Lecture Points - Part I

M Parameswaran

October 20, 2022

Contents

1 Concept of Development 4
1.0.1 Why do we study economic development . . . . . . . . . 7

2 A Theoretical Framework 9
2.0.1 Development of the Social System . . . . . . . . . . . . 9
2.0.2 Marx and New Institutionalism . . . . . . . . . . . . . . 13
2.0.3 Theory of Induced Innovation . . . . . . . . . . . . . . . 15
2.0.4 Historical Path-dependency . . . . . . . . . . . . . . . . 19

3 Theories of Population Growth 20

3.1 Demographic Transition . . . . . . . . . . . . . . . . . . . . . . 21
3.2 Economic theories of population growth . . . . . . . . . . . . . 22
3.2.1 Malthusian theory of population . . . . . . . . . . . . . 22
3.2.2 Household utility maximisation model . . . . . . . . . . 24

4 Ricardian model of development 26

5 Economic Growth: Empirics 29

6 Economic Growth: Theory 33

6.1 Solow Growth Model . . . . . . . . . . . . . . . . . . . . . . . . 34
6.1.1 Solow Model Without Technological Progress . . . . . . 35
6.1.2 Technology and Solow Model . . . . . . . . . . . . . . . 41
6.1.3 Convergence . . . . . . . . . . . . . . . . . . . . . . . . . 44

7 Growth Accounting 50
7.1 Standard Primal Growth Accounting . . . . . . . . . . . . . . . 50
7.2 Dual Approach to growth accounting . . . . . . . . . . . . . . . 53
7.3 Problems with growth accounting . . . . . . . . . . . . . . . . . 54
7.3.1 An increasing returns model with spillovers . . . . . . 54

1
CONTENTS 2

8 Econometrics of Income Convergence 57

8.1 Neoclassical Growth model** . . . . . . . . . . . . . . . . . . . 59
8.1.1 Convergence Tests . . . . . . . . . . . . . . . . . . . . . 61
8.1.2 Time series tests . . . . . . . . . . . . . . . . . . . . . . 64

9 Growth Accounting and Income Convergence 67

9.1 Growth Accounting** . . . . . . . . . . . . . . . . . . . . . . . . 67
9.1.1 Standard Primal Growth Accounting . . . . . . . . . . . 67
9.1.2 Dual Approach to growth accounting . . . . . . . . . . . 70
9.1.3 Problems with growth accounting . . . . . . . . . . . . . 71
9.2 Income Convergence** . . . . . . . . . . . . . . . . . . . . . . . 74

10 Solow type growth models: limitations 84

11 New Growth Models 93

11.1 Romer (1986) Model . . . . . . . . . . . . . . . . . . . . . . . . . 93
11.2 Romer (1990b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
11.3 Semiendogeneous and Schumpeterian Growth Models . . . . . 106

12 Evolution of growth theory 111

12.1 Technological Progress and Growth Models . . . . . . . . . . . 111
12.2 Trade and Technological Progress . . . . . . . . . . . . . . . . . 119
12.3 Neutrality of Technological Progress . . . . . . . . . . . . . . . 121
12.4 Representation of Technical Progress . . . . . . . . . . . . . . . 122
12.5 Classification of Technological Progress . . . . . . . . . . . . . 125
12.5.1 Hicks’ Classification . . . . . . . . . . . . . . . . . . . . 125
12.5.2 Harrod’s Classification . . . . . . . . . . . . . . . . . . . 128
12.5.3 Solow’s Classification: . . . . . . . . . . . . . . . . . . . 130
12.5.4 Some properties of Output per worker production function131

13 Histry vs Expectations 133

13.1 Externalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
13.2 Complementarities . . . . . . . . . . . . . . . . . . . . . . . . . 136
13.2.1 Coordination Failure . . . . . . . . . . . . . . . . . . . . 139
13.2.2 History versus expectations . . . . . . . . . . . . . . . . 143
13.3 Increasing returns . . . . . . . . . . . . . . . . . . . . . . . . . . 144
13.3.1 Increasing returns and market entry . . . . . . . . . . . 144
13.3.2 Increasing Returns and Market Size . . . . . . . . . . . 147
List of Figures

2.1 Interrelated Developments in the social system . . . . . . . . . 10

2.2 A model of political market for a public good . . . . . . . . . . 18

3.1 Malthusian Population Theory . . . . . . . . . . . . . . . . . . 23

3.2 HH utility maximisation and Number of children . . . . . . . 24

4.1 Ricardo model of economic development . . . . . . . . . . . . . 28

6.1 Cobb-Douglas Production Function . . . . . . . . . . . . . . . . 36

6.2 Basic Solow Diagram . . . . . . . . . . . . . . . . . . . . . . . . 38
6.3 Solow Diagram with Production function . . . . . . . . . . . . 39
6.4 Comparative statics in Solow Model . . . . . . . . . . . . . . . 40
6.5 Economic Growth in the Simple Model . . . . . . . . . . . . . . 42
6.6 Unconditional Convergence . . . . . . . . . . . . . . . . . . . . 45
6.7 Conditional Convergence . . . . . . . . . . . . . . . . . . . . . . 47

10.1 Growth of India’s GDP . . . . . . . . . . . . . . . . . . . . . . . 87

12.1 Representation of Technological Progress . . . . . . . . . . . . 123

12.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
12.3 Hicks’ Neutral Technical Change . . . . . . . . . . . . . . . . . 127
12.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
12.5 Harrod-neutral Technical Change . . . . . . . . . . . . . . . . . 129
12.6 Per worker Production Function (y “ f pkq) . . . . . . . . . . . . 132

13.1 Cost curves with complementarities in adoption . . . . . . . . 138

13.2 Linkages and Coordination . . . . . . . . . . . . . . . . . . . . 141
13.3 History Vs Expectations . . . . . . . . . . . . . . . . . . . . . . 144
13.4 Increasing Returns and Market entry . . . . . . . . . . . . . . 146

3
Chapter 1

Concept of Development

• Development of a society essentially means assuring quality life to its

members. However, quality of life (QoL) is a broader concept; it includes
a full range of factors that people value in achieving and thus clearly
goes beyond material prosperity (see: Sen, 1988).
• An approach to assess development across societies using GDP per
capita, which essentially summarises people’s command over resources,
has mainly three limitations (see: Stiglitz, Sen, and Fitoussi, 2009),
apart from the usual problems associated with its distribution and
computation.
1. Resources are the means that are transformed into well-being in
ways that differ across people. Individuals with greater capaci-
ties for enjoyment or greater abilities for achievement in valuable
domains of life may be better off even if they command a fewer
resources.
2. Many of the resources are not marketed, and even when they
are, prices will differ across individuals, making it problematic to
compare real income across individuals.
3. Many determinants of human well-being are aspects of people’s
life-circumstances, they cannot be described as resources with
imputable prices, even if people make trade-offs among them.
• Though command over resources are considered as insufficient met-
ric for quality of life, which other metric should be used instead for
assessing the quality of life depends on the philosophical perspective
taken. However, irrespective of the philosophical perspective taken,
development is usually considered to be multidimensional in nature
and it also lacks straightforward link with opulence.

4
CHAPTER 1. CONCEPT OF DEVELOPMENT 5

• Given the multidimensional nature of development, three approaches

have been devised in the literature to assess the quality of life (QoL).
They are subjective approach, capability approach, and an approach
based on welfare economics and theory of fair share.1

• Subjective approach is based on the long philosophical tradition that

individuals are the best judge of their own situation. It uses the self
reporting of people about their own state. One of the important con-
tributions of research based on subjective approach is that it elicited
a better understanding of the determinants of QoL, such as income,
health status, education, security and so on.

• In capability approach, a person’s life is evaluated in terms of the

functioning he/she is free to choose, where functioning is a broad term
used to refer to the activities and situations that people consider to
be important to achieve, such as health, education, and political voice.
In this perspective well-being of a person is a summary index of the
person’s functionings.

• While the above two approaches closely connected respectively to the dis-
ciplines of psychology and moral philosophy, the third one is connected
to economics (Stiglitz, Sen, and Fitoussi, 2009). Welfare economics and
the theory of fair allocation suggest a method to include non-market
aspects of QoL into a broader measure of well-being on the basis of
individual preferences.

• Further, the foundations of the capability approach, which has strong

roots in philosophical notions of social justice, reflect a focus on human
ends and on respecting the individual’s ability to pursue and realise
the goals that he or she values; a rejection of the economic model of
individuals acting to maximise their self-interest heedless of relation-
ships and emotions; an emphasis on the complementarities between
various capabilities; and a recognition of human diversity, which draws
attention to the role played by ethical principles in the design of the
“good” society (Stiglitz, Sen, and Fitoussi, 2009).

• The three approaches to assess quality life have obvious differences,

but also certain similarities. For example, subjective well-being is
sometimes claimed to encompass all capabilities, in so far as these refer
to attributes and freedoms that people value (implying that enhancing
their capabilities will improve people’s subjective states). However,
1
For more details see Stiglitz, Sen, and Fitoussi (2009).
CHAPTER 1. CONCEPT OF DEVELOPMENT 6

proponents of the capability approach also emphasise that subjective

states are not the only things that matter, and that expanding people’s
opportunities is important in itself, even if this does not show up in
greater subjective well-being. Similarly, both the capability and the fair
allocation approaches rely on information on the objective attributes of
each person, while differing in the ways in which these are weighted and
aggregated. While the choice between these approaches is ultimately
a normative decision, they all point to the importance of a number
of features that go beyond command over resources(Stiglitz, Sen, and
Fitoussi, 2009, p.42).

• Research on QoL based on the above mentioned approaches emphasises

a range of features in people’s life that are important either intrinsically
as objective expressions of a good life or instrumentally to achieve
valuable subjective state or objective goal.

• However, it is important to note that what are the objective features or

dimensions of good quality life inevitably depends on value judgements.
It is true that valuation of various aspects of life differ from person
to person (value-heterogeneity) and the process of change involved in
development also change the valuation of people (value-endogeneity).

• Though value-heterogeneity and value-endogeneity are serious issues,

it has been argued that the actual extent of agreement on aspects
determining life’s quality is indeed quite large (see: Sen, 1988; Stiglitz,
Sen, and Fitoussi, 2009). For instance, Stiglitz, Sen, and Fitoussi (2009)
considers following dimensions to be essential for having quality life in
modern societies.

1. Health: Health is a basic factor deciding both length and quality

of life. It has both intrinsic and instrumental worth. Better health
not only allow people to enjoy their life, but it also enable them to
work and earn an income.
2. Education: Education is essential for having a quality life in
modern society. Apart from its contribution in improving peoples’
productivity and earning, educated people typically have better
health status, more social connection and greater engagement in
civic and political life.
3. Personal Activities: The way people spend their time and the
nature of their personal activities have greater impact on the qual-
ity of their life, irrespective of the income generated. Further, the
CHAPTER 1. CONCEPT OF DEVELOPMENT 7

activities in which people engage have an effect on their subjective

well-being.
4. Political voice and governance: Ability to participate in polit-
ical discourse, and to dissent as well as to speak up against the
authority on matters that one perceives to be wrong are essen-
tial aspects of political freedom. Apart from giving a corrective to
the public policy, this, along with their ability participate in the
governance of the country, also improves the self-esteem of the
people.
5. Social Connections: Being a social animal, social connections
and interactions improve the quality of life in many ways. Apart
from providing many services like insurance and security, social
connections and interactions help people to have a broader and
realistic attitude to life, helping them to cop up with many anxieties
and stresses in life, enabling them to have better life-evaluations
and health status.
6. Environmental Conditions: Apart from providing sustainability,
better environmental conditions contribute to life quality in many
ways. Clean air, water and surroundings positively contribute to
the health of the people. Further, people also value environmental
amenities or disamenities and they take this into consideration
when they make their choices such as where to live.
7. Personal insecurity: Personal insecurity includes external fac-
tors that put at risk the physical integrity of each person: crime,
accidents, natural disasters, and climate changes are some of the
most obvious factors.
8. Economic insecurity: Uncertainty about future income and ma-
terial conditions emerging out of events such as employment loss,
wage cut, and illness, can drastically reduce the quality of life.

• Economic Development: Compared to the concept of development,

economic development is a narrower concept. Besides rise in income
and consequent increase in the material standard of living, it includes
dimensions such as absence of poverty and under-nutrition and mod-
ernisation of the economy in terms of technological progress.

1.0.1 Why do we study economic development

• Why are we studying development? There are mainly two reasons.
CHAPTER 1. CONCEPT OF DEVELOPMENT 8

• One reason that there exists extreme inequality in income across coun-
tries. The picture is highly unequal even if we use PPP exchange rate
and also correct for the availability of many non-market goods and
services in poor countries. The growing inequality in income across
countries/regions is contributing to increased tensions in the interna-
tional relations. It is one of the prominent factor behind the recent rise
in ethnic conflict and terrorism, which is undermining peace, security
and political stability among a number of countries.

• The second reason is the prevalence of wide spread poverty and destitu-
tion in a large number of countries. This is also reflected in the high
rate of infant mortality rate, low life expectancy, illiteracy and so on.
This implies that escaping from this dire poverty and destitution should
be the major goal of the poor countries.

• To move to high income status, it is essential that poor countries have

to adopt production technology ensuring high level of productivity. How-
ever, this needs supporting social and cultural institutions and values.
So one purpose of studying development economics is understand this
aspect, i.e. which institutions and cultures promote economic progress,
and what we can do with those cases where the prospects is not that
rosy.

• Another issue is the optimum mix of government, market and commu-

nity, i.e. choice of appropriate economic system. These three organ-
isations coordinate the division of labour among people - market by
competition, government by coercion, and community by cooperation.
Each of them have their own merits and demerits, so optimum combi-
nation is essential for maximising the social productivity. This is an
issue we need to ponder. The argument is that if developing countries of
today are to catch up with the developed economies, they must develop
effective economic systems suitable to their unique cultural and social
traditions.
Chapter 2

A Theoretical Framework for

Development

• Achieving and maintaining high economic growth requires major changes

in the social organisations and people’s value system.

• Understanding the process by which quantitative expansions in eco-

nomic variables (such as capital and income) interact with culture and
institutions to evolve a social system that supports major growth in
per capita income is argued to be the ultimate goal of development
economics (Hayami and Godo, 2005).

• Here we present a framework suitable for analysing the complex rela-

tionship among economic, cultural and institutional changes.

2.0.1 Development of the Social System

A model of dialectic social development
• The Figure 2.1 presents a model of the evolution of the social sys-
tem through dialectic interactions between economic and cultural-
institutional variables1 .

• The lower part of the figure presents economic sector as a subsystem

of society. This subsystem consists of interactions between technology
and ‘resources’,– broadly defined as ‘factors of production’, including
1
Dialectic interaction means change through conflict of opposing forces. The meaning
given in the dictionary is “the existence or action of opposing social forces, concepts, etc”.

9
CHAPTER 2. A THEORETICAL FRAMEWORK 10

Cultural-
Culture Institutions
Institutional
(Value system) (Rules)
Subsystem

Resources Technology Economic

(Production (Production Subsystem
Factors) Function)

Figure 2.1: Interrelated Developments in the social system

natural resources, labour and capital. Technology helps to combine var-

ious resources to produce output. Improvements in technology enables
production of more valuable products from the given resources.

• Per capita income or average income increases through the expansion

of resources, such as accumulation of human and physical capital, as
well as through improvements in technology. However, improvement in
technology and expansion of resources are not independent; expansion
of one can stimulate the expansion of the other.

• The productivity of an economic subsystem, consisting of its resource

endowments and technology, is conditioned by culture and institutions
in society.

• Culture is narrowly defined to imply the value system of the people in

the society. While for institutions North (1990, p.3) gives the following
definition: “Institutions are the rules of the game in a society or, more
formally, are the humanly devised constraints that shape human in-
teraction”. He also emphasises the key implications of institutions “In
CHAPTER 2. A THEORETICAL FRAMEWORK 11

consequence, they structure incentives in human exchange, whether po-

litical, social or economic”. See also Acemoglu, Robinson, and Johnson.
(2005).
• Similar to technology and resources, culture and institutions thus de-
fined are inseparably related. The rules that contradict the morals
or value system of the people would not be sanctioned socially and, if
stipulated formally, would not function effectively. An example is the
case of slavery, which is an institution having legitimacy once, has no
social approval now. Other examples include caste system and racial
discrimination. Some societies’ value system does not tolerate them,
while some others’ does.
• Culture and institutions indicated in the upper section of the Figure
2.1 as components of the social system exert significant influence on
the economic subsystem located in the lower section. The example
include saving rate, i.e., saving out of current income, determined by
the peoples’ future preferences over present consumption, which is a
part of their value system. Preferences of the people regarding saving
would change when a society moves from underdeveloped to developed
state, partly due to the the decline in informal social security. Similarly,
as economic subsystem develops it also exert significant influence on
value system in the society, for instance regarding women employment.
• Weber’s (1930) argument that Protestant ethic as a source of modern
capitalist development.
• While resource accumulation and technological progress are conditioned
by culture and institutions, changes in the latter are also induced by the
former. That is the process of social development is through dialectic
interactions between economic and cultural-institutional subsystems.

A Historical Example:
• While resource accumulation and technological progress are conditioned
by institutions, changes in the latter are also induced by the former.
Such a process of social development through dialectic interactions
between the economic and social-institutional subsystems may be un-
derstood more concretely by tracing the transition from the hunting
and gathering economy to the agricultural (and pastoral) economy.
• The major factor inducing epochal changes in the human history was
the increased scarcity of natural resources under the pressure of pop-
CHAPTER 2. A THEORETICAL FRAMEWORK 12

ulation growth, (see Cohen (1977)). As population grew (though very

gradually), an inevitable state reached in which exploitation of the
wild resources began to exceed their reproductive capacity, the hunter-
gathering economy could not be sustained.

• Then it became imperative for hunters to augment the reproduction pro-

cess by raising animals instead of killing and eating them immediately,
and for gatherers to plant nuts and cereals for future harvests.

• An economic basis for the increased reproduction was the accumulation

of capital. A limited set of capital items was required for hunting and
gathering, such as stones, knives, clubs, and bows and arrows. A larger
capital stock was required for shifting to the agriculture-based system,
especially in the forms of reared animals, standing crops and trees,
and opened and cultivable farmlands. Capital requirement further
increased as agricultural production system advanced to the stage at
which it began to rely on man-made infrastructure, such as irrigation
and drainage facilities and animal power.

• Similarly, to covert animals and plants to productive capital, it was nec-

essary to accumulate knowledge to identify useful plants and animals,
to plant seeds and ways to tame and use animals. This advancement
in knowledge came through centuries of trail and error by the human
beings.

• However, this advancement in knowledge was though necessary, but

not sufficient for development of agriculture-based economic system.
This development required a major institutional change: establishment
of property rights over productive resources.

• The basic rule in the hunter-gatherer society was the free access to the
natural resources, under which all the resources were the property of
every one and no one has any particular property right over anything.
Under this rule, a person engaging in agriculture has the difficulty of
preventing others from capturing the plants and animals he was trying
to grow.

• Therefore, individuals did not have incentive to invest in agriculture

and animal husbandry. So the prerequisite for the formation of an
agricultural economy was the establishment of a new social order of
clearly defined property rights by which the person who made efforts to
invest in productive capital could exclude others from its use (Demsetz,
1967; Alchian and Demsetz, 1973).
CHAPTER 2. A THEORETICAL FRAMEWORK 13

• In the course of agricultural development, property rights were first

assigned to standing crops and domestic animals and later to agricul-
tural land, enhancing the owners of agricultural land to improve and
preserve land quality.

• Strong social norms, supported by religious institutions, were also

introduced to ensure the enforcement of property rights. This gives an
example for the economic and social development through interactions
between economic forces and cultural and social institutions.

• With the emergence of industrial society, knowledge, particularly tech-

nological knowledge become an important factor of production, resulted
in the emergence of property right over knowledge, i.e., patent system.
These developments are same as the emergence of property right over
animals and agricultural land.

2.0.2 Marx and New Institutionalism

• The theoretical framework outlined above has strong similarity with
the perspective on evolution of the social system described by Karl Marx
and Friedrich Engels.

• It is a theory of socio-economic development according to which changes

in material conditions (technology and productive capacity OR forces of
production) are the primary influence on how society and the economy
are organised.

• Social classes and the relationship between them, along with the po-
litical structures and ways of thinking in society, are founded on and
reflect contemporary economic activity.

• Historical materialism started from a fundamental underlying reality

of human existence: that in order for human beings to survive and
continue existence from generation to generation, it is necessary for
them to produce and re-produce the material requirements of life

• The concept of ‘mod of production’ which consists of forces of produc-

tion and relations of production. The main modes of production Marx
identified generally include primitive communism or tribal society (a
prehistoric stage), ancient society, feudalism, and capitalism.

• In each of these social stages, people interact with nature and produce
their living in different ways. Any surplus from that production is
CHAPTER 2. A THEORETICAL FRAMEWORK 14

allotted in different ways. Ancient society was based on a ruling class of

slave owners and a class of slaves; feudalism was based on landowners
and serfs; and capitalism based on the capitalist class and the working
class. The capitalist class privately owns the means of production
(e.g.,factories, mines, shops and banks) while the working class live by
exchanging their labour power with the capitalist class for wages.

• The economic subsystem and cultural-institutional subsystem in 2.1

correspond to broadly with what they term ‘infrastructure’ (or forces of
production) and ‘superstructure’ (relations of production).

• In this system the core of the superstructure is the property right

relations of the production factors (the so called relations of production)
while infrastructure is the technology deciding the capacity of material
production from the available resources (forces of production).

• In this scheme the evolution of the forces of production (technology)

decides the evolution of the relations of production.

• This view on the formation of institutions evolved in response to the

economic demands is analogous to the theory of induced institutional
innovation.

• In Marxian system a major time lag between increase in material

production capacity and changes in institutions was assumed. This
made changes in the social system discontinuous and abrupt. However,
accumulation of capital and evolution of technology is continuous.

• The key institution in Marxian system is the property right relations.

Production relations decides all other relations and structure in the
society. Each social system is identified on the basis of who owns the
factors of production namely capital.

• In this system, institutions namely relations of production that once

facilitated improvement in the forces of production ( economic develop-
ment) over time turns out to be the hindrance for further development
of the forces of production (or technology).

• Marxian system gave prime importance to technology and capital accu-

mulation and ignored the importance of natural resources.

• On the other hand, new institutional historians stress the influence of

changes in the relative resource endowments and price due to popula-
tion growth and other factors (North and Thomas 1973; North 1981).
CHAPTER 2. A THEORETICAL FRAMEWORK 15

• Further, the nature of institutional evolution may not be abrupt and

through violent measures. Institutions can change or adjusts gradu-
ally. Thus institutions can have cumulative adjustment through such
means like informal adjustments and re-interpretation of laws and
codes (David and North 1970).

• In this context it should be noted that some institutions may not be

able to change or adjust if it is inseparably intertwined with other
institutions and this is usually called ‘historical path dependency’. In
this case it needs total institutional change (The case of caste system in
India; Akerlof 1984).

• Regarding the forces behind institutional change we can make the

following observations. Usually the benefits and its distribution among
the members of a society is very uncertain, leading to weak support
for institutional change. On the other hand losses can be certain and
therefore, arise strong opposition to change (Fernadez and Rodrik 1991).
It is therefore, not uncommon to observe that a society continues to
be trapped in economic stagnation and poverty under dysfunctional
system bound by strong social inertia for the preservation of established
institutions (Basu et al 1987).

• Thus it is likely that changes in institutions and more so in culture

lag significantly behind changes in the material production base, and
resulting contradiction could often create strong social and political
tensions, culminating in major disruptions.

2.0.3 Theory of Induced Innovation

• The theoretical framework developed in the previous section is general,
but not very operational for economic analysis in the sense that the
implied hypothesis are broad for empirical testing.

Induced Technological Innovation

• Changes in the resource endowment induce changes in technology. A

standard economic theory on this relationship is called the theory of
“induced technological innovation” in the tradition of JR Hicks.

• The Hicksian theory presupposes a mechanism in which endowment

of one factor (eg. capital) becomes more abundant relative to another
CHAPTER 2. A THEORETICAL FRAMEWORK 16

factor (say labour), resulting in a decline in the relative price of capita, a

change in technology is induced towards using more capital and saving
labour for given relative factor prices.

• Such biased change in technology stems from the efforts of profit-seeking

entrepreneurs to reduce cost of production by substituting more abun-
dant (hence cheaper) resources for scarcer (hence dearer) resources.

Induced Institutional Innovation

• The theory of induced technological innovation is explained above in

terms of producers’ cost-minimising behaviour in the tradition of main-
stream economics. This theoretical structure appears to be relevant
to modern market economies in which technological innovations are
carried out mainly by large firms with R&D capacities.

• For the references on the debate on induced technological progress see

the end note 4 in the book.

• A major modification for the theory of induced innovation to cover

both primitive and high stages of development would be to combine
the theory of technological innovation with the theory of institutional
innovation. Because, whatever be the expected profitability of a new
technological innovation and however rational the producer may be, it
may not be possible for him alone to carry out the innovation. Example:
the development of agriculture along with the development of property
right institutions.

• Collective action is required not only to create institutions for promot-

ing private investment incentives, but also to undertake large scale
investment in social overhead capital such as flood control of rivers
and building of irrigation systems. Appropriate institutions must be
developed to organise people effectively for such collective action.

• Now the question is what is the mechanism that helps organise collec-
tive action to facilitate technological progress and capital accumulation
in a society.

• The most naive model would be to assume that collective action is organ-
ised when aggregate social profit a change in institutional arrangement
exceeds the cost of organising the collective action to realise a change.
CHAPTER 2. A THEORETICAL FRAMEWORK 17

• However, if such naive mechanism of institutional innovation always

operated all the economies would have developed smoothly and no great
income gap would ever have emerged between developed and developing
economies.

• This implies that to understand the causes of poverty and underde-

velopment one has to look at the reasons for the failure of induced
institutional innovation to work effectively.

Logic of Political Market

• The supply of public goods in response to social needs is determined

through a political process at equilibria between demands for and sup-
plies of these public goods from various interest groups, which might be
called political market, in analogy with economic market for goods and
services.

• Usually public goods are under-supplied because of the non-rival and

non-excludable nature of many public goods. So the organisers of the
public action may not get the whole gain and free-rider problem also
exists.

• For the supply of public goods some one must take charge of organising
collective action. Collective action is organised at various levels. For the
supply of “global public goods” widely applicable to a large number of
people in society, it is often necessary to set up a mechanism of coercion
in the of “state”.

• The collective action aimed to form and manipulate coercive power of

the state is called ‘politics’ or ‘political movement’. The organiser of the
political movement is called ‘political leader or ‘politician’.

• Returns to the leader for this cost of organising collective action collec-
tive action for the supply of public goods would be the strengthening
of his power base due to increased support from people who get the
benefits from the provision of public good. Unless the increment in his
utility arising from his strengthened political power was expected to
exceed his cost, he would not attempt to organise the collective action.

• Such behaviour of the political leader is modelled in Figure 2.2 in the

tradition of public choice theory or the economics of politics.
CHAPTER 2. A THEORETICAL FRAMEWORK 18

MR1

Politician's Marginal Revenue/Cost

MC0'

MR0
MC0
i
MC1
j

h k

MC0
MR1
MC1
MR0
MC0'
o
Q0 Q1' Q1 Q1''
Quantity of Public Good

Figure 2.2: A model of political market for a public good

• Line MR represents the decreasing marginal revenue of the politician

for increasing the supply of a public good. The marginal revenue can be
in the form of his utility, support base or the number of votes he may
get.

• Leader’s marginal cost (MC) is defined as the marginal dis-utility of his

time and effort in organising the collective action.

• The supply of public good at which the leader maximises his utility
is OQ0 , which may not be the socially optimal supply of public good.
There can be under-supply or excess supply of public good (agricultural
subsidy, LPG subsidy).

• Because of the improvement in technology both MC and MR curves

can shift. This can induce institutional innovation - like property right,
global warming and politics in US. Here read page 24-25.

• The basic question is how can modern education and information media
promote efficiency of the induced innovation mechanism involving polit-
ical processes? This is an important issue in the development economics.
The general question is how can we make sure that politician would
deliver the things required for the development of the society?
CHAPTER 2. A THEORETICAL FRAMEWORK 19

2.0.4 Historical Path-dependency

• Due to the scale economies enjoyed by the existing system historical
path dependencies can occur. Increasing returns and externalities can
prevent the migration to a new system. Transition from nomadic culture
to settled agriculture must have faced this problem, because of the scale
of nomadic culture in the initial days. In some places because of some
historical reasons settled agriculture become dominant.

• There exists multiple equilibria and to which equilibrium the society

will move depends to a large extent on its ‘historical paths’ (David 1985,
Arthur 1988, North 1990 cha.1).

• Useful insights on the emergence of multiple equilibria are also pro-

vided by evolutionary game theory (Mailath 1992; Kandori, Mailath
and Rab 1993; Aoki 2001). According to this theory, human beings can
perceive future possibilities only within narrow range based on their
own past experience, and also they tend to be concerned with short term
profit more than long-term well-being. Institutions are considered to be
the equilibria in human relationship that are reached through a trial
and error process. Since people’s choice set is bounded by their past
experience, institutions in a society are usually different from those
in other societies with different histories, even if they are surrounded
by the same economic conditions. As such, whether institutions be-
ing adopted in a society are efficient in organising economic activities
depends largely on accidents of history. In the long run, through the
working of adaptation and natural selection analogous to the process of
biological evolution, societies under efficient institutions will prosper
and expand, whereas those under inefficient will shrink and eventually
disappear.
Chapter 3

Economic Theories of
Population Growth

• When low-income economies try to escape from stagnation and set out
for modern economic development, the first problem they normally face
is acceleration in population growth and consequent relative exhaustion
of natural resources.

• When compared to the problem faced by the today’s developed countries

during their development process (or during their high economic growth
phase) the population problem today’s developing economies facing has
two important differences.

1. The growth rate of population experienced by the today’s developed

countries during their higher growth phase though higher than
the growth rate during the pre-industrialisation era, it was around
1 per cent a year and never exceeded 1.5 per cent. On the other
hand, the population growth in today’s developing economies is
much higher than this, around 2 per cent a year and in the case
of some low-income economies it is as high as 3% or more during
1980-2000.
2. The population growth in today’s developed countries during their
modern economic growth was mainly an endogenous process, i.e. it
is stimulated by economic growth or resource expansion. On the
other hand, the population growth in today’s developing countries
is, to a large extent, is exogenous, encouraged by imported health
care facilities and knowledge. For instance, Sub-Saharan Africa
recorded significant increases in life expectancy and decreases in

20
CHAPTER 3. THEORIES OF POPULATION GROWTH 21

infant mortality rates for the quarter century from 1965 to 1990,
despite decline in per capita calorie intake.

• The population growth without concomitant income growth (exoge-

nously induced population growth) has serious implications for the
developing countries including faster resource exhaustion, environmen-
tal problems and so on.

• This faster increase in the population in developing countries is usually

termed as ‘population explosion’ and this is mainly a post World War I
phenomenon.

3.1 Demographic Transition

• In this section our objective is to have a better understanding of the
population explosion in developing countries. For this, it is useful to
conceptualise natural population growth (NR) as equivalent to the
birth-rate (BR) minus the death-rate (DR).

• The rate of growth of population in an economy is this natural rate

with adjustment for migration to and from other regions. However,
in current world migration possibilities are very less as most of the
lands for new settlements were closed, we can ignore this part and
take natural growth rate of population in a country equals to birth rate
minus death rate.

• The theory of demographic transition explains the evolution of natural

growth rate in terms of evolution of BR and DR. The conventional theory
assumes following three stages.

1. High BR and DR and therefore low natural growth rate in the

pre-modern period.
2. In the first phase, with the emergence of modern economic growth,
death rate begins to fall and birth rate remains high (or even
increase as happened in England due to expansion of resource and
facilities), resulting in accelerating natural growth rate.
3. In the second phase, death rate stops declining further, but birth
rate remains high, resulting very high population growth.
4. In the third phase, birth rate begins to decline at a faster pace than
death rate, resulting in the declining natural growth rate.
CHAPTER 3. THEORIES OF POPULATION GROWTH 22

5. In the final stage, the pre-modern pattern of high birth-rate and

death-rate and therefore, low natural growth rate is replaced by
low birth-rate and low-death rate and therefore low natural rate.
In this stage, population gets stabilised.
• Now we have to think about the reasons for proposed behaviour at each
stage. The reasons for the behaviour in the third phase appears to be
that the cost for parents to have many children began to have exceed
their utility during that period (utility maximisation behaviour). Fur-
ther, child labour was prohibited, primary schooling became widespread,
minimum age to enter labour market was enacted. All these increased
the cost of child and reduced its monetary benefit.
• The utility of children as source of old-age security is also declined since
increased income, combined with development of the financial sector,
allowed people to make own arrangements to meet old age requirements
though savings and insurance. Further, the decline in infant mortality
rate, which reduced the family’s need for more children to ensure a
minimum number of surviving ones.

3.2 Economic theories of population growth

• The purpose is to explain demographic transition observed above in
terms of economic theories.

3.2.1 Malthusian theory of population

• The Malthusian theory can be summarised as follows: as with other
animals, human beings have a natural instinct to bear children to a
physical maximum; under this ‘fixity of passion’ people tend to multiply
at an exponential rate; where the production of food is constrained
by the fixed endowment of natural resources, especially land, and can
increase only arithmetically, whatever slack of food supply per capita
beyond a subsistence level may exist will eventually be used up by
increased population; further increase in population are bound to be
checked by famines, pests and wars of desperate competition for limited
food supply; thus it is not possible that levels of living and income
per capita for the majority of people can remain beyond a subsistence
minimum in the long run. The theory may be illustrated with the help
of the Figure 3.1
CHAPTER 3. THEORIES OF POPULATION GROWTH 23

Population growth rate

G

H
o _
W Wage
G

Figure 3.1: Malthusian Population Theory

• In the Figure, the line GG represents a relationship between wage rate

and population growth rate. The line cuts the horizontal axis at W̄ ,
which represents the subsistence wage rate. W̄ is barely sufficient for
the labourer and his family for a living and therefore keeps average
family size and population constant.

• Whenever, wage rate is higher than W̄ , it induces a population growth

rate and which increases the supply of labour force and eventually
eliminate the excess demand for labour. This brings wage rate back to
W̄ .

• When the wage rate is less than W̄ , the subsistence wage rate, pop-
ulation would decline and labour supply and demand equilibrium is
maintained at the subsistence wage rate.

• So in the Malthusian model sustained deviation from the subsistence

wage rate W̄ , never occurs.

• Though many people, including David Ricardo, accepted Malthus’ theory

of populations, it could not stand the test of subsequent history. In the
phase 3, which corresponds to low population growth, one can see
sustained increase in the real wage rate.

• This indicates that association between population growth and wage

rate is not linearly rising as represented by the line GG, but turns to
CHAPTER 3. THEORIES OF POPULATION GROWTH 24

MU1 MD2
MU0

Parent's marginal utility/disutility

MU2 MD1
MD0

b
c

MD2
MU1
MD1 MU0
MD0
MU2
o n2 n0 n1
Number of Children

Figure 3.2: HH utility maximisation and Number of children

be downward-sloping towards H after a certain threshold is reached as

indicated by the dotted line.

• Even though Malthus model did not stand the empirical test for the
later stage of development, it was relevant to English economy during
1770s and 1780s, when it was developed. During this period employ-
ment opportunities expanded with beginning of Industrial Revolution
following the Agricultural Revolution. Even if the wage rate per hour
may not have increased very significantly, the household income level
increased from increased working hours and employment of females
and children. Such a condition induced people in the labour class to
marry earlier and produce more children. This kind of response of
population to income growth through adjustment in the marriage can
be observed in other countries also.

• To predict the future course of demographic change in developing

economies, a more general model should be envisaged that is able
to explain both the empirical relevance of the Malthusian theory for
the early phase and its divergence from reality in the later phase of
development. Attempts to build such a model have used an approach of
maximising the utility function common to household members.

3.2.2 Household utility maximisation model

• We follow the model of Leibenstein approach as it is relatively easy to
understand, even though the Becker model is little more general in
CHAPTER 3. THEORIES OF POPULATION GROWTH 25

incorporating an explicit choice among consumption goods and number

and quality of children in parent’s utility function.

• The model assumes that parents are the sole decision makers in the
household and husband and wife have the same utility function. Their
marginal utilities and marginal dis-utilities from having an additional
child are represented by M U and M D respectively.

• Sources of utility: (1) Instinctive pleasure and love of children, and

satisfaction of having heirs, (2) expected income from children for the
household (c) security for the parents during the old age.

• Sources of dis-utility: (1) Physical and physiological hardships in bear-

ing and rearing children (2)cost paid for looking after the child, (3)
opportunity cost of parents’ labour used for child bearing and caring.
The marginal dis-utility increases with the number of children.

• increased mobility, reduced infant mortality rate, social security system.

• Changes in values and institutions are essential, particularly women

education. So there is a lag in the response of population growth to
economic development, inertia.
Chapter 4

Ricardian model of development

• As we know the developments of the human society has been realised

through developments in technology and institutions that facilitate
substitution of man-made capital for natural resources.

• Malthus’ theory focusing on the side of human behaviour driven by

animal instincts without due regard for capital formation activities
could be a theory of population, but could hardly be called a theory of
development. However, note that Malthus theory predicted that in the
long run real income for a majority of the people would not go above the
subsistence level. To that extent it is concerned with development.

• It is first David Ricardo who delineated a mechanism in which economic

development can be constrained by the scarcity of natural resources by
building an explicit theory of economic development.

• Ricardo’s theory identified capital accumulation in modern industries

as the driving force of economic growth. ‘Capital’ in his view was the
‘wage fund’ defined as as the sum of payments to labour in advance of
sale of commodities produced by the labour applied as well as payments
for the purchase of tools and structures complementary to the use of
labour. Therefore, demand for labour increases proportionately with
the increase in the wage fund. While supply of labour is determined
by the number of labourers existing who are willing to work full time
regardless of the wage rate. This implies that labour supply is constant
in the short run (defined as the period during which population is
constant).

• Therefore, investments, additions to the wage-fund, increases the de-

mand for labour and given the inelastic supply of labour in the short

26
CHAPTER 4. RICARDIAN MODEL OF DEVELOPMENT 27

run, wage rate increases. The increase in the wage rate above subsis-
tence wage rate induces population to increase in the Malthusian sense,
which in turn increases labour supply and bring don the wage rate to
subsistence level. Therefore, in the long run labour supply is considered
to be perfectly elastic.

• Thus in the long run wage cost to the industry does not rise, so profit
increases in proportion to the increase in capital. Since the rate of profit
does not decline, incentive to reinvest the profit is maintained. So that
production and capital continue to increase in the modern sector.

• However, the subsistence wage for industrial workers depends on food

prices. Agriculture cannot escape from decreasing returns in production
since it is constrained by the endowment of the land.

• Increased demand for food due to population growth brings low fertile
lands into cultivation, increasing the marginal cost of food production
and thereby the food prices.

• Extension of cultivation to inferior quality lands and the consequent

increase in the food prices cultivation in superior quality lands more
profitable. This increases the rents of superior quality lands.

• As food prices increase, the money wage to the industrial workers need
to be raised to maintain the subsistence living. As the wage cost raises,
profit does not continue to increase proportionately with the increase in
capital.

• Thus as food demand continues to increase due to capital accumulation

and consequent employment growth, food prices would eventually raised
to a level at which the rate of profit will become so low as to provide
no incentive for further investment. Economic growth will stop at this
moment.

• The Ricardo model can be presented as a model in modern economics as

given in Figure 4.1. The important features of the figure are:

1. Horizontal supply curve of labour at subsistence wage rate OW1 in

the long run as represented by LS. However labour force remains
constant in the short run – for one period SS represents the short-
run supply curve.
2. DD0 represents the demand for labour corresponding to the capital
stock Ko . The long run equilibrium in the initial period is at point
CHAPTER 4. RICARDIAN MODEL OF DEVELOPMENT 28

A with employment of OL0 at the subsistence wage rate. The

total value of output of the industry is represented by area ADOL0
of which AW1 OL0 is paid to the workers and the remaining area
ADW1 become profit or return to capital.
3. The shaded area is the rent paid to the owners of the first-grade
land.

• Solutions to Ricardian trap that industrialisation in the early stages

may face. Increase agricultural productivity as importing food is not a
solution due to insufficient foreign exchange, higher international food
prices etc.

(a)cLabourcmarketcforcindustry (b)cProductcmarketcforcagriculture
SS
D
HS
Wagecrate

d2
Corncprice

d1
W2
G E d0

W1 LS P2
A B C

D2 P1
D0 D1
d2
d0 d1
O L0 L1 L2 O O1 Q2
Employment Corncoutoutc/consumption

Figure 4.1: Ricardo model of economic development

• The mechanism of fixed land resources endowments that constrain

economic growth in the early stage of industrialisation is called ‘the
Ricardian trap’.
Chapter 5

Economic Growth: Empirics

• Economic growth is typically understood as well as measured in terms

of the growth of per capita Gross Domestic Product (GDP). Sustained
economic growth at positive rate is fairly a recent phenomenon in the
human history, most of it having occurred in the last 200 years.

• Most part of the human history was characterised by Malthusian phe-

nomena, technological progress and population growth were minuscule
by modern standards and average growth rate of per capita income in
various regions of the world were even slower due to the offsetting effect
of population growth on the expansion of resources per capita.

• During the Malthusian epoch, resource expansion due to technological

progress and land reclamation were utilised mainly for maintaining an
increased population, with minor long run effect on income per capita.
Technological progress and availability arable land generated temporary
gains in income per capita leading eventually to a larger population, but
not richer population. Technologically superior countries had ultimately
denser population, but their standard of living did not reflect the degree
of their technological advancement(Galor, 2010, p.57)1 .

• In the post Malthusian era, the pace of technological progress increased

substantially along with the process of industrialisation, triggering a
take off from the Malthusian trap. Growth rate of income per capita
increased significantly, but the positive Malthusian effect of income
per capita on population growth was still maintained, resulting in a
sizeable increase in the growth rate of population.
1
The case of China and India.

29
CHAPTER 5. ECONOMIC GROWTH: EMPIRICS 30

• The rate of expansion of resources was, however, high enough to main-

tain a positive growth in the income per capita. The average growth
of income per capita in the world jumped from 0.05% per year during
1500-1820 to 0.53 per cent during 1820-1870 and to 1.3% per year dur-
ing 1870-1913. However, the timing of the take-off and its magnitude
differed across regions and countries.

• The acceleration in the technological progress during the post-Malthusian

regime and the associated increase in income per capita stimulated ac-
cumulation of human capital in the form of literacy rates, schooling and
health. The increase in the investment in human capital was facilitated
by the rise in income per capita, as well as by qualitative changes in the
economic environment that increased the demand for human capital,
encouraging the households to invest in the education of their offspring.

• In the first phase of the Industrial Revolution, in which human capital

had a very limited role in the production process, education was moti-
vated by a variety of reasons, such as religion, enlightenment, social
control, moral conformity, socio-political stability, social and national
cohesion, and military efficiency. The extensiveness of public educa-
tion was therefore not correlated with the industrial development and
it differed across countries according to social, political and religious,
cultural and institutional factors.

• In the second phase of Industrial Revolution, however, the demand for

education increased, reflecting the increasing skill requirements in the
process of industrialisation (Galor, 2010, p.62)2 .

• The acceleration in the rate of technological progress in the second

phase of the industrialisation and its interaction with human capital
formation, triggered a demographic transition, paving the way to a
transition to an era of sustained economic growth for the change in the
growth rate of population). In the post demographic transition period,
the rise in the aggregate income due to technological progress and factor
accumulation was no longer offset by population growth, permitting
sustained growth in per capita income in regions that experienced
sustained technological progress and accumulation of physical and
human capital (Galor, 2010, p.62).
2
In this context Galor and Moav (2006) argues that economic interest of the capitalists
were a significant driving force behind the implementation of educational reforms.
CHAPTER 5. ECONOMIC GROWTH: EMPIRICS 31

• As we have already noted above the timing and magnitude of take-off

from the Malthusian trap was not uniform across regions and coun-
tries. The take-off from the Malthusian epoch and the transition to
post-Malthusian regime occurred in Western Europe, the Western Off-
shoots (that is USA, Canada, Australia and New Zealand) and Eastern
Europe at the beginning of the 19th century, whereas in Latin America,
Asia and Africa it happened towards the beginning of the 20th century.
This variation in the timing and pace of the take-off resulted in con-
siderable rise the per capita income differences between rich and poor
countries/region in the world.

• The variation in the timing and pace of the take-off from the Malthu-
sian era across countries reflects the initial differences in geographical
factors and historical accidents and their manifestation in diversity in
institutional, demographic and cultural factors, trade patterns, colo-
nial status, and public policy. In particular, with the emergence of
technology driven demand for human capital in the second phase of
industrialisation, the prevalence of human capital-promoting institu-
tions determined the extensiveness of human capital formation, the
timing of the demographic transition and the pace of the transition from
stagnation to growth (Galor, 2010, p.58-59). This is the background of
the evolution of the modern growth theory which probes into the causes
behind the different growth experiences of countries.

• Economic growth is typically measured as the change in per capita

Gross Domestic Product (GDP). Sustained long-term economic growth
at positive rate is fairly a recent phenomenon in human history most of
it having occurred in the last 200 years. According to Maddison (2007)
estimates per capita GDP in the world economy was no higher than in
the year 1000 than in the year one, and only 53 per cent higher in 1820
than in 1000, implying an annual average growth rate of 1/19% during
the 820 years. By 2000, world percapita GDP had risen to more than 8.5
times its 1820 value. The main task of the growth theory is to explain
the variation in the living standards across time and countries(Howitt
and Weil, 2010).

• “By modern standards, all countries were poor in 1700 but since then
sustained growth, first in the United Kingdom and parts of the Western
Europe and more recently in the United States and parts of the Asia
Pacific region has resulted in large cross-country differences living
standards. In 2000, average GDP per worker in some countries was
about one-fifth of that in the United States, while more than 40% of
CHAPTER 5. ECONOMIC GROWTH: EMPIRICS 32

the world’s population lived in countries with average levels of GDP

per worker of no more than ten per cent of that in the United States”
(Durlauf and Johnson, 2010, p.52).
Chapter 6

Economic Growth: Theory

• The technical core of modern growth theory rests upon technical change,
specialisation, factor substitution, and factor accumulation with various
recent theories emphasising the effects on these of trade, institutions,
inequality, political economy, geography and population size. However,
all these were the concerns of classical economists, even if they used a
different vocabularyCameron (2010).

• The key difference between neoclassical and modern growth economics

is its domain of explanation: whereas neoclassical theory attempted to
understand the long run behaviour of advanced industrialised economies,
the new growth economics attempts to understand worldwide growth
pattern. As a result, the difference between the advanced industri-
alised economies and the rest of the world take on primary importance”
(Durlauf and Johnson, 2010, p.54).

• The focus of growth economics has shifted to identifying features of

international divergence rather than international convergence. There-
fore, the effort to identify the patterns that characterise the differences
in cross-country growth experiences has led to empirical research that
focuses on the identification of particular factors in generating the diver-
gence. The theoretical work in growth economics moved away from the
traditional emphasis on factor accumulation and towards the analysis
of a wide range of social, historical, geographical and political factors as
sources of cross-country heterogeneity(Durlauf and Johnson, 2010). See
also (Brock and Durlauf, 2001).

• The fundamental question raised is why some countries are rich and
why some are poor? Adam Smith’s book An Inquiry into the Nature and
Causes of Wealth of Nations, is an early attempt to answer this question.

33
CHAPTER 6. ECONOMIC GROWTH: THEORY 34

Economic historians like David Landes also explored the same question
in his lecture to the American Economic Association titled “Why are We
So Rich and and They So Poor”1 .

• Modern examination of the issue of economic growth started with

Solow’s famous work (Solow, 1956). Solow’s work made clear the role
of capital accumulation and technological progress in economic growth.
During the 1960’s and 1970’s, though the issue of economic growth was
at the centre of attention, formal theoretical analysis of some of the
important aspects were postponed because of methodological reasons.

• Theoretical work on economic growth got a bigger push in the 1980s

with the works Romer (1986a) and Lucas (1988), they respectively em-
phasised the role of “ideas” and “human capital” in economic growth.
Taking advantage of developments in the economics of imperfect compe-
tition, Romer introduced economics of technology to macroeconomists.
These developments led to the emergence of what is now called new
growth theory or endogenous growth theory. The new growth literature
stimulated lot of empirical research testing its predictions.

6.1 Solow Growth Model

• As we have noted above, the first formal analysis of economic growth is
Solow (1956). This section explains the Solow model in detail. As any
theoretical model, this model is also an abstract representation of the
reality, capturing only very important aspects of an economy.2

• As any economic model, we also start with following set of simplifying

assumptions. First, the world consists of countries producing only a
single homogeneous good (output). Conceptually as well as for testing
the model using empirical data, it is convenient to think of this output
as units of country’s Gross Domestic Output (GDP). One implication of
this assumption is exclusion of international trade. Second, technology
is exogenous, implying that technology available to firms is unaffected
by their own actions.
1
See Landes (1990).
2
For more details on economic models see: Rodrik (2015); Rubinstein (2006).
CHAPTER 6. ECONOMIC GROWTH: THEORY 35

6.1.1 Solow Model Without Technological Progress

• The Solow model is built around two equations, a production function
and a capital accumulation equation. The production function describes
how inputs, such as capital and labour are combined to produce output.
To simplify the model, we assume that there are only two categories of
inputs, namely labour pLq and capital pKq. We denote output by Y . The
production function is of Cobb-Douglas form as given below:
•
Y “ F pK, Lq “ K α L1´α (6.1)
where 0 ă α ă 1.
• The production function (11.1) exhibits diminishing returns to individ-
ual factors of production and constant returns to scale. Firms in the
economy pay workers a wage w for each unit of labour and pay r in
order to rent a unit of capital for one period.
• We assume that perfect competition prevails in the economy and so
firms are price-takers. Normalising the price of output of our economy
to unity, profit maximising firms solve the following problem.
•
max F pK, Lq ´ rK ´ wL
K,L

This gives us the following first order conditions:

BF Y
w“ “ p1 ´ αq
BL L

BF Y
r“ “α
BK K
Notice that w.L ` r.K “ Y , i.e payments to the inputs (factors of produc-
tion) completely exhaust the value of output produced, so that there are
no economic profit to be earned.
• The second key equation of the Solow model is the equation describing
capital accumulation. It is given by:
K9 “ sY ´ dK (6.2)
Where K9 “ BKBt
and d is rate of depreciation. The equation (6.2) says that
change in the capital stock K9 is equal to the amount of gross investment
sY less the amount of depreciation that occurs during the production
process, dK.
CHAPTER 6. ECONOMIC GROWTH: THEORY 36

• The third equation describes the evolution of labour force in the economy.
Here we assume that labour force participation rate in the economy is
constant and the population growth rate is given by the parameter n.
9
This implies that growth rate of labour force, LL , is also n. So labour
force in the economy at time t is

Lt “ L0 ent (6.3)

• As we are interested in the evolution of per capita output, we reformu-

late the production function in terms of output per worker 3 .

• Let y and k respectively denote output per worker ( YL ) and capital

per worker ( K
L
). Now the production function (11.1) can be written as
follows:

y “ kα (6.4)

y
y

Figure 6.1: Cobb-Douglas Production Function

• Figure 6.1 plots the production function (6.4). The figure shows that
each additional unit of capital per worker increases output at a lesser
rate because of the operation of diminishing returns.
3
The growth rates of per capita income and output per worker are equal because there
exists a proportional relationship between population and labour force due to the assumption
of constant worker participation rate. Therefore, the evolution of per capita income can be
analysed through output per worker.
CHAPTER 6. ECONOMIC GROWTH: THEORY 37

• Capital accumulation equation can also be written in terms of capital

per person as follows. Writing k in logarithmic terms and deriving its
growth rate as follows.

log k “ log K ´ log L

B log k k9 K9 L9
“ “ ´
Bt k K L
Now we can write the capital accumulation equation in per worker
terms:

k9 “ sy ´ pn ` dqk

• The above equation says that the change in the capital per worker in
each period is determined by three terms. Investment per worker, sy,
increases k, while depreciation per worker dk reduces k. Population
growth also reduces k. In each period nL new workers are around, who
were not there during the last period. If there are no new investment
and no depreciation, capital per worker declines by the amount nk.

Solving Solow Model

• Solving a model means obtaining the values of endogenous variables,
given the values of exogenous variables and parameters. Ideally, we
would like to be able to express each endogenous variables as a function
of only exogenous variables and parameters. So that we can examine
the evolution of endogenous variables at every point of time.

• In the present context, to solve the Solow model we adopt diagrammatic

approach, which is simple and provide greater insights into the solution.

• The two key equations describing the evolution of two endogenous

variables in our model are:
y “ kα

k9 “ sy ´ pn ` dqk

Now assume that the economy starts with a given stock of capital
per worker, k̂ and a given population growth rate, depreciation and
investment rate. The basic question is how does output per worker
CHAPTER 6. ECONOMIC GROWTH: THEORY 38

evolve over time in this economy. One could also ask questions like,
how does output per worker compares in the long run between two
economies that have different investment rates? These questions are
analysed in the Solow diagram given in Figure 6.2.

(n+d)k

sy

k^ k* k

Figure 6.2: Basic Solow Diagram

• The curve sy represents the amount of investment per person at various

levels of income. As sy “ sk α , this curve has the same shape of the
production function plotted in Figure 6.1, but it is shifted down by the
factor s. The second line pn ` dqk, represents the amount of new invest-
ment per labour required to keep k constant in the face of increasing
population and depreciation. The difference between these two curves
is the change in the amount of capital per worker. When the difference
is positive, the economy is increasing its capital per worker.

• When capital per worker is increasing we say that capital deepening is

occurring. When the change in the per worker capital is zero, but the
total capital stock K is growing (because of population growth) we say
that capital widening is taking place.

• At k̂ investment exceeds what is required to keep the capital per worker

constant in the face of depreciation and population growth. The capital
deepening will continue until k “ k ˚ . At this point the amount of capital
per worker remains constant and this point is called steady state.

• If the economy starts with a capital stock per worker larger than k ˚ ,
the amount of investment is less than what is required to keep capital-
labour ratio constant. The term k9 is negative and therefore capital per
worker begins to decline. This decline would continue until capital per
worker falls to k ˚ .
CHAPTER 6. ECONOMIC GROWTH: THEORY 39

• Note that the Solow diagram determines the steady state value of
capital per worker. The production function (6.4) then determines the
steady state value of output per worker y ˚ as a function of k ˚ . This is
shown in figure 6.3.

{
y* (n+d)k
Consumption

sy
sy*

k* k

Figure 6.3: Solow Diagram with Production function

Comparative statics
• Comparative statics are used to understand the response of the model
to changes in the parameters. Two important parameters in the Solow
model are saving rate s and population growth rate n. We consider
two cases, an increase in the investment rate and an increase in the
population growth rate.

Increase in investment rate: As shown in the part a of Figure 6.4

an increase in the investment rate shift the sy curve upward to sy 1 .
It shows that the new steady state is associated with higher level of
capital stock per worker k ˚˚ . From the production function we know
that a higher level of capital per worker is associated with higher level
of output; the economy is now richer than it was before.
•

Increase in the population growth rate: Suppose that the econ-

omy has reached a steady state, but then because of immigration, for
example, the population growth rate of the economy increased from n
CHAPTER 6. ECONOMIC GROWTH: THEORY 40

(n'+d)k
(n+d)k
(n+d)k
sy
s'y

sy

k** k* k
k* k** k
(b) Change in Population Growth
(a) Change in Saving Rate
Rate

Figure 6.4: Comparative statics in Solow Model

to n1 . The effect of this increase in the population growth rate on k and

y is depicted in the part b of Figure 6.4. The pn ` dqk curve rotates up
and to the left to the new curve pn1 ` dqk. The new steady state level
of capital stock is k ˚˚ less than the old one, k ˚ . So higher population
growth rate is associated with lower level of output per person.

Properties of the steady state

• At steady state k9 “ sk α ´ pn ` dqk and setting this equation equal to
zero yields.
ˆ ˙1{1´α
˚ s
k “
n`d

Substituting into the production function gives the steady state output
per worker, y ˚ .
ˆ ˙α{1´α
˚ s
y “
n`d

• The model predicts that countries having higher rate of investment will
also have higher level of per capita income and countries having higher
population growth rate tend to be poorer. This prediction seems to be
supported by data Jones (2001).
CHAPTER 6. ECONOMIC GROWTH: THEORY 41

Economic Growth in the Simple Model

• At steady state level of income there is no per capita income growth.
Output per worker (and therefore output per person, as we have as-
sumed constant labour force participation rate) is constant in the steady
state. Output, Y , is growing at the rate of population growth rate.
An economy that begins with a stock of capital per worker below the
steady state value will experience growth rate in k and y along the
transition path to the steady state. This can be shown using the capital
accumulation equation.

k9 “ sy ´ pn ` dqk

“ sk α ´ pn ` dqk
k9
“ sk α´1 ´ pn ` dq
k

y
“ s ´ pn ` dq (6.5)
k
• From the equation (6.5) it is clear that as k rises, the growth rate of
k gradually declines4 . As the growth rate of y is proportional to the
growth rate of k, the growth rate of y also declines gradually. This is
depicted in figure 6.5. The curve sy{k slopes downward because the
higher the level of capital per worker lower the average product of
capital y{k because of diminishing returns to capital accumulation. In
the figure, the difference between horizontal line pn ` dq and sy{k line
9
is k{k. The growth rate of k is positive to the left of k ˚ and negative to
the right.

6.1.2 Technology and Solow Model

• In Solow model sustained growth in per capita income is achieved by in-
troducing technological progress. Technological progress is incorporated
by adding a technology variable A to the production function.

Y “ F pK, ALq “ K α pALq1´α (6.6)

4 9 can also be proved by differentiating k{k
9
This inverse relationship between k and k{k
9
Bpk{kq
with respect to k. Bk “ pα ´ 1qsk α´2 . This term is negative as 0 ă α ă 1.
CHAPTER 6. ECONOMIC GROWTH: THEORY 42

.
k/k

(n+d)

y
s k_

k* k

Figure 6.5: Economic Growth in the Simple Model

Entered in this way technology variable A is said to be “labour augment-

ing” or “Harrod-Neutral”. Technological progress makes labour more
efficient.5

• Technological progress is assumed to be exogenous. This means that we

are not modelling the process of technological progress, instead simply
recognise for the moment that there is technological progress and make
the assumption that A is growing at a constant rate:

A9
“ g ðñ At “ A0 egt
A
where g is a parameter representing rate of technological progress.
Note that this assumption of technological progress is unrealistic and
relaxing this assumption is one of the major accomplishment of “new”
growth theory.

• In the model with labour augmenting technology, capital per efficiency

unit of labour is

K
κ“ (6.7)
AL
and output per efficiency unit of labour is
5
In this formulation, technological progress enhances the effectiveness of labour force,
which is equivalent to an increase in population.
CHAPTER 6. ECONOMIC GROWTH: THEORY 43

Y K α pALq1´α
φ“ “ “ κα (6.8)
AL AL
• The differential equation representing capital accumulation in the
model with technology can be derived in the same way as equation
(6.2). The rate at which new saving raises κ is the rate of saving per
efficiency unit of labour sφ. Depreciation causes κ to decrease at the
rate d. In addition, growth in the number of efficiency unit of labour at
n ` g rate also causes κ to fall at the rate of pn ` gqκ. The net change in
κ is the result of all these three forces. Taking log on both sides of equa-
tion (6.7) and differentiating with respect to time, we get the following
differential equation showing capital accumulation in the model with
technology.
κ9 “ sκα ´ pd ` n ` gqκ

This is almost identical to the previous k,9 except the growth rate of
labour force n has now been replaced by the growth rate n ` g, growth
in the number of efficiency units of labour.
• As before κ will approach a unique steady-state value κ˚ in the long
run, while φ approaches its steady state φ˚ “ pκ˚ qα . Although output
per efficiency units does not grow in the long run, the same is no longer
true of output per person. Writing output per person, y, as follows and
taking log derivative with respect time we get,

Y
y“ “ Aφ “ Aκα
L
lny “ lnA ` αlnκ
y9 A9 κ9
“ `α
y A κ
• In the long run when κ approaches κ˚ , the time derivative κ9 approaches
zero, so the growth rate of output per person, gy approaches the exoge-
nous rate of technological change, g.

gy Ñ g, as t Ñ 8

But in the short run, as before, the growth rate can also raise above g
temporarily as a result of an increase in the saving rate, s which raises
the rate of increase in the capital stock per efficiency labour, according
to the fundamental differential equation.
CHAPTER 6. ECONOMIC GROWTH: THEORY 44

• Intuitively, the growth rate of output per person does not fall to zero
because as capital accumulates, the tendency for output/capital ratio
to fall because of the diminishing return to capital is continually offset
by technological progress. The economy approaches a steady state in
which the two opposing forces of diminishing returns and technological
progress exactly offset each other and output/capital ratio is constant6 .
Although the height of the steady state path will be determined by the
parameters such as saving rate s, population growth rate n, and rate
of depreciation d, the only parameter determining the long run growth
rate is the exogenous technological progress, g.

6.1.3 Convergence
• Solow model predicts that, in the long run an economy would end
up in its steady state level of per capita income, y ˚ and capital per
efficiency labour, k ˚ . Steady state values of y and k depend on the
exogenous structural parameters of the model, namely s, d, n and the
level of technology.

• If the structural parameters are same across countries, the steady state
values of k and y would be same for all of them and the model predicts
that, in the long run, all countries would converge to a common steady
state level of per capita income, y ˚ and capital per efficiency labour, k ˚ .
This concept of convergence of countries or regions to a common steady
state is called unconditional convergence or absolute convergence.

• Absolute convergence would occur irrespective of the initial state of

these economies, as measured by their starting levels of per capita in-
come (or equivalently their initial levels of capital per efficiency labour).
This implies that, if the structural parameters, s, n, d and level and
rate of growth of technology, governing the evolution of the economy
are similar across countries, history in the sense of different initial
conditions does not matter.
6
Note that in macroeconomics ‘balanced growth’ refers to classes of equilibrium growth
paths, while in development economics the term refers to a particular development strategy.
The concept of balanced growth is central element of macroeconomics. It refers to an
equilibrium in which major aggregates, usually but not exclusively output and capital stock,
grow at the same rate over time and the real interest rate is constant. Most textbook growth
models are constructed in a way that delivers this outcome. This is motivated partly by
theoretical convenience but also by historical observation. The conventional wisdom is that
the real interest rate and the capital-output ratio are surprisingly stable over long spans of
time, at least in the developed countries.
CHAPTER 6. ECONOMIC GROWTH: THEORY 45

• Figure 6.6 illustrates the concept of unconditional convergence.

D
B
F

log(Per capita Income) A

Time

Figure 6.6: Unconditional Convergence

• The figure 6.6 plots the log of per capita income against time, so that
constant rate of growth of income (such as that experienced at the
steady-state) appears as a straight line. The line CD plots the time path
of log per capita income at the steady state. The path EF represents a
country that starts below the steady state level of capital per efficiency
labour. According to the model this country will initially experience a
rate of growth that exceeds the steady state level and its time path of
(log) per capita income will move asymptotically towards the CD line as
shown. Over time its rate of growth will decelerate to the steady-state
level.

• Similarly, a country that starts close to the steady state, say A, will
experience a relatively lower rate of growth because its time path AB
of (log) income flattens out to converge to the line CD from above.
This description of convergence allows us to derive one of its testable
implications. It says that there exists a strong negative relationship
between growth rates of per capita income and the initial value of per
capita income7 .

• Assume that we have a sample of n countries, we can use the following

7
This is also evident from the fact that in the Solow model smaller values of k are
associated with higher growth rate of k. It implies that economies with lower capital per
person tend to grow faster in per capita.
CHAPTER 6. ECONOMIC GROWTH: THEORY 46

regression model to test the hypothesis of absolute income convergence.

gyi “ α ´ βyi0 ` ui
Where gyi is the ith country’s growth rate of per capita income over a
period of time and yi0 is the initial year per capita income. α and β
are the regression parameters and ui is the error term. A statistically
significant negative β suggests absolute income convergence among the
countries.
• A number studies empirically verified the hypothesis of unconditional
convergence. As long time series data on large number of countries,
the most appropriate data set to verify this hypothesis, is difficult to
get, empirical studies used to types of data sets; (1) long time series
data on small number of countries and (2) short time series on a large
number of countries. Important studies examined the unconditional
convergence include Baumol (1986) and De Long (1988).
• The empirical studies show that the prediction of the Solow model
of absolute convergence seems to be rejected by the data; suggesting
modifications to the model in order to be consistent with data. However,
the model in its simplest form is useful as it provides some directions for
modification. The assumption, which makes growth rates sensitive to
income level in Solow model, is diminishing returns to capital8 . As one
can observe, the assumption of diminishing returns to capital seems to
fit well with the real world. Physical capital does need labour to operate.
We are far from a world of complete automation. There is little doubt
that in the absence of other inputs or technical progress, straightforward
accumulation of machines would fail to provide corresponding increase
in output.
• One obvious weak basis of the hypothesis of absolute convergence is
the assumption that structural parameters, namely saving rate, rate of
technological progress, population growth rate and depreciation rate,
are same across countries. This assumption flies in the face of facts;
countries certainly differ in many, not all, of these aspects. However,
relaxation of this assumption does not invalidate the Solow’s prediction
of convergence of countries to their own steady states determined by
country specific values of parameters. The concept of convergence that
accommodates heterogeneity in structural parameters and therefore
different steady states across countries is called conditional convergence.
8
Note that in Harrod-Domar model, which assumes constant rate of returns to capital,
growth rate is neutral to income level.
CHAPTER 6. ECONOMIC GROWTH: THEORY 47

Conditional Convergence:
• To discuss the concept of conditional convergence, we retain the assump-
tion that knowledge flows freely across countries making technological
knowledge same for all, but allow other parameters, s, d and n to differ.

• Solow model, s, n and d have only level effects and the growth rate of per
capita income in the long run is determined entirely by technological
progress, which we assume same across countries. This makes the
model to predict convergence in growth rates. Although long run per
capita income varies from country to country, the long run rate of growth
of per capita income is same for all countries.

• Unlike the hypothesis of absolute convergence, it is not necessary for

the conditional convergence that poor countries should grow faster.
The hypothesis only says that countries which are much below of their
steady-state level of income grow faster. This idea can be explained
using the Figure 6.7.

B
log(per capita income)

A F

Time

Figure 6.7: Conditional Convergence

• The Figure 6.7 plots log of per capita income against time. Countries
starting at points A and B have same steady state points and countries
starting at points E and G are approaching to another steady state point.
As we assume same rate of technological progress across countries and
therefore same rate of growth of per capita income, at steady state
levels of income time paths of log incomes are parallel to each other.

• Now note that initial income of country starting at point E is less

than that of country starting at point C, still the growth rate of second
CHAPTER 6. ECONOMIC GROWTH: THEORY 48

country is higher than that of first one because of its higher steady state
level of income. Thus in the case of conditional convergence, it is not
necessary that poor countries should grow faster than rich countries. It
only implies that countries staying much below their steady state level
of income would grow faster.

Beta and Sigma Convergence:

• In the empirical literature on convergence across countries and regions
two concepts of convergence can be seen (see Barro and Sala-I-Martin
(1992)). The first concept of convergence, known as β convergence, takes
place if a poor economy tends to grow faster than a rich one, so that
the poor country tends to catch up with rich ones in terms of levels of
per capita income. The second concept is related to the cross sectional
dispersion of per capita income. Convergence occurs if the standard
deviation of the log of per capita income of a group of countries or regions
declines over time. This notion of convergence is called σ convergence.
As shown below, even if there is absolute convergence, dispersion of log
per capita income need not decline over time.

• Suppose that absolute convergence holds for a group of economies i “

1...N , where N is a large number and we have data for per capita output
corresponding to a discrete time, for instance annul data. The real per
capita income for economy i can then be approximated by the process:

log yit “ a ` p1 ´ bq log yit´1 ` uit (6.9)

Where a and b are constants, with 0 ă b ă 1 and uit is disturbance

term. The condition that b ą 0 implies absolute convergence because
the annual growth rate, logpyit {yit´1 q is inversely related to logpyit´1 q. A
higher coefficient b corresponds to greater tendency toward convergence.
The disturbance term picks up temporary shocks to the production
function, saving rate, and so on. We assume that uit has zero mean, the
same variance σu2 for all economies and is independent over time and
across economies.

• One measure of dispersion of per capita income among countries is the

sample variance, Dt , of log yit :

N
1 ÿ
Dt “ . rlogpyit q ´ µt s2
N i“1
CHAPTER 6. ECONOMIC GROWTH: THEORY 49

Where µt is the sample mean of the logpyit q. Now we can derive the
evolution of Dt using the equation (6.9).

Dt « p1 ´ bq2 .Dt´1 ` σu2

Let D˚ denotes the steady state level of dispersion, it can be expressed

as follows:
D˚ “ σu2 {rp1 ´ p1 ´ bq2 s
Hence the steady state dispersion falls with b (the strength of the con-
vergence effect) but rises with the variance σu2 of the disturbance term.
The results show that convergence of the first kind (poor countries tend-
ing to grow faster than rich ones) tends to generate convergence of the
second kind (reduced dispersion of per capita income) but this process
could be offset by new disturbances that tend to increase dispersion.9

9
Read Hart (1995) for more on this.
Chapter 7

Growth Accounting

• Growth accounting provides a breakdown of observed economic growth

into components associated with growth in factor inputs and a residual
that reflects technological progress and other elements (Barro, 1999).

• Generally, growth accounting exercise is viewed as a preliminary step

for the analysis of fundamental determinants of economic growth. The
growth accounting exercise can be particularly useful if the fundamental
determinants that matter for factor growth rates are substantially
independent from those that matter for technological change.

• The topics we discuss include primal and dual approaches to growth

accounting, growth accounting when increasing returns are present due
to spillovers, and the interpretation of Solow residual as a measure of
technological change.1

7.1 Standard Primal Growth Accounting

• Consider an neoclassical production function

Y “ F pA, K, Lq (7.1)

where A is the level of technology, K is the capital stock, and L is the

quantity of labour. Capital and labour can be disaggregated among
types or qualities as in Jorgenson and Griliches (1967).
1
Barro (1999) also discusses growth accounting in the context of taxes and multiple inputs
are used.

50
CHAPTER 7. GROWTH ACCOUNTING 51

• The growth rate in output can be partitioned into components associ-

ated with factor accumulation and technical progress. Differentiating
equation (9.1) with respect to time yields, after division by Y and rear-
rangement of terms,

Y9
ˆ ˙ 9 ˆ ˙ 9
FK K K FL L L
“g` ¨ ` ¨ (7.2)
Y Y K Y L

where FK and FL are the factor (social) marginal products and g - the
growth due to technological change – is given by
ˆ ˙ 9
FA A A
g” ¨ (7.3)
Y A

If the technology factor appears in Hicks-neutral way, so that F pA, K, Lq “

9
A ¨ F̃ pK, Lq, then g “ A{A.

• The rate of technological progress g can be calculated from equation

(9.2) as a residual,

Y9
ˆ ˙ 9 ˆ ˙ 9
FK K K FL L L
g“ ´ ¨ ´ ¨ (7.4)
Y Y K Y L

• However, equation (9.4) is impractical because it requires knowledge of

the social marginal products, FK and FL . In practice, the computations
typically assume that the social marginal products can be measured by
observed factor prices.

• If factors are paid their social marginal products, so that FK “ R(the

rental price of capital) and FL “ w (the wage rate), then standard primal
estimate of the rate of technological progress follows from equation (9.4)
as
Y9 K9 L9
ĝ “ ´ sK ¨ ´ sL ¨ (7.5)
Y K L
where sK ” RK{Y and sL ” wL{Y are the respective shares of each
factor payment in total product.

• The value ĝ is often described as an estimate of total factor productivity

(TFP) growth or Solow residual.

• The condition sK ` sL “ 1 or Y “ RK ` wL must hold if all the income

associated with the gross domestic product Y is attributed to one of the
factors, restricted here to capital and labour.
CHAPTER 7. GROWTH ACCOUNTING 52

• In an international context, some net factor income may accrue to

foreign-owned factors, and RK `wL would include this net factor income
also.
• The equation of output Y to total factor income is consistent with
the equality between the factor prices and marginal products if the
production function, F p¨q, is of constant returns to scale in K and L, so
that Y “ FK K ` FL L holds.
• Using sk ` sL “ 1, equation (9.5) can be rewritten in intensive form as
y9 k9
ĝ “ ´ sK
y k
where y ” Y {L and k ” K{L are quantities per unit of labour.
• Jorgenson and Griliches (1967); Jorgenson, Gollop, and Fraumeni (1987)
stressed the importance of disaggregating inputs by quality classes.
Failure to include input quality (of both labour and capital) in this way
would overestimate the estimated total factor productivity or Solow
residual.
• An alternative approach to measure productivity growth is to regress
growth rate of output Y9 {Y on the growth rates of inputs, K{K9 and
9L{L, in the form of equation (9.2). The intercept then measures g,
` ˘ ` ˘
and the coefficients on the factor growths measure FKY K and FYL L ,
respectively. The main advantage of this approach is that it dispenses
with the assumption that the factor social marginal products coincide
with the observable factor prices – that is, FK “ R and FL “ w.
• However, the disadvantages of the regression approach are the follow-
ing,
– The variables K{K9 9
and L{L cannot usually be regarded as exoge-
nous with respect to variations in g. As we have already seen
factors get accumulated in response to technological progress. This
makes OLS estimates inconsistent.
9
– If K{K 9
and L{L are measured with error, then estimated `coeffi-
FK K
˘
cients` of these
˘ variables would be inconsistent estimates of Y
and FYL L respectively. This problem can especially serious for the
growth rates of capital input, as correct measurement of actually
used capital stock is sometimes not possible. And this usually
results in downward estimates for the coefficient of the growth rate
of capital.
CHAPTER 7. GROWTH ACCOUNTING 53

– The regression framework has to be extended to allow for changes

in factor shares and TFP growth rates over time.

7.2 Dual Approach to growth accounting

• In the dual approach, the Solow residual is computed from growth rates
of factor prices rather than factor quantities. It can be derived directly
from the equality between output and factor incomes.

Y “ RK ` wL (7.6)

• Differentiation of both sides of equation (9.6) with respect to time gives,

after division by Y and rearrangement of terms, the following
˜ ¸ ˜ ¸
Y9 R9 K9 w9 L9
“ sK ¨ ` ` sL ¨ ` (7.7)
Y R K w L

where sK and sL are again the factor income shares.

• Now the estimated TFP growth rates are given by
˜ ¸ ˜ ¸ ˜ ¸
Y9 K9 L9 R9
ˆ ˙
w9
ĝ “ ´ sK ´ sL “ sK ` sL (7.8)
Y K L R w

• The intuition for the dual estimate on the right-hand side of the equa-
tion (9.8) is that rising factor prices (for factors of given quality) can
be sustained only if output is increasing for given inputs (because of
technological progress). Therefore, appropriately weighted average pf
the growth of the factor prices measures the extent of TFP growth.
• It is important to note that the derivation of equation (9.8) uses only
the condition that Y “ RK ` wL. No assumption is made about the
relation of factor prices to social marginal products or about the form of
the production function. However, ĝ “ g in equation (9.8) if the factors
are paid their social marginal product.
• If Y “ RK ` wL holds then primal and dual estimates of TFP growth
inevitably coincide. Note that Y “ RK ` wL when the underlying
production function exhibits constant returns to scale and factors are
paid their social marginal product. For an application of dual approach
see Hsieh (2002).
CHAPTER 7. GROWTH ACCOUNTING 54

• In some cases - notably when factor prices deviate from social marginal
products – the estimated value ĝ from equation (9.8) would deviate from
the true value, g. However, the error, g ´ ĝ from the dual approach will
be the same as that from the primal approach(Barro, 1999).

7.3 Problems with growth accounting

• A key assumption in the growth accounting exercise is that factor
prices coincide with social marginal products. If this assumption is
violated, the estimated value ĝ calculated from equation (9.5) – or the
corresponding dual estimate from equation (9.8) – deviates from the
true contribution g of technical change to economic growth.

7.3.1 An increasing returns model with spillovers

• A number of authors – including Griliches (1979); Romer (1986b); Lu-
cas (1988)have constructed models of economic growth with increasing
returns arising from spillovers. Romer’s analysis is a generalisation of
Arrow’s (1962a) learning by doing model, in which efficiency of produc-
tion increases with cumulative experience.
• In a simple version of the Romer’s model, output Yi depends not only
on the standard private input Ki and Li , but also on the economy wide
capital stock K. The idea is that investment in Ki by the ith firm
generates knowledge. A part of this knowledge gets into the public
domain due to its non-rival and non-excludable character. Firms can
make use of this knowledge to increase their productivity . This makes
productivity at firm level depends on aggregate capital stock.
• The idea can be represented with a Cobb-Douglas production function
as follows;
Yi “ AKiα K β L1´α
i
where 0 ă α ă 1, and β ě 0. For given K, this production function
exhibits constant returns to scale in the private inputs Ki and Li . If
β ą 0, then spillover effect is present.
• In Griliches (1979) version of the production function in the above
equation, Ki represents firm i’s specific knowledge capital, whereas K
(modelled as sum of the Ki ) is the aggregate level of knowledge in an
industry.
CHAPTER 7. GROWTH ACCOUNTING 55

• In the Lucas (1988) version, Ki is the firm’s employment of human

capital, and K is the aggregate (or average) level of human capital in
an industry or country. In this case spillovers involve benefits from
interacting with smart people.
• Returning to the Romer interpretation of the above production function,
each firm behaves competitively taking as given the economy wide factor
prices R and w and the aggregate capital stock K. So private marginal
products are equated to the factor prices.
• This yields the following

R “ αYi {Ki and w “ p1 ´ αq ¨ Yi {Li

• The factor income share are therefore given, as usual, by

sK “ α and sL “ 1 ´ α

• Since R and w are given and same for all firms, in equilibrium each
firm adopts same capital-labour ratio ki , but the scale of each firm is
indeterminate. The above production function can be written as follows
•
Yi “ Akiα k β Li Lβ
where k ” K{L. The equilibrium condition ki “ k, then implies

Yi “ Ak α`β Li Lβ

which can be aggregated across firms to get

Y “ Ak α`β L1`β

• Finally, the condition k ” K{L results in the following economy-wide

production function
Y “ AK α`β L1´α

• This expression relates aggregate output Y to the aggregate inputs K

and L. If β ą 0, then increasing returns to scale apply economy-wide.
• The above equation shows that the correct way to do the growth ac-
counting with aggregate data is to compute
˜ ¸ ˜ ¸
A9 Y9 K9 L9
ĝ “ “ ´ pα ` βq ¨ ´ p1 ´ αq ¨
A Y K L
CHAPTER 7. GROWTH ACCOUNTING 56

9
• Hence, sL “ 1´α is the correct weight for L{L, but the coefficient sK “ α
9
understates by β ě 0 the contribution of K{K.

• This understatement occurs becuse - with the assumed investment-

based spillovers of knowledge - the social marginal product of capital
pα ` βq ¨ Y {K exceeds the private marginal product αY {K (This private
marginal product does equal the factor price R).

• Also note that the weights on the factor inputs growth in the above
equation add to 1 ` β, which exceeds one by β ą 0 because of the
underlying increasing returns to scale. The increasing returns arise
because ideas about hwo to produce efficiently are fundamentally non-
rival (and spillover freely and instantaneously across firms).

• The interpretation of K – the factor that receives a weight above its

income share in the growth accounting equation above – depends on
the underlying model.

• Griliches (1979) identifies K with knowledge creating activities such

as R&D. Romer (1986b) stresses the physical capital itself. Lucas
(1988) emphasises human capital in the form education. It is, of course,
also possible to have spillover effects that are negative, such as traffic
congestion and environmental damage.

• Empirical implementation of the above equation is difficult because

the proper weights on the factor growth rates cannot be inferred from
income shares; specifically, no direct estimates are available for the
coefficient β. If one instead computes the standard Solow residual
within this model, the one gets

A9 K9 Y9 K9 L9
g̃ “ `β¨ “ ´α¨ ´ p1 ´ αq ¨
A K Y K L

• Thus the standard calculation includes the growth effect from spillovers
9
and increasing returns – β ¨ pK{Kq – along with the rate of exogenous
9
technological progress A{A in the Solow residual. In this case Solow
residual gets underestimated.
Chapter 8

Econometrics of Income
Convergence

• The neoclassical growth model, originating with Solow (1956), has

profoundly affected the way in which economists conceptualising long-
run interrelationships between macroeconomies.1

• Concept of convergence: Given the same preferences and technology

across economies, the differences between countries in per capita output
are transitory.

• However, starting with Romer (1986b) and Lucas (1988), a body of

theoretical literature challenged the strong cross-country implications
of the neoclassical model.

• The proponents of the ‘new growth theories’ have stressed the failure of
the per capita output to equalise across first and third world economies
as well as the failure of the growth rate in developing economies to
exceed those in advanced industrialised economies to as evidence to
press the point that there is little evidence of poor economies catching
up with richer ones.

• In terms of theory, this part of the literature argued that the funda-
mental factor in growth is the presence of non-convexities in production
which can create an non-diminishing relationship between an economies
initial conditions and its output growth rates over arbitrary long hori-
zons2 .
1
Larger part of this note is taken from Bernard and Durlauf (1996).
2
See Azariadis and Drazen (1990) and Durlauf (1993). These authors specifically show
how production complementarities can interact with market incompleteness to generate

57
CHAPTER 8. ECONOMETRICS OF INCOME CONVERGENCE 58

• The striking difference in the empirical implications of the neoclassical

and new growth models with respect to income convergence across
countries have led to a literature formally testing the convergence
hypothesis.

• The empirical test of convergence fall into two categories.

• The first class of test studies the cross-section correlation between initial
per capita output levels and subsequent growth rates for a group of
countries. A negative correlation is taken as evidence of convergence as
it implies that, on average, countries with low per capita initial incomes
are growing faster than those with high initial per capita incomes.
Similarly, other work has explored whether the cross-section variance
of the the per capita output in log for a set of countries has decreased
over time.

• The second set of tests has examined the long-run behaviour of dif-
ferences in per capita output across countries. These test interpret
convergence to mean that these differences are always transitory in
the sense that long-run forecasts of the difference between any pair of
countries converges to zero as the forecast horizon grows. Convergence,
according to this approach has strong implication that output differ-
ences between two economies cannot contain unit roots or time trend
and the weak implication is that output levels in the two economies
must be cointegrated.

• These different testing framework have tended to give contrary results

when applied to output series. Cross section tests generally rejects the
no convergence null in the case of advanced industrialised economies
as well as in large cross-sections after controlling for variables such as
population growth or saving rates (Baumol, 1986; Dowrick and Nguyen,
1989; Barro, 1991; Mankiw, Romer, and Weil, 1992). On the other hand,
time series tests generally accepted the no convergence null for a range
of data, for instance see Bernard and Durlauf (1995).

• In this lecture our objective is to explore the properties of various tests

of the convergence hypothesis.

• Bernard and Durlauf (1996) consider two definitions of the convergence

which directly focus on the transience or permanence of contemporary
multiple equilibria in long-term output paths, which implies that similarly placed economies
need not converge.
CHAPTER 8. ECONOMETRICS OF INCOME CONVERGENCE 59

output differences. Both types of convergence are implications of the

neoclassical growth model. Further, the paper also relates these defini-
tions to cross section and time series tests which have been employed
in the empirical growth literature.
• The paper suggests that time series tests are based on a stricter notion
of convergence than cross-section tests. Further, cross-section tests turn
out to be unable to distinguish between neoclassical growth model and
certain new growth alternatives.
• The paper also shows that different testing approaches also make dif-
ferent assumptions about the statistical properties of the data. The
cross-section tests assume that data under analysis are generated by
economies far from a steady state, time series tests assume that the
data possess well-defined population moments in either levels or first
differences. Inference from the time series approach may be invalid
when based on data which are far from a limiting distribution.
• The paper illustrates how the cross-section and time series approaches
to convergence make different assumptions both about what one means
by convergence and about he properties of economies under study and
therefore how tests within the two frameworks can lead to very different
conclusions concerning cross-country output relationship.

8.1 Neoclassical Growth model**

• Following Solow (1956), a typical neoclassical stochastic growth model
can be written as follows. Output Yt obeys a production function of the
following form
Yt “ At F pKt´1 , Ht´1 Lt´1 , ξt q (8.1)
where At denotes the level of productivity, Kt and Ht denotes physical
and human capital respectively, Lt denotes labour and ξt is a productiv-
ity shock. At can embody both deterministic and stochastic technical
change.
• Physical and human capital obey the laws of motion
Kt “ p1 ´ δK qKt´1 ` SK Yt (8.2)
Ht “ p1 ´ δH qHt´1 ` SH Yt (8.3)
where SK and SH denotes saving rate and δK and δH denotes deprecia-
tion rates.
CHAPTER 8. ECONOMETRICS OF INCOME CONVERGENCE 60

• Labour grows at a constant rate n

Lt “ p1 ` nqt L0 (8.4)

• Finally, some restrictions are placed on the function F p¨, ¨, ¨q. It is

assumed to exhibit nonincreasing returns to scale. Further, the function
obeys the Inada-type conditions.
BF p0, H, L, ξq BF pK, 0, L, ξq
“ “8 (8.5)
BK BH
BF p8, H, L, ξq BF pK, 8, L, ξq
“ “0 (8.6)
BK BH

• Under these conditions, the long-run behaviour of the economy will

be independent of initial conditions in the following sense. Consider
A´1 ´1
t Lt Yt , the level of output normalised by the levels of productivity
and labour. The limiting behaviour of this stochastic process is indepen-
dent of the initial conditions in the economy.

• Proposition 1. Convergence in the neoclassical growth model. For

any economy obeying the neoclassical growth model that was defined by
equations (9.9)– (9.14), limtñ8 ProbpA´1 ´1
t Lt Yt |K0 , H0 , L0 q is independent
of K0 , H0 and L0 .

• Defining Convergence:

• Two definitions of convergence which capture some of the implications

of the neoclassical growth model for the transience of contemporaneous
output differences.

• These definitions characterise convergence between a pair of economies

i and j, convergence between members of a set of I economies may be
defined analogously by requiring that every pair within the set exhibits
convergence. Throughout Ft denotes all information available at time t.

• The first definition considers the behaviour of the output differences be-
tween two economies over a fixed time interval and equates convergence
with the tendency of the difference to narrow.
Definition 8.1.1. Convergence as catching up. Countries i and j con-
verge between dates t and t ` T if the (log) per capita output disparity
at t is expected to decrease in value. If yi,t ą yj,t ,

Epyi,t`T ´ yj,t`T |Fq ă yi,t ´ yj,t (8.7)

CHAPTER 8. ECONOMETRICS OF INCOME CONVERGENCE 61

• The second definition ask whether the long run forecasts of output
differences tent to zero as the forecasting horizon increases. This defini-
tion is violated if history matters, i.e., the effect of a shock on output
differences persist in the indefinite future.

Definition 8.1.2. Convergence as equality of long-term forecasts at a

fixed time. Countries i and j converge if the long-term forecasts of (log)
per capita output for both countries are equal at a fixed time t,

lim Epyi,t`k ´ yj,t`k |Fq “ 0 (8.8)

kñ8

• The relationship between these two definitions is quite straightforward.

In fact, we can order these definitions in terms of the range of restric-
tions placed on the behaviour of output differences.

• Proposition 2. Relationship between definitions. Definition 2 ñ Defini-

tion 1 (for some T).

• From the perspective of empirical work, both definitions are useful as

they each represent implications of the neoclassical growth model.

• Proposition 3. Relationship between the neoclassical growth model

and convergence definitions. Any pair of economies which obey equations
(9.9) to (9.14) and possess identical savings rates, population growth
rates, production functions and probability distributions of shocks will
exhibit convergence in the sense of both Definition 1 and Definition 2.

8.1.1 Convergence Tests

• Cross section tests

• First set of convergence test examines how an economy’s average growth

comoves with initial income. Defining the average growth rate giT “
T ´1 pyi,T ´ yi,0 q for each of I economies, these authors have examined

gi,T “ α ` βyi,0 ` ϵi,T (8.9)

where T is the fixed horizon and Epϵi,T |F0 q “ 0.3

3
In some formulations of cross-section test this regression equation is modified to include
a set of control variables Xi .
CHAPTER 8. ECONOMETRICS OF INCOME CONVERGENCE 62

• In this equation a negative value of β is taken as evidence for the

convergence process, treating β ě 0 as the no convergence hypothesis.

• This requirement may be rewritten as a constraint on the mean

ř output
differences between two time series. Note that gi,T “ T ´1 Tt“1 ∆yi.t
where ∆yi,t “ yi,t ´ yi,t´1 , now (9.17) implies that
T
ÿ T
ÿ
´1 ´1
T ∆yi,t ´ T ∆yj,t “ βpyi,0 ´ yj,0 q ` ϵi,T ´ ϵj,T (8.10)
t“1 t“1

• If yi,0 ´ yj,0 ą 0 then the requirement that β is negative implies that the
expected value of T ´1 Tt“1 ∆yi,t ´ T ´1 Tt“1 ∆yj,t , is negative.
ř ř

• This suggests that cross-section β tests examine whether the average

change in the per capita output of an initially poorer country exceeds
that of an initially richer country.
´1
řI ´1
řI
• Equivalently, letting yĎ i,0 “ I i“1 yi,0 and g
Ěi,T “ I i“1 gi,T , now the
OLS estimator β̂ can be written as

I
ÿ
β̂ “ ϕi ψi (8.11)
i“1

where
2
pyi,0 ´ yĎ
i,0 q
ϕi “ řI (8.12)
Ě0q2
pyi,0 ´ yi,
i“1
pgiT ´ gĚi,T q
ψi “ (8.13)
pyi,0 ´ yĎ
i,0 q

• The above equations shows that β̂ equals a weighted average of the

ratio of differences of growth rates from the sample means to differences
of initial incomes from the sample mean.

• Cross section test therefore requires that a weighted average of coun-

tries with above average initial income grow at a slower rate than the
mean growth for the cross section.

• Implies that the first moments of the stochastic process governing

growth rates differ for initially rich and poor economies.
CHAPTER 8. ECONOMETRICS OF INCOME CONVERGENCE 63

• The derivation shows how the cross-section tests may be interpreted

with respect to Definition 1. Suppose that β̂ ă 0 Since β̂ is a weighted
average of ψi ’s a negative β̂ means that output differences between some
pairs of countries have declined over the sample. Hence, information
set consisting exclusively of a constant, some pairs of countries are
converging in the sense of Definition 1.

• However, this cross-section tests cannot identify groupings of countries

which are converging. This class of test is ill-designed to analyse data
where some countries are converging and others are not.

• Further, this test does not provide evidence on whether countries are
converging in the sense of Definition 2. To see this, consider a cross-
section law of motion for growth,where economies converge to one of n
long run steady states, as given below.

gi,T “ α ` βpyi,0 ´ γi q ` ϵi,T (8.14)

where β ă 0 and γi “ µn if the economy is converging to equilibrium n.

• In the above law of motion, if various economies are distributed across

N long-run steady states, then convergence as characterised by Def-
inition 2 does not hold, although countries associated with the same
equilibrium are converging.

• Suppose equation (9.17) is estimated in order to test for convergence.

From the perspective of equation (9.22), equation (9.17) is a misspecified
regression where regressor yi,0 ´ γi has been replaced by yi,0 . In other
words, γi is an omitted regressor.

• The biased OLS estimate β̂ from equation (9.17) may still be negative.
To see this, assume that I is large so that we can replace sample mo-
ments with population moments. Note that β̂ estimated from equation
(9.17) will follow
ˆ ˙
covpγi , yi,0 q
β̂ “ β 1 ´ (8.15)
varpyi,0 q

• To identify the direction of this limit we can rewrite this limit as

ˆ ˙
covpγi , yi,0 q covpyi,0 , γi ´ yi,0 q
β 1´ “ ´β (8.16)
varpyi,0 q varpyi,0 q
CHAPTER 8. ECONOMETRICS OF INCOME CONVERGENCE 64

• The sign of covpyi,0 , γi ´ yi,0 q{varpyi,0 q is ambiguous and depends upon

the distribution of initial incomes around the different equilibria. The
distribution of initial incomes for a cross-section of economies is not a
part of the set of restrictions imposed on data by any class of growth
models.

• When this covariance is negative, the estimated coefficient β̂ will also

be negative, which means that the econometrician will erroneously
conclude that all countries in the cross-section are converging when
countries are converging to different steady states. This leads us to the
following proposition.

• Proposition 4. Convergence definitions and cross-section tests. A negative

coefficient β̂ in the cross-section growth regression (9.17) is compatible
with a class of structural models which violate Definition 2 of conver-
gence.

• Proposition 4 calls into question the utility of cross-section tests in

adjudicating theoretical disputes in the growth literature. From the per-
spective of new growth models, the relevant notion of convergence would
appear to be Definition 2, since this definition equates convergence with
the independence of limiting behaviour from initial conditions, whose
failure is a hallmark of that class of models.

8.1.2 Time series tests

• The second approach employed by many studies such as Bernard and
Durlauf (1995) has relied on the time series properties of output series.
Time series approaches to convergence check for the compatability of
yi,t ´ yj,t with time-invariant Wold representation of the form
8
ÿ
yi,t ´ yj,t “ κi,j ` πi,j,r ϵi,j,t´r (8.17)
r“0

such that the κi,j “ 0 and πi,j,r is square summable.4

• As stated in proposition 5, the presence of either a deterministic or a

unit root component in yi,t ´ yj,t is a violation of Definition 2, as either
component implies that forecasts of output differences do not converge
to zero in expected value as the forecast horizon becomes arbitrary long.
4
ř8
Square summability implies that r“0 |πr |2 ă 8.
CHAPTER 8. ECONOMETRICS OF INCOME CONVERGENCE 65

• Proposition 5: Relationship between convergence definitions and time

series tests. If yi,t ´ yj,t contains either a non-zero mean or a unit root,
then Definition 2 of convergence is violated.

• Note that if yi,t ´ yj,t is a zero-mean stationary process, then equation

(9.25) immediately implies that T ´1 Tt“1 ∆yi,t ´ T ´1 ∆yj,t is also a zero-
ř
mean stationary process.

• Relationship between tests and transition and versus steady

state behaviour

• Despite their links to related convergence definitions, cross-section

and time series tests places different implications on the data. At one
level, if the two types of tests are applied to the same data set, they
must necessarily be inconsistent. This holds because, as a comparison
of (9.18) and (9.25) shows, cross-section test requires that the first
difference of cross-country output differences possesses a nonzero mean,
whereas time series tests require the same series to possess a zero
mean.

• Proposition 6: Incompatibility of cross-section and time series tests.

UnderřTcross-section tests,
řT convergence requires that the expected value of
´1 ´1
T t“1 ∆yi,t ´ T t“1 ∆yj,t is negative if yi,0 ´ yj.0 is positive, whereas
under ř time series testsřconvergence requires that the expected value
of T ´1 Tt“1 ∆yi,t ´ T ´1 Tt“1 ∆yj,t is zero regardless of yi,0 ´ yj,0 . Thus,
there exists a set of values for the sample moments ∆yi,t and ∆yj,t which
implies convergence under one test yet implies no convergence under the
other.

• In order to understand this dissimilarity in the data restrictions of the

two classes of tests, it is necessary to observe that the two approaches
to convergence have fundamentally different views concerning the prop-
erties of the data under question.

• In cross-section tests, one assumes that the data are in transition to-
wards a limiting distribution and convergence is interpreted as meaning
that initial output differences between economies dissipate over a fixed
time period.

• Cross-section tests turn out to place much weaker restrictions on the

behaviour of growth across countries than the associated time series
tests. As a result, cross-section test can reject a no convergence null
CHAPTER 8. ECONOMETRICS OF INCOME CONVERGENCE 66

hypothesis for data generated by economies with different long-run

steady states, such as those considered by Romer (1986b).

• In time series tests, one assumes that the data are generated by
economies near their limiting distributions and convergence is inter-
preted to mean that initial conditions have no (statistically significant)
effect on the expected value of output differences. Time series tests
do not spuriously reject the no convergence null for data generated by
multiple long-run equilibria. However, the time series approach re-
quires that economies under analysis are near their long-run equilibria
since the test assume that the sample moments of the data accurately
approximate the limiting moments for the data under analysis. The
test therefore may invalid if the data are largely driven by transition
dynamics.

• This implies that time series test would have poor power properties
when applied to data from economies in transitions.

• Therefore, a given approach is appropriate depending upon whether one

regards the data as better characterised by transition or steady states
dynamics.
Chapter 9

Growth Accounting and Income

Convergence

9.1 Growth Accounting**

• Growth accounting provides a breakdown of observed economic growth
into components associated with growth in factor inputs and a residual
that reflects technological progress and other elements (Barro, 1999).

• Generally, growth accounting exercise is viewed as a preliminary step

for the analysis of fundamental determinants of economic growth. The
growth accounting exercise can be particularly useful if the fundamental
determinants that matter for factor growth rates are substantially
independent from those that matter for technological change.

• The topics we discuss include primal and dual approaches to growth

accounting, growth accounting when increasing returns are present due
to spillovers, and the interpretation of Solow residual as a measure of
technological change.1

9.1.1 Standard Primal Growth Accounting

• Consider an neoclassical production function

Y “ F pA, K, Lq (9.1)
1
Barro (1999) also discusses growth accounting in the context of taxes and multiple inputs
are used.

67
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE68

where A is the level of technology, K is the capital stock, and L is the

quantity of labour. Capital and labour can be disaggregated among
types or qualities as in Jorgenson and Griliches (1967).

• The growth rate in output can be partitioned into components associ-

ated with factor accumulation and technical progress. Differentiating
equation (9.1) with respect to time yields, after division by Y and rear-
rangement of terms,

Y9
ˆ ˙ 9 ˆ ˙ 9
FK K K FL L L
“g` ¨ ` ¨ (9.2)
Y Y K Y L

where FK and FL are the factor (social) marginal products and g - the
growth due to technological change – is given by
ˆ ˙ 9
FA A A
g” ¨ (9.3)
Y A

If the technology factor appears in Hicks-neutral way, so that F pA, K, Lq “

9
A ¨ F̃ pK, Lq, then g “ A{A.

• The rate of technological progress g can be calculated from equation

(9.2) as a residual,

Y9
ˆ ˙ 9 ˆ ˙ 9
FK K K FL L L
g“ ´ ¨ ´ ¨ (9.4)
Y Y K Y L

• However, equation (9.4) is impractical because it requires knowledge of

the social marginal products, FK and FL . In practice, the computations
typically assume that the social marginal products can be measured by
observed factor prices.

• If factors are paid their social marginal products, so that FK “ R(the

rental price of capital) and FL “ w (the wage rate), then standard primal
estimate of the rate of technological progress follows from equation (9.4)
as
Y9 K9 L9
ĝ “ ´ sK ¨ ´ sL ¨ (9.5)
Y K L
where sK ” RK{Y and sL ” wL{Y are the respective shares of each
factor payment in total product.

• The value ĝ is often described as an estimate of total factor productivity

(TFP) growth or Solow residual.
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE69

• The condition sK ` sL “ 1 or Y “ RK ` wL must hold if all the income

associated with the gross domestic product Y is attributed to one of the
factors, restricted here to capital and labour.

• In an international context, some net factor income may accrue to

foreign-owned factors, and RK `wL would include this net factor income
also.

• The equation of output Y to total factor income is consistent with

the equality between the factor prices and marginal products if the
production function, F p¨q, is of constant returns to scale in K and L, so
that Y “ FK K ` FL L holds.

• Using sk ` sL “ 1, equation (9.5) can be rewritten in intensive form as

y9 k9
ĝ “ ´ sK
y k

where y ” Y {L and k ” K{L are quantities per unit of labour.

• Jorgenson and Griliches (1967); Jorgenson, Gollop, and Fraumeni (1987)

stressed the importance of disaggregating inputs by quality classes.
Failure to include input quality (of both labour and capital) in this way
would overestimate the estimated total factor productivity or Solow
residual.

• An alternative approach to measure productivity growth is to regress

growth rate of output Y9 {Y on the growth rates of inputs, K{K9 and
9L{L, in the form of equation (9.2). The intercept then measures g,
` ˘ ` ˘
and the coefficients on the factor growths measure FKY K and FYL L ,
respectively. The main advantage of this approach is that it dispenses
with the assumption that the factor social marginal products coincide
with the observable factor prices – that is, FK “ R and FL “ w.

• However, the disadvantages of the regression approach are the follow-

ing,
9
– The variables K{K 9
and L{L cannot usually be regarded as exoge-
nous with respect to variations in g. As we have already seen
factors get accumulated in response to technological progress. This
makes OLS estimates inconsistent.
9
– If K{K 9
and L{L are measured with error, then estimated `coeffi- ˘
cients of these variables would be inconsistent estimates of FKY K
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE70
` ˘
and FYL L respectively. This problem can especially serious for the
growth rates of capital input, as correct measurement of actually
used capital stock is sometimes not possible. And this usually
results in downward estimates for the coefficient of the growth rate
of capital.
– The regression framework has to be extended to allow for changes
in factor shares and TFP growth rates over time.

9.1.2 Dual Approach to growth accounting

• In the dual approach, the Solow residual is computed from growth rates
of factor prices rather than factor quantities. It can be derived directly
from the equality between output and factor incomes.

Y “ RK ` wL (9.6)

• Differentiation of both sides of equation (9.6) with respect to time gives,

after division by Y and rearrangement of terms, the following
˜ ¸ ˜ ¸
Y9 R9 K9 w9 L9
“ sK ¨ ` ` sL ¨ ` (9.7)
Y R K w L

where sK and sL are again the factor income shares.

• Now the estimated TFP growth rates are given by

˜ ¸ ˜ ¸ ˜ ¸
Y9 K9 L9 R9
ˆ ˙
w9
ĝ “ ´ sK ´ sL “ sK ` sL (9.8)
Y K L R w

• The intuition for the dual estimate on the right-hand side of the equa-
tion (9.8) is that rising factor prices (for factors of given quality) can
be sustained only if output is increasing for given inputs (because of
technological progress). Therefore, appropriately weighted average pf
the growth of the factor prices measures the extent of TFP growth.

• It is important to note that the derivation of equation (9.8) uses only

the condition that Y “ RK ` wL. No assumption is made about the
relation of factor prices to social marginal products or about the form of
the production function. However, ĝ “ g in equation (9.8) if the factors
are paid their social marginal product.
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE71

• If Y “ RK ` wL holds then primal and dual estimates of TFP growth

inevitably coincide. Note that Y “ RK ` wL when the underlying
production function exhibits constant returns to scale and factors are
paid their social marginal product. For an application of dual approach
see Hsieh (2002).
• In some cases - notably when factor prices deviate from social marginal
products – the estimated value ĝ from equation (9.8) would deviate from
the true value, g. However, the error, g ´ ĝ from the dual approach will
be the same as that from the primal approach(Barro, 1999).

9.1.3 Problems with growth accounting

• A key assumption in the growth accounting excercise is that factor
prices coincide with social marginal products. If this assumption is
violated, the estimated value ĝ calculated from equation (9.5) – or the
corresponding dual estimate from equation (9.8) – deviates from the
true contribution g of technical change to economic growth.

An increasing returns model with spillovers

• A number of authors – including Griliches (1979); Romer (1986b); Lu-

cas (1988)have constructed models of economic growth with increasing
returns arising from spillovers. Romer’s analysis is a generalisation of
Arrow’s (1962a) learning by doing model, in which efficiency of produc-
tion increases with cumulative experience.
• In a simple version of the Romer’s model, output Yi depends not only
on the standard private input Ki and Li , but also on the economy wide
capital stock K. The idea is that investment in Ki by the ith firm
generates knowledge. A part of this knowledge gets into the public
domain due to its non-rival and non-excludable character. Firms can
make use of this knowledge to increase their productivity . This makes
productivity at firm level depends on aggregate capital stock.
• The idea can be represented with a Cobb-Douglas production function
as follows;
Yi “ AKiα K β L1´α
i
where 0 ă α ă 1, and β ě 0. For given K, this production function
exhibits constant returns to scale in the private inputs Ki and Li . If
β ą 0, then spillover effect is present.
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE72

• In Griliches (1979) version of the production function in the above

equation, Ki represents firm i’s specific knowledge capital, whereas K
(modelled as sum of the Ki ) is the aggregate level of knowledge in an
industry.

• In the Lucas (1988) version, Ki is the firm’s employment of human

capital, and K is the aggregate (or average) level of human capital in
an industry or country. In this case spillovers involve benefits from
interacting with smart people.

• Returning to the Romer interpretation of the above production function,

each firm behaves competitively taking as given the economy wide factor
prices R and w and the aggregate capital stock K. So private marginal
products are equated to the factor prices.

• This yields the following

R “ αYi {Ki and w “ p1 ´ αq ¨ Yi {Li

• The factor income share are therefore given, as usual, by

sK “ α and sL “ 1 ´ α

• Since R and w are given and same for all firms, in equilibrium each
firm adopts same capital-labour ratio ki , but the scale of each firm is
indeterminate. The above production function can be written as follows

•
Yi “ Akiα k β Li Lβ
where k ” K{L. The equilibrium condition ki “ k, then implies

Yi “ Ak α`β Li Lβ

which can be aggregated across firms to get

Y “ Ak α`β L1`β

• Finally, the condition k ” K{L results in the following economy-wide

production function
Y “ AK α`β L1´α

• This expression relates aggregate output Y to the aggregate inputs K

and L. If β ą 0, then increasing returns to scale apply economy-wide.
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE73

• The above equation shows that the correct way to do the growth ac-
counting with aggregate data is to compute
˜ ¸ ˜ ¸
A9 Y9 K9 L9
ĝ “ “ ´ pα ` βq ¨ ´ p1 ´ αq ¨
A Y K L

9
• Hence, sL “ 1´α is the correct weight for L{L, but the coefficient sK “ α
9
understates by β ě 0 the contribution of K{K.
• This understatement occurs becuse - with the assumed investment-
based spillovers of knowledge - the social marginal product of capital
pα ` βq ¨ Y {K exceeds the private marginal product αY {K (This private
marginal product does equal the factor price R).
• Also note that the weights on the factor inputs growth in the above
equation add to 1 ` β, which exceeds one by β ą 0 because of the
underlying increasing returns to scale. The increasing returns arise
because ideas about hwo to produce efficiently are fundamentally non-
rival (and spillover freely and instantaneously across firms).
• The interpretation of K – the factor that receives a weight above its
income share in the growth accounting equation above – depends on
the underlying model.
• Griliches (1979) identifies K with knowledge creating activities such
as R&D. Romer (1986b) stresses the physical capital itself. Lucas
(1988) emphasises human capital in the form education. It is, of course,
also possible to have spillover effects that are negative, such as traffic
congestion and environmental damage.
• Empirical implementation of the above equation is difficult because
the proper weights on the factor growth rates cannot be inferred from
income shares; specifically, no direct estimates are available for the
coefficient β. If one instead computes the standard Solow residual
within this model, the one gets

A9 K9 Y9 K9 L9
g̃ “ `β¨ “ ´α¨ ´ p1 ´ αq ¨
A K Y K L

• Thus the standard calculation includes the growth effect from spillovers
9
and increasing returns – β ¨ pK{Kq – along with the rate of exogenous
9
technological progress A{A in the Solow residual. In this case Solow
residual gets underestimated.
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE74

9.2 Income Convergence**

• The neoclassical growth model, originating with Solow (1956), has
profoundly affected the way in which economists conceptualising long-
run interrelationships between macroeconomies.

• Concept of convergence: Given the same preferences and technology

across economies, the differences between countries in per capita output
are transitory.

• However, starting with Romer (1986b) and Lucas (1988), a body of

theoretical literature challenged the strong cross-country implications
of the neoclassical model.

• The proponents of the ‘new growth theories’ have stressed the failure of
the per capita output to equalise across first and third world economies
as well as the failure of the growth rate in developing economies to
exceed those in advanced industrialised economies to as evidence to
press the point that there is little evidence of poor economies catching
up with richer ones.

• In terms of theory, this part of the literature argued that the funda-
mental factor in growth is the presence of nonconvexities in production
which can create an non-diminishing relationship between an economies
initial conditions and its output growth rates over arbitrary long hori-
zons2 .

• The striking difference in the empirical implications of the neoclassical

and new growth models with respect to income convergence across
countries have led to a literature formally testing the convergence
hypothesis.

• The empirical test of convergence fall into two categories.

• The first class of test studies the cross-section correlation between initial
per capita output levels and subsequent growth rates for a group of
countries. A negative correlation is taken as evidence of convergence as
it implies that, on average, countries with low per capita initial incomes
are growing faster than those with high initial per capita incomes.
2
See Azariadis and Drazen (1990) and Durlauf (1993). These authors specifically show
how production complementarities can interact with market incompleteness to generate
multiple equilibria in long-term output paths, which implies that similarly placed economies
need not converge.
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE75

Similarly, other work has explored whether the cross-section variance

of the the per capita output in log for a set of countries has decreased
over time.

• The second set of tests has examined the long-run behaviour of dif-
ferences in per capita output across countries. These test interpret
convergence to mean that these differences are always transitory in
the sense that long-run forecasts of the difference between any pair of
countries converges to zero as the forecast horizon grows. Convergence,
according to this approach has strong implication that output differ-
ences between two economies cannot contain unit roots or time trend
and the weak implication is that output levels in the two economies
must be cointegrated.

• These different testing framework have tended to give contrary results

when applied to output series. Cross section tests generally rejects the
no convergence null in the case of advanced industrialised economies
as well as in large cross-sections after controlling for variables such as
population growth or saving rates (Baumol, 1986; Dowrick and Nguyen,
1989; Barro, 1991; Mankiw, Romer, and Weil, 1992). On the other hand,
time series tests generally accepted the no convergence null for a range
of data, for instance see Bernard and Durlauf (1995).

• In this lecture our objective is to explore the properties of various tests

of the convergence hypothesis.

• Bernard and Durlauf (1996) consider two definitions of the convergence

which directly focus on the transience or permanence of contemporary
output differences. Both types of convergence are implications of the
neoclassical growth model. Further, the paper also relates these defini-
tions to cross section and time series tests which have been employed
in the empirical growth literature.

• The paper suggests that time series tests are based on a stricter notion
of convergence than cross-section tests. Further, cross-section tests turn
out to be unable to distinguish between neoclassical growth model and
certain new growth alternatives.

• The paper also shows that different testing approaches also make dif-
ferent assumptions about the statistical properties of the data. The
cross-section tests assume that data under analysis are generated by
economies far from a steady state, time series tests assume that the
data possess well-defined population moments in either levels or first
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE76

differences. Inference from the time series approach may be invalid

when based on data which are far from a limiting distribution.

• The paper illustrates how the cross-section and time series approaches
to convergence make different assumptions both about what one means
by convergence and about he properties of economies under study and
therefore how tests within the two frameworks can lead to very different
conclusions concerning cross-country output relationship.

Neoclassical Growth model**

• Following Solow (1956), a typical neoclassical stochastic growth model

can be written as follows. Output Yt obeys a production function of the
following form
Yt “ At F pKt´1 , Ht´1 Lt´1 , ξt q (9.9)
where At denotes the level of productivity, Kt and Ht denotes physical
and human capital respectively, Lt denotes labour and ξt is a productiv-
ity shock. At can embody both deterministic and stochastic technical
change.

• Physical and human capital obey the laws of motion

Kt “ p1 ´ δK qKt´1 ` SK Yt (9.10)
Ht “ p1 ´ δH qHt´1 ` SH Yt (9.11)

where SK and SH denotes saving rate and δK and δH denotes deprecia-

tion rates.

• Labour grows at a constant rate n

Lt “ p1 ` nqt L0 (9.12)

• Finally, some restrictions are placed on the function F p¨, ¨, ¨q. It is

assumed to exhibit nonincreasing returns to scale. Further, the function
obeys the Inada-type conditions.

BF p0, H, L, ξq BF pK, 0, L, ξq
“ “8 (9.13)
BK BH
BF p8, H, L, ξq BF pK, 8, L, ξq
“ “0 (9.14)
BK BH
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE77

• Under these conditions, the long-run behaviour of the economy will

be independent of initial conditions in the following sense. Consider
A´1 ´1
t Lt Yt , the level of output normalised by the levels of productivity
and labour. The limiting behaviour of this stochastic process is indepen-
dent of the initial conditions in the economy.

• Proposition 1. Convergence in the neoclassical growth model. For

any economy obeying the neoclassical growth model that was defined by
equations (9.9)– (9.14), limtñ8 ProbpA´1 ´1
t Lt Yt |K0 , H0 , L0 q is independent
of K0 , H0 and L0 .

• Defining Convergence:

• Two definitions of convergence which capture some of the implications

of the neoclassical growth model for the transience of contemporaneous
output differences.

• These definitions characterise convergence between a pair of economies

i and j, convergence between members of a set of I economies may be
defined analogously by requiring that every pair within the set exhibits
convergence. Throughout Ft denotes all information available at time t.

• The first definition considers the behaviour of the output differences be-
tween two economies over a fixed time interval and equates convergence
with the tendency of the difference to narrow.

Definition 9.2.1. Convergence as catching up. Countries i and j con-

verge between dates t and t ` T if the (log) per capita output disparity
at t is expected to decrease in value. If yi,t ą yj,t ,

Epyi,t`T ´ yj,t`T |Fq ă yi,t ´ yj,t (9.15)

• The second definition ask whether the long run forecasts of output
differences tent to zero as the forecasting horizon increases. This defini-
tion is violated if history matters, i.e., the effect of a shock on output
differences persist in the indefinite future.

Definition 9.2.2. Convergence as equality of long-term forecasts at a

fixed time. Countries i and j converge if the long-term forecasts of (log)
per capita output for both countries are equal at a fixed time t,

lim Epyi,t`k ´ yj,t`k |Fq “ 0 (9.16)

kñ8
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE78

• The relationship between these two definitions is quite straightforward.

In fact, we can order these definitions in terms of the range of restric-
tions placed on the behaviour of output differences.

• Proposition 2. Relationship between definitions. Definition 2 ñ Defini-

tion 1 (for some T).

• From the perspective of empirical work, both definitions are useful as

they each represent implications of the neoclassical growth model.

• Proposition 3. Relationship between the neoclassical growth model

and convergence definitions. Any pair of economies which obey equations
(9.9) to (9.14) and possess identical savings rates, population growth
rates, production functions and probability distributions of shocks will
exhibit convergence in the sense of both Definition 1 and Definition 2.

Convergence Tests

• Cross section tests

• First set of convergence test examines how an economy’s average growth

comoves with initial income. Defining the average growth rate giT “
T ´1 pyi,T ´ yi,0 q for each of I economies, these authors have examined

gi,T “ α ` βyi,0 ` ϵi,T (9.17)

where T is the fixed horizon and Epϵi,T |F0 q “ 0.3

• In this equation a negative value of β is taken as evidence for the

convergence process, treating β ě 0 as the no convergence hypothesis.

• This requirement may be rewritten as a constraint on the meanřToutput

´1
differences between two time series. Note that gi,T “ T t“1 ∆yi.t
where ∆yi,t “ yi,t ´ yi,t´1 , now (9.17) implies that
T
ÿ T
ÿ
´1 ´1
T ∆yi,t ´ T ∆yj,t “ βpyi,0 ´ yj,0 q ` ϵi,T ´ ϵj,T (9.18)
t“1 t“1

• If yi,0 ´ yj,0 ą 0 then the requirement that β is negative implies that the
expected value of T ´1 Tt“1 ∆yi,t ´ T ´1 Tt“1 ∆yj,t , is negative.
ř ř

3
In some formulations of cross-section test this regression equation is modified to include
a set of control variables Xi .
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE79

• This suggests that cross-section β tests examine whether the average

change in the per capita output of an initially poorer country exceeds
that of an initially richer country.
´1
řI ´1
řI
• Equivalently, letting yĎ i,0 “ I i“1 yi,0 and g
Ěi,T “ I i“1 gi,T , now the
OLS estimator β̂ can be written as

I
ÿ
β̂ “ ϕi ψi (9.19)
i“1

where
2
pyi,0 ´ yĎ
i,0 q
ϕi “ řI (9.20)
Ě0q2
pyi,0 ´ yi,
i“1
pgiT ´ gĚi,T q
ψi “ (9.21)
pyi,0 ´ yĎ
i,0 q

• The above equations shows that β̂ equals a weighted average of the

ratio of differences of growth rates from the sample means to differences
of initial incomes from the sample mean.

• Cross section test therefore requires that a weighted average of coun-

tries with above average initial income grow at a slower rate than the
mean growth for the cross section.

• Implies that the first moments of the stochastic process governing

growth rates differ for initially rich and poor economies.

• The derivation shows how the cross-section tests may be interpreted

with respect to Definition 1. Suppose that β̂ ă 0 Since β̂ is a weighted
average of ψi ’s a negative β̂ means that output differences between some
pairs of countries have declined over the sample. Hence, information
set consisting exclusively of a constant, some pairs of countries are
converging in the sense of Definition 1.

• However, this cross-section tests cannot identify groupings of countries

which are converging. This class of test is ill-designed to analyse data
where some countries are converging and others are not.

• Further, this test does not provide evidence on whether countries are
converging in the sense of Definition 2. To see this, consider a cross-
section law of motion for growth,where economies converge to one of n
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE80

long run steady states, as given below.

gi,T “ α ` βpyi,0 ´ γi q ` ϵi,T (9.22)
where β ă 0 and γi “ µn if the economy is converging to equilibrium n.
• In the above law of motion, if various economies are distributed across
N long-run steady states, then convergence as characterised by Def-
inition 2 does not hold, although countries associated with the same
equilibrium are converging.
• Suppose equation (9.17) is estimated in order to test for convergence.
From the perspective of equation (9.22), equation (9.17) is a misspecified
regression where regressor yi,0 ´ γi has been replaced by yi,0 . In other
words, γi is an omitted regressor.
• The biased OLS estimate β̂ from equation (9.17) may still be negative.
To see this, assume that I is large so that we can replace sample mo-
ments with population moments. Note that β̂ estimated from equation
(9.17) will follow
ˆ ˙
covpγi , yi,0 q
β̂ “ β 1 ´ (9.23)
varpyi,0 q
• To identify the direction of this limit we can rewrite this limit as
ˆ ˙
covpγi , yi,0 q covpyi,0 , γi ´ yi,0 q
β 1´ “ ´β (9.24)
varpyi,0 q varpyi,0 q

• The sign of covpyi,0 , γi ´ yi,0 q{varpyi,0 q is ambiguous and depends upon

the distribution of initial incomes around the different equilibria. The
distribution of initial incomes for a cross-section of economies is not a
part of the set of restrictions imposed on data by any class of growth
models.
• When this covariance is negative, the estimated coefficient β̂ will also
be negative, which means that the econometrician will erroneously
conclude that all countries in the cross-section are converging when
countries are converging to different steady states. This leads us to the
following proposition.
• Proposition 4. Convergence definitions and cross-section tests. A negative
coefficient β̂ in the cross-section growth regression (9.17) is compatible
with a class of structural models which violate Definition 2 of conver-
gence.
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE81

• Proposition 4 calls into question the utility of cross-section tests in

adjudicating theoretical disputes in the growth literature. From the per-
spective of new growth models, the relevant notion of convergence would
appear to be Definition 2, since this definition equates convergence with
the independence of limiting behaviour from initial conditions, whose
failure is a hallmark of that class of models.

Time series tests

• The second approach employed by many studies such as Bernard and

Durlauf (1995) has relied on the time series properties of output series.
Time series approaches to convergence check for the compatability of
yi,t ´ yj,t with time-invariant Wold representation of the form
8
ÿ
yi,t ´ yj,t “ κi,j ` πi,j,r ϵi,j,t´r (9.25)
r“0

such that the κi,j “ 0 and πi,j,r is square summable.4

• As stated in proposition 5, the presence of either a deterministic or a

unit root component in yi,t ´ yj,t is a violation of Definition 2, as either
component implies that forecasts of output differences do not converge
to zero in expected value as the forecast horizon becomes arbitrary long.

• Proposition 5: Relationship between convergence definitions and time

series tests. If yi,t ´ yj,t contains either a non-zero mean or a unit root,
then Definition 2 of convergence is violated.

• Note that if yi,t ´ yj,t is a zero-mean stationary process, then equation

(9.25) immediately implies that T ´1 Tt“1 ∆yi,t ´ T ´1 ∆yj,t is also a zero-
ř
mean stationary process.

• Relationship between tests and transition and versus steady

state behaviour

• Despite their links to related convergence definitions, cross-section

and time series tests places different implications on the data. At one
level, if the two types of tests are applied to the same data set, they
must necessarily be inconsistent. This holds because, as a comparison
of (9.18) and (9.25) shows, cross-section test requires that the first
4
ř8
Square summability implies that r“0 |πr |2 ă 8.
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE82

difference of cross-country output differences possesses a nonzero mean,

whereas time series tests require the same series to possess a zero
mean.

• Proposition 6: Incompatibility of cross-section and time series tests.

Under
ř cross-section tests, convergence requires that the expected value of
T ´1 Tt“1 ∆yi,t ´ T ´1 Tt“1 ∆yj,t is negative if yi,0 ´ yj.0 is positive, whereas
ř
under řtime series testsřconvergence requires that the expected value
of T ´1 Tt“1 ∆yi,t ´ T ´1 Tt“1 ∆yj,t is zero regardless of yi,0 ´ yj,0 . Thus,
there exists a set of values for the sample moments ∆yi,t and ∆yj,t which
implies convergence under one test yet implies no convergence under the
other.

• In order to understand this dissimilarity in the data restrictions of the

two classes of tests, it is necessary to observe that the two approaches
to convergence have fundamentally different views concerning the prop-
erties of the data under question.

• In cross-section tests, one assumes that the data are in transition to-
wards a limiting distribution and convergence is interpreted as meaning
that initial output differences between economies dissipate over a fixed
time period.

• Cross-section tests turn out to place much weaker restrictions on the

behaviour of growth across countries than the associated time series
tests. As a result, cross-section test can reject a no convergence null
hypothesis for data generated by economies with different long-run
steady states, such as those considered by Romer (1986b).

• In time series tests, one assumes that the data are generated by
economies near their limiting distributions and convergence is inter-
preted to mean that initial conditions have no (statistically significant)
effect on the expected value of output differences. Time series tests
do not spuriously reject the no convergence null for data generated by
multiple long-run equilibria. However, the time series approach re-
quires that economies under analysis are near their long-run equilibria
since the test assume that the sample moments of the data accurately
approximate the limiting moments for the data under analysis. The
test therefore may invalid if the data are largely driven by transition
dynamics.

• This implies that time series test would have poor power properties
when applied to data from economies in transitions.
CHAPTER 9. GROWTH ACCOUNTING AND INCOME CONVERGENCE83

• Therefore, a given approach is appropriate depending upon whether one

regards the data as better characterised by transition or steady states
dynamics.
Chapter 10

Growth models with diminishing

returns to capital: Limitations

• “The central problem in the theory of economic development is to under-

stand the process by which a community which was previously saving
and investing 4 or 5 per cent of its national income or less, converts
itself into an economy where voluntary saving is running at about 12
to 15 per cent of national income or more. This is the central problem
because the central fact of development is rapid capital accumulation
(including knowledge and skills with capital)” (Lewis, 1954, p.155).
• This is the perception prevailed in the 1950s. What we can see that sav-
ing rate of countries increased a lot, but not growth rate. For instance
the case of India.
• In a model with exogenous technological progress, such as Solow (1954),
Ramsey (1928), Cass (1965) and Koopmans (1965), the rate of return on
investment and and the rate of growth of per capita output are expected
to be a decreasing functions of the level of the per capita capital stock.
k9 y
“ s ´ pn ` dq
k k
• Over time wage rates and capital-labour ratios across different coun-
tries are expected to converge. Therefore, initial conditions or current
disturbances have no long-run effect on the level of output and consump-
tion. Example: a negative shock on capital stock and the consequent wage,
reduction in the stock of capital. income
conver-
• Theories of long-run growth assume away any variation in output at- gence
tributable to business cycles. Therefore, it is important to filter cycles

84
CHAPTER 10. SOLOW TYPE GROWTH MODELS: LIMITATIONS 85

Table 10.1: Productivity Growth for Leading Countries

Lead Country Interval Growth rate (in %)
Netherlands 1700-1785 -0.07
United Kingdom 1785-1820 0.5
United Kingdom 1820-1890 1.4
United States 1890-1979 2.3
Source:Romer (1986a).

out of data and isolate that component of the GDP movement that
growth theory tries to explain. So we take average growth rate over a
very long period.

Behaviour of Growth Rates

• Now we ask whether there is anything in the data that should cause
economists to choose a model with a diminishing returns, falling rates
of growth, and convergence across countries rather than an alternative
without these features.

• Now consider the growth rate of leader countries, that is countries

having highest level of productivity compared to all other countries.
Growth for a country, which is not a leader, will reflect at least in part
the process of imitation and transmission of existing of knowledge,
whereas the growth rate of the leader gives some indication of the
expansion of the growth at the frontier of the knowledge. See Table
10.1.

• Similar evidence is apparent from data for individual countries over

short horizons. Table reports the growth rate of per capita GDP for the
United States over five sub-periods from 1800-1978.

The Case of less developed countries

• Reynolds (1983) gives an overview of the pattern of development in less

developed countries.

• Given the paucity of precise data for less developed countries, he fo-
cuses on the “turning points” at which a country first begins to exhibit
a persistent upward trend in per capita income. The timing of this
CHAPTER 10. SOLOW TYPE GROWTH MODELS: LIMITATIONS 86

Table 10.2: Per Capita Growth in the United States

Interval Growth Rate (in%)
1800-1840 0.58
1840-1880 1.44
1880-1920 1.78
1920-1960 1.68
1960-1978 2.47
Note: Average Annual compound growth rate.
Source: Romer (1986a), Table 2.

transition and the pace of subsequent growth are strongly influenced

by the variation in the world economy. Although all less developed
countries are affected by the worldwide economy, the effects are not
uniform. The key observation is that those countries having more ex-
tensive prior development appear to benefit more from periods of rapid
worldwide growth and suffer less during any slowdown. That is growth
rate appear to be increasing not only as a function of calender time but
also as a function of the level of development.

• The observation that more developed countries appear to grow relatively

faster extends to a comparison of industrialised versus less developed
countries as well. In the period between 1950 and 1980, Reynolds
(1983) reports that the median growth rate of per capita income for his
sample of 41 less developed countries was 2.3 per cent, “clearly below
the median for the OECD countries for the same period” (p.975).

• The growth rate of India for the last 100 years. Where the growth
rate is very low at low levels of per capita income, and growth rate is
increasing along with income levels.

The issue of convergence

• If growth rates are not negatively correlated with the level of per capita
output or capital, there should be no tendency for the dispersion in the
(logarithm of the) level of per capita income to decrease over time.

• Examining the dispersion in the logarithm of the level of per capita

income, not dispersion in the level itself, is the correct way to test for
the convergence in the growth rates. If the rate of growth were
constant across countries that start from different levels, the dispersion
CHAPTER 10. SOLOW TYPE GROWTH MODELS: LIMITATIONS 87

16
15
1.80 0.29 1.19 0.58 -1.30 3.54 4.92 7.86
log GDP
14
13
12
1900-01

1908-09

1917-18

1930-31

1945-46

1953-54

1978-79

2001-02

2009-10
Year

Figure 10.1: Growth of India’s GDP

in the logarithm of the levels will stay constant, but the dispersion in
the levels will increase((Romer, 1986b, p.1012)

• Study by Baumol(1985) finds no evidence on decreasing dispersion over

time in his full sample of countries. Actually he found a weak tendency
towards divergence. See the table taken from Easterly and Levine
(2001).
0123456719 4 94 
.CP=@!%#f#T%*#/gf#
,#YZWX
h$$-#0
_05## **/YZWX'i)
g-#--# !`
,*-#--# !X
^#-#--# ![
,#-#--# X!Y
%#-#--# X!X
jk34lf##-#& #0**#YZ!
mkn5o4lhp#&,002e#
&52#*!

!"#$%&'()
*##+#,&-#$!.$/
&#$$00*#1*#2
0##0###$-/0-#3--
*#/!4#*#5#
**-*#*#-&3*#0#$
&##1-!
6789:;<=>?@A9BCD;9E89FGHI;J96K9LH;9M@GK>K;@N;O
PQ;9BCD;HG9RDDQSQ=C;>HN96K
T#0U&&#0!.#-**#$(
#VVWX#**#$#YZWVY(##!Z[!
.#*##\##-*!]X
CHAPTER 10. SOLOW TYPE GROWTH MODELS: LIMITATIONS 88

The evidence from Growth Accounting

Concept of Growth Accounting

•
y “ Ak α n1´α
where y is the national output per person, A is the technological progress,
and k is the physical capital stock per person, n is the number of labour
inputs per person (reflecting work patterns, human capital, and the
like) and α is the production function parameter (equal to the share of
capital in the national output under perfect competition)(Easterly and
Levine, 2001).

• Output growth is then divided into components attributable to changes

in factors of production. Rewriting above equation in growth rates:

p∆y{yq “ p∆A{Aq ` αp∆k{kq ` p1 ´ αqp∆n{nq

• Economists believe that virtually all technical change is endogenous,

the outcome of deliberate actions taken by economic agents. If so and if
the production function exhibits constant returns to scale, one would
expect to be able to account for the rate of growth of output in terms of
rates of growth of all inputs (Romer, 1986b).

• The issue of measuring inputs correctly in the growth accounting exer-

cise.

• Though productivity growth is important in determining economic

growth, it does not mean that accumulation is unimportant. For some
countries accumulation of capital is very important for realising eco-
nomic growth through technological progress.

• A growing body of research suggests that, even after physical and

human capital accumulation are accounted for, something else accounts
for the bulk of cross-country differences in the level and growth rate
of gross domestic (GDP) per capita. Economists typically refer to the
something else as total factor productivity (TFP)(Easterly and Levine,
2001).

• Different theories offer very different conceptions of TFP. These range

from changes in technology (the instructions for producing goods and
services) to the role of externalities, changes in the sectoral composition
 %$%:;<%4%&%)&=%85>%
8?8%=@A%011B3'7%&8%&&))=%8%%9)%@
'C%5&&&8D&%8$&%4%,:;<
$$&)$%&%%4%%@
%&%'C%9%&/01123,>%8
?8%=@A%011B3&$$$)5&%$$%
CHAPTER 10. SOLOW TYPE GROWTH MODELS: LIMITATIONS 89

EFGH!'I%%$%8J+$$?%&&C8-8A%&
0%$%3
I%$ I%$%85

6$)5  K#< A * :;<
!"#1LBMB
N$% 'L. 2'L. L L 22
J%)5 '1 O'O L  2O
C5 '1 2'. L P OL
Q '1 1'2. 2 P LP
R%8>8) ' 'B. LB  2P
R%8I%& 'L. L'.. L PL 
!"#1O.M1.
N$% 'LP '2. 2  L
J%)5 'L. 'P. 21 M L1
C5 ' L'. L1  L
Q 'LP O' 2B L P1
R%8>8) '1 P'L1 2P ML 2P
R%8I%& 'L '. L2 LP 
*+)%$1L.M.
+% '2L 'O. L PO 
S= 'L2 O'L. 2 P. P1
A% '2P '. L PO L.
T%9$ 'O1 O'. L. P B
(%%=% '22 2'P. 2B L 1
6&+&1OOM1.
U>,A 'B B'. LP P .
I% '2 '2. B P M2
>%,?%' 'P .'P LO LP P
7,A .'P1 1'. L. L. P.
VWX5Y4ZN!"#,A&%&,A))&,8Q%&01.38[%50113\
*+)%$,6&011.3\6&+&,/01123'

of production, adoption of low cost production methods, and institutional

arrangements.

• Easterly and Levine (2001) argues that factor accumulation does not
account for the bulk of the cross-country differences in the level or
growth rate of GDP per capita; TFP does. In the search for the secret
of long run economic growth, high priority should go to rigorously
defining TFP, empirically dissecting it, and identifying the policies and
institutions most conducive to its growth. They also argue that there
are huge and growing differences in GDP per capita; divergence – not
conditional convergence – is the big story. An emphasis on TFP growth
with increasing returns to technology is more consistent with divergence
CHAPTER 10. SOLOW TYPE GROWTH MODELS: LIMITATIONS 90

than are models of factor accumulation with decreasing returns, no scale

economies and some fixed factor of production. Note that the returns to
capital are not falling. See the table taken from Easterly and Levine
(2001).

• The literature shows that growth accounting exercise repeatedly failed

to account for the growth of output in terms of growth of inputs. This
implies that aggregate production functions are best described as ex-
0hibiting
12 345789  588
increasing 95to
returns 578 8 1986b).
scale(Romer,

phq95.0Jr/+4)*.'55+,3)23-O.r/+4)*.i0)78.=;.c752(7
|u
{u
}~€‚ƒ„ †ƒ‡ˆ†~‰†Š‹

zu „†“‘”• ’
tu
vu – ˆ‘’ƒ—‚ ~†
wu
xu
yu
sxu
swu
svu
stu
˜x
™™š w v t z { | Œ  ›œ
xyž
ŽŠ‰†ƒ†‰‘’† Ÿ ¡¢œ£¡¤¥¦£™§¥̈©£œª¡£«©¥¬š™ª¥˜­®¥z¯{°
0123456719 4 94 
bScQOB$d!e!  "! "#$"%&"
'()*+,-*.)/012)2+30(.-/+4)*.055+,3)23-.67080 ,/78.)*0).90/).+:.0.5+,3)/;<8.-/+4)*
/0)7.)*0).60;.=7.0))/2=,)0=(7.)+.:05)+/.055,6,(0)#2+3>.)*)28.8),0 1;.,878.!?0/)20357
175+69+82)2+3.)+.5+38)/,5).231250)+/8.+:.)*0).90/).+:."%%@"A5&B.1( 2::7/73578
23.75+3+625.-/+4)*./0)78.055%+&',3)7( 1.:+/.=;.0 "%%@"A&(B.12::7/7%3&5'78.23(.3CD.031
:05)+f/.-/+4)*.EF+378.0GGHIJ.'88,623-.)*0).K.L.JM>.)*7.:+((+423-.*+(18.:+/.)*7
5/+88N87256)24+732.B+ :.5+,3)/278O 4$8 4$? 4$4
EPI fQRS
 2  4 7 2 B
    ETBUBI.L.QRSETVWXUVWXI.Y.EZJMI\]^ET_U_I`.Y 4$g
4 $6 8 4[ $B 
26472B# [EZJM4I$\2ab ? ETVWXUVWX>.T_4U $_
8BI`J 74$?8
c75+69+823-.)*7.8+,/578.+:.-/+4)*.05/+88.5+,3B)4/278.,823-.12::7/47$3B).10)0.87)8
 2  4 7   4 $6  4 $
8*+48.)#*>F0:),.,5*)/+878N5*!+:,3!)"/$;.?0/20)2!+"$38.23.3CD.-/+4)*.055+,3).:+/.6+/7.)*0).2Z
97/573).F+:).+,,)9,).!-"/$+4)*.,823-.0()7/30)2?7.10)0.87)8.E)0=(7.[IJ.d*7.5/+88N5+,3)/;
?0/20)2+h3i.j235k.49l*M;8250"(G."5092)0(.0(+"3#7e7 "!f 5(,1 !2 '3<%
-.)8*$76.5+?0/2"035+7."42 !!)"*.3 CD .-/+4)
=! ##*e
37?7/.05>5+!,3!)G"8.1:+2/2.?6+ 3!"/ 7 .)*!"03).[g .97/573).+:.)(
*7$.5/+88N5+,3)/;.?0/20)2+3.23.97/
5092)0.h D.-/+4)*J
i7870/5*7/8.0(8+.235+ /9+/0)7.*,60 3!."5092)0+(8.20)!25+53,.6,
!#,(00)")22++33.8!2."30)8+O.$1%7&5'+69
+( 82N)*
)2+3.)7!f7!/52878">./74/
2)23-).)*7.!?0#/20+357)N175+""69 *,7j( ))!!!"$-!(
EMI . !/*0E!TB
#24!]^
122#3!"ETVW
!UB)I.[LE.!Z]^ J5HI\ab
!XI.Y.EZJ)H)I![\!]^
XEUTVWVW
XUVW X>(   
!ET"kUkI%`.&Y'()
!")!(+!)
.T k U k I `>
4*7/"7.T
klk.!28.:0)52)1+#/.0"5!5,6, !"(0)2+3.97/.24+
 "          
64/7m278/>.1)7:!2371.08.)*7.0?7)/0-7.-
9  # ! :  !  !  ! 3 $; !  " !  )
*./0)7*
/!+'4)
+:.9*;82"50(.5+0"9!2)0!(".97/.4+  ! " /m
 7/.0!"31.)71,!50)2+( 30(.0))!023)67   3
).97/.4+   /m7)/J.n<%

  

582$:62N*
97=! ##
50((;>.Tklk>L.!ET_ !l_1.Y2.2T?o3loIU"[>."4* 7(" /7.o.28.71),)!50!)2+3!0"(.0)%)0&2'367 3( ).97/.4+ /m7/)JH#
24023!>.12:!:7/73).0),)*+/"8.",*87.12::7/73,).47)2-)*!)8!>.)*+!,"-*.)*28!.)7318.!3+).)+.5*03-7@A'
HJ'- (23-8J
26472B$;"))!!!"%&'(C!).)*7.=0825.:231 
!"C"!!+!!!"!:""*,))!!!"
CHAPTER 10. SOLOW TYPE GROWTH MODELS: LIMITATIONS 91

Factor Movement

• factor movement toward the richest areas reinforces the concentration

of economic activity. Each factor of production moves to where it is
already abundant.
• Labour migration is overwhelmingly toward the richest countries. The
three richest countries alone, the United States of America, Canada,
Switzerland, half the net immigration of all countries reporting net
immigration (Easterly and Levine, 2001).
• The pattern of migration is reflecting the differential rate of return to
education in rich and poor countries.
• Countries in the richest quintile are the net recipients of migrants. Only
8 of the 90 countries of the bottom 4/5 of the sample are net recipients
of the migrants.
• Embodied in this flow of labour are the flows of human capital towards
the rich countries, the famous brain drain.
• Carrington and Detragiache (1998) found that in 51 of 61 developing
countries in their sample, people with a university education were more
likely to emigrate to the United States than people with a secondary
education. In all 61 countries, migration rates to the US were lower for
people with a primary education or less than for people with a secondary
or university education.
• Capital also flows mainly to areas that are already rich. In 1990, the
richest 20% of the world population received 92% of gross portfolio
capital inflows, whereas the poorest 20% received 0.1%. The richest
20% of the world population received 79% of foreign direct investment,
and the poorest 20% received 0.7%. Altogther, the richest 20% of the
world population received 88% of the gross private capital gross inflows,
and the poorest 20 per cent received 1 per cent.
• The high concentration of income, reinforced by the flow of all factors
toward the richest areas, is inconsistent with the neoclassical growth
model.
• These facts seem to be consistent with the productivity explanation of in-
come differences than with factor accumulation story. If rich area is rich
because technology (A) more advanced, then all factors of production
will tend to move towards the rich area, reinforcing concentration.
CHAPTER 10. SOLOW TYPE GROWTH MODELS: LIMITATIONS 92

• Technological spillovers between agents would lead to endogenous

matching of rich agents with each other, and those matching will rein-
force the matching of poor people with other poor people (as in the O-ring
story of Kremer (1993) or the inequality model of Bénabou (1996))
Chapter 11

New Growth Models

11.1 Romer (1986) Model

• This model offers an alternative view of long run prospects for growth.
In a fully specified competitive equilibrium, per capita output can grow
without bound, possibly at a rate that is monotonically increasing over
time.
• The rate of return on capital and rate of investment may increase rather
than decrease with increases in the capital stock. Hence, the level of
per capita output in different countries need not converge; growth may
be persistently slower in less developed countries and may even fail to
take place at all.
• Above results are not based on any kind of exogenously specified techni-
cal change or differences between countries. Preferences and technology
are stationary and identical.
• The model can be viewed as an equilibrium model of endogenous tech-
nological change in which long-run growth is driven primarily by the ac-
cumulation of knowledge by forward-looking, profit maximising agents.
• New knowledge is assumed to be a product of a research technology
that exhibits diminishing returns to scale.
• Knowledge has natural externality.
• These three elements – externalities, increasing returns in the pro-
duction of output, and decreasing returns in the production of new
knowledge – combine to produce a well-specified competitive equilib-
rium model of growth.

93
CHAPTER 11. NEW GROWTH MODELS 94

• Diminishing returns in the production of knowledge is required to

ensure that consumption and utility do not grow too fast.

Background of increasing returns to scale

• The idea that increasing returns are central to the explanation of the
long-run growth of economies are are as old as Adam Smith’s example
pin factory(Romer, 1986b).

• Later Alfred Marshall introduced the idea of internal and external

economies of scale. Subsequently Young (1928) emphasised the role
of industry or economy wide increasing returns to scale in the growth
process.

• Because of the technical difficulties posed by dynamic models, Mar-

shall’s concept of increasing returns to scale that are external to the
firm but internal to the industry was most widely used in static models,
especially in the field of international trade.

• Following Smith, Marshall, and Young, most authors justified the ex-
istence of increasing returns on the basis of increasing specialisation
and division of labour. It is now clear that these changes in the or-
ganisation of production cannot be rigorously treated as technological
externalities. Formally, increased specialisation opens new markets and
introduces new goods. All producers in the industry may benefit from
the introduction of these goods, but they are goods, not technological
externalities.

• In models, increasing returns is modelled as firms’ unit cost is decreas-

ing in industry output.

• One application of externality induced increasing returns in the context

of growth is Arrow (1962a). In his model productivity of a given firm is
assumed to be an increasing function of cumulative aggregate invest-
ment for the industry. Source of increasing returns is learning by doing
and consequent knowledge spillovers.

Technical difficulties of models with increasing returns to scale

• First issue is the existence of competitive equilibrium. If the increasing

returns to scale are external to the firm, an equilibrium exists.
CHAPTER 11. NEW GROWTH MODELS 95

• The second problem, unique to the dynamic optimising models, con-

cerns the existence of a social optimum and finiteness of the objective
functions. In a standard optimising growth model that maximises a
discounted sum or integral over an infinite horizon, the presence of in-
creasing returns raises the possibility that feasible consumption paths
may grow so fast that the objective function is not finite. Arrow’s(1962)
solved this problem by assuming that marginal product of capital de-
creases with fixed supply of labour. That is labour supply is acting as a
constraint on the growth of output.

Details of the Romer (1986b) model

• The model assume that knowledge is a capital good with an increasing

marginal product.

• Production of the consumption good is assumed to be globally convex,

not concave, as a function of the stock of knowledge when all other
inputs are held constant.

• A finite-valued social optimum is guaranteed to exists because of the di-

minishing returns in the research technology, which imply the existence
of a maximum, technologically feasible rate of growth for knowledge.
This in turn implies the existence of a maximum feasible rate of growth
for per capita output. Over time the rate of growth of output may be
monotonically increasing, but it cannot exceed this upper bound.

• That is the diminishing returns in the research will limit the rate
of growth of the state variable. A general proof that restrictions on
the rate of growth of the state variable are sufficient to prove the
existence of an optimum for a continuous-time maximisation problem
with noncovexities is given in Romer (1986a).

• Assume that production of consumption good is a function of the state

of knowledge, denoted by k, and the set of additional factors such as
physical capital, labour and so forth, denoted by a vector x.

• The newly produced private knowledge can be only partially kept secret
and cannot be patented. We can represent the technology of firm i in
terms of a twice continuously differentiable production function F that
depends on the firm specific inputs ki and xi and on the aggregate level
of knowledge in the economy. If N isř the number of firms, define this
aggregate level of knowledge as K “ N i“1 ki .
CHAPTER 11. NEW GROWTH MODELS 96

• The first major assumption on the production function F pki , K, xi q is

that, for any fixed value of K, F is concave as a function of ki and xi .
Without this assumption, a competitive equilibrium will not exists in
general.
• Once concavity is granted, there is little loss of generality in assuming
that F is homogeneous of degree one as a function of ki and xi , when K is
held constant; any concave function can be extended to be homogeneous
of degree one by adding an additional factor to the vector x if necessary
(Rockafellar, 1970).
• By homogeneity of F in ki and xi , and by the assumption that F is
increasing in the aggregate stock of knowledge, K, it follows that F
exhibits increasing returns to scale. For any λ ą 1,

F pλki , λK, λxi q ą F pλki , K, λxi q “ λF pki , K, xi q

Lucas (1988)
• The author is considering an alternative, or at least a complementary,
engine of growth to the technological change that drives growth in the
Solow Model.
• Lucas is doing this by adding what Schultz(1963) and Becker (1964) call
human capital to the model. This is very close technically to similarly
motivated models of Arrow (1962a), Uzawa(1965) and Romer (1986b),
Romer (1986b).
• By an individual’s human capital means simply his general skill level
so that a worker with human capital hptq is the productive equivalent
of two workers with 12 hptq each or a half-time worker with 2hptq. The
theory of human capital focuses on the fact that the way an individual
allocates his time over various activities in the current period affects
his productivity, or his hptq level in future periods. Introducing hu-
man capital into th model, then, involves spelling out both the way
human capital levels affect current production and the way current
time allocation affects the accumulation of human capital.
• Suppose there are N workers in total with skill levels h ranging from
0 to şinfinity. Let there be N phq workers with skill level h, so that
8
N “ 0 N phqdh.
CHAPTER 11. NEW GROWTH MODELS 97

• Suppose a worker with skill h devotes the fraction uphq of his non-leisure
time to current production, and the remaining 1 ´ uphq to human capital
accumulation.
ş8 Then the effective workforce in production is the sum
N e “ 0 uphqN phqhdh of th skill-weighted man hours devoted to current
production.

• In addition to the effects of an individual’s human capital on his own

productivity, what Lucas calls internal effect of human capital – Lucas
also consider an external effect. Let the average level of skill or human
capital, defined by ş8
hN phqdh
ha “ ş08
0
N phqdh

• To simplify the analysis considerably, we treat all workers in the econ-

omy as being identical. In this case, if all workers have skill level h and
all choose the time allocation u, the effective workforce is just N e “ uhN
and the average skill level ha is just h. We will use the notation ha for
average skill level to emphasis the distinction between internal and
external effects.

• Production function

Y “ AKptqβ ruptqhptqN ptqs1´β ha ptqγ

where the term ha ptqγ is intended to capture the external effects of

human capital, and where the technology level A is now assumed to be
constant.

• Note that the above production function exhibits increasing returns to

scale with respect to all inputs. The increasing returns to scale is due
to externality from human capital accumulation.

• Accumulation of human capital

9
hptq “ hptqδr1 ´ uptqs

Note that production of human capital depends on upon the already

achieved level of human capital. This implies that scale matters, higher
the level of human capital already achieved, higher the addition to the
human capital.

•
9
hptq
“ δp1 ´ uq “ ν
h
CHAPTER 11. NEW GROWTH MODELS 98

• The model shows that

K9
ˆ ˙
cptq
9 1´β`γ
“ “ ν
cptq K 1´β

Spillovers and increasing returns: A Cobb-Douglas

case
• As we have already seen, in the simple version of the Romer’s model,
output Yi depends not only on the standard private input Ki and Li , but
also on the economy wide capital stock K. The idea is that investment
in Ki by the ith firm generates knowledge. A part of this knowledge
gets into the public domain due to its non-rival and non-excludable
character. Other firms can make use of this knowledge to increase their
productivity, making firm’s productivity depends on aggregate capital
stock.

• The idea can be illustrated with using Cobb-Douglas production function

as follows;
Yi “ AKiα K β L1´α
i

where 0 ă α ă 1, and β ě 0. For given K, this production function

exhibits constant returns to scale in the private inputs Ki and Li . If
β ą 0, then spillover effect is present.

• In Griliches (1979) version of the production function in the above

equation, Ki represents firm i’s specific knowledge capital, whereas K
(modelled as sum of the Ki ) is the aggregate level of knowledge in an
industry.

• In the Lucas (1988) version, Ki is the firm’s employment of human

capital, and K is the aggregate (or average) level of human capital in
an industry or country. In this case spillovers involve benefits from
interacting with smart people.

• Returning to the Romer interpretation of the above production function,

each firm behaves competitively taking as given the economy wide factor
prices R and w and the aggregate capital stock K. So private marginal
products are equated to the factor prices.

• This yields the following

R “ αYi {Ki and w “ p1 ´ αq ¨ Yi {Li

CHAPTER 11. NEW GROWTH MODELS 99

• The factor income share are therefore given, as usual, by

sK “ α and sL “ 1 ´ α

• Since R and w are given and same for all firms, in equilibrium each
firm adopts same capital-labour ratio ki , but the scale of each firm is
indeterminate. The above production function can be written as follows

•
Yi “ Akiα k β Li Lβ
where k ” K{L. The equilibrium condition ki “ k, then implies

Yi “ Ak α`β Li Lβ

which can be aggregated across firms to get

Y “ Ak α`β L1`β

• Finally, the condition k ” K{L results in the following economy-wide

production function
Y “ AK α`β L1´α

• This expression relates aggregate output Y to the aggregate inputs K

and L. If β ą 0, then increasing returns to scale apply economy-wide.

Specialisation and Increasing Returns (Romer, 1987)

• A production function using both labour and intermediate inputs that
is analogous to the utility function of Dixit and Stiglitz (1977) is
ż ˆ ˙
xpiq
Y pL, xq “ L g di
R` L

where g is an increasing strictly increasing concave function with gp0q “

0. In the special case considered by Dixit and Stiglitz (1977) and Ethier
(1982), g is a power function gpxq “ xα , with 0 ă α ă 1. Then Y takes
the more familiar form
ż
1´α
Y pxq “ L xpiqα di
R`
CHAPTER 11. NEW GROWTH MODELS 100

Suppose that production technology of all the intermediate inputs xi

are identical and the price per unit of intermediate input can be derived
from the production function for Y and is given below.

Pi “ αL1´α xα´1
i

Given the above shadow price and the assumption that all intermediate
inputs are produced using identical technologies implies that same
quantity of each variety is produced in the economy. Let xi “ x̄ and
also assume that there are N varieties. Then the total quantity of
intermediate inputs is X “ N x̄. Now we can write the production
function as follows:
ż ˆ ˙α ˆ ˙α
1´α X 1´α X
Y pL, N, Xq “ L “L N
R` N N

Rearranging we get the economy wide production function as follows.

Y pK, N, Xq “ N 1´α pL1´α X α q

• Note that the above production function exhibits constant returns to

scale with respect to X and L, that is with respect to total quantity of
capital and labour. If an increase in the scale of the economy (i.e. X and
L) also results in an increase in the variety of inputs (N ), the production
function exhibits increasing returns to scale.

11.2 Romer (1990b)

• The paper presents a model in which economic growth is driven endoge-
nously by accumulating knowledge motivated by profit.

• The paper argues that the raw materials that we use have not changed,
but as a result of trial and error, experimentation, refinement, and
scientific investigation, the instructions that we follow for combining
raw materials have become vastly more sophisticated. One hundred
years ago, all we could do to get visual stimulation from iron oxide was
to use it as a pigment. Now we put it on plastic tape and use it to make
videocassette recordings.

• The argument presented in the paper is based on the following three

premises.
CHAPTER 11. NEW GROWTH MODELS 101

• The first is that technological change - improvement in the instructions

for mixing together raw materials - lies at the heart of economic growth.
As a result, the model presented in the paper resembles the Solow
(1956) model with technological change. Technological change provides
the incentive for continued accumulation of both physical and human
capital, and together, capital accumulation and technological change
account for the observed growth in output per person.

• The second premise is that technological change arises in large part

because of intentional actions taken by people who respond to market
incentives. Thus, as we already mentioned above, the model is one
of endogenous rather than exogenous technological change. This does
not mean that everyone who contributes to technological change is
motivated by market incentives. The premise here is that market
incentives nonetheless play an essential role in the process whereby
new knowledge is translated into goods with practical value.

• The third and most fundamental premise is that instructions for work-
ing with raw materials are inherently different from other economic
goods. Once the cost of creating a new set of instructions has been in-
curred, the instructions can be used over and over again at no additional
cost. Developing new and better instructions is equivalent to incurring
a fixed cost. This property is taken to be the defining characteristic of
technology.

• Once these three premises are granted, it follows directly that an equi-
librium with price taking cannot be supported. In the model a firm
incurs fixed design or research and development costs when it creates
a new good. It recovers those costs by selling the new good for a price
that is higher than its constant cost of production. Since there is free
entry into this activity, firms earn zero profit in a present value sense.

• On the basis of results from the static theory of trade with differentiated
goods (see, e.g., Helpman and Krugman (1985)), one should expect that
fixed costs lead to gains from increases in the size of the market and
therefore to gains from trade between different countries.

• The model defines an equilibrium that allows for both monopolistic

competition and external effects arising from knowledge spillovers.

• Economists studying public finance have identified two fundamental

attributes of any economic good: the degree to which it is rivalrous and
the degree to which it is excludable.
CHAPTER 11. NEW GROWTH MODELS 102

• Rivalry is a purely technological attribute. A purely rival good has the

property that its use by one firm or person precludes its use by another;
a purely nonrival good has the property that its use by one firm or
person in no way limits its use by another.
• Excludability is a function of both the technology and the legal system.
A good is excludable if the owner can prevent others from using it. A
good such as the code for a computer program can be made excludable by
means of a legal system that prohibits copying or by means of encryption
and copy protection schemes.
• By definition, public goods are both non-rival and non-excludable. Be-
cause they are non-excludable, they cannot be privately provided or
traded in markets. Public goods can be introduced into a model of
price-taking behaviour by assuming the existence of a government that
can levy taxes. Basic scientific research is an example of a public good
that could be provided in this way and that is relevant for modelling
growth. The interesting case for growth theory is the set of goods that
are non-rival yet excludable.
• The third premise cited above implies that technology is a non-rival
input. The second premise implies that technological change takes place
because of the actions of self-interested individuals, so improvements in
the technology must confer benefits that are at least partially excludable.
The first premise therefore implies that growth is driven fundamentally
by the accumulation of a partially excludable, non-rival input.
• Non-rivalry has two important implications for the theory of growth.
First, non-rival goods can be accumulated without bound on a per
capita basis, whereas a piece of human capital such as the ability to
add cannot. Each person has only a finite number of years that can be
spent acquiring skills. When this person dies, the skills are lost, but any
nonrival good that this person produces-a scientific law; a principle of
mechanical, electrical, or chemical engineering; a mathematical result;
software; a patent; a mechanical drawing; or a blueprint lives on after
the person is gone. Second, treating knowledge as a nonrival good
makes it possible to talk sensibly about knowledge spillovers, that is,
incomplete excludability.
• These two features of knowledge-unbounded growth and incomplete
appropriability - are features that are generally recognised as being
relevant for the theory of growth. Also non-rival feature is also linked
to the presence of non-convexities.
CHAPTER 11. NEW GROWTH MODELS 103

• If a nonrival input has productive value, then output cannot be a

constant-returns-to-scale function of all its inputs taken together. The
standard replication argument used to justify homogeneity of degree
one does not apply because it is not necessary to replicate nonrival
inputs.

• Suppose a firm invest 1000 engineering hours to develop a 20 Mb hard

disk drive for computers. Suppose it can produce it can produce 20000
MB of disk space (1000 units) every year using a factory worth |100000
and 100 workers. By just doubling its rival input, factory and workers,
it can double its output to 40000 MB disk space in a year (2000 units).
Suppose, the firm could have invested 2000 engineering hours and
develop a 30 MB hard disk, which can be produced using same factory
and workers. In this case doubling of all factors, engineering hours,
factory and workers would result in an output of 60000 MB disk space,
three times of the original output.

• More formally, if F pA, Xq represents a production process that depends

on rival inputs X and nonrival inputs A, then by a replication argument,
it follows that F pA, λXq “ λF pA, Xq. If A is productive as well, it follows
that F cannot be a concave production function because F pλA, λXq ą
λF pA, Xq.

• Because of the properties of homogeneous functions, it also follows that

a firm with these kinds of production possibilities could not survive as
a price taker. If disk drives sold for marginal cost, annual revenue for
the firm would just equal interest payments on the capital and wage
payments to workers. More generally, since
BF pA, Xq
F pA, Xq “ X ¨
BX
It follows that
BF pA, Xq BF pA, Xq
F pA, Xq ă A ¨ `X ¨
BA BX
This implies that if all inputs were paid their value marginal product,
the firm would suffer losses.

• This point has been made many times before (Schumpeter 1942; Arrow
1962b; Shell 1966, 1967, 1973; Nordhaus 1969; Wilson 1975). Previous
growth models have avoided this difficulty in various ways. Solow (1956)
treats A as an exogenously provided public input (i.e., an input that
CHAPTER 11. NEW GROWTH MODELS 104

is both nonexcludable and nonrival). Shell (1966, 1967) treats it as a

public input that is provided by the government. In each case, the factor
A receives no compensation, and every individual firm is assumed to
be free to exploit the entire stock of A. Lucas (1988) assumed in effect
that it is production of human capital rather than physical capital that
generates this nonrival, nonexcludable good.

• The advantage of the interpretation that knowledge is compensated out

of quasi rents is that it allows for intentional private investments in
research and development. The difficulty is that it violates the logic of
the replication argument. If the input A is truly the result of research
and development, it is a nonrival good. In this case, the function F(.)
must be homogeneous of degree one in X alone and the Dasgupta and
Stiglitz (1988) argument applies. If a nonrival input is even partially
excludable, non-convexities are present and a decentralised equilibrium
cannot be sustained.

• The four basic inputs in this model are capital, labor, human capital,
and an index of the level of the technology. Capital is measured in
units of consumption good. Labor services L are skills such as eye-
hand coordination that are available from a healthy physical body and
measured by counts of people. As used here, human capital H is a
distinct measure of the cumulative effect of activities such as formal
education and on-the-job training.

• The model used here separates the rival component of knowledge, H,

from the nonrival, technological component, A. Because it has an
existence that is separate from that of any individual, A can grow
without bound. In the specific formulation used in the paper, each new
unit of knowledge corresponds to a design for a new good, so there is no
conceptual problem measuring A. It is a count of the number of designs.

• The formal model of the economy has three sectors. The research sector
uses human capital and the existing stock of knowledge to produce new
knowledge. Specifically, it produces designs for new producer durables.

• An intermediate-goods sector uses the designs from the research sector

together with forgone output to produce the large number of producer
durables that are available for use in final goods production at any
time. A final-goods sector uses labor, human capital, and the set of
producer durables that are available to produce final output. Output
can be either consumed or saved as new capital.
CHAPTER 11. NEW GROWTH MODELS 105

• To keep the dynamic analysis simple and highlight the effects of in-
terest, several simplifying assumptions are used. The first is that the
population and the supply of labor are both constant. This rules out
an analysis of fertility, labour force participation, or variation in hours
worked per worker.

• The second is that the total stock of human capital in the population is
fixed and that the fraction supplied to the market is also fixed.

• Also, the plausible assertion that research is relatively human capital-

and knowledge-intensive is translated into an extreme specification
in which only knowledge and human capital are used to produce new
designs or knowledge. Labour and capital do not enter at all. These
kinds of restrictions will reduce the analysis of the dynamics of this
system to a system of equations that can be explicitly solved by doing
algebra.

• Final output Y in this model is expressed as a function of physical

labor L, human capital devoted to final output Hy, and physical capital.
The unusual feature of the production technology assumed here is that
it disaggregates capital into an infinite number of distinct types of
producer durables.

• Only a finite number of intermediate inputs, the ones that have already
been invented and designed, are available for use at any time. Thus
if x “ txi ui“18 , is the list of inputs used by a firm that produces final
output, there is some value A such that xi “ 0; @i ą A.

• The final goods production function is;

ż8
Y pHY , L, xq “ HYα Lβ xpiq1´α´β di (11.1)
0

• Total capital K is the cumulative foregone output. Thus Kptq evolves

according to,

9
Kptq “ Y ptq ´ Cptq
Where Cptq is the aggregate consumption at time t. K can also be
written as,
ż8 żA
K“η xi “ η xi
i“1 i“1
CHAPTER 11. NEW GROWTH MODELS 106

• As noted above research output depends on the amount of human capital

devoted to research. It also depends on the stock of knowledge available
to a person doing research. The model assumes that anyone engaged
in research has free access to the entire stock of knowledge. This is
feasible because knowledge is a non-rival input. All researchers can
take advantage of A at the same time. The output of researcher j is
therefore δH j A. Summing across all people engaged in research, the
aggregate stock of knowledge (design) evolves according to,
A9 “ δHA A
HA is the total human capital employed in research. The crucial feature
of the specification used here is that knowledge enters into production
in two distinct ways. A new design enables the production of a new good
that can be used to produce output. A new design also increases the
stock of knowledge and thereby increases the productivity of human
capital in the research sector. The owner of a design has property right
over its use in production of new producer durable, but not over its use
in research. The research technology also implies that,
A9
g “ “ δHA
A
• The model shows that
C9 Y9 Y9 A9
g“ “ “ “ “ δHA
C Y Y A
Thus the economy would grow at the rate of technological progress, g.
The most interesting positive implication of the model is that an econ-
omy with larger stock of human capital will experience faster growth.
This suggests that free international trade may accelerate growth, when
it expands the stock of human capital.
• The growth rate is increasing in the stock of human capital, but it does
not depend on the total size of the labor force or the population. In a
limiting case that may be relevant for historical analysis and for the
poorest countries today, if the stock of human capital is too low, growth
may not take place at all.

11.3 Semiendogeneous and Schumpeterian Growth

Models
• Jones (1995b,a) observed that since the early 1950s the number of
CHAPTER 11. NEW GROWTH MODELS 107

scientists and engineers engaged in R&D in the United States has

grown more than five-fold without resulting in any increase in the
growth rate of output per person or TFP. The coexistence of upward
trend in R&D labour and no trend in TFP growth has effectively refuted
the first generation of R&D -based theories, according to which more
R&D labour should induce more TFP growth (Ha and Howitt, 2007).

• Theorists have responded to the challenge by developing a second gen-

eration of R&D-based theories that are consistent with these findings.
The new theory comes in two varieties, with similar foundations, but
radically different long run implications.

• One is the “semi-endogenous” theory of Jones (1995a), Kortum (1997)

and Segerstrom (1998), which modifies the original theory by incorpo-
rating diminishing returns to the stock of knowledge in R &D. That
is, as technology develops and become increasingly complex, sustained
growth in R&D labour becomes necessary just to maintain a given rate
of TFP growth.

• Semi-endogenous growth theory has a stark long-run prediction, namely

the long-run rate of TFP growth, and hence the long-run growth rate
of per capita income depends on the rate of population growth, which
ultimately limits the growth rate of R&D labour, to the exclusion of all
economic determinants. That is why semi-endogenous growth theory.

• The other variety of second-generation R&D-based theory consists of the

fully endogenous “Schumpeterian” models of Aghion and Howitt (2008),
Dinopoulos and Thompson (1998), Peretto (1998), Howitt(1999). These
authors all builds on the Young’s (1998) insights that as the economy
grows, proliferation of product varieties reduces the effectiveness of
R&D aimed at quality improvement, by causing it to be spread more
thinly over a large number of different sectors.

• Schumpeterian theory retains the original assumption of constant re-

turns to the stock of knowledge in R&D, and therefore implies that the
long-run rate of TFP growth will be governed by the same economic
factors as in the first-generation R&D based theories, with the sole ex-
ception that the size of a country’s labour force no longer has a positive
scale effect on long-run growth. The theory is consistent with the ob-
served co-existence of stationary TFP growth and growing R&D labour,
because growing R&D labour is required to counter the deleterious
effect of product proliferation on the productivity of R&D.
CHAPTER 11. NEW GROWTH MODELS 108

• Note that the central idea of endogenous growth theory is that long-run
economic growth has economic determinants, which in R&D-based the-
ories consist of the policies, regulations, and institutions that impinge
on the incentive to create and adopt new technologies. Schumpeterian
theory confirms this idea, and semi-endogenous theory denies it.

Different versions of R&D-based growth theory

• The central component of any R&D-based growth models is a knowledge-
creation function, according to which the flow A9 of new knowledge
depends on R&D input X and other variables.

• In these models A measures productivity as well as knowledge, so that

specifying a knowledge-creation function is equivalent to specifying
a productivity-growth function, according to which the growth rate
9
gA ” A{A of TFP is a function of R&D input and other variables.

• The key difference between competing models can be interpreted for-

mally as different assumptions on these two related functions, which
are typically specified as Cobb-Douglas.

• First-generation fully Endogenous Models: In models developed by

Aghion and Howitt (1992), Grossman and Helpman (1990) and Romer
(1990b) the productivity growth function is:

gA “ λX σ

where 0 ă σ ď 1. Equivalently the knowledge creation function is

A9 “ λX σ A

• R&D input X is specified as either the flow N of labour allocated to

R&D,or the productivity-adjusted flow R{A of R&D expenditure on
capital and labour. The model all assume a constant population size
and possess a balanced-growth equilibrium in which X is constant.
Long-run growth thus depend on policies that determine the long-run
level of R&D input.

• The unit exponent on A on the right-hand side of the above knowledge

production function imposes constant returns to knowledge in the pro-
duction of new knowledge. Constant returns are what allows sustained,
non-explosive endogenous growth.
CHAPTER 11. NEW GROWTH MODELS 109

• This specification of knowledge production function is not supported by

the empirical data.
• Semi-endogenous growth Models: To avoid the above mentioned
problem, Jones (1995a) developed an elegant semi-endogenous model
with decreasing returns to knowledge.

gA “ λX σ Aϕ´1 , 0 ă ϕ ă 1

where again X is the R&D input. Taking logs and differentiating both
sides of the above equation with respect to time yields:
g9A
“ p1 ´ ϕqpγgX ´ gA q
gA
σ
where γ ” 1´ϕ
and gx is the exponential growth of R&D input.
• This semi-endogenous growth model is compatible with the observation
of positive trend growth in R&D input, because as long as 0 ă ϕ ă 1 and
the time path of gx is bounded the above differential equation yields a
bounded solution for TFP growth. In particular, if gX is constant, or
approaches a constant, then

gA ptq Ñ γgX

Fully endogeneous growth Models with product proliferation

• The other version of R&D-based theory that is compatible with growing
population and growing R&D labour is the Schumpeterian version.
These models retain the assumption of constant returns to knowledge,
but take into account the deleterious effect on productivity growth
arising from product proliferation, as pointed out by Young (1998) by
specifying gA “ λpX{Qqσ , where Q is a measure of product variety. In
these models, Q is typically proportional to the size L of population, at
least in the long run. So we have

gA “ λpX{Qqσ , or gA “ λpX{Lqσ

• An economic interpretation of the above equation is that a larger popu-

lation increases the number of people who can enter an industry with
a new product, this resulting in more horizontal innovations, which
dilutes R&D expenditure over a larger number of separate projects.
That is the growth-enhancing effect of rising R&D input is just offset
by the deleterious effect of product proliferation.
CHAPTER 11. NEW GROWTH MODELS 110

• This theory is consistent with the observed coexistence of stationary

TFP growth and rising R&D input. It also retains most of the long
run implications of the first-generation theory. In particular, it implies
that anything that raises the fraction of society’s resources allocated to
R&D will raise the long run TFP growth rate. We call it this branch of
theory as second generation fully endogenous theory or “Schumpeterian”
theory.

• Schumpeterian theory differs from semi-endogenous theory in terms of

the necessary conditions it implies for sustaining long-run productivity
growth. According to semi-endogenous growth theory society must
sustain growth in R&D input. But according to Schumpeterian theory,
society must just sustain the fraction of resources it allocates to R&D
Chapter 12

Evolution of growth theory

12.1 Technological Progress and Growth Mod-

els
The role of technological progress in sustaining economic growth is formally
modelled first in Solow (1956) and Swan (1956). The Solow-Swan model
consists of an economy-wide production function of the Cobb-Douglas form
with constant returns to scale and diminishing returns to each input1 . This
production function is combined with a constant saving rate rule to gener-
ate a simple general equilibrium model of the economy. In this economy,
at very low level of capital per worker, capital per worker and per capita
income grow faster due to the higher marginal productivity of capital. As
the capital accumulation continues, diminishing returns to capital sets in,
slowing down the rate of accumulation and ultimately placing the economy
in the steady state with no growth in per capita output2 . The basic Solow-
Swan model, therefore, unable to explain the long run growth of the now
developed economies. It is explained by incorporating technological progress
into the model. Technological progress halts the marginal product of capital
from falling and thereby sustains the pace of capital accumulation. In these
1
Cass (1965) and Koopmans (1965) made saving decision endogenous and this extension
also preserved the predictions of the Solow-Swan model. One implication of Solow-Swan
model is that economies having less capital stock per worker (relative to their long-run
capital stock) tend to grow faster than economies having more capital per worker, leading
to conditional convergence in income levels. Conditional convergence means convergence
in income levels conditional on the differences in factors such as technology, saving rates
and human capital stock. This is one of the most empirically tested propositions of the
neoclassical growth theory(Ray, 1999; Barro and Sala-I-Martin, 2004).
2
In this literature, capital is defined in a narrow sense, referring mainly to machinery
and equipment(Grossman and Helpman, 1991a)

111
CHAPTER 12. EVOLUTION OF GROWTH THEORY 112

models, however, technological progress is exogenous and hence, they are

incapable of explaining the long run growth of economies (Barro and Sala-I-
Martin, 2004). Further, the Solow-Swan model, based on exogenous technical
change, also failed to account for many of the empirical realities like lack of
cross-country convergence, international pattern of migration and capital
movement (Romer, 1986a, 1994; Lucas, 1988).
The assumption that technological progress is exogenous to the economic
system found itself very difficult to reconcile with reality3 . Quoting Schmook-
ler (1966, p.199), who took the view that economic factors have a dominant
influence on the pattern of technical change,
Despite the popularity of the idea that scientific discoveries
and major inventions typically provide the stimulus for inventions,
the historical record of important inventions in petroleum refin-
ing, paper making, railroading and farming revealed not a single,
unambiguous instance in which either discoveries or inventions
played the roles hypothesised. Instead, in hundreds of cases the
stimulus was the recognition of a costly problem to be solved or a
potentially profitable opportunity to be seized; in short, a technical
problem or opportunity evaluated in economic terms. In a few
cases, sheer accident was credited.
Similarly, Dosi (1988) concludes his survey of sources and patterns of
industrial innovation that technological progress reflects an interplay of tech-
nological opportunities created by scientific discoveries and inducements for
applied research that emerge from market opportunities. Not only invention
and innovation are determined by the economic factors, diffusion of technol-
ogy also closely depends on them. For instance,Griliches (1957) argues that
the timing of the diffusion process of hybrid-seed can be explained very well
in economic terms and shows that behaviour of both farmers and hybrid-seed
producers was firmly founded in expectations of profit. Another study by
Mansfield (1961), which examines the speed with which twelve important
innovations diffused among firms in four industries, also shows that rate of
imitation is a direct function of the profitability of a given innovation and
decreasing function of the size of investment required for its installation.
3
Though two prominent scholars, namely Marx in the 19th century and Schumpeter in
the 20th century, emphasised the critical role of technological progress in the dynamics of
capitalism, it was relatively neglected in the mainstream economics literature. Various
explanations were advanced for this, the most frequent being the ‘black box’ explanations-
that technical change was outside the specialised competence of most economists and had
to be tackled by engineers and scientists. This argument went well with the assumption
(although erroneous) that technology could be treated as exogenous manna from heaven
(Freeman, 1994).
CHAPTER 12. EVOLUTION OF GROWTH THEORY 113

The limitations of the Solow-Swan model as well as the desire to have

formal models that can accommodate growth experience of developed as well
as under developed countries prompted the theorists to formulate models in
which technological progress is endogenous. Earlier attempts to incorporate
a theory of technological progress into the neoclassical growth models proved
difficult, because most of them assume a competitive market. Technological
progress involves creation of new ideas or knowledge, which are non-rival and
have features of public good4 . Inclusion of a non-rival commodity as a factor
of production makes the resulting production function exhibit increasing
returns to scale, which conflicts with the assumption of perfect competition.
A firm, having increasing returns to scale production function, cannot survive
as a price taker, since the compensation of non-rival old ideas in accordance
with their current marginal cost of production - zero - does not provide the
appropriate reward for the research effort that underlies the creation of new
ideas Romer (1990b); Barro and Sala-I-Martin (2004).
The earlier attempts to endogenous technological progress include Arrow
(1962b), Sheshinski (1967) and Shell (1967). In Arrow (1962b) and Sheshinski
(1967) knowledge is generated through learning by doing and, due to its non-
rival character, it is assumed to instantaneously diffuse throughout the
economy5 . Shell (1967) incorporates the view that innovation is driven by
basic research. This model includes a public research sector and profit-
seeking firms that use knowledge produced by the former in production.
The public research sector is financed by the government out of its tax
revenue and therefore the size of this sector is largely exogenous, reflecting
the government’s willingness to impose tax. In these models, competitive
equilibrium can be maintained, because the knowledge produced either
through learning by doing or through public research enters into the public
domain and therefore receives no compensation (Romer, 1986a, 1990b).
In Romer (1986a), the production function that consisting of firm’s own
technological knowledge and a stock of economy-wide knowledge as argu-
ments exhibits increasing returns to scale at the aggregate level. The econ-
omy wide stock is formed from the spillovers from individual firms’ knowledge
stock. Existence of externalities in innovation makes the aggregate produc-
tion function exhibit increasing returns to scale, and this is an important
novelty of endogenous growth models. The presence of externalities means
that if one firm (say) doubles its inputs, inputs of all other firms also increase,
resulting in more than proportionate increase in the aggregate output. Romer
4
These features of knowledge and their implications shall be discussed in detail below.
5
In Arrow (1962a), knowledge is generated in the private sector through learning by
doing in the production process and considered as an increasing function of the accumulated
physical capital investment in the economy.
CHAPTER 12. EVOLUTION OF GROWTH THEORY 114

(1986a) shows that if spillovers are strong enough, private marginal product
of (physical or human) capital can remain permanently above the consumers’
discount rate and thereby sustain the capital accumulation. This model has
no explicit micro economic foundation for the production of knowledge (in a
separate research sector) itself and hence, does not address the question of
market structure and market price for technological change in an explicit
way (Verspagen, 1992).
Later developments in the endogenous growth literature incorporated a
more detailed theory of technological progress into the growth models. In
Romer (1990b), Grossman and Helpman (1991a) and Aghion and Howitt
(1992), growth is driven by technological progress generated through inten-
tional investment by the profit-seeking firms in R&D and spillovers from it.
Research is treated as an ordinary economic activity that requires input of
resources and responds to profit opportunities6 .
In these models, spillover of technological knowledge plays an important
role in determining the growth rate of economies. Spillovers arise from the
two important characteristics of knowledge as a commodity, namely non-
rivalry and partial excludability7 . A non-rival good has the property that its
use by one firm or person in no way limits its use by another. Technological
knowledge is a non-rival commodity in that same knowledge can be used in
different applications as well as in different locations at the same time. A good
is excludable if the owner can prevent others from using it and therefore,
excludability is a function of both technology and legal system. In this
literature, knowledge is considered as a partially excludable commodity in
that inventions usually generate two types of knowledge, namely knowledge
that are specific to the product or process invented (specific information)
and knowledge that are more general (general information). The first type
of knowledge can be the knowledge specific to the newly invented product
and the second type consists of general scientific principles or theories on
which the new product is based. The originators have control over the use
of first type of knowledge through instruments like patenting. The second
type has wider application and over the use of which the originators have no
control. The originators, therefore, have difficulty in extracting payments
6
Although the beginning of the endogenous growth literature as a topic, irrespective of its
conceptual or intellectual legacy, is only dating back, in acknowledged published form, to
Romer (1986a), the growth of this literature is tremendous after that, making it difficult
to cover the entire literature. Fine (2000, p.246) observes, “Over the past three years, the
number of articles explicitly drawing upon endogenous growth theory almost certainly
borders on a thousand. Equally significant, they are spread over 50 or more journals.”
7
For details on these characteristics of knowledge and their implications see Romer
(1990b).
CHAPTER 12. EVOLUTION OF GROWTH THEORY 115

from others for the use of second type of knowledge (Romer, 1990b). Grossman
and Helpman (1991a, p.16-17) brings out these distinctions more clearly as
follows,
It may be useful to distinguish between two types of outputs
that are (jointly) produced in the industrial research laboratory.
Commercial research generates both specific technical information,
which allows a firm to manufacture a particular product or engage
in particular production process, and more general information
with wider applicability. Firms may be able to keep secret the
detailed information concerning product attributes and production
techniques. And even if they cannot, the applicable patent laws
can be relied upon to prevent others from copying specific product
designs or unique processes. So product specific information may
be an excludable commodity in many cases. General information
is much less likely to be so, both because it is harder to prevent
the spread of universal principles and because it is more difficult
to invoke existing legal structures to enforce proprietorship over
such information.
In addition to this, Grossman and Helpman (1991a), Aghion and Howitt
(1992) highlight another form of externality when innovators, who bring
out successive generations of similar products and each begins where its
predecessors left off, make use of the whole knowledge contained in the
previous generation of products. For example, a new entrant into the per-
sonal computer industry seeking to improve upon the state of the art need
not make its own progression from the abacus to the analogue computer to
the digital computer to the personal computer. Instead, it can inspect the
latest generation of products available on the market and extract much of
the accumulated knowledge embodied in them (Grossman and Helpman,
1994). Besides generating spillovers, another implication of non-rival char-
acter of knowledge relevant to growth theory is that non-rival good can be
accumulated without bound on a per capita basis, whereas a piece of human
capital such as ability to add cannot. Each person has only a finite number
of years that can be spent to acquire skills and these skills are lost when the
person dies. On the other hand, any non-rival good that a person produces,
for instance a scientific law, lives on after the person goes.
The R&D based growth models make a distinction between research sector
and other sectors of the economy. The returns to R&D, in these models, come
in the form of monopoly rents in imperfectly competitive product markets.
This literature adopts two approaches to product innovation depending upon
whether the innovative product bears a vertical or a horizontal relationship
CHAPTER 12. EVOLUTION OF GROWTH THEORY 116

with existing products. The innovative product would bear a horizontal rela-
tion if it serves new functions and thereby increases the variety of products
available (increasing product variety). It would have a vertical relation if it
performs similar functions of existing products, but provides greater qual-
ity8 (rising product quality). The rising product quality approach can also be
interpreted as describing a series of process innovations. With this interpre-
tation, each technological breakthrough reduces the costs in some product
line. Process innovation and product innovation are similar here because
each represents a means by which producers can provide greater “services”
at a given cost. The innovation directed towards increasing product quality
has a distinct Schumpeterian flavour in as much as successful innovators
displace extant industry leaders. The results of Grossman and Helpman
(1991a) show that many details of the equilibrium growth path do not depend
on the form of innovation that drives growth.
The inputs into the research activity are human capital and the existing
stock of scientific knowledge. Larger the already available stock of knowledge,
to which researchers have access through spillovers, higher the productivity
of human capital in the research sector. The increase in the general stock
of scientific knowledge, thus, reduces the cost of innovation and thereby
maintains the private incentive to invest in R&D in the face of declining
returns to marginal innovation. In other words, the non-appropriable benefits
from R&D keep the state of knowledge moving forward and thereby sustain
the economic growth (Grossman and Helpman, 1991a). Due to the inherent
uncertainty of innovation process, growth is uneven and stochastic at micro
level. Firms continually race to bring out next generation of products and
there may be long periods without success in some industries. Meanwhile
other industries may experience research breakthrough and aggregation
mask this micro-level turbulence and the macro economy grows at a steady
pace (Romer, 1990a; Grossman and Helpman, 1994).
Grossman and Helpman (1991a) uses production function specified by
Ethier (1982) in which final good is produced from a variety of intermediate
inputs and productivity in the final good production sector increases with the
number of intermediate commodities9 . The R&D directed towards increasing
the variety of intermediate inputs, therefore, can enhance the productivity of
the final goods production. This literature has adopted a general equilibrium
perspective and the existence of externalities and cumulative nature of inno-
vation make initial conditions important in determining the final outcome.
Further, the presence of externalities and consequent market failure also
8
See Grossman and Helpman (1991a, ch.3 and 4)
9
See Grossman and Helpman (1991a, p.47)
CHAPTER 12. EVOLUTION OF GROWTH THEORY 117

create room for government policy.

The R&D based growth literature argues that there are many reasons to
believe that investment in research can drive growth, despite its low share
as per cent of domestic product of industry even in developed countries. First,
what is generally recorded as R&D represents only a portion of the resources
that firms spend on learning to produce new goods or methods. Learning
on the shop floor–consisting of many small improvements in design and
technique–is also important in the overall picture of technological advance.
Moreover, knowledge is cumulative, with each idea building on the last,
whereas machines depreciate and must be replaced. In this sense, every
knowledge produced at the margin makes a contribution to the productivity,
while a certain portion of investment may be to replace depreciation. Finally,
social rates of return on R&D may substantially exceed the private rates of
return (Grossman and Helpman, 1994)10 .
There are also endogenous growth models focusing on other sources of
growth such as human capital formation and learning by doing. In Lucas
(1988), human capital formation, which is having positive externalities, is
the source of growth. Human capital can be accumulated either through
deliberate investment of resources to acquire skills or through production
experience. In Young (1991), the source of technological progress is knowledge
generated through learning by doing and spillovers from it.
It would be relevant to elaborate on the concept of learning by doing, as
we use models based on this idea extensively. Learning by doing means
accumulation of knowledge through production experience. This is identified
as an important source of technological progress11 . It may be conceived of
as the exploration and actualisation of the productive potential of new tech-
nologies through production experience (Young, 1991). It not only enhances
understanding of the technology and thereby increases productivity through
better control but also helps incremental innovations to take place at the
shop floor (David, 1975; Rosenberg, 1976, 1982).
Learning by doing and subsequent technological progress emerge mainly
10
Computations in Grossman and Helpman (1994) show that business R&D need not
absorb vast resources for innovation to be the engine of a reasonably rapid growth.
11
A number of studies confirmed the importance of learning by doing in increasing the
labour productivity in industries such as airframe production, shipbuilding, machine tools
and textiles. These studies include Hirsch (1956), Alchian (1963), Rapping (1965) and David
(1970). Hollander (1965, cited in Freeman(1994)p.474) in his study of technical change in
Du Pont’s rayon plants and Townsend (1976 cited in Freeman 1994, p.474) in his study of
technical change in British coal industry found that majority of the innovations did not
come from formal R&D. Instead, most of the hundreds of the small improvements to the
equipment and the organisation of work came from production engineers, systems engineers,
technicians, managers, maintenance personal and from production workers.
CHAPTER 12. EVOLUTION OF GROWTH THEORY 118

due to three features of technology, especially of a new complex technol-

ogy. First, labour cannot effectively be trained prior to the operation of new
technology; workers get more effective training through experience in pro-
duction. Second, technologies are typically systems of elements that can be
integrated in various ways and achieving proper integration requires exper-
imentation that in turn is an art based on experience. Third, technologies
are circumstantially sensitive and much of the requisite knowledge about
local circumstances and how technology responds to them while using can
only be acquired through experimentation. All these imply that learning by
doing from a given technology would disappear after some time when there
is nothing to learn or improve upon and its continuation depends on some
breakthrough innovation. Further, the intensity and productivity enhancing
effect of learning by doing also depend on other factors, especially on the em-
ployment of skilled workers, who can understand and modify the technology
(Teitel, 1984).
The endogenous growth literature is based on the view that innovation
and the consequent technological progress do not fall like manna from heaven,
but it is a social process. The intensity and direction of people’s innovative
activities are conditioned by the laws, institutions, customs, and regulations
that affect their incentive and ability to appropriate rents from newly created
knowledge, to learn from each others’ experience, to organise and finance
R&D, to pursue scientific careers, to enter markets currently dominated by
powerful incumbents, to accept working with new technologies, and so forth.
Clearly, these factors are also influenced by the technological progress. It is,
thus, viewed that economic growth involves a two-way interaction between
technology and economic life; technological progress transforms the very
economic system that creates it. The endogenous growth theory seeks a
better understanding of this interplay between technical change and various
structural characteristics of the economy and society and how such interplay
results in economic growth (Aghion and Howitt, 2003).
This literature incorporates many features of technical progress at the
micro level, like intentional investment in R&D by profit seeking firms, exis-
tence of externalities in the production and diffusion of innovation, learning
by doing and the role of imperfect market structure in inducing innovation.
It uses these ideas to explain what was left unexplained in the old growth
literature, namely technological progress, to account for the observed pattern
of growth of nations12 . Although, these ideas are not new to the develop-
12
In this context it is relevant to see Prescott (1997) which shows that differences in
physical and intangible capital cannot account for the observed differences in per capita
income across countries, instead differences in total factor productivity (TFP) can do. There-
CHAPTER 12. EVOLUTION OF GROWTH THEORY 119

ment economists, the specific contributions of this literature could be the

followings13 . The originality lies more in bringing these ideas to the fore and
packaging them in the most advanced form of mathematical models (Fine,
2000). The formal treatment made these ideas more clear as well as brought
out the implicit assumptions in these arguments. To quote Bardhan (1995,
p.2985),
A much more substantive contribution of the new growth the-
ory is to formalise endogenous technical progress in terms of a
tractable imperfect competition framework in which temporary
monopoly power acts as a motivating force for private innovations.
Another contribution is that this literature provides a useful framework
for analysing the effect of various factors like international trade, product
market competition and government policy on economic growth. As we shall
see below, the extension of these models to include international movement
of goods, capital and ideas yielded a theoretical framework that is rich in
predictions and consistent with a host of observed phenomena (Grossman
and Helpman, 1994). To quote Bardhan (1995, p.2986) again “the major
impact of this literature on development theory has been in the area of trade
and technological diffusion in an international economy”.

12.2 Trade and Technological Progress

The important features of endogenous growth theory such as imperfect com-
petition, presence of externalities and increasing returns are also present
in the new trade theory, making the integration of trade and growth more
easy and natural (Fine, 2000). The incorporation of trade into this literature
revealed a number of ways through which it can affect technological progress
of the domestic industry14 . This result is different from the neoclassical
growth models, where trade has no growth effects and has only level effects15 .
fore, it concludes that to understand the income differences we need a theory of total factor
productivity (TFP).
13
Most of the ideas such as increasing returns, learning by doing were also stressed by
the development economists of 1940s and 1950s; on this see Krugman (1993) and Thirlwall
(2002).
14
In this context, it should be noted that there exists another set of literature analysing
the effect of technological progress on the pattern of trade. For a review of this literature see
Grossman and Helpman (1995).
15
In the neoclassical model, trade openness generates gains through better resource
allocation, improved X-efficiency, etc. These are short run gains and therefore, exhaust after
some time, placing the economy at higher level of income after trade openness (Young, 1991;
Edwards, 1993).
CHAPTER 12. EVOLUTION OF GROWTH THEORY 120

Theoretical models examining the effect of trade on technological progress

within the endogenous growth framework include Grossman and Helpman
(1991a), Grossman and Helpman (1991b), Young (1991), Rivera-Batiz and
Romer (1991), Smulders and Klundert (1995), Ben-David and Loewy (1998),
Ben-David and Loewy (2000) and Traca (2002). In these models, trade can
affect technological progress by: (a) changing the structure or specialisation
pattern of the domestic industry, (b) facilitating spillover of technology from
the trade partner countries and (c) affecting firms’ R&D investment. These
channels are explained below in the context of manufacturing industry, where
one can expect their presence with greater probability.

Trade-induced structural change and technological progress By the

structure of the manufacturing industry, we mean its composition. Trade
usually expands those sectors of the manufacturing industry where the
country has comparative advantage by reallocating resources away from
other sectors. This trade-induced expansion of some sectors and contraction
of others have implications for technological progress of the manufacturing
industry. These implications derive from the fact that various sectors differ
in their potential to generate technological progress through investment in
R&D, learning by doing and so on (Grossman and Helpman, 1991a; Young,
1991; Lucas, 1988, 1993). If the trade-induced resource allocation is from
sectors having more potential for learning to those having less potential, the
industry as a whole would experience a lower rate of technological progress
in the long run. Grossman and Helpman (1991a), Young (1991) and Lucas
(1988), and Lucas (1993) modelled this aspect of trade, respectively focusing
on R&D, learning by doing and human capital accumulation as the source of
growth. In these models, if trade expands sectors having higher potential to
generate technological progress through any of these sources, the industry
would experience a higher growth. Obviously, which sector would expand
or contract depends on the comparative advantage of the country in various
sectors. Comparative advantage in turn depends on the relative endowment
of various factors of production, including technology and human capital.

Trade facilitated technology spillovers and technological progress

In the endogenous growth literature, it is postulated that trade is one of
the important channels facilitating spillover of technology among the trade
partner countries (Grossman and Helpman, 1991a,b; Rivera-Batiz and Romer,
1991; Ben-David and Loewy, 1998, 2000). Since the lion’s share of world
R&D is located in developed countries, developing countries are considered
as the major beneficiaries of trade-facilitated technology spillovers (Helpman,
CHAPTER 12. EVOLUTION OF GROWTH THEORY 121

Melitz, and Yeaple, 2004). Both import and export are assumed to transfer
technological knowledge. Import of capital goods can transmit benefits of new
technology from exporting to importing countries. Likewise, import of final
manufactured goods from technology leader countries to developing countries
allows the latter country producers to get familiarity with technologically
superior products. This would help them to obtain useful insights to improve
their products. Similarly, export gives a chance for the developing country
firms to interact with their foreign buyers and learn about new ways to
improve the products and production process. Due to the tacitness and
circumstantial sensitivity of most of the technology, it is, however, argued
that certain level of technological capability is essential for firms to efficiently
absorb technology spillovers (Cohen and Levinthal, 1989). Further, it is
also pointed out that technological spillovers might be confined to low and
medium technology industries, where tacitness and complexity of technology
is lower.

Trade and R&D Investment Trade can affect firms’ R&D investment
through several ways and these include import competition, export, tech-
nology import and trade related technology spillovers. Formal theoretical
analysis of the direction of the effect of import competition, however, is not
clear and is sensitive to modelling assumptions on domestic market structure,
cost structure of firms, ease of entry and exit in the domestic industry and so
on. Export can encourage innovation efforts by allowing firms to produce on
a large scale and thereby to exploit increasing returns to scale made possible
by fixed investment like R&D. Hughes (1986) argues that export will have a
positive effect on R&D because elasticity of export demand with respect to
R&D is likely to be greater than that of the domestic demand. Technology
import - both embodied and disembodied - can affect innovation effort of the
firm. This is one of highly debated issues in the literature on the technological
progress of the developing countries (Evenson and Westphal, 1995). Whether
technology import encourages or discourages innovative effort is not clear
and depends on many factors.

12.3 Neutrality of Technological Progress

In a world of diminishing returns to factors, technological progress is consid-
ered to be the driver of long run economic growth16 . We define technology as
the ‘social pool of knowledge of the industrial arts’ and technological progress
16
This note is largely based on the material covered in the chapter 7 of Jones (1976).
CHAPTER 12. EVOLUTION OF GROWTH THEORY 122

refers to the increase in the stock of this knowledge. Application of the im-
proved knowledge in the production process bring about technical progress,
which results in either of the following.

1. More output is produced using the same quantities of inputs, or for

producing the same quantity of output, less quantities of inputs are
used.

2. Existing outputs undergo qualitative improvement; or

3. Totally new products are produced.

An invention which leads to a new technique for producing an existing

good is referred to as a process invention, whereas an invention which changes
the form of the existing goods or generates totally new goods is referred to as
a product invention. Once an invention is actually put into practice in the
economy a product or process innovation is said to have occurred.
Attempts at the classification of technical progress have, on the whole,
stemmed from an interest in its effect upon the distribution of income be-
tween capital and labour. A number of possible classifications have been
proposed but modern interest was revived by the realisation that most forms
of technical progress appeared to be inconsistent with the conceptions of
steady-state growth employed in the analysis of many simple growth models.

12.4 Representation of Technical Progress

In a one-good model of growth, the only possible effect of technological
progress is to allow more of that commodity to be produced given the in-
puts of capital and labour. To quote Salter:

The common characteristics of all such (technological) advances

is that they lead to a new production function which is superior
to its predecessor in the sense that less of one or more factors of
production is required to produce a given output (Salter, 1966,
p.21).

Thus in terms of our per-worker production function, technical progress

implies that the production function shifts upwards as in Figure 12.1
In the Figure, the production function prior to the technological progress
is represented by the curve f pk, t0 q. Following technical progress, the curve
shifts upwards to the new position f pk, t1 q such that at each level of the
CHAPTER 12. EVOLUTION OF GROWTH THEORY 123

y D
f(k,t1)
B

f(k,t0)
E
A

O k

Figure 12.1: Representation of Technological Progress

f(k)

O k

Figure 12.2:

capital-labour ratio (except zero) more output per worker can be produced
than previously.
The situation portrayed in Figure 12.1 is the conventional textbook presen-
tation of technical change. Atkinson and Stiglitz (1969), however, emphasised
that there was no reason to suppose that the whole curve would be shifted
upwards by technical progress. They argued that the whole curve would not
be shifted upwards by technical progress. They point out that the basic idea
underlying the smooth neoclassical per-worker production function is that
there exists a large number of different processes of production which can be
approximated by a smooth curve. They argue that technical progress in any
one of the separate processes of production need not affect any of the other
processes and that, as a consequence, the effect of technical change would
be to produce a ‘bulge’ in the per-worker production function rather than to
shift the whole curve. Their argument is illustrated in Figure 12.2.
The most general method of representing the effect of technological
progress in a model of economic growth involves rewriting the aggregate
CHAPTER 12. EVOLUTION OF GROWTH THEORY 124

production function as:

Y “ F pK, L, tq (12.1)
The production function (12.1) includes, apart from K and L, a time
variable t, representing the evolution of technology. This implies that output
produced by a fixed combination of capital and labour increasing in time, an
indirect representation of technological progress.
The per-worker production function can be written as

y “ f pk, tq (12.2)
Although equations (12.1) and (12.2) constitute the most general forms of
the aggregate production function in the presence of technological progress,
a different formulation is widely used in the literature. In this approach,
technical progress is said to be factor-augmenting. Technical progress shifts
the production function such that more output is produced even though the
stock of capital and the labour force may not have increased. It is as if the
factors of production had been augmented. In this formulation the aggregate
production function is written as

Y “ F pAptqK, BptqLq (12.3)

In equation (12.3) output, Y, is no longer a simple function of the quantities
of capital and labour. The stock of capital,K, and labour, L, are multiplied
by factors A and B which are both functions of time. The expressions AptqK
and BptqL are usually referred to as effective capital and effective labour
9
respectively. if Aptq is positive, then as time goes on, the effective capital
stock increases even though the actual capital stock may have remained
9
constant. Similarly, if Bptq is positive, then the effective labour force is
increasing even if the actual number of labour force is constant. In concrete
terms, this form of factor augmenting technical progress implies, for example
that ten persons can do the amount of work previously performed by twelve
and/or five machines can produce output that previously required six.
9 is positive and Bptq “ 1 then technical change is said to be purely
If Aptq
capital-augmenting. Conversely, if Aptq “ 1 and Bptq9 is positive then technical
change is said to be purely labour-augmenting. Finally, if Aptq 9 9
“ Bptq ą0
then technical change is said to be equally capital and labour augmenting.
It is crucial to note that the factor-augmenting representation of technical
change does not imply anything about causation or the source of technical
improvement. If, for example, technical progress can be represented as being
purely labour augmenting, we cannot infer that there has been an intrinsic
CHAPTER 12. EVOLUTION OF GROWTH THEORY 125

change in the quality of labour force. In Solow’s worlds: ‘it could in fact be
an improvement in the design of the typewriter that gives one secretary the
strength of 1.04 secretaries after a year has gone by’

12.5 Classification of Technological Progress

The interest in classifying technical progress as labour-saving, neutral or
capital-saving stems from a concern for its effect upon the distribution of
income. Three main classificatory schemes, due to Hicks (1963), Harrod
(1948), Solow (1959) exist in the literature and they all can be interpreted in
terms of the effect of technical progress upon the relative shares of capital
and labour.

12.5.1 Hicks’ Classification

We can classify inventions accordingly as their initial effects are
to increase, leave unchanged or diminish the ratio of the marginal
product of capital to that of labour. We may call these inventions
“labour-saving”, “neutral” and “capital-saving” respectively (Hicks,
1963, p.121)

Labour Saving Technical Progress: An upward shift, representing a

technical progress, in the per-worker production function is said to be labour
saving, if at any constant value of the capital-labour ratio, the ratio of the
marginal product of capital to the marginal product of labour has increased.

Capital Saving Technical Progress: An upward shift, representing a

technical progress, in the per-worker production function is said to be capital
saving, if at any constant value of the capital-labour ratio, the ratio of the
marginal product of capital to the marginal product of labour has decreased.

Hicks-neutral Technical Progress: An upward shift, representing a

technical progress, in the per-worker production function is said to be Hicks-
neutral, if at any constant value of the capital-labour ratio, the ratio of
the marginal product of capital to the marginal product of labour remains
constant.
Assume that capital-labour ratio K L
remains constant. If we denote the
marginal product of capital and labour before the onset of technical progress
CHAPTER 12. EVOLUTION OF GROWTH THEORY 126

as FK p0q and FL p0q respectively, and the same marginal product after techni-
cal progress as FK ptq and FL ptq, then Hicks definition can be summarised as
follows.
FK ptq FK p0q
1. Technical progress is labour-saving if FL ptq
ą FL p0q
.
FK ptq FK p0q
2. Technical progress is Hicks-neutral if FL ptq
“ FL p0q
.
FK ptq FK p0q
3. Technical progress is capital-saving if FL ptq
ă FL p0q
.

Assuming a competitive market and firms’ optimisation imply that marginal

product of capital is equal to the rental rate on capital and the marginal
product of labour is equal to the wage rate. Hence a labour saving innovation
increases the ratio r{w, while capital-saving innovation decreases r{w. As
we will see this has important implications on the relative shares of capital
and labour in the national income and also allows us to interpret Hicks’
classification of innovation in terms of relative shares of labour and capital in
the national income. If we define the relative shares of capital and labour in
r.K
the national income as Π “ w.L , then given the K{L, a labour-saving saving
innovation would increase the relative share of capital in the national income,
a neutral technical change would leave Π unchanged and a capital saving
technical progress would reduce Π. Hence Hicks’ classification of technologi-
cal progress can also be stated in terms of its impact on the relative shares of
capital and labour in the national income.

Labour-saving Technical Change: Technical progress is said to be labour-

saving in Hicks sense if at any constant value of capital labour ratio, the
9 ą 0).
ratio of capital share to that of labour, Π “ r.K{w.L, is increasing (i.e. Π

Capital-saving Technical Change: Technical progress is said to be capital-

saving in Hicks’ sense if at any constant value of capital-labour ratio, the
9 ă 0).
ratio of capital share to that of labour, Π “ r.K{w.L, is decreasing, (i.e. Π

Neutral Technical Progress: Technical Progress is said to be neutral in

Hicks’ sense if at any constnat value of capital-labour ratio, the ratio of share
of capital to that of labour, Π “ r.K{w.L, remains constant (Π 9 “ 0).
Hicks’ neutral technical progress can be illustrated using the Figure
12.3. Assume that, the production function prior to technological progress,
f pk, t0 q, and f pk, t1 q denotes the the production function after the technological
progress. The economy has a capital labour ratio of k ˚ . Given the capital-
labour ratio, the marginal product of capital in the production function
CHAPTER 12. EVOLUTION OF GROWTH THEORY 127

f pk, t0 q is measured by the slope of the tangent RA and the wage-rental ratio,
z “ w{r is equals to the distance OR (see Appendix Section A for a proof
of this statement). If technical progress causes the per worker production
function shifts upward to f pk, t1 q, then Hicks’ neutrality requires that, at k ˚
the ratio of marginal product of capital to the marginal product of labour or
the ratio rental rates on capital to the wage rate (r{w) must remain constant.
Thus a Hicks’ neutral shift from f pk, t0 q to f pk, t1 q requires that the tangent to
the new production function at the capital-labour ratio k ˚ must originate from
the point R, such that the distance ORp“ w{rq remains the same after the
shift. These conditions are seen to be satisfied in the Figure 12.3 and the shift
from C to D in the production function therefore represents Hicks-neutral
technical progress.

y
B
f(k,t1)

D A

f(k,t0)
C

R O k* k
Figure 12.3: Hicks’ Neutral Technical Change

It can be proved that Hicks’ neutral technical progress is exactly equiva-

lent to the idea of equally capital and labour augmenting technical progress.
Therefore if technical progress is happening at the rate of g, the production
function can be written as
Y “ egt F pK, Lq
Hicks’ classification of technical progress was devised specifically within the
context of a theory of wages and distribution. It is not, however, particularly
useful in the context of growth models that have been discussing.
As any upward shift in the per-worker production function will always
implies a new stable balanced growth path involving a higher capital-labour
CHAPTER 12. EVOLUTION OF GROWTH THEORY 128

y
(n+d)k

y** f(k,t1)

f(k,t0)
y*

O k* k** k

Figure 12.4:

ratio and level of output per labourer (see Figure 12.4). Hicks’ classification
of technical progress is restricted to the comparison of points involving a
constant capital-labour ratio and will therefore not useful in the context of
growth models having steady-state. Harrod’s alternative classification of
technical progress is devised for use in models of a steadily growing economy.

12.5.2 Harrod’s Classification

Harrod (1948, 1937) introduced his classification of technical progress. Ac-
cording to him

‘I define a neutral technical advance as one which, at a constant

rate of interest, does not disturb the value of the capital coeffi-
cient’, and ‘A stream of inventions, which are neutral as defined,
will, provided that the rate of interest is unchanged, leave the
distribution of the total national product as between labour (in the
broadest sense) and capital unchanged’(Harrod, 1948, pp.23)

The capital coefficient is defined as ‘the ratio of the value of capital in use
to income per period’(Harrod, 1948, pp.22), i.e. the capital-output ratio. In
competitive conditions and assuming absence of risk, the rate of interest is
equal to the marginal product of capital. Also notes that Harrod compares
points at which the capital-output ratio is constant as opposed to Hicks
procedure which compares points at which capital-labour ratio is constant.

Harrod’s Neutral: An upward shift representing technical progress, in

the per-worker production function, is said to be Harrod-neutral if, at any
CHAPTER 12. EVOLUTION OF GROWTH THEORY 129

constant value of the capital-output ratio, the marginal product of capital

remains unchanged.
Similar to Hicks’ classification of technological progress, Harrod’s classi-
fication can also be expressed in terms of the relative shares of capital and
labour (Π “ r.K{w.Lq in the national income.

Harrod’s Labour-saving Technological Progress: Technical progress is

said to be labour saving in Harrod’s sense if at any constant value of capital-
output ratio,(K{Y ) the ratio of relative shares, Π “ r.K{w.L, is increasing.

Harrod’s Capital-saving Technological progress: Technical progress is

said to be capital saving in Harrod’s sense if at any constant value of capital-
output ratio,(K{Y ) the ratio of relative shares, Π “ r.K{w.L, is decreasing.

Harrod’s Neutral Technical Progress: Technical progress is said to be

Harrod-neutral if, at any constant value of the capital-output ratio, the ratio
of relative shares, Π “ r.K{w.L, remains constant, (i.e. Π 9 “ 0 when K{Y is
constant).
Both Hicks’ and Harrod method of classifying technical progress can be
reduced to a study of the effect of upon the relative shares of capital and
labour resulting from the shift in the aggregate production function. But their
classification differ in which points on the old and new production functions
are compared – Harrod’s method being applicable to the comparison of points
with equal capital–output ratios, while Hicks’ system is applicable to the
comparison of points with equal capital–labour ratios.
A Harrod-neutral shift in the per-worker production function can be
illustrated using the Figure 12.5.

y Z
M'

D f(k,t1)
y**
M
M'
y* f(k,t0)
B
M

O k* k** k

Figure 12.5: Harrod-neutral Technical Change

CHAPTER 12. EVOLUTION OF GROWTH THEORY 130

In Figure 12.5, f pk, t0 q represents the conventional per-worker production

function prior to the onset of the technical progress. If the economy is initially
operating at a capital-labour ratio equal to k ˚ , the capital-output ratio is
equal to the inverse of the slope of the line OBD (slope of OBD “ Bk ˚ {Ok ˚ “
y{k “ Y {Kq and marginal product of capital is equal to the slope of the
tangent M M at the point B on f pk, t0 q. Technical progress now shifts the per
worker production function to the new position f pk, t1 q. If the shift is to be
Harrod-neutral two conditions must be satisfied:

1. The marginal product of capital must remain the same as that given by
the slope of the tangent M M , and
2. The capital-output ratio must be equal to that determined by the slope
of the line OBZ.

These conditions seen to be satisfied in the movement form the point B,

with capital-labour ratio of k ˚ , to the point D, with capital-labour ratio of k ˚˚ .
It can be shown that Harrod-neutral technical progress is equivalent to
what we call a purely labour augmenting technical progress (see: Robinson,
1938; Uzawa, 1961; Sato and Beckmann, 1968). Thus if the technical progress
is Harrod-neutral and increasing at the constnat proportional rate g, then
aggregate production function may be written as

Y “ F pK, BptqLq
where
9
Bptq
“g
Bptq
and utilising the cocept of continuous growth

Y “ F pK, egt Lq
Harrod’s neutral technical progress because of its equivalence with an in-
crease in the labour force, is particulalry useful to incorporate in the models
of growth.

12.5.3 Solow’s Classification:

Solow neutral: An innovation is Solow neutral if the relative input shares
L.FL
K.FK
, remain unchanged for a given labour/output ratio. This definition can
be shown to imply a production function of the form

Y “ F pK.Bptq, Lq
CHAPTER 12. EVOLUTION OF GROWTH THEORY 131

where Bptq is the index of the technology, T9 ptq ě 0. Production functions

of this form is called capital augmenting because technological progress
increases production in the same way as an increase in the stock of capital
(see: Barro and Sala-I-Martin, 2004, p.52).

12.5.4 Some properties of Output per worker production

function
Let the production function takes the following form

y “ f pkq (12.4)
where y is the output per worker (y “ YL ) and k is the capital per worker
(k “ K
L
). The figure 12.6 presents the production function and some of its
properties. Now consider the point A, the slope of the line CA equals the
marginal product of capital at point A of the production function f pkq. If
marginal productivity theory of distribution is accepted, the slope equals the
rental rate, r. The slope of the line CA is given by CD{DA.
CD
slope of the tangent CA “ r “
DA
As DA equals OE, which is the capital labour ratio, k˚ associated with
the tangent CA,
CD CD
r“ “
OE k˚
or CD “ r ˆ k˚
Now r ˆ k˚ is the rental rate multiplied by the amount of capital per labourer.
Therefore, the distance CD measures the amount of rental income per
labourer. Since OD measures the total amount of output per labourer, on
the basis of constant returns to scale and the consequent Euler’s theorem of
product exhaustion, wages per labourer or wage rate is given by

Wages per Worker “ OD ´ CD “ OC

The slope of the tangent CA can be written in an alternative way
OC
Slope of CA “ r “
OB
As OC measures the wage rate, w
w
r“
OB
CHAPTER 12. EVOLUTION OF GROWTH THEORY 132

y = f(k)
y* D A

C
E
B O k* k
Figure 12.6: Per worker Production Function (y “ f pkq)

or,
w
OB “
r
Summarising the results associated with the use of per worker continuous
production function in Figure 12.6:

1. Rental per worker are measured by the distance CD

2. Wages per worker are measured by the distance OC
w
3. The ratio wages to rental rate, r
, is measured by the distance OB.
Chapter 13

Histry vs Expectations

“it is sometimes not possible to uncover the logic (or illogic) of the world
around us except by understanding how it got that way” (David, 1985, p.332).

A casual reading of economic development across countries reveals that

they differ enormously in terms of income per person, some are very rich
and some are extremely poor. Growth rates of income also differ greatly
across countries, suggesting persistence, if not worsening, of current income
differences. Some countries are growing faster and thereby catching up with
the rich countries, and some others are experiencing either stagnation or
decline in their per capita income for a long time. The crucial question is
why do countries differ in their per capita income and growth rate. The
neoclassical growth models implies that automatic mechanism will take an
underdeveloped economy out of poverty. In this model an economy is poor due
to lack of capital, implying that in poor economies rate of return to capital is
high, generating strong incentives to foreign investment and domestic capital
formation. However, a casual look at the experience of many underdeveloped
countries since World War II reveals that both foreign direct investment and
domestic capital formation as a share of production is very low and there is
no automatic mechanism(Rodríguez-Clare, 1996). If technological progress is
the source of long run growth, why does technological progress differ across
societies. People all over the world are similar, having similar physical and
mental configuration and same set of hopes and desires. If this is the case,
why are some societies more innovative and productive than some others,
and why do same rate of investment generate different growth rate across
countries. In this note, we examine the role of history and expectations
people hold of their own future in shaping the economic status of a society1 .
1
There are other explanations for the lack of income convergence among countries. These
include human capital by Barro, Mankiw, and Sala-I-Martin (1995), externalities by Lucas Jr

133
CHAPTER 13. HISTRY VS EXPECTATIONS 134

History and expectations influence the economic status of society if there are
complementarities and increasing returns in production. Complementarities
and increasing returns are closely related to the concept of externalities. A
clear understanding of these concepts, therefore, requires a firm grip on
the concept of externalities. Hence, next section discusses the concept of
externality in detail.

13.1 Externalities
We say that an economic situation involves a Consumption externality if
one consumer cares directly about another agent’s production or consumption.
Similarly a production externality arises when the production possibilities
of one firm are influenced by the choices of another firm or consumer (Varian,
1999). There are four types of externalities. (1) An individual’s satisfaction
may depend not only on the quantities of products the person consumes and
services he renders but also on the satisfaction of other individuals. (2) A
person’s satisfaction may be influenced by the activities of producers not only
thorough producers’ demand for the individual’s services and supply of the
products the individual buys but also in ways that do not operate through
the market mechanism; examples include noises or pollution emanated from
the nearby factory. (3) Producers’ output may be influenced by the actions
of persons in ways other than through their offer of services to production
and their demand for firm’s products. This is a counter part of the previous
case and one example is firms benefiting from inventions by individuals. (4)
The output of the individual producer may depend not only on his input of
productive resources, but also on the activities of other firms.
The last one, i.e. externalities in the production sphere, is called external
economies–with which we are concerned in this chapter–essentially mean ser-
vices (and dis-services) rendered free (without compensation) by one producer
to another2 . It is agreed that external economies are a cause for divergence
between private benefit and social benefit and thus for the failure of perfect
competition to lead to an optimum situation. External economies or direct
interdependence among producers exists whenever the output (x1 ) of a firm
depends not only on the factors of production, (l1 , c1 , ...) utilised by it but also
on the output (x2 ) and factor utilisation (l2 , c2 , ..) of other firms. In symbols,

x1 “ F pl1 , c1 , ...; x2 , l2 , ...q

(1990), economic policy by Parente and Prescott (1994); Sachs and Warner (1995).
2
A fine example is the bee keeper benefiting from the nearby apple farm and the apple
farm benefiting from the bee keeping farm through better pollination.
CHAPTER 13. HISTRY VS EXPECTATIONS 135

Externalities and Property Right

The crucial feature of externalities is that these are goods people care
about but lacks market because property rights are not well defined.
Suppose A and B are staying in the same room. There is a problem if
A believes that he has the right to smoke and B also believes that he
has the right to have clean air. This problem can be solved by assigning
property rights either to A-the right to smoke-or to B-the right to have
clean air. A well defined property right enables people to trade their
rights to produce externalities in the same way they trade rights to
produce and consume ordinary goods. In this type of solution, the amount
of externality produced depends on the assignment of property rights,
i.e. whether the smoker or non-smoker has the righta . Generally, it is
assumed that all the interactions between consumers and producers take
place thorough the market. So they need to care only the market prices
and their production and consumption possibilities. In such a context
market mechanism is capable of achieving pareto optimalityb . However,
when externalities are present market mechanism fails to achieve pareto
efficiency.
a
When the preferences are quasi-linear,i.e. the demand for goods causing externality
does not depend on the distribution of income, efficient amount of externalities is
invariant to assignment of property right (see: Coase, 1960)
b
Pareto optimality is an equilibrium condition where one agent can improve his
welfare only by reducing some other’s welfare

In the literature the concept of external economies has been used in entirely
two different contexts; (1) in equilibrium theory, and (2) in the theory of
industrialisation in underdeveloped countries (see: Scitovsky, 1954). The
concept stands for very different things in two contexts. In the case of
underdeveloped economies the concept of external economies are invoked
whenever the profits of one producer are affected by the actions of other
producers. In symbols,

Π1 “ Gpx1 , l1 , c1 , ...; x2 , l2 , c2 , ...q

Which shows that profit of firm one depends not only on its own output and in-
puts, but also on those of other firms. The presence of variables to the right of
the semicolon suggests the existence of external economies. This definition of
external economies obviously includes direct or non-market interdependence
among producers as discussed above. However, it is much broader than the
concept of direct interdependence, as in addition to direct interdependence, it
CHAPTER 13. HISTRY VS EXPECTATIONS 136

also includes interdependence among producers through market. This kind

of interdependence may be called pecuniary externalities, to distinguish it
from the technological external economies of direct interdependence. An
example of pecuniary externalities can be given as follows. Expansion of an
industry producing intermediate inputs can reduce the price of its output
and raises the prices of its inputs. Lower price of the intermediate inputs
(output of the expanding industry) benefits the user industries. Similarly,
increased factor prices benefits the suppliers of these factors. When these
benefits accrue to firms in the form higher profit, they are pecuniary external
economies. Similarly, expansion of an industry would generate more income
in the form of profit, wage, and rent. This increased income would enhance
demand for the products of other industries and thereby their profit. This
type of pecuniary externalities are also called demand spillovers (see: Murphy,
Shleifer, and Vishny, 1989). Unlike technological externalities, pecuniary
externalities does not have a place in equilibrium theory (Scitovsky, 1954).

13.2 Complementarities
Complementarity is a special case of externality in which cost of doing one
activity declines as more and more people do the same activity. Large scale
adoption makes dividing the whole process into large number of sub-processes
and undertake each of them on a large scale economically viable. This deeper
specialisation generates increasing returns at the society level. One impli-
cation of complementarities is that it makes the cost or benefit of adopting
a system by an individual depends on how many other individuals have
adopted the system. Similar to other type of externalities, complementarities
also creates divergence between individual and social gains. The concept of
complementarities can be better explained using the example of typewriter
keyboard. As all of you have noticed the top left row of your computer or
typewriter key board starts with the letters q,w,e,r,t,y,... The QWERTY ar-
rangement was evolved mainly to avoid tangles or jam among typewriter
levers carrying letter imprint at their head. This arrangement is not efficient
in terms of typing speed. A more efficient keyboard arrangement, DSK (Dvo-
rak Simplified Keyboard), was patented in 1932 by August Dvorak and W.L.
Dealey. The DSK keyboard was found to be faster than QWERTY in many
typing competitions (David, 1985). However QWERTY system is still used in
today’s computer keyboards3 . To understand why the inefficient system still
3
During the 1940s, experiments by the US Navy showed that increased cost of retraining a
group of typist in DSK keyboard would be offset by its increased efficiency within the first ten
days of their subsequent employment. Apple introduced DSK in its IIC computers. However,
CHAPTER 13. HISTRY VS EXPECTATIONS 137

dominates the market, we have to see the players existed in the typing and
document preparation industry. There were mainly three agents functioned
in the market; they were (1) Typewriter manufacturing firms, (2) Typewriting
training institutes, and (3) Firms employing typists to prepare documents.
Given that all document firms are hiring QWERTY trained typists, there
is little incentive for the typewriter manufacturing firms to produce more
efficient typewriters with, say, DSK keyboards. As all of the typing schools
train in QWERTY system, there is no incentive for the firms to invest in
typewriters with alternative keyboards (this requires costly re-training of
their typist). As there is no more efficient, say DSK typewriters, there is no
incentive for the typing schools to train people in this system. The emergence
of QWERTY can be characterised as follows. Suppose buyers of typewriters
have no inherent preference for particular keyboard, but care only the dis-
tribution of typists trained across different keyboards. Suppose that typists
are also heterogeneous in their preferences for learning alternative keyboard
systems, but attentive to the distribution of the stock of machines according
to the keyboard style. Now imagine the members of this heterogeneous
population deciding in random order what kind of typing training to acquire.
It may be seen that, with unbound decreasing cost of selection, each stochas-
tic decision is in favour of QWERTY (David, 1985). From the view point
of the formal theory of stochastic processes, the emergence of QWERTY is
equivalent to a Generalised Polya Urn Scheme (see the box).

The Polya Urn Scheme

In a simple scheme of Generalised Polya Urn, an urn containing

balls of various colours is sampled with replacement, and every
drawing of a ball of a specified colour results in a second ball of the
same colour being added to the urn. The probability that balls of
a specified colour will be added is an increasing (linear) function
of the proportions in which the respective colours were presented
within the urn. Theorem by Arthur (1983); Arthur, Ermoliev, and
Kaniovski (1985, cited in David (1985)) says that when a gener-
alised form of such process (characterised by unbound increasing
returns) is extended indefinitely the proportional share of one of
the colour will, with probability of one, converge to unity.

this superior design also met the same rejection as the previous seven improvements on the
QWERTY typewriter keyboard that were patented in the US and Briton during the years
1909-24 (David, 1985).
CHAPTER 13. HISTRY VS EXPECTATIONS 138

A QWERTY

DVORAK

N
Number of people

Figure 13.1: Cost curves with complementarities in adoption

In figure 13.1, the average cost of using both DSK and QWERTY is
declining with level of adoption. Of course, the average cost of DSK is
lower than that of QWERTY at all levels of adoption, indicating its greater
efficiency. An individual, while choosing a system would compare the cost of
adopting QWERTY having an adoption level of N with that of DSK having
zero adoption level, adopt QWERTY. If the individual chooses DSK typewriter,
he has to bear a huge cost of getting it manufactured (because there might
not be any DSK typewriters already manufactured) and training people
in the new system. Because of some historical reasons, QWERTY came
into the market first, encouraging the production of QWERTY keyboard
typewriters and establishment of training schools in QWERTY keyboard.
Development of these complementary facilities reduced the cost of adoption of
QWERTY keyboard and helped to establish itself in the market. If the society
wants to move towards more efficient system, what is needed is an incredible
amount of coordination among the typewriter manufactures, typing schools
and document firms. The system has multiple equilibria, everybody using
DSK is also a stable state, provided that the society could somehow reach
there. The society can bypass the historically decided lower level equilibrium,
if agents make their choices in a forward-looking way, rather than myopically.
If all agents believe that DSK, being an efficient system, would dominate in
the market and on that basis make their choices, the final outcome would be
the efficient one, dictated by expectations of forward-looking agents. To quote
David (1985, p.335) “Intuition suggests that if choices were made in a forward-
looking way, rather than myopically on the basis of comparisons among the
currently prevailing costs of different systems, the final outcome could be
CHAPTER 13. HISTRY VS EXPECTATIONS 139

influenced strongly by expectations”. Another example, where we can see this

historical ‘locked in’ and consequent path-dependence, is computer software,
where complementaries are very strong.

13.2.1 Coordination Failure

When complementarities are present, there is no unique equilibrium; either
the economy can get locked-in a lower level equilibrium or it can settle at a
better higher level equilibrium. In an underdeveloped economy, pervasive
complementaries demand coordinated action among economic agents to take
it out of low level equilibrium trap. The genesis of this argument can be traced
to a classic paper by Rosenstein-Rodan (1943). The recent developments in
the literature use this idea and characterise under-development as lack
of coordination in the presence of widespread complementarities(Murphy,
Shleifer, and Vishny, 1989; Matsuyama, 1991; Krugman, 1991; Rodríguez-
Clare, 1996; Baland and Francois, 1996; Ciccone and Matsuyama, 1996).
The coordination failure literature views underdevelopment as an outcome
of massive coordination failure, in which several investments do not take
place because other complementary investments are not there, and the latter
investments are not made because the former investments are not made.
This argument may seem circular. This view of development can explain
why similar economies behaved differently depending upon the events in
their history. The situation is similar to the QWERTY example. The choice
of “investing” and “not investing” is analogous to adopting QWERTY (less
efficient system) or DSK. Here the crucial point is that the gain to investing
depends positively upon investments made by others.

In an underdeveloped economy complementarities and consequent need

for coordination arise because of (1) linkages, and (2) demand spillovers.
Input-output relationships among industries generates linkages. Figure 13.2
presents a simplified representation of linkages among industries. The arrow
from one industry to another suggests that expansion of one industry facili-
tate the expansion of another industry by easing the conditions of production
in the latter industry. In this case, multiple equilibria are possible, one hav-
ing all linkages depressed and another in which all the linkages are active.
Linkages can be either backward or forward linkages. In Figure 13.2 steel
industry facilitates the development of other industries, such as railways,
by increasing the supply of steel and/or lowering its price. This is a forward
linkage, one that works by easing the supply of one product and is captured
in the figure by the direction of arrows. The steel industry has backward
CHAPTER 13. HISTRY VS EXPECTATIONS 140

Coordination Vs Price signals

In a competitive market economy prices are the signalling device in-
forming each person of the peoples’ economic decisions. Market prices,
however, reflect the economic situation as it is and not as it will be. Price
signals, therefore, are more useful for coordinating current production
decisions having immediate effect and guided by short-run considera-
tions. They are not suitable for coordinating investment decisions, as
they are based on expectations about a long future period and also have
very delayed effect on production and consumption. Investment decisions
should be governed not by what the present economic situation is, but
by what the future economic situation is expected to be. The proper coor-
dination of investment decisions, therefore, would require a signalling
device to transmit information about present plans and future conditions
expected to realise through present plans. Therefore, the need for either
a centralised investment planning or some communication mechanism
in addition to the price mechanism as a signalling device is called for
(Scitovsky, 1954).

linkage with coal industry, expansion of the former raises the demand for the
latter’s product.
Demand spillovers means expansion of one sector enlarges the size of
market of other sectors. The market for the products of other sectors expand
because of increased income (in the form wage, profit, rent, and interest)
generated in the expanding sector. This is also related to the argument
that one impediment of industrialisation in underdeveloped countries is the
small size of the market. When domestic market is small and world trade
is not free and costless, firms may not be able to generate enough sales to
make adoption of increasing returns to scale technologies profitable (Murphy,
Shleifer, and Vishny, 1989).
The issue of linkages has important bearing on policy. The question,
in the context of an under developed economy, is how it can be assisted to
move to the better equilibrium. Rosenstein-Rodan (1943) introduced the idea
of big push, a policy of coordinated and simultaneous investment in many
sectors. This coordinated simultaneous investment in various sectors not
only take care of the input-output linkages among sectors, but also allows
adoption of increasing returns to scale technologies as income generated in
one sector become a source of demand for goods in other sectors(Murphy,
Shleifer, and Vishny, 1989, p.1025). This strategy, however, requires large
volume of resources for investment (mainly from the public sector), which
in normal cases is difficult for an underdeveloped economy to mobilise, even
CHAPTER 13. HISTRY VS EXPECTATIONS 141

Iron Railways

Coal Steel Exports

Mining Machinery Shipping

Consumer
Goods

Figure 13.2: Linkages and Coordination

CHAPTER 13. HISTRY VS EXPECTATIONS 142

with foreign aid. Second,informational requirement of this strategy is very

high, proper implementation requires detailed quantitative information on
sectoral linkages. Without proper information, investment in steel can be
very high and investment in coal is insufficient, resulting in the suboptimal
performance of both industries.

Importance of Domestic Marketa .

The importance of large domestic market for industrialisation when firms

can access the world market is questionable. Its importance can be well
appreciated for the development of non-tradebles like infrastructure.
When world trade is free and costless, industry faces a world market, the
size of which cannot constrain the adoption of increasing returns to scale
technologies. However, the available evidence from the industrialisation
experience of various countries shows that domestic market is important.
Chenery and Syrquin (1975); Chenery, Robinson, and Syrquin (1986) pro-
vide the most detailed evidence. Studying a number of growing countries
over a long period of time, Chenery, Robinson, and Syrquin (1986) shows
that domestic demand accounts for a larger share of increase in domestic
industrial output. In countries with population of over 20 million, expan-
sion of domestic demand accounts for 72-74% of the increase in domestic
industrial output (Chenery, Robinson, and Syrquin, 1986, p.156). In such
countries, when per capita income is between 200-800 1964 US$, the
share of industry in GNP is five to six percentage points higher than in
countries with population under 20 million, with difference concentrated
in industries with important economies of scale, such as basic metals,
paper, chemicals and rubber products (Chenery and Syrquin, 1975, p.78).
Further, it has also been found that intensive export of manufactures
began only after the industry became established in the domestic market
(Chenery and Syrquin, 1975, p.101)
a
This box is based on the discussion in Murphy, Shleifer, and Vishny (1989)

Apart from the implementation difficulties of big push strategy, it also

does not recognise the fact that the desirable state is also an equilibrium; so if
state can device proper incentives, the market forces can lead the economy to
that desirable equilibrium (Ray, 1999). The idea of using market forces along
with state policy to correct the coordination failure is the basis of Hirschman’s
(1958) unbalanced growth strategy. This strategy involves investment in
certain selected key sectors of the economy, and as the linkages generated by
these key sectors make their presence felt, the market will respond to the
CHAPTER 13. HISTRY VS EXPECTATIONS 143

unbalanced situation by making other investments available. The key sectors

can be selected on the basis of many criteria, which includes (1) number and
strength of linkages. Industries having more and strong linkages should get
priority in investment, (2) scale of investment; industries requiring large
scale of investment to begin with need to get priority as private sector may
find it difficult to mobilise capital for investment in these industries, (3)
intrinsic profitability of the sector. Highway project is highly profitable from
the point of society, but it is not profitable for the private sector as it cannot
extract the reasonable return from investment.

13.2.2 History versus expectations

We have found that presence of complementarities would lead to multiple
equilibria and history can have a role in fixing the equilibrium. Apart from
history, expectations of economic agents also can decide the equilibrium. If
economic agents believe that the economy would move to the better equi-
librium and make their choices on that basis, then the economy would end
up at the higher level equilibrium. This self-fulfilling prophecy, that takes
place when economic agents makes their choices in a forward looking way
rather than in a myopic manner, can be explained using a simple example in
Figure 13.3. Suppose there are two sectors; one traditional sector (OLD) and
a technologically sophisticated new sector (NEW). The old sector employs
unskilled labour and NEW sector skilled labour. Therefore, to enter into the
NEW sector the labourers have to invest in skill formation. In the NEW
sector increased employment leads to greater skill formation and thereby
to higher productivity through learning by doing and knowledge spillovers.
Therefore, in this sector, wage rate is an increasing function employment.
Suppose size of the labour force AB is fixed in the economy and initial level
of employment is OA in OLD sector and OB in NEW sector, indicating very
low level industrialisation. In this allocation of labour force between sectors,
wage is higher in the OLD sector. Now assume that every agent believes
that economy would soon industrialise and wage rate in the sector NEW
would go above that in OLD. If agents, on the basis of their expectations, are
considering future relative wages also while making decision on whether to
invest in skill acquisition to enter sector NEW, industrialisation would take
place. Now more and more people would enter into the sector NEW and this
would further strengthen the expectation of industrialisation. Employment
of OB 1 in NEW would equalise wage in two sectors and and further employ-
ment in the sector NEW would make the current relative wage in the sector
higher. Thus a self-fulfilling optimism could make a take off possible. On the
CHAPTER 13. HISTRY VS EXPECTATIONS 144

other hand if every agent is myopic and discounts future returns completely,
history can have a dominant role.

Per Capita Sectoral Return

A A' O B B'
Number of People Number of People
OLD NEW

Figure 13.3: History Vs Expectations

13.3 Increasing returns

A production activity displays increasing returns to scale if per unit cost is
declining with the scale of production. Increasing returns can take place
at aggregate (economy/sector/industry) level as well as the plant level. The
source of increasing returns at the aggregate level include externalities and
specialisation and the source at plant level include indivisibilities and greater
specialisation. Young (1928) emphasised the importance of increasing returns
in development. However, the ability to exploit increasing returns in the
production of a product depends on the size of the market for the product.
If the market is small, a profitable product never make it into the market.
Second, more subtly the size of the market may itself depend on the ability
to exploit increasing returns, greater production generates more income and
thereby higher demand. In this section we discuss these two issues.

13.3.1 Increasing returns and market entry

Similar to complementarities, increasing returns internal to the plant also
prevent the entry of socially more efficient products and technologies. This
CHAPTER 13. HISTRY VS EXPECTATIONS 145

can be explained using an example. Consider a developing country where

mother tongue is not English. Currently, computers in this country are using
an operating system (OS) designed and developed in an English speaking
country. The problem with the current OS is that it is very costly as it is
imported from a developed country, and second, a large section of the society
not knowing English language is unable to use computers. In response to this
a domestic software company developed a new operating system that uses
native languages and also has features specific to the requirements of the
people of this country. Further, the cost of development of the new operating
system is much lower as it is developed by the native programmers4 . Since
there is no language barrier for the new OS and it is cheap, the domestic
firm expected that it would have a larger market and ultimately replaces the
prevailing OS. Despite these advantages the firm found it difficult to sell the
new OS.
The difficulty is that, since the cost of designing and developing the new
OS is fixed, to realise the low per unit cost and thereby low price it has to
sell very large number of licenses. The situation can be illustrated using the
Figure 13.4. The current number of licenses of old OS in the market is Q
with an average cost of a and unit price p. Though domestic firm has lower
cost curve, initially the firm is at the higher point of this cost curve, with the
current number of licenses of (near) zero, the per unit cost is c which exceeds
a. The entry would not have been a problem if computer users instantly
switch their operating system to the new one. In software field, however, this
is not possible because (1) consumers are already learned and accustomed
to the old OS and they need some time to learn the new OS if they desire
so, and (2) several application softwares need to be made compatible to the
new OS, this takes time. Further to attract more and more customers, the
new OS must build some confidence among potential customers through
experience. This implies that during the transition time, where customers
are progressively switching to the new OS, the firm has to bear a loss. This
is easy enough to see from the figure. The current price of a licence is p, then
the firm must charge this price from day one itself to attract customers. This
implies that until the number of licenses reaches Q˚ , the firm has to function
at a loss. Losses would not be a problem if capital markets were perfect, then
the firm can get a loan to cover the interim losses. If capital market is not
complete or missing, the firm may find it difficult to finance the loss, and may
result in closing down of a socially desirable project(Ray, 1999).
Presence of both increasing returns at plant level and positive exter-
4
Here, for simplicity, we assume that the firm plans to release new and improved version
of the software in every year and that development costs of every version are equal.
CHAPTER 13. HISTRY VS EXPECTATIONS 146

Average cost/Price
Cost Curve of old OS

c
Cost Curve of new OS

p
a

Q* Q Number of Licenses

Figure 13.4: Increasing Returns and Market entry

nalities at the society level can make entry of innovative new technologies
difficult. In a market economy most of the technologies are developed by
the profit oriented private sector at a cost. So if somehow it is known or
perceived that the new technologies may find it difficult to invade the existing
market, the incentive to develop that technology is seriously diminished(Ray,
1999). This is particularly serious in developing country context. Most of
the technologies being used in the developing countries are designed and
developed in advanced countries and therefore they are more suitable for
their needs and environment. Extensive use of these technologies in the
developing world can become a barrier in the development of alternative
technologies suitable for the developing countries. A topical example is the
effort to develop and diffuse a computer operating system suitable for Indian
languages and conditions. The widespread use of Windows, with which users
are familiar, are actually preventing the migration to the new operating
system. This barrier become more strong, if the product is evolved through
extensive use and feedbacks from the customers, as usually the case of soft-
ware development. Complementarity is also very strong in this case as other
application softwares also need to be made compatible to the new operating
system5 .
5
Centre for Development of Advanced Computing has developed an operating system
suitable for Indian conditions, called BOSS, based on Linux. Note that this is a public sector
initiative.
CHAPTER 13. HISTRY VS EXPECTATIONS 147

13.3.2 Increasing Returns and Market Size

An important feature of modern production process is its roundaboutedness,
that is use of a variety of intermediate inputs and services to produce final
goods. These specialised inputs themselves may be produced using another
set of specialised inputs and services. Thus, the production of a commodity
has been reduced to a large number of steps, each of them have high degree
of sophistication and uses a number of intermediate inputs and skilled labour.
These sophisticated inputs are extremely costly to produce if they are serving
only a very tiny market. The division of labour is an important source of
wealth of nations and has been recognised in Adam Smith’s Wealth of Na-
tions. The specialised inputs include both specialised services, like banking,
consulting, auditing, wholesale services, transportation and machine repair
and specialised raw materials and machinery. Production of a final product
generally involves a number of functions. When the size of the market for the
final product is small, it may not be economical to establish a separate firm
to produce some intermediate inputs or to perform some specialised services.
However, when the market for the final product become large it is viable to
separate the whole production processes into a number of specialised ones
and perform them in separate firms and industries. At this stage it becomes
economically viable to design and produce highly specialised machinery and
tools for performing each task, as the market for these tools and machines
is big enough to make their separate production viable. This means that
division of labour is limited by the extend of the market (Stigler, 1951).
Generally increased division of labour is a feature of developed countries.
In poor economies few resources are allocated to produce specialised inputs
and most firms produce goods using techniques that rely intensively on direct
‘raw labour’. In this context one could ask why poor economies are not import-
ing specialised inputs produced in developed countries to benefit from the
division of labour existing there. The reasons for this include (1) non-tradable
nature of goods like banking, auditing, consulting, transportation and ma-
chine repairing, (2) in many cases employing imported physical intermediate
inputs is not possible due to the absence local provision of after sale and
repair services, (3) when inputs are to be imported it is possible that they
may not arrive in time or with correct specifications, forcing the firm to hold
high inventories of such inputs. This suggests that increased productivity
through deeper division of labour is possible only when these specialised
inputs are produced locally (Rodríguez-Clare, 1996).
Assuming increasing returns to scale in the production of intermediate
inputs and their non-tradable nature, Rodríguez-Clare (1996) shows the
presence of multiple equilibrium. The lower level equilibrium is characterised
CHAPTER 13. HISTRY VS EXPECTATIONS 148

Input variety and Productivity

Assume that final good Q is produced from a set of differentiated inter-

mediate inputs including producer services and n varieties of them are
available. Let the production function is,
„ż n ȷ1{α
α
Q“ xpjq dj (13.1)
0

0ăαă1
The production function (13.1) exhibits constant returns to scale. That is,
if the firm producing final goods, increases all available inputs by certain
proportion, the output would also increase by the same proportion. The
elasticity of substitution between any two inputs in the above technology
is ϵ “ 1{p1 ´ αq ą 1. The important property of the production technology
(13.1), which is relevant here, is that in this factor productivity raises
with the number of available varieties. To see this clearly suppose that
all intermediate inputs were produced according to the same constant-
returns-to-scale production function. In a symmetric equilibrium all
produced inputs would bear the same price and manufactures of final
goods (consumer goods) employ equal quantities xpjq “ x of each. Then
(13.1) implies that Q “ n1{α x. The same quantity of resources is needed
to produce a bundle of differentiated inputs of a given size, regardless of
its composition, so we can use X “ nx to measure the resources embodied
in the final goods. The final output per unit of primary input (i.e., total
factor productivity) is given by Q{X “ np1´αq{α . With 0 ă α ă 1, we see
that productivity of a given stock of resources rises with the number of
available varieties. Ethier (1982) ascribes this property of the technology
to the gains from increasing degree of specialisation in production.

by shallow division of labour in which low variety of specialised inputs are

produced. The low variety of specialised inputs leads to production of goods
using techniques that do not require a wide variety of specialised inputs
and use raw labour. The use of raw labour results in low productivity and
low income, which in turn keep the size of the market small for the final
goods. This in turn limits the size of the market for specialised inputs and
the incentives to undertake their production. In the desirable higher level
equilibrium, a large variety of intermediate goods are produced with deep
division of labour. Both wages and rate of return to capital are higher in this
equilibrium and it is also pareto superior to the equilibrium with shallow
CHAPTER 13. HISTRY VS EXPECTATIONS 149

division of labour. Faini (1984) shows that increasing returns to scale in the
production of non-traded intermediate inputs would give rise to a cumulative
divergence of growth rate across regions. Also the pattern of export and
investment in a backward region will be biased towards sectors with very low
service inputs requirements and therefore very low multiplier effect on the
local economy. A cumulative and circular causation process may therefore
inhibit growth in the backward regions. The small size of the service sector is
thus a barrier to industrial development. On the other hand limited demand
for the non-traded inputs by relatively underdeveloped industrial sector
becomes an obstacle to the expansion of service sector6 .
Increased division of labour contributes to economic growth through two
sources (1) Increasing returns in the intermediate input production, (2) In-
creasing returns due to the increase in the variety of inputs. This implies that
doubling the production budget in an industry not only leads to an increase
in the size of the industry and thereby exploitation of increasing returns
in the intermediate input industry, but also to greater variety of inputs. It
has been shown that increase in the variety of inputs generates increasing
returns to scale in an otherwise constant returns to scale technology (see:
Grossman and Helpman, 1991a; Ethier, 1982).

6
Faini (1984) argues that there is extensive evidence in support of the prevalence of
increasing returns to scale in the production of many producer services. Many professional
services such as consulting, auditing, engineering and architectural, are often related to the
provision of information. This by itself is likely to be a source of decreasing cost.
Bibliography

A CEMOGLU, D., J. A. R OBINSON, AND S. J OHNSON. (2005): “Institutions

as a Fundamental Cause of Long-Run Growth,” in Handbook of Economic
Growth 1A, ed. by P. Aghion, and S. Durlauf, chap. 6, pp. 386–472. North
Holland.
A GHION, P., AND P. H OWITT (1992): “A Model of Growth through Creative
Destruction,” Econometrica, 60(2), 323–351.
(2003): Endogeneous Growth Theory. The MIT Press, Cambridge M
A.
(2008): The Economics of Growth. MIT Press, Cambridge MA.
A LCHIAN, A. (1963): “Reliability of Progress Curves in Airframe Production,”
Econometrica, 31, 679–693.
A LCHIAN, A. A., AND H. D EMSETZ (1973): “The Property Right Paradigm,”
Journal of Economic History, 33(1), 16–27.
A RROW, K. (1962a): “The Economic Implications of Learning by Doing,”
Review of Economic Studies, 29, 155–173.
A RROW, K. J. (1962b): “Economic Welfare and Allocation of Resources for
Innovation,” in The Rate and Direction of Inventive Activity, ed. by R. R.
Nelson. Princeton University Press for NBER, Princeton.
A RTHUR , W. B. (1983): “On Competing Technologies and Historical Small
Events: The Dynamics of Choice under Increasing Returns,” Technological
innovation program workshop paper, Department of Economics, Stanford
University.
A RTHUR , W. B., Y. M. E RMOLIEV, AND Y. M. K ANIOVSKI (1985): “Strong
Laws for a Class of Path-Dependent Urn Processes,” in Proceedings of the
International conference on Stochastic Optimisation, Kiev. Spriger-Verlog,
Munich.

150
BIBLIOGRAPHY 151

A TKINSON, A. B., AND J. E. S TIGLITZ (1969): “A New View of Technological

Change,” The Economic Journal, 79(315), 573–578.

A ZARIADIS, C., AND A. D RAZEN (1990): “Threshold Externalities in Economic

Development,” Quarterly Journal of Economics, 105, 501–526.

B ALAND, J.-M., AND P. F RANCOIS (1996): “Innovation, monopolies and the

poverty trap,” Journal of Development Economics, 49(1), 151–178.

B ARDHAN, P. (1995): “The Contribution of Endogenous Growth Theory to

the Analysis of Development Problems: An Assessement,” in Handbook of
Development Economics- Vol 3B., ed. by J. Behrman, and T. N. Srinivasan,
p. ?? North–Holland Amsterdam.

B ARRO, R. J. (1991): “Economic Growth in a Cross Section of Countries,” The

Quarterly Journal of Economics, 106(2), 407–443.

(1999): “Notes on Growth Accounting,” Journal of Economic Growth,

4(2), 119–137.

B ARRO, R. J., N. G. M ANKIW, AND X. S ALA -I-M ARTIN (1995): “Capital

Mobility in Neoclassical Models of Growth,” American Economic Review,
85(1), 103–115.

B ARRO, R. J., AND X. S ALA -I-M ARTIN (1992): “Convergence,” Journal of

Political Economy, 100(2), 223–251.

(2004): Economic Growth- Second Edition. Prentice Hall India Private

Ltd, New Delhi.

B AUMOL , W. J. (1986): “Productivity Growth, Convergence, and Welfare:

What the Long-Run Data Show,” American Economic Review, 76(5), 1072–
1085.

B EN -D AVID, D., AND M. B. L OEWY (1998): “Free Trade, Growth and Conver-
gence,” Journal of Economic Growth, 3, 143–170.

(2000): “Knowledge Dissemination, Capital Accumulation, Trade and

Endogenous Growth,” Oxford Economic Papers, 52, 637–650.

B ÉNABOU, R. (1996): “Heterogeneity, Stratification, and Growth: Macroe-

conomic Implications of Community Structure and School Finance,” The
American Economic Review, 86(3), 584–609.
BIBLIOGRAPHY 152

B ERNARD, A. B., AND S. N. D URLAUF (1995): “Convergence in International

Output,” Journal of Applied Econometrics, 10(2), 97–108.

(1996): “Interpreting Test of the convergence hypothesis,” Journal of

Econometrics, 71(1-2), 161–173, Derivation of equation 16 is given in the
supplementary meterial directory.

B ROCK , W. A., AND S. N. D URLAUF (2001): “Growth Empirics and Reality,”

World Bank Economic Review, 15(2), 229–272.

C AMERON, G. (2010): “Classical economics and economic growth,” in Eco-

nomic Growth, ed. by S. N. Durlauf, and L. E. Blume, pp. 9–15. Palgrave,
Macmillan.

C ARRINGTON, W. J., AND E. D ETRAGIACHE (1998): “How Big Is the Brain

Drain?,” Working Paper 98/102, International Monetary Fund.

C ASS, D. (1965): “Optimum Growth in an Aggregative Model of Capital

Accumulation,” Review of Economic Studies, 32, 569–596.

C HENERY, H., S. R OBINSON, AND M. S YRQUIN (1986): Industsrialisation

and Growth: A Comparative Study. Oxford University Press, New York.

C HENERY, H. B., AND M. S YRQUIN (1975): Patterns of Development, 1950-

1970. Oxford University Press (for World Bank).

C ICCONE , A., AND K. M ATSUYAMA (1996): “Start-up costs and pecuniary

externalities as barriers to economic development,” Journal of Development
Economics, 49(1), 33–60.

C OASE , R. H. (1960): “The Problem of Social Cost,” Journal of Law and

Economics, 3, 1–44.

C OHEN, M. N. (1977): The food crisis in prehistory. Overpopulation and the

origins of agriculture. Yale University Press, New Haven, CT.

C OHEN, W. M., AND D. A. L EVINTHAL (1989): “Innovation and Learning:

The two faces of R&D,” Economic Journal, 99(397), 569–596.

D ASGUPTA , P., AND J. S TIGLITZ (1988): “Learning-by-Doing, Market Struc-

ture and Industrial and Trade Policies,” Oxford Economic Papers, 40(2),
246–268.

D AVID, P. A. (1970): “Learning by Doing and Tariff Protection,” Journal of

Economic History, 30, 521–601.
BIBLIOGRAPHY 153

(1975): Technical Choice, Innovation and Economic Growth. Cam-

bridge University Press, Cambridge.

(1985): “Clio and the Economics of QWERTY,” American Economic

Review: Papers and Proceedings, 75(2), 332–337.

D E L ONG, J. B. (1988): “Productivity Growth, Convergence, and Welfare:

Comment,” American Economic Review, 78(5), 1138–1154.

D EMSETZ , H. (1967): “Toward a Theory of Property Rights,” The American

Economic Review, 57(2), 347–359.

D INOPOULOS, E., AND P. T HOMPSON (1998): “Schumpeterian Growth with-

out Scale Effects,” Journal of Economic Growth, 3(4), pp. 313–335.

D IXIT, A. K., AND J. E. S TIGLITZ (1977): “Monopolistic Competition and

Optimum Product Diversity,” The American Economic Review, 67(3), 297–
308.

D OSI , G. (1988): “Sources, Procedures and Mirco economic effects of innova-

tion,” Journal of Economic Literature, 26(3), 1120–1171.

D OWRICK , S., AND D.-T. N GUYEN (1989): “OECD Comparative Economic

Growth 1950-85: Catch-Up and Convergence,” The American Economic
Review, 79(5), 1010–1030.

D URLAUF, S. N. (1993): “Nonergodic Economic Growth,” The Review of

Economic Studies, 60(2), 349–366.

D URLAUF, S. N., AND P. A. J OHNSON (2010): “Empirical regularities in

economic growth,” in Economic growth, ed. by S. N. Durlauf, and L. E.
Blume, pp. 51–56. Palgrave Macmillan.

E ASTERLY, W., AND R. L EVINE (2001): “It’s Not Factor Accumulation: Styl-
ized Facts and Growth Models,” The World Bank Economic Review, 15(2),
177–219.

E DWARDS, S. (1993): “Openness, Trade Liberalisation and Growth in Devel-

oping Countries,” Journal of Economic Literature, 31, 1358–1393.

E THIER , W. J. (1982): “National and International Returns to Scale in the

Modern theory of International Trade,” American Economic Review, 72(3),
389–405.
BIBLIOGRAPHY 154

E VENSON, R. E., AND L. E. W ESTPHAL (1995): “Technological Change and

Technology Strategy,” in Handbook of Development Economics: Volume
3A, ed. by J. Behrman, and T. N. Srinivasan, pp. 2209–2299. Elsevier,
Amsterdam.

FAINI , R. (1984): “Increasing Returns, Non-Traded Inputs and Regional

Development,” Economic Journal, 94(374), 308–323.

F INE , B. (2000): “Endogenous Growth Theory: A Critical Assessement,”

Cambridge Journal of Economics, 24, 245–265.

F REEMAN, C. (1994): “The Economics of Technical Change,” Cambridge

Journal of Economics, 18, 463–514.

G ALOR , O. (2010): “Economic growth in the very long run,” in Economic

growth, ed. by S. N. Durlauf, and L. E. Blume, pp. 57–67. Palgrave Macmil-
lan.

G ALOR , O., AND O. M OAV (2006): “Das Human-Kapital: A Theory of the

Demise of the Class Structure,” Review of Economic Studies, 73(1), 85–117.

G RILICHES, Z. (1957): “Hybrid Corn: An Exploration in the economics of

Technological Change,” Econometrica, 25, 501–522.

(1979): “Issues in Assessing the Contribution of Research and De-

velopment to Productivity Growth,” The Bell Journal of Economics, 10(1),
92–116.

G ROSSMAN, G., AND E. H ELPMAN (1990): “Trade, Innovation, and Growth,”

American Economic Review, 80(2), 86–91.

G ROSSMAN, G. M., AND E. H ELPMAN (1991a): Innovation and Growth in

the Global Economy. MIT Press, Cambridge MA.

(1991b): “Trade, Knowledge Spillovers and Growth,” European Eco-

nomic Review, 35, 517–586.

(1994): “Endogenous Innovation in the Theory of Growth,” Journal of

Economic Perspective, 8, 23–44.

(1995): “Technology and Trade,” in Handbook of International Eco-

nomics, ed. by G. M. Grossman, and K. Rogoff, pp. 1279–1337. Elsevier
Science, Amsterdam.
BIBLIOGRAPHY 155

H A , J., AND P. H OWITT (2007): “Accounting for Trends in Productivity and

R&D: A Schumpeterian Critique of Semi-Endogenous Growth Theory,”
Journal of Money, Credit and Banking, 39(4), 733–774.

H ARROD, R. F. (1937): “Review of Joan Robinson’s “Essays in the theory of

Employment”,” Economic Journal, 47(186), 326–330.

(1948): Towards a Dynamic Economics: Somce Recent Developments

of Economic Theory and their applications to Policy. Macmillan, London.

H ART, P. E. (1995): “Galtonian Regression Across Countries and Convergence

of Productivity,” Oxford Bulletin of Economics and Statistics, 57, 287–293.

H AYAMI , Y., AND Y. G ODO (2005): Development Economics: From Poverty to

the Wealth of Nations (Third Edition). Oxford University Press, New Delhi.

H ELPMAN, E., AND P. K RUGMAN (1985): MarketStructureandForeign Trade:

Increasing Returns, ImperfectCompetition,and theInternational Economy.
MIT Press, Cambridge MA.

H ELPMAN, E., M. J. M ELITZ , AND S. R. Y EAPLE (2004): “Export versus FDI

with Heterogeneous Firms,” American Economic Review, 94(1), 300–316.

H ICKS, J. R. (1963): The Theory of Wages, 2nd edn,. Macmillan, London.

H IRSCH , W. Z. (1956): “Firm Progress Ratios,” Econometrica, 24, 136–143.

H IRSCHMAN, A. O. (1958): The Strategy of Economic Development. Yale

University Press, New Haven, CT.

H OLLANDER , S. (1965): The Sources of Increased Efficiency: A Study of

DuPont Rayon Plants. MIT Press, Cambridge MA.

H OWITT, P., AND D. N. W EIL (2010): “Economic growth,” in Economic growth,

ed. by S. N. Durlauf, and L. E. Blume, pp. 38–50. Palgrave Macmillan.

H SIEH , C.-T. (2002): “What Explains the Industrial Revolution in East Asia?
Evidence from the Factor Markets.,” American Economic Review, 92(3),
502 – 526.

H UGHES, K. (1986): Export and Technology. Cambridge University Press,

Cambridge, UK.

J ONES, C. I. (1995a): “R & D-Based Models of Economic Growth,” Journal of

Political Economy, 103(4), 759–784.
BIBLIOGRAPHY 156

(1995b): “Time Series Tests of Endogenous Growth Models,” The

Quarterly Journal of Economics, 110(2), 495–525.

(2001): Introduction to Economic Growth (Second Edition).

W.W.Norton & Company, New York.

J ONES, H. G. (1976): An Introduction to Modern Theories of Economic

Growth. McGraw Hill Book Company, New York.

J ORGENSON, D., F. M. G OLLOP, AND B. M. F RAUMENI (1987): Productivity

and US Economic Growth. Harvard University Press, Cambridge M A.

J ORGENSON, D. W., AND Z. G RILICHES (1967): “The Explanation of Produc-

tivity Change,” The Review of Economic Studies, 34(3), 249–283.

K OOPMANS, T. (1965): “On the Concept of Optimal Economic Growth,” in

The Econometric Approach to Development Planning. North–Holland, Ams-
terdam.

K ORTUM , S. S. (1997): “Research, Patenting, and Technological Change,”

Econometrica, 65(6), 1389–1419.

K REMER , M. (1993): “The O-Ring Theory of Economic Development,” Quar-

terly Journal of Economics, 108, 551–575.

K RUGMAN, P. (1991): “History Versus Expectations,” Quarterly Journal of

Economics, 106(2), 651–667.

K RUGMAN, P. R. (1993): “Towards a Counter-Counterrevolution in Devel-

opment Theory,” in Proceedings of the World Bank Annual Conference on
Development Economics 1992, pp. 15–38. The World Bank, Washington
DC.

L ANDES, D. S. (1990): “Why are We So Rich and They So Poor?,” American

Economic Review, 80(2), 1–13.

L EWIS, W. A. (1954): “Economic Development with unlimited supplies of

Labour,” The Manchester Scholl of Economic and Social Studies, 22, 139–
191.

L UCAS, R. E. J. (1988): “On the Mechanics of Economic Development,”

Journal of Monetary Economics, 22, 3–42.

(1993): “Making the Miracle,” Econometrica, 61, 251–272.

BIBLIOGRAPHY 157

L UCAS J R , R. E. (1990): “Why Doesn’t Capital Flow from Rich to Poor

Countries?,” American Economic Review: Papers and Proceedings, 80(2),
92–96.

M ADDISON, A. (2007): The World Economy (Two-in-One Edition): Volume

1 - A Millennial Perspective, Volume 2 - Historical Statistics. Academic
Foundation, New Delhi, indian edition edn.

M ANKIW, N. G., D. R OMER , AND D. N. W EIL (1992): “A Contribution to

the Empirics of Economic Growth,” The Quarterly Journal of Economics,
107(2), 407–437.

M ANSFIELD, E. (1961): “Technical Change and Rate of Immitation,” Econo-

metrica, 29, 741–766.

M ATSUYAMA , K. (1991): “Increasing Returns, Industrialization, and Indeter-

minacy of Equilibrium,” Quarterly Journal of Economics, 106(2), 617–650.

M URPHY, K. M., A. S HLEIFER , AND R. V ISHNY (1989): “Income Distribu-

tion, Market Size, and Industrialization,” Quarterly Journal of Economics,
104(3), 537–564.

N ORTH , D. C. (1990): Institutions, institutional change and economic perfor-

mance. Cambridge university press.

PARENTE , S. L., AND E. C. P RESCOTT (1994): “Barriers to Technology

Adoption and Development,” Journal of Political Economy, 102(2), 298–
321.

P ERETTO, P. F. (1998): “Technological Change and Population Growth,”

Journal of Economic Growth, 3(4), 283–311.

P RESCOTT, E. C. (1997): “Needed: A Theory of Total Factor Productivity,”

Working Paper 242, Federal Reserve Bank of Minneapolis, Federal Reserve
Bank of Minneapolis.

R AMSEY, F. (1928): “A Mathematical Theory of Saving,” Economic Journal,

38, 543–559.

R APPING, L. (1965): “Learning and World War II Production Functions,”

Review of Economics and Statistics, 47, 81–86.

R AY, D. (1999): Development Economics. Oxford University Press, New Delhi.

BIBLIOGRAPHY 158

R EYNOLDS, L. G. (1983): “The Spread of Economic Growth to the Third

World: 1850-1980,” Journal of Economic Literature, 21, 941–980.

R IVERA -B ATIZ , L. A., AND P. M. R OMER (1991): “Economic Integration and

Endogenous Growth,” Quarterly Journal of Economics, 106(2), 531–555.

R OBINSON, J. (1938): “The Classification of Inventions,” Review of Economic

Studies, 5(2), 139–142.

R OCKAFELLAR , R. T. (1970): Convex Analysis. Princeton University Press,

Princeton, N J.

R ODRÍGUEZ -C LARE , A. (1996): “The division of labour and economic devel-

opment,” Journal of Development Economics, 49(1), 3–32.

R ODRIK , D. (2015): Economics Rules: The Rights and Wrongs of the Dismal
Science. W. W. Norton & Company.

R OMER , P. (1986a): “Cake Eating, Chattering, and Jumps: Existence Results

for Variational Problems,” Econometrica, 54(4), 897–908.

R OMER , P. M. (1986b): “Increasing Returns and Long-Run Growth,” Journal

of Political Economy, 94(5), 1002–1037.

(1987): “Growth based on increasing returns due to specialisation,”

American Economic Review: Papers and Proceedings, 77(2), 56–62.

(1990a): “Capital, Labour, and Productivity,” Brookings Papers on

Economic Activity. Microeconomics, 1990, 337–367.

(1990b): “Endogenous Technological Change,” Journal of Political

Economy, 98(5), s71–s102.

(1994): “The Origins of Endogenous Growth,” Journal of Economic

Perspective, 8(1), 3–22.

R OSENBERG, N. (1976): Perspectives on Technology. Cambridge University

Press, Cambridge.

(1982): Inside the Black Box: Technology and Economics. Cambridge

University Press, Cambridge.

R OSENSTEIN -R ODAN, P. N. (1943): “Problems of Industrialisation of Eastern

and South-Eastern Europe,” Economic Journal, 53(210/211), 202–211.
BIBLIOGRAPHY 159

R UBINSTEIN, A. (2006): “Dilemmas of an Economic Theorist,” Econometrica,

74(4), 865–883.

S ACHS, J. D., AND A. M. WARNER (1995): “Economic Convergence and

Economic Policies,” Working Paper 5039, National Bureau of Economic
Research, Cambridge MA.

S ALTER , W. E. G. (1966): Prodcutivity and Technical Change 2nd edn. Cam-

bridge University Press, Cambridge.

S ATO, R., AND M. J. B ECKMANN (1968): “Neutral Inventions and Production

Functions,” Review of Economic Studies, 35(1), 57–66.

S CHMOOKLER , J. (1966): Innovation and Economic Growth. Harvard Uni-

versity Press, Cambridge.

S CITOVSKY, T. (1954): “Two Concepts of External Economies,” Journal of

Political Economy, 62(2), 143–151.

S EGERSTROM , P. S. (1998): “Endogenous Growth without Scale Effects,” The

American Economic Review, 88(5), pp. 1290–1310.

S EN, A. (1988): “The Concept of Development,” in Handbook of Development

Economics Vol 1., ed. by H. Chenery, and T. N. Srinivasan, chap. 1, pp. 9–26.
North Holland, Amsterdam.

S HELL , K. (1967): “A Model of Inventive Activity and Capital Accumulation,”

in Essays on the Theory of Optimal Economic Growth, ed. by K. Shell, pp.
31–52. The MIT Press, Cambridge.

S HESHINSKI , E. (1967): “Optimal Accumulation with Learning by Doing,”

in Essays on the Theory of Optimal Economic Growth., ed. by K. Shell, pp.
31–52. The MIT Press, Cambridge.

S MULDERS, S., AND T. V. D. K LUNDERT (1995): “Imperfect Competition,

Concentration and Growth with firm specific R&D,” European Economic
Review, 39(1), 139–160.

S OLOW, R. (1954): “The Survival of Mathematical Economics,” Review of

Economics and Statistics, 36(4), 372–374.

(1956): “A Contribution to the Theory of Economic Growth,” Quarterly

Journal of Economics, 70(1), 65–94.
BIBLIOGRAPHY 160

S OLOW, R. M. (1959): “Investment and Technical Change,” in Mathematical

Methods in the Social Sciences, ed. by K. J. Arrow, S. Karlin, and P. Suppes.
Stanford University Press, Palo Alto, CA.

S TIGLER , G. J. (1951): “The Division of Labor is Limited by the Extent of the

Market,” Journal of Political Economy, 59(3), 185–193.

S TIGLITZ , J., A. K. S EN, AND J.-P. F ITOUSSI (2009): “Report by the Commis-
sion on the Measurement of Economic Performance and Social Progress,”
Available from the URL: http://www.stiglitz-sen-fitoussi.fr/
documents/rapport_anglais.pdf, accessed September 23, 2009.

S WAN, T. W. (1956): “Economic Growth and Capital Accumulation,” Economic

Record, 32, 334–361.

T EITEL , S. (1984): “Technology Creation in Semi-industrial Economies,”

Journal of Development Economics, 16, 36–61.

T HIRLWALL , A. P. (2002): The Nature of Economic Growth. Edward Elgar,

Cheltenham, UK.

T RACA , D. A. (2002): “Import as Competitive Discipline: The role of produc-

tivity gap,” Journal of Development Economics, 69, 1–21.

U ZAWA , H. (1961): “Neutral Inventions and the Stability of Growth Equili-

birum,” Review of Economic Studies, 28(2), 117–124.

VARIAN, H. R. (1999): Intermediate Microeconomics. WW Norton & Company,

New York.

V ERSPAGEN, B. (1992): “Endogenous Innovation in the Neoclassical Growth

Models: A Survey,” Journal of Macro Economics, 14, 631–662.

W EBER , M. (1930): The Protestant Ethic and the Spirit of Capitalism. Rout-
ledge, London.

Y OUNG, A. (1991): “Learning by doing and Dynamic effect of International

Trade,” Quarterly Journal of Economics, 106(2), 369–405.

Y OUNG, A. (1998): “Growth without Scale Effects,” Journal of Political

Economy, 106(1), pp. 41–63.

Y OUNG, A. A. (1928): “Increasing Returns and Economic Progress,” Economic

Journal, 38(152), 527–542.