of the four less-developed countries you selected with the high-income economies?
Discuss the evidence.
3 More than one development economist has opined that ‚Äúpoor countries are poor because
they are poor.‚ÄĚ What might at first sound like a vacuous and not very helpful observation
actually captures the essence of the neoclassical theory of economic growth. Looking at
note 2 below and the material in the text related to it, explain in words why saying ‚Äúpoor
countries are poor because they are poor‚ÄĚ might in fact be a very rich statement reflecting
some basic facts of real world economies that may impact on their ability to grow.
4 How can increased human capital accumulation which results in a higher average number
of years of education contribute to sustaining a high rate (i.e., percentage change) of
economic growth over time? What specific positive externalities might be expected from
workers with more education interacting with other workers with education or even
with workers with little education? Can you see how two (or more) workers might be
more productive together than they would be if they did not interact? Can you give
some concrete examples of how and when such interaction have resulted in such posi-
tive externalities? (One example: Have you ever studied with someone else? Can both
learn more by such sharing than by studying alone? Is this a positive externality, i.e., a
5 How can a country try to improve its rate of ‚Äútechnical efficiency change‚ÄĚ? What public
policies might be required? What ‚Äúinitial conditions‚ÄĚ might be important if an economy
is to have a positive rate of technical efficiency change? Is ‚Äúknowledge or technology‚ÄĚ
something that all economies can use with the same effectiveness? Why, or why not?
6 What does it mean to say technological change is endogenous to an economy? How is this
different from viewing technological change as exogenous? Does the idea that technology
is country-specific fit better into the exogenous or endogenous technological perspective?
Can technological change be both exogenous and endogenous at the same time?
7 Looking back at Table 8.1, how many years will it take the low- and middle-income
countries to double their per capita income, assuming they continue to grow at the
average annual rate which prevailed during 1970‚Ä“2005? How many years will it take for
the high-income economies to double their per capita income, using their average annual
growth rate for 1970‚Ä“2005? How long would it take Sub-Saharan Africa to double its
income per person if it grew at the average annual rate which prevailed from 1970‚Ä“2005?
(Hint: using the Rule of 70, the doubling time in years = 70/annual % change; in finance,
the Rule of 72 is often used in the same way. In fact, the Rule of 70 gives a very good
estimate of doubling for percentage rates of 5 percent or less. The Rule of 72 gives better
estimates for percentage changes between 5 and 10 percent. If growth rates are negative,
what is being determined is the time to halve a value, and for negative values the Rule
of 70 is more accurate.)
8 Consider Table 8.4 on ‚Äútechnical efficiency change.‚ÄĚ What might be some reasons
explaining the negative value for South Korea? Does negative technical efficiency
change necessarily mean an economy is inefficient or becoming more so? What does
negative technical efficiency mean, then? (Hint: remember, technical efficiency change
has to do with to what degree a country is approaching its production possibilities fron-
tier, a frontier that is a moving target as a result of on-going knowledge creation and
9 Draw a ‚Äúfuzzy‚ÄĚ or ‚Äúflexible‚ÄĚ production function showing how accumulating more
264 The Process of Economic Development
human capital (the variable on the horizontal axis) can result in different levels of per
capita output (the variable on the vertical axis) depending on the spillover effects and
positive externalities to the economy that human capital accumulation provides. Think,
for example, of a string in the shape of a production function that is not fixed in place,
but is capable of sliding upward as spillovers occur at any given level of human capital
accumulation. Are there increasing returns to your input?
1 In Solow-type neoclassical models, technology is treated as if it were a pure public good avail-
able to all countries and firms freely and with the same effect on productivity when applied to the
production process, regardless of any differences in organizational or institutional structures or in
the levels of education or skill levels of the work force.
Income convergence among nations will not occur in the Solow model, however, if a poor
country does not save and invest at the same rate as higher income countries, that is, unconditional
convergence is not predicted by the neoclassical model. Being a model where growth is based
primarily on capital accumulation, that is, on additional investment, which is equal to saving in
the neoclassical formulation, the Solow model only predicts convergence of poor nations to rich
nations if the poor nations save and invest a proportion of their GDP which is similar to that of
the richer nations, assuming comparable rates of population growth. Income convergence is thus
conditional on the similarity of the fundamental variables affecting the level of income per capita
shown in equation 4.5 in Chapter 4.
2 The level of savings in period t, St, is related to the level of aggregate income. In simple models, this
relation is often stated as St = a + sYt, where Yt is income in period t, a > 0 is some constant amount
saved out of income each period t, and 0 < s < 1, is the marginal propensity to save (MPS), that is,
the fraction of each additional unit of income which is saved rather than spent. It is easy to see that
if the constant, a, and the marginal propensity to save, s, and income, Yt, are all small values, then
so too will be St. To the extent that saving is required for investment, then a low level of aggregate
savings contributes to a low level of aggregate investment and a lower rate of economic growth in
future and a lower income per person. Other, more complex, and perhaps more realistic, theories
of savings, such as the ‚Äúpermanent income hypothesis‚ÄĚ and the ‚Äúrelative income hypothesis‚ÄĚ can
be read about in any intermediate macroeconomics text, but the ‚Äúlow income‚Ä“low saving‚ÄĚ relation
remains in force within these more complex formulations. This is another example of a ‚Äúvicious
circle of poverty.‚ÄĚ
Also of interest are the Kaleckian or Kaldorian-type savings functions which disaggregate
savings behavior by classes, usually the working and the capitalist class, each of which is presumed
to have a different propensity to save. In such models, the higher income, capitalist class is assumed
to save more than workers, and thus income inequality is presumed to functionally contribute to a
higher level of investment and greater economic growth. In many developing nations, however, this
may be an unwarranted assumption. To the degree that conspicuous consumption plays an impor-
tant role among the remnants of the agro-export and financial elites, the rate of domestic saving
may be driven down, as will the rate of capital formation, by a high degree of income inequality.
This possibility is considered later in this chapter.
3 The literature is vast and often quite challenging. For a useful collection of original readings, see
Stiglitz and Uzawa (1969).
4 Even though there is evidence of relative convergence of incomes in South Asia and East Asia and
the Pacific, both regions have low levels of income per person in an absolute sense compared to
the high-income economies. For example, in 1970, GDP per person in East Asia and the Pacific
was $175 versus $13,372 in the high-income economies. East Asia‚Ä™s GDP per person was but 1.3
percent of that of the high-income economies. In 2005, relative incomes were $1,352 and $28,304,
respectively (all data is in 2000 US dollars, i.e., adjusted for inflation), and the ratio of East Asia‚Ä™s
GDP per person to that of the high-income economies had increased to 4.8 percent because of the
more rapid rate of GDP growth in East Asia (6.0 percent per year) compared to the high-income
economies (2.2 percent).
Though relative incomes converged somewhat between these two regions, the gap in absolute
Endogenous growth theories and new strategies for development 265
incomes actually widened. Whereas in 1970, GDP per person in the richer economies exceeded that
in East Asia by $13,197 the absolute gap in average GDP in 2005 had grown to $26,952. If East
Asia were to continue to grow at an average yearly rate of 6.0 percent, it would take until the year
2058 for East Asia‚Ä™s income to reach the level that the high-income economies had reached in 2005.
It would take until 2089 for GDP per person to converge for both regions at a value of $173,460 per
person, assuming that the annual growth rates of 6.0 and 2.2 percent are maintained.
More sobering is to realize that with a GDP per person of $562 in 2005, and a slower rate of
growth per year than East Asia (2.7 percent), South Asia‚Ä™s income would not converge to that of the
high-income economies until 2808!
5 This is due to the assumption of diminishing returns to capital. As the amount of capital, K, in use
increases, its marginal product, MPK = df(K)/dK, decreases. If we use the short-hand of assuming
profit, r, to be the return to capital paid in units of production, then the return to capital when there
are K0 units of capital, r0, is greater than where there are K1 units of capital in use, if K0 < K1. Graphi-
cally, the slope of the production function gets flatter, and r decreases, as K increases.
6 Plosser (1993: 63‚Ä“4), for example, notes that recent estimates of the Solow model show that if
physical capital differences were to account for the fact that per capita income in the United States
is about twenty times that of Kenya, the United States capital stock per person would have to be
on the order of 8,000 times greater than in Kenya, when in fact it is ‚Äúonly‚ÄĚ about twenty-six times
larger. This leads to the suggestion that there must be factors other than the stock of physical capital
and of rates of investment which account for the wide income differentials between developed and
7 See Warsh 2006, especially Chapters 15‚Ä“24, for a very insightful, informed, and eminently read-
able intellectual history of the rise of endogenous growth theory and its impact on thinking in
8 The Solow-type model might be called an exogenous growth model, in the sense that once a nation
reaches its ‚Äústeady state‚ÄĚ level of income per capita as determined in equation 4.5 of Chapter 4,
income per capita will only grow in the future at a pace determined by the rate of exogenous, that
is, externally determined, technological change taking place at the world level. On the other hand,
in the endogenous growth models, the rate of change of both short- and long-run income per capita
are internal, that is endogenous, to the workings of each economy. In these models, the pace of
economic growth and of technological change depend on that economy‚Ä™s specific organizational
structure, labor types and skills, institutions, government, and an entire range of factors we shall be
discussing in this chapter.
9 In particular, there were the suggestive and profound works of Young (1928) and Arrow (1962).
Rereading Young today one sees how old ideas are rediscovered.
10 Human capital accumulation is any improvement in the quality of labor, be it the result of increased
education, on-the-job learning, better health care, interaction with other workers with accumu-
lated human capital, or other influences which improve labor‚Ä™s productivity without adding more
physical capital to production.
11 Endogenous growth models eventually accepted some form of imperfect competition as a real-
world reality that needed to be built into the theories. There was no other way to understand why
firms would undertake expensive research and development and invest in the search for new
knowledge unless there was some way to capture a sufficient part of the expected returns from
such purposeful action so as to warrant the initial investment. Assuming perfect competition also
assumed away such purposeful behavior with sufficient individual returns through the assumption
of perfect information imbedded in the concept. Thus endogenous growth theories became even
more complicated theoretically, though more real, by presuming that imperfect competition was
12 Controversy over the superiority of endogenous growth models versus neoclassical models
continues. endogenous growth theories ask economists to let go of the concept of diminishing
returns in the short run for some variable inputs, especially human and research capital, a large
theoretical leap given the power of neoclassical theory. For a taste of the strength of the passions,
and perhaps egos, involved on both sides of this debate, see the mostly non-technical symposium
on endogenous growth in the Journal of Economic Perspectives, Winter 1994. Leading proponents
of the endogenous model, such as Paul Romer, square off against skeptics, like Robert Solow, who
defends a revised neoclassical model. Warsh‚Ä™s (2006) book also reveals how much was and is at
stake in this debate for the reputations of many economists.
266 The Process of Economic Development
13 It is perhaps worth noting that the World Bank study was initiated at the insistence of, and financed
by, the government of Japan as a result of ‚Äúa determination on Japan‚Ä™s part to get the World Bank
to pay greater attention to the distinctive features of the East Asian development experience, which
stood in marked contrast to development approaches the Bank was then advocating‚ÄĚ (Fishlow and
Gwin 1994: 3). As many commentators have noted, there is much in the Bank‚Ä™s report that would
seem to reflect a criticism of past World Bank policy and an acceptance of at least some aspects of a
more activist approach for certain kinds of state involvement in the support of more rapid economic
growth and human development.
14 Solow (1994: 49‚Ä“51) has argued that the elimination of diminishing returns to the variable inputs
is one of the weakest links in the endogenous growth models. He shows that if marginal returns are
increasing, then an endogenous growth model would predict an infinite level of per capita income in
the finite future, a clear impossibility. Typical of Solow, known for his humorous side, he comments
on this ‚Äúknife-edge‚ÄĚ instability of the increasing returns assumption: ‚ÄúIt is one thing to say that
[income] will eventually exceed any bound. It is quite another to say that it will exceed any stated
bound before Christmas.‚ÄĚ So, if there are increasing returns to any factor, the theoretical model
explodes to infinity. If there are diminishing returns, on the other hand, the endogenous model is
reduced to the Solow model, and there will be a finite, steady-state income level. Only when there
are constant returns to the variable factors does the possibility of sustained increases in long-run per
capita income make theoretical sense in the endogenous growth model. Solow, however, does not
think that the real world is likely to be so precise as to be of the constant-return variety.
Lau (1996: 90), however, finds increasing returns to scale on the order of 1.6 for developing
economies and approximately constant returns for more mature economies, suggesting sustained
growth is possible. Perhaps the real world is more precise than Solow would give it credit for, and
it may be possible to have ‚Äúconstant‚ÄĚ increasing returns. Perhaps increasing returns at the rate
found by Lau do not ‚Äúexplode‚ÄĚ the model, nor lead to infinite income by Christmas, but rather
permit long-term growth of income per person for fairly long periods of time, certainly longer than
would be possible in a world of diminishing returns. Park and Ryu (2006: 251, Table 3) estimate
something very close to constant returns over a thirty-year time frame, but with mildly increasing
returns in earlier periods for some of the East Asian economies.
Graphically, an endogenous growth production function is a bit like a piece of clay or Plasticine.
It is not fixed in place but is capable of shifting upward and stretching without any new investment
or labor being added to production. It does not have just one income level for each level of K or L;
since the economy‚Ä™s production function is capable of stretching, over time, the attainable level of
15 Increasing returns are thus accruing to the economy as a whole, not to any particular individual firm
through at least partial spillover of new proprietary knowledge to other producers. This is an impor-
tant point, since if the increasing returns were internal that would lead to one firm monopolizing
the industry. Increasing returns are external to any firm and accrue to the economy as a whole.
However, these theories do not presume that there is perfect competition, but rather that there is
some form of imperfect competition that leads firms to look for new products, more specialization,
and new ways to produce in the hopes of maximizing their profits.
16 As Lucas (1988) observed, individuals who have accumulated human capital tend to gravitate
toward locations where human capital is abundant not toward where it is scarce. Thus individuals
in rural areas with more education migrate to cities, and highly educated persons in less-developed
nations often migrate from the cities of their own country to the cities of more-developed nations ‚Ä“
the so-called brain drain ‚Ä“ to associate with others with abundant human capital. The rate of return
(i.e., the private benefits in increased income) to migrants is higher in their new setting than it
was in the human-capital-poor regions from which they migrated. This is an extremely important
This is strong, if casual, evidence for the positive externality effect of human capital; the produc-
tivity of one individual‚Ä™s human capital is increased, not decreased, when it is combined with
more human capital inputs similar or higher in quality. That is, this is evidence for an absence of
diminishing returns to human capital accumulation and, via positive externalities, for increasing
returns to such inputs when combined with complementary human capital. Myrdal, of course, made
much the same point in discussing ‚Äúvirtuous circles‚ÄĚ (see Chapter 6), though his theory was purely
literary, not mathematical and rigorous like the endogenous growth theory literature.
One estimate of the social rate of return to an additional year of secondary schooling is 13 percent,
Endogenous growth theories and new strategies for development 267
while another finds that each additional year of schooling above three to four years added to the
average years per person contributes about 5 percent to total output (Rowen 1996: 103).
17 Learning-by-doing effects are not limited to industrial and manufacturing pursuits or even to the
productivity of those who ostensibly are doing the learning-by-doing. In a study of the introduction
of high-yield seed varieties in agriculture, Foster and Rosenzweig (1995: 1205) found that ‚Äúfarmers
with experienced neighbors are significantly more profitable than those with inexperienced neighbors,
as they experienced positive learning-by-doing spillover benefits from those with more experience.
Such spillover learning effects to other farmers suggest that subsidies to early users of new tech-
nologies may help to not only increase total output and efficiency of those particular producers, but
also to contribute to the attainment of the social optimum that is even higher due to the ‚Ä˜teaching‚Ä™
they impart to neighboring farms.‚ÄĚ
The original formulation of the theory of learning-by-doing is due to another Nobel Prize-win-
ning economist, Kenneth Arrow. Arrow‚Ä™s work fits very neatly with the endogenous growth theo-
ries, as one of the characteristics of his model is that it is non-linear and that opens the possibility of
non-diminishing returns as productivity of a given input is able to expand over time in the process
18 The constant, a, can be interpreted as the output‚Ä“capital ratio, that is, Y/K, which indicates on
average how much additional output can be produced from each additional unit of capital. The
value, a, may also be interpreted as the incremental output‚Ä“capital ratio, that is, even though a is a
constant at any point in time, it is not necessarily fixed through time. As an economy evolves and
grows, the value of a can change, rising with increasing returns, falling with decreasing returns,
though at any point in time, it can be taken as a constant value. A changing value of a would result
in a ‚Äúflexible‚ÄĚ production function if graphed.
19 If well-trained individuals in a poor country migrate to higher-income economies where their skills
are worth more because they can be combined with those of other workers with high levels of skills,
a ‚Äúlock-in‚ÄĚ of a low-level income poverty trap may be more likely. Already richer countries actu-
ally benefit from the spillover of learning of these individuals, the training of which took place in a
poorer economy, as their human capital can contribute to more rapid growth rates.
20 Given that there are high initial costs associated with creating new knowledge or new products
or processes of production, i.e., of more specialization, these fixed costs alone are sufficient to
generate increasing returns, since higher levels of output are associated with decreasing costs of
production. See Jones (1998: Chapter 4) for a simple and clear exposition.
21 In other words, new knowledge is not a pure public good available for adoption by any firm or
economy. Some spillover is likely to occur, but for most new knowledge it is possible for its creator
to exclude others from using it, at least for a time. Such knowledge thus shares one characteristic of
a public good: no one‚Ä™s use of such knowledge prevents anyone else from using it (i.e., knowledge
is a non-rival good). But knowledge also at least in part shares, for a time and to some degree, an
important characteristic of a private good, that is, it is possible to exclude others from using it.
Whether knowledge is therefore called an impure public good or a quasi-public good or a club good
is immaterial. What is important in the endogenous growth models is that new knowledge or new
technology or new goods or a new production process is created by someone or some firm with an
expectation of a return on that uncertain investment.
Knowledge is different from human capital. Human capital is all the knowledge imbedded in
human beings that must have been learned in some way. Such knowledge is non-transferable.
Knowledge is the un-imbedded raw material that can be utilized by purposeful human action. It is
this world pool of knowledge, to the extent that an economy can tap into it, that creates the possi-
bility of increasing returns over the future.
22 For technological models of endogenous growth that are quite consistent with the Ayresian
approach or Myrdal‚Ä™s ‚Äúcumulative causation‚ÄĚ considered in Chapter 6, see the very suggestive
work of Grossman and Helpman (1991, 1994). This theoretical work puts the technological process
at the forefront of the development process, a view we argue is essential in Chapter 13.
23 The primary school enrolment rate is defined as the number of all students enrolled in primary
school as a percentage of primary-school-aged children. The ratio can exceed 100 percent due to
the presence of students older or younger than the normal primary-school age.
24 There is a good reason to exclude the share of national output going to investment as a variable
explaining the rate of economic growth, however. There is a feedback from economic growth to
investment and back to income and output that makes these two variables interdependent rather
268 The Process of Economic Development
than independent. This is Rodrik‚Ä™s rationale for not including the investment rate, variable 5, as an
independent variable in his own re-estimation of the World Bank model (Rodrik 1994: 19‚Ä“20). Also
see Grier (2003: 394) who models investment as endogenous.
25 Again, the HPAEs are the eight high-performance Asian economies of Japan, Hong Kong, South
Korea, Singapore, Taiwan, Indonesia, Malaysia, and Thailand.
26 Recall, from our discussion in Chapter 6 Ayres‚Ä™s emphasis on ceremonialism as a retarding factor
that can obstruct the creation, the spread, and the adaptation of new technology and the spread of
new knowledge. In Latin America, such backward ceremonial institutional structures might include
the extreme concentration of land ownership and income, monopoly power in certain industries,
military dictatorships during certain periods and pseudo-democratic authoritarian structures.
The adjusted R 2 = 0.67 was obtained when the investment/GDP ratio coefficient was included as
an explanatory variable, but since Rodrik felt it inappropriate to include investment as a separate
explanatory factor for economic growth given the feedback between the two, we have reported his
regression 3 results only (Rodrik 1994: 20).
28 Why should land and income inequality play a retarding role in growth performance? Two responses
have been forthcoming; one concentrates on aggregate demand. Studies of Japan and Taiwan have
indicated that robust demand by farmers for domestic manufactured goods contributed strongly
to economic growth during the formative stages of the industrialization process. Thus a ‚Äúbetter‚ÄĚ
distribution of land and income would contribute to a larger domestic market. The second reason
for a link between higher inequality and lower economic growth has to do with social stability;
societies with a greater degree of equality are likely to be more politically stable. When political
instability is a recurring event, investment of all types, and hence economic growth, is likely to be
reduced. Rodrik‚Ä™s study pointed out that the World Bank did recognize income distribution and land
distribution as important factors in the HPAEs success, but argued that the Bank‚Ä™s study ‚Äú[lacked]
a serious discussion of equity as a precondition of growth‚ÄĚ (Rodrik 1994: 26).
29 In this example, technological change is not neutral. Assuming equal units on both axes, produc-
tion possibilities frontier FF2 shows that technology has had a greater impact on the production of
consumption goods, C, relative to the impact on capital goods, K.
30 It is not enough to simply have access to the best technologies or the most modern capital equip-
ment if greater efficiency is to be attained. Technology transfer or technology purchases do not
guarantee that a country has the capacity to make effective use of such technology. As Nelson and
Winter‚Ä™s analogy suggests, buying the best and most advanced tennis racket available is not the
same thing as having mastered the skills necessary to play the game. That is where human capital
accumulation, R&D expenditures, industrial organization, and a whole range of ‚Äúinputs‚ÄĚ to produc-
tion enter into determining success or failure of the development effort.
31 It may be that the very rapid economic transformation of South Korea, particularly the ‚Äúcompressed‚ÄĚ
nature of the industrialization process compared to the already developed countries, has meant that
the ability to keep up with international best practice has been difficult. On the other hand, it may be
that the measures of technical efficiency change for countries with rapid growth do not accurately
measure such advances toward best practice. With very rapid physical capital accumulation, new
machines typically will embody the latest technology, so some of what might be counted as tech-
nical efficiency in slower growing economies is imputed to increases in physical capital in rapidly
32 Research continues on these issues. An interesting recent contribution, though somewhat outside
the endogenous growth framework, is Jones (2007), who offers a new theoretical explanation for
the widening income gaps between rich and poor countries. His model is based upon the concept
of industrial linkages considered in Chapter 5 and the complementary nature of inputs in produc-
tion. The latter concept allows for the introduction of a range of institutional realities that can help
or hinder production in modern economies. One of the attractions of this new theory is that it does
not require that there be constant or increasing returns to explain wide income differences between
Arrow, Kenneth J. 1962. ‚ÄúThe Economic Implications of Learning by Doing,‚ÄĚ Review of Economic
Studies 29 (June): 155‚Ä“73.
Endogenous growth theories and new strategies for development 269
Barro, Robert J. 1991. ‚ÄúEconomic Growth in a Cross Section of Countries,‚ÄĚ Quarterly Journal of
Economics 106 (May): 407‚Ä“43.
‚Ä”‚Ä”. 1993. ‚ÄúHuman Capital and Economic Growth,‚ÄĚ pp. 199‚Ä“216 in Federal Reserve Bank of Kansas
City, Policies for Long-Run Economic Growth. Kansas City: Federal Reserve Bank.
Fishlow, Albert and Catherine Gwin. 1994. ‚ÄúLessons from the East Asian Experience,‚ÄĚ pp. 1‚Ä“12 in
Albert Fishlow et al., Miracle or Design?: Lessons from the East Asian Experience. Washington,
D.C: Overseas Development Council.
Foster, Andrew D. and Mark R. Rosenzweig. 1995. ‚ÄúLearning by Doing and Learning from Others:
Human Capital and Technical Change in Agriculture,‚ÄĚ Journal of Political Economy 103 (December):
Grier, Robin. 2003. ‚ÄúToothless Tigers? East Asian Economic Growth from 1960 to 1990,‚ÄĚ Review of
Development Economics 7 (August): 392‚Ä“405.
Grossman, Gene M. and Elhanan Helpman. 1991. Innovation and Growth in the Global Economy.
Cambridge, MA: MIT Press.
‚Ä”‚Ä”. 1994. ‚ÄúEndogenous Innovation in the Theory of Growth,‚ÄĚ Journal of Economic Perspectives 8
Jones, Charles I. 1998. Introduction to Economic Growth. New York: W.W. Norton & Co.
‚Ä”‚Ä”. 2007. ‚ÄúThe Weak Link Theory of Economic Development,‚ÄĚ draft copy, University of California,
Berkeley and NBER.
Landau, Ralph, Timothy Taylor and Gavin Wright. 1996. ‚ÄúIntroduction,‚ÄĚ pp. 1‚Ä“18 in Ralph Landau,
Timothy Taylor, and Gavin Wright (eds.), The Mosaic of Economic Growth. Stanford: Stanford
Lau, Lawrence J. 1996. ‚ÄúThe Sources of Long-Term Economic Growth: Observations from the Experi-
ence of Developed and Developing Countries,‚ÄĚ pp. 63‚Ä“91 in Ralph Landau, Timothy Taylor, and
Gavin Wright (eds.), The Mosaic of Economic Growth. Stanford: Stanford University Press.
Lucas, Robert E. 1988. ‚ÄúOn the Mechanics of Economic Development,‚ÄĚ Journal of Monetary
Economics 22 (July): 3‚Ä“42.
Park, Jungsoo and Hang K. Ryu. 2006. ‚ÄúAccumulation, Technical Progress, and Increasing Returns in
the Economic Growth of East Asia,‚ÄĚ Journal of Productivity Analysis 25 (June): 243‚Ä“55.
Plosser, Charles I. 1993. ‚ÄúThe Search for Growth,‚ÄĚ pp. 57‚Ä“86 in Federal Reserve Bank of Kansas City,
Policies for Long-Run Economic Growth. Kansas City: Federal Reserve Bank.
Reynolds, Lloyd G. 1986. Economic Growth in the Third World: An Introduction. New Haven, CT:
Yale University Press.
Rodrik, Dani. 1994. ‚ÄúKing Kong Meets Godzilla: The World Bank and ‚Ä˜The East Asian Miracle‚Ä™,‚ÄĚ
pp. 13‚Ä“53 in Albert Fishlow et al., Miracle or Design?: Lessons from the East Asian Experience.
Washington, D.C: Overseas Development Council.
Romer, Paul M. 1994. ‚ÄúThe Origins of Endogenous Growth,‚ÄĚ Journal of Economic Perspectives 8
Rowen, Henry S. 1996. ‚ÄúWorld Wealth Expanding: Why a Rich, Democratic, and (Perhaps) Peaceful
Era Is Ahead,‚ÄĚ pp. 92‚Ä“125 in Ralph Landau, Timothy Taylor, and Gavin Wright (eds.), The Mosaic
of Economic Growth. Stanford: Stanford University Press.
Solow, Robert M. 1994. ‚ÄúPerspectives on Growth Theory,‚ÄĚ Journal of Economic Perspectives 8
Stiglitz, Joseph E. and Hirofumi Uzawa (eds.). 1969. Readings in the Modern Theory of Economic
Growth. Cambridge, MA: MIT Press.
UNDP (United Nations Development Programme). 1994. Human Development Report 1994. Oxford:
Oxford University Press.
Warsh, David. 2006. Knowledge and the Wealth of Nations: A Story of Economic Discovery. New York:
W.W. Norton & Co.
World Bank. 1983. World Development Report 1983. Oxford: Oxford University Press.
‚Ä”‚Ä”. 1991. World Development Report 1991. Oxford: Oxford University Press.
‚Ä”‚Ä”. 1993a. The East Asian Miracle. Oxford: Oxford University Press.
270 The Process of Economic Development
‚Ä”‚Ä”. 1993b. World Development Report 1993. Oxford: Oxford University Press.
‚Ä”‚Ä”. 1995. World Development Report 1995. Oxford: Oxford University Press.
‚Ä”‚Ä”. 2003. World Development Report 2003. Oxford: Oxford University Press.
Young, Allyn A. 1928. ‚ÄúIncreasing Returns and Economic Progress,‚ÄĚ The Economic Journal 38
Young, Alwyn. 1995. ‚ÄúThe Tyranny of Numbers: Confronting the Statistical Realities of the East Asian
Growth Experience,‚ÄĚ The Quarterly Journal of Economics 110 (August): 641‚Ä“80.
9 The initial structural transformation
Initiating the industrialization process
after reading and studying this chapter, you should better understand:
‚ÄĘ the need for industrialization to accelerate the pace of economic growth and
‚ÄĘ the structural changes which easy import substitution industrialization (ISI) can
initiate as the first stage of industrialization;
‚ÄĘ the role of government in speeding-up industrialization;
‚ÄĘ the nature of transitional inefficiencies and competition from imported goods;
‚ÄĘ the static and dynamic welfare effects of infant industry tariffs;
‚ÄĘ the role that development banks, financing, and other initiatives can play in
‚ÄĘ the potential benefits and dangers of pursuing an ISI strategy;
‚ÄĘ why easy ISI is almost assuredly necessary but definitely not sufficient for escaping
past adverse path dependence.
Introduction: an industrialization imperative?
Achieving a higher level of development appears to be inextricably intertwined with the
industrialization of an economy. One chronicler of the nature of structural change in the
development process has written that the issue is not whether industrialization is necessary for
development but only ‚Äúwhen and in what manner it will take place‚ÄĚ (Syrquin 1988: 218).
The structural transformation that accompanies industrialization alters not only the phys-
ical landscape of nations via urbanization, internal labor migration and the establishment of
a complex of, typically, urban business enterprises. It also alters many of the cultural, social,
and other institutional arrangements that have stamped a particular society and made it what
it is to that point in time.
With industrialization, nations become more homogeneous in terms of what is consumed,
what is read, what is seen on the television and at the cinema, and what is learned in schools
and in universities. Of course, many cultural differences remain even after industrialization,
since they arise from distinct historical experiences woven into the fabric of individual soci-
eties in their language, literature, law, folklore, religion, music, cuisine, and embedded in
shared attitudes, perspectives, and practices.
Industrialization and development provide unmistakable benefits to society, but sacri-
fices also are required, as is true whenever choices must be made. One inevitable cost is an
272 The Process of Economic Development
often wrenching change in social values and the everyday rhythms of life in less-developed
nations. Industrialization induces adjustments that often are disruptive of ancient patterns
and ways of doing things. Frequently, opposition to industrialization arises from powerful
groups who benefit from the status quo and who thus oppose the possibility of a new order. In
other circumstances, opposition to industrialization and the changes it sets in motion comes
from religious or cultural institutions which see a spiritual or traditional way of life being
threatened by the material concerns of a business society and the private acquisitiveness and
individualism on which it thrives.
Opposition to industrialization is natural and expected, and the debate over how to develop,
how fast, and who is to benefit are important issues for any nation to consider and evaluate.
There is no a priori reason why all nations must follow the same industrialization path or must
make the same sacrifices on the road to a higher living standard and an improvement in their
level of human development. Since many decisions about development can have irreversible
effects on the local environment or on social structures, it is wise for societies to engage in an
ongoing dialogue over the present and the future. Too often, in evaluating the expected effects
of development, economic and business concerns ‚Ä“ perhaps because they are more amenable
to measurement ‚Ä“ are treated as if they are more important than are spiritual, cultural, environ-
mental, or religious interests which are intrinsically more subjective and hence more difficult
to quantify. That should not make them less important to consider, however.
Reconciling conflicting interests may slow progress, but each society must evaluate as
objectively as possible to what extent trade-offs made to sustain traditional dimensions of
social, political, and cultural life are worth the sacrifice. In some cases, caution in balancing
growth and development versus other interests is justified. In other circumstances, an unnec-
essary slow-down in progress for the majority is simply the consequence of catering to
narrow special interests.
Nonetheless, all nations and economies are linked and interdependent to varying degrees,
making the general forces and pressures of the global capitalist economy difficult, perhaps
increasingly impossible, for nations to circumvent. Given these economic bonds and the real
need for less-developed nations to make progress along both economic and human develop-
ment dimensions (think back to the Millennium Development Goals introduced in Chapter
1), higher levels of industrialization ‚Ä“ even with its attendant costs ‚Ä“ would seem to be abso-
The data at the top of Table 9.1 suggest the close relation between the growth rate of
industry and the rate of growth of total national output.1 There would seem to be a positive
correlation between the rate of growth of total output and the pace of industrial growth,
though the direction of causation is certainly not proven by this data.
If one disaggregates further to look at the link between industrial growth and the growth of
total GDP for specific countries, the relationship is also positive, as shown for China, India,
and South Korea in the last lines of the table. More rapid industrial growth tends to be asso-
ciated with more robust GDP growth, though again the direction of causation is not defined,
and it most likely is two-way.
Looking at China, for example, as the rate of industrial growth rose from 9.5 percent
annually in the 1980s to 13.5 percent in the 1990s, GDP growth increased from 9.3 percent
to 10.4 percent per year. When industrial growth slowed down after 2000, so too did the GDP
growth rate. Korea‚Ä™s pattern also shows a positive relationship between industrial and GDP
growth rates; when industrial growth fell from 1980 to 2005, GDP growth also declined.
India‚Ä™s growth of industry fell and then rose, and so too did the GDP growth rate. Again, this
does not prove causation from industrial growth to GDP growth; the direction of causation
The initial structural transformation 273
Table 9.1 Industrialization and economic growth (annual percentage growth of constant dollar GDP
1980‚Ä“90 1990‚Ä“2000 2000‚Ä“5 1980‚Ä“90 1990‚Ä“2000 2000‚Ä“5
East Asia and Pacific 10.9 9.2 7.5 8.4
Latin America and Caribbean 0.6 3.2 1.0 1.1 3.4 1.9
Middle East/North Africa 4.1 3.0 2.9 3.1 4.0 4.1
South Asia 5.6 6.8 5.6 5.2 6.4
Sub-Saharan Africa 1.1 1.6 4.4 1.9 4.3
9.5 13.5 10.6 9.3 10.4 9.5
7.1 5.6 5.8 5.5 6.8
10.9 6.5 4.9* 6.1 4.7*
Source: World Bank, World Development Indicators Online.
could be in the other direction. It is likely, though, that the causation is reciprocal, or at least
that is what economic theory would suggest. But the important point is that the two growth
rates are positively related.
In fact, not only are the growth rates of industry and GDP directly related, but so are their
relative magnitudes. When the growth rate of industry is low, so is the growth rate of GDP
(and vice versa). Just consider Latin America and the Caribbean or Sub-Saharan Africa in
the 1980s. On the other hand, when the industrial growth rate is high, so is the GDP growth
rate. Look at East Asia and the Pacific or China. There is thus not only a positive relation-
ship between the growth rates of industry and GDP that can be observed in Table 9.1; these
growth rates are closely related to one another. When one is low, the other is low; when one
is high, the other is high as well.
Besides the positive relationship between industrial expansion and economic growth,
there is yet a further reason for pursuing industrialization. As considered in Chapter 3 and
then again in Chapter 6 in the discussion of the Prebisch-Singer hypothesis, countries which
predominantly produce and export primary products and import the bulk of their manufactured
goods are apt to experience instability in their terms of trade, that is, in the purchasing power
of their exports. In fact, there has been a long-term deterioration in the purchasing power
of these exports relative to the manufactured goods imported. Industrialization is therefore
a means to create not only a more productive domestic economic structure which raises
domestic incomes; it also can help to create the possibility for an import and export pattern
more similar to that of the already-developed nations, one that tends to be more stable in
terms of the purchasing power of a country‚Ä™s exports (see Focus 9.1).
There is thus a twofold industrial imperative: first, higher levels of industrialization can
contribute to higher levels of income, and secondly, industrialization sets the stage for a
desired structural transformation of import and export patterns.
Structural change and economic growth
It is not simply the expansion of the industrial and manufacturing sectors that is critical to the
pace of development. The process of growth and development is the unfolding consequence
of a quantitative and a qualitative reorientation of the entire economic and social structure
of a nation. It involves basic changes in the education level of the population, in what they
274 The Process of Economic Development
know and can do, in business organization and in the population‚Ä™s way of thinking about
their relation to the world around them. It is as much ideological as it is economic. We know,
too, from Chapter 8 on endogenous growth that higher levels of development require the
accumulation and use of more of the ever-expanding world stock of knowledge.
The size of the agricultural (primary) sector tends to shrink with economic growth as
rural workers move into the industrial (secondary) and services (tertiary) sectors. This
transformation entails the migration of labor from the rural countryside to urban areas where
former rural workers become urban workers available to run the machinery of industry and
to work in all the supporting firms, shops, and institutions, including government, which an
expanding economy requires.
During successful industrial transformations, agricultural production becomes more efficient
and intensive in its use of capital, both physical and human, and in the use of knowledge and
technology which increase the productivity of workers remaining in that sector. An industrial
revolution without an agricultural transformation will be but a partial success (see Chapter
11). As we saw in Focus 8.2, a change in agricultural land distribution patterns in the direction
of greater equality seems to be associated with more rapid progress as well. As an economy
proceeds to ever-higher paths of development, there is continuing structural transformation
as workers shift from the secondary sector toward the service sector to work in commerce,
transportation, trade, government, finance, educational institutions, and so on.
Over time, then, more developed economies tend to have the majority of their labor force
employed in industry and services. As well, the greatest part of total income is generated in
these two sectors, while the relative importance of agriculture shrinks. All three sectors of
production show a tendency toward converging levels of worker productivity over time if
development efforts are successful, as the labor force is used to its fullest effectiveness. The
spread of technology and human capital accumulation generates a trend toward homogeneity
among the primary, secondary and tertiary sectors in terms of the level of output per worker
(Chenery 1979: 18‚Ä“21; also see Kuznets 1971, on structural change patterns that induce and
accompany economic growth and development; Exercise 2 at the end of the chapter gives
you some practice in examining these issues).
Table 9.2 shows the structural shift of labor usage from primary production toward
secondary and tertiary activities that occurs at higher levels of development. High-income
economies have less than 5 percent of employed workers in agriculture; in the low-income
economies, this proportion is nearly 60 percent in the most recent period. A strong inverse
relation between the share of the labor force in agriculture and the level of per capita income is
Table 9.2 Labor force distribution, by sectora
GDP/personb Agriculture Industry Services
1980 2001 1980 2001 1980 2001 1980 2001
17,282 26,460 8.6 4.0 33.5 26.3 57.8 69.5
1,245 1,809 62.0 36.2 18.8 19.5 17.3 40.7
310 338 66.0 59.2 14.0 14.2 18.1 26.5
Source: World Bank, World Development Indicators Online.
a As percentage of total labor force.
b GDP per person in constant 2000 US dollars.
c For the low-income economies, because of data limitations, the data is for 1990 and 1995 only.
The initial structural transformation 275
quite evident. The larger the share of the total labor force engaged in agriculture, the lower
the level of income. At higher levels of development, the share of the labor force engaged in
agriculture is relatively small.
Likewise, there is an equally strong positive relation between the share of the labor force in
industry and in services and the level of development. The smaller the share of the employed
labor force in industry and services, the lower the level of income of the country or region.
The larger that share, the higher the level of aggregate income and development. Part of the
explanation for this has to do with the increase in total output that accompanies the shift
of labor from low-productivity agriculture to, typically, higher-productivity industry and
service activities. As one of the fundamental structural transformations first introduced at the
end of Chapter 1, this labor transfer results in a one-time boost to GDP as an economy‚Ä™s labor
force is distributed in a more optimal manner (again, Exercise 2 at the end of the chapter
illustrates the gains in output that can come from a redistribution of the labor force from low-
to high-productivity activities).
The slow pace of both agrarian transformation and of industrialization of the low-income
nations provides strong evidence of the compelling correlation for the agricultural and
industrial transformations that contribute to greater development as measured by income or
by the human development index (HDI).
Development thus involves both an industrial revolution and a reorganization of the
agrarian sector.2 Success in such a transformation requires the proper additions to human and
physical capital and the attention to endogenous technological change considered in previous
chapters and to which we will return in later chapters.
An industrial transformation aimed at raising a nation‚Ä™s level of development that fails
to foment an effective agrarian transformation eventually will falter and fail, just as it will
be less effective without the attention to technology and to the quality of inputs that have
been identified in endogenous growth theories as central to sustained growth in income per
capita.3 All sectors of an economy must become progressively more technological and more
productive as the shift in labor use from primary to secondary to tertiary uses moves forward.
This typically involves a slow, then accelerating, reduction in the significance of primary
production within the overall productive structure of the economy. One development econ-
omist, Benjamin Higgins, has referred to this structural transformation from agricultural
dominance to industrial and service production primacy as the strategy of ‚Äúgetting rid of
farmers‚ÄĚ (Higgins and Higgins 1979).4
The Lewis dual-economy model of structural
transformation: surplus labor
How does this structural evolution that shifts labor from agriculture to industry come about
in practice? What sets this labor migration process in motion? A classic description, already
presented in Chapter 5, was provided by Nobel Prize-winning economist Sir W. Arthur
Lewis. Here we only briefly review the main points of his argument.
If there is surplus labor in agriculture, and by surplus labor we simply mean a lower level
of productivity of labor in agriculture than in industry (or services), then it is possible for
the more productive and higher-wage industrial (or service) sector to attract labor from the
countryside to urban areas by paying wages slightly above rural wages. Shifting workers
from lower-productivity activities in agriculture to higher-productivity industrial (or service)
jobs will result in an increase in total national output and, of course, income per person.5
Lewis observed that the typical less-developed nation was dualistic, not only in the sense
276 The Process of Economic Development
of having two key sectors, agriculture and industry, but also in the more fundamental sense
that these sectors had little interconnection. There was a traditional low-productivity,
low-technology agriculture sector where the great bulk of the population lived, worked,
and produced most of what they consumed. There also existed, or there could be created,
an incipient industrial sector where production was profit-oriented, more capital-using and
technology-driven, and where worker productivity was higher than in the traditional sector.
Those working in the modern sector bought food and perhaps some other primary inputs
from the traditional sector. But the most important link between the traditional and modern,
industrial capitalist sector was via the provision of labor from the primary sector to industry
via labor migration.
It was in this labor supply link between the two sectors that Lewis found the
transformation dynamic that could contribute to greater growth and development via
expanded industrialization. Lewis maintained that with continuing reinvestment from
the profits of the modern sector, the transfer of surplus labor from agriculture to industry
could be accomplished relatively smoothly, especially if the capital‚Ä“labor ratio did not rise
significantly in industry and if the growth of the industrial sector was financed at a low cost
from the profits of private industrialists (for further details review the discussion around
Figures 5.1(a) and 5.1(b)). It was believed there would be sufficient new employment for
rural migrants, as formal sector industries would be able to grow at a rate sufficient to
absorb the labor flow from agriculture.
The Lewis model has been a powerful theory for understanding how the structural
transformation of an agriculture society into a more industrial society might be accomplished.
If all works more or less as Lewis envisaged, it would seem that the labor migration process
is cumulative once it has begun, as the higher wages of industry attract rural workers until an
equilibrium of productivity and wages is reached between sectors.
Now we shall consider how countries might foment such a migration of labor from
agriculture to industry if it has not begun spontaneously on its own or if the pace of
industrialization is deemed too low.
Initiating the structural transformation process: industrialization
Is there a particular development strategy that less-developed nations can follow that can
help to initiate or accelerate the transition of labor from agriculture to industry and contribute
to the positive transformation of the productive structure of the economy from an agricultural
to an industrial base?
Historically, import substitution industrialization (ISI) has been the means by which
governments have supported this process. In this chapter, we consider in some detail how ISI
can propel an agriculture-based economy into a more modern, industrially-based economy.
We will also note the pitfalls to be avoided if the potential transformative benefits of this
instrumental strategy are to be reaped.
The first stage of industrialization: easy import substitution
With the exception of Great Britain, the first modern industrializing nation, all subsequent
successful efforts by nations to modernize have involved elements of import substitution
industrialization (ISI) as a means to promote the expansion of a domestic industrial sector.
ISI involves the establishment of firms within the domestic economy that produce for local
The initial structural transformation 277
consumption at least some of the manufactured goods currently being imported. With ISI,
domestic production replaces ‚Ä“ substitutes for ‚Ä“ some imported goods, hence the name of the
strategy: industrialization that substitutes for imports, ISI.
Which commodities are the most likely candidates for new or expanded domestic
production when beginning ISI?
Typically, ISI begins with the production of relatively simple non-durable consumer goods,
such as socks, beverages (soft drinks, beer, canned and bottled juices), furniture, shoes,
cloth diapers, toys, building materials, and so on. These are goods with pre-existing market
demands within the local economy that are currently being met by imports. The technology
and know-how for producing such goods tends to be relatively rudimentary, of low cost, and
frequently available ‚Äúoff-the-shelf‚ÄĚ on the world market with few or no restrictions on use,
as, for example, for sewing machines or bottling equipment. The demands on the skills and
knowledge of the labor force is moderate, which is important since many industrial workers
will be recent arrivals from rural areas with but modest levels of education and other training,
certainly within a factory setting.
A focus on producing non-durable consumer goods is also consistent with the Lewis
model. The surplus labor coming from agriculture will be absorbed more easily if industrial
production is concentrated in labor-intensive processes, which is the case for many non-durable
consumer goods. A focus on labor-intensive production can help to avoid problems of urban
unemployment and underemployment by providing sufficient new employment to the flow
of rural migrants. Consumer non-durable production that is labor-using conserves scarce
physical and financial capital and makes use of the less-developed economy‚Ä™s most abundant
factor of production, its labor.
Concentrating on producing goods similar to those being imported removes some of the
risk for local entrepreneurs undertaking these projects since there is a pre-existing demand
for low-cost goods, such as T-shirts or socks. Given a known import demand, local producers
will be better able to make reasonable calculations as to the probability of the profits that can
be expected from initiating or expanding production given local wage and cost conditions,
the level of technology and human capital accumulation, and the expertise level of local
workers, managers, and entrepreneurs.
This first stage of ISI with the focus on producing simple consumer non-durable goods for
the domestic market to replace imported goods has been called easy, primary, or horizontal
ISI.6 We will most often refer to this stage of industrialization as simply easy ISI.
It is not necessarily desirable nor recommended that easy ISI be initiated for all non-durable
consumer goods products being imported. Nor does easy ISI need to take place simultaneously
for many goods via a ‚Äúbig push‚ÄĚ in all sectors (Chapter 5). Some non-durable consumer
goods may be better candidates for easy ISI than others. Some imports may present better
opportunities for substitution later rather than sooner. Factors that need to be considered in
determining the pace and reach of easy ISI include:
1 the size of the domestic market, including population size, average income, and income
2 the size and skills of the existing pool of potential entrepreneurs who would operate new
3 the educational attainment and available skills of the labor force and the speed of migra-
tion of labor from agriculture to industry;
4 the availability of finance for the purchase of technology, needed physical capital and
other inputs, and for the needed training of managers and workers;
278 The Process of Economic Development
5 the potential for growth of demand over the future for each import substitution good and
the learning-by-doing potential from production; and
6 possible linkage and spin-off and spillover effects (positive externalities in production)
that might stimulate production or knowledge expansion in other industries, including
the establishment of new industries, as the consequence of initiating production in any
particular easy ISI industry.
Whether a ‚Äúbalanced‚ÄĚ or ‚Äúunbalanced‚ÄĚ ISI strategy in the sense discussed in Chapter 5 is
followed will depend on the evaluation in each country of the feasibility of and the expected
gains and costs accruing to either ‚Äúblanket‚ÄĚ or more ‚Äúselective‚ÄĚ ISI promotion of non-durable
consumer goods production. If selective ISI is chosen, as will be the case in most instances,
then the timing of the promotion and expansion of particular industries and firms can be
decided by taking into consideration the factors listed above and by making strategic deci-
sions based on those constraints.
Japan, India, and the larger countries in Latin America (Brazil, Mexico, Argentina, Chile,
and Colombia) followed an easy ISI strategy of industrial transformation beginning as far
back as the mid-1800s in the case of Mexico and by the last third of the century for the others
(Amsden 2001: 43). In Japan, the conscious decision to industrialize via ISI was part and
parcel of the Meiji restoration after 1868, and ‚Äúall modern industries were started as import
substitutes‚ÄĚ (ibid.: 174).
ISI on an expanded scale in Latin America was to some degree forced upon the coun-
tries by the breakdown in normal trade patterns as a result of the successive crises of the
First World War, the Great Depression, and then the Second World War. As a result of these
disruptions, it was difficult to purchase imported consumer goods because of supply inter-
ruptions. It was equally difficult to sell agricultural and other primary exports to traditional
markets to earn the foreign exchange required to purchase imports during these crisis periods
(see Focus 9.1). As a result of these interruptions to world trade, local manufacturing firms
were able to emerge and expand by producing for the domestic market behind the artificial
barrier of these external international crises that shielded them from import competition (see
the discussion on the structuralists in Chapter 6 for more details).
FOCUS 9.1 THE EXPORT STRUCTURE
A fundamental shared characteristic of many less-developed countries is the predomi-
nance of primary product exports as a percent of total merchandise exports. This, of
course, means that manufactured goods are a small fraction of total exports.
Successful structural transformation, as outlined at the end of Chapter 1 and discussed
in this chapter as well, not only alters what is produced domestically by shifting an
economy from an agricultural focus toward industrial and service enterprises. Long-term
development also requires a makeover in the export and import profiles of an economy. To
avoid declining terms of trade, the export composition must be modified so that traditional
primary product exports are replaced by manufactured good exports or higher value-
added non-traditional primary product exports (like cut-flowers, strawberries, or shrimp)
not as systematically subject to decreases in their purchasing power.
The following table reminds us of the often heavy concentration of primary product
exports in total exports that still persists for many less-developed economies, particularly
in Latin America and Africa. Primary exports include: agricultural raw materials, food, fuels
and ores and metals.
The initial structural transformation 279
Primary product exports as % of merchandise exports GNI per capita
1965 1980 1990 2000 2004 2004
Bolivia 95.7 97.1 95.2 71.0 86.3 960
Brazil 91.6 61.4 46.9 39.5 45.9 3,000
Chile 85.9 90.3 87.4 81.4 85.9 4,930
China ‚Ä“ ‚Ä“ 26.5 11.6 8.4 1,500
Colombia 93.0 79.6 74.2 67.5 61.6 2,010
Costa Rica 84.2 65.7 65.6 34.4 37.2 4,470
C√īte d‚Ä™Ivoire 94.7 ‚Ä“ ‚Ä“ 84.9 ‚Ä“ 760
Ghana 97.4 98.0 ‚Ä“ 85.2 ‚Ä“ 380
India 51.0 41.0 27.6 21.3 26.1 630
Indonesia ‚Ä“ 97.6 64.5 42.9 43.9 1,130
Korea 40.6 10.1 6.4 9.2 7.8 14,040
Malaysia 93.9 81.0 45.8 18.8 23.1 4,520
Mexico 83.6 88.1 56.5 16.5 20.0 6,930
Nicaragua 94.5 86.2 91.6 92.1 88.5 830
Nigeria 97.7 ‚Ä“ ‚Ä“ 99.8 ‚Ä“ 430
Pakistan 63.9 50.9 20.9 15.0 14.6 600
Taiwan 58.5 ‚Ä“ 7.4 4.8 7.3 15,350
Thailand 95.2 71.0 35.6 22.0 ‚Ä“ 2,490
Venezuela 98.2 98.3 89.6 90.9 88.4 4,030
UK 17.2 26.3 18.8 16.8 17.9 33,630
US 34.8 32.1 21.2 12.8 14.4 41,440
Examining the above figures carefully, it is possible to detect some correlation between
the percentage of primary exports and the level of per capita income in 2004. A larger
percent of primary exports to total exports tends to be associated with a lower per capita
income.7 Consider Bolivia, C√īte d‚Ä™Ivoire, Ghana, Nicaragua, and Nigeria; primary exports
comprise a relatively large share of their total manufactured exports. Income per capita is
relatively low in all these economies.
The correlation is not perfect, however. There are countries like Venezuela with a large
share of primary exports and yet income per capita is not as low as one might expect
from that fact alone. Why? Does it help to know that Venezuela‚Ä™s major export is oil?
Some primary product exports, like petroleum, may not be as associated with low average
incomes as is the case for other primary product exports.
Pakistan also stands out as an exception, too, but for the opposite reason. Pakistan has
both relatively low per capita income and a small share of primary product exports. The
bulk of Pakistan‚Ä™s exports were concentrated in manufactured goods, but the majority of
these were textile fibers, textiles, and socks, which tend to be generally low-value-added,
low-income generating exports.
Clearly, then, having manufactured exports alone is not the whole of the story for
becoming more developed. The concentration of exports and their diversity and type also
are important. Apparently there are ‚Äúgood‚ÄĚ manufacturing exports and not-so-good, just
as there are ‚Äúgood‚ÄĚ primary exports (like petroleum or cut-flowers, for example), while
others that are not so valuable (think bananas or peanuts).
China is an interesting intermediate case. In 1990, China‚Ä™s GNI per person was $320
(not shown in the table); as primary exports declined over time and manufactured goods
exports grew, average GNI rose to $1,500 in 2004. India‚Ä™s progress toward raising its
income per capita has been slower, and that may be connected to the larger share of
primary exports compared to China.
Source: World Bank, World Development Indicators Online
280 The Process of Economic Development
What easy ISI can do and what it cannot do: a foreshadowing
of what‚Ä™s to come
For the easy ISI strategy to work so that LDCs are less likely to be subject to long-term
declining terms of trade, it is necessary that the pattern of domestic production and exports
be altered. A reorganization of the internal structure of production toward a higher degree of
industrialization is insufficient for long-term progress without a fundamental reorientation of
what less-developed nations export to the world market.
It is important to keep this qualification in mind as we continue through this chapter. Easy
ISI can help set the stage for successful structural transformation in an economy; by itself, it
is insufficient, as we hope to make very clear. But easy ISI is necessary to get industrialization
under way. Without that, there is no possibility of an evolution in the export structure.
It is incorrect for countries to accept the existing export structure as immutable and
definitive, as if it reflected an optimal outcome of international specialization in production,
as described by Ricardo‚Ä™s theory of comparative advantage discussed in Chapter 4. Rather,
existing patterns of trade relations might better be viewed as a single frame of a motion
picture, a static snapshot. It is not the whole of the movie. Initiating easy ISI is one means
to begin to create new comparative advantage in more dynamic product lines that promise
greater gains in productivity and income growth over the future in both domestic and, if
successful, in international markets.
To advance the metaphor, easy ISI is not a rapid ‚Äújump-cut‚ÄĚ to a more efficient productive
structure compared to foreign producers, but rather it is akin to a scene within the whole of a
movie that ultimately advances toward the desired conclusion: an evolving industrial structure,
alterations in the use of labor by sector, and evolving patterns of imports and exports that can
instrumentally contribute to a progressively higher average standard of living. All of the parts
are essential to the final outcome, just as all the scenes of a film are integral to the whole.
The array of productive inputs a society possesses at any particular moment is its current
resource endowment. However, every nation is capable of modifying and making more
productive and more extensive its reproducible resources, particularly human capital,
knowledge, technology, and physical capital. expansion in the stock and quality of these
assets helps to increase the pace of economic expansion and leads to higher levels of income,
as Chapter 8 argued.
every economy has a constantly changing resource endowment on which it can build for
the future. A country‚Ä™s resource endowment is not a given. Productive resource endowments
are fluid, altering as population grows, as education takes place, as research and development
occurs leading to technology and knowledge advances, as governments improve their
operations, and so on. Of course, the pace and direction of change in endowments can be
quite different across economies.
Public policy has an impact, for good or bad, on an economy‚Ä™s potential resource endowment.
For development to continue, a society must alter and make more productive its resource base
so as to keep pace with the maturation of the world economy and its changing demand patterns
and new knowledge about how to produce and what to produce. Countries that fail to keep pace
will find their average incomes lagging and their growth rates lower than they could be.
The products and services in which a nation has a comparative advantage are, from this
perspective, subject to a substantial degree of control and design through the particular
decisions made regarding investments in human capital and knowledge, technology, better
organizational methods, and other policies under the full or partial control of government
policy-makers and individuals.
The initial structural transformation 281
Government and easy ISI
In an industry where no, or only a few, local firms produce goods similar to imported
commodities, stimulating domestic production will typically require some sort of government
intervention that can provide insulation to local firms from direct foreign competition. This
is necessary because in most less-developed nations few if any local firms are likely to be
able, at least initially, to produce efficiently enough to compete with imports. As relative
late-comers to industrialization, new firms in less-developed economies are likely to be at
a substantial disadvantage vis-√†-vis existing foreign firms which already produce for the
international market. A lack of financing for new projects, few trained entrepreneurs, low
skill levels of workers and fierce competition from imports all work against the spontaneous
expansion of the manufacturing sector in most LDCs.
Without an industrialization strategy supported by government, a thriving domestic industrial
sector may continue to be an impossibility, particularly in this era of very open international
markets with low barriers to trade among nations. New firms face start-up problems that tend
to make their costs higher than for comparable firms in other countries that already supply
the international market. One specific obstacle prospective manufacturers encounter is in
attempting to borrow funds for projects in economies with very imperfect markets in banking
and finance (see Focus 9.2). Thus without a government-sponsored industrial initiative, the
high probability of market failure through institutional inadequacy will result in a sub-optimal
level of industrialization and a sacrifice of potential overall economic welfare in most LDCs.
New firms anywhere, but especially those in less-developed economies, are likely to incur
higher per unit production costs for some time compared to foreign firms already successfully
producing for the international market in the same product line. These higher unit costs are
the consequence of what can be called transitional inefficiencies. Such inefficiencies can
‚ÄĘ inexperienced management and labor;
‚ÄĘ low levels of technological development, technological effectiveness, and knowledge
‚ÄĘ a lack of experience with accounting, financial, marketing, supply networks, and other
critical inputs to the production process;
‚ÄĘ low output levels that prevent the attainment of scale economies;
‚ÄĘ a low average level of human capital accumulation and educational attainment;
‚ÄĘ inefficient financial markets; and
‚ÄĘ government red-tape and other bureaucratic inefficiencies.
The list of factors that might contribute to transitional inefficiencies compared to foreign
firms supplying the import demand for comparable goods could be extended almost ad
infinitum and may differ among economies.
What is not in doubt is the effect of transitional inefficiencies: they raise the unit costs of
production for new firms in a less-developed economy relative to established producers in
other countries which export to markets around the world.
Transitional inefficiencies are reduced and overcome through ‚Äúlearning-by-doing,‚ÄĚ a real-
time process that can be fostered only by initiating production so that workers, managers,
282 The Process of Economic Development
FOCUS 9.2 CREDIT AND MARKET FAILURE
Some economists argue that if there are sufficient expected profits to justify the creation of
an ISI firm, private firms and entrepreneurs will recognize such opportunities and borrow
against future earnings to establish such enterprises. If that were truly the case, there
would be no need for government intervention to initiate industrialization.
However, even if such a future profit calculation is made by private entrepreneurs,
any market failure problem in financial markets because of a weak or poorly functioning
banking and financial intermediation system, or the lack of an equity capital or bond
market, may make the realization of such projects unlikely by rendering borrowing difficult
or even impossible.
This is an example of institutional inadequacy quite common in less-developed econo-
mies as a result of incomplete and poorly functioning markets. The result of such market
failure will be an inefficient allocation of society‚Ä™s resources, both currently and into the
future, and a sub-optimal level of economic and human development. When there is such
market failure in financial markets, there is a strong case to be made for government reme-
diation if industrialization is to have half a chance at success.
A further complication enters the picture when it is asked whether it is even reasonable
to expect prospective entrepreneurs in LDCs to be able to make the expected profitability
calculations for establishing new enterprises. This is an issue concerning the availability
of information. Joseph Stiglitz and Kenneth Arrow, both Nobel Prize-winning economists,
have written extensively on the economics of information.
Information is not free; it is a ‚Äúgood‚ÄĚ much like any other. There is a cost to both
producing information and to obtaining it. In less-developed nations with only a rudimen-
tary industrial structure, is it reasonable to expect potential entrepreneurs to be able, at
a sensible private cost, to identify the expected future profit stream of an investment that
no one has yet made?
If an economy lacks the required social and physical infrastructure, appropriate human
capital resources, supply networks, sources of financing, and so on, is it a reasonable
possibility to think prospective entrepreneurs will be able to estimate what profits they
might make if they did set up a firm?
If the answer to these questions is no, then there is a valid economic rationale for
government action that corrects for market failure and for information costs that distort
the private calculation of future benefits. Such government intervention has been shown
to be able to improve upon what the market outcome would have been, that is, to increase
economic efficiency. An easy ISI strategy can be one part of an effort to help the market
work better over the long term, thus rendering such government intervention to correct for
market failure less necessary over time.
Source: Stiglitz 1992
and entrepreneurs have the opportunity to improve their efficiency level in the actual process
of producing. Transitional inefficiencies also can be addressed via more formal and typi-
cally economy-wide processes such as technology acquisition and licensing, improvements
in managerial and worker education to increase the quantity and quality of the stock of
human capital, and through organizational innovations like equity and bond markets and an
expansion in financial intermediation via banks and other such institutions that reduce the
costs of borrowing. Further, the operations of government vis-√†-vis the business sector can
be streamlined and made more efficient and transparent, and infrastructure, such as roads
and communications networks, with substantial positive externalities, can be modernized.
Some, perhaps all, of these initiatives are essential for surmounting transitional inefficiencies
in new firms.
The initial structural transformation 283
Figure 9.1 shows the effect of transitional inefficiencies in production: average and
marginal costs will be higher for new domestic producers compared to established foreign
producers providing a comparable import product. Particularly for production at less than the
socially optimal level ‚Ä“ where unit costs of production are minimized at the bottom of the
average total cost (ATC) curve ‚Ä“ when economies of scale have been fully realized, average
and marginal costs of producing are higher than they otherwise might be because of the
effects of transitional inefficiencies.
At relatively low levels of production like Qcurrent for an easy ISI firm just beginning
production, unit costs of production (along ATCnew) are higher than they would be at the same
level of output for existing producers providing imports to the local market (ATCestablished).
Compared to the per unit costs of production ATCO at the optimal level of production, QO,
for an already established foreign firm, the combination of transitional inefficiencies and low
production levels (i.e. unexploited economies of scale) when beginning ISI result in higher
costs of production per unit of output for new producers of ISI goods. These higher per
unit costs translate into higher prices for new domestic producers compared to experienced
foreign firms providing a country‚Ä™s imports.
For many small nations and even for some larger ones, the limited size of the domestic
market as a result of a small population, low average incomes, and a high degree of income
inequality can constitute a significant barrier to attaining the cost reductions consistent
with optimal production levels, even if the sources of the transitional inefficiencies are
removed (that is, domestic output still may be less than QO in Figure 9.1 because of limited
Q Q Output
Figure 9.1 Average costs of production, new versus established firms.
284 The Process of Economic Development
domestic demand).8 However, the domestic market is not the only potential outlet for the
output of easy ISI firms. There is no inherent reason why easy ISI goods that are first
produced for local consumption cannot be exported to foreign markets at some point in
the future. This would contribute to an expansion in demand that could help producers
achieve lower costs per unit through scale economies once and if transitional inefficien-
cies are overcome. We shall see that this is precisely what successful developing econo-
The critical issue for us to explore is how a new firm can confront the barrier of transitional
inefficiencies and be able to survive in the face of competition from imported goods. If the
firm does survive, it may be able to eventually compete head-to-head with existing foreign
firms. In other words, we need to consider how transitional inefficiencies might be overcome
so that a new domestic firm in an LDC can become as efficient as existing producers else-
where. At the end of the day, efficiency must be one of the primary goals of any successful
structural transformation process if government intervention in the market is ultimately to
Infant industry tariffs: overcoming transitional inefficiencies
The higher per unit costs of production for new domestic producers compared to estab-
lished foreign producers shown in Figure 9.1 translate into higher prices to consumers
for the goods produced by domestic firms compared to the prices of imports of the same
commodity, all else the same. Given the choice between a higher priced (and probably,
initially, lower quality) domestically produced good and a foreign imported good at a lower
price, there is little question which product most consumers will choose, particularly low-
income consumers. In the face of external competition and given the inevitability of tran-
sitional inefficiencies, new domestic producers can hardly be expected to compete at the
beginning of easy ISI. They may choose to not even initiate production under such circum-
stances, since the chances of success against established foreign firms supplying the import
market is so stiff.
It is the difference in average and marginal costs of production between potential domestic
producers and foreign firms ‚Ä“and the price differential the cost disparity implies ‚Ä“that under-
lies the argument for imposing infant industry tariffs on imported goods if domestic indus-
trialization is to be stimulated. Raising the final price of imported non-durable consumer
goods by adding an appropriate import tax on top of the price charged by the firm will make
the prices of domestically produced import substitutes more attractive to consumers in the
The need for such protection from external competition at the beginning of the
industrialization process is rooted in the difficulty for new producers to acquire the knowledge
associated with the entire process of production (also see Amsden 2001: 5‚Ä“6).
The fundamental rationale for protection is found in the tacitness of technology, which
implies that internationally competitive levels of productivity cannot be reached without
experience-based learning which entails comparatively high costs that must in some
way be financed.
(Evenson and Westphal 1995: 2284)10
A protective tariff is one way to finance this learning process. As a tax on a specific
import, it raises the price of the imported good thus making the higher-cost domestic output
The initial structural transformation 285
better able to compete, assuming no substantial consumer bias in favor of foreign goods.
The final price of the imported good to the consumer will be pushed above that of the
domestic import-substitute. This will augment the demand for the domestic import-substitute
as substitution by consumers takes place away from imports and toward domestic goods,
thus permitting the domestic producer to slide down the average cost curve (ATCnew in
Figure 9.1) as higher rates of domestic production fill the gap in demand left by the decrease
in imported goods.
From this perspective, infant industry tariffs may best be looked at as one form of invest-
ment in the economic progress of an economy. Prices are increased, yes, but that is necessary
to give new domestic firms a chance to expand production and learn to produce more effi-
ciently. For future growth possibilities, past and present ‚Äúmanufacturing experience matters‚ÄĚ
in learning to produce efficiently and without starting up and expanding industrialization,
such learning cannot take place. As Bruton (1989: 1607) writes: ‚ÄúAnother way of looking
at this cost [of infant economy protection] is possible: the reduced availability of goods and
services can be considered an investment.‚ÄĚ
There is no alternative to actually initiating production and participating in the
learning-by-doing required if the beneficial path dependence of an industrial-based
economy is to have any chance of being forged (Amsden 2001: Chapter 5). Easy ISI
may be an essential investment in the future human and economic development of late-
developing economies, as fundamental as investing in education, infrastructure, and new
An alternative to a tariff on imports to protect new easy ISI enterprises from lower-cost
foreign imports is a subsidy provided to domestic firms. A subsidy would lower the private
average costs of production of new enterprises and could provide the same degree of protec-
tion from imported goods as a tariff, and it would do so without necessarily increasing the
price of the good to the final consumer as a tariff will do. However, a subsidy creates a drain
on public resources which are already likely to be in short supply in most less-developed
nations. For this reason, infant industry tariff protection is a near universal feature of ISI
programs (other forms of non-tariff barriers also are used, such as regulations, quotas, and
Gerschenkron (1962: 44) noted that the more ‚Äúbackward‚ÄĚ a nation is the more likely it is
that industrialization will need ‚Äúto proceed under some organized direction; depending on
the degree of backwardness, the seat of such direction could be found in private investment
banks, in investment banks acting under the aegis of the state, or in bureaucratic controls.‚ÄĚ
This expresses Gerschenkron‚Ä™s important insight on the ‚Äúsubstitutability‚ÄĚ of instruments to
stimulate industrialization already mentioned in note 2 to this chapter. It will usually be
government providing ‚Äúorganized direction‚ÄĚ to the industrialization process, but there are
Tariff protection is but one instrument often used to promote easy ISI. Undervalued
exchange rates, low interest rates extended to particular borrowers, directed credit alloca-
tion via state-owned investment banks (see Focus 9.3), technology and R&D assistance,
education and training programs, planned government purchases of private domestic firm
output, and the creation of para-state firms are but a few of the alternatives available to public
policy-makers to initiate, sustain, and stimulate the industrialization process. The goal is to
create the desired structural transformation from an agriculturally-based economy to a more
productive industrial and service economy. Successful developing countries actually use a
combination of these instruments to promote industrialization, as we shall see here and in
the next chapter.
286 The Process of Economic Development
FOCUS 9.3 DEVELOPMENT bANKS AND ISI
Development banks, which are state-owned and state-directed non-commercial banks,
have been an important tool for directing a nation‚Ä™s savings and borrowed funds toward
higher-risk projects which can help foment development in some economies.
One of the most successful development banks has been Mexico‚Ä™s Nafinsa. In 1940,
Nafinsa was directed by the Mexican government to pursue the following objectives:
(1) promote industrialization; (2) promote the production of intermediate and capital
goods; (3) invest in infrastructure; (4) help stimulate and develop indigenous entrepre-
neurial talent; (5) build confidence within the Mexican private sector; and (6) reduce the
role of direct foreign investment in industry.
Nafinsa went through two major stages: aggressive promotion of industrialization from
1940 to 1947 and then promotion of infrastructure and heavy industry from 1947 to the
early 1960s. In making its investments, Nafinsa emphasized potential linkage effects.
An emphasis on linkages is well known today, but in the early 1940s, Nafinsa engaged
in innovative policy-making. In a classic study of Nafinsa, Calvin Blair emphasized the
‚Äúsystematic‚ÄĚ nature of Nafinsa‚Ä™s investments:
Nafinsa established in 1941 a department of promotion and began to make system-
atic studies of industrial development projects. With a predilection for manufacturing,
it promoted enterprises in practically every sector of the Mexican economy over the
course of the next several years. The roster of firms aided by loan, guarantee, or
purchase of stocks and bonds reads like a ‚Äúwho‚Ä™s who‚ÄĚ of Mexican business.
In addition to promoting para-state firms, Nafinsa engaged extensively in lending long-
term capital to the private sector and in forming partnership investments with both the
private sector and international firms. By 1961, Nafinsa‚Ä™s investments were supporting
533 industrial firms, and its long-term investments were twice as large as the sum of such
loans deriving from the private banking system. Between 1940 and 1980, the Mexican
economy grew at an average annual rate of 6 percent, after adjusting for inflation. Nafin-
sa‚Ä™s role was a major ingredient in what was then known as the ‚ÄúMexican Miracle.‚ÄĚ
Nafinsa continued to be a prominent development agency through the 1980s, but its
‚Äúgolden age‚ÄĚ was in the 1940s and 1950s, when the private sector‚Ä™s reluctance to commit
funds to industry was particularly acute.
By the early 1990s, all but thirteen of the hundreds of state-owned firms which Nafinsa
had helped to create had been privatized, merged, or liquidated. Nafinsa still continues
to play a role, now channeling funds into the export sector to promote the expansion of
In Brazil, a development bank known as BNDES was created to channel credit to stra-
tegic sectors. At the height of the so-called ‚ÄúBrazilian Miracle‚ÄĚ from 1975‚Ä“9, BNDES was
responsible for 48‚Ä“56 percent of all industrial investment. Known for its ‚Äúunusually effective
bureaucracy,‚ÄĚ which was also well-paid, BNDES was the prime catalyst for Brazil‚Ä™s strong
process of successful industrialization for decades. Brazil‚Ä™s economy grew at the annual rate
of 6 percent from 1947 to 1962, with industrial growth leading the way at an annual rate of
9‚Ä“10 percent. After a pause, extremely high rates of growth began again in 1968 and lasted
through the 1970s. It was precisely in the areas prioritized by BNDES and other key state
entities such as automobiles, steel, and public utilities where the fastest growth occurred.
[BNDES and other key executive groups] brought together managers of domestic and
foreign private industry and senior bureaucrats from various ministries ‚Ä¦These groups
planned the output and investment strategy of prioritized sectors and concomitantly
The initial structural transformation 287
created appropriate public policies to support private sector needs for credit, imports,
and domestic inputs. ‚Ä¦ the state elite worked closely with business to direct and
facilitate private production. These groups worked well ‚Ä¦ because planners and
implementers were the same people.
Sources: Blair 1964: 213; Kohli 2004; Krieckhaus 2002
The contribution of an infant industry tariff to the industrialization process
The effect of the infant industry tariff on the domestic market price, consumption, domestic
output and on the welfare of a less-developed nation imposing an infant industry tariff can
be shown using Figure 9.2. It is assumed here for simplicity that the import good currently
enters the domestic market with a zero tariff.
The current (or potential) domestic supply curve of socks is SD. SW is the world supply
curve of socks, which is drawn as perfectly elastic, reflecting the assumption that with free
trade the less-developed country can purchase any quantity of socks it wishes at a fixed free-
trade price, PF. In other words, the less-developed economy is assumed to be a price-taker for
socks on the international market.
DD is the domestic demand curve for socks which reflects the per capita income, income
distribution, population size and preferences of consumers. Given the premise of free trade
prior to initiating the ISI program, the price of socks to domestic consumers is equal to the
SW + t
a e SW
0 C C1 C2 C3 CF
Figure 9.2 Impact of an infant industry tariff.
288 The Process of Economic Development
world market price, PF. The total quantity of socks demanded in and supplied to the domestic
market will be CF, determined by the intersection of the SW supply curve and the domestic
demand curve, DD.
At the world market price PF, local firms will produce quantity C0 of socks, determined by
where PF crosses the domestic supply curve, SD.11 The quantity of imported socks in Figure
9.2 is equal to CF ‚ą’ C0 (total quantity demanded minus domestic production), which also
represents the potential market for import substitution.
If the less-developed economy imposes an infant industry tariff of a fixed amount, t, this
will shift the world supply curve of socks to SW + t.12 With the tariff, the price of the imported
good to consumers will be shifted upward to the new equilibrium price PT, where the new
supply curve, SW + t, is equal to the domestic demand curve, DD.
Domestic producers are now willing to supply a larger portion of the domestic market at
the higher price PT, potentially up to quantity C1. The price of the socks produced by domestic
firms, assuming they are of equal quality and in that sense a perfect substitute for the foreign
good, could be as high as PT. However, it is possible that some socks, equal to any quantity
less than C1 in Figure 9.2, might be sold by domestic producers at a price below PT but above
P0. This price-cutting behavior by domestic producers would reduce their profit, but it may
be necessary in those cases where the domestically produced good is not a perfect substitute
for the foreign import as a result of quality differences or a strong bias in favor of imported
goods by consumers. As shown in Figure 9.2, there continue to be imports of foreign socks
even with the tariff. Imports are equal to the quantity C3 ‚Ä“ C1, that is total quantity demanded
minus the quantity supplied by domestic enterprises.
What the imposition of the tariff does is to provide space for domestic firms to expand
production for the local market by shielding them from the competition and lower prices of
foreign imports. It also gives them breathing room so that production can be initiated and
expanded so that the possibility of learning to be more efficient has a chance of success.
Static and dynamic welfare effects of an infant industry
tariff: to do or not to do?
In analyzing the impact of a tariff, t, economists typically show the welfare loss to society by
examining the impact of the tariff on total consumer surplus. But there is more to the issue
than any possible immediate welfare loss, as we argue below. Let us first examine the effect
of the infant industry tariff on social welfare.
Prior to imposing the infant industry tariff on socks, total consumer surplus was equal to
the large triangle PMPFf in Figure 9.2. Consumer surplus measures the additional amount
consumers would have been willing to pay to buy quantity CF of the good but did not have to
pay because of the fact that all units of the good could be bought for price PF. The consumer
surplus is thus the triangle-shaped area above the prevailing price of the good, PF, and below
the demand curve, DD.
What is the effect of imposing the infant industry tariff? With the tariff, total consumer
surplus now is equal to the smaller triangle PMPTd. It is easy to see that there has been a
loss in consumer surplus equal to the area PFfdPT, i.e. the area of the trapezoid. That area
is simply the area left after subtracting the smaller triangle of consumer surplus after the
tariff from the larger triangle of consumer surplus that existed with free trade. Domestic
consumers clearly have been made worse off by the tariff. They are paying a higher price
for the good, are consuming less, and receive less consumer surplus than was the case at the
free market price.
The initial structural transformation 289
Who gains, if anyone, from imposing the infant industry tariff? How can such a tariff be
justified if consumers, many of whom are poor in an LDC, are made worse off by paying
higher prices and having less to consume?
First, not all of what consumers lose is a pure loss to society. Part of the lost consumer
surplus as a result of the tariff is collected by government as tariff revenues. These revenues
are equal to the area represented by the rectangle bcde, which is equal to the tariff, t, times
the quantity of imports remaining after the imposition of the tariff, C3 ‚ą’ C1.
Second, domestic firms producing the import substitute also gain from the tariff. After all,
the purpose of imposing an infant industry tariff in the first instance is to make it possible for
domestic firms to begin to produce so as to learn to compete with foreign producers. With the
tariff and the higher price of imports, domestic producers are able to increase their level of
production from C0 to C1. As a consequence, they receive additional producer surplus equal
to the area PFabPT.13
So, some of what consumers lose in consumer surplus from the tariff goes to government
as new tariff revenues and some of the lost surplus is received by domestic producers as
additional producer surplus at the higher level of domestic production, C1. Still, looking at
the area of lost consumer surplus in Figure 9.2, it is clear that after accounting for the new
tariff revenues and the increased producer surplus, some of the lost consumer surplus is not
received by anyone in society. A portion of the lost consumer surplus is a pure loss.
This loss is called the deadweight loss. It is equal to the sum of the areas of the two small
triangles, abc + def. 14 This is the remaining area of the lost (pre-tariff) consumer surplus that
is not transferred either to domestic firms as producer surplus or to government as increased
tariff revenues. economists argue that there is thus a net loss in total social welfare from the
imposition of the infant industry tariff compared to the free trade situation as measured by
the value of the deadweight loss.
What, then, can be the justification for infant industry tariffs except as a means to increase
tariff revenues or to augment the producer surplus of local firms? Consumers are worse
off, paying higher prices for fewer socks. How can such a tariff (or other market-distorting
policies ‚Ä“ see Focus 9.4) ever be defensible?
FOCUS 9.4 TAIWAN‚Ä™S EXPERIENCE WITH ISI IN TEXTILES
In the early 1950s, the Taiwanese government paid particular attention to the textile industry
as part of the core of a loosely formulated plan for industrial development. The first textile
producers were mostly relocated mainlanders, so the industry did not arise de novo.
Nevertheless, a whole battery of market-distorting and even market-replacing methods
was used to establish the industry quickly. The market-distorting methods included tariffs
and quantitative restrictions on imports of yarn and finished products, restrictions on the
entry of new producers to prevent ‚Äúexcessive‚ÄĚ competition, and controlled access to raw
materials. From 1951 to 1953, a government agency, with help from the US, replaced
market allocation altogether. It supplied raw cotton directly to the spinning mills, advanced
all working capital requirements, and bought up all production ‚Ä“ and did basically the
same at the weaving stage.
The supply response was dramatic. Between 1951 and 1954, production of cotton yarn
went up by over 200 percent and woolen yarn rose by over 400 percent. By mid-1953
Taiwan was more than self-sufficient in yarn and cloth. There was no need for imports of
these goods, and domestic producers, workers, and consumers were the winners from
such promoted development.
Source: Wade 1990: 79
290 The Process of Economic Development
Remember, the purpose of an infant industry tariff is to assist an ISI industry in getting
started and to give it time to overcome its transitional inefficiencies. If these new domestic
firms succeed in becoming more efficient, then the domestic supply curve, SD, will shift
outward and downward toward SD1 (and curve SD will no longer exist). If the domestic sock
industry becomes internationally competitive, the domestic supply curve will shift all the
way to SD2 in Figure 9.2 (and neither SD nor SD1 will exist).
What would cause the domestic supply curve to shift in this fashion? First, an increase in
the number of domestic producers who are able to emerge and produce behind the protective
wall provided by the tariff could shift the supply curve outward. Second, and of greater
importance, the domestic supply curve would be shifted outward as a consequence of:
1 increased efficiency of domestic enterprises through the use of ‚Äúcutting-edge‚ÄĚ or ‚Äúbest-
2 the more effective application of whatever technology already is available, that is, via
positive ‚Äútechnical efficiency change‚ÄĚ;
3 better training of managers and workers in schools or special institutes;
4 ‚Äúlearning-by-doing‚ÄĚ on the job, whereby both workers and managers become more
efficient with practice, adaptation and even trial-and-error in the process of producing
5 the application of more effective management and financial techniques and quality
6 improvements in banking and financial institutions to facilitate financing of production;
7 the improvement of infrastructure, such as roads, communications, ports, power, and so on.
All of these factors, and others could be added, would contribute to reducing marginal and
average costs of production with the outcome being a shift in the domestic sock industry‚Ä™s
total supply curve toward SD2.
It is easy to see from the graph that with the full elimination of transitional inefficiencies
so that SD2 intersects DD at CF, consumer surplus will be equal what it was prior to the
imposition of the tariff (area PFPMf) if price is again at PF. If that happens, however, the
domestic producer surplus is larger than in the pre-ISI free-trade regime as the supply
curve has shifted outward allowing domestic firms to capture revenues previously received
by foreign firms.15 With domestic production at CF there would be both more industrial
employment and a higher level of GDP produced by the country than was the case when the
free-trade regime prevailed, assuming a shift of labor from agriculture to industry to produce
the new output.
What the imposition of an infant tariff, t, can facilitate, then, is the attainment of dynamic
welfare gains that, accruing over time, can quite easily swamp the static welfare losses that
the tariff initially imposed on the economy. As a means to promote the transformation of an
economy from agriculture to industry and to shift workers from lower to higher productivity
employment, infant industry ISI tariffs can foster significant dynamic welfare gains which
can easily outweigh any short-run deadweight loss of uncompensated consumer surplus.
With industrialization, a country can realize both higher total income and higher income per
person. Seeing the deadweight loss as one of the costs of development, as an investment in
the future, assuming the ISI industries become more efficient, helps to put everything into
The use of an infant industry tariff is one example of how ‚Äúgetting prices wrong‚ÄĚ (in this
case, raising prices of socks above the international market level with a tariff) can yield
The initial structural transformation 291
positive development outcomes. Rather than abdicating control over the economy to past
adverse path dependence and to current unfettered market forces, the conscious forging of
an easy ISI sector, as in Korea, Taiwan, Brazil, Mexico, and many other large less-developed
nations, can be a means to bend domestic resource allocation, production and labor usage in
more productive directions via state intervention which eventually is less necessary as the
market system develops.
Such ‚Äúgoverning‚ÄĚ of the market, as Wade (1990) calls it, does lead to market-distorting
and perhaps even market-replacing policies, at least for a time. That should not be the issue;
what should be asked is: What are the effects of such policies? When these policies are
carried out with care and monitored for results and where vested interests have difficulty
in influencing state actions, the practice of ‚Äúgetting prices wrong‚ÄĚ by governing the market
rather than surrendering to the neoclassical and orthodox policy of ‚Äúgetting prices right‚ÄĚ
and accepting existing market forces as optimal can result in substantial gains in output,
income, and human development by accelerating the pace of industrialization. There are
an increasing number of modern examples that support this view, from Korea to Taiwan to
Chile to China and beyond.
The elimination of infant industry protection: when is enough
How long should infant industry protection from lower-cost imports be extended to domestic
firms? In other words, how long should it take until ISI firms are able to compete head-to-
head with foreign imports without the artificial benefit of tariff protection?
From Figure 9.2, once transitional inefficiencies are overcome and the domestic supply
curve has increased to SD2, tariff protection becomes redundant and unnecessary. In that
case, domestic industries can compete directly with foreign imports and at world prices,
since domestic firms will have attained the same level of efficiency at output CF as producers
In fact, as firms begin to achieve greater efficiency through learning-by-doing the need
for a tariff at the original level, t, is diminished. It is perilous, perhaps, to put a number to
this, but in the non-durable consumer goods industries in which ISI begins, seven to ten
years would seem to be a reasonable target date for ending protection and for expecting
domestic producers to have overcome the basic transitional inefficiencies encountered at the
beginning of easy ISI.16 For more complicated products with longer learning curves or where
technology is more complex, somewhat longer transition periods may be warranted. For some
goods the transitional period may be less than seven years. In instances where a society‚Ä™s
overall level of human capital accumulation and its technological and R&D capacity needs
to be improved through relatively large social investments in education and health care, the
period of tariff protection may need to be lengthened. This will allow time for the necessary
‚Äúinitial endowments‚ÄĚ to be formed so as to provide domestic firms a reasonable opportunity
to become competitive with international enterprises.
Whatever the specific plan for the termination of infant industry protection, what is critical
is that a timetable for the phasing-out of tariff protection be part of government policy and
be announced to potential firms when infant industry tariffs are introduced. In this way,
domestic producers do not become complacent, thinking and acting as if protection is to be a
permanent fixture of the economic landscape that permits higher-cost, less efficient domestic
producers to prosper, immune from outside competition.
Domestic producers must anticipate, with certainty, that protective tariffs will end at a
292 The Process of Economic Development
determinate date, and they must either be competitive with foreign imports at that time by
having overcome any transitional inefficiencies (as on supply curve SD2 in Figure 9.2), or
suffer the consequences when foreign imports are permitted to re-enter the domestic market
on an equal footing with domestic production.
The phasing-out of tariffs can be uniform over the protected period or accelerating. For
example, if the domestic beverage industry is initially provided a 25 percent protective tariff,
to be phased out over five years, then 5 percent of tariff protection could be removed at the end
of year 1, a further 5 percent at the end of year 2, and so on until all compensatory protection
has been lifted (there still might be a minimal tariff on beverages after the protective tariff
is removed to cover the administrative costs of the customs service). Alternative schemes
might leave the full 25 per cent protection for two years; remove 5 percent at the end of
year 3; remove 10 percent at the end of year 4; and at the end of year 5 remove the final 10
percent of protection. Whatever the phasing-out pattern, it should be clear to enterprises that
it is non-negotiable after start-up and that the end period for eliminating tariff protection is
These tariffs are called infant industry tariffs for a reason; they are applied to new firms
and industries with short-term inefficiencies. The language is evocative of a presumed reality
and that is that infants do (and should!) mature and that transitional inefficiencies are indeed
transitory. When the firms mature and have overcome those start-up inefficiencies, there is
no reason to maintain the tariffs any longer.
If tariffs on ISI industries are not phased out, then they can become a substantial internal
barrier to progress. Maintaining infant industry tariffs too long not only threatens further
industrialization, but also the possibility of reaching higher levels of growth, development,
and social and human welfare that are the motivating forces for initiating easy ISI in the
first place. If tariff protection is not withdrawn from domestic producers, there is a strong
likelihood that the static deadweight losses shown in Figure 9.2 will persist as permanent
deadweight losses, thus sacrificing the potential dynamic welfare gains that should be the
motivating force behind implementing infant industry tariffs. No one can be in favor of such
an outcome, and as we shall see below and in the next chapter, in those countries that did
not draw down infant industry tariff protection, economic progress was compromised and
Infant industry tariff protection can only be a phase, and a temporary one at that, on the
path to fuller industrialization and a higher standard of living.
The importance of embedded state autonomy to successful ISI
One of the legitimate concerns many economists have had about countries that pursue an