Directory UMM :Data Elmu:jurnal:UVW:World Development:Vol28.Issue9.Sep2000:
World Development Vol. 28, No. 9, pp. 1583±1596, 2000
Ó 2000 Published by Elsevier Science Ltd.
Printed in Great Britain
0305-750X/00/$ - see front matter
www.elsevier.com/locate/worlddev
PII: S0305-750X(00)00050-4
Structural Change and Labor Productivity Growth in
Latin American Manufacturing Industries 1970±96
JORGE KATZ *
Economic Commission for Latin America and the Caribbean, Santiago, Chile
Summary. Ð This paper examines recent changes in the pattern of production specialization
attained by Latin American manufacturing industries. Comparing Argentina, Brazil, Colombia,
Chile and Mexico to the United States, the paper identi®es ``catching up'' and ``lagging behind''
industries during 1970±96. It considers the evolution of macro and micro forces that aect whether
an industry manages to close down the gap with the international technological frontier. Evidence
is presented showing that recent structural reforms did not result in a major discontinuity with the
past. Rather, ``path dependency'' forces have acted as a major factor explaining dierential
performance among industries after recent changes in the global incentive regime. Ó 2000
Published by Elsevier Science Ltd.
Key words Ð productivity growth, technological gaps, path dependency, structural reforms, Latin
America
1. INTRODUCTION
Recent studies (Katz, Benavente, Crespi &
Stumpo, 1997; Katz & Vera, 1997, 1999)
examining some of the major changes that have
taken place in the Latin American industrial
sector during the last two decades have shown a
major change in the pattern of production
specialization of most economies. There has
been a shift in favor of nontradable sectorsÐ
those producing telecommunication, energy or
®nancial services, for exampleÐas well as
toward natural resource-processing industriesÐproducing iron and steel, petrochemicals, nonferrous minerals, ®shmeal, vegetable
oil, pulp and paper, etc. In addition, electronic
and garment maquiladoras (assembly plants), as
well as automobile producers, have grown quite
rapidly throughout this period in response to
ad hoc industrial policies applied by local
economic authorities.
Large domestically-owned conglomerates, as
well as subsidiaries from transnational corporations (TNCs), have set up very modern,
highly capital-intensive new production plants
in the above mentioned ®elds. Leaving aside the
case of nontradable goods and services, we
notice that many of these new production
facilities have become major exporters of
industrial commodities to highly competitive
world markets. Latin American ®rms act in
such markets as ``price takers,'' with little
bargaining power and attaining low unit pro®t
margins. Given the high degree of volatility
shown by world prices of industrial commodities, the external sector of most Latin American
economies now stands on rather weak foundations which need to be taken into consideration when we examine the long-term
sustainability of the recent restructuring of the
regional production fabric.
In contrast to the above, labor-intensive
industries producing ®nal consumer goods such
as footwear or clothing, and engineering and
R&D-intensive sectors producing capital
goods, ®ne chemicals, 1 or scienti®c instruments, have been shrinking and contracting
(with the exception of the electronic maquila).
Following trade liberalization and market
deregulation eorts the former have had a hard
time competing with low-cost producers from,
* The present paper has been written as part of a larger
research program undertaken with ®nancial support
from the German Technical Cooperation Agency (GTZ)
and the International Cooperation Agency of the
Netherlands. To both these agencies, as well as to
W. Peres and N. Reinhardt for their useful comments on
an earlier version of the paper, the author wishes to
express his gratitude. The views hereby advanced are his
exclusive responsibility.
1583
1584
WORLD DEVELOPMENT
say, China or Vietnam, where wage levels are
just a fraction of those typically paid in Latin
America, while the latter have been unsuccessful in keeping up with the pace of technological
modernization in areas characterized by large
R&D and engineering expenditure, short
product life cycles and high rates of technological obsolescence. As these industries have
continued to incorporate the use of microprocessors, CNC technologies, new biotechnological know how, etc. at a breath-taking pace,
Latin American ®rms have found themselves
lagging further and further behind the worldÕs
technological frontier. The capability of these
®rms to compete in the international marketplace in such ®elds has lagged behind as a result
of these trends.
In many countries of the region this process
of change in the pattern of production
specialization actually began during the late
1970s, i.e. prior to the so-called Debt Crisis,
when the pace of domestic market-led growth
began to slow-down. In othersÐparticularly
ChileÐthe restructuring toward natural
resource-processing industries clearly resulted
from the major transformation in the global
incentive regime which obtained after the military takeover of 1973. It is important therefore
to understand that the change in the pattern of
production specialization we are about to
examine in this paper did not come about in
each country of the region as a result of trade
liberalization and market deregulation eorts.
In many cases, such change was already in the
making well before structural reform eorts
actually began. 2
It is true, however, that in both cases a new
relative price structure, less supportive of
import substitution industrialization activities,
has gained ground during the 1990s. Concomitantly, the ¯ow of direct foreign investment
and the process of capital formation have
accelerated, under the impact of both a more
liberalized and deregulated domestic institutional and economic environment and of a
buoyant international capital market. Property
rights in the area of natural resources have been
strengthened and many institutions prototypical of more developed industrial societiesÐ
such as patent laws granting wider and stronger
intellectual property rightsÐhave been gradually adopted by most Latin American countries.
In short, the regionÕs pattern of production
specialization and its linkages with world
markets are going through a major, and as yet
un®nished, transformation. Central to this
process is a retreat to resource-based processing
industries, to nontradable goods and services,
to labor-intensive electronic and garment
maquiladoras, and to vehicles and transport
equipment receiving special ad hoc treatment
from various governments in the region, notably Mexico, Brazil, Argentina, Colombia,
Venezuela and Chile.
Our purpose here is to examine how the
emerging new industrial structure has
performed in terms of labor productivity
growth vis-
a-vis more developed industrial
economies. Is the Latin American industrial
sector ``catching up'' or ``lagging behind''
relative to the worldÕs technological frontier?
For the purpose of answering this question we
shall use productivity growth in the US
manufacturing sector as our benchmark.
For a study of this sort we should ideally be
using estimates of total factor productivity
growth. Figures for gross capital formation at
the three-digit level of aggregation are
unavailable, however, making it impossible to
calculate such index. We have therefore decided
to perform the comparison on the basis of
labor productivity growth. Using for this
purpose a data base recently put together at
ECLAC we have examined the aggregate
performance of the industrial sector of nine
Latin American countries vis-a-vis
the United
States over 1970±96. We continue with a more
detailed comparison examining labor productivity growth in 27 industries at the three-digit
level of aggregation of the International Standard Industrial Classi®cation of All Economic
Activities (ISIC) for ®ve countriesÐArgentina,
Brazil, Chile, Colombia and MexicoÐfor
1970±96. Special attention is given to the 1990s,
as this is the period in which structural reform
eorts were strengthened.
In Section 2 we brie¯y discuss the analytical
framework upon which the present research is
based. In Section 3 we report our main statistical results. Section 3 also looks at some meso
and micro aspects of recent changes in industrial production structure. In particular, we pay
attention to the entry and exit of ®rms to and
from the market, as well as to changes in the
pattern of production specialization. Section 3
further summarizes our view concerning the
relationship between changes in the macropolicy incentive regime and changes in the
structure and behavior of manufacturing
industry. Micro-to-macro linkages frequently
remain unexplored in the literature on the
STRUCTURAL CHANGE AND LABOR PRODUCTIVITY GROWTH
impact of economic reforms in the region. We
believe this to be a major shortcoming. In our
view, such linkages are crucial for the understanding of the impact of structural reforms. A
few suggestions for further research are
advanced in the ®nal paragraphs of the paper.
2. THE ANALYTICAL FRAMEWORK OF
THE PRESENT RESEARCH
Contemporary studies on economic growth
show at least two dierent ways of approaching
the issues we want to examine here. On the one
hand, received literature oers us ``top-down''
explanations of growth of the type Solow
developed in the 1950s and 1960s, which have
recently reemerged within the framework of
modern growth theory in the work of Barro
and Sala-i-Martin (1995), Romer (1992), and
others. These studies are based on the conventional microfoundations of the ``representative
®rm'' and on neoclassical ``growth accounting''
principles that rely on assumptions of perfect
information, ``generic'' production functions,
``well-behaved'' ®rms, equilibrium growth
paths, perfectly competitive markets, and
production factors being paid their marginal
productivity.
The explanation of why productivity growth
is attained in such a stylized analytical framework is extremely simple. Historical and institutional factors virtually play no role at all. It is
precisely such degree of simpli®cation of reality
and the elimination of market imperfections,
uncertainty, etc. that allows us to have aggregate production, investment, consumption and
saving functions on the basis of which to look
at the behavior of the global economy.
Contemporary growth theory has in recent
years incorporated increasing returns to scale
and externalities speci®ed at the industry level,
i.e. outside the realm of any one speci®c ®rm.
This allows the analyst to maintain the
assumption of competitive behavior at the level
of the individual production unit, and therefore
to save the microfoundations of conventional
price theory, even while increasing returns and
externalities are being incorporated into our
formal thinking about the growth process. But
as Nelson has recently pointed out, modern
growth theory still faces major diculties
bringing on board both the role of institutions
and of production organization, as both such
topics still appear dicult to tackle through
formal mathematical modeling. (Nelson, 1997).
1585
Over the course of the last two decades an
increasing number of economists have
approached the study of innovation and
productivity growth from an ``evolutionary''
perspective (e.g., Nelson & Winter, 1982; Dosi
et al., 1988; Freeman, 1994; Metcalfe, 1997).
These authors see productivity growth as the
outcome of a process of natural selection
among ®rms, rather than as the result of
conventional maximizing behavior. The success
and failure of dierent ®rms, imperfect information, uncertainty, trial and error are all
central features of a screening mechanism
which involves both quasi-genetic ``mutations''
(Nelson, 1995) and historical, cultural and
institutional forces conditioning the economyÕs
growth path over time. Here, growth is not the
outcome of equilibrium behavior but is instead
a ``cultural construct'' deeply embedded in the
technological and institutional foundations of
society. The growth story attains a historical
and institutional atmosphere that is simply
absent from the neoclassical account of the
facts. Whereas the latter is based on conventional price theory, the ``evolutionary'' story
has a large component of ``social anthropology'' that goes well beyond the scope of the
professional economistÕs formal tool box.
The evolutionary account of why labor
productivity growth takes place in any given
environment considers competition as a ``selection'' mechanism winnowing ®rms, as well as
inter®rm dierences in structure, strategy and
accumulated technological capabilities. It also
considers the gradual transformation (through
learning processes) of the institutional and
regulatory environment in which dierent
countries and industries operate through time.
Historical and institutional variables play a
central role in¯uencing what individual agents
want, know and actually do, whereas in the
neoclassical model they always know ex ante all
that needs to be known, and do exactly all that
has to be done, in order to maximize bene®ts.
In the neoclassical model exogenous forces
induce robotic-type responses on the part of
®rms and consumers. Information is perfectly
accessible to them and production functions are
of a generic sort. In the evolutionary model,
behavior is ``adaptive,'' i.e., based on trial and
error, while information has to be gradually
attained through learning processes, via the
accumulation of experience. Production functions are improved upon as a result of engineering and R&D activities. Some companies
are better than others at producing incremental
1586
WORLD DEVELOPMENT
units of technological and organizational
knowledge and therefore do better through
time. Market structure and performance
change endogenously re¯ecting dierential
processes of knowledge accumulation across
®rms.
Our research draws a great deal on recent
``evolutionary'' thinking. Macro, meso and
micro variables, on the one hand, and the coevolution of economic, institutional, and technological forces, on the other, seem to us to be
of crucial importance for the understanding of
the impact trade liberalization and market
deregulation eorts are having today in Latin
America. We need to take into account industry-speci®c variables and institutions in order
to understand how each particular sector reacts
to changes in macroeconomic and regulatory
signals. Moreover, we think that a micro-tomacro feedback mechanism (loop) needs to be
incorporated in our growth models, re¯ecting
the way in which the ``new'' features of
the production structure in¯uenceÐthrough
exports and investmentÐthe long-term sustainability of the new pattern of production
specialization attained by the economy after
structural reforms. The profession has not so far
adequately explored many of these micro-tomacro relations.
3. LABOR PRODUCTIVITY GROWTH IN
LATIN AMERICAN MANUFACTURING
INDUSTRY DURING 1970±96
Using a database recently prepared at
ECLACÕs division of Production, Productivity
and Management covering 27 industries at the
ISIC three-digit level of aggregation, we have
estimated labor productivity growth (measured
as value-added per capita) for Argentina,
Brazil, Colombia, Chile, Costa Rica, Jamaica,
Mexico, Peru, Uruguay for 1970±96. Similar
data have been put together for the US manufacturing sector in order to serve as a benchmark for comparison in order to identify
``catching up'' and ``lagging behind'' countries
and industries. While using US data might not
always be our best possible point of reference, it
nevertheless establishes an interesting and
accessible benchmark for comparison.
We begin by looking at the aggregate picture
for each one of the above mentioned countries.
We then proceed to the three-digit level of
aggregation, in order to attain a more detailed
picture of the relative performance of dierent
branches of manufacturing production in the
region vis-a-vis
their US counterpart.
(a) Labor productivity growth in manufacturing
production in Latin America and the US
Our results in Table 1 show that, as far as
manufacturing production is concerned, only
three Latin American countriesÐArgentina,
Colombia and MexicoÐhave managed to
attain a rate of growth of labor productivity
over the whole 1970±96 period higher than the
one attained by the US industrial sector. In
other words, only these three countries of the
region have managed partially to reduce the
relative labor productivity gap with the United
States during the nearly three decades covered
by our information.
Table 1. Output, employment and labor productivity growth in Latin American and US manufacturing production in
1970±90 and 1990±96a
Manufacturing output
Argentina
Brazil
Chile
Colombia
Costa Ricab
Jamaicab
Mexico
Peru
Uruguay
United
States
a
Employment
Labor productivity
1970±96
1990±96
1970±96
1990±96
1970±96
1990±96
1.18
2.81
2.76
3.98
4.39
0.11
3.79
1.17
0.61
2.39
4.87
2.26
6.40
3.52
n.a.
n.a.
2.27
5.09
)1.46
5.04
)2.62
0.95
1.51
1.24
4.83
1.66
0.91
2.85
0.37
0.35
)3.15
)6.41
3.49
)0.22
n.a.
n.a.
)0.03
1.97
)8.58
0.30
3.80
1.86
1.25
2.74
)0.44
)1.55
2.88
)1.68
0.24
2.04
8.02
8.67
2.91
3.74
n.a
n.a
2.30
3.12
7.12
4.74
Source: Industrial Performance Analysis Program (PADI), ECLAC Division of Production Productivity and
Management.
b
Data until 1992.
STRUCTURAL CHANGE AND LABOR PRODUCTIVITY GROWTH
The data also show that the rate of growth of
labor productivity in manufacturing production increased quite signi®cantly during the
1990s in all of the countries under consideration. This is particularly the case for Argentina, Brazil and Uruguay, where the
acceleration of labor productivity growth in
manufacturing production during the 1990s
was especially strong. We notice, however, that
in all three countries the faster rate of labor
productivity growth of the 1990s is more the
re¯ection of a very high rate of labor
displacement from manufacturing industry
than of a high rate of expansion of industrial
production per se. In fact, manufacturing
growth was not particularly strong in most of
the countries under examination throughout
the 1990s. In other words, the pattern we are
trying to describe here does not seem to be one
of rapid rates of productivity growth deriving
from a major outward expansion of output, but
rather of major ``labor saving'' restructuring
eorts carried out by an industrial sector that is
not growing very rapidly through time. 3
Table 2 presents the ratio of labor productivity in manufacturing production in each of
the countries under examination and in the
United States, for 1970, 1980, 1990 and 1996.
None of the Latin American countries had
attained more than about one-third of US
labor productivity in 1970. By the end of the
period under consideration, only three countriesÐArgentina, Colombia and MexicoÐexhibit signs of having partially reduced the labor
productivity gap with the United States. Interestingly, much of the (small) improvement
attained by Colombia and Mexico came about
during the two decades prior to trade liberal-
Table 2. Relative labor productivity in Latin American
manufacturing production vis-
a-vis the United Statesa
a
Country
1970
1980
1990
1996
Argentina
Brazil
Chileb
Colombia
Costa Rica
Jamaicac
Mexicod
Peru
Uruguaya
0.42
0.28
0.25
0.29
)
0.26
0.32
0.33
0.35
0.41
0.26
0.24
0.25
)
0.16
0.30
0.25
0.22
0.55
0.29
0.23
0.37
0.15
0.16
0.44
0.16
0.20
0.67
0.37
0.20
0.34
0.14
0.13
0.38
0.15
0.22
Source: PADI, ECLAC.
Data until 1995.
c
Data until 1992.
d
Data until 1994.
b
1587
ization and market deregulation eorts, i.e. in
1970±90. In the case of Argentina, about half of
the relative improvement in labor productivity
vis-a-vis
the United States took place during
1970±90, i.e. prior to structural reforms, while
the other half was obtained during the 1990s,
i.e. concomitantly with trade liberalization and
market deregulation eorts. Our ®gures also
indicate that aggregate manufacturing labor
productivity increased quite signi®cantly in
Argentina during 1990±96 but that this was not
the case for Mexico and Colombia. Brazil, on
the other hand, does not show signi®cant
improvement in labor productivity vis-
a-vis the
United States during 1970±90, but achieved a
major relative improvement during 1990±96.
This improvement appears to be basically the
outcome of a strong labor displacement
phenomena.
In the cases of Chile (in spite of the fact that
labor productivity has increased signi®cantly
since the opening up and deregulation of the
economy) and Costa Rica, no signs of relative
improvement can be detected over time.
Finally, the relative positions of Peru, Uruguay
and Jamaica have clearly deteriorated.
(b) Labor productivity growth in Latin
American and US industries during 1970±96
For a more detailed level of analysis, we have
estimated labor productivity growth for 27
dierent branches of manufacturing production
for Argentina, Brazil, Chile, Colombia and
Mexico for 1970±96. We have then estimated
through a simple logarithmic regression the
long-term trend of each series. Dividing the
coecient thus obtained by a similarly calculated coecient for US industries covering the
same period of time gives us an index of relative
labor productivity. The results are presented in
Table 3.
Table 3 clearly shows Latin American
industries ``catching up'' or ``falling behind''
the US labor productivity standard. An index
larger than one indicates that the industry of
reference has managed to grow faster in terms
of labor productivity than its US counterpart
throughout the period, whereas an index
smaller than one depicts the opposite situation.
Given the fact that labor productivity in most
of these industries was not more than 30% or
40% of US labor productivity in our base
year (1970), a large indexÐsay, around twoÐ
indicates that the gap was signi®cantly
reduced, even though, in absolute terms, labor
1588
WORLD DEVELOPMENT
Table 3. Relative labor productivity in Latin American manufacturing production vis-
a-vis the United States, 1970±96,
27 branches of industrya
ISIC
311
313
314
321
322
323
324
331
332
341
342
351
352
353
354
355
356
361
362
369
371
372
381
382
383
384
385
390
Category
Food manufacturing
Beverage industries
Tobacco manuf.
Textiles
Wearing and apparel
Leather products
Footwear
Wood products
Furniture
Pulp and paper
Printing/publishing
Chemicals
Chemical products
Petroleum re®neries
Petroleum and coal
Rubber products
Plastics
Pottery and china
Glass products
Non-metallic minerals
Iron and steel
Nonferrous metals
Metal products
Non-electrical mach
Electrical machinery
Transport equipment
Scienti®c instruments
Other industries
Argentina
1.10
1.04
0.74e
1.67
1.17
1.38
0.78e
0.55e
2.69f
0.99
1.21
1.92
1.98
1.22
1.85
1.55
0.81
1.33
1.91
2.35f
2.54f
1.28
2.07f
1.91
2.68f
2.00f
1.29
0.52e
Brazilb
1.14
0.72e
0.21e
1.43
1.20
0.93
1.13
0.87
1.40
1.26
0.86
1.18
0.60e
1.57
2.07f
2.55f
1.25
1.20
1.92
1.28
1.97
2.50f
1.78
1.12
1.97
1.33
1.48
0.76e
Chilec
e
0.67
0.91
0.76e
0.77e
0.75e
0.45e
0.65e
0.97
1.13
1.10
1.43
1.79
0.97
3.35f
2.14f
0.41e
0.51e
0.45e
1.67
1.68
1.33
0.43e
1.22
1.31
0.94
0.76e
1.22
0.92
Colombia
Mexicod
0.93
0.79e
0.28e
1.23
1.30
0.58e
1.03
0.94
0.85
1.12
0.89
1.09
0.86
0.28e
2.10f
1.36
1.50
2.24f
1.57
1.36
2.82f
1.92
1.79
0.75e
0.99
2.07f
3.27f
1.26
1.21
0.83
0.38e
0.75e
1.85
0.97
0.72e
0.94
0.96
1.03
1.03
0.88
0.58e
0.30e
1.55
1.24
1.25
2.47f
1.60
1.39
1.54
2.39f
1.39
0.72e
1.76
1.81
3.81f
2.19f
a
Source: PADI, ECLAC.
1996/1970.
c
1995/1970.
d
1994/1970.
e
6 0:80 ``falling behind'' industries.
f
P 2:0 ``catching up'' industries.
b
productivity could continue to be below the US
level in 1996. These industries are considered to
be ``catching up'' to the US standard. On the
other hand, a number, say, equal to or smaller
than 0.80 suggests that the relative gap between
the two has expanded signi®cantly through
time, and the industry is ``falling behind.''
Six dierent branches of industryÐmarked
with ``f'' in the table for identi®cationÐappear
in Argentina as ``catching up'' with the US
standard during 1970±96. Five industries
attained a similarly ``successful'' performance
in Colombia, four in Mexico, three in Brazil
and two in Chile. On the other hand, four
sectorsÐmarked ``e'' in Table 3Ðshow clear
signs of ``falling behind'' the US standard: in
Argentina and Brazil, ®ve in Colombia, six in
Mexico and 11 in Chile. Argentine industries
have done the best and Chilean ones the worst
in ``catching up'' with the US standard. Brazil,
Colombia and Mexico fall somewhere in
between these two extremes.
What explains the dierences in the productivity performance of individual industries
within each country? One hypothesis is that the
fastest-growing industries were the ones that
did better in terms of reducing the original
labor productivity gap with their US counterpart, while the ``lagging behind'' industries
tended to be those that attained a worse than
average growth performance. In order to test
this hypothesis, we estimated in a cross-section
regression the ordinary least squares (OLS)
relationship between the coecients reported in
Table 3 and the observed rate of growth of each
industry in each country throughout the period
under consideration. The results, shown in
Table 4, tend to con®rm our a priori expecta-
STRUCTURAL CHANGE AND LABOR PRODUCTIVITY GROWTH
1589
Table 4. Crossindustry regressions between relative productivity growth and the rate of growth of output, 1970±96
Argentina
Brazil
Chile
Colombia
Mexico
Regression coecient
S.E.
R2
F-test
Degrees of freedom
0.77
1.13
0.68
0.89
1.17
0.13
0.06
0.14
0.25
0.21
0.59
0.95
0.55
0.33
0.57
33.56
419.05
22.39
12.59
31.97
23
22
20
25
24
tion that the interindustry variance in the rate
of growth constitutes an important part of the
explanation of why some sectors ``catch up''
with labor productivity levels in the United
States, while other sectors ``lag behind'' in
relative terms.
An adequate explanation of interindustry
dierences in the degree of productivity
convergence must therefore include an explanation of why some sectors grow faster than
others through time. This could be the consequence of both macro and sector-speci®c
forces. On the one hand, we could a priori
expect macro forces to play a signi®cant role
inducing some sectors in the economy to grow
faster than others. Indeed, the NEM is predicated on an expectation that trade liberalization will induce a shift in resource allocation
along comparative advantage lines, i.e. toward
labor intensive and/or natural resource-intensive industries. On the other hand, sector
speci®c forcesÐthe ``localized'' institutional
and regulatory environment in which each
particular industry operates, the ad hoc treatment each sector manages to get from local
economic authorities, etc.Ðcould also be
expected to in¯uence interindustry dierences
in growth performance. Thus, both macro and
sector-speci®c variables may in¯uence why
some sectors ``catch up'' and others ``fall
behind'' in the process of economic growth.
We have previously noted that the aggregate
productivity performance of dierent countries
in the region vis-
a-vis the United States does
not appear to be exclusively determined by
recent trade liberalization and market deregulation eorts. The data previously presented
show that part of the ``catching up'' attained by
the industrial sector of Argentina, Colombia
and Mexico during the period under consideration was already ``in the making'' during
1970±90 and that it would be somewhat
misleading to assign it exclusively to recent
market-oriented reforms. Do the three-digit
data con®rm such a picture? In order to throw
light upon this question, we have calculated the
same index of relative industry productivity
performance for 1970±90. Our results are
presented in Table 5.
By comparing Tables 3 and 5 we con®rm our
previous ®nding, i.e. that in many cases the
``catching up'' or the ``falling behind'' of certain
industries was already ``in the cards'' before
trade liberalization and market deregulation
measures were actually introduced by the
various governments in the region. ``Path
dependency'' appears to be a strong structural
feature of production systems, which induces us
a priori to expect that industries that did well in
terms of ``catching up'' with international labor
productivity standards after trade liberalization
and market deregulation eorts would tend to
be those that were already doing wellÐrelative
to US labor productivity growthÐbefore
structural reforms. We would also expect that
industries that ``lagged behind'' prior to changes in the global incentive regime would tend to
be the same as those that ``lagged behind'' US
labor productivity growth rates after structural
reforms. Consider the case for Argentina. From
the six industries that appear as ``catching up''
sectorsÐISIC 332, 369, 371, 381, 383, and 384
in Table 3Ðfour of them were already doing
quite well in ``catching up'' with the US
productivity standard before trade liberalization and market deregulation eorts. On the
other hand, all four of the ``lagging behind''
industries of 1970±90 continued to be ``failure
stories'' after trade liberalization and market
deregulation eortsÐISIC 314, 324, 331, and
390.
In order to test in a simple way the idea that
industries that did better than average in the
pastÐin terms of closing the relative labor
productivity gap with their US counterpartÐ
tend to be the same as those that did better than
average under the NEM, we have estimated the
simple product moment correlation coecient
between the two columns for each country in
Tables 3 and 5. The resultsÐr 0.80 for
1590
WORLD DEVELOPMENT
Table 5. Relative labor productivity of Latin American industries vis-
a-vis the United States, 1970±90, 27 branches of
industrya
ISIC
311
313
314
321
322
323
324
331
332
341
342
351
352
353
354
355
356
361
362
369
371
372
381
382
383
384
385
390
Food manufacturing
Beverage industries
Tobacco manuf.
Textile manufactures
Wearing and apparel
Leather products
Footwear
Wood
Furniture
Pulp and paper
Printing/publishing
Chemicals
Chemical products
Petroleum re®neries
Petroleum and coal
Rubber products
Plastics
Pottery and china
Glass products
Non-metallic minerals
Iron and steel
Nonferrous metals
Metal products
Non-electrical machinery
Electrical machinery
Transport equipment
Scienti®c instruments
Other industries
Argentina
Brazil
Chile
Colombia
Mexico
0.96
0.87
0.72b
1.40
0.80b
0.78b
0.62b
0.57b
1.88
0.85
1.09
1.78
1.80
0.60b
1.29
1.00
0.60b
0.65b
1.52
1.19
1.70
1.15
1.75
1.47
1.79
1.01
1.78
0.76b
0.91
0.78b
0.36b
1.48
1.06
1.17
1.0
0.65b
1.0
1.20
0.74b
1.12
0.70b
1.44
1.80
0.95
0.86
0.80
1.37
0.85
1.42
1.53
1.08
1.03
1.14
0.83
1.46
0.86
0.72b
0.89
1.41
0.85
0.80
1.12
0.65b
1.45
1.00.
1.61
1.27
3.10b
1.15
3.56b
1.47
0.66b
0.73b
0.25b
2.10b
1.50
1.30
0.68b
1.14
1.00
1.00
1.00
0.81
0.66b
1.19
0.94
1.11
1.48
1.01
1.06
1.63
1.39
0.89
1.50
1.08
1.42
1.01
0.36b
2.84b
1.39
1.40
1.96
1.69
1.69
3.34b
2.10b
1.28
0.79b
1.20
1.44
2.76b
1.47
1.34
1.14
0.69b
1.01
2.44c
2.20b
.93
1.29
1.12
1.40
1.35
1.16
0.79b
0.41b
2.00b
1.21
1.45
2.45b
1.68
1.69
2.17b
2.37b
1.40
0.74b
1.82
1.81
4.35b
2.39b
a
Source: PADI, ECLAC.
P 2:0 ``Catching up'' industries.
c
6 80 ``Falling behind'' industries.
b
Argentina, 0.70 for Brazil, 0.83 for Chile, 0.86
for Colombia and 0.94 for MexicoÐcon®rm
the fact that the ``path dependency'' hypothesis
has strong explanatory power in all ®ve of the
countries. In other words, even if it is indisputable that the market-oriented structural
reforms of the 1990s introduced stronger
competitive discipline into local markets for
manufacturing products, it is somewhat
misleading to argue that said reforms introduced a major discontinuity with respect to the
past when we look at which industries tended
to ``catch up'' and which ones tended to ``fall
behind'' during the ISI period and under the
NEM. On average, they tend to be the same,
regardless of the change in the global incentive
regimen.
We leave here our examination of the ``relative convergence'' question and whether or not
recent market-oriented structural reforms
constitute the explanation of why some countries and industries did better than others
throughout the 1990s. Having presented
statistical evidence in support of the ``path
dependency'' view of the growth process we
now turn to the exploration of other meso and
micro features of the process of structural
change we are hereby trying to describe. The
``entry'' and ``exit'' of ®rms to and from the
market and changes in the pattern of production specialization appear as two dimensions
demanding further exploration if we are to
account correctly for interindustry and intercountry dierences in labor productivity
growth in recent years.
(c) Inter and intraindustry changes in Latin
American manufacturing production during
1970±96
The way in which the structure of industry
responds to changes in the incentive regimeÐ
both to macroeconomic signals as well as to
sector-speci®c institutional and regulatory
STRUCTURAL CHANGE AND LABOR PRODUCTIVITY GROWTH
forcesÐinvolves at least two dierent adjustment mechanisms that need to be dierentiated
from each other and examined separately. On
the one hand, each individual industry must be
seen as going through a complex process of
change in structure and performance as a result
of the ``entry'' and ``exit'' of ®rms. On the other
hand, the relative weight of dierent industries
has to be imagined as changing through time as
a result of changes in the global incentive
regime.
Most attention has so far focused on the
interindustry restructuring process anticipated
in response to the NEM. The devaluation of
the exchange rate, a higher interest rate, lower
taris, etc which belong in the structural
adjustment policy package were clearly expected to aect industries dierently depending
on their market orientation, resource intensity, and import content. In the 1980s,
Latin American import-substituting industries
producing for local markets simultaneously
had to face the contraction of domestic demand
due to macroeconomic stabilization policies,
and the massive arrival of cheaper (and often
better) foreign substitutes. On the other hand,
natural resource-processing industries, catering
for exports, bene®ted from a highly elastic
external demand as well as from the devaluation of the local currency. Under such circumstances we could expect a priori that the former
group of industries would be the one that
suered the most as a result of trade liberalization and market deregulation eorts, while
resource-processing industries and foodstus
producers would bene®t the most from such
measures. Looking at the ®ve countries under
examination we notice that three of themÐ
Argentina, Colombia and ChileЮt very well
with the above a priori expectation. The eect is
somewhat less clear in the case of Mexico, due
to the major expansion of the auto industry and
by the electronic maquiladoras after 1986.
Table 6 illustrates the extent to whichÐwith
the already mentioned exception of MexicoÐ
the production of foodstus and of natural
resource-based industrial commodities has
become central to the pattern of Latin American production specialization. It also shows,
however, that much of the change in that
direction took place before the 1990s,
con®rming the fact that much of what we
presently observe in the region was not exclusively the outcome of recent market-oriented
structural reforms. Longstanding historical
trends seem to underlie the currently observed
1591
pattern of production specialization. The
aggregate weight of raw material-processing
industries and of the foodstu sector increased
signi®cantly over time in all of these countries
except Mexico. On the other hand, the data
also show that the relative weight of the auto
industry has increased in Argentina, Colombia
and Mexico, and maintained its relative share
in Brazil. Together with a clear retrenchment to
natural comparative advantages, the current
pattern of production specialization also seems
to be the result of ad hoc industrial policies (as
in the case of the automobile industry) that
have strongly in¯uenced the pattern of industrial restructuring of Latin America over the
last two decades.
In parallel to the above we have to consider
the issue of intraindustry ``®rm demography,''
which has to do with dierences in ®rmsÕ
adaptive capabilities and strategies within each
industry. Some ®rms are clearly more ``outward-going'' and ``proactive'' in their ``reading'' of the signals emerging from the global
and sector-speci®c environment in which they
operate, while others are more conservative,
``passive'' and ``defensive'' in their reaction.
Moreover, given that we are dealing with situations in which signi®cant market failures
prevail, hindering access to capital and technologyÐparticularly among small and medium
enterprisesÐwe have to accept that some actors
could have had easier access to the resources
they needed for restructuring, while others
probably had greater diculties getting ®nance
and know-how for their technological upgrading. It should not, therefore, come as a surprise
that some ®rms did well and gained ground
within their sector of activity while others
``lagged behind,'' or even disappeared altogether from the marketplace.
Large inter®rm dierences in production
eciency can be observed when we examine
any particular ®eld of manufacturing production. Capital goods and organizational technologies dier a great deal from one ®rm to the
next, and the intraindustry degree of heterogeneity might well be much higher than the one
we would a priori expect on the basis of
conventional price theory, which predicates
similar patterns of reaction from similarly
endowed ®rms. Using British data, for example, Salter cited inter®rm dierences in unit
production costs on the order of 7:1 (Salter,
1960).
In explaining these dierences, Salter
presented a simple ``demographic'' model
1592
Table 6. Structure of industrial value-added in selected Latin American countries, 1970, 1990, 1996a; b
I
II
III+IV
V
a
Brazil
Chile
Colombia
Mexico
1990
1996
1970
1990
1986
1970
1990
1996
1970
1990
1996
1970
1990
1996
15.6
9.9
36.2
38.2
14.3
8.5
46.7
30.5
13.1
12.1
45.7
29.0
18.8
9.9
35.8
35.5
22.9
7.0
39.6
30.5
22.8
8.7
42.4
26.1
14.9
7.7
43.2
34.2
10.1
2.3
55.5
32.0
10.2
2.0
56.2
31.6
10.7
2.9
45.7
40.7
9.6
4.3
51.1
34.9
10.5
6.5
51.2
31.8
13.3
5.5
46.8
34.4
12.3
9.5
46.8
31.4
13.9
10.8
46.5
28.8
Source: PADI, ECLAC.
The groups used here are as follows (with reference numbers added): I. metalworking and engineering intensive industries, except motor vehicles (ISIC 381, 382, 383,
385); II. motor vehicles (ISIC 384); III. food, beverages and tobacco (ISIC 311, 313, 314); IV. natural resource-processing industries (ISIC 341, 351, 354, 355, 356, 371,
372); V. labor-intensive industries (ISIC 321, 322, 323, 324, 331, 332, 342, 352, 361, 362, 369, 39).
b
WORLD DEVELOPMENT
Argentina
1970
STRUCTURAL CHANGE AND LABOR PRODUCTIVITY GROWTH
describing the ``evolutionary'' dynamics of any
given sector of production. New ®rms entering
the market bring with them more ecient
production technologies which allow them to
reduce production costs and market prices. As
a result, their entry triggers changes in market
structure and performance which eventually
force the less ecient ®rms in the sector to leave
the market when they can no longer cover their
variable costs of production out of the market
price.
Although SalterÕs model provides a useful
®rst approximation of an ``evolutionary'' story
of the sort we want here to develop, it is much
too simple for our purposes. The model
assumes that an equilibrium situation prevails
only to be disrupted by the arrival of a superior
new technology that puts in motion a ``steady
state'' demographic transition. Imagine now a
much more turbulent situation in whichÐas a
result of trade liberalization and market deregulationÐchanges take place not just in the
technology of the sector but in many other
variables such as the price of imported substitutes, the rate of interest, the cost of the capital
goods, etc. Each and every one of the above
changes aects the expected rate of return on
investment, altering both risks and opportunities as seen by the individual entrepreneur.
It is to be expected that the response of
dierent ®rms will vary a great deal depending
on how each one perceives ongoing events and
also as a result of their dierent access to
capital and technology markets. Some entrepreneurs will adopt a ``defensive'' attitude and
refrain from taking risks, while others will
respond proactively and seek to expand their
share in the industryÕs output, turning the new
circumstances into their bene®t. Many ®rms
will simply die and exit the market. A small
number might be expected to react ``defensibly''
introducing ``minor'' organizational changes in
their daily routines, trying to cope with stronger competitive threads but without committing themselves to major new investments.
Finally, only a very small minority could be
expected to react ``proactively'' introducing
major changes in their daily routines, or
investing in new production facilities. The
industryÕs overall labor productivity would
improve both as a result of new and more
modern plants entering the market, but also as
a result of ``defensive'' and ``minor'' adaptations on the part of preexisting ®rms, and of the
death of many others that simply could not
cope with the new rules of the game.
1593
Does the empirical evidence support this
kind of a priori thinking? We believe it does.
We cannot, within the scope of the present
paper, discuss at any length dierent industry
studies available to us at ECLAC that clearly
support our ``appreciative story.'' We will,
therefore, limit ourselves to discuss one particular example: the recent evolution of the
Argentine Iron and Steel industry. We have
previously seen that labor productivity in this
sector has improved dramatically in the last
three decades. The Iron and Steel sector is one
of the six industrial sectors in which Argentina
has managed to close down signi®cantly the
relative productivity gap with the United States
during the course of the period hereby under
consideration. 4 Similar cases could be found in
relation to the Chilean pulp and paper sector,
to the Brazilian or Argentine vegetable oil
industry, etc. (Katz, 1999).
Concomitantly with the drop in aggregate
manufacturing investment as a percentage of
GDP, investment in the Iron and Steel industry
increased signi®cantly in Argentina during the
1980s, con®rming the fact that sector-speci®c
forces played an important role in explaining
company behavior in this sector throughout
that decade. Two large ``state of the art''
integrated steel mills came on stream by the
end of the decade. The entry to the market of
these two plants signi®cantly altered the
structure and performance of the sector, forcing a large number of small and medium-size
mills to close down. Most of these ®rms were
actually taken over by the two large domestic
conglomerates that now control the industry.
Furthermore, the privatization of a large stateowned steel millÐSomisaÐalso occurred
during those years as part of the overall
industry restructuring process. One of the
above mentioned domestic conglomerates took
over Somisa, scrapped a large part of its
outmoded equipment, and absorbed the valuable parts of its inventories into its own operation. Table 7 presents the relevant information
on this case.
This example strongly con®rms our previous
argument that sector-speci®c forces might at
certain points in time play a crucial role (even
more relevant than the one played by macroeconomic variables), in explaining company
behavior and industry performance. The interplay of both sets of forces needs to be examined
on a case-by-case basis if we are correctly to
understand why some ®rms invest and others
do not, while some industries ``catch up'' and
1594
WORLD DEVELOPMENT
Table 7. Number of plants in ArgentinaÕs iron and steel industry, 1975±92a
a
1975
1980
1985
1990
1992
Integrated
Semi-integrated
Rollings mills
2
10
47
4
5
38
4
2
36
4
2
33
4
1
21
Industry-wide total
59
47
42
39
26
Source: Azpiazu and Basualdo (1995).
others ``lag behind'' in the process of economic
growth. It is only at the meso and micro level
that the dynamics of growth can be understood, as it involves the co-evolution of macro
and sector-speci®c economic, institutional and
technological forces interacting in complex
manners that cannot be easily perceived if we
proceed exclusively from a macro perspective.
We can now combine the two parts of our
argument in order to advance an ``appreciative'' account of the recent process of industrial
restructuring Latin American countries have
been going through. Both macroeconomic and
sector-speci®c variables have simultaneously
been at work inducing changes in the pattern of
production specialization attained by each one
of the economies in the region as well as the
intraindustry pattern of ®rm survival and
production organization. Some sectors and
®rms have been able to adapt themselves better
than others to changes in the global incentive
regime and in the sector-speci®c institutional
and regulatory environment. The evidence
presented in this paper leads us to believe that
those that did so tend to be industries that were
already doing well, i.e. where accumulated
technological capabilities have been the larger.
In addition, those related to the processing of
natural resources, i.e. closer to natural
comparative advantages, and/or those that
bene®ted the most from ad hoc preferential
treatment on the part of economic authorities,
such as the auto industry. In addition, the
``entry'' and ``exit'' of ®rms to and from the
market has been instrumental in signi®cantly
changing the structure and performance of each
industry as well as the observed degree of
business concentration.
A micro-to-macro approach to these issues
provides considerable insights for the understanding of the highly heterogeneous pattern of
industrial restructuring Latin American countries have been going through over the course of
the last two decades as well as of the dierent
patterns of success and failure that countries and
industries have been experiencing through time.
4. CONCLUSIONS AND SUGGESTIONS
FOR FUTURE RESEARCH
We have argued throughout this paper that
the growth path of any given economy
appears to be determined by the co-evolution
of macro, meso and microeconomic forces
emerging from the economic, institutional and
technological spheres of human interaction.
The way in which recent changes in the global
incentive regime and institutional environment
have aected dierent sectors in the economy and, vice versa, the mechanism by which
sector-speci®c changes are currently in¯uencing the long-term sustainability of recent
structural reforms, is still largely uncharted
territory within the realm of economic
analysis.
The empirical evidence presented in previous
sections of this paper suggests that, throughout
the process of adjustment to the NEM, dierent
sets of forces have been simultaneously at
work. On the one hand, the historically-accumulated production capabilities and degree of
maturity that dierent countries and industries
have managed to attain prior to recent marketoriented structural reforms have undoubtedly
in¯uenced the success or failure of dierent
countries and industries. On the other hand, an
intraindustry ``selection'' process that has
winnowed ®rms has also been at work. Many
®rms have been forced to leave the market,
which now features a much higher rate of
participation of large domestically-owned
conglomerates and of local subsidiaries of
transnational corporations. Numerous SMEs
have been forced to close down, in many cases
not as a result of their long-term ineciency,
but as a consequence of imperfect factor
markets which precluded their access to longterm ®nance and engineering and managerial
know-how. New ®rms entering the market have
brought with them ``world class'' equipment
and product designs, having therefore signi®cantly closed the gap with the worldÕs technological frontier.
STRUCTURAL CHANGE AND LABOR PRODUCTIVITY GROWTH
Finally, the pattern of production specialization has gradually shifted in the direction of
natural comparative advantagesÐraw material
processing industries and labor-intensive
maquiladorasÐas well as in favor of specially
protected sectors such as in the case of the auto
industry.
Many new problems of structure and
behavior have emerged during the course of the
above mentioned industrial restructuring
process. Among them, the following seem to be
worth mentioning as requiring further research:
(a) new forms of structural unemployment
which result from the rapid transition toward
CNC technologies featuring a strong laborsaving bias; (b) chronic long-term imbalances
in the external sector of the economy related
both to a rapidly expanding demand for
imported capital goods and to a pattern of
production specialization which strongly
emphasizes highly volatile markets for indus-
1595
trial commodities; (c) a growing degree of
business concentration deriving from the fact
that many small and middle-size enterprises
have been forced to leave the market, which
now remains in the hands of large domestic
conglomerates and/
Ó 2000 Published by Elsevier Science Ltd.
Printed in Great Britain
0305-750X/00/$ - see front matter
www.elsevier.com/locate/worlddev
PII: S0305-750X(00)00050-4
Structural Change and Labor Productivity Growth in
Latin American Manufacturing Industries 1970±96
JORGE KATZ *
Economic Commission for Latin America and the Caribbean, Santiago, Chile
Summary. Ð This paper examines recent changes in the pattern of production specialization
attained by Latin American manufacturing industries. Comparing Argentina, Brazil, Colombia,
Chile and Mexico to the United States, the paper identi®es ``catching up'' and ``lagging behind''
industries during 1970±96. It considers the evolution of macro and micro forces that aect whether
an industry manages to close down the gap with the international technological frontier. Evidence
is presented showing that recent structural reforms did not result in a major discontinuity with the
past. Rather, ``path dependency'' forces have acted as a major factor explaining dierential
performance among industries after recent changes in the global incentive regime. Ó 2000
Published by Elsevier Science Ltd.
Key words Ð productivity growth, technological gaps, path dependency, structural reforms, Latin
America
1. INTRODUCTION
Recent studies (Katz, Benavente, Crespi &
Stumpo, 1997; Katz & Vera, 1997, 1999)
examining some of the major changes that have
taken place in the Latin American industrial
sector during the last two decades have shown a
major change in the pattern of production
specialization of most economies. There has
been a shift in favor of nontradable sectorsÐ
those producing telecommunication, energy or
®nancial services, for exampleÐas well as
toward natural resource-processing industriesÐproducing iron and steel, petrochemicals, nonferrous minerals, ®shmeal, vegetable
oil, pulp and paper, etc. In addition, electronic
and garment maquiladoras (assembly plants), as
well as automobile producers, have grown quite
rapidly throughout this period in response to
ad hoc industrial policies applied by local
economic authorities.
Large domestically-owned conglomerates, as
well as subsidiaries from transnational corporations (TNCs), have set up very modern,
highly capital-intensive new production plants
in the above mentioned ®elds. Leaving aside the
case of nontradable goods and services, we
notice that many of these new production
facilities have become major exporters of
industrial commodities to highly competitive
world markets. Latin American ®rms act in
such markets as ``price takers,'' with little
bargaining power and attaining low unit pro®t
margins. Given the high degree of volatility
shown by world prices of industrial commodities, the external sector of most Latin American
economies now stands on rather weak foundations which need to be taken into consideration when we examine the long-term
sustainability of the recent restructuring of the
regional production fabric.
In contrast to the above, labor-intensive
industries producing ®nal consumer goods such
as footwear or clothing, and engineering and
R&D-intensive sectors producing capital
goods, ®ne chemicals, 1 or scienti®c instruments, have been shrinking and contracting
(with the exception of the electronic maquila).
Following trade liberalization and market
deregulation eorts the former have had a hard
time competing with low-cost producers from,
* The present paper has been written as part of a larger
research program undertaken with ®nancial support
from the German Technical Cooperation Agency (GTZ)
and the International Cooperation Agency of the
Netherlands. To both these agencies, as well as to
W. Peres and N. Reinhardt for their useful comments on
an earlier version of the paper, the author wishes to
express his gratitude. The views hereby advanced are his
exclusive responsibility.
1583
1584
WORLD DEVELOPMENT
say, China or Vietnam, where wage levels are
just a fraction of those typically paid in Latin
America, while the latter have been unsuccessful in keeping up with the pace of technological
modernization in areas characterized by large
R&D and engineering expenditure, short
product life cycles and high rates of technological obsolescence. As these industries have
continued to incorporate the use of microprocessors, CNC technologies, new biotechnological know how, etc. at a breath-taking pace,
Latin American ®rms have found themselves
lagging further and further behind the worldÕs
technological frontier. The capability of these
®rms to compete in the international marketplace in such ®elds has lagged behind as a result
of these trends.
In many countries of the region this process
of change in the pattern of production
specialization actually began during the late
1970s, i.e. prior to the so-called Debt Crisis,
when the pace of domestic market-led growth
began to slow-down. In othersÐparticularly
ChileÐthe restructuring toward natural
resource-processing industries clearly resulted
from the major transformation in the global
incentive regime which obtained after the military takeover of 1973. It is important therefore
to understand that the change in the pattern of
production specialization we are about to
examine in this paper did not come about in
each country of the region as a result of trade
liberalization and market deregulation eorts.
In many cases, such change was already in the
making well before structural reform eorts
actually began. 2
It is true, however, that in both cases a new
relative price structure, less supportive of
import substitution industrialization activities,
has gained ground during the 1990s. Concomitantly, the ¯ow of direct foreign investment
and the process of capital formation have
accelerated, under the impact of both a more
liberalized and deregulated domestic institutional and economic environment and of a
buoyant international capital market. Property
rights in the area of natural resources have been
strengthened and many institutions prototypical of more developed industrial societiesÐ
such as patent laws granting wider and stronger
intellectual property rightsÐhave been gradually adopted by most Latin American countries.
In short, the regionÕs pattern of production
specialization and its linkages with world
markets are going through a major, and as yet
un®nished, transformation. Central to this
process is a retreat to resource-based processing
industries, to nontradable goods and services,
to labor-intensive electronic and garment
maquiladoras, and to vehicles and transport
equipment receiving special ad hoc treatment
from various governments in the region, notably Mexico, Brazil, Argentina, Colombia,
Venezuela and Chile.
Our purpose here is to examine how the
emerging new industrial structure has
performed in terms of labor productivity
growth vis-
a-vis more developed industrial
economies. Is the Latin American industrial
sector ``catching up'' or ``lagging behind''
relative to the worldÕs technological frontier?
For the purpose of answering this question we
shall use productivity growth in the US
manufacturing sector as our benchmark.
For a study of this sort we should ideally be
using estimates of total factor productivity
growth. Figures for gross capital formation at
the three-digit level of aggregation are
unavailable, however, making it impossible to
calculate such index. We have therefore decided
to perform the comparison on the basis of
labor productivity growth. Using for this
purpose a data base recently put together at
ECLAC we have examined the aggregate
performance of the industrial sector of nine
Latin American countries vis-a-vis
the United
States over 1970±96. We continue with a more
detailed comparison examining labor productivity growth in 27 industries at the three-digit
level of aggregation of the International Standard Industrial Classi®cation of All Economic
Activities (ISIC) for ®ve countriesÐArgentina,
Brazil, Chile, Colombia and MexicoÐfor
1970±96. Special attention is given to the 1990s,
as this is the period in which structural reform
eorts were strengthened.
In Section 2 we brie¯y discuss the analytical
framework upon which the present research is
based. In Section 3 we report our main statistical results. Section 3 also looks at some meso
and micro aspects of recent changes in industrial production structure. In particular, we pay
attention to the entry and exit of ®rms to and
from the market, as well as to changes in the
pattern of production specialization. Section 3
further summarizes our view concerning the
relationship between changes in the macropolicy incentive regime and changes in the
structure and behavior of manufacturing
industry. Micro-to-macro linkages frequently
remain unexplored in the literature on the
STRUCTURAL CHANGE AND LABOR PRODUCTIVITY GROWTH
impact of economic reforms in the region. We
believe this to be a major shortcoming. In our
view, such linkages are crucial for the understanding of the impact of structural reforms. A
few suggestions for further research are
advanced in the ®nal paragraphs of the paper.
2. THE ANALYTICAL FRAMEWORK OF
THE PRESENT RESEARCH
Contemporary studies on economic growth
show at least two dierent ways of approaching
the issues we want to examine here. On the one
hand, received literature oers us ``top-down''
explanations of growth of the type Solow
developed in the 1950s and 1960s, which have
recently reemerged within the framework of
modern growth theory in the work of Barro
and Sala-i-Martin (1995), Romer (1992), and
others. These studies are based on the conventional microfoundations of the ``representative
®rm'' and on neoclassical ``growth accounting''
principles that rely on assumptions of perfect
information, ``generic'' production functions,
``well-behaved'' ®rms, equilibrium growth
paths, perfectly competitive markets, and
production factors being paid their marginal
productivity.
The explanation of why productivity growth
is attained in such a stylized analytical framework is extremely simple. Historical and institutional factors virtually play no role at all. It is
precisely such degree of simpli®cation of reality
and the elimination of market imperfections,
uncertainty, etc. that allows us to have aggregate production, investment, consumption and
saving functions on the basis of which to look
at the behavior of the global economy.
Contemporary growth theory has in recent
years incorporated increasing returns to scale
and externalities speci®ed at the industry level,
i.e. outside the realm of any one speci®c ®rm.
This allows the analyst to maintain the
assumption of competitive behavior at the level
of the individual production unit, and therefore
to save the microfoundations of conventional
price theory, even while increasing returns and
externalities are being incorporated into our
formal thinking about the growth process. But
as Nelson has recently pointed out, modern
growth theory still faces major diculties
bringing on board both the role of institutions
and of production organization, as both such
topics still appear dicult to tackle through
formal mathematical modeling. (Nelson, 1997).
1585
Over the course of the last two decades an
increasing number of economists have
approached the study of innovation and
productivity growth from an ``evolutionary''
perspective (e.g., Nelson & Winter, 1982; Dosi
et al., 1988; Freeman, 1994; Metcalfe, 1997).
These authors see productivity growth as the
outcome of a process of natural selection
among ®rms, rather than as the result of
conventional maximizing behavior. The success
and failure of dierent ®rms, imperfect information, uncertainty, trial and error are all
central features of a screening mechanism
which involves both quasi-genetic ``mutations''
(Nelson, 1995) and historical, cultural and
institutional forces conditioning the economyÕs
growth path over time. Here, growth is not the
outcome of equilibrium behavior but is instead
a ``cultural construct'' deeply embedded in the
technological and institutional foundations of
society. The growth story attains a historical
and institutional atmosphere that is simply
absent from the neoclassical account of the
facts. Whereas the latter is based on conventional price theory, the ``evolutionary'' story
has a large component of ``social anthropology'' that goes well beyond the scope of the
professional economistÕs formal tool box.
The evolutionary account of why labor
productivity growth takes place in any given
environment considers competition as a ``selection'' mechanism winnowing ®rms, as well as
inter®rm dierences in structure, strategy and
accumulated technological capabilities. It also
considers the gradual transformation (through
learning processes) of the institutional and
regulatory environment in which dierent
countries and industries operate through time.
Historical and institutional variables play a
central role in¯uencing what individual agents
want, know and actually do, whereas in the
neoclassical model they always know ex ante all
that needs to be known, and do exactly all that
has to be done, in order to maximize bene®ts.
In the neoclassical model exogenous forces
induce robotic-type responses on the part of
®rms and consumers. Information is perfectly
accessible to them and production functions are
of a generic sort. In the evolutionary model,
behavior is ``adaptive,'' i.e., based on trial and
error, while information has to be gradually
attained through learning processes, via the
accumulation of experience. Production functions are improved upon as a result of engineering and R&D activities. Some companies
are better than others at producing incremental
1586
WORLD DEVELOPMENT
units of technological and organizational
knowledge and therefore do better through
time. Market structure and performance
change endogenously re¯ecting dierential
processes of knowledge accumulation across
®rms.
Our research draws a great deal on recent
``evolutionary'' thinking. Macro, meso and
micro variables, on the one hand, and the coevolution of economic, institutional, and technological forces, on the other, seem to us to be
of crucial importance for the understanding of
the impact trade liberalization and market
deregulation eorts are having today in Latin
America. We need to take into account industry-speci®c variables and institutions in order
to understand how each particular sector reacts
to changes in macroeconomic and regulatory
signals. Moreover, we think that a micro-tomacro feedback mechanism (loop) needs to be
incorporated in our growth models, re¯ecting
the way in which the ``new'' features of
the production structure in¯uenceÐthrough
exports and investmentÐthe long-term sustainability of the new pattern of production
specialization attained by the economy after
structural reforms. The profession has not so far
adequately explored many of these micro-tomacro relations.
3. LABOR PRODUCTIVITY GROWTH IN
LATIN AMERICAN MANUFACTURING
INDUSTRY DURING 1970±96
Using a database recently prepared at
ECLACÕs division of Production, Productivity
and Management covering 27 industries at the
ISIC three-digit level of aggregation, we have
estimated labor productivity growth (measured
as value-added per capita) for Argentina,
Brazil, Colombia, Chile, Costa Rica, Jamaica,
Mexico, Peru, Uruguay for 1970±96. Similar
data have been put together for the US manufacturing sector in order to serve as a benchmark for comparison in order to identify
``catching up'' and ``lagging behind'' countries
and industries. While using US data might not
always be our best possible point of reference, it
nevertheless establishes an interesting and
accessible benchmark for comparison.
We begin by looking at the aggregate picture
for each one of the above mentioned countries.
We then proceed to the three-digit level of
aggregation, in order to attain a more detailed
picture of the relative performance of dierent
branches of manufacturing production in the
region vis-a-vis
their US counterpart.
(a) Labor productivity growth in manufacturing
production in Latin America and the US
Our results in Table 1 show that, as far as
manufacturing production is concerned, only
three Latin American countriesÐArgentina,
Colombia and MexicoÐhave managed to
attain a rate of growth of labor productivity
over the whole 1970±96 period higher than the
one attained by the US industrial sector. In
other words, only these three countries of the
region have managed partially to reduce the
relative labor productivity gap with the United
States during the nearly three decades covered
by our information.
Table 1. Output, employment and labor productivity growth in Latin American and US manufacturing production in
1970±90 and 1990±96a
Manufacturing output
Argentina
Brazil
Chile
Colombia
Costa Ricab
Jamaicab
Mexico
Peru
Uruguay
United
States
a
Employment
Labor productivity
1970±96
1990±96
1970±96
1990±96
1970±96
1990±96
1.18
2.81
2.76
3.98
4.39
0.11
3.79
1.17
0.61
2.39
4.87
2.26
6.40
3.52
n.a.
n.a.
2.27
5.09
)1.46
5.04
)2.62
0.95
1.51
1.24
4.83
1.66
0.91
2.85
0.37
0.35
)3.15
)6.41
3.49
)0.22
n.a.
n.a.
)0.03
1.97
)8.58
0.30
3.80
1.86
1.25
2.74
)0.44
)1.55
2.88
)1.68
0.24
2.04
8.02
8.67
2.91
3.74
n.a
n.a
2.30
3.12
7.12
4.74
Source: Industrial Performance Analysis Program (PADI), ECLAC Division of Production Productivity and
Management.
b
Data until 1992.
STRUCTURAL CHANGE AND LABOR PRODUCTIVITY GROWTH
The data also show that the rate of growth of
labor productivity in manufacturing production increased quite signi®cantly during the
1990s in all of the countries under consideration. This is particularly the case for Argentina, Brazil and Uruguay, where the
acceleration of labor productivity growth in
manufacturing production during the 1990s
was especially strong. We notice, however, that
in all three countries the faster rate of labor
productivity growth of the 1990s is more the
re¯ection of a very high rate of labor
displacement from manufacturing industry
than of a high rate of expansion of industrial
production per se. In fact, manufacturing
growth was not particularly strong in most of
the countries under examination throughout
the 1990s. In other words, the pattern we are
trying to describe here does not seem to be one
of rapid rates of productivity growth deriving
from a major outward expansion of output, but
rather of major ``labor saving'' restructuring
eorts carried out by an industrial sector that is
not growing very rapidly through time. 3
Table 2 presents the ratio of labor productivity in manufacturing production in each of
the countries under examination and in the
United States, for 1970, 1980, 1990 and 1996.
None of the Latin American countries had
attained more than about one-third of US
labor productivity in 1970. By the end of the
period under consideration, only three countriesÐArgentina, Colombia and MexicoÐexhibit signs of having partially reduced the labor
productivity gap with the United States. Interestingly, much of the (small) improvement
attained by Colombia and Mexico came about
during the two decades prior to trade liberal-
Table 2. Relative labor productivity in Latin American
manufacturing production vis-
a-vis the United Statesa
a
Country
1970
1980
1990
1996
Argentina
Brazil
Chileb
Colombia
Costa Rica
Jamaicac
Mexicod
Peru
Uruguaya
0.42
0.28
0.25
0.29
)
0.26
0.32
0.33
0.35
0.41
0.26
0.24
0.25
)
0.16
0.30
0.25
0.22
0.55
0.29
0.23
0.37
0.15
0.16
0.44
0.16
0.20
0.67
0.37
0.20
0.34
0.14
0.13
0.38
0.15
0.22
Source: PADI, ECLAC.
Data until 1995.
c
Data until 1992.
d
Data until 1994.
b
1587
ization and market deregulation eorts, i.e. in
1970±90. In the case of Argentina, about half of
the relative improvement in labor productivity
vis-a-vis
the United States took place during
1970±90, i.e. prior to structural reforms, while
the other half was obtained during the 1990s,
i.e. concomitantly with trade liberalization and
market deregulation eorts. Our ®gures also
indicate that aggregate manufacturing labor
productivity increased quite signi®cantly in
Argentina during 1990±96 but that this was not
the case for Mexico and Colombia. Brazil, on
the other hand, does not show signi®cant
improvement in labor productivity vis-
a-vis the
United States during 1970±90, but achieved a
major relative improvement during 1990±96.
This improvement appears to be basically the
outcome of a strong labor displacement
phenomena.
In the cases of Chile (in spite of the fact that
labor productivity has increased signi®cantly
since the opening up and deregulation of the
economy) and Costa Rica, no signs of relative
improvement can be detected over time.
Finally, the relative positions of Peru, Uruguay
and Jamaica have clearly deteriorated.
(b) Labor productivity growth in Latin
American and US industries during 1970±96
For a more detailed level of analysis, we have
estimated labor productivity growth for 27
dierent branches of manufacturing production
for Argentina, Brazil, Chile, Colombia and
Mexico for 1970±96. We have then estimated
through a simple logarithmic regression the
long-term trend of each series. Dividing the
coecient thus obtained by a similarly calculated coecient for US industries covering the
same period of time gives us an index of relative
labor productivity. The results are presented in
Table 3.
Table 3 clearly shows Latin American
industries ``catching up'' or ``falling behind''
the US labor productivity standard. An index
larger than one indicates that the industry of
reference has managed to grow faster in terms
of labor productivity than its US counterpart
throughout the period, whereas an index
smaller than one depicts the opposite situation.
Given the fact that labor productivity in most
of these industries was not more than 30% or
40% of US labor productivity in our base
year (1970), a large indexÐsay, around twoÐ
indicates that the gap was signi®cantly
reduced, even though, in absolute terms, labor
1588
WORLD DEVELOPMENT
Table 3. Relative labor productivity in Latin American manufacturing production vis-
a-vis the United States, 1970±96,
27 branches of industrya
ISIC
311
313
314
321
322
323
324
331
332
341
342
351
352
353
354
355
356
361
362
369
371
372
381
382
383
384
385
390
Category
Food manufacturing
Beverage industries
Tobacco manuf.
Textiles
Wearing and apparel
Leather products
Footwear
Wood products
Furniture
Pulp and paper
Printing/publishing
Chemicals
Chemical products
Petroleum re®neries
Petroleum and coal
Rubber products
Plastics
Pottery and china
Glass products
Non-metallic minerals
Iron and steel
Nonferrous metals
Metal products
Non-electrical mach
Electrical machinery
Transport equipment
Scienti®c instruments
Other industries
Argentina
1.10
1.04
0.74e
1.67
1.17
1.38
0.78e
0.55e
2.69f
0.99
1.21
1.92
1.98
1.22
1.85
1.55
0.81
1.33
1.91
2.35f
2.54f
1.28
2.07f
1.91
2.68f
2.00f
1.29
0.52e
Brazilb
1.14
0.72e
0.21e
1.43
1.20
0.93
1.13
0.87
1.40
1.26
0.86
1.18
0.60e
1.57
2.07f
2.55f
1.25
1.20
1.92
1.28
1.97
2.50f
1.78
1.12
1.97
1.33
1.48
0.76e
Chilec
e
0.67
0.91
0.76e
0.77e
0.75e
0.45e
0.65e
0.97
1.13
1.10
1.43
1.79
0.97
3.35f
2.14f
0.41e
0.51e
0.45e
1.67
1.68
1.33
0.43e
1.22
1.31
0.94
0.76e
1.22
0.92
Colombia
Mexicod
0.93
0.79e
0.28e
1.23
1.30
0.58e
1.03
0.94
0.85
1.12
0.89
1.09
0.86
0.28e
2.10f
1.36
1.50
2.24f
1.57
1.36
2.82f
1.92
1.79
0.75e
0.99
2.07f
3.27f
1.26
1.21
0.83
0.38e
0.75e
1.85
0.97
0.72e
0.94
0.96
1.03
1.03
0.88
0.58e
0.30e
1.55
1.24
1.25
2.47f
1.60
1.39
1.54
2.39f
1.39
0.72e
1.76
1.81
3.81f
2.19f
a
Source: PADI, ECLAC.
1996/1970.
c
1995/1970.
d
1994/1970.
e
6 0:80 ``falling behind'' industries.
f
P 2:0 ``catching up'' industries.
b
productivity could continue to be below the US
level in 1996. These industries are considered to
be ``catching up'' to the US standard. On the
other hand, a number, say, equal to or smaller
than 0.80 suggests that the relative gap between
the two has expanded signi®cantly through
time, and the industry is ``falling behind.''
Six dierent branches of industryÐmarked
with ``f'' in the table for identi®cationÐappear
in Argentina as ``catching up'' with the US
standard during 1970±96. Five industries
attained a similarly ``successful'' performance
in Colombia, four in Mexico, three in Brazil
and two in Chile. On the other hand, four
sectorsÐmarked ``e'' in Table 3Ðshow clear
signs of ``falling behind'' the US standard: in
Argentina and Brazil, ®ve in Colombia, six in
Mexico and 11 in Chile. Argentine industries
have done the best and Chilean ones the worst
in ``catching up'' with the US standard. Brazil,
Colombia and Mexico fall somewhere in
between these two extremes.
What explains the dierences in the productivity performance of individual industries
within each country? One hypothesis is that the
fastest-growing industries were the ones that
did better in terms of reducing the original
labor productivity gap with their US counterpart, while the ``lagging behind'' industries
tended to be those that attained a worse than
average growth performance. In order to test
this hypothesis, we estimated in a cross-section
regression the ordinary least squares (OLS)
relationship between the coecients reported in
Table 3 and the observed rate of growth of each
industry in each country throughout the period
under consideration. The results, shown in
Table 4, tend to con®rm our a priori expecta-
STRUCTURAL CHANGE AND LABOR PRODUCTIVITY GROWTH
1589
Table 4. Crossindustry regressions between relative productivity growth and the rate of growth of output, 1970±96
Argentina
Brazil
Chile
Colombia
Mexico
Regression coecient
S.E.
R2
F-test
Degrees of freedom
0.77
1.13
0.68
0.89
1.17
0.13
0.06
0.14
0.25
0.21
0.59
0.95
0.55
0.33
0.57
33.56
419.05
22.39
12.59
31.97
23
22
20
25
24
tion that the interindustry variance in the rate
of growth constitutes an important part of the
explanation of why some sectors ``catch up''
with labor productivity levels in the United
States, while other sectors ``lag behind'' in
relative terms.
An adequate explanation of interindustry
dierences in the degree of productivity
convergence must therefore include an explanation of why some sectors grow faster than
others through time. This could be the consequence of both macro and sector-speci®c
forces. On the one hand, we could a priori
expect macro forces to play a signi®cant role
inducing some sectors in the economy to grow
faster than others. Indeed, the NEM is predicated on an expectation that trade liberalization will induce a shift in resource allocation
along comparative advantage lines, i.e. toward
labor intensive and/or natural resource-intensive industries. On the other hand, sector
speci®c forcesÐthe ``localized'' institutional
and regulatory environment in which each
particular industry operates, the ad hoc treatment each sector manages to get from local
economic authorities, etc.Ðcould also be
expected to in¯uence interindustry dierences
in growth performance. Thus, both macro and
sector-speci®c variables may in¯uence why
some sectors ``catch up'' and others ``fall
behind'' in the process of economic growth.
We have previously noted that the aggregate
productivity performance of dierent countries
in the region vis-
a-vis the United States does
not appear to be exclusively determined by
recent trade liberalization and market deregulation eorts. The data previously presented
show that part of the ``catching up'' attained by
the industrial sector of Argentina, Colombia
and Mexico during the period under consideration was already ``in the making'' during
1970±90 and that it would be somewhat
misleading to assign it exclusively to recent
market-oriented reforms. Do the three-digit
data con®rm such a picture? In order to throw
light upon this question, we have calculated the
same index of relative industry productivity
performance for 1970±90. Our results are
presented in Table 5.
By comparing Tables 3 and 5 we con®rm our
previous ®nding, i.e. that in many cases the
``catching up'' or the ``falling behind'' of certain
industries was already ``in the cards'' before
trade liberalization and market deregulation
measures were actually introduced by the
various governments in the region. ``Path
dependency'' appears to be a strong structural
feature of production systems, which induces us
a priori to expect that industries that did well in
terms of ``catching up'' with international labor
productivity standards after trade liberalization
and market deregulation eorts would tend to
be those that were already doing wellÐrelative
to US labor productivity growthÐbefore
structural reforms. We would also expect that
industries that ``lagged behind'' prior to changes in the global incentive regime would tend to
be the same as those that ``lagged behind'' US
labor productivity growth rates after structural
reforms. Consider the case for Argentina. From
the six industries that appear as ``catching up''
sectorsÐISIC 332, 369, 371, 381, 383, and 384
in Table 3Ðfour of them were already doing
quite well in ``catching up'' with the US
productivity standard before trade liberalization and market deregulation eorts. On the
other hand, all four of the ``lagging behind''
industries of 1970±90 continued to be ``failure
stories'' after trade liberalization and market
deregulation eortsÐISIC 314, 324, 331, and
390.
In order to test in a simple way the idea that
industries that did better than average in the
pastÐin terms of closing the relative labor
productivity gap with their US counterpartÐ
tend to be the same as those that did better than
average under the NEM, we have estimated the
simple product moment correlation coecient
between the two columns for each country in
Tables 3 and 5. The resultsÐr 0.80 for
1590
WORLD DEVELOPMENT
Table 5. Relative labor productivity of Latin American industries vis-
a-vis the United States, 1970±90, 27 branches of
industrya
ISIC
311
313
314
321
322
323
324
331
332
341
342
351
352
353
354
355
356
361
362
369
371
372
381
382
383
384
385
390
Food manufacturing
Beverage industries
Tobacco manuf.
Textile manufactures
Wearing and apparel
Leather products
Footwear
Wood
Furniture
Pulp and paper
Printing/publishing
Chemicals
Chemical products
Petroleum re®neries
Petroleum and coal
Rubber products
Plastics
Pottery and china
Glass products
Non-metallic minerals
Iron and steel
Nonferrous metals
Metal products
Non-electrical machinery
Electrical machinery
Transport equipment
Scienti®c instruments
Other industries
Argentina
Brazil
Chile
Colombia
Mexico
0.96
0.87
0.72b
1.40
0.80b
0.78b
0.62b
0.57b
1.88
0.85
1.09
1.78
1.80
0.60b
1.29
1.00
0.60b
0.65b
1.52
1.19
1.70
1.15
1.75
1.47
1.79
1.01
1.78
0.76b
0.91
0.78b
0.36b
1.48
1.06
1.17
1.0
0.65b
1.0
1.20
0.74b
1.12
0.70b
1.44
1.80
0.95
0.86
0.80
1.37
0.85
1.42
1.53
1.08
1.03
1.14
0.83
1.46
0.86
0.72b
0.89
1.41
0.85
0.80
1.12
0.65b
1.45
1.00.
1.61
1.27
3.10b
1.15
3.56b
1.47
0.66b
0.73b
0.25b
2.10b
1.50
1.30
0.68b
1.14
1.00
1.00
1.00
0.81
0.66b
1.19
0.94
1.11
1.48
1.01
1.06
1.63
1.39
0.89
1.50
1.08
1.42
1.01
0.36b
2.84b
1.39
1.40
1.96
1.69
1.69
3.34b
2.10b
1.28
0.79b
1.20
1.44
2.76b
1.47
1.34
1.14
0.69b
1.01
2.44c
2.20b
.93
1.29
1.12
1.40
1.35
1.16
0.79b
0.41b
2.00b
1.21
1.45
2.45b
1.68
1.69
2.17b
2.37b
1.40
0.74b
1.82
1.81
4.35b
2.39b
a
Source: PADI, ECLAC.
P 2:0 ``Catching up'' industries.
c
6 80 ``Falling behind'' industries.
b
Argentina, 0.70 for Brazil, 0.83 for Chile, 0.86
for Colombia and 0.94 for MexicoÐcon®rm
the fact that the ``path dependency'' hypothesis
has strong explanatory power in all ®ve of the
countries. In other words, even if it is indisputable that the market-oriented structural
reforms of the 1990s introduced stronger
competitive discipline into local markets for
manufacturing products, it is somewhat
misleading to argue that said reforms introduced a major discontinuity with respect to the
past when we look at which industries tended
to ``catch up'' and which ones tended to ``fall
behind'' during the ISI period and under the
NEM. On average, they tend to be the same,
regardless of the change in the global incentive
regimen.
We leave here our examination of the ``relative convergence'' question and whether or not
recent market-oriented structural reforms
constitute the explanation of why some countries and industries did better than others
throughout the 1990s. Having presented
statistical evidence in support of the ``path
dependency'' view of the growth process we
now turn to the exploration of other meso and
micro features of the process of structural
change we are hereby trying to describe. The
``entry'' and ``exit'' of ®rms to and from the
market and changes in the pattern of production specialization appear as two dimensions
demanding further exploration if we are to
account correctly for interindustry and intercountry dierences in labor productivity
growth in recent years.
(c) Inter and intraindustry changes in Latin
American manufacturing production during
1970±96
The way in which the structure of industry
responds to changes in the incentive regimeÐ
both to macroeconomic signals as well as to
sector-speci®c institutional and regulatory
STRUCTURAL CHANGE AND LABOR PRODUCTIVITY GROWTH
forcesÐinvolves at least two dierent adjustment mechanisms that need to be dierentiated
from each other and examined separately. On
the one hand, each individual industry must be
seen as going through a complex process of
change in structure and performance as a result
of the ``entry'' and ``exit'' of ®rms. On the other
hand, the relative weight of dierent industries
has to be imagined as changing through time as
a result of changes in the global incentive
regime.
Most attention has so far focused on the
interindustry restructuring process anticipated
in response to the NEM. The devaluation of
the exchange rate, a higher interest rate, lower
taris, etc which belong in the structural
adjustment policy package were clearly expected to aect industries dierently depending
on their market orientation, resource intensity, and import content. In the 1980s,
Latin American import-substituting industries
producing for local markets simultaneously
had to face the contraction of domestic demand
due to macroeconomic stabilization policies,
and the massive arrival of cheaper (and often
better) foreign substitutes. On the other hand,
natural resource-processing industries, catering
for exports, bene®ted from a highly elastic
external demand as well as from the devaluation of the local currency. Under such circumstances we could expect a priori that the former
group of industries would be the one that
suered the most as a result of trade liberalization and market deregulation eorts, while
resource-processing industries and foodstus
producers would bene®t the most from such
measures. Looking at the ®ve countries under
examination we notice that three of themÐ
Argentina, Colombia and ChileЮt very well
with the above a priori expectation. The eect is
somewhat less clear in the case of Mexico, due
to the major expansion of the auto industry and
by the electronic maquiladoras after 1986.
Table 6 illustrates the extent to whichÐwith
the already mentioned exception of MexicoÐ
the production of foodstus and of natural
resource-based industrial commodities has
become central to the pattern of Latin American production specialization. It also shows,
however, that much of the change in that
direction took place before the 1990s,
con®rming the fact that much of what we
presently observe in the region was not exclusively the outcome of recent market-oriented
structural reforms. Longstanding historical
trends seem to underlie the currently observed
1591
pattern of production specialization. The
aggregate weight of raw material-processing
industries and of the foodstu sector increased
signi®cantly over time in all of these countries
except Mexico. On the other hand, the data
also show that the relative weight of the auto
industry has increased in Argentina, Colombia
and Mexico, and maintained its relative share
in Brazil. Together with a clear retrenchment to
natural comparative advantages, the current
pattern of production specialization also seems
to be the result of ad hoc industrial policies (as
in the case of the automobile industry) that
have strongly in¯uenced the pattern of industrial restructuring of Latin America over the
last two decades.
In parallel to the above we have to consider
the issue of intraindustry ``®rm demography,''
which has to do with dierences in ®rmsÕ
adaptive capabilities and strategies within each
industry. Some ®rms are clearly more ``outward-going'' and ``proactive'' in their ``reading'' of the signals emerging from the global
and sector-speci®c environment in which they
operate, while others are more conservative,
``passive'' and ``defensive'' in their reaction.
Moreover, given that we are dealing with situations in which signi®cant market failures
prevail, hindering access to capital and technologyÐparticularly among small and medium
enterprisesÐwe have to accept that some actors
could have had easier access to the resources
they needed for restructuring, while others
probably had greater diculties getting ®nance
and know-how for their technological upgrading. It should not, therefore, come as a surprise
that some ®rms did well and gained ground
within their sector of activity while others
``lagged behind,'' or even disappeared altogether from the marketplace.
Large inter®rm dierences in production
eciency can be observed when we examine
any particular ®eld of manufacturing production. Capital goods and organizational technologies dier a great deal from one ®rm to the
next, and the intraindustry degree of heterogeneity might well be much higher than the one
we would a priori expect on the basis of
conventional price theory, which predicates
similar patterns of reaction from similarly
endowed ®rms. Using British data, for example, Salter cited inter®rm dierences in unit
production costs on the order of 7:1 (Salter,
1960).
In explaining these dierences, Salter
presented a simple ``demographic'' model
1592
Table 6. Structure of industrial value-added in selected Latin American countries, 1970, 1990, 1996a; b
I
II
III+IV
V
a
Brazil
Chile
Colombia
Mexico
1990
1996
1970
1990
1986
1970
1990
1996
1970
1990
1996
1970
1990
1996
15.6
9.9
36.2
38.2
14.3
8.5
46.7
30.5
13.1
12.1
45.7
29.0
18.8
9.9
35.8
35.5
22.9
7.0
39.6
30.5
22.8
8.7
42.4
26.1
14.9
7.7
43.2
34.2
10.1
2.3
55.5
32.0
10.2
2.0
56.2
31.6
10.7
2.9
45.7
40.7
9.6
4.3
51.1
34.9
10.5
6.5
51.2
31.8
13.3
5.5
46.8
34.4
12.3
9.5
46.8
31.4
13.9
10.8
46.5
28.8
Source: PADI, ECLAC.
The groups used here are as follows (with reference numbers added): I. metalworking and engineering intensive industries, except motor vehicles (ISIC 381, 382, 383,
385); II. motor vehicles (ISIC 384); III. food, beverages and tobacco (ISIC 311, 313, 314); IV. natural resource-processing industries (ISIC 341, 351, 354, 355, 356, 371,
372); V. labor-intensive industries (ISIC 321, 322, 323, 324, 331, 332, 342, 352, 361, 362, 369, 39).
b
WORLD DEVELOPMENT
Argentina
1970
STRUCTURAL CHANGE AND LABOR PRODUCTIVITY GROWTH
describing the ``evolutionary'' dynamics of any
given sector of production. New ®rms entering
the market bring with them more ecient
production technologies which allow them to
reduce production costs and market prices. As
a result, their entry triggers changes in market
structure and performance which eventually
force the less ecient ®rms in the sector to leave
the market when they can no longer cover their
variable costs of production out of the market
price.
Although SalterÕs model provides a useful
®rst approximation of an ``evolutionary'' story
of the sort we want here to develop, it is much
too simple for our purposes. The model
assumes that an equilibrium situation prevails
only to be disrupted by the arrival of a superior
new technology that puts in motion a ``steady
state'' demographic transition. Imagine now a
much more turbulent situation in whichÐas a
result of trade liberalization and market deregulationÐchanges take place not just in the
technology of the sector but in many other
variables such as the price of imported substitutes, the rate of interest, the cost of the capital
goods, etc. Each and every one of the above
changes aects the expected rate of return on
investment, altering both risks and opportunities as seen by the individual entrepreneur.
It is to be expected that the response of
dierent ®rms will vary a great deal depending
on how each one perceives ongoing events and
also as a result of their dierent access to
capital and technology markets. Some entrepreneurs will adopt a ``defensive'' attitude and
refrain from taking risks, while others will
respond proactively and seek to expand their
share in the industryÕs output, turning the new
circumstances into their bene®t. Many ®rms
will simply die and exit the market. A small
number might be expected to react ``defensibly''
introducing ``minor'' organizational changes in
their daily routines, trying to cope with stronger competitive threads but without committing themselves to major new investments.
Finally, only a very small minority could be
expected to react ``proactively'' introducing
major changes in their daily routines, or
investing in new production facilities. The
industryÕs overall labor productivity would
improve both as a result of new and more
modern plants entering the market, but also as
a result of ``defensive'' and ``minor'' adaptations on the part of preexisting ®rms, and of the
death of many others that simply could not
cope with the new rules of the game.
1593
Does the empirical evidence support this
kind of a priori thinking? We believe it does.
We cannot, within the scope of the present
paper, discuss at any length dierent industry
studies available to us at ECLAC that clearly
support our ``appreciative story.'' We will,
therefore, limit ourselves to discuss one particular example: the recent evolution of the
Argentine Iron and Steel industry. We have
previously seen that labor productivity in this
sector has improved dramatically in the last
three decades. The Iron and Steel sector is one
of the six industrial sectors in which Argentina
has managed to close down signi®cantly the
relative productivity gap with the United States
during the course of the period hereby under
consideration. 4 Similar cases could be found in
relation to the Chilean pulp and paper sector,
to the Brazilian or Argentine vegetable oil
industry, etc. (Katz, 1999).
Concomitantly with the drop in aggregate
manufacturing investment as a percentage of
GDP, investment in the Iron and Steel industry
increased signi®cantly in Argentina during the
1980s, con®rming the fact that sector-speci®c
forces played an important role in explaining
company behavior in this sector throughout
that decade. Two large ``state of the art''
integrated steel mills came on stream by the
end of the decade. The entry to the market of
these two plants signi®cantly altered the
structure and performance of the sector, forcing a large number of small and medium-size
mills to close down. Most of these ®rms were
actually taken over by the two large domestic
conglomerates that now control the industry.
Furthermore, the privatization of a large stateowned steel millÐSomisaÐalso occurred
during those years as part of the overall
industry restructuring process. One of the
above mentioned domestic conglomerates took
over Somisa, scrapped a large part of its
outmoded equipment, and absorbed the valuable parts of its inventories into its own operation. Table 7 presents the relevant information
on this case.
This example strongly con®rms our previous
argument that sector-speci®c forces might at
certain points in time play a crucial role (even
more relevant than the one played by macroeconomic variables), in explaining company
behavior and industry performance. The interplay of both sets of forces needs to be examined
on a case-by-case basis if we are correctly to
understand why some ®rms invest and others
do not, while some industries ``catch up'' and
1594
WORLD DEVELOPMENT
Table 7. Number of plants in ArgentinaÕs iron and steel industry, 1975±92a
a
1975
1980
1985
1990
1992
Integrated
Semi-integrated
Rollings mills
2
10
47
4
5
38
4
2
36
4
2
33
4
1
21
Industry-wide total
59
47
42
39
26
Source: Azpiazu and Basualdo (1995).
others ``lag behind'' in the process of economic
growth. It is only at the meso and micro level
that the dynamics of growth can be understood, as it involves the co-evolution of macro
and sector-speci®c economic, institutional and
technological forces interacting in complex
manners that cannot be easily perceived if we
proceed exclusively from a macro perspective.
We can now combine the two parts of our
argument in order to advance an ``appreciative'' account of the recent process of industrial
restructuring Latin American countries have
been going through. Both macroeconomic and
sector-speci®c variables have simultaneously
been at work inducing changes in the pattern of
production specialization attained by each one
of the economies in the region as well as the
intraindustry pattern of ®rm survival and
production organization. Some sectors and
®rms have been able to adapt themselves better
than others to changes in the global incentive
regime and in the sector-speci®c institutional
and regulatory environment. The evidence
presented in this paper leads us to believe that
those that did so tend to be industries that were
already doing well, i.e. where accumulated
technological capabilities have been the larger.
In addition, those related to the processing of
natural resources, i.e. closer to natural
comparative advantages, and/or those that
bene®ted the most from ad hoc preferential
treatment on the part of economic authorities,
such as the auto industry. In addition, the
``entry'' and ``exit'' of ®rms to and from the
market has been instrumental in signi®cantly
changing the structure and performance of each
industry as well as the observed degree of
business concentration.
A micro-to-macro approach to these issues
provides considerable insights for the understanding of the highly heterogeneous pattern of
industrial restructuring Latin American countries have been going through over the course of
the last two decades as well as of the dierent
patterns of success and failure that countries and
industries have been experiencing through time.
4. CONCLUSIONS AND SUGGESTIONS
FOR FUTURE RESEARCH
We have argued throughout this paper that
the growth path of any given economy
appears to be determined by the co-evolution
of macro, meso and microeconomic forces
emerging from the economic, institutional and
technological spheres of human interaction.
The way in which recent changes in the global
incentive regime and institutional environment
have aected dierent sectors in the economy and, vice versa, the mechanism by which
sector-speci®c changes are currently in¯uencing the long-term sustainability of recent
structural reforms, is still largely uncharted
territory within the realm of economic
analysis.
The empirical evidence presented in previous
sections of this paper suggests that, throughout
the process of adjustment to the NEM, dierent
sets of forces have been simultaneously at
work. On the one hand, the historically-accumulated production capabilities and degree of
maturity that dierent countries and industries
have managed to attain prior to recent marketoriented structural reforms have undoubtedly
in¯uenced the success or failure of dierent
countries and industries. On the other hand, an
intraindustry ``selection'' process that has
winnowed ®rms has also been at work. Many
®rms have been forced to leave the market,
which now features a much higher rate of
participation of large domestically-owned
conglomerates and of local subsidiaries of
transnational corporations. Numerous SMEs
have been forced to close down, in many cases
not as a result of their long-term ineciency,
but as a consequence of imperfect factor
markets which precluded their access to longterm ®nance and engineering and managerial
know-how. New ®rms entering the market have
brought with them ``world class'' equipment
and product designs, having therefore signi®cantly closed the gap with the worldÕs technological frontier.
STRUCTURAL CHANGE AND LABOR PRODUCTIVITY GROWTH
Finally, the pattern of production specialization has gradually shifted in the direction of
natural comparative advantagesÐraw material
processing industries and labor-intensive
maquiladorasÐas well as in favor of specially
protected sectors such as in the case of the auto
industry.
Many new problems of structure and
behavior have emerged during the course of the
above mentioned industrial restructuring
process. Among them, the following seem to be
worth mentioning as requiring further research:
(a) new forms of structural unemployment
which result from the rapid transition toward
CNC technologies featuring a strong laborsaving bias; (b) chronic long-term imbalances
in the external sector of the economy related
both to a rapidly expanding demand for
imported capital goods and to a pattern of
production specialization which strongly
emphasizes highly volatile markets for indus-
1595
trial commodities; (c) a growing degree of
business concentration deriving from the fact
that many small and middle-size enterprises
have been forced to leave the market, which
now remains in the hands of large domestic
conglomerates and/