Figure 7.2(b) The South Korean state: the developmental state.
The state as a potential agent of transformation 233
defined by institutional linkages between the state and society which are strong, versatile,
and transformative. In short, state autonomy is embedded in society.
Consequently, the South Korean state constitutes a developmental state, with an endog-
enous basis of transformation capable of channeling both public and private investment to
strategic sectors. Numerous attempts have been made to distort or discredit the record of the
Korean state, particularly in the aftermath of the Asian crisis of 1997, because it constitutes
the greatest challenge to the neoliberal theory. However, Ha-Joon Chang has demonstrated
that the Asian crisis did not arise from any endemic failings of state intervention. Rather, it
was the turn away from industrial policy after 1993 and the weakening of the state apparatus
which facilitated duplicative investments in the 1993‚Ä“7 period (Chang 2000). As we will see
in the following chapters, the concepts of the state developed here have wide application and
profound implications for the process of development.
Summary and conclusions
Our journey through the debate over the role of the state in the developmental process has
led to the insights offered by Peter evans and many others. The archetype constructions
of the non-cohesive, fragmented intermediate and developmental states will be of use in
understanding the role states play in given societies. Can intermediate states become devel-
opmental states? Evans argues that such a metamorphosis can be a daunting, but not an
existing public bureaucracies, certainly in Latin America and even to some degree in
Africa, are not the irredeemable cesspools of incompetence that they are sometimes
painted by neoliberal rhetoric. There are dedicated individuals working in public service
in almost all countries ‚Ä¦ Most state bureaucracies, at least in Latin America, contain
key agencies that at least in certain periods have displayed many of the institutional
traits of ‚ÄúWeberian bureaucracies‚ÄĚ. Such agencies have performed with levels of effi-
ciency comparable to those [of] their East Asian counterparts, despite being surrounded
by a public sector that is much less effectual overall.
(Evans 1998: 80)
He warns that: ‚ÄúThe concept of embedded autonomy is a useful analytical guidepost, not
an engineering formula that can be applied, with a few easy adjustments‚ÄĚ (Evans 1995: 244).
The goal of state policy must be to induce private investment and to channel such investment
to strategic sectors. But embeddedness is easier to describe than attain. ‚ÄúConnecting state
and society is the more difficult problem. Capacity without connection will not do the job‚ÄĚ
(ibid.: 244). Evans also warns us that the state-led successes in East Asia will not be easy to
follow, because of the fact that state and society were well connected there as a result of what
may be rather unique historical circumstances. Still, evans illuminates a path to follow, while
showing there is no simple road map to development. At the same time, Atul Kohli, after
exhaustively examining the role of the state in Brazil, India, Korea, and Nigeria, reminds us
that we ignore at our own peril the fact that:
There are hardly any significant examples in the developing world, now or in the
recent past, where industrialization has proceeded very far without state intervention.
The underlying reason is simple but powerful: Private investors in late-late-developing
countries need organized help, help that effective states are most able to provide to
234 The Process of Economic Development
overcome such obstacles as capital scarcity, technological backwardness, rigidities in
labor markets, and to confront the overwhelming power of foreign corporations and
of competitive producers elsewhere. ‚Ä¦ state intervention in more or less successful
industrializers varies, not so much by quantity as by type and quality. It is therefore
patterns of state intervention in the economy that are key to explaining successful late-
(Kohli 2004: 377)
Questions and exercises
1 Using the categories of non-cohesive, fragmented intermediate, and developmental
states, review the discussion of the critique of state policies and practices offered by
the neoliberals. How do these categories help you better understand the role of the state
and the dangers of generalizing too broadly on the weakness of state intervention from
2 Evans emphasizes that states have to play specific roles in order to promote develop-
ment. Review the discussion of Deepak Lal and determine if the issues raised by Lal
can be understood with greater precision with the use of Evans‚Ä™s categories. Why, or
3 Review the case of Nigeria and Lord Bauer‚Ä™s analysis of West Africa, along with the
discussion of non-cohesive state formation, as presented in this chapter. What are the
lessons to be learned regarding the role of the state versus the role of the market in this
4 Professor Krueger has maintained that the solution to rent-seeking on the part of state
functionaries is the shrinkage of the state. Peter Evans‚Ä™s perspective is quite different.
Compare their contrasting approaches to rent-seeking. Where do they agree, and when
and why do they disagree?
5 Under what social and economic conditions are markets likely to fail? When is there
likely to be government failure? Is macroeconomic inefficiency (for example, a high
level of unemployment and underemployment) due to market failure or government
6 Review Figure 7.1 and write a two-page essay on the state using all the terms and
concepts from this figure to explain what is unique and important regarding the concept
of the developmental state. How would you change this figure if you were to explain the
idea of the fragmented intermediate state?
1 The classic study of the term ‚Äúliberalism‚ÄĚ is Girvetz (1963).
2 Lugard was an administrator of great renown and influence. He served as High Commissioner
of Northern Nigeria from 1900 to 1906, and then as Governor of Nigeria from 1912 to 1919.
After retirement he published a landmark work on colonial administration, The Dual Mandate.
According to P.J. Cain and A.G. Hopkins, those who followed Lugard into colonial careers ‚Äúwere
inclined to idealize rural Africa, to identify with ‚Ä˜natural‚Ä™ pastoralists and cultivators, and to view
urbanized and supposedly ‚Ä˜detribalized‚Ä™ Africans with a mixture of disdain and alarm‚ÄĚ (Cain and
Hopkins 1993: 218).
3 Some mining occurred. Tin accounted for about 10 percent of Nigeria‚Ä™s exports before the Second
4 Independence in 1960, unfortunately, did not carry with it a deep and profound consolidation of
the hopeful trends and forces which had surfaced in the mid-1950s. From 1954 to 1960, Nigeria
The state as a potential agent of transformation 235
experienced a very modest increase in annual per capita income of 0.32 percent. From 1960 to
1966, there was an annual decline of 0.72 percent in per capita income. In his interesting discussion
of the Nigerian case, Tom Kemp argued that the Nigerian elite could not capture the momentum of
the developmental strategy which was colonial rule‚Ä™s last effort in Nigeria. Nigeria‚Ä™s leaders had
sprung from the prosperous peasantry, and their largely positive historical experience with market
forces led them to adopt a relatively uncritical and passive posture regarding the limitations of
markets. Thus, while the elite dabbled in developmental projects, their lack of determination and
commitment to a strategy of ‚Äúgoverning the market,‚ÄĚ as it is now termed, led to a downward spiral
of dependence and decay:
Already privileged economically and in the possession of educational advantages, they
accepted personal advancement and enrichment as valid goals, modelling their expectations
very much upon the life-style of expatriate europeans in colonial times. The way was thus
open for cooperation and collusion between politicians and civil servants on the one hand and
indigenous entrepreneurs, of whom there was no lack of aspirants, and foreign businessmen
on the other. Speculation and corruption became endemic in the new state, intertwined with
regional and ethnic rivalries and favoritism which were to plague the country, bringing insta-
bility and a tragic war within less than a decade.
(Kemp 1989: 184)
5 One particular target has been state-owned industries, or para-statals. The hypothesis of the
neoliberals is that any activity which conceivably could be undertaken in a market environment
will be more efficient if it is operated on a ‚Äúfor profit‚ÄĚ basis. (Implicitly this approach assumes that
the less-developed country has an efficient set of markets already in place.) Thus, in order to cut
the state to its ‚Äúproper‚ÄĚ size, the neoliberals have strongly advocated privatization of state-owned
Research into the issue of efficiency of state-owned industries is limited. However, one
summary of the literature for Mexico, which has carried out the largest wave of privatizations
in the less-developed world, indicates that many of the state-owned industries were no more
inefficient than private-sector firms, while other state firms were operated ‚Äúinefficiently‚ÄĚ on
purpose in order to subsidize private sector accumulation and/or to achieve other broad political
goals. In other words, even though the employees and managers of these firms were government
employees, that fact alone did not suffice to explain the efficiency level of the firm (Cypher 1990:
6 The entire theme of the dynamic relationship between the state and the market has been brilliantly
explored in a twentieth century classic, Karl Polanyi‚Ä™s The Great Transformation (Boston: Beacon
7 We will discuss the impact of openness on the domestic economy further in subsequent chapters,
especially 9 and 10. Some neoliberals have emphasized the possibility of accelerating technical
change under the ‚Äúdo-or-die‚ÄĚ imperative, while others have suggested that under such pressures
indigenous entrepreneurial capabilities might blossom.
8 Rent-seeking is often subsumed under the more general category of ‚Äúdirectly unproductive profit-
seeking‚ÄĚ (DUP) activities, such as lobbying, smuggling, bribery, monopoly, or any other activity
which generates profits, but produces no goods or services directly.
9 Notice that the predatory state describes some of the features noted by Krueger, earlier in this
chapter, under the heading of the absolutist state. Here, in contrast to Krueger, the dysfunc-
tional state is viewed as an outgrowth of a specific historical condition. It is the ‚Äútraditional,‚ÄĚ or
pre-capitalist (semi-feudal/semi-capitalist) society, perhaps imbued with the ethos of merchant
capital, which results in the absolutist state. Krueger assumes that by shrinking the state the
society will be strengthened, because there is assumed to exist an ordered, structured, market-
based economy. Evans avoids the dichotomous assumption ‚Äúbad state/good civil society.‚ÄĚ Here a
chaotic socio-economic system is complemented by a predatory state apparatus. Destroying such
a state while leaving the socio-economic system untouched would not solve the basic underlying
10 Notice the surface similarities between Evans‚Ä™s intermediate state and Krueger‚Ä™s factional state.
Krueger finds rent-seeking, which distorts the intention of government policy by locking in place
subsidies and other forms of vested interests. Her solution is shrinking the state. For evans, rents
236 The Process of Economic Development
arise not from too much intervention, but from too little effective intervention, because of a lack of
an adequately trained, professional civil service.
While far from the ideal, the intermediate state exhibits ‚Äúpockets of efficiency‚ÄĚ in the state
sector, something Krueger rules out by assumption. Such ‚Äúpockets‚ÄĚ are hardly trivial; they form
the basis for the creation of whole new industries. They redefine the production base, creating new
forms of comparative advantage. Krueger‚Ä™s approach cannot accommodate such activities, which
Evans presents in some detail. And, therefore, the new political economy approach has difficulty
explaining the economic development of such nations as Brazil, India, or Mexico, all important
instances of the intermediate state.
11 The concept of demiurge is not the same as what Lal called ‚Äúdirigiste dogma.‚ÄĚ Lal refers to a situa-
tion wherein the government seeks to supplant the market, rather than supplement it. The demiurge
function is limited to the creation of state-owned enterprises, or what are sometimes known as para-
Adelman, Irma. 2001. ‚ÄúFallacies in Development Theory,‚ÄĚ pp. 103‚Ä“34 in Gerald Meier and Joseph
Stiglitz (eds.), Frontiers of Development Economics. Oxford: Oxford University Press.
Amsden, Alice. 1989. Asia‚Ä™s Next Giant. Oxford: Oxford University Press.
‚Ä”‚Ä”. 2001. The Rise of ‚Äúthe Rest.‚ÄĚ Oxford: Oxford University Press.
Bauer, P.T. 1984. ‚ÄúRemembrance of Studies Past,‚ÄĚ pp. 27‚Ä“43 in Gerald Meier and Dudley Seers (eds.),
Pioneers in Development. Oxford: Oxford University Press.
Beatty, Jack. 2007. Age of Betrayal: The Triumph of Money in America, 1865‚Ä“1900. New York:
Bhagwati, Jagdish and Anne Krueger. 1973. ‚ÄúExchange Control Liberalization and Development,‚ÄĚ
American Economic Review 63 (June): 419‚Ä“27.
Cain, P.J. and A.G. Hopkins. 1993. British Imperialism: Crisis and Deconstruction 1914‚Ä“1990.
Chang, Ha-Joon. 2000. ‚ÄúThe Hazard of Moral Hazard: Untangling the Asian Crisis,‚ÄĚ World Development
‚Ä”‚Ä”. 2002. Kicking away the Ladder: Development Strategy in Historical Perspective. London:
‚Ä”‚Ä”. 2007. Bad Samaritans: Rich Nations, Poor Policies, and the Threat to the Developing World.
London: Random House.
Cheng, Tun-jen. 1990. ‚ÄúPolitical Regimes and Development Strategies,‚ÄĚ pp. 139‚Ä“78 in Gary Gereffi
and Donald Wyman (eds.), Manufacturing Miracles. Princeton: Princeton University Press.
Chibber, Vivek. 2003. Locked in Place: State Building and Late Industrialization in India. Princeton:
Princeton University Press
Cypher, James. 1990. State and Capital in Mexico. London and Boulder, CO: Westview Press.
‚Ä”‚Ä”. 2001a. ‚ÄúDeveloping Disarticulation within the Mexican Economy,‚ÄĚ Latin American Perspectives
28 (May): 11‚Ä“37.
‚Ä”‚Ä”. 2001b. ‚ÄúNAFTA‚Ä™s Lessons: From Economic Mythology to Current Realities,‚ÄĚ Labour Studies
Journal 26 (Spring): 5‚Ä“21.
‚Ä”‚Ä”. 2005. ‚ÄúThe Political Economy of the Chilean State in the Neoliberal Era: 1973‚Ä“2005,‚ÄĚ Canadian
Journal of Development Studies 24 (4): 763‚Ä“80.
Delgado Wise, Ra√ļl and James Cypher. 2007. ‚ÄúThe Strategic Role of Mexican Labor under NAFTA:
Critical Perspectives on Current Economic Integration,‚ÄĚ Annals of the American Academy of Political
and Social Science 610 (March): 120‚Ä“42.
Evans, Peter. 1995. Embedded Autonomy: States and Industrial Transformation. Princeton, NJ: Princeton
‚Ä”‚Ä”. 1998. ‚ÄúTransferable Lesson?: Re-examining the Institutional Prerequisites of East Asian
Economic Policies,‚ÄĚ Journal of Development Studies 34 (August): 66‚Ä“86.
Fieldhouse, David K. 1967. The Colonial Empires. New York: Delacorte Press.
The state as a potential agent of transformation 237
Fine, Ben. 2006. ‚ÄúThe New Development Economics,‚ÄĚ pp. 1‚Ä“20 in Jomo K.S. and Ben Fine (eds.), The
New Development Economics: After the Washington Consensus. London: Zed Books.
Girvetz, Harry K. 1963. The Evolution of Liberalism. New York: Collier Books.
Haberler, Gottfried. 1987. ‚ÄúLiberal and Illiberal Development Policy,‚ÄĚ in Gerald Meier (ed.), Pioneers
in Development, 2nd series. Oxford: Oxford University Press.
Helleiner, Gerald. 1966. Peasant Agriculture, Government, and Economic Growth in Nigeria. Home-
wood, IL: Irwin.
Herring, Ronald. 1999. ‚ÄúEmbedded Particularism: India‚Ä™s Failed Developmental State,‚ÄĚ pp 306‚Ä“34 in
Meredith Woo-Cumings (ed.), The Developmental State. Ithaca: Cornell University Press.
Hopkins, A.G. 1973. An Economic History of West Africa. London: Longman.
Kahn, Joseph and David Barbaoza. 2007. ‚ÄúAs Unrest Rises China Broadens Workers‚Ä™ Rights,‚ÄĚ New York
Times (June 30): A1, A5.
Kahn, Mushtaq Husain. 2006. ‚ÄúCorruption and Governance,‚ÄĚ pp. 200‚Ä“21 in Jomo K.S. and Ben
Fine (eds.), The New Development Economics: After the Washington Consensus. London: Zed
Kemp, Tom. 1989. Industrialization in the Non-Western World. London: Longman.
Kohli, Atul. 1999. ‚ÄúWhere do High-Growth Political Economies Come From? The Japanese Lineage
of Korea‚Ä™s ‚Ä˜Developmental State‚Ä™,‚ÄĚ pp. 93‚Ä“136 in Meredith Woo-Cumings (ed.), The Developmental
State. Ithaca: Cornell University Press.
‚Ä”‚Ä”. 2004. State-Directed Development. Cambridge: Cambridge University Press.
Krueger, Anne. 1993. Political Economy of Policy Reform in Developing Countries. Cambridge, MA:
Lal, Deepak. 1985. The Poverty of Development Economics. Cambridge, MA: Harvard University
Lipton, Michael. 1984. ‚ÄúComment [on Bauer],‚ÄĚ pp. 40‚Ä“50 in Gerald Meier and Dudley Seers (eds.),
Pioneers in Development. Oxford: Oxford University Press.
Little, Ian, Tibor Scitovsky and Maurice Scott. 1970. Industry and Trade in Some Developing Coun-
tries. Oxford: Oxford University Press.
Meier, Gerald. 1990. ‚ÄúTrade Policy and Development,‚ÄĚ pp. 155‚Ä“69 in Maurice Scott and Ian Little
(eds.), Public Policy and Economic Development. Oxford: Oxford University Press.
Myint, Hla. 1987. ‚ÄúNeoclassical Development Analysis: Its Strengths and Limitations,‚ÄĚ pp. 107‚Ä“36 in
Gerald Meier (ed.), Pioneers in Development, 2nd series. Oxford: Oxford University Press.
Morris, Cynthia and Irma Adelman. 1988. Comparative Patterns of Economic Development: 1850‚Ä“
1910. Baltimore: Johns Hopkins Press.
Pempel, T.J. 1999. ‚ÄúThe Developmental Regime in a Changing World Economy,‚ÄĚ pp. 137‚Ä“81 in
Meredith Woo-Cumings (ed.), The Developmental State. Ithaca: Cornell University Press.
Pingl√©, Vihba. (1999). Rethinking the Developmental State. New York: St. Martin‚Ä™s Press.
Rodrik, Dani. 1999. The New Global Economy and Developing Countries: Making Openness Work.
Baltimore: Johns Hopkins Press.
Schneider, Ben Ross. 1999. ‚ÄúThe Desarrollista State in Brazil and Mexico,‚ÄĚ pp. 276‚Ä“305 in Meredith
Woo-Cumings (ed.), The Developmental State. Ithaca: Cornell University Press.
Shapiro, Helen and Lance Taylor. 1990. ‚ÄúThe State and Industrial Strategy,‚ÄĚ World Development 18
Srinivasan, T.N. 1984. ‚ÄúComment [on Bauer],‚ÄĚ pp. 51‚Ä“8 in Gerald Meier and Dudley Seers (eds.),
Pioneers in Development. Oxford: Oxford University Press.
Stiglitz, Joseph. 2001. ‚ÄúMore Instruments and Broader Goals: Moving toward the Post-Washington
Consensus,‚ÄĚ pp. 17‚Ä“56 in Ha-Joon Chang (ed.), The Rebel Within: Joseph Stiglitz and the World
Bank. London: Anthem Press.
Toye, John. 1987. Dilemmas of Development. Oxford: Basil Blackwell.
UN Development Programme. 2003. Human Development Report 2003. New York: Oxford University
‚Ä”‚Ä”. 2006. Human Development Report 2006. New York: Oxford University Press.
238 The Process of Economic Development
Wade, Robert. 1990. ‚ÄúIndustrial Policy in East Asia: Does It Lead or Follow the Market?,‚ÄĚ in Gary
Gereffi and Donald Wyman (eds.), Manufacturing Miracles. Princeton: Princeton University Press.
Weisman, Steven. 2006. ‚ÄúDrive on Corruption Angers Some at World Bank,‚ÄĚ New York Times
(September 14): C1, C4.
World Bank. 1993. The East Asian Miracle. Oxford: Oxford University Press.
‚Ä”‚Ä”. 2002. World Development Report 2002. Oxford: Oxford University Press.
‚Ä”‚Ä”. 2006. World Development Report 2006. Oxford: Oxford University Press.
8 Endogenous growth theories and
new strategies for development
after reading and studying this chapter, you should better understand:
‚ÄĘ the income convergence controversy;
‚ÄĘ the basic structure of endogenous growth models and how they differ from the
classical and neoclassical growth theories;
‚ÄĘ the importance of human capital, learning-by-doing, specialization, research and
development, and other positive externalities that can permit rapid growth rates
over the long term;
‚ÄĘ the significance of constant and increasing returns in the endogenous growth models
and how these contribute to long-run growth and increased income levels;
‚ÄĘ the possible effects of inequality of income and land distribution on the rate of
‚ÄĘ the importance of social infrastructure and other alterable initial endowments to
the rate of economic growth;
‚ÄĘ the significance of ‚Äútechnical efficiency change‚ÄĚ to economic growth.
Near the end of Chapter 4, the Solow neoclassical growth theory was introduced. It has been
interpreted as predicting that the per capita incomes of economies will tend to converge to the
same level over time. This happens as lower income nations grow faster than higher income
nations, assuming they all have access to the same technology and share similar savings,
investment, and population growth rates.1 Solow‚Ä™s theoretical structure validated the policy
recommendations of the early development economists and their policy-oriented theories,
such as the ‚Äúbig push,‚ÄĚ ‚Äúbalanced growth,‚ÄĚ and ‚Äúunbalanced growth‚ÄĚ strategies considered
in Chapter 5. You will remember that these focused on the expansion of the industrial capital
stock and an increase in the rate of savings as the means to promote more rapid economic
growth and higher income per capita.
The primacy of physical capital accumulation, the K variable in the Solow-type aggregate
production function, has been a focus for development economists since the time of Adam
Smith. In this view, countries need to save and invest sufficiently so as to augment their total
physical capital stock if they are to reach higher income levels. A higher level of investment
boosts the attainable level of income by increasing the productivity of each worker, who
has more physical capital with which to work. Of course, it was recognized that many poor
240 The Process of Economic Development
nations would have trouble accumulating physical capital, as their low level of income meant
limited savings to finance new investments. Because of low domestic incomes and savings,
the level of capital accumulation might remain insufficient to achieve higher levels of per
However, as discussed in Chapter 5, flush with the success of post-Second World War
reconstruction in europe and Japan, most early development economists believed that
domestic saving and investment could be complemented by external financing. Total
saving and investment, and hence total income, need not be constrained by a shortage of
domestic resources in less-developed nations. Rather, saving is equal to S = SD + SF, where
S is total saving, SD is domestic saving by households, business, and government and SF is
foreign saving in the form of loans, aid, and foreign direct investment. The level of invest-
ment of a country was not believed to be rigidly determined by that economy‚Ä™s own limited
ability to save, thus breaking the link between low income and low savings that seemed to
constrain the possibilities of the poorest of the less-developed nations to a ‚Äúvicious circle
of poverty‚ÄĚ where being poor becomes the equilibrium outcome. External resources could
add to and complement domestic financing for investment and thus spur economic growth
The income convergence controversy
The 1950s and 1960s were periods of relative optimism among development economists
and policy-makers in evaluating the prospects for the newly independent former colonies
that emerged as new nations after the Second World War. The economic models of the day
suggested that growth was essentially a technical problem to be solved through larger injec-
tions of physical capital and secondarily through measures to slow population expansion.
economists had no shortage of recommendations as to the strategic means for adding to that
stock of capital, as we learned in Chapters 4 and 5.3
Further, the convergence of world income levels predicted by Solow-type neoclassical
growth models reassured economists and policy-makers alike that world poverty might
reasonably be expected to be eradicated within the not-too-distant future if countries could
just undertake the right policies. The relatively rapid rates of growth of some countries in
the less-developed world compared to the already-developed nations in the period after the
Second World War seemed to corroborate these hopeful theoretical prognostications, at least
for a time.
As Table 8.1 shows, data to support income convergence to that of the high-income
economies, particularly for some regions, can be found. The rate of growth of income person
(as measured by GDP per capita) in East Asia and the Pacific has been consistently a multiple
of income growth in the high-income economies over the entire 1970‚Ä“2005 period and in
every sub-period. South Asia‚Ä™s rate of growth of income since 1980 also exceeded that of the
high-income economies. These regions‚Ä™ growth experience seemed to support the income
convergence hypothesis of neoclassical economics as their more rapid growth rates brought
their incomes relatively closer to those of the high-income countries.4
However, there is at the same time evidence of income divergence between poor and rich
regions. The income growth of the fifty-four low-income economies was close to, but still
less than, the growth rate of the fifty-six high-income economies as shown in the last two lines
of the table. As a result, average real income levels of high- and low-income economies were
further apart in 2005 than they had been in 1970. Sub-Saharan Africa‚Ä™s growth in GDP per
person was virtually nil over more than three decades, meaning a widening of the income gap
Endogenous growth theories and new strategies for development 241
Table 8.1 Comparative growth rates
2005 GNI Average annual growth rate of GDP per capita in constant 2000 dollars
per capitaa (percentage)
1970‚Ä“5 1975‚Ä“80 1980‚Ä“5 1985‚Ä“90 1990‚Ä“5 1995‚Ä“2000 2000‚Ä“5 1970‚Ä“2005
East Asia and 1,627 3.8 5.2 5.9 5.8 8.9 5.3 7.3 6.0
Latin America 4,008 3.4 2.9 ‚ą’1.6 ‚ą’0.16 1.7 1.6 0.9 1.3
2,241 1.5 ‚ą’1.2 1.4 2.4 1.1b
Middle east n.a. 0.7 2.2
South Asia 684 ‚ą’0.03 1.3 3.1 3.6 3.0 3.5 4.6 2.7
745 ‚ą’1.7 ‚ą’0.35 ‚ą’1.4 0.1
Sub-Saharan 2.0 0.07 0.8 2.0
580 0.4 0.9 2.3 1.7 2.9 4.1
Low-income 2.7 2.0
35,131 2.4 1.9 2.9 1.7 1.5
High-income 2.8 2.0 2.2
Source: World Bank, World Development Indicators Online.
a In current US dollars; GNI is gross national income, what used to be called gross national product.
between that region and every other. Nor did Latin America and the Caribbean or the Middle
east enjoy income convergence to the high-income economies. Further, these regions lost
ground compared to East Asia and the Pacific in terms of relative income levels among the
less-developed economies themselves.
The economic growth data thus provides contradictory evidence for income convergence
and for the neoclassical hypothesis that one should expect faster growth rates for countries
with lower incomes. In some cases, for example, East Asia and the Pacific, the neoclas-
sical theory would seem to fit. For other regions, like Sub-Saharan Africa, just the opposite
appears true. As one of the world‚Ä™s poorest regions, Sub-Saharan Africa has had the worst
overall growth performance in the less-developed world. The neoclassical model would have
predicted that it would have been precisely such a low-income region that would have had
the greatest opportunities for profits from increased capital accumulation. Its growth rate
would have been expected to be the highest, not the lowest, as it actually was.
The way we have interpreted the data in Table 8.1, however, needs to be approached with
caution. It represents what is called the crude unconditional income convergence prediction.
It is an interpretation based on the belief that in a relatively open international economy,
where both physical and financial capital are relatively mobile, if capital is scarce in some
countries, like those in Sub-Saharan Africa, then the lower quantity of capital will correspond
to a higher expected rate of return to investors.5 It is thought that international capital should
flow toward such capital-scarce nations, thus increasing the rate of investment relative to
total output in those economies. This would then increase the potential level of GDP in those
countries and contribute to the convergence of incomes among nations as the stock of capital
in the less-developed nations increased relative to the higher-income economies. Higher
growth rates would put these economies on the path to a higher level of income, too.
242 The Process of Economic Development
As discussed in Chapter 4 and in note 1 in this chapter, the Solow-type neoclassical theory
actually anticipates income convergence only when countries have similar savings rates and
similar population growth rates ‚Ä“ that is, income convergence is conditional on countries
sharing the same fundamentals (you‚Ä™ll remember that it is already assumed they all share
the same technology, which is exogenous to each economy). Income convergence is not
dependent upon international capital flows in this view, the magnitude of which could never
be sufficient to achieve convergence. Convergence of incomes is dependent on the similarity
of savings and investment rates and population growth. Only then would incomes converge
to the same levels. So, it is not a comparison of growth rates between regions or countries
that reveals whether convergence of income is likely. Rather it is the share of saving and
investment out of income that determines the future equilibrium income and hence is the
determinant as to whether incomes will converge in the neoclassical perspective.
Sub-Saharan Africa has had low rates of investment and savings as a share of GDP
compared to other regions (though close to those in Latin America), as can be seen from
Table 8.2. Thus, even if population growth rates were similar between this region and the
higher-income economies (which they are not; they are higher: see Chapter 12), we should not
expect to see a convergence of income per capita between Sub-Saharan Africa and the high-
income economies as a result of the large differential in the rates of investment and saving.
In other words, the necessary conditions for Solow-type income convergence to occur are
absent. The low growth rates of GDP in Sub-Saharan Africa relative to other regions, and the
low absolute level of GDP per person, are not a surprise within the neoclassical framework.
It is what would be expected from the fundamental Solow income equation from Chapter 4
(equation 4.5) because of differences in the accumulation of physical capital and the low-
level equilibrium of income toward which those economies are transitioning. Low growth
rates in the Solow model indicate that an economy is approaching its equilibrium income
level, and in Sub-Saharan Africa that is a low-level equilibrium income.
On the other hand, East Asia and the Pacific have experienced a dramatic increase in
investment and saving rates, such that investment has equaled 30 percent or more of income
in many years. One would expect convergence of income levels with more developed regions
on this basis, as savings and investment rates actually exceed those in the high-income econ-
omies. Growth rates of per capita income in East Asia did increase more rapidly, as shown
Table 8.2 Saving and investment by region (as a percentage of GDP)
1970 1980 1990 2000 2005
S I S I S I S I S I
East Asia and Pacific 33.0 33.0 35.0 33.5 35.8 31.6 42.9 37.7
Latin America 21.3 21.8 23.0 24.5 21.6 19.4 19.0 19.9 22.3 19.5
Middle East and N. Africa 26.4 29.0 18.7 26.0 25.9 23.8 26.1 27.4
South Asia 13.5 15.4 13.3 18.7 22.3 23.9 26.7 30.7
Sub-Saharan Africa 19.9 22.4 25.9 24.8 18.8 17.7 19.5 17.8 18.1 19.2
21.6 22.2 24.9 26.6 25.4 25.2 25.2 24.0 26.6
Low- and middle-income 28.7
13.6 15.5 13.4 18.4 17.7 21.1 21.2 22.6 25.1
Low-income only 28.8
25.6 25.2 23.6 24.6 22.6 22.9 21.5 21.9 19.9 20.5
Source: World Bank, World Development Indicators Online.
Endogenous growth theories and new strategies for development 243
in Table 8.1, so the gap between East Asian incomes and incomes in more-developed nations
In fact, as the Solow model would predict (again, refer back to equation 4.5 in Chapter 4),
if East Asian saving and investment rates were to remain as high as they are for a sufficiently
long period of time, per capita income in the region would eventually be even higher than
those of all other regions, when the steady-state equilibrium level of income was reached,
since the region‚Ä™s level of investment and saving is the highest in the world (assuming similar
population growth rates)! And growth rates of income would have to be high for some period
of time so that the higher level of income could be attained, though as the equilibrium level
of income was approached, growth rates of income would be expected to slow.
However, there remains something slightly disconcerting about the statistics contained in
Tables 8.1 and 8.2. The differences in investment and savings rates, and in the aggregate size
of the physical capital stock they imply, are not really that great and are unable to explain
the persistence of the large absolute income differentials between regions. Why is the absolute
level of income so low in Sub-Saharan Africa relative to the high-income areas given that
the differences in savings and investment rates shown in Table 8.2 are not that dramatic and
that new investment there should be more effective in raising incomes compared to higher-
income economies, according to the neoclassical growth perspective?6
What other forces might be at work besides the rates of saving and investment (and popu-
lation growth rates) that might be affecting not only the growth rates of GDP per person but
also the absolute level of income in different economies that lead to the wide gap in observed
levels of development?
Income convergence/divergence, path dependence, and poverty traps
economist Lloyd Reynolds found what seemed to be path-dependent patterns of growth
rates in his study of ‚Äúturning points‚ÄĚ in economic development. Some countries with certain
shared characteristics performed better over time in terms of their income per capita growth
rates than other economies with different characteristics (Reynolds 1986: 79‚Ä“81). If there
is path dependence resulting from specific growth-inhibiting characteristics that are cumu-
lative in their effects, then catch-up by the poorest countries, which have had the lowest
average growth rates and much lower incomes, obviously would be impossible. For example,
the negative impact of Myrdal-type ‚Äúbackwash effects,‚ÄĚ or ‚Äúvicious circles‚ÄĚ discussed in
Chapter 6, might account for the consistent lack of progress in Sub-Saharan Africa and slow
progress in other regions, despite investment and savings rates that would have led one to
believe that incomes should be higher and rising more rapidly than they have. There seems
to be ‚Äúsomething‚ÄĚ about the Sub-Saharan economies that resists progress. In Chapter 6, we
considered what some of those factors restraining progress might be.
What seems to happen is that countries that already have attained a higher level of income
are more likely to grow faster, for reasons which are considered below. Countries with
low absolute incomes are more likely to find themselves on a lower growth path, so that
the income gap between poor and rich nations widens rather than shrinks. Consider the
Of 114 economies with data on GDP per person in both 1970 and 2000, those with incomes
below $1,000 per capita in 2000 (in constant 2000 US dollars) were clustered around an
annual 1970‚Ä“2000 growth rate of 0 percent. For economies with incomes exceeding $1,000
per person in 2000, 1970‚Ä“2000 growth rates were clustered around 2 percent (calculations
by the authors from World Bank, World Development Indicators Online data). Economies
244 The Process of Economic Development
with low incomes had low growth rates, while those with higher incomes had higher growth
rates, on average.
Such differences can also generate a widening gap between countries within the less-
developed world as well, with the poorest remaining at the bottom of the heap, while some
formerly poor nations ‚Äúsomehow‚ÄĚ succeed in raising their living standards. All this is contrary
to the predictions of the neoclassical models and the hopes of development economists. How
well countries are performing today seems to be the best predictor of how they will perform
in the future, which does not seem to be a very hopeful prognosis for the poorest of the poor.
Nor does this tell us why this occurs. Only that it seems to be the pattern.
There is thus some evidence that convergence of incomes might be a reasonable hypothesis
for some of the already better-off less-developed countries, but not for the poorer economies.
Many of these economies appear to be in a low-level equilibrium income trap such that their
incomes are diverging from world averages (Jones 1998:56‚Ä“88). The Solow model thus may
apply quite well to developed countries or for some upper-tier less-developed economies, but
for understanding economic development for the bulk of the world‚Ä™s economies, especially
the poorest, it would appear that more is needed.
It is important to remember that the past does not have an absolute stranglehold on the
future, though. Some economies have found the means to escape the ravages of past adverse
path dependence and what must have looked like a low-level equilibrium income trap at
some point in their past. Yet they have succeeded in jumping to a higher standard of living and
more rapid economic growth. The faster growth rates and rapidly rising incomes observed
in some countries or regions that had been relatively low-income in the not-too-distant past,
countries like South Korea and Taiwan in East Asia, for example, and now perhaps China,
too, and their ability to sustain these high rates of growth over time are reminiscent of the
‚Äúvirtuous circles‚ÄĚ and ‚Äúcumulative causation‚ÄĚ effects that many development and heterodox
economists observed at work in their earlier historical studies (discussed in Chapters 5 and 6).
It is important to remember that progress has been made in countries that previously looked
as hopeless as, say, economies in Sub-Saharan Africa today. Low income does not have to be
a permanent feature of economic life if the key to moving from a low income path to a higher
income path can be found. That key apparently is not just the savings and investment rate of
the Solow-type models, however. There has to be more to the explanation.
China‚Ä™s and even India‚Ä™s recent growth paths may hold promise for other less-developed
economies mired in poverty (see Focus 8.1). It is possible for now-poor economies to escape
from adverse path dependence and what seems to be a low-income trap. Sustained increases
in income per person seem to have been maintained over a number of decades in a growing
number of formerly poor economies. The challenge for development economics is finding as
many of the critical variables as possible that explain this, variables that can be manipulated
by economic policy so as to foment positive cumulative change over the long term for more
of the world‚Ä™s population.
The pressing question, though, is whether high or rising growth rates of income ‚Ä“ such
as those China has attained or those of the East Asian economies shown in Table 8.1 ‚Ä“ can
be maintained over the long run. Why is this in question? It is an issue, because Solow-type
neoclassical growth models imply that high rates of economic growth cannot be persistent.
Because of the law of eventually diminishing returns, it is argued, as an economy gets closer
to its equilibrium level of income determined by the savings rate and population growth, the
percentage growth rate will decrease until it eventually reaches zero when the equilibrium
income is attained. After that, income per person will only grow if there is a change in exog-
enous technology or if the savings rate changes or the population growth rate changes.
Endogenous growth theories and new strategies for development 245
FOCUS 8.1 CHINA AND INDIA ON THE RISE: INCOME
In Chapter 2, China‚Ä™s recent growth experience was considered. Here we examine both
China‚Ä™s and India‚Ä™s growth history specifically as they apply to the income convergence
2005 GNI Average annual growth rate of GDP per capita in constant 2000 dollars
per capita (percentage)
US$ 1970‚Ä“5 1975‚Ä“80 1980‚Ä“5 1985‚Ä“90 1990‚Ä“5 1995‚Ä“2000 2000‚Ä“5 1970‚Ä“2005
China 1,740 3.7 5.0 9.3 6.2 10.9 7.6 8.8 7.3
India 720 0.6 0.8 3.1 4.0 3.3 4.0 5.2 3.0
United 37,600 1.9 1.8 1.9 3.0 1.4 2.7 2.1 2.1
United 43,740 1.8 2.6 2.3 2.3 1.2 2.9 1.7 2.1
China‚Ä™s overall growth rate of GDP per person since 1970 has been more than three times
that of both the UK and the US, so income per capita has been converging toward US and
UK levels. China‚Ä™s 1970 per capita GDP was $122, while that of the US was $18,150 (the
values are all in constant US dollars, i.e. adjusted for inflation). US 1970 GDP per person was
nearly 149 times as large as in China; by 2005, US average GDP was ‚Äúonly‚ÄĚ 26 times larger
than China‚Ä™s, as GDP per person increased to $1,445 in China compared to $37,574 in the
US. The relative income gap, as measured by GDP per capita, was closing. Relative income
convergence was taking place. (Think about this, though: What was happening to the abso-
lute GDP per person gap between the US and China between 1970 and 2005? Calculate the
difference for the two years.)
The average GNI for each country is also shown for 2005, as GNI is the amount of income
that is actually available to be spent in an economy, as opposed to the income created (i.e.
GDP), as you will remember from Chapter 2. India, too, was closing the relative GDP gap
with both the US and the UK, though the pace of convergence in India has been significantly
slower. In 2005, US output per person was still more than 64 times larger than average GDP
in India, down from almost 87 times greater in 1970, as a result of the slightly larger growth
rate of GDP in India compared to that in the US over the time period.
There thus has been some measurable income convergence between the two most
populated economies in the world (China and India) and the third most populous (the
United States). That is undoubtedly good news, as this economic growth led also to a
higher average level of development in both India and China. Chapter 2 looked at the
positive direction of changes in key indicators, including the Human Development Index
(HDI), for both of these economies, and these showed essential progress being made in
improving living standards as incomes grew.
There remains, however, a large gap in absolute incomes between China and India and
the US and the UK that has become wider, not narrower, over time. It is the absolute level
of income that is available for consumption and investment, not relative income. If the
1970‚Ä“2005 growth rates in GDP per person shown above were to be maintained over the
future, it would not be until the year 2052 for China and 2146 for India that the US 2005
level of GDP per capita would be attained.
Of course, if US GDP continued to increase, as it would be likely to do, there still would
not be full income convergence by those dates as US income would have increased. If
China, India and the US continued with the same rates of economic growth into the future
as prevailed over the period 1970‚Ä“2005, actual convergence of GDP per capita with the
US would occur in 2071 for China and in the year 2479 for India!
246 The Process of Economic Development
It is important to remember that in the Solow model, incomes will be expected to
converge to the same level only if the so-called "fundamentals" are the same. Specifi-
cally, the rate of savings and the population growth rates must be the same for incomes to
converge to the same equilibrium level (see equation 4.5 in Chapter 4). The following table
shows s/n ‚Ä“ the savings rate divided by the population growth rate ‚Ä“ for China, India, the
UK, and the US for various years.
1970 1980 1990 2000
China 10.5 27.9 26.0 49.5
India 6.5 6.9 11.2 13.8
United Kingdom 68.0 124.1 50.3 9.0
United States 15.7 20.6 14.4 14.7
Using the 2000 values of s/n, we would actually predict from the Solow model that
China‚Ä™s per capita income sometime in the future would be the highest, followed by the
US, India, and the UK, everything else the same. Of course, the s/n value shows substantial
variation (using 1970, 1980, or 1990 values, the UK would be predicted to eventually have
the highest equilibrium value rather than the lowest if we look only at the 2000 values), so
this shows that not only growth rates might vary as the steady-state level of income per
person is approached, but also the level of that equilibrium income is subject to variation
within the Solow model as the values of s and n are not constant over time.
Source: World Bank, World Development Indicators Online
If, however growth rates are sustained over time, an economy would be moving toward
ever higher levels of income. Sustained growth rates imply that there is no equilibrium
target level of income towards which an economy is transitioning. Growth rates would be
influencing the future level of income, an outcome that is not possible within a Solow-type
neoclassical growth framework (Lucas 1988: 12‚Ä“13). Since the real-world evidence does
show growth rates remaining high for many economies for relatively long periods of time,
how can this be explained in an economic model?
Endogenous growth models7
It was not only the question as to whether and how economies could maintain rapid growth
rates over the longer-term that vexed many development economists. There was also the
relatively slow progress of most Sub-Saharan African economies, as well as lagging growth
rates in other regions like Latin America, that led a number of economists to question the
validity of any growth model which predicts eventual income convergence based on simple
fundamentals like the rate of saving or investment. This resulted in a critical re-examination
of the policy recommendation to accumulate ever more physical capital, that is, to save or
borrow more, that flowed from the neoclassical economic formulations. As one economist
The idea that capital investment is essential to the long-run state of growth of a nation is a
common, if somewhat vague, axiom of most policy discussions of economic growth and
development. Yet for the better part of a generation the preeminent theory of economic
growth developed by the Nobel Prize winning economist Robert Solow and the data
summarized by the important contributions of Edward Denison, John Kendrick, Solow,
and others have provided us with virtually no basis for making such claims. Perhaps
Endogenous growth theories and new strategies for development 247
even more striking was the fact that theory seemed unable to explain the extreme and
persistent differences in living standards or growth rates across countries.
(Plosser 1993: 57)
The empirical research on growth using the neoclassical framework typically found that a
significant portion of the growth rate of a country, often well over 50 percent, could not be
accounted for by changes in the use of physical capital and labor, leaving what came to be
called the ‚ÄúSolow residual‚ÄĚ as the major contributor to economic growth. In other words, the
K, capital, and L, labor, inputs in the Solow growth equation back in Chapter 4 were not good
predictors of either levels of income per person or growth rates of income when sophisticated
statistical analyses were undertaken.
Accumulating more physical capital through more savings did not explain much of what
caused economies to develop in the real world. Increases in income per person were assumed
to thus be the result of exogenous technology, but the Solow model had no theory of how
or why technology changed, and almost everyone agreed that it was technology in some
broad sense that had been responsible for a very significant proportion of the progress in
most economies. Instead, something outside the Solow model was found responsible for
the bulk of economic growth; savings rates and population growth and the accumulation
of physical capital had an effect, but they were not the major players when the statistical
analyses were done. Something was missing from the story that the neoclassical model was
trying to tell.
All the diverse factors and influences that might reasonably be attributed to generating the
Solow residual ‚Ä“ such as the effects of education, technology, business organization, research
and development efforts, culture, growing international trade, local politics, and so on ‚Ä“
invited much speculation, but empirical models that might have helped to untangle, classify,
and identify these possible influences on economic growth rates were slow to appear. What
economists were left with for quite some time was a theory that seemed to say that ‚Äúsome-
thing in the air‚ÄĚ was responsible for most economic growth and higher incomes, but we are
just not sure of what that might be.
Then, in the late 1980s, so-called endogenous growth models began to emerge in the
economics literature. Endogenous growth theories do not assume, nor do they find, physical
capital accumulation to be the dominant factor in spurring economic growth or in explaining
differences in income levels among nations. Perhaps most controversially, these models jetti-
soned the assumption of diminishing returns for at least some of the inputs to production. This
effectively turns a nation‚Ä™s short-run production function into a dynamic, constantly-evolving
relationship, rather than a ‚Äúfixed-in-time‚ÄĚ function where given quantities of physical inputs
impose limits on how much can be produced and on income levels. And, lastly, the rate of
growth of per capita income is not constrained by exogenous technological change but is
internally, that is, endogenously, determined by forces specific to each economy.8
In endogenous growth models, a higher level of investment, properly defined, not only
increases per capita income, but it can sustain high and even rising rates of income growth
over the future. This is something simply not possible within the traditional neoclassical
growth model, which finds that a steady-state income level, determined by the rate of saving
and the population growth rate, is the equilibrium outcome of the growth process. On the path
to this income level, growth rates would be expected to slow down the closer current income
per person is to its steady-state value. Once that equilibrium level of income is reached,
there would be no further economic growth in a Solow-type model, unless there were to be a
change in savings and investment rates or a new burst of exogenous technology.
248 The Process of Economic Development
In endogenous growth models, it is possible for countries to continue to grow quickly for
long periods, even when they have already achieved relatively high incomes. There is not
some target equilibrium income level that is fixed by the current level of inputs to production.
There is, in fact, no steady-state income level per se. There is no pre-determined maximum
level of income that can be reached based on the rate of savings and investment and the limits
of diminishing returns. Income per person is more of a moving target. As Young (1928: 535)
long ago put it, ‚Äúthe more appropriate conception is that of a moving equilibrium.‚ÄĚ
In an endogenous growth framework, rapid growth rates can be sustained without an increase
in the rate of saving or investment, an astonishing and impossible result in Solow-type
models. Growth is not just a stage of development, as it is in Solow models, a stage that ends
once equilibrium income is attained. Instead, growth can become a permanent feature of an
As Warsh (2006: 207) writes in describing Paul Romer‚Ä™s struggle to develop one of the
earliest endogenous growth models:
the strictly empirical problem Romer was addressing was the reality that growth seems
to have been speeding up for more than a century instead of slowing down, as had been
expected [from a Solow-type model]. He reasoned that it must have to do with the
internal dynamic of science: the more you learn, the faster you learn things. If knowl-
edge was the source of increasing returns, then accumulating more of it should mean
faster growth ‚Ä“ which was, in fact, the record of the preceding two hundred years.
This insight about knowledge and how it is both cumulative and dynamic is a key to grasping
the essence of what endogenous growth theories have to say about how economies make
endogenous growth models also can quite easily explain a widening gap in income
between poorer and richer nations, because they break the link between the rate of economic
growth and the law of diminishing returns. This effectively removes the ceiling on income
per person for any particular rate of savings and investment that is imbedded in neoclassical
theory, as represented by the production function shown in Figure 4.3 in Chapter 4. Countries
that already are developed can continue to grow and grow rapidly over the future.
Endogenous growth models are reminiscent of ‚Äúcumulative causation‚ÄĚ and ‚Äúvirtuous
cycles‚ÄĚ discussed in Chapter 6, and they are consistent with a shift to a new, more beneficial
path dependence which contributes to rising income levels. In fact, some of what seemed
new in endogenous growth models was already there in Smith‚Ä™s view of economic growth,
especially his concepts of specialization and the division of labor on an ever-widening scale.
endogenous growth models actually incorporate theories and concepts from a rich earlier
economic literature on decreasing costs and increasing returns that, while not quite ignored,
was not fully appreciated for its importance in contributing to economic growth.9
In most endogenous growth models, one of the important factors of production contrib-
uting to growth has been found to be both the rate of accumulation of, as well as the initial
stock of, human capital.10 Another key input is ‚Äúresearch‚ÄĚ capital, that is, research and
development and the creation of knowledge, as the quotation from Warsh above suggests.
While these models share some superficial similarities with the capital- and saving-centered
neoclassical growth models in their form, they do not predict convergence of income levels,
even among countries which share similar rates of saving, investment, and population growth
rates. They also treat research and development and the creation of new knowledge as a
purposeful economic activity, pursued in the real world by profit-driven firms and individuals
Endogenous growth theories and new strategies for development 249
operating within a specific institutional context. The development of new technology and
new products is an internally driven process that is endogenous to every economy, and it is
this purposeful pursuit of profit within a particular institutional context that helps to explain
how economic growth occurs over the long run and why there are differences in income
levels and growth rates among real-world economies.
This insight fits better with how knowledge and technology have progressed historically.
The Industrial Revolution marks a turning point in economic and human history, as we saw
in Chapter 3. Why did economic growth accelerate for some countries, for those who became
capitalist, with the Industrial Revolution? It was not due to exogenous technological change
but rather to the decisions and actions of thousands, perhaps millions, of decision-makers,
who, in the pursuit of income and profits via the market system, experimented and searched
for the means to cut costs and gain an advantage over other producers. This is another sense in
which economic growth is an endogenous process; in capitalist societies, there is an incentive
to invent and for some to expend energy and resources to attempt to create more efficient ways
to produce as a means to increase their individual income. Adam Smith saw this process as it
emerged, and in his own way described it as the forces of accumulation and the ever-present
drive for a further division of labor, especially through the use of machines.11
endogenous growth models also place a quite different emphasis on what is required to
boost a country‚Ä™s economic growth and development possibilities compared to the recommen-
dations derived from the capital- and saving-centered neoclassical-type models by focusing
more on institutions, knowledge-creation, and education. These theories have profoundly
affected the way many economists think about policy and how they identify the most severe
barriers to development.12
A key document signaling a shift in emphasis in thinking among economists was the World
Bank study, The East Asian Miracle (World Bank 1993a).13 This critical examination of the
‚Äúhigh-performance Asian economies‚ÄĚ ‚Ä“ the HPAEs ‚Ä“ of Japan, Hong Kong, South Korea,
Singapore, Taiwan, Indonesia, Malaysia, and Thailand was built around insights flowing
from the endogenous growth theory‚Ä™s methodology for identifying the crucial policy vari-
ables in the growth process. Such endogenous growth models help to explain the power of
path dependence on growth rates and why growth can be cumulative, as well as suggesting
what is needed if countries are to jump from lower paths of growth and development to
higher paths. In other words, endogenous growth models can explain why some countries
grow quickly, reaching higher equilibrium income levels, while others get stuck in what look
like low-level equilibrium poverty traps.
A simple endogenous growth model with externalities
The major conceptual difference between the Solow-type neoclassical growth models and
the endogenous growth models is the presumption in the endogenous models that there are
not necessarily diminishing returns to all the reproducible factors of production, K, the stock
of physical capital, to H, the stock of human capital, nor to technology (or more broadly,
the intangible concept of knowledge), R. Rather it is assumed that constant, or perhaps
even increasing, marginal returns to at least some of these inputs are possible.14 How is this
endogenous growth models typically assume that there are positive externalities to human
capital accumulation, research capital, and, perhaps, to some physical capital accumulation
to the extent that new capital embodies the latest knowledge, as it is quite likely to do.
Diminishing returns to H, R and maybe even K are avoided through society-wide spillover
250 The Process of Economic Development
effects.15 When the social benefits from, for example, human capital accumulation exceed
the private benefits, there are positive secondary and tertiary effects from any increase in a
country‚Ä™s average education level or enrolment ratios that reverberate through the economy.
More educated and presumably more productive workers not only produce more at their
own tasks, but they also interact synergistically with their workmates so that the productivity
of other workers also rises, even though their level of education, that is, the quantity of
enhanced labor, remains unchanged.16
Higher average levels of education in an economy also contribute to ‚Äúlearning-by-doing‚ÄĚ
effects, that is, to the capacity for workers to build upon their previous learning and training
so that the same level of human capital input is actually able to improve its productivity
over time in the process of producing goods and services on the shop floor, or wherever
production takes place. Learning-by-doing contributes to increases in the potential level of
total output of a given level of labor input without the need for an increase in any additional
inputs and without any additional increase in investment. The presumption is that the higher
the average level of human capital accumulation in an economy, the stronger will be such
effects, again breaking the link between increases in employment and human capital accu-
mulation and diminishing returns.17
In the endogenous growth theory view, the ability to use technology, to develop new
knowledge and new products, and the skills of the labor force that complement knowledge
creation and its application are formed and shaped by each particular economy. In other
words, growth is an endogenous process, coming from within each particular economy, with
each having a different production function reflecting different quantities and qualities of its
inputs and their ability to adapt, develop and use knowledge about how to produce within
A very general endogenous growth production function for a representative economy
would look like (Romer 1994: 16):
Y = F(R, K, H) (8.1)
where Y is total output, R is research and development (R&D) done by all firms in the
economy, K is the accumulated physical capital stock and H is the accumulated stock of
human capital. For analytical purposes, this formulation has at times been operationalized as
a particular linear aggregate production function, often called an ‚ÄúAK‚ÄĚ production function,
for obvious reasons, shown in equation 8.2.
Yt = aKt (8.2)
where K is redefined as a measure of the combined stock of human, physical, and research
capital and a is a constant multiplier.18 In this very simple formulation, there are constant
returns to scale to K in production (since XYt = aXKt, where X is any finite number), as well
as constant marginal returns in the short-term, since dY/dK = a > 0 and d2Y/dK2 = 0, so that
the marginal product curve, MPK, is a horizontal line with a constant value a. Since increased
increments of K are not less effective than prior additions to K, both a rising per capita in-
come and a non-decreasing rate of growth of per capita income are possible.
In a slightly more complex formulation that captures a bit more of the ‚Äúendogeneity‚ÄĚ of
the growth process, we can write the aggregate production function as
Yt = A(K)tKt (8.3)
Endogenous growth theories and new strategies for development 251
where A(K)t is the ‚Äúinduced or endogenous technological change‚ÄĚ imparted to the economy
by the stock of physical, human and research capital particular to that country. In the produc-
tion function shown in definition 8.3, different economies will have distinct A(K) values,
depending on the feedback mechanisms affecting knowledge adaptation and technological
change specific to that economy. These are reflections of differences in human capital accu-
mulation, the spread of specialization and the introduction of new products, micro- and
macro-policies of business and government organization, social and physical infrastructure
capacity, and so on. In the original Solow-neoclassical formulation, by comparison, tech-
nology, A, was ‚Äúexogenous‚ÄĚ to all economies, affecting them identically like ‚Äúmanna from
heaven.‚ÄĚ Technology was assumed to grow at the same rate for all countries, regardless of
their own resources, policies, or actions. Such technology was not subject to policies or deci-
sions internal to any economy and did not grow at different rates for each economy.
In the endogenous growth models, technological progress as represented by A(K) is
dependent on the accumulation and spread of knowledge within each economy. It is
dependent on the rate of capital formation, broadly defined. This includes physical, human,
and research capital, and it also includes the organizational and institutional structures of
that economy. Such structures affect an economy‚Ä™s capacity to effectively utilize the world
supply of knowledge in production, to adapt it, and eventually to add to that knowledge.
Thus, in the endogenous growth formulation, the level of technology and the rate of its
application are not determined externally to the operation of that economy. Given this formu-
lation, countries that accumulate more H and undertake more R&D, R, will be more likely to
be able to continue to grow, and even to accelerate their economic growth rates, over time,
compared to nations which accumulate these inputs at slower rates.19 Indeed, the pace of any
individual country‚Ä™s technological progress is conditional on:
1 the level and type of education of the labor force and on the level and types of invest-
ment the society makes in R&D;
2 certain government policies, for example, tax credits for R&D, worker training and
education, patent and copyright laws, and so on;
3 the economy‚Ä™s and society‚Ä™s organizational and institutional capabilities formed over
time in both the private and public sectors.
Technology is not the A of the Solow-model, available equally and identically to all coun-
tries as if it were a costless public good. It is, at least in part, a private good that is costly to
produce for each country.20 The profits of such technology can be appropriated, to a degree,
by the creators of such new knowledge, which is typically assumed to be a private firm oper-
ating in an environment of imperfect competition. Newly created knowledge will affect the
production process and will spill over into the rest of the economy over time, resulting in
even more new processes, to new products, to broader and deeper specialization both internal
to firms and economy-wide, and in enhanced efficiency that contributes to the ability of an
economy to continue to grow without necessarily facing diminishing returns.21
What the endogenous growth models attempt to explain is really nothing more than how a
capitalist economy functions. Robert Lucas, an important contributor to the literature, wrote
(quoted in Warsh 2006: 237):
A few centuries ago, some of us moved into a phase of sustained growth while others
did not, and out of this ill-understood process emerged the unequal world we know
252 The Process of Economic Development
The ‚Äúa few centuries ago‚ÄĚ comment marks the beginning of the Industrial Revolution for
Western European nations and their offshoots, like the US, Canada and Australia. As capi-
talism took root and advanced in those economies, the drive to earn profits created a strong
incentive for the creation of new knowledge that could reduce costs and provide a competi-
tive advantage for one producer over another. This is nothing more than the process that
Adam Smith so presciently described in The Wealth of Nations and with so much hope for
the spread of progress (see Young 1928: 528‚Ä“31, who discusses how Smith‚Ä™s insights lead
directly to a consideration of external increasing returns and specialization of industries on
an expanding scale and how this is a characteristic of capitalist economies). It is this process
that endogenous growth models seek to explain, including the failure of progress in the still-
Figure 8.1 shows an endogenous-growth production function implied by statement 8.3.
Also shown for comparison is a typical, Solow-type neoclassical production function which
exhibits diminishing returns to K. The neoclassical production function will shift upward for
all economies when there are changes in exogenous technology, but along any given produc-
tion function there are diminishing returns to the variable input shown on the horizontal axis
(be it capital, K, or labor, L) as can be seen by the smaller slope of the production function
at higher levels of K.
However, for the endogenous growth function, A(K)tKt, there are no diminishing returns
to the variable and reproducible factors of production (be they K, or human capital, H, or
research and development, R). It would be even better to show the endogenous growth produc-
tion function as being ‚Äúflexible,‚ÄĚ i.e. as not fixed in position but capable of ‚Äústretching‚ÄĚ as
A(K ) t K t
Figure 8.1 An endogenous growth production function.
Endogenous growth theories and new strategies for development 253
spillovers from new knowledge begin to affect the nature of the productive structure of an
While further investment cannot increase total output and income per person, YN/L, above
YN, in the neoclassical world with its assumption of diminishing returns, in the endogenous
growth formulation additional investment in K, H, and R can increase output beyond YE
along the A(K) production function without reaching a pre-determined maximum. Figure 8.1
shows what an exogenous growth production function would look like if there are constant
returns to the reproducible inputs to production along A(K)tKt, though increasing returns
could also be shown. (What would the graph look like for increasing returns?)
Given the possibility of non-diminishing returns, convergence of income levels will not
occur automatically. Higher initial stocks of both capital, K, and research and development,
R, contribute to even higher rates of growth in future. Different economies will evolve in
a path-dependent fashion with distinct growth rates depending on past accumulation and
assimilation of knowledge. In economies that have accumulated a critical mass of these
inputs, spillovers will lead to more learning-by-doing, new and more specialized products,
and decreasing costs of producing. This occurs not because the world pool of ‚Äúbest practice‚ÄĚ
technological knowledge has expanded exogenously. Rather this outcome is as a consequence
of the interaction of the reproducible inputs in these economies in ways that contribute to
greater efficiency, higher incomes, and to a changing economic structure in response to the
effects of knowledge transmission by firms operating in an imperfectly competitive world. 22
For economies that have failed to accumulate a sufficient stock of human, research, and
physical capital, the spillover benefits will be smaller or nearly non-existent. This will be mani-
fested in a lower level of income per person and lower growth rates of income and the possi-
bility of a low-level equilibrium income trap. As Lucas (1988: 25) concluded after analyzing
the dynamics of his endogenous growth model: ‚Äúan economy beginning with low levels of
human physical capital will remain permanently below an initially better endowed economy.‚ÄĚ
In endogenous growth theory, the key inputs to production are not perfect substitutes for
one another. Physical capital, human capital and knowledge are, rather, complementary
inputs. This ‚Äúmeans the higher the capital stock, the more technology can increase produc-
tivity over the future. Instead of a one-time boost to productivity, higher rates of saving and
capital investment increase the rate at which productivity rises. There is no steady state of
growth ‚Ä¦ the key inputs of growth ‚Ä“ skilled labour, sophisticated capital, new forms of
technology ‚Ä“ are not independent of each other but are positively interdependent‚ÄĚ (Landau
et al. 1996: 6; also see Lau 1996).
endogenous growth models suggest that government policies can affect the rate of long-
term economic growth by impacting the accumulation of both physical and human capital
and the effort dedicated to research and development and the creation of new knowledge.
Such policies are extremely important in boosting the long-run rate of growth and the level
of income by shaping future path dependence. Markets for saving and borrowing are often
absent, or less-than-perfect, in many less-developed economies. Consequently, a purely
market-based development strategy will fail to adequately tap a society‚Ä™s potential to the
extent that investment is financed by private savings and investment decisions. Further, since
firms that invest in labor training and in research and development often create positive
externalities in production that spread to other producers, not all the profits from their invest-
ment activities will accrue to the firms making such investments.
Therefore, left to their own devices, private firms are likely to under-invest in socially
desirable labor training, in research and development, and other such non-fully appropriable
investments, since the private and social benefits of such expenditures diverge. Consequently,
254 The Process of Economic Development
government action will be necessary to subsidize or otherwise augment such investment
activities by private firms if the socially desired level of augmented capital accumulation is
to be attained. (The theory of positive externalities and the case for government action in the
face of this type of market failure are explored in detail in Chapter 12.)
At the very least, the endogenous growth models and their insistence on the existence of
pervasive positive externalities and of the spread of knowledge suggest a wider arena for
public policy action than is immediately evident from the simple Solow model.
Measuring the impact of key inputs in endogenous growth models:
The work on endogenous growth often has been highly theoretical. But there also has been
an applied econometric and empirical literature, too. Robert J. Barro‚Ä™s early research was
especially influential in this regard, but there are many other contributions, including some
useful review essays (Barro 1991, 1993; Romer 1994; World Bank 1993a: Chapter 1).
The endogenous growth research in its growth accounting form measures the contribu-
tion of various inputs to production. Table 8.3 presents the results of some of this research
drawing on the World Bank‚Ä™s study (1993a) of the HPAEs (based on Barro‚Ä™s methodology)
and a follow-up study by Rodrik (1994) incorporating additional factors affecting economic
growth. This research is suggestive of what those working early-on in this area considered
important factors to the growth process. These empirical models were based on ideas from
the endogenous growth literature, including the role of human capital and institutions in
the broadest sense. What the empirical studies attempt to do is to measure how much each
included variable ‚Äúaccounts for‚ÄĚ or contributes to economic growth.
The regression coefficients in Table 8.3 show the contribution of the various inputs to
per capita income growth. They can be interpreted as follows. For the first variable, ‚ÄúRela-
tive GDP,‚ÄĚ the World Bank‚Ä™s negative coefficient suggests, as the Solow-type neoclassical
models predict, that countries with lower incomes will grow faster, holding all other factors
constant, and thus incomes would be expected to converge. For example, a country with a
1960 GDP equal to 40 percent of 1960 US GDP would be predicted to grow 1.28 percent
Table 8.3 estimates of input contributions to per capita economic growth
World Bank coefficient Rodrik coefficient
1 Relative GDP, 1960 ‚ą’0.0320b ‚ą’0.38b
2 Primary school enrolment, 1960 2.66a
3 Secondary school enrolment, 1960 0.0069
4 Population growth, 1960‚Ä“85 0.0998
5 Average investment/GDP, 1960‚Ä“85 0.0285
6 HPAEs 0.0171b
7 Latin America ‚ą’0.0131b
8 Sub-Saharan Africa ‚ą’0.0099a
9 Gini coefficient for land, around 1960 ‚ą’5.22b
10 Gini coefficient for income, around 1960 ‚ą’3.47
Adjusted R2 0.4821 0.53
Sources: World Bank 1993a: 51, Table 1.8; Rodrik 1994: 20, Table 3.
a Statistically significant at the 0.05 level.
b Statistically significant at the 0.01 level.
Endogenous growth theories and new strategies for development 255
(‚ą’40% √— ‚ą’0.0320) faster per year than the United States over the period 1960‚Ä“85, ceteris
This ‚Äúother factors remaining the same‚ÄĚ assumption is quite important; it is an ‚Äúas if‚ÄĚ
assumption, used in interpreting each variable independently. It assumes when we look at the
first coefficient that the only difference between a poor nation and a rich nation is the level
of relative income. All other possible differences between nations on all other variables are
assumed away. Over time, then, the income per capita of poorer nations would be expected
to converge to that of richer nations, provided the only difference between countries is this
initial difference in incomes as a result of differences in the initial stock of physical capital,
as discussed previously. Thus conditional convergence is confirmed in the World Bank
model, since the first coefficient shows that being poor relative to the United States does tend
to raise growth rates of output, all else constant.
However, actual convergence of income is not presumed to occur simply because of initial
income differentials. There are other inputs to production in the World Bank model, especially
the level and type of human capital, that can and do result in quite different rates of economic
expansion. In fact, as we shall see from examining the remainder of the coefficients in the
first column of Table 8.3, income convergence is predicted to occur in the World Bank study
only if a poorer country has a higher than average stock of human capital which will allow it
to make ever better use of the world pool of technological knowledge and to take advantage
of the positive externalities which result from higher human capital investments within their
own economic setting.
The second statistically significant variable in the World Bank model in Table 8.3 is the
initial stock of accumulated human capital, as measured by primary school enrolment rates
in 1960.23 An increase of 10 percent in the primary school enrolment rate would, everything
else the same, increase the annual predicted rate of growth of the average country by 0.27
percent (10% √— 0.0272). Thus Peru, with a primary school enrolment rate in 1960 of 83
percent, would have been expected to have grown 0.95 percent faster each year than Mozam-
bique, which had a 1960 primary school enrolment rate of 48 percent, again assuming all
other variables the same (World Bank 1993a: 196‚Ä“7, Table 25).
Variables 3, 4, and 5 in Table 8.3 were not found to be statistically significant in the World
Bank model, that is, their coefficient values were not able to be confirmed as being different
from zero, though the average investment variable (variable 5) is statistically significant
when variables 6, 7, and 8 are not included in the estimates.24 Thus, the endogenous growth
model suggests the unimportance, by itself, of the level of physical capital investment.
What variable 6 tells us is that if a country was one of the HPAEs,25 that status added 1.7
percent per year to the growth of real per capita GDP. Unfortunately, exactly why HPAE
status had this impact is unexplained by the variable itself: it is actually a residual, or dummy,
variable that captures all the intangible factors that affect economic growth but which them-
selves are not directly estimated in the World Bank model. These might include: good public
administration; better organization of production at the firm level; increased product and
process specialization; appropriate macroeconomic policies; more efficient use of the world
pool of technological knowledge; more and better R&D expenditures; education appropriate
to modern economic growth; effective institutional and financial organizations; and so on. In
a broad sense, the reasons for the better performance of the HPAEs was likely due to a more
appropriate and facilitating institutional environment that was conducive to rapid economic
growth and development at both the micro and macro levels. The East Asian economies have
had weak ceremonial structures since the 1950s that do not impede, and in some cases posi-
tively promoted, the growth process.
256 The Process of Economic Development
When one compares the values for variables 7 and 8 (also dummy variables) in the World
Bank model, these estimates tell quite a different story from that told by the HPAE dummy
coefficient. Simply being a Latin American country actually reduced a country‚Ä™s predicted
growth rate by 1.31 percent per year over the entire 1960‚Ä“85 period, all else constant.26
For countries in Sub-Saharan Africa, their growth rates were predicted to be lower by 0.99
percent per year, all else constant. Thus if a country in Latin America were to be compared
with one of the HPAE countries, even if each were identical with respect to all other meas-
ured inputs to production in Table 8.3, the Latin American economy would be predicted to
have an annual growth rate per capita of 3.02 percent lower per year than the comparable
HPAE economy (= variable 6, the HPAE dummy, minus variable 7, the Latin American
dummy, or = 1.71 percent ‚ą’ (‚ą’1.31 percent)). For the average Sub-Saharan African economy,
its annual growth rate would be predicted to be 2.70 percent below that of the average HPAE
economy, all else the same.
Let‚Ä™s return to the interpretation of variable 5, average investment as a share of GDP.
The fact that this coefficient is statistically insignificant should be interpreted with care:
this does not prove that new capital formation is unimportant. Rather, the investment co-
efficient suggests that capital formation per se is a necessary, but not sufficient, condition for
economic growth. In the case of the high-growth Asian economies, the necessary and suffi-
cient conditions have been met. There, a given mass of physical capital equipment is more
productively utilized than in Latin America or Africa. This is likely due to organizational and
institutional factors not isolated by the World Bank study but captured in the dummy vari-
ables 6‚Ä“8 rather than in the contribution of investment.
Note, too, from the value of the adjusted R2 that the variables shown in Table 8.3 can explain
only about 48 percent of the growth rate of per capita income of the average economy. That
still leaves quite a lot out of the picture. In fact, more than half of what determines growth
remains unexplained, much as the Solow-residual left half or more of the growth rate unex-
plained by the inputs to production.
This unexplained difference in growth rates between regions or nations unaccounted for by
differences in relative income levels, human capital, or investment is, of course, of extreme
importance. This variation in performance tells us that some countries perform better with
the same endowments of labor, physical capital, and human capital than do other economies.
They are more efficient in producing output than other economies that are seemingly identical
in the so-called fundamental economic variables, that is, initial physical and human capital
stocks. Thus simply accumulating capital, be it physical or human, is not the whole story of
what promotes a higher level of economic development in the World Bank‚Ä™s version of the
endogenous growth model. It is but part of what is necessary to stimulate economic growth
and to qualitatively change the nature of path dependence. There is still a lot missing from
Accumulating human capital may be necessary for achieving higher rates of economic
growth, but it surely is not sufficient, or more of the variation in income would be explained
by the World Bank‚Ä™s analysis. There are other factors, such as the entire macroeconomic
environment of the economy, the types of human capital accumulated and their effectiveness
in using knowledge and physical capital, and other positive externalities associated with the
production process and the creation of knowledge that interact to contribute to aggregate
growth through all kinds of spillovers and learning. And, surely, there are both micro and
macro organizational and institutional forces at work. But none of these additional factors
is identified explicitly in the World Bank estimation. They are implicit only, which is not
helpful to policy-makers.
Endogenous growth theories and new strategies for development 257
Other empirical studies have included more explanatory variables. For example, Barro
(1991) uses the number of revolutions per year as a proxy measure of political stability; it
has the expected negative effect on growth. So, too, do fertility rates, though weakly. Other
analyses add R&D expenditures as an explanatory variable, and this raises the explanatory
power of the model. There is a danger of adding too many different variables to any growth
accounting model. One wants a theory with some generality, not a grab-bag of everything
that might affect growth, since virtually anything can conceivably do so. What is important
is to identify those factors that are most significant in influencing growth and for affecting
the nature of path dependence. For endogenous growth theory in the World Bank-type
endogenous growth format, these have been found to be human capital accumulation, capital
accumulation (if only weakly supported), R&D expenditures, political stability, fertility, and
openness to international trade.
Other endogenous factors: income and wealth distribution again
Dani Rodrik, of Harvard University and the National Bureau of Economic Research
(NBER), redid the World Bank study, altering slightly the explanatory variables, but other-
wise following the structure of the original model (Rodrik 1994). By excluding the invest-
ment rate, secondary education, and the population growth rate (which were not statistically
significant in the World Bank model shown in Table 8.3), and by including Gini coefficient
measures for the degree of inequality in land ownership and in income distribution, Rodrik
was able to explain more of the growth rate in per capita income ‚Ä“ between 53 and 67
percent, as opposed to 48 percent for the World Bank estimate.27 Since data on land and
income distribution were available for only forty countries, fewer than the sample used by
the World Bank, Rodrik‚Ä™s study is somewhat less comprehensive. Nonetheless, these find-
ings do contribute to our understanding about other factors that may be affecting growth rates
beyond the traditional variables used by the World Bank and other researchers.
Taking the Asian nations as a group, there is a significantly lower degree of inequality in
land and income distribution than in other less-developed nations (see Focus 8.1). Rodrik‚Ä™s
findings suggest that less inequality in land distribution is associated with higher economic
growth. For example, a reduction in the Gini coefficient for land distribution from 0.5 to
0.4, implying less inequality, would be predicted, from coefficient 9, to increase the rate of
growth of per capita income by 0.52 percent per year, a not insignificant amount for countries
with low growth rates and low incomes.
From Chapter 2, Focus 2.4, we have already seen that a high degree of income inequality
can be a substantial burden on economic growth rates, so it is valuable to have Rodrik‚Ä™s
independent confirmation of precisely how adverse the effects of inequality are. Even
though the Gini coefficient for income inequality was significant only at the 10 percent
confidence level (variable 10), its negative value, like the statistically significant negative
value for land distribution, does imply an inverse relation between inequality and per capita
Since both of these Gini coefficient measures were for 1960 (or close to that year), lesser
inequality may be a significant initial condition for attaining a higher growth rate and a higher
level of income, just as the stock of human capital was found to be (variable 2 in Table 8.3).
Countries may need to reach a threshold instrumental value in inequality if future success in
economic and human development is to be speeded up (which turns the traditional Kuznets
curve we considered in Chapter 2 on its head). This may be due partly to the fact that if the
income distribution is highly skewed, individuals may be unable to reap the private benefits
258 The Process of Economic Development
FOCUS 8.2 INEQUALITY AND GROWTH
The East Asian HPAEs individually and as a group have had substantially less inequality
in both their income and land distribution patterns than is the case for most other less-
developed nations, as the following table, from Rodrik (1994: 18), shows. All Gini coef-
ficient values are for the year closest to 1960.
Gini coefficient for
Hong Kong ‚Ä“ 0.49
Japan 0.47 0.40
South Korea 0.39 0.34
Malaysia 0.47 0.42
Taiwan 0.46 0.31
Thailand 0.46 0.41
Average, all eight HPAEs 0.45 0.39
Argentina 0.87 0.44
Brazil 0.85 0.53
India 0.52 0.42
Kenya 0.69 0.64
Mexico 0.69 0.53
Philippines 0.53 0.45
Average, selected others 0.68 0.50
It will be remembered from Chapter 2 that Gini coefficients closer to 1 (or 100) imply greater
inequality, while values closer to 0 suggest greater equality of the respective distribution. It is
quite impressive to note the substantially greater degree of equality in both land and income
distribution in the HPAE economies compared to other less-developed nations. It is certainly
relevant to examine, as Rodrik does in his study of the sources of growth shown in Table
8.3, the significance of these differences on the rate of economic growth. Rodrik found that
higher degrees of inequality were harmful to the level and to the pace of economic growth,
as shown by the negative values on variables 9 and 10 in Table 8.3. Inequality may be neces-
sary to provide incentives, but there is a danger if inequality is too great.
of increasing their human capital, thus reducing the potential level of GDP and reducing the
growth rate. An income distribution with ‚Äútoo much‚ÄĚ inequality can act as an institutional
barrier to progress and contribute to a low-level equilibrium income by limiting opportunities
and aggregate demand.
Part of the reason for the lower growth rates of income for economies in Latin America and
in Sub-Saharan Africa was therefore likely due to the greater degree of initial inequality in those
economies. While Rodrik‚Ä™s work leaves unexplained other reasons for differences in income
levels among economies, as evidenced by the adjusted R2 value, the inclusion of the degree of
inequality adds explanatory power as a further insight of importance for nations that remain
relatively poor beyond those of the World Bank model.28 Further, as Rodrik (1994: 22) argues,
once initial levels of schooling and equality are taken into account, there appears to be
nothing miraculous about the HPAE‚Ä™s growth experience ‚Ä¦ Around 90 percent or more
of the growth of Korea, Taiwan, Malaysia, and Thailand can be accounted for by these
economies‚Ä™ exceptionally high levels of primary school enrolment and equality around 1960.
Endogenous growth theories and new strategies for development 259
In other words, Rodrik does not believe that the East Asian economies have resorted to
any ‚Äúmiracle‚ÄĚ strategies, as the title of the World Bank study might suggest, in shifting from
a less-developed path to one taking them toward a more developed status. What the East
Asian economies achieved and what contributed to their economic growth is tangible and
reproducible and was ‚Äúby design‚ÄĚ. Public policies were devised to create the key ‚Äúfundamen-
tals‚ÄĚ ‚Ä“ in this case, human capital and greater equality ‚Ä“ that provided the essential initial
conditions conducive to future progress. In the case of primary and secondary education and
increased equality, these are examples of social infrastructure or social capital. They are just
as important to economic growth, and to human development, as are physical infrastruc-
ture, such as ports, roads, water and power grids, and communications systems. Too often,
perhaps, social infrastructure has not been accorded sufficient attention by policy-makers in
less-developed economies. The evidence now seems clear, and more will be adduced later,
in support of more productive public policies for economic growth.
Other research supports this conclusion that there was no ‚Äúmiracle‚ÄĚ that explains the
rapid pace of economic progress in the East Asian economies since the 1950s. Young (1995)
and Grier (2003) both argue that rapid economic growth was due to accumulating the right
kind of inputs in sufficient quantities. More human capital accumulation paid off with its
efficiency-enhancing, spillover effects. Greater capital accumulation that embodied new
knowledge and technological innovation contributed to economic growth to the extent that
the complementary human capital inputs had been formed to work with it. There were no
miracles that led to productivity increases that cannot be explained primarily by more of the
right inputs, that is, not just more labor and physical capital, but better trained and educated
workers who could work with new physical capital embodying new knowledge in ways that
expanded output even faster.
That there was no ‚Äúmiracle‚ÄĚ of growth in East Asia is actually very good news for those
economies that remain mired in poverty. It means that what some successful, formerly less-
developed economies have done to shift to a higher growth path can be followed by others.
The lessons are clear thanks to the on-going efforts of economists to clarify the specific
inputs to production: provide funding and incentives for the accumulation of more human
capital; create financial and legal institutional structures conducive to innovation in products
and processes; and provide incentives for investments in new physical capital that embodies
new knowledge. From these broad strategies comes more rapid economic progress.
Technical efficiency change
Part of the story told by the endogenous growth theories has to do with the effectiveness with
which a country‚Ä™s endowments ‚Ä“ human capital, physical capital, other resources, knowledge,
etc. ‚Ä“ are utilized in the production process. And in today‚Ä™s world, with such rapidly expanding
knowledge and technology creation, countries must be ‚Äúdoing the right things‚ÄĚ if growth rates
are to be maintained and higher levels of income are to be reached.
One way effectiveness can be measured is by what is called technical efficiency change.
The idea behind this concept is illustrated in Figure 8.2, which shows a standard production
possibilities frontier (PPF). Technological change can be represented by an outward shift of
the frontier from FF1 to FF2.29
First, let‚Ä™s consider an economy that had originally been producing at A on FF1 and then,
after the technological advance, moves to B on the new, higher production possibilities
frontier, FF2. This economy would be keeping pace exactly with the rate of technological
advance and international ‚Äúbest practice‚ÄĚ as represented by the maximum outputs that can
260 The Process of Economic Development
Figure 8.2 Technological change versus technical efficiency change.
be produced from the inputs on the production possibilities frontier. The potential rate of
economic growth is determined by the rate of technological change and new knowledge
creation, whatever its source may be. The actual rate of technological and knowledge change
attained by an economy, however, is endogenous, depending on past and present investment,
educational, R&D, legal, and a whole range of other decisions that determine how efficient
the economy is relative to what is possible as represented by the PPF (World Bank 1993a:
An economy that moves from production point A to production level B is keeping pace
with what is possible. It is growing at the pace of technological change (represented by the
shift in the PPF) and is operating efficiently (on the PPF). For such an economy, there is zero
technical efficiency change, as the economy operates at maximum efficiency both before and
after a change in technology.
Now consider an economy beginning at a position like C, inside of FF1, which moves to
D, inside FF2 after the change in technology. By approaching closer to the new production
possibilities frontier compared to its position relative to the former PPF, we can say this
economy has experienced positive technical efficiency change. The level of international best
practice is being approached as the nation‚Ä™s productive resources are better able to capture
the advantages of increases in knowledge than in the past. For such countries, their rate of
economic growth will exceed the average rate of world technological change measured by
the shift outward of the PPF.
Such economies are better able to make use of their current resource endowments in ways
that enhance the overall rate of economic expansion. This permits growth rates that exceed
the pace of overall, world best-practice technological change. In other words, such countries
are able to combine their existing resources in more efficient combinations by being better
able to capture the benefits of increases in knowledge and technology. This may be due
Endogenous growth theories and new strategies for development 261
to improvements in the stock of human capital accumulated, learning-by-doing, continued
improvements in the macroeconomic or microeconomic management of the economy, or to
any of a number of possible causes, internal and external, as discussed earlier in the chapter.
Table 8.4 summarizes the World Bank‚Ä™s estimates of how various countries and regions
did relative to international ‚Äúbest practice‚ÄĚ technology, that is, in terms of technological
efficiency change and progress toward reaching the ever-moving optimal production possi-
bilities frontier resulting from constant knowledge creation at the international level.
Of the HPAEs, Hong Kong, Japan, Taiwan, and Thailand were making progress toward
using, and using more effectively, the best available technology.30 This happened, the World
Bank suggests (and Chapter 12 will consider this reasoning in detail) because these countries
had accumulated an appropriate stock of human capital that permitted them to learn from,
adapt, and use new technologies, new ideas, and new production processes discovered else-
where, perhaps even adding to best practice methods adapted to their own economies as a
consequence of trial-and-error.
Singapore, Malaysia, Indonesia, and even apparently South Korea, however, were slipping
behind world best practice. Graphically this would mean that their relative position compared
to the outward-shifting production possibilities frontier in Figure 8.2 was to be further inside
the outward-shifting frontier, that is, further away from the expanding pool of technological
knowledge and its application to the production process compared to production prior to
technological advances that shifted the PPF.31 Negative technical efficiency change does not
imply that an economy is not growing. It means that it is not growing as quickly as it might
have given the advances in knowledge that occurred. Such economies are not keeping pace
with the expansion of best-practice technological change, but they may still be growing.
For Latin America and Sub-Saharan Africa, there also was a growing shortfall from the
most efficient production methods. This situation was most severe in the case of Sub-Saharan
Africa, which effectively was falling further and further behind in the ability to use its
resources in ways that could positively augment production and human development. This
reflects a weakness in the creation of human capital and in R&D expenditures, and perhaps
is also a consequence of the inequalities in land and income distribution that rendered the
Sub-Saharan economies unable to use, adopt, or adapt very effectively what the world pool
of technology had to offer.
Additionally, the productivity gap between Sub-Saharan Africa and Latin America and
the HPAEs was widening because of these differences in technological capacity, making
even relative, unconditional convergence of incomes impossible without a substantial turn-
Table 8.4 Estimates of technical efficiency change, 1960‚Ä“1989
Technical efficiency change
Latin America ‚ą’1.4217
Sub-Saharan Africa ‚ą’3.4539
Hong Kong 1.9714
South Korea ‚ą’0.2044
Source: World Bank 1993a: 69.
262 The Process of Economic Development
around in the level of technical efficiency change. That will require changes in the factors
that the new growth theories see as fundamental to stimulating growth. More saving or
investment alone is not sufficient. There must be appropriate human capital accumulation
and sufficient research and development so that a critical threshold of domestic techno-
logical competency is reached. It is also likely that better macroeconomic policies need to
be introduced, that changes in industrial organization and business practices that reward
innovation be fomented, and, following Rodrik, amelioration of severe inequality also may
be a prerequisite for future progress. These are themes we will touch on in the remaining
chapters, building on the endogenous growth theories and the identification of some of the
fundamental policy variables.
endogenous growth theories have contributed to our understanding of how some countries
can maintain high economic growth rates over long periods of time. They also help to clarify
why levels of income remain low in far too many economies and why others have been able
to escape adverse past path dependence and attain higher levels of both economic growth and
incomes. In often complex mathematical formulations, economists have attempted to model
the workings of market economies with all of their real-world imperfections.
In a very general way, endogenous growth models focus on how capitalist economies
expand over time. They show how new knowledge, new products, and a finer division of labor
within and between firms can contribute to both economic growth and higher income levels.
For such models, the institutional milieu of each economy is extremely important, though
it cannot always be modeled explicitly. Where the institutional environment ‚Ä“ government,
finance, the legal system, openness of the economy to trade, incentives to entrepreneurship,
the educational system, and so on ‚Ä“ is conducive to expansion, an economy is more likely to
prosper and can do so for long periods of time. And we know if the institutional environment
is propitious by observing what is occurring at the macro-level of the economy. While this
may sound like circular reasoning, it is not, since part of what is done empirically with these
models is to examine these institutional structures.
In other economies where economic growth rates and levels of income are both low, the
confluence of obstructive institutions and the failure to accumulate the appropriate human
and physical capital resources can be seen to be the restraining forces to progress. It is in
this way that theory can be an aid to policy by identifying what is most crucial for economic
growth and human welfare. On these counts, endogenous growth theory has made important
contributions to our understanding of these mechanisms.32
Questions and exercises
1 Distinguish between ‚Äúconditional convergence‚ÄĚ and ‚Äúunconditional convergence‚ÄĚ of
per capita income. On what does ‚Äúconditional‚ÄĚ convergence of income depend in the
neoclassical model, i.e., what are the conditions that must hold for incomes to converge?
Does either the Solow-type growth model or the endogenous growth model predict
unconditional convergence of incomes amongst nations? What would unconditional
convergence of incomes imply for incomes of countries in the future?
2 Table 8.1 considers income convergence by looking at aggregate income by region.
Using the World Bank website (http://www.worldbank.org), find income growth figures
for four different less-developed economies and for the United States, the UK, and Japan
Endogenous growth theories and new strategies for development 263
for as many years as feasible. Put the data into a table format so it is easy to compare