Endogenous Growth: The Economics of Ideas
The issue of convergence has raised as an important question the possible
importance of differences in technology (knowledge) across the economies
of the world. The Solow model attempts to explain per capita income level
and growth differences assuming that technology is a pure public good and is
therefore freely available to all countries irrespective of their level of development.
An increasing number of economists, and most economic historians
and development economists see significant technology gaps as the crucial
problem facing poor countries. Such an approach emphasizes the need for
policies to be adopted that will close ‘idea gaps’ between nations (P. Romer,
1993).
Paul Romer’s 1986 model explains technological progress as an unintentional
by-product of capital accumulation by individual firms. Subsequently,
Romer (1990), dissatisfied with his initial approach, proceeded to develop a
second strand of new growth theory. Endogenous innovation models embrace
a neo-Schumpeterian framework of endogenous technological change based
on three premises (Grossman and Helpman, 1991, 1994; Crafts, 1996; Aghion
and Howitt, 1998). First, as in the Solow model, the basic driving force
behind economic growth is technological change, that is, improvements in
knowledge about how we transform inputs into outputs in the production
process. Second, technological change is endogenous, being determined by
the deliberate activities of economic agents acting largely in response to
financial incentives. Third, the defining characteristic of ideas/knowledge is
that ‘once the cost of creating a new set of instructions has been incurred, the
instructions can be used over and over again at no additional cost’ (Romer,
1990). Therefore ideas are non-rivalrous outputs and their use by one firm or
person does not in any way reduce their availability to other firms or persons.
Ideas are also ‘partially excludable’, where excludability is defined as the
ability of the owners of a good to prevent other economic agents from using it
without payment. As Romer (1990) notes, ‘excludability is a function of the
technology and the legal system’. Given Romer’s second premise that technological
change results from the purposeful actions of self-interested
economic agents, improvements in technology (new ideas) must generate
benefits to individuals that are at least ‘partially excludable’, for example by
having patent laws.
Romer’s insights have led to a burgeoning of research into the economics
of ideas (Jones, 2002, 2005). The three premises discussed above have two
important implications for the theory of economic growth. First, because
ideas are non-rivalrous, they can be accumulated without limit on a per capita
basis. Second, because of incomplete excludability (appropriability), knowledge
creation involves substantial spillovers of benefits (externalities) which
cannot be entirely captured by the economic agents who produce the ideas.
The ‘unbounded’ growth and ‘incomplete appropriability’ features of the
economics of ideas imply that ‘output cannot be a constant-returns-to-scale
function of all its inputs taken together’. Romer’s analysis implies increasing
returns to scale and by implication microfoundations based on the presence
of imperfect competition (see Romer, 1994a).
While a non-rivalrous good such as a new idea involves a fixed cost of
production, which is often substantial, once the new knowledge has been
created there is zero marginal cost involved with any further use of the new
idea. A new design is costly to produce but once in existence it can be used as
often as desired and in as many contexts as desired. It is for this reason that
legal mechanisms such as patents and copyrights exist in order to grant
investors monopoly rights over a new idea, at least for a time, so that they can
earn a reward for their new ideas (Kremer, 1998; Mazzoleni and Nelson,
1998). The importance of this issue has been illustrated by North (1990), who
argues that the economic development of Western Europe did not seriously
begin until the development of property rights ensured that individuals could
reap some of the benefits of their ‘ideas’ and helped to speed up the pace of
technological change (Crafts, 1995). The era of modern economic growth,
beginning with the Industrial Revolution in Britain,
occurred when the institutions protecting intellectual property rights were
sufficiently well developed that entrepreneurs could capture as a private return
some of the enormous social returns their innovations would create … history
suggests that it is only when the market incentives were sufficient that widespread
innovation and growth took hold. (Jones, 2001a)
In the case of the USA the framers of the US Constitution were eager to
‘promote the progress of science and useful arts’. Therefore an intellectual
property clause providing for copyright and patent rights appears in the first
article of the Constitution and by 1810 the USA ‘far surpassed Britain in
patenting per capita’ (Khan and Sokoloff, 2001). The failure of China to lead
the first Industrial Revolution has also been attributed to that country’s inability
to establish a free market, institutionalize property rights, provide an
environment conducive to emulation and innovation, and to absorb foreign
technology (Landes, 1998). Thus according to the new breed of endogenous
growth models, ‘the government has great potential for good or ill through its
influence on the long-term rate of growth’ (Barro, 1997). Economic growth
can be influenced not only by policies that affect trade regimes, technology
transfer, the provision of infrastructure and financial markets, but also by
policies that affect taxation and incentives, the protection of intellectual
property rights and the maintenance of law and order.
By developing an endogenous theory of technological change Romer has
challenged both the traditional and augmented versions of the Solow neoclassical
growth model (see below, section 11.16). In the neoclassical model
technology is assumed to be exogenous and hence available without limitation
everywhere across the globe. Romer (1995) rejects this assumption on
the basis of ‘overwhelming evidence’ that technology is not a pure public
good. The neoclassical model emphasizes ‘object gaps’, differences in physical
and human capital, in explaining income per capita differentials across
nations. While Mankiw (1995) believes that much of the variation in living
standards can be explained by differences in the quantities of human and
physical capital, in contrast Romer (1993) emphasizes ‘idea gaps’, productivity
differences resulting from technology gaps, as the main source of divergent
living standards.
Parente and Prescott (1994, 1999, 2000) also attribute differences in international
incomes to technology gaps. In their research they have found evidence
to suggest that these productivity gaps are not caused by fundamental differences
in the stock of available knowledge that developing countries have
access to. Instead, Parente and Prescott argue that there exist barriers in the
form of society-imposed constraints which prevent firms in many developing
countries from adopting better production methods, and many of these constraints
‘are put in place to protect the interests of groups vested in current
production processes’. As a result they conclude that most differences in
international incomes ‘are the result of differences in total factor productivity’.
Parente and Prescott (2005) conclude that ‘changes in a country’s
institutions that result in large increases in the efficiency with which resources
can be used in production give rise to growth miracles’.
Romer’s position has received recent support from the research of Easterly
and Levine (2001), who find that the ‘residual’ (total factor productivity)
rather than factor accumulation can explain most of the cross-country income
and growth differentials. Their data show that while factor accumulation is
persistent, growth is not. Nelson and Pack (1999), in their discussion of the
Asian miracle and modern growth theory, also stress the importance of the
entrepreneurship, innovation and learning that these economies had to undertake
before they could successfully assimilate new technologies. In their view
the accumulation of human and physical capital is a necessary but far from
sufficient part of this process. What is crucial for success is the establishment
of a policy environment that nurtures learning, and for economists to better
understand the learning process taking place during the assimilation of new
ideas and technologies they need ‘a better theory of firm behaviour in such
situations’.
Historical experience demonstrates that the creation and transmission of
ideas has undoubtedly been an important determinant of current living standards
(Rosenberg, 1994; Mokyr, 2005). If Romer is correct and the poor
countries do suffer from idea gaps rather than object gaps, then a significant
part of worldwide poverty can be eliminated ‘at relatively low cost’ via
technological ‘catch-up’. A clear implication of this analysis is that nations
which isolate themselves from the free flow of ideas, or erect barriers to the
adoption of new technologies, will suffer relative stagnation since trade policies
and openness affect innovation and growth. Foreign direct investment
can act as a significant channel for the diffusion of new innovations and
ideas, thereby enhancing the growth process (Grossman and Helpman, 1990;
Romer, 1994b; Sachs and Warner, 1995; Proudman and Redding, 1997;
Edwards, 1998; Parente and Prescott, 2000). Therefore, at least potentially,
poor economies have the most to gain from reducing restrictions to international
trade, encouraging inward FDI flows and investing in human capital
because by doing so they can gain access to the stock of world knowledge
(World Bank, 1998/9). While in the neoclassical model the removal of inefficiencies
caused by trade barriers will produce level effects on production
possibilities but no sustained growth effects, in endogenous growth models
the growth effects of increasing economic integration are likely to be much
more important.
A further implication of Paul Romer’s research is that for the USA to
maintain its leadership position, government policies must continue to support
a high level of R&D activities in both private and public institutions.
Given the well-documented large divergence between social and private rates
of return from R&D expenditures, the government has a vital role to play in
preventing underinvestment in this activity. In a recent investigation of the
optimal rate of R&D investment in the USA Jones and Williams (1998)
conclude that the private rate of return to R&D in the USA is of the order of
7–14 per cent, while a ‘conservative estimate’ of the social rate of return is 30
per cent. Therefore Jones and Williams conclude that optimal R&D spending
as a share of GDP is ‘more than two to four times larger than actual spending’
(see also Jones and Williams, 2000).
In contrast to the supply-side view of the growth of knowledge, ideas and
technological change, Schmookler (1966) argues that technological change is
primarily demand-induced. Unlike Romer’s model, where a key input to the
development of new technology is the supply of previous innovations (see
Jones, 2005), Schmookler sees the stimulus to technological change and innovation
as the need to solve current technological problems; that is, technological
change is demand-driven and dependent on the usefulness of new ideas. In a
recent discussion of Schmookler’s work, Kelly (2002) concludes that the supply-
and demand-side influences on technology are complementary.
An important deficiency of recent endogenous growth theories is that they
lose the prediction of conditional convergence, a prediction which Barro
(1997) argues has a ‘strong empirical regularity in the data for countries and
regions’. To rectify this flaw Barro and Sala-i-Martin (1997) have developed
a model that combines elements of endogenous growth with the convergence
implications of the Solow model. Their model has the following elements:
1. in the long run the rate of growth in the world economy is driven by
technological discoveries in the leading economies;
2. follower economies share in the new innovations via a process of imitation;
3. since imitation is generally cheaper than innovation, ‘most countries
prefer to copy rather than invent’;
4. the relatively low cost of imitation implies that the follower economies
will grow relatively faster than the leader economies and converge, at
least part way, towards the leaders;
5. as the amount of uncopied innovations decreases, the costs of imitation
will tend to rise and therefore the follower’s growth rate will tend to slow
down;
6. therefore, the Barro/Sala-i-Martin model generates a form of conditional
convergence based on the diffusion of technology across countries and
resembles the predictions of the Solow model;
7. in the long run ‘all economies grow at the rate of discovery in the leading
places’.
The Barro/Sala-i-Martin hybrid model therefore establishes a framework where
long-run growth is driven endogenously by the discovery of new ideas in the
‘leading-edge’ economies, but also retains the convergence properties of the
neoclassical growth model via the impact of the imitation behaviour of follower
countries.
The issue of convergence has raised as an important question the possible
importance of differences in technology (knowledge) across the economies
of the world. The Solow model attempts to explain per capita income level
and growth differences assuming that technology is a pure public good and is
therefore freely available to all countries irrespective of their level of development.
An increasing number of economists, and most economic historians
and development economists see significant technology gaps as the crucial
problem facing poor countries. Such an approach emphasizes the need for
policies to be adopted that will close ‘idea gaps’ between nations (P. Romer,
1993).
Paul Romer’s 1986 model explains technological progress as an unintentional
by-product of capital accumulation by individual firms. Subsequently,
Romer (1990), dissatisfied with his initial approach, proceeded to develop a
second strand of new growth theory. Endogenous innovation models embrace
a neo-Schumpeterian framework of endogenous technological change based
on three premises (Grossman and Helpman, 1991, 1994; Crafts, 1996; Aghion
and Howitt, 1998). First, as in the Solow model, the basic driving force
behind economic growth is technological change, that is, improvements in
knowledge about how we transform inputs into outputs in the production
process. Second, technological change is endogenous, being determined by
the deliberate activities of economic agents acting largely in response to
financial incentives. Third, the defining characteristic of ideas/knowledge is
that ‘once the cost of creating a new set of instructions has been incurred, the
instructions can be used over and over again at no additional cost’ (Romer,
1990). Therefore ideas are non-rivalrous outputs and their use by one firm or
person does not in any way reduce their availability to other firms or persons.
Ideas are also ‘partially excludable’, where excludability is defined as the
ability of the owners of a good to prevent other economic agents from using it
without payment. As Romer (1990) notes, ‘excludability is a function of the
technology and the legal system’. Given Romer’s second premise that technological
change results from the purposeful actions of self-interested
economic agents, improvements in technology (new ideas) must generate
benefits to individuals that are at least ‘partially excludable’, for example by
having patent laws.
Romer’s insights have led to a burgeoning of research into the economics
of ideas (Jones, 2002, 2005). The three premises discussed above have two
important implications for the theory of economic growth. First, because
ideas are non-rivalrous, they can be accumulated without limit on a per capita
basis. Second, because of incomplete excludability (appropriability), knowledge
creation involves substantial spillovers of benefits (externalities) which
cannot be entirely captured by the economic agents who produce the ideas.
The ‘unbounded’ growth and ‘incomplete appropriability’ features of the
economics of ideas imply that ‘output cannot be a constant-returns-to-scale
function of all its inputs taken together’. Romer’s analysis implies increasing
returns to scale and by implication microfoundations based on the presence
of imperfect competition (see Romer, 1994a).
While a non-rivalrous good such as a new idea involves a fixed cost of
production, which is often substantial, once the new knowledge has been
created there is zero marginal cost involved with any further use of the new
idea. A new design is costly to produce but once in existence it can be used as
often as desired and in as many contexts as desired. It is for this reason that
legal mechanisms such as patents and copyrights exist in order to grant
investors monopoly rights over a new idea, at least for a time, so that they can
earn a reward for their new ideas (Kremer, 1998; Mazzoleni and Nelson,
1998). The importance of this issue has been illustrated by North (1990), who
argues that the economic development of Western Europe did not seriously
begin until the development of property rights ensured that individuals could
reap some of the benefits of their ‘ideas’ and helped to speed up the pace of
technological change (Crafts, 1995). The era of modern economic growth,
beginning with the Industrial Revolution in Britain,
occurred when the institutions protecting intellectual property rights were
sufficiently well developed that entrepreneurs could capture as a private return
some of the enormous social returns their innovations would create … history
suggests that it is only when the market incentives were sufficient that widespread
innovation and growth took hold. (Jones, 2001a)
In the case of the USA the framers of the US Constitution were eager to
‘promote the progress of science and useful arts’. Therefore an intellectual
property clause providing for copyright and patent rights appears in the first
article of the Constitution and by 1810 the USA ‘far surpassed Britain in
patenting per capita’ (Khan and Sokoloff, 2001). The failure of China to lead
the first Industrial Revolution has also been attributed to that country’s inability
to establish a free market, institutionalize property rights, provide an
environment conducive to emulation and innovation, and to absorb foreign
technology (Landes, 1998). Thus according to the new breed of endogenous
growth models, ‘the government has great potential for good or ill through its
influence on the long-term rate of growth’ (Barro, 1997). Economic growth
can be influenced not only by policies that affect trade regimes, technology
transfer, the provision of infrastructure and financial markets, but also by
policies that affect taxation and incentives, the protection of intellectual
property rights and the maintenance of law and order.
By developing an endogenous theory of technological change Romer has
challenged both the traditional and augmented versions of the Solow neoclassical
growth model (see below, section 11.16). In the neoclassical model
technology is assumed to be exogenous and hence available without limitation
everywhere across the globe. Romer (1995) rejects this assumption on
the basis of ‘overwhelming evidence’ that technology is not a pure public
good. The neoclassical model emphasizes ‘object gaps’, differences in physical
and human capital, in explaining income per capita differentials across
nations. While Mankiw (1995) believes that much of the variation in living
standards can be explained by differences in the quantities of human and
physical capital, in contrast Romer (1993) emphasizes ‘idea gaps’, productivity
differences resulting from technology gaps, as the main source of divergent
living standards.
Parente and Prescott (1994, 1999, 2000) also attribute differences in international
incomes to technology gaps. In their research they have found evidence
to suggest that these productivity gaps are not caused by fundamental differences
in the stock of available knowledge that developing countries have
access to. Instead, Parente and Prescott argue that there exist barriers in the
form of society-imposed constraints which prevent firms in many developing
countries from adopting better production methods, and many of these constraints
‘are put in place to protect the interests of groups vested in current
production processes’. As a result they conclude that most differences in
international incomes ‘are the result of differences in total factor productivity’.
Parente and Prescott (2005) conclude that ‘changes in a country’s
institutions that result in large increases in the efficiency with which resources
can be used in production give rise to growth miracles’.
Romer’s position has received recent support from the research of Easterly
and Levine (2001), who find that the ‘residual’ (total factor productivity)
rather than factor accumulation can explain most of the cross-country income
and growth differentials. Their data show that while factor accumulation is
persistent, growth is not. Nelson and Pack (1999), in their discussion of the
Asian miracle and modern growth theory, also stress the importance of the
entrepreneurship, innovation and learning that these economies had to undertake
before they could successfully assimilate new technologies. In their view
the accumulation of human and physical capital is a necessary but far from
sufficient part of this process. What is crucial for success is the establishment
of a policy environment that nurtures learning, and for economists to better
understand the learning process taking place during the assimilation of new
ideas and technologies they need ‘a better theory of firm behaviour in such
situations’.
Historical experience demonstrates that the creation and transmission of
ideas has undoubtedly been an important determinant of current living standards
(Rosenberg, 1994; Mokyr, 2005). If Romer is correct and the poor
countries do suffer from idea gaps rather than object gaps, then a significant
part of worldwide poverty can be eliminated ‘at relatively low cost’ via
technological ‘catch-up’. A clear implication of this analysis is that nations
which isolate themselves from the free flow of ideas, or erect barriers to the
adoption of new technologies, will suffer relative stagnation since trade policies
and openness affect innovation and growth. Foreign direct investment
can act as a significant channel for the diffusion of new innovations and
ideas, thereby enhancing the growth process (Grossman and Helpman, 1990;
Romer, 1994b; Sachs and Warner, 1995; Proudman and Redding, 1997;
Edwards, 1998; Parente and Prescott, 2000). Therefore, at least potentially,
poor economies have the most to gain from reducing restrictions to international
trade, encouraging inward FDI flows and investing in human capital
because by doing so they can gain access to the stock of world knowledge
(World Bank, 1998/9). While in the neoclassical model the removal of inefficiencies
caused by trade barriers will produce level effects on production
possibilities but no sustained growth effects, in endogenous growth models
the growth effects of increasing economic integration are likely to be much
more important.
A further implication of Paul Romer’s research is that for the USA to
maintain its leadership position, government policies must continue to support
a high level of R&D activities in both private and public institutions.
Given the well-documented large divergence between social and private rates
of return from R&D expenditures, the government has a vital role to play in
preventing underinvestment in this activity. In a recent investigation of the
optimal rate of R&D investment in the USA Jones and Williams (1998)
conclude that the private rate of return to R&D in the USA is of the order of
7–14 per cent, while a ‘conservative estimate’ of the social rate of return is 30
per cent. Therefore Jones and Williams conclude that optimal R&D spending
as a share of GDP is ‘more than two to four times larger than actual spending’
(see also Jones and Williams, 2000).
In contrast to the supply-side view of the growth of knowledge, ideas and
technological change, Schmookler (1966) argues that technological change is
primarily demand-induced. Unlike Romer’s model, where a key input to the
development of new technology is the supply of previous innovations (see
Jones, 2005), Schmookler sees the stimulus to technological change and innovation
as the need to solve current technological problems; that is, technological
change is demand-driven and dependent on the usefulness of new ideas. In a
recent discussion of Schmookler’s work, Kelly (2002) concludes that the supply-
and demand-side influences on technology are complementary.
An important deficiency of recent endogenous growth theories is that they
lose the prediction of conditional convergence, a prediction which Barro
(1997) argues has a ‘strong empirical regularity in the data for countries and
regions’. To rectify this flaw Barro and Sala-i-Martin (1997) have developed
a model that combines elements of endogenous growth with the convergence
implications of the Solow model. Their model has the following elements:
1. in the long run the rate of growth in the world economy is driven by
technological discoveries in the leading economies;
2. follower economies share in the new innovations via a process of imitation;
3. since imitation is generally cheaper than innovation, ‘most countries
prefer to copy rather than invent’;
4. the relatively low cost of imitation implies that the follower economies
will grow relatively faster than the leader economies and converge, at
least part way, towards the leaders;
5. as the amount of uncopied innovations decreases, the costs of imitation
will tend to rise and therefore the follower’s growth rate will tend to slow
down;
6. therefore, the Barro/Sala-i-Martin model generates a form of conditional
convergence based on the diffusion of technology across countries and
resembles the predictions of the Solow model;
7. in the long run ‘all economies grow at the rate of discovery in the leading
places’.
The Barro/Sala-i-Martin hybrid model therefore establishes a framework where
long-run growth is driven endogenously by the discovery of new ideas in the
‘leading-edge’ economies, but also retains the convergence properties of the
neoclassical growth model via the impact of the imitation behaviour of follower
countries.
No comments:
Post a Comment