its oil wealth has never been a natural incentive for its industrialisation, nor has it been in any way an incentive for growth since the oil price increase of 1973.35

35 See especially Di John (2003); see also Palma (2003). It must have taken some doing for oil-rich Venezuela to have today an income per capita well below what it had at the time of the first oil price shock three decades ago!



42

3.5 China, India and Turkey

It is important to mention in this of necessity extremely brief review of countries that have swam against the de-industrialisation tide the cases of China, India and Turkey (see Figure 26). From the point of view of this study, the most prominent characteristics of these three processes of industrialisation are: (i) in the case of China (as in the NICs-1 and as opposed to the Latin American “maquila” model), its industrialisation has been associated with a particularly rapid increase in income per capita;36 (ii) in the case of India, its industrialisation has been characterized by an especially high level of manufacturing employment;37 and (iii) in the case of Turkey, its industrialisation path has only been able to bring this country above the primary-commodity track, despite providing a large volume of manufactures for exports.

Figure 26 – China, India & Turkey: manufacturing employment and Income p.c., 1950-98

Cn = China; In = India; and Tu = Turkey.

36 For important insights on China’s economic success, see Sen (1999). 37 See, for example, Bhaduri and Nayyar (1996).


43

3.6 South Africa: still searching for an industrialisation path?

Finally, South Africa seems to be a country with difficulties in terms of ‘making up its mind’ regarding which industrialisation path to follow; see figure 27.

Figure 27 – South Africa: manufacturing employment and income p.c., 1950-98

As is well known, during the 1970s South Africa made a strong industrialisation push, but the last period of Apartheid had a disastrous effect on growth; in turn, the transition to democracy was one of uncertainty and open debate over economic policies. The graph seems to suggest that South Africa still has to decide whether to follow an ‘mf’ or a ‘pc’ industrialisation path!



44

4. ‘De-industrialisation’ – does it matter ?

One of the consequences of the process of de-industrialisation has been to reopen one of the age-old debates in economics: is a unit value added in manufacturing equal to one in primary commodities or services? In particular, is this so from the point of view of the level and sustainability of long-term growth? This debate has re-emerged because, even though it is a well-known fact that over the long-term course of economic development the structure of employment changes substantially, relative changes in scale and speed occurring since the 1960s inmost industrialised countries and in many medium- and high-income developing countries constitute a phenomenon without precedent.

Although a detailed discussion of this issue is beyond the scope of this paper, from the point of view of this debate one can basically classify growth theories into three basic camps. However, in order to do this, it is necessary to introduce first a distinction between two concepts: ‘activity’ and ‘sector’. Examples of the former are research and development (R&D) and education, and an example of the latter are manufacturing and agriculture. Taking this distinction into account, these three ‘growth camps’ are: (i) those (mainly traditional neo-classical models) that treat economic growth as a process that is both ‘activity-indifferent’ and ‘sector-indifferent’; (ii) those (mainly new growth models) that postulate instead that growth is ‘activity-specific’ but ‘sector-indifferent’; and, finally, (iii) those (mainly post Keynesian theories) that argue that economic growth is ‘activity-neutral’ but ‘sector-specific’.

In the first camp (growth as activity- and sector-indifferent), one finds the Solow-type growth models (both the traditional ones, including the Krugman-type critique of the Korean model, and the later “augmented” models), and the branch of endogenous growth theories that associates growth with increasing returns which are activity-indifferent. This would include, for example, early “AK” models (although this would depend on how one interprets them), as well as more recent endogenous growth models in which changes in the rate of growth are the result of the cumulative effect of market imperfections arising in the process of technical change. However, these imperfections, and the associated increasing returns, are somehow seen as stemming directly from within the production function (rather than being based -- as is characteristic of the new growth theories classified here as belonging in the second camp -- on any specified mechanism, such as the use of R&D or the production of human capital).38

In the second growth camp (growth as activity-specific but sector-indifferent in new endogenous growth models), one finds in particular the by now classic examples such as Romer (1990) and the neo-Schumpeterian version of Aghion and Howitt (1998). In these models, as well as some of the later endogenous growth models, increasing

38 See Barro and Sala-i-Martin (2004), and Blankenburg (2000 and 2004) for overviews ofnew growth theories.


45

returns, though generated by research-intensive activities, are explicitly not associated with the size, depth or strength of the manufacturing sector as such, or with the process of capital accumulation within the manufacturing sector, nor do they allow for specific effects from the manufacturing sector on the R&D activities.

Essentially, growth models in this second camp are similar to the general new growth-theory approach of the first one – namely, they model growth as a function of market imperfections that somehow create increasing returns in the process of technical change. However, the crucial difference between them and the more general models of the first camp is simply that they explicitly attribute the increasing returns to R&D, while more general models do not address this issue (i.e., the issue of the “growth-smoking gun”). This is why the Romer et al. models are classified here as activity-specific (e.g., ‘R&D-specific’, human-capital-specific, etc.); however, these models are explicitly not sector-specific, as they do not allow for any way in which capital accumulation in the manufacturing sector might have a feedback effect on R&D (other than via exchange between the two, but not via a Kaldorian-style argument of capital accumulation embedding or embodying technical change).

Thus, Aghion and Howitt, for example, accommodate (physical) capital accumulation by arguing that it is ‘complementary’ to innovation (i.e., knowledge production) through its effect on the profitability of research. They admit, however, that “there are many reasons for thinking that policies that favour capital accumulation will generally also stimulate innovation and raise the long-run growth rate”, adding in a footnote that “(a)nother reason, which does not however fit easily into the present framework, is that capital goods embody technologies.” (1998, p. 102; emphasis added). Although the effects of research on profitability is a very particular concept which they develop, the quote clearly indicates that “embodiment” à la Kaldor, Thirlwall and other (mainly Post Keynesian) models of the third camp cannot be accommodated in their framework. They are, therefore, different from both the first and third camps because, while paying attention to R&D (as somehow being the main source of growth), they do not link it specifically with manufacturing.

In the third camp (activity-neutral but sector-specific growth theories), the approaches to economic growth found in Kalecki, Hirschman, Kaldor, Thirlwall, Pasinetti, Prebisch and (arguably) Schumpeter stand out.39 In these sector-specific growth theories – which continue a long tradition that goes all the way back to Smith and Hume40 – there are specific capital accumulation effects on growth stemming from the

39 It could be argued that these growth theories are also activity-specific rather than just activity-neutral; however, in Kaldor’s work and in those of the other authors mentioned above (with some exceptions) the internal organisation of manufacturing – e.g., the role of R&D versus that of other (what has here been called) “activities” – is usually not central to their analysis. 40 “Husbandry […] is never more effectively encouraged than by the increase of manufactures” (Hume, 1767, Vol. III; quoted in Reinhart, 2003). Again, in the words of Governor Keith of Massachusetts (expressed it in 1738):

[…] although Spain, by possessing the Mines of Mexico and Peru, may be said to be richer in that respect than any other Nation; […] and tho’ it may furnish the Spaniards with all the Product of other Men’s Labour, which the most exquisite Luxury can desire, in the main it destroys Industry, by encouraging



46

manufacturing sector. In these models, the pattern of growth, increasing returns and the whole dynamics of economic growth are crucially dependent on the economic activities being developed (i.e., on the structure of output). The crucial difference between this camp and the previous two (but in particular between it and the first) is that in its more classical theories, issues such as the capacity to generate and diffuse technological change, productivity growth potentials, ability to move up the ‘technology ladder’, externalities, synergies, balance-of-payments sustainability, gains from trade, and, in the case of developing countries, ultimately their capacity for ‘catching up’ were directly linked to the size, strength and depth of the manufacturing sector. In the more recent theories of this economic growth camp, increasing returns, ‘indivisibilities’, complementary capital, public goods, property rights, entrepreneurial capacity, transaction costs and the structure of incentives are also (directly or indirectly, explicitly or implicitly) related to the structure of output.

Therefore, from the point of view of the possible growth-effect of deindustrialisation, the first ‘growth camp’ does not look at de-industrialisation as a particularly relevant growth-issue other than on the question of whether other sectors of the economy would be able to absorb the labour force that is displaced from manufacturing. Furthermore, for these growth theories, even if the discovery of natural gas did produce some structural changes in the Dutch economy, to label these transformations a ‘disease’ must be seen as a misleading dramatisation!

From the point of view of the second camp, de-industrialisation in mature economies may or may not have a specific impact on growth; it would all depend on the specific form that this de-industrialisation takes. For example, it could actually result in a stimulus for growth if the dynamic of the “upward” deindustrialization is one characterised by the reallocation of resources within manufacturing into more R&D-intensive products. However, in the case of deindustrialization in middle-income countries it is difficult to imagine how this approach to growth can view “premature” (or “downward”) de-industrialisation – particularly when it includes a backward movement in the level of processing of primary commodities for exports, as in some Latin American countries – could in any way be good for long-term growth.

Finally, it goes without saying that in the third approach to economic growth both de-industrialisation – especially if it has the additional component related to the Dutch Disease – and the present difficulties of generating and implementing new and imaginative ‘palliative’ trade and industrial policies (given the current international ideological and institutional climates) are unambiguously major issues for growth in industrialised countries and developing countries alike. For example, one interpretation of the industrialised countries’ remarkable slowdown in productivity growth since the mid-1970s based on this approach would be that it could be precisely

Sloth and Indolence, which inevitably must introduce both a Neglect and Contempt of the Arts and Sciences; whereas an industrious Commonwealth, who keeps her subjects employed in Manufactures, and Foreign Trade, by continually furnishing Spain with such Things as there is a constant Demand for, to supply that People’s Conveniency, and feed their Pleasures, must needs in Return command as great a share of Spanish Bullion as they want […]. William Keith, History of the British Plantations in America (1738), quoted in Reinhart, 2003).


47

the result of ‘wrong’ policies (particularly in the 1980s) and ‘wrong’ structural change (e.g., “financialisation”) excessively intensifying the otherwise natural processes of de-industrialisation. Another obvious example would be the likely damaging long term growth-effects for developing countries of ‘premature’ de-industrialisation (such as in the Southern Cone of Latin America and Brazil) – not just for the speed of their economic growth but (crucially) for its sustainability.



48

5. Conclusions

If one makes a distinction between ‘relative’ and ‘absolute’ de-industrialisation, this paper only deals systematically with the former – i.e., the analysis has concentrated on the study of the shrinkage of the manufacturing sector (in terms of employment) relative to the rest of the economy in the context of a process of structural change brought about either by endogenous forces (e.g., a movement towards service industries in mature economies), or by exogenous ones (e.g., changes in economic policy in middle-income countries, and the discovery of natural resources). For reasons of space it has not been possible to analyse properly recent experiences of ‘absolute’ de-industrialisation: the decline of the manufacturing sector within the context of a collapse in national income, such as the one that took place in the former Soviet republics and in parts of Sub-Saharan Africa.41

The data and analysis presented in this paper have provided significant confirmation of the inverted-U-type of trajectory of manufacturing employment with respect to income per capita. However, it has also shown that this relationship has causes that are far more complex than has so far been recognized. Firstly, the relationship between these two variables is not stable over time. Moreover, there is convincing evidence that the original impulse for de-industrialisation was not the fact that some countries had already reached the level at which the curve begins to slope downward, but is instead closely related to a continuous fall over time of the actual inverted-U relationship for middle- and high-income countries. Furthermore, the 1980s combination of neo-liberal-politics and deflationary-monetarist-economics had a remarkably downward effect on the relationship between manufacturing employment and income per capita in industrialised economies.42

Secondly, the regressions presented above have also identified a huge drop in the turning point of the regressions that relate manufacturing employment to per capita income since the 1980s. Since the beginning of that decade, there has been a dramatic reduction in the level of income per capita from which the downturn in manufacturing employment began: from US$ 21,000 in 1980 to less than US$ 10,000 in 1990 (both measured in 1985 international US dollars). This rapid lowering of the turning-point of the regressions since 1980 is crucial to an understanding of the evolving nature of de-industrialisation: where by the 1980s, no country – not even the United States, the country with the highest income per capita in the sample – had reached a level of income per capita anywhere near the inflection point of the

41 In section three of the paper, ‘absolute’ de-industrialisation was also called ‘reversed’ de-industrialisation. 42 The former aimed at rolling back the welfare state, the radical transformation of industrial relations (in order to ‘discipline’ labour), and the creation of a new stream of rents (e.g., privatisations) in order to give a new impetus to capitalist accumulation (or, perhaps, just as part of the long awaited revenge of the rentier); the latter necessary for dealing with the newly discovered Friedman-type obsession with inflation.


49

inverted-U curve, by 1990 there were more than 30 countries (including first-tier NICs) whose income per capita was beyond that critical point in the respective curve.

Thirdly, the data and analysis presented in this paper have allowed us to develop a new, more specific (and, it is to be hoped, more useful) way of looking at the Dutch Disease. There is a group of countries that exhibits a specific additional degree of de-industrialisation (additional to the de-industrialisation caused by the three de-industrialising sources already mentioned above, that is).

This added de-industrialisation is associated either with a sudden surge in exports of primary commodities or the development of a successful service-exporting sector (mainly tourism or finance). From this perspective, the Dutch Disease is a process in which a country undergoes a change in its group of reference, from one corresponding to countries that need to generate a trade surplus in manufacturing to one corresponding to those able to generate a trade surplus in primary commodities or services. When this is the case, the country experiencing this “disease” moves along two different paths of de-industrialisation: the first, which is common to the countries in the original group; and, in addition, a second surge of de-industrialisation resulting from the change in reference group. In this context, the Dutch Disease should only be regarded as the “excess” degree of de-industrialisation associated with the latter movement.

Fourthly, this “disease” also spread to some Latin American countries; but the key issue in this case is that it was not brought on by the discovery of natural resources or the development of a service-exporting sector, but rather by a drastic switch in the economic policy regime.43 This was basically the result of a radical programme of trade and financial liberalisation, in the context of an overall process of economic reform and institutional change, leading to a sharp reversal of their (State-led) ISI industrialisation strategy. Brazil and the three Southern Cone countries with the highest incomes per capita (Argentina, Chile and Uruguay) were the Latin American countries that experienced the highest levels of de-industrialisation while also being among the countries in the region that both had previously industrialised most rapidly and had implemented the most drastic policy reforms. From this point of view, the major difference in the consequences of neoliberal politics and monetarist economics between Latin America and industrialised countries is that in the latter the crucial transformation took place in industrial relations, the welfare State, etc., while in the Latin American countries – as they were hit at a much lower level of income per capita – it also hampered their transition towards a more mature form of industrialisation (i.e., self-sustaining in a Kaldorian sense).

To begin with, ISI policies had achieved a degree of manufacturing employment that ‘normally’ corresponds to a situation in which the countries concerned seek to generate a trade surplus in manufacturing (although Latin American countries were never actually able to achieve this). In turn, a radical shift in the ‘policy regime’ (mostly implemented after the 1982 debt crisis) brought about the end of industrial

43 See, for example, ECLAC (2003), and Palma (2003).



50

and trade policies and, in particular, changes in relative prices, in real exchange rates, in the institutional framework of the economies, in the structure of property rights and in market incentives in general. This shift led them to abandon their industrialisation agenda, bringing them back to their ‘natural Ricardian position’; i.e., a position associated with comparative advantages more in accordance with their traditional resource endowment.

Fifthly, Finland, Sweden, Malaysia and, to a lesser extent, other Southeast Asian countries rich in natural resources (such as Thailand, Indonesia and the Philippines) prove that from the perspective of manufacturing employment, there is no such thing as the so-called “curse of natural resources”. It seems blatantly clear that countries rich in natural resources or high potential for developing strong export-services activities have sufficient degrees of freedom to allow them to follow trade and industrial policies to continue developing a strong manufacturing industry – let alone implementing policies that would avoid the ‘Dutch Disease’.

However, as the Latin American experience in particular shows, it would seem that as globalisation progresses, there are fewer and fewer countries left with the political will to take advantage of such degrees of freedom and undertake policies that promote or maintain manufacturing capacity. This not only because the new international institutional order is rapidly trying to narrow down these degrees of freedom, but also because of the obvious role of fundamentalism in economic ideologies, and the fact that the newly develop neo-liberal structure of property rights has (at least so far) been perfectly capable of generating alternative massive (non-manufacturing) rents from which domestic political and economic elites have been able to profit.

However, whether a process of structural change that includes ‘premature’ de-industrialisation can ever deliver rapid and sustainable economic growth is another matter altogether; so is the issue of whether the current ‘premature’ de-industrialisation taking place in the Southern Cone of Latin America and in Brazil will turn out to be simply a paradigmatic case of policy-induced ‘uncreative destruction’.


51

Endnote 1 – Country classification

The classification of countries was based on their 1998 position; all Latin American countries are classified in the ‘primary commodity’ group (including those that export maquila manufactures due to the problems discussed above, in particular the low levels of domestic value-added of these exports).

Non-primary commodity (and export-services) group

Primary commodity and export-services group

Austria Korea, Rep. Argentina Malawi

Belgium Malaysia Australia Malta

Denmark Morocco Bolivia Mauritius

Egypt Pakistan Botswana Mexico

Finland Philippines Brazil Mozambique

France Portugal Canada Namibia

Germany Singapore Chile Netherlands

India Spain Colombia New Zealand

Indonesia Sri Lanka Congo Nicaragua

Ireland Sudan Costa Rica Niger

Israel Sweden Cote d'Ivoire Norway

Italy Taiwan Cyprus Panama

Japan Thailand Dominican Republic Paraguay

Jordan Tunisia Ecuador Peru

El Salvador Puerto Rico

Gabon Reunion

Ghana South Africa

Greece Switzerland

Guatemala Tanzania

Guyana Turkey

Haiti Uganda

Honduras United

Iceland Kingdom

Kenya United States

Liberia Uruguay

Present in all samples

Venezuela

Bangladesh Algeria Guinea-

China Angola Bissau

Syria Burkina Faso Lesotho

Burundi Luxembourg

Cameroon Madagascar

Cape Verde Mali

Central African Republic Mauritania

Chad Myanmar

Comoros Nigeria

Ethiopia Rwanda

Gambia Senegal

Guinea Swaziland

Zimbabwe

Missing in the 1960 sample, but present in all other samples



52

Endnote 2 – Econometric results: Ordinary Least Squares Estimation

Parameters’ point estimation

Reg. 1 Reg. 2 Reg. 3 Reg. 4 Reg. 5

Intersect -16.71 -15.96 -14.98 -17.78 -16.47 Ln Y pc 4.189 3.889 3.660 4.491 4.204 Ln Y pc sq -0.218 -0.195 -0.183 -0.242 -0.228

Regression 1 corresponds to the ‘average’ regression for 1960 (see Figure 3); Reg. 2 corresponds to that for 1970; Reg. 3 for 1980; Reg. 4 for 1990; and Reg. 5 for 1998. Ln Y pc = log of income per capita; Ln Y pc sq = square of the log of income per capita.

‘t’ values


Intersect -3.5 -4.9 -4.3 -5.5 -5.0 Ln Y pc 3.3 4.5 4.1 5.5 5.0 Ln Y pc sq -2.6 -3.5 -3.2 -4.7 -4.3

‘p’ values


Intersect 0.001 0.000 0.000 0.000 0.000 Ln Y pc 0.002 0.000 0.000 0.000 0.000 Ln Y pc sq 0.011 0.001 0.002 0.000 0.000

Regression statistics


R-bar-sq 0.65 0.72 0.67 0.64 0.57 F (2, 78) 75.1 -- -- -- -- F (2,102) -- 135.5 107.0 92.2 70.1 ‘p’ of F 0.000 0.000 0.000 0.000 0.000

R-bar-sq is the adjusted coefficient of determination; and F is the F statistic. All regressions pass the tests for homoscedasticity and for the normality of residuals at the 5% level of significance.


53

Parameters’ point estimation


Intersect -14.65 -15.35 -14.82 -17.90 -16.73 ‘pc’dummy -0.334 -0.434 -0.419 -0.274 -0.243 Ln Y pc 3.724 3.848 3.760 4.637 4.401 Ln Y pc sq -0.189 -0.196 -0.194 -0.255 -0.245

Regression 6 corresponds to the regression for 1960 when an intercept-dummy is added for countries able to generate a trade surplus in primary commodities or export-services (see Figure 7 and 8); Reg. 7 corresponds to that for 1970; Reg. 8 for 1980; Reg. 9 for 1990; and Reg. 10 for 1998. ‘p c ’ dummy is the intercept-dummy.

‘t’ values


Intersect -3.2 -5.2 -5.0 -7.2 -6.9 ‘pc’ dummy -2.9 -4.1 -4.0 -2.8 -2.5 Ln Y pc 3.1 4.9 5.0 7.3 7.2 Ln Y pc sq -2.4 -3.9 -4.0 -6.4 -6.4

‘p’ values


Intersect 0.002 0.000 0.000 0.000 0.000 ‘pc’ dummy 0.006 0.000 0.000 0.000 0.014 Ln Y pc 0.003 0.000 0.000 0.000 0.000 Ln Y pc sq 0.020 0.000 0.000 0.000 0.000

Regression statistics


R-bar-sq 0.67 0.77 0.74 0.74 0.70 F (3, 77) 55.1 -- -- -- -- F (3,101) -- 114.1 99.9 99.4 81.3 ‘p’ of F 0.000 0.000 0.000 0.000 0.000

All regressions pass the tests for homoscedasticity and for the normality of residuals at the 5% level of significance.



54

Bibliography

Aghion, P. and P. Howitt (1998), Endogenous Growth Theory, MIT Press.

Barro, R. and X. Sala-i-Martin (2004), Economic Growth, McGraw-Hill.

Bhagwati, J. (2003) ‘Testimony’, Subcommittee on Domestic and International

Monetary Policy, Trade and Technology, US House of Representatives

Committee on Financial Services, April 1, 2003.

Blankenburg, S. (2000), ‘Knowledge, economic growth and the role of policy’, ESRC Centre for Business Research, Department of Applied Economics, Working Paper 185.

------------------ (2004), Knowledge in the ‘New Endogenous Growth Theory’: A factor of production, a good and an externality, unpublished Ph D thesis, Cambridge University.

Bhaduri, A. and D. Nayyar (1996) The Intelligent’s Person Guide to Liberalization, Penguin Books.

Caputo, O., (1996), ‘La sobreproducción mundial de cobre creada por Chile. Su impacto en la economía nacional’, ARCIS-CETES.

Carrillo, J. (2002) ‘Foreign direct investment and local linkages: experiences and the role of policies. The case of the Mexican television industry in Tijuana’, mimeo, UNCTAD.

Chang, H-J. (ed.) Rethinking Development Economics, Anthem Press.

COMTRADE (2002), CD-Room, ECLAC and World Bank.

Di John, J. (2003), Growth as Interactive Process of Political and Economic Strategy: The Political Economy of Industrial Policy and State Capacity in Venezuela, 1920-2000, unpublished Ph D dissertation, Cambridge University.

ECLAC (2002) Foreign Investment in Latin America and the Caribbean. 2002 Report.

Hume, D. (1767) History of England, vol. III, Cambridge University Press.

ILO Database, at http://www.ilo.org.

Kaldor, N. (1967), ‘Problems of Industrialization in Underdeveloped Countries’, in Strategic Factors of Economic Development, Cornell University Press.

Lagos, J. F. (2000), Chile, de Exportador de Cobre Refinado a Gran Exportador de Concentrados: la segunda fase exportadora ?, mimeo, Santiago de Chile.


55

La Jornada, at http://www.jornada.unam.mx.

Lavanderos, J. (2001), El Cobre NO, es de Chile, el cobre no es de Chile, USACH.

Mandeng, O. (1998) Change and Adaptability in International Trade, unpublished M Sc thesis, Cambridge University.

Mortimore, M. (2000) ‘Libre comercio, integración y el futuro de la industria maquiladora. Producción global y trabajadores locales’. mimeo, COLEF, Tijuana.

Ocampo, J. A., R. Bajraj, and J. Martin (2001), Una Década de Luces y Sombras. América Latina y el Caribe en los años noventa, ECLAC.

Palma, J. G. (1996), ‘Does it make a difference to export micro-chips rather than potato-chips? Comparing export structures in East Asia and Latin America', mimeo, Cambridge; and UNCTAD Working Paper, 1998.

-------------, (2000), ‘Trying to “tax and spend” oneself out of the Dutch Disease’. The Chilean economy from the War of the Pacific to the Great Depression’, in E Cárdenas, J A Ocampo and R Thorp (eds.) An Economic History of Latin America, Vol. 1, The Export Age. The Latin American Economies in the Late Nineteenth and Early Twentieth Centuries, Palgrave.

--------------, (2002), ‘Trade Liberalisation in Mexico: its impact on growth, employment and wages’, ILO Employment Paper N. 55, http://www.ilo.org.

--------------, (2003), ‘Latin America during the second half of the twentieth century: from the ‘age of extremes’ to the age of ‘end of history’ uniformity’, in H-J. Chang (ed.), 2003.

--------------, (2004a), ‘The economic consequences of different regional leadership: the rôle of the US and Japan in the economic development of Latin America and East Asia’, in J L Fiori, O Poder Americano, Editora Vozes, Rio de Janeiro.

--------------, (2004b), “Flying-geese and waddling-ducks: the different capabilities of East Asia and Latin America to ‘demand-adapt’ and ‘supply-upgrade’ their export productive capacity”, in M Cimoli, G Dosi and J Stiglitz (eds.), Industrial policy in Developing Countries, forthcoming, Oxford University Press.

Pesaran, H. (1984), Macroeconomic policy in an oil-exporting economy with foreign exchange controls, Economica.

Pesaran, H. et al. (2000), ‘Neglected heterogeneity and dynamics in cross-country savings regressions’, in Krishnakumar, J. and Ronchetti, E. (eds.), Panel Data Econometrics - Future Direction, Elsevier Science.

Pieper, U (2000), ‘De-industrialisation and the social and economic sustainability nexus in developing countries: cross-country evidence on productivity an employment’, Journal of Development Studies, Vol. 36.



56

Reinhart, E. S. (2003) ‘Increasing poverty in a globalised world: Marshall Plans and Morgenthau Plans as mechanisms of polarization of world incomes’, in H-J. Chang (ed.), 2003.

Romer, P. M. (1990), ‘Endogenous technical change’, Journal of Political Economy, Vol. 98.

Ross, J. (2000), Development Theory and Economic Growth, University of Michigan Press.

Rowthorn, R. (1994), ‘Korea at the Cross-roads’, Centre for Business Research, Cambridge University, Working Paper no. 11.

--------------- (1997), ‘Manufacturing in the World Economy’, Economie Appliqueé, No. 4.

--------------- (1999), ‘The Political Economy of Full Employment in Modern Britain’, The Kalecki Memorial Lecture, Department of Economics, University of Oxford, 19 October 1999.

Rowthorn, R. and J. Wells (1987), De-industrialisation and Foreign Trade, Cambridge University Press.

Rowthorn, R. and R. Ramaswamy (1999), ‘Growth, trade and deindustrialization’, IMF Staff Papers, Vol. 46.

Sen, A. (1999) Beyond the Crisis: development strategies in Asia, Institute of South East Asian Studies.

Solow, R. (1997) ‘What is labour market flexibility ? What is it good for? Keynes Lecture, British Academy. Proceedings of the British Academy.

Stiglitz, J. (2002), Globalization and its Discontents, W. W. Norton Inc.

----------- (2004), “Chile perdió soberanía al firmar TLC con EEUU”, in El Mostrador (www.elmostrador.cl), 15 December, 2003.

The Economist (2001) ‘Chile could do more to become less dependent on copper’, 29 November.

Tomić, R. (1985), La política minera chilena, Santiago de Chile, CEP, 1985.

Tomić, C. (1999), ‘El cobre debe ser refinado en Chile’, mimeo, Santiago de Chile. Trade and Development Report (2002), UNCTAD.

Palma,G.2004DeindustrialisationDutchDisease

Documents

Transcript of Palma,G.2004DeindustrialisationDutchDisease