Cjengt

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/5208241
Endogenous Growth Theory: A Critical Assessment
Article in Cambridge Journal of Economics · February 2000

Source: RePEc
CITATIONS READS
98 10,794
1 author:
Ben Fine
SOAS, University of London
371 PUBLICATIONS 9,915 CITATIONS
SEE PROFILE
All content following this page was uploaded by Ben Fine on 23 January 2017.
The user has requested enhancement of the downloaded file.

Cambridge Journal of Economics 2000, 24, 245–265
CRITICAL SURVEY
This is the latest in our series of Critical Survey articles. The aim of the series is to report on recent develop -
ments, to provide an assessment of alternative approaches and to suggest lines of future enquiry. The
intention is that the articles should be accessible not only to other academic researchers but also to students
and others more practically involved in the economy. Earlier Survey articles include Chris Freeman on
‘The Economics of Technical Change’, Allin Cottrell on ‘Post-Keynesian Monetary Economics’, Herbert
Hovenkamp on Law and Economics in the United States’, Warren Samuels on ‘Institutional
Economics’, Philip Arestis on ‘Post-Keynesian Economics’, Geoff Ingham on ‘Economics and Sociology’,
Sheila Dow on ‘Mainstream Economic Methodology’, Lionel Orchard and Hugh Stretton on ‘Public
Choice’, Andrew Glyn on ‘Does Profitability Really Matter?’, Ron Martin on ‘The New “Geographical
Turn” in Economics’ and Sergio Cesaratto on ‘Savings and Economic Growth in Neoclassical Theory’.
Endogenous growth theory: a critical

assessment
Ben Fine*
A critical assessment is made of endogenous growth theory from the perspective of

recent developments within economics as a discipline. These include its increasing
mathematical formalisation, its focus upon microfoundations, the casual use of
econometrics to test models, and the incorporation of factors that have traditionally
been outside mainstream economics. It is found that the theory focuses upon market
imperfections and technological progress in a variety of ways, and reconstructs their
impact upon the macroeconomy as a growth rather than as a level effect. Whilst the
growth rate is endogenised, this leads to the problem of multiple equilibria and
extremely complex dynamics. The generalisation from exogenous growth theory is,
however, associated with even more extreme assumptions and analytical distance
from the socioeconomic processes of growth itself.
Key words: Endogenous growth

JEL classifications: P16, O40
1. Introduction
As a topic, irrespective of its intellectual antecedents, endogenous growth theory is
extremely recent, only dating back in acknowledged published form to Romer’s (1986)
Manuscript received 12 January 1998; final version received 2 August 1998.

Address for correspondence: Ben Fine, Department of Economics, SOAS, Thornhaugh Street, Russell
Square, London WC1 0XG, UK; emailbf@soas.ac.uk
* SOAS. This is a much shortened version of Fine (1998), in which ample and detailed reference is made to
the literature to support and illustrate the arguments. Thanks to many colleagues and anonymous referees for
comments.
© Cambridge Political Economy Society 2000

246 B. Fine
article, which is usually coupled with Lucas’ (1988) contribution. 1 Within a decade, the
literature has itself taken off into sustained growth. 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 economics journals.
There are textbooks for endogenous growth theory, such as Grossman and Helpman
(1991) and Barro and Sala-i-Martin (1995).2 Surveys and special issues and sections in
journals proliferate.
It is now possible to posit a rough and ready ideal-type contribution to the literature. An
opening section might provide cursory discussion of some earlier contributions and an
overview of what is to come. Next comes a mathematical model. Typically, it includes a
production function for aggregate output, dependent on capital and labour as inputs.
This is, however, modified from the standard one-sector growth model by allowing for
productivity increase to be generated endogenously. This can depend, for example, upon
constant returns at the level of the individual firm but with positive spin-offs between
them. Alternatively, productivity might result from increasing returns to scale, whether
directly within the production function or through the application of produced R&D or
through the production of human capital.
It will be seen that the number of mechanisms for generating productivity increase is
without bound. What is important is that it depends upon choices made within the
model—how many resources are devoted to human capital or R&D, for example. In this
way, productivity increase and the growth rate both become endogenous in contrast to
these being exogenous as in the old growth theory.3 There is, however, another sense in
which the term endogenous is appropriate to the new growth theory. For, necessarily
associated with the variability in the growth rate as an effect, there is some variability in the
factors or parameters that generate growth. They become subject to choice—how much
productivity increase tomorrow, for example, in return for sacrifices in the form of
investment in R&D today?
Not surprisingly in view of more general developments within (mathematical) eco-
nomics, the model is liable to eschew a constant savings rate and to endogenise saving
by intertemporal utility maximisation over an infinite (or overlapping generation)
horizon. In many models, because of the endogenous productivity effects, the savings rate
is too low for Pareto-efficiency since individuals take no account of their spin-off effects
on others. This can then lead our article to discuss the efficiency effects and policy impli-
cations. Inevitably, standard results from growth theory concerning the efficacy of the
market no longer pertain—again, hardly surprising given the presence of market imper-
fections.
Such discussions depend upon solving the model, which is doubly demanding because
of the relatively sophisticated time-dependent optimisation and the potential for out-
comes to incorporate multiple equilibria and complex dynamics around steady-state
solutions. For much empirical work—an alternative to examining policy implications—
the details can be set aside and an equation extracted from the model for the purposes
of regression. The motivation here is to explain why growth rates differ, so that the
regression is usually applied to cross-country data with growth per capita as the
1
Solow (1991) marks the origins of endogenous growth theory with Romer’s doctoral thesis of 1983,
emerging as Romer (1986), and Lucas’ Marshall Lectures of 1985, giving rise to Lucas (1988).
2
Most recently, see Aghion and Howitt (1998), which unwittingly but comprehensively exposes the limita-
tions upon which the new growth theory is based.
3
For an account of the old growth theory, see Sen (1970).
Endogenous growth theory: a critical assessment 247
dependent variable. In this way, the impact of the selected mechanism or factor for gener-
ating endogenous growth can be assessed empirically.
On this basis, the scope for contributions is enormous. This had also been true of the
old growth theory, even though it was more or less confined to estimating the various
contributions to output from increases in factor inputs and an unexplained residual
credited to technical progress (so-called total factor productivity, TFP). Studies could
prosper as allowed by the data generated by the passage of time, and the extensive appli-
cation to more sectors, countries or more finely disaggregated factor inputs. The new
growth theory can also incorporate an expanding realm of theory, although much of the
empirical work is similar to the old TFP calculations, except that factor inputs, sufficiently
broadly interpreted to include growth-generating variables, can now explain—on a micro-
economic, market-imperfection basis—rather than simply measure productivity increase.
Individual saving, for example, might not only add to output by increasing the capital
stock, it also adds to productivity through generating production externalities.
As our representative article closes by summarising its main points, the new growth
theory is readily seen to embody much that is typical of contemporary economics as a
discipline. Consequently, the purpose of this paper is not only to provide an overview of
the literature, but also to set it against broader developments within economics. It fur-
nishes an excellent point from which to assess the state and momentum of the discipline.
This still leaves open the vantage point from which to make such an assessment. Else-
where (Fine, 1997),1 I have argued that economics is currently experiencing a revolution.
It is doing so by increasingly colonising the other social sciences from which it had
previously been insulated by its assumptions and methods; its axiomatic formalism as
in model-building; its methodological individualism with the narrow motivation of
economic agents restricted to utility maximisation; the notion of the ‘non-economic’,
including social institutions, as exogenously given as in technology and preferences; the
heavy reliance upon equilibrium as an organising principle; and the use of econometrics as
an unproblematic test of theories, not least in lieu of conceptual rigour and depth.
Paradoxically, these apparently insurmountable barriers between economics and the
other social sciences have been consolidated rather than weakened as economics thrusts
outwards to new analytical terrains. This is, however, often disguised by the informal,
non-technical ways in which notions such as human capital have been deployed across the
social sciences without careful regard to their substantive conceptual content and
analytical origins. Far from abandoning its traditional methods, mainstream neoclassical
economics has extended them in two complementary ways. The first, perhaps most
readily associated with Gary Becker, 2 has been simply to presume that individual utility
maximisation that had previously been confined predominantly to the market economy,
despite the generality of Bentham, should be applied as an analytical principle to every
aspect of life. The second, in a sense the more revolutionary, continues to be based upon
individual optimisation but now takes social institutions and structures as endogenous,
where previously they were assumed to be exogenous. In the light of the new micro-
economics based on transactions costs and imperfect and asymmetric information,
individuals may choose to create social structures, whether in labour, financial or
other markets. In this vein is created a new institutional economics, a new political
economy, a new development economics, a new financial economics, a new theory of
1
See also Fine (1999A, 1999C, 1998C).
2
See especially Becker (1996).
248 B. Fine
rural institutions, etc. New growth theory, at least in part, belongs to these new waves of
economics.
2. New growth for old?

How does the new growth theory fit into this expanding world of novelty? In the con-
cluding section, the relations between endogenous growth theory and other social
sciences will be assessed. For the moment, consider its position from within economics
alone. The central factor here is the microeconomic basis of endogenous growth theory.
This is true of the dependence of growth upon the optimising savings behaviour of
representative individuals rather than upon a systemic savings rate taken as a parameter
for the economy. Significantly, however, this is the only way in which demand tends to
enter the model. The microeconomics is primarily concerned with the supply side. This is
apparent in two different respects. First, endogenous growth theory is concerned with the
sources of productivity increase. As such, it has drawn upon microeconomic theories, as
in the use of resources to produce R&D, economies of scale and scope, the use of ‘human
capital’ as an input, or the externalities that spill over from one firm or agent to another.
These theories have nothing to do as such with an economy as a whole, other than in the
trivial sense of requiring at least two economic agents in order for exchange to arise.
Indeed, often implicitly and sometimes explicitly, the literature takes a microeconomic
theory and simply interprets it as (long-run) macroeconomics.
Secondly, by the same token, macroeconomics itself over the past two decades has
increasingly been organised around microeconomics and general equilibrium, most
notably in the use of representative individuals and the new Keynesian microeconomics,
in which there is a mixture of imperfections across market competition, differential access
to and use of information, and imperfect market clearing.1 In short, endogenous growth
theory is heavily implicated in drawing upon and strengthening the traditional micro-
economic foundations of neoclassical economics.
In what sense, then, is it new? A number of contributors have reasonably argued that
most of the ideas deployed are not new and are to be found in the old neoclassical growth
theory or other schools of thought.2 The originality lies more in bringing these ideas to the
fore and packaging them in the most advanced form of mathematical models.3 At a more
mundane level, the claim to newness resides, at least terminologically, in substituting
endogenous for exogenous. What was previously taken as given is now explained. It is
crucial to recognise, however, that this is a simple moving of the explanatory boundaries
outwards although, as will be seen, there is also some countervailing shrinkage in defer-
ence to model tractability. This is made especially clear by Hammond and Rodriguéz-
Clare (1995), who open their survey as follows:
1
For an assessment of the evolution of macroeconomics in these terms, see Fine (1998B, ch. 2).
2
A standard reference within neoclassical economics for a precedent for the new growth theory is Arrow
(1962). See also Solow (1997). For anticipation of the new growth from within the UK Cambridge tradition,
see Kaldor (1996). Note that Harcourt (1997) views Kaldor’s work and Stiglitz (1994), a representative of the
new information-theoretic neoclassical economics, as ‘not that far apart’. For a critique of this stance, and its
dangers for radical political economy, especially in the context of the World Bank’s Post-Washington con-
sensus inspired by Stiglitz, see Fine (1999B).
3
See Stern (1996, p. 74) for example: ‘Our summary of the endogenous growth theories . . . led us to the
conclusion that many of them constituted only modest advances on the earlier theories.’ As will be seen later,
assessments from other than a neoclassical perspective can even view new growth theory as a backward step.
Progress in economic science often takes the form of explaining what was previously inexplicable.
That is, variables which had earlier been treated as exogenous become endogenized. Their values
become determined, at least in principle, within an economic model.
In other words, new growth theory shares crucial aspects of methodology with its aged
parent. It requires a division between an unexplained exogenous and an explained
endogenous.1 In this respect, a standard history is now usually told of old growth theory,
presumably to new generations of students. Rarely is reference made to classical political
economy or to Marx. Rather, growth theory begins with Harrod–Domar and moves
forward to Solow/Swan. Here, it stagnates sometime in the late 1960s until endogenous
growth theory brings the subject back to life both by allowing sources of productivity
increase to be explained theoretically as well as by addressing the apparent differences in
per capita growth rates between countries, which ought to have been the same with a free
flow of production functions (or knowledge) between countries.
But the meteoric rise of endogenous growth theory does not derive simply from its
having made variable what was previously fixed. There is also the lifting of some of the
theoretical and empirical limitations imposed by the old theory. To be explicit, neo-
classical exogenous growth theory is essentially a theory of the contribution that increases
in factor inputs make to output in a perfectly functioning economy. Both theoretically and
empirically, it does not address how other factors might contribute to growth, nor what
impact might be made by market imperfections.
From this point of departure, endogenous growth theory has evolved on the basis of two
complementary sources. One has been to model the sources of productivity increase in
various ways. Quite apart from technological spillovers or learning-by-doing, which are
presumed to arise spontaneously out of accumulation itself, productivity increase has
been modelled by focusing upon the different stages in the generation and use of new
knowledge. Private and public resources can be devoted to R&D with greater or lesser
benefits to individual capitalists. Productivity increase can accrue through the quality and
range of intermediate inputs as well as in the greater output from a given level of inputs.
In particular, as a special case, the quality of labour can be enhanced through the
accumulation of human capital. This has itself been modelled in various ways, depending
upon how it is produced and how it is used—is it attached to work experience, leisure, or
public and private resources devoted to education and, once accumulated, how does it
affect overall levels of production from given inputs?
A crucial consequence of endogenous sources of productivity increase is that they
involve market imperfections. This has provided the second central component of
endogenous growth theory. With all agents optimising on the basis of given prices, an
associated equilibrium is not Pareto-efficient. This is a simple result of the externalities or
socially increasing returns to scale involved. Generally, it follows that the competitive
outcome induces a level of saving that is below the optimum, since private agents take no
account of the knock-on effects of corresponding levels of investment. Once there is an
endogenous growth mechanism in place, however, attention can focus on any market
imperfection that affects the saving rate and, hence, long-run prospects.
It is now possible to characterise endogenous growth theory in shorthand form as well
1
The division between exogenous and endogenous is often erroneously taken to be an essential aspect of
any theory. Such a myth is perpetuated within, and favours, mainstream economics in view of its axiomatic
methods in which explanation is reduced to internal workings on the basis of externally given parameters.
Concepts such as exploitation, power, class and globalisation are rendered devoid of substantive meaning,
together with their historically and socially specific roots.
250 B. Fine
as in its intellectual thrust. As recognised by many of its practitioners, it is simply a market
imperfections theory of technical change in which, in contrast to static general equilib-
rium or exogenous growth theory, the impact of the imperfections is felt on the rate of
growth rather than upon the level of output or welfare for a given rate of growth. The last
point, the transformation from level to rate of growth of output, is the only novelty. For
market imperfections are now deemed to have cumulative effects over time. Otherwise,
endogenous growth theory draws upon the two previously outlined traditions for its
conceptual content: the theory of technical change and the theory of market imper-
fections.
Each source provides a rich vein of raw analytical material. In examining the course of
the previously mentioned current revolution in economic thought, I have argued that it
has proceeded both by pillaging and simplifying the other social sciences for ideas in order
to accommodate them within a framework of methodological individualism and by
declaring as new what has previously been well known to other scholars. Such is the
general feature of the contribution of endogenous growth theory to the understanding of
productivity change. As suggested below, Schumpeter is one of the chief sources in this
regard. His work, both for initiating ideas, as in monopoly rents for innovation, and for
outcomes, as in clusters of innovations, has been plundered and formally reconstructed
on an individualistic basis.1 More generally, endogenous growth theory has drawn upon
the insights on productivity increase attached to Kaldor and Adam Smith, for example,
but all of the ideas used in endogenous growth theory are readily invented through casual
knowledge of technology invention, adoption and adaptation.
As regards the incorporation of market imperfections, endogenous growth theory is
essentially cannibalistic. The discipline can plunder itself for the sources of Pareto-
inefficient outcomes and translate these into sources for growth as opposed to deadweight
losses. Not surprisingly, particularly favoured sources are those that concern imperfect
competition, since innovation involves temporary monopoly rents, and any capital market
imperfections that affect the level, composition or use of savings and investment.
In view of the analytical sources for endogenous growth theory, there are two reasons
why it should give rise to conceptual backwardness. Where there is dependence upon
other social sciences or heterodoxy within economics, as in theories of technical change,
the content is generally stripped of its broader historical and social framework in defer-
ence to the requirements of the axiomatic model-building associated with methodological
individualism. The limitations imposed by drawing upon previous notions of market
imperfections are more technical in nature. It will be shown in Section 3 that, in the con-
text of the multiple equilibria and the dynamics attached to endogenous growth theory,
simple extensions of any basic model lead to extremely complex outcomes. On the one
hand, limited generalisations of basic assumptions suffice to generate a wealth of out-
comes. On the other hand, models become intractable beyond the point of the most
simplified models of the economy.
This gives rise to a paradox. Whilst the mathematical techniques and modelling
attached to the new growth theory are a level or two higher than those of the old, some of
the economic assumptions are often as simple if not simpler. It follows that in many
1
For a more rounded account of Schumpeter’s work, see the national systems of innovation literature, for
which Fine (1993) provides an exposition and assessment. It is only surprising that it should have taken so
long for Schumpeter to have been incorporated within mainstream neoclassical economics, since he is
credited with having coined the expression ‘methodological individualism’, and he favoured both it and
modelling for analytical purposes. See Machlup (1978).
fundamental respects endogenous growth theory represents a continuity rather than a
break with exogenous growth theory. For, many of the criticisms of the latter carry over,
even with greater force, especially since the scope for generalising endogenous growth
theory is limited by the greater technical demands involved in the corresponding models.
The standard assumption of a single sector for the economy is open to the devastating
consequences of the Cambridge critique. This was developed as an explicit response to
the old growth theory and as part of a more general assault on neoclassical economics. It
showed, convincingly even to its opponents, that the one-sector growth model and the
economic intuitions associated with it, are illegitimate. Standard results derived for the
relationship between output, capital and distribution (wages and profits) from a produc-
tion function do not carry over to models with more than one sector, and calculations of
TFP are fundamentally flawed.1
Endogenous growth theory has proceeded as if oblivious to such conundrums. Even
where more than one sector is assumed, often surreptitiously as in a separate sector to
produce human capital, very special assumptions are made about the production func-
tions or preferences concerned, such as Cobb–Douglas production functions and con-
stant intertemporal discount rates. Otherwise, there are severe problems of multiple
equilibria and instability which are, in any case, extremely common.
Endogenous, like exogenous, growth theory also tends to be organised around steady-
state balanced growth. This has a number of drawbacks, although some of this can be
accommodated within the more complex dynamics of endogenous growth theory. First,
in reality, growth is neither steady nor balanced. It is irregular and the composition of
output shifts significantly both in the early stages of industrialisation and in the mature
stages of post-industrial society. For this reason, there is considerable and justifiable
scepticism about the applicability of endogenous growth theory to the problems of
development, once compositional shifts in output are considered rather than aggregate
growth alone.2
Secondly, even with one sector for output, from a technical point of view, the require-
ment of balance in growth equilibrium is extremely demanding, since many variables
are constrained to grow at the same rate. The problem is that, in the absence of what
are often special assumptions, endogenising the growth rate can readily lead to its
becoming an explosively fast rate or to its being eroded to the exogenous rate over time,
thereby undermining the long-term thrust of the theory. In some respects, this is simply
Harrod’s existence problem—does the natural equal the warranted rate—raised to a
higher plane.
Thirdly, and again endogenous growth theory does occasionally address these issues on
a piecemeal basis (as discussed in Section 5 where government policy can be endogenised,
for example), economic growth is not synonymous with (economic) development, which
witnesses a range of social change with interactions with the economy—proletarianis-
ation, welfarism, urbanisation, demographic transition, etc.3 In the hands of endogenous
1
For the most detailed exposition, see Harcourt (1972, 1976) and Fine (1980, ch. 5) for a simpler expos-
ition. Harcourt (1995, ch. 4) places the Cambridge critique in context.
2
For this reason, not surprisingly for Syrquin (1998, p. 171): ‘Endogenous growth theory and its empirical
implementation could become highly relevant and influential for development economics, but that . . .
potential remains largely unrealised.’ And for Stern (1996, p. 74): ‘Growth in steady state is not a useful
description of agricultural change, either in terms of the detail of how occurrences, such as the “Green
Revolution” take place, or in terms of the changing role of agriculture in the economy.’
3
See Brinkmann (1995) for a critical discussion of the distinction between development and growth as
understood by neoclassical economics.
252 B. Fine
growth theory, these can be endogenised by expanding the scope of the model but with
two reservations. The model becomes too complex to handle, not least if economic agents
have to strategise consistently about the future consequences of their actions in both the
economic and the endogenous non-economic arena. The model is also precisely that—
once set in motion from initial conditions, it has a life of its own with a path that can only
be subsequently changed by random shocks. There is, otherwise, no scope for historical
contingency nor for differentiation between one period and another and from one society
to another.
Fourthly, quite apart from its fragmented forays into the socioeconomic factors that are
otherwise taken as exogenous, new growth theory is built up on the basis of method-
ological individualism in which agents optimise or otherwise engage in activity according
to more or less arbitrary behavioural patterns. Such an approach precludes the
endogeneity of social forces, structures and relations, whether these be economic or
otherwise, unless derived from aggregation over individuals. This is in sharp contrast to
classical political economy with, for example, Smith’s emphasis on the interaction
between a growing division of labour within the constraints imposed by the extent of the
market, and to Marx’s theory of the accumulation of capital as the driving force behind
productivity increase. In each of these cases, as for Ricardo’s theory of differential rent,
the value theory constructed is concerned to derive prices on the basis of a changing
technology.1 How is value formed when technology is changing—whether it be due to
growing division of labour, movement onto worse land, or the changing composition of
capital. A significant continuity between exogenous and endogenous growth theory is that
the latter incorporates shifting technology only on the basis of the same principle, even if
more complex in practice—that of present value discounting of streams of utility.
3. Opening a can of dynamics

As observed, endogenous growth theory has been heavily influenced by microeconomics.
Indeed, like much macroeconomics in the most recent period, it can be thought of as
microeconomics parading as macroeconomics, especially in the context of representative
economic agents whether serving as producers or consumers. Consequently, there is a
persistent analytical thrust and logic to proceed by disaggregating the economy in
conjunction with whatever market imperfections have already been incorporated. Having
treated your economy as if it were made up of a single production function and optimising
consumer, it makes sense to generalise or become more realistic by differentiating
between sectors or consumers.
To some extent, this has already been implicitly embarked upon in being a growth
theory which examines the economy over time. This means that production and con-
sumption at different times do have a different impact on the economy even if they do not
represent the supply and demand in entirely different markets as, in a one good world
for example, the goods involved are otherwise identical apart from timing. But, for
endogenous growth theory, consume or produce now or not and you affect not only the
resources available in the future but also their productivity.
More explicitly, the economic model is liable to be disaggregated by sector, even if in
limited and specific ways. This is inevitably the case where human capital or technical
1
For a full discussion of this point, see Fine (1982).
advance is produced with resources devoted to the purpose. The model will then have two
or more sectors, those attached to production of goods and those attached to producing
productivity, however directly. There can also be more than one physical good, whether
for capital, consumption, as an intermediate input, or for trade. In addition, it is possible
to disaggregate by differentiating between agents, whether as consumers or producer, a
point which will mainly be taken up in subsequent sections.
Another factor motivating endogenous growth theory is its capacity to draw upon a
wider range of model outcomes, not least in explaining why growth rates can differ even
with free flow of technology. Exactly what results the models are supposed to be able to
yield is open to question. Before endogenous growth theory, for example, with the
exception of those such as Goodwin and Kalecki, it was found to be mathematically
impossible for a single model to generate both growth and cycles, given the limitations of
difference equations when applied to Harrod–Domar through the use of a multiplier-
accelerator model. Both for analytical and empirical convenience, the short and long runs
have been perceived to be independent of one another. For endogenous growth theory,
the short and long runs are related to one another in the sense that the two are chosen
together without the latter being given independent of the former. Put another way, as
economic agents choose the growth path, they are also going to choose the path taken to
attain it and, of necessity, they will choose a different path if they choose a different
growth rate. In addition, in this context, it is possible for random shocks or initial
conditions—which are not chosen—to affect the long-run path.
It follows that the existence and stability of the endogenous growth path must be
determined together. In the first instance, this begs the question of the uniqueness of the
endogenous growth path. Not surprisingly, in view of the highly restrictive conditions
necessary for a static general equilibrium to be unique and stable, endogenous growth
theory is replete with uniqueness and stability problems even with the slightest opening up
of its microeconomic assumptions. Consider, first, the issue of multiple equilibria. As
Hahn (1990) observes in the context of growth theory, these are always liable to occur
in the presence of increasing returns. Even the slightest generalisation of a simple
endogenous growth theory suffices not only to allow for different rates of growth but also
to generate multiple equilibria. A common source in the literature for multiple equilibria
is through some form of reinforcement, with the coexistence of virtuous and vicious
circles as possible outcomes, in the context of market failure of some sort. In fertility
models, for example, parents may choose many children when productivity is low in order
to provide for their own present or future utility. If productivity is high, parents can
choose to have fewer children and are more than compensated through higher produc-
tivity growth. In models with research and development, multiple equilibria have arisen
out of its interaction with imperfect competition (high or low levels of rental returns to
innovation); and the labour market can be high human capital, high capital-intensity or
low–low, depending on agents’ choices (whether matching high or low skills are created
simultaneously). There can also be multiple equilibria arising out of threshold effects,
with the strength of externalities generating productivity assumed to be S-shaped or
limited at both low and high levels of development but high in-between.
The counterpart to multiple, or even to unique, equilibria is the dynamics to which they
are attached. The simple dynamics of the one-sector model gives way, at one extreme, to
the complex potential for oscillations around a variety of steady-state equilibria, with
outcomes depending upon initial conditions. It is also possible to generate a wide range
of outcomes for growth as well as for catch-up, overtaking and leapfrogging, as new
254 B. Fine
technologies with high potential for endogenous growth become introduced in low wage,
low productivity regions.1
4. Statistical conundrums
So far, the focus has been upon the analytical content of endogenous growth theory. But it
is commonly acknowledged that a major source of momentum to the topic has been
provided by the comparison of, and explanation for, differences in growth rates. No
doubt, this has been inspired by external events, such as the success of the newly
industrialised countries, for the apparently unacceptable conclusion from exogenous
growth theory of equal per capita growth rates is far from new. What is new is that the
proposition should be contested in the context of a theory of endogenous technical
progress. The capacity to do this has been greatly enhanced by the ready and cheap
availability of data sets and computing power.2 Initially at least, as in what has been
termed Barro-type regressions, following the seminal work of Barro (1991), the con-
vergence of per capita growth rates has been tested by taking them as a dependent variable
with per capita income as the independent variable in cross-section analyses. With a
negative sign in the regression taken as evidence both of convergence and as supportive of
exogenous growth, other independent variables can be thrown in either to correct for
country differences or, to the contrary, as indicative of sources of endogenous growth.
The result has been an explosion of econometric literature around endogenous growth
theory. Fischer (1994) reports from Levine and Renelt (1992) that 40 cross-sectional
studies of the new growth theory had already been conducted with well over 50 different
regressors. An issue that tends to be overlooked is that of the stability of the regression—
whether particular variables remain significant as others are added or omitted. Levine and
Zervos (1994) confirm an earlier analysis of cross-section studies, showing that regression
results depend very heavily on what independent variables are included, with financial
development and the black market exchange premium proving robust, positively and
negatively for their effect on growth respectively, but the same not applying to govern-
ment expenditure, fiscal deficit, inflation and trade. Even these limited results within, as
will be seen later in this section, what is a flawed statistical approach, leave the direction of
causation open to question.3
Following the work of Levine and his collaborators, a cloud of pessimism has darkened
the prospects of empirically discovering systematic sources of growth. This situation is
sharply illustrated by what appears to be an unwitting self-parody from one of the most
active new growth accountants. Sala-i-Martin (1997) attempts to retrieve robust results in
a paper entitled ‘I Just Ran Four Million Regressions’. 4 Essentially, his argument is that
1
Further, quite apart from dynamics and multiple equilibria, endogenous growth theory has occasionally
endogenised the economic structure itself, as structural change and crisis become endogenised in a way that
corresponds to the shift from Fordism to post-Fordism, as dedicated mass production gives way to flexibility
in response to higher levels of income and more sophisticated patterns of demand. See You (1994), Cornwall
and Cornwall (1994), Driver (1996), Lombardini (1996) and Lordon (1997).
2
See Summers and Heston (1988, 1991).
3
Summaries of results from regressions are provided in Sala-i-Martin (1996A, 1996B), but also Ochoa
(1996) in a more critical discussion. Econometric work continues to flourish without bound.
4
In principle, this is an understatement, as Sala-i-Martin (1997, p. 6) seeks to economise by use of ad hoc
methods—always retaining three of the variables fixed in the regressions. Otherwise: ‘If I tested one variable
and allowed the other 62 to be combined in groups of 6, I would have to estimate 61 million regressions per
variable tested. This would sum to a total of 3·9 billion regressions. My computer can estimate about 2,000
regressions per minute so it would take about 4 years to estimate all these models.’ Note, for variables
combined in groups of eight, 30 billion regressions are needed over a 30-year period.
variables need not be significant in every combination of others included within a
regression. Rather, they should be significant for a sufficient number of regressions.
This all begins to look like statistics without theory other than as an initiating impulse.
But such procedures are commonplace in the use of econometrics. A theory is used to
derive a simple equation to which a range of modifications, including the addition of error
terms, are made prior to statistical testing. There are serious problems with this. First, the
independent variables in this context will inevitably be related to one another, since the
correlates of growth are systematically connected, quite apart from the mutuality of
dependent and independent variables. Education, for example, is both a source and a
consequence of growth, as well as of trade performance, etc. Second, the econometrics is
highly selective in terms of the relations that it does examine as opposed to those that it
does not. Insofar as it only focuses on growth rate outcomes as opposed, for example, to
the processes by which those outcomes are achieved, there is a neglect of the models’
implications, which may not be borne out by the data. Simple regressions between growth
rates and per capita income, for example, may suggest convergence even though changes
in TFP might suggest otherwise with growing productivity differences between countries.
Third, it is important to stress how stochastic variation is generated in the model since, for
endogenous growth theory, random shocks can have a persistent and varied impact over
time, depending upon how they arise and how they are transmitted.
These and other issues are becoming the bread and butter for a more sophisticated
econometrics, which has paid closer attention to time-series and panel data estimation
over the most recent period. As a consequence, endogenous growth theory has been
a fertile application for advances in econometrics in which complex dynamics and
stochastic properties can be more fully examined. The base-line conclusion of such work
is that you cannot have your cake and eat it. For the favourable consequences from
endogenous growth theory in terms of the richer variety of empirical outcomes for which it
allows (differences in growth rates) is necessarily coupled with a dynamics that renders
simple Barro-type regressions irrelevant if not erroneous. Thus, in the good old exogenous
growth theory, the steady-state balanced growth path could easily be identified, as could
the relatively simply dynamics around it. Moreover, it could be readily used to estimate
sources of growth other than those from factor inputs alone. Other implications from the
theory itself did not need to be tested—the inverse relation between capital–labour ratios
and wages, for example. Indeed, the limited scope for different growth paths within the
old growth theory can be used as a basis for criticising it because of its failure to match the
conventional wisdom attached to the empirical experiences of growth, those associated
with Kaldor and Kuznets for example.
Nonetheless, as suggested by other experience within economics, as for the Cambridge
controversy over the measurement of capital and TFP for exogenous growth, the presence
of unsavoury theoretical and empirical implications for conventional wisdom and prac-
tices seems to pose no impediment.1 The theory and measurement of endogenous growth
and convergence by standard and erroneous methods is almost certainly assured in the
classroom and will be applied by generations of new students for the foreseeable future.
More specifically, it is apparent that even the simplest extensions of the models of
endogenous growth have readily given rise both to multiple equilibria and to complex
dynamics to or around them. If this result had been taken as a starting point for applied
1
Most recently, the reinterpretation of the Asian miracle as an otherwise unexceptional accumulation of
factor inputs has been innocently founded on totally flawed calculations of TFP.
256 B. Fine
work, much of the recent econometric investigation of growth might have been con-
siderably different. For it has been standard practice to employ Barro-type regressions
and to deduce both convergence and rejection of endogenous as an alternative to
exogenous growth theory for appropriate and significant coefficients in a simple
regression. First, observe, however, that the two theories of growth are not being tested
against one another properly. This is simply demonstrated by Quah (1995), for example,
who legitimately points out that if changes in growth rates were randomly but identically
distributed across countries, then there would, by virtue of Galton’s regression to the
mean, be a correlation between higher growth rates and lower income. For those
countries that had experienced high (low) growth in the past would inevitably tend to
grow slower (faster) in the future. This is a simple demonstration that testing growth
models in the context of convergence involves some model of growth as well as assump-
tions about the sources of disturbances both for a representative country and across
countries. In addition, in comparing growth rates across countries, we need to know
whether each is moving on or towards its steady state and whether from above or below.
At a more complex level, then, convergence cannot be tested without some underlying
model incorporating both the sources of growth and the sources and impact of stochastic
disturbances. Necessarily, this has led to estimates and comparisons of the properties of
the various time series involved. Endogenous and exogenous growth models not only
generate different growth rates but also different patterns of growth over time. Most of
the studies in this vein have shared four features in common. First, they reject simple
Barro-type regressions as a way of testing either for some definition of convergence and/or
for endogenous versus exogenous growth, some explicitly recognising that the two
hypotheses are not synonymous. Secondly, they have studied the time series properties of
the growth process, even if often drawing upon panel data across countries, or from
regions within countries, for statistical testing. Thirdly, these studies are often supportive
of convergence and exogenous growth when examining time-series properties. This is
because the latter need to reveal persistent effects from productivity or other shocks, and
these, not surprisingly, are rarely confirmed, since it would require any deviation of the
economy from a growth path to send it onto another growth path. Fourthly, however,
whilst the discussion has moved beyond simple Barro-type regressions, even if these
remain popular, much of the empirical literature continues to proceed as if the theoretical
results concerning multiple equilibria simply do not exist. In short, Quah (1996, p. 1053)
concludes:
With hindsight, the key point . . . is obvious. Convergence concerns poor countries catching up with
rich ones. What one wants to know here is, what happens to the entire cross-sectional distribution of
economies, not whether a single economy is tending towards its own, individual steady state.
However, it is the latter that has preoccupied the traditional approach.
But a more general, and slightly different conclusion emerges from this literature. It is that
simple models either cannot even address simple questions or that they become too
complicated even before relatively few explanatory variables are introduced. Endogenous
growth theory is in a shambles of its own making as the formal models intended to paint
the broad-brush features of growth have become ensnared in unduly demanding
statistical issues relative to the original intention.
5. The favoured terrain of applications
Given that endogenous growth theory draws upon the microeconomics of market imper-
fections and technical change, it has enormous resources upon which to draw. To some
extent, this makes its content arbitrary and subject to the whims of individual contrib-
utors. Nonetheless, the theory has reflected old traditions and the continuing momentum
of economics as a discipline. Here, five separate areas are delineated.
5.1 Asset-stripping Schumpeter

Since one strand of endogenous growth theory has been concerned with modelling
productivity change, it is hardly surprising that the literature should have rediscovered
Schumpeter and, occasionally, Adam Smith, for whom a growing division of labour as a
source of productivity increase is readily reinterpreted in terms of economies of scale,
scope and intermediate product range. Three particular aspects of Schumpeter’s work
figure prominently.1 One is the notion of clusters of innovation that are heavily associated
with externalities in the production and/or use of knowledge, as in technological spill-
overs. Secondly, reference is made to imperfect competition and the extent to which the
benefits of innovation accrue in the form of rents to the firms concerned. Here, there is a
balance of effects. For, if innovation spreads quickly so does the general level of product-
ivity. On the other hand, this means that the incentive to innovate, in terms of temporary
surplus profits, is undermined. At one extreme, for example, if diffusion of innovation
were costless and timeless, no one would bother to invent. At the other extreme, if all
innovations were monopolised, there could be no general productivity increase from
spillover. Now, the processes of technical change are extremely complex and varied,
ranging over invention, innovation, adoption, adaptation, learning by doing this, that and
the other, and they depend upon how resources are used, financed or spillover, etc. It is
even possible to eschew steady-state growth and to model waves of destruction as an
efficient response to the need to sacrifice production in the short run in order to gain
higher productivity into the future on the basis of new products, inputs or production
processes. Thirdly, Schumpeter is a suggestive source of dynamics, with waves of creation
and, at times, destruction. But, in this and in other areas in which mainstream economics
deals with the future, the unknown or the invented are reduced to risk rather than
uncertainty. A firm, for example, and so the economy as a whole, chooses to go through
upheavals as a means of profit-maximisation on the basis of knowable outcomes attached
to probability distributions around the future. As for Keynes’s notion of expectations in
macroeconomics, so for Schumpeter’s understanding of long-run growth, their analytical
rhetoric is adopted but any systemic content is systematically reduced to the optimising
behaviour of aggregated individual agents.
In short, endogenous growth theory as a theory of productivity increase has two bounti-
ful and overlapping sources for ideas, which can be incorporated in one or another way
into a formal model and be tested empirically. Either ideas, usually piecemeal and most
notably from Schumpeter, can be purloined from earlier literature, or a more or less
arbitrary but highly specific source of technical change is casually invented through some
economic mechanism attached to human capital, produced R&D, spillover, or whatever.
On this basis, the theory is able to broaden its scope of application and even putatively
confront a range of policy issues as revealed in following sections.
1
Solow (1991) significantly entitles a section, ‘Formalizing Schumpeter’.
258 B. Fine
5.2 New trade theory
New trade theory is little more than old infant industry arguments extended to a broader
canvas. Are the benefits of protecting domestic economic activity outweighed or not by
the disadvantages of discouraging external economic influences? Of course, in the
simplest version of this issue, it is a matter of setting, for example, increasing returns to
scale in domestic production against the availability of temporarily cheaper imports.
Consequently, all the ingredients of endogenous growth theory are inherently present,
given the market imperfections necessarily attached to scale economies.
It is, then, a simple and natural step for trade and endogenous growth theory to flourish
together, just as Edgeworth boxes translate seamlessly from the domestic to the inter-
national economy. The only conceptual issue, one that is rarely addressed explicitly, is
how to distinguish one country from another. Traditionally, the answer has been, as in the
Heckscher–Ohlin theory and its off-shoots, the extent of factor endowments and mobility,
with the country or nation-state otherwise treated as an optimising agent or set of agents
insofar as the interests of capital and labour, for example, are distinguished. For endogenous
growth theory, this has two implications. First, the closed economy model is simply
carried over in some form to a trade model, with individual agents designated as
countries. And secondly, the exact nature of the model depends upon market imper-
fections and the corresponding sources of productivity increase deployed. As a corollary,
it follows that a theory of international trade has not been constructed in anything other
than name because of the weakness with which separate countries have been specified and
distinguished. This is apparent in that exactly the same analysis can be brought to bear
upon intra-country growth and trade as in study of uneven regional development.
Once again, whether tied to more or less sophisticated econometrics, endogenous
growth theory in the context of trade is characterised by a proliferation of models based on
simplistic microeconomic notions with complex and varied outcomes. Policy impli-
cations, especially as regards trade liberalisation, depend upon the mechanisms through
which endogenous growth is generated and the market imperfections to which they are
attached. It is hardly surprising that free trade tends to be favoured insofar as this will
allow for the greatest scope for spillover and other effects to accrue, although it is possible
for either poor or rich countries to lose, depending upon the relative impact of scale
economies, catch-up and first-mover advantages. Whilst trade and other policy measures
can be considered together, as in levels of education expenditure, endogenous growth
theory in the context of trade policy rarely considers the portfolio of policies that might
promote comparative advantage in particular sectors. Trade policy, more closely asso-
ciated with productivity increase, tends to be considered in isolation from that other
concern of the Washington consensus, fiscal balance, even if this has itself been a focus for
endogenous growth theory in its own right in the context of intertemporal optimisation of
the incidence of government expenditure and taxation.
5.3 Accounting for the state

As in the case of the new econometrics of growth, the literature is heavily influenced by the
fashions of the time. This is also true of much theoretical work as will be evident in the
following applications. For, an important theoretical development in the passage between
Solow/Swan and Romer has been the rise to prominence of intertemporal optimisation
instead of the behavioural assumption of a constant saving rate. Consequently, growth
theory, as it deals with the long run, has been attracted by problems in which inter-
temporal optimisation was already in place.
Thus, fiscal issues have increasingly been set on a microfoundations basis with the state
optimising intertemporally in the context of other optimising agents and in the presence of
market imperfections. The standard result in public finance is that taxation should not be
levied on capital, the accumulable factor of production, since this will reduce the incentive
to save, and hence, the level of investment and growth. Instead, taxation should be levied
on fixed or unalterable factors of production, such as labour, since the level of supply will
not be affected. Such considerations continue to apply, even with greater force, in case of
models of endogenous growth since the laissez-faire level of saving is already too low
relative to the social optimum because of externalities or whatever.
Such conclusions no longer hold once a number of standard assumptions are dropped.
First, if human capital is involved, this can be accumulated and is no longer an
exogenously given factor of production. Secondly, there may be imperfections in par-
ticular asset markets, especially those for human capital in which it may not be possible to
borrow and consume now on the basis of future higher earnings after human capital has
been accumulated. Thirdly, this is especially so if individuals are not treated as if they
were infinitely long-lived (or ‘dynastic’ in optimising for future generations). In over-
lapping generation models, children both serve to provide utility directly in the present
and in the future, and are a source of future earnings. The desired, let alone the optimal,
path of consumption may not be sustainable if it is not possible to borrow on children’s
future earnings. In short, prospective productivity increases may lead to the desire to
bring consumption forward from the income to be earned later. This may be constrained
by the form taken by assets. The exact outcome, however, depends upon how produc-
tivity increasing (human) capital is accumulated and passed on, how it contributes to
productivity, and how utility is distributed across generations.
In each of these cases, as in static models, there is a role for government subsidies where
social exceed private returns and, where there is expenditure, there must also be taxation.
In addition, savings behaviour will be affected by how government taxes, and productivity
can also be affected directly by government expenditure on education and R&D. Models
of the fiscus, then, in the context of endogenous growth theory are little more than an
exercise in intertemporal optimisation, with outcomes dependent upon how productivity
is generated and how the welfare of future generations is linked to those of the present so
that appropriate taxes and subsidies can be calculated. Such literature, apart from
qualifying standard results derived in the absence of market imperfections, is notable for
its failure to consider the political and practical issues attached to fiscal policy which seem
to have been set aside. As in other applications, and the core models themselves,
endogenous growth theory depends upon gross simplification despite the technical
complexity that results.
5.4 Money, finance and growth

The same applies to the inclusion of money and finance into endogenous growth theory.
Their introduction has been marked by three different approaches—one includes real
money holdings in the intertemporal utility function, another requires there to be cash
holdings for production purposes, and the third relates money to transaction costs. Whilst
each of these reflects relatively crude understandings of money at a microeconomic level,
they are readily translated into a macroeconomic context. With inflation resulting from
excessive increases in the money supply, the result is equivalent to a tax on future income,
which places a premium on present consumption, serving to decrease the saving and
growth rates.
260 B. Fine
The role of financial intermediation has been modelled in a more sophisticated way
with deeper roots in microeconomic considerations, exploiting the role of money as an
asset in the presence of market and informational imperfections. As in other applications,
high- and low-level equilibria can be induced by matching sophisticated/crude levels of
financial services with specialisation or not, and high/low productivity in investment,
thereby modelling the relationship between finance and industrialisation. It takes little
sophistication, then, to extend theories of financial assets to endogenous growth in terms
of the levels and effectivity of saving, borrowing and investment. Any short-run negation
of the neutrality of money, as in most macroeconomics, especially in the presence of
micro- as macro- imperfections, is readily translated into growth effects through a variety
of indirect mechanisms as the level and composition of saving and investment are
affected.
5.5 The political economy of growth

In a previous illustration, endogenous growth theory has been linked to fiscal questions in
which the government is either taken as a benevolent optimiser or an arbitrary tax
authority. Not surprisingly, models have also been developed in the style of the new
political economy in which government policy does itself become endogenous. As such
models depend upon the balance of vested interests, they inevitably involve differentiation
between individuals. Some analyses are relatively simple, associating lower investment
with greater social instability, and the latter with inequality. At a more sophisticated level,
an endogenous growth model is coupled with an endogenous political process, in which
heterogeneous agents determine policy (for example, by reference to the tax level pre-
ferred by the median voter), with such policies subsequently affecting growth and stratifi-
cation and so on. More generally, it has been argued that inequality is detrimental to
growth in a democracy because resulting redistribution through the fiscus reduces the
saving rate (rather than raising the saving rate through higher propensities to save from
those on higher incomes). Further, if the process of endogenous growth disadvantages
some economic interests, unskilled workers, for example, then there is an incentive for
that group to use its political power to obstruct change. In other words, politics is simply
economics by other means.
6. Concluding remarks
From the review to be found in the previous sections, it is obvious to the point of tedium
that endogenous growth theory is rapidly expanding its scope both extensively and
chaotically, if not arbitrarily, in ways that strongly reflect the established norms of the
discipline, both in its formal modelling and in its econometric techniques. The wide
variety of models that can be deployed and the different degrees of sophistication with
which they can readily be tested empirically suggest that endogenous growth theory has
extremely promising prospects into the indefinite future.
In this light, it is worth dwelling on two features of the literature. First, as it has been
built upon microfoundations but aims at addressing the macroeconomy, it is always liable
to be hopping between the two, with differentiation or disaggregation by sectors or agents
to a greater or lesser degree. For Solow (1991), by analogy with the natural sciences and
Stephen Hawking’s A Brief History of Time, it is a matter of getting the balance right
between what is taken as exogenous and endogenous and realising the limitations of a
partial theory as approximation. There can be little doubt of the exaggerated claims for so
partial an understanding as is provided by endogenous growth theory. Solow (1991) is
particularly concerned with the long run in which so much is taken as given or in which
the grandly endogenous, such as stages of capitalism and shifting social institutions, are
tied exclusively to the most simplistic optimising behaviour. Such cautionary notes, even
if within the neoclassical paradigm, are most welcome but are equally likely to be ignored
as models based on simple intuitions are worked out mathematically and tested empiri-
cally against the conveniently available large data sets across regions and time.1
Second, endogenous growth theory inevitably leads to policy discussion. As Bertola
(1994, p. 103) observes, ‘almost every paper on endogenous growth addresses issues
of policy’. Yet, no policy consensus has emerged nor are the practical implications of
endogenous growth theory readily applicable in practice. It is surprising analytically, if not
ideologically, that endogenous growth theory should not have been more readily attached
to state intervention, given its dependence on market imperfections. This is most notable
in trade theory, where the advantages of international spillovers have to be set against the
benefits of protected markets. Yet, it is precisely the abstract, formal and highly aggre-
gated content of endogenous growth theory, together with its multifarious sources and
effects which, in contrast to monetarism and Keynesianism, say, lead to policy ambiguity
and imprecision—the more so with the slightest degree of complexity. The movement to
detail necessary for policy formulation—levels of education or R&D expenditure—more
or less render endogenous growth theory impracticable. It tells us what we know
already—market imperfections, R&D, education, etc.—are important to growth.2
The micro/macro bridge and the policy imprecision characteristic of endogenous
growth theory allow it to prosper through proliferation in models and competing
theoretical, policy and empirical conclusions. Perhaps this is just as well for the orthodoxy
for, in some respects, endogenous growth theory is highly radical in its implications for the
mainstream, despite lying centrally within it. Endogenous growth theory is for the long
run what the theory of the second best is for the short run. The broad thrust of the new
growth theory is highly unpalatable to conventional thinking and to many of its policy
nostrums. There is a case for state intervention to correct market imperfections in the
long-run, which is not given; the latter cannot be readily disentangled from the short run,
given path-dependence undermining the notion of natural rates of unemployment and
growth, let alone that these are determined by fundamentals on the supply side; even the
short-run non-neutrality of money can affect the growth rate and, thereby, undermine its
long-run neutrality.
By the same token, in the wake of the new revolution in economics discussed in the
opening remarks, the potential for new growth theory to provide a platform to launch an
internal critique against the orthodoxy, has its counterpart in posing a profound challenge
against radical political economy. It has previously done so in two different ways. First,
the stumbling block for orthodox growth theory has been taken to be its inability to
explain the simplest facts about economic growth, whether it be the patterns of con-
vergence and divergence or the stylised facts associated with Kaldor. Now, it is apparent
that models can be built which have the potential to accommodate short- and long-run
patterns of variables to order. More or less any stylised facts can be modelled. Where
stylised facts used to be an embarrassment to the orthodoxy, they are now a welcome
1
See also Solow (1997). For a brief critical assessment of the limitations of endogenous growth theory, see
Pasinetti (1994).
2
See Solow (1997) for the policy limitations of the new growth theory, with an implicit focus on the futility
of basing macro propositions on micro insights of limited validity.
262 B. Fine
challenge and invitation. It is clear in retrospect that the conventional wisdom on such
matters is an informal counterpart to results derived from ahistorical formal models.1
The second challenge, then, to radical political economy lies in the fact that endogenous
growth theory has been able to incorporate the very variables that neoclassical economics
has been criticised for omitting, those relating to shifting productivity, institutions and
market imperfections more generally. Of course, it can be argued that this is advantageous
to radical political economy, since the mainstream is acknowledging the importance of its
insights, not least in breaking down the rigid separation both empirically and theoretically
between the short run and the long run, and in broadening the scope of the (endogenous)
economic. The consequence, however, as has happened in the economics of labour
markets for example, is more likely to be the merging of the radical and mainstream which
become increasingly indistinguishable. 2 In this light, the tradition laid down by Kaldor,
and the same applies to Schumpeter, is ambiguous in its support for political economy.3
For, in the past, they could have been claimed as bulwarks against the orthodoxy, with
their lack of formalism leaving open the role of broader socioeconomic determinants.
Now their ideas have been captured by a neoclassical economics that closes off the
indeterminacy that had previously been left unspecified.
These remarks are well-illustrated by a small but significant stream within the literature
emanating from the political economy camps. These tend to welcome the new growth
theory for recognising factors that have long been the concern of political economy, but
they also deplore what is often a lack of or only cursory acknowledgement of their intellec-
tual debt, as well as the failure to embrace more fully from antecedents the specific
insights or their broader analytical context. A telling illustration is provided by Kurz
(1997), who employs the device of an imagined conversation between Ricardo and Smith
on the limitations of the new theory.4
A natural, legitimate but insufficient response by radical political economy to these
issues is to appeal to the informality of its own approach and the corresponding ability to
deal with historically and socially specific circumstances. The latter, however, have to be
wedded to a deeper understanding of socioeconomic processes. At the theoretical level,
outcomes must be understood as the complex and contradictory consequences of the
underlying socioeconomic processes attached to the accumulation of capital. At the
empirical level, this implies that trends or other movements in economic variables should
not be seen as the more or less complex balance of mechanically realised socioeconomic
processes, as in Verdoorn-type analysis, but as their historically and socially contingent
resolution. Simply referring to endogenous productivity, monopoly, institutions, money
and finance, the patterns of growth and cycles, conflict, inequality, etc. is no longer
sufficient to defend radical political economy against the neoclassical orthodoxy. The
latter can accommodate them all and more. It is more a matter of going back to
methodological first principles, and placing the social, the historical and the forces within
the economy at the forefront in order to be able to combat both methodological indi-
vidualism and the analytical strategy of creeping endogeneity. The orthodoxy, especially
in endogenous growth theory, has side-stepped these issues by insisting upon more and
more sophisticated mathematics and statistics, a departure of assumptions even from
1
See Wulwick (1993) for the notion of endogenous growth theory as the excessive formalisation of Kaldor,
both mathematically and statistically.
2
See Fine (1998B, ch. 4).
3
For an account of, and referencing to, Kaldor’s contribution, see Harcourt (1997).
4
See also, for example, the dissenting contributions collected in Coricelli et al. (1998).
basic descriptive narrative, and number-crunching on these flimsy foundations. Such
weaknesses in method must be exposed and opposed rather than embraced on the basis of
alternative assumptions, models and variables.
Finally, if endogenous growth theory is gobbling up radical political economy rather
than deferring to it, how is it relating to the other social sciences? It is apparent from the
previous sections that endogenous growth theory is able to incorporate the political
(voting) and the social (stratification). Simply renaming the unit comprising the economic
agent, as already seen, allows endogenous growth theory to spread to geography and to be
deployed in models of urban agglomeration, decentralisation and core–periphery models.
The environment has also provided scope for the application of endogenous growth
theory, not least because of its potential attachment to intertemporal optimisation. Can,
for example, productivity increase proceed fast enough to compensate for, or hold back,
ecological decay, and how should taxes induce optimality in the context of both environ-
mental and growth externalities?
Despite these forays into the other social sciences, endogenous growth theory advances
for the moment only with difficulty, not least because of the extreme formalism with
which it is characterised. On the other hand, as observed, endogenous growth theory has
already been extended to a political economy of growth in which interest groups, voting,
conflict and inequality have been incorporated, and other disciplines have previously
demonstrated an interest in the non-economic determinants of growth without neces-
sarily referring to endogenous growth theory.1 Thus, although endogenous growth theory
has yet to cut deep inroads into the other social sciences,2 and its impact has not yet been
marked in more general discourse, 3 it might only be a matter of time for the technically
demanding to filter across—as happened in the case of human capital. In short, as far as
the revolution in economics is concerned, endogenous growth theory might not be in the
vanguard, but it is certainly liable to be one of the new wave of following colonisers.
Bibliography
Aghion, P. and Howitt, P. 1998. Endogenous Growth Theory, Cambridge, MIT Press
Arrow, K. 1962. The economic implications of learning by doing, Review of Economic Studies, vol.
28, no. 1, 155–73
Barro, R. 1991. Economic growth in a cross-section of countries, Quarterly Journal of Economics, vol.
106, no. 2, 407–43
Barro, R. and Sala-i-Martin, X. 1995. Economic Growth, New York, McGraw-Hill
Becker, G. 1996. Accounting for Tastes, Cambridge, Harvard University Press
Bertola, G. 1994. Flexibility, investment, and growth, Journal of Monetary Economics, vol. 34, no. 2,
215–38
Brinkmann, R. 1995. Economic growth versus economic development, Journal of Economic Issues,
vol. XXIX, no. 4, 1171–81
Coricelli, F. DiMatteo, M. and Hahn, F. (eds) 1998. New Theories in Growth and Development,
London, Macmillan
Cornwall, J. and Cornwall, W. 1994. Growth theory and economic structure, Economica, vol. 61,
no. 242, 237–51
1
As in Haggard and Webb (1993).
2
In the context of the industrial revolution, Crafts (1995, p. 772) judges that: ‘Growth theorists have . . .
found useful ways of formalizing ideas long discussed by economic historians, and the way may now be open
for some fruitful interaction between economics and economic history.
3
As Crafts (1996, p. 30) observes: ‘In a speech in the autumn of 1994, the Shadow Chancellor, Gordon
Brown, referred to “post-neo-classical endogenous growth theory”. The press seized upon this phrase and
lampooned Mr. Brown.’
264 B. Fine
Crafts, N. 1995. Exogenous or endogenous growth—the industrial revolution reconsidered, Journal
of Economic History, vol. 55, no. 4, 745–72
Crafts, N. 1996. Post-neoclassical endogenous growth theory—what are its policy implications,
Oxford Review of Economic Policy, vol. 12, no. 2, 30–47
Driver, C. 1996. Stagnation and a problem of transition—arguments and proposals, Cambridge
Journal of Economics, vol. 20, no. 5, 553–64
Fine, B. 1980. Economic Theory and Ideology, London, Edward Arnold
Fine, B. 1982. Theories of the Capitalist Economy, London, Edward Arnold
Fine, B. 1993. Economic development and technological change: from linkage to agency, in
Liodakis, G. (ed.), Society, Technology and Restructuring of Production, Athens, V. Papazissis
Fine, B. 1997. The new revolution in economics, Capital and Class, no. 61, Spring, 143–8
Fine, B. 1998A. Endogenous growth theory: a critical assessment, SOAS Working Paper in
Economics, no. 80
Fine, B. 1998B. Labour Market Theory: a Constructive Reassessment, London, Routledge
Fine, B. 1998C. The triumph of economics: or ‘rationality’ can be dangerous to your reasoning, in
Carrier, J. and Miller, D. (eds), Virtualism: the New Political Economy, London, Berg
Fine, B. 1999A. From Bourdieu to Becker: economics confronts the social sciences, International
Papers in Political Economy, vol. 5, no. 3, 1–43
Fine, B. 1999B. The developmental state is dead—long live social capital, Development and Change,
vol. 30, no. 1, 1–19
Fine, B. 1999C. A question of economics: is it colonising the social sciences? Economics and Society,
vol. 28, 403–25
Fischer, R. 1994. Growth: the role of macroeconomic factors, in Baldassarri, M. et al. (eds),
International Differences in Growth Rates: Market Globalization and Economic Areas, London,
Macmillan
Greenaway, D., Bleaney, M. and Stewart, I. 1991. Companion to Contemporary Economic Thought ,
London, Routledge
Grossman, G. and Helpman, E. 1991. Innovation and Growth in the Global Economy, Cambridge,
MA, MIT Press
Haggard, S. and Webb, S. 1993. What do we know about the political economy of economic policy
reform, The World Bank Research Observer, vol. 8, no. 2, 143–68
Hahn, F. 1990. Solowian Growth Models, in Diamond, P. (ed.), Growth/Productivity/Unemployment:
Essays to Celebrate Bob Solow’s Birthday, Cambridge, MA, MIT Press
Hammond, P. and Rodrigu\O(e)z-Clare, A. 1995. On endogenising long-run growth, in Andersen,
T. and Moene, K. (eds), Endogenous Growth, Oxford, Blackwell, reproduced from Scandinavian
Journal of Economics, 1993, vol. 95, no. 4
Harcourt, G. 1972. Some Cambridge Controversies in the Theory of Capital, Cambridge, Cambridge
University Press
Harcourt, G. 1976. The Cambridge controversies: old ways and new horizons—or dead end? Oxford
Economic Papers, vol. 28, no. 1, 25–65
Harcourt, G. 1995. Capitalism, Socialism and Post-Keynesianism, Aldershot, Edward Elgar
Harcourt, G. 1997. Economic theory and economic policy: two views, Discussion Paper, no. 369,
Australian National University, Centre for Economic Policy Research
Kaldor, N. 1996. Causes of Growth and Stagnation in the World Economy, Cambridge, Cambridge
University Press
Kurz, H. 1997. What could the ‘new’ growth theory teach Smith or Ricardo? Economic Issues, vol. 2,
part 2, 1–20
Levine, R. and Renelt, R. 1992. A sensitivity analysis of cross-country growth regressions, American
Economic Review, vol. 82, no. 4, 942–63
Levine, R. and Zervos, S. 1994. What we know about policy and growth from cross-country
analysis, in Pasinetti, L. and Solow, R. (eds), Economic Growth and the Structure of Long-Term
Development, London, Macmillan
Lombardini, S. 1996. Growth and Economic Development, Cheltenham, Edward Elgar
Lordon, F. 1997. Endogenous structural change and crisis in a multiple time-scales growth model,
Journal of Evolutionary Economics, vol. 7, no. 1, 1–21
Lucas, R. 1988. On the mechanics of economic development, Journal of Monetary Economics, vol.
22, no. 1, 3–42
Machlup. F. 1978. Methodology of Economics and Other Social Sciences, New York, Academic Press
Ochoa, O. 1996. Growth, Trade and Endogenous Technology: a Study of OECD Technology, London,
Macmillan
Pasinetti, L. 1994. The structure of long-term development: concluding comments, in Pasinetti, L.
and Solow, R. (eds), Economic Growth and the Structure of Long-Term Development, London,
Macmillan
Quah, D. 1995. Galton’s fallacy and tests of the convergence hypothesis, in Andersen, T. and Moene,
K. (eds), Endogenous Growth, Oxford, Blackwell, reproduced from Scandinavian Journal of
Economics, 1993, vol. 95, no. 4
Quah, D. 1996. Twin Peaks: growth, and convergence in models of distribution dynamics, Economic
Journal, July, vol. 106, 1045–55
Romer, P. 1986. Increasing returns and long-run growth, Journal of Political Economy, vol. 94, no. 5,
1002–37
Sala-i-Martin, X. 1996A. The classical approach to convergence analysis, Economic Journal, July,
vol. 106, 1019–36
Sala-i-Martin, X. 1996B. Regional cohesion—evidence and theories of regional growth and
convergence, European Economic Review, vol. 40, no. 6, 1325–52
Sala-i-Martin, X. 1997. ‘I Just Ran Four Million Regressions’, mimeo
Sen, A. (ed.) 1970. Growth Economics, London, Penguin
Solow, R. 1991. Growth theory, in Greenaway, D. et al. (eds), Companion to Contemporary Economic
Thought, London, Routledge
Solow, R. 1997. Learning from Learning by Doing: Lessons for Economic Growth, Stanford, Stanford
University Press
Stern, N. 1996. Growth theories old and new: and the role of agriculture in economic development,
FAO Economic and Social Development Paper, no. 136, Rome
Stiglitz, J. 1994. Whither Socialism? Cambridge MA, MIT Press
Syrquin, M. 1998. Modern economic (endogenous) growth and development, in Coricelli, F. et al.
(eds), New Theories in Growth and Development, London, Macmillan
Summers, R. and Heston, A. 1988. A new set of international comparisons of real product and price
levels estimates for 130 countries, 1950–1985, Review of Income and Wealth, vol. 34, no. 1, 1–25
Summers, R. and Heston, A. 1991. The Penn World Table (Mark 5: an expanded set of inter-
national comparisons, 1950–1988), Quarterly Journal of Economics, vol. 106, no. 2, 327–68
Wulwick, N. 1993. What remains of the growth controversy? Review of Political Economy, vol. 5, no.
3, 321–43
You, J. 1994. Macroeconomic structure, endogenous technical change and growth, Cambridge
Journal of Economics, vol. 18, no. 2, 213–233
View publication stats

Cjengt

Uploaded by

Copyright:

Available Formats

Cjengt

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cjengt

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Endogenous Growth Theory: A Critical Assessment

Article in Cambridge Journal of Economics · February 2000

The user has requested enhancement of the downloaded file.

Endogenous growth theory: a critical

A critical assessment is made of endogenous growth theory from the perspective of

Key words: Endogenous growth

Manuscript received 12 January 1998; final version received 2 August 1998.

© Cambridge Political Economy Society 2000

2. New growth for old?

3. Opening a can of dynamics

5.1 Asset-stripping Schumpeter

5.3 Accounting for the state

5.4 Money, finance and growth

5.5 The political economy of growth

View publication stats

You might also like