OXFORD REVIEW OF ECONOMIC POLICY, VOL. 12, NO.

OXFORD REVIEW OF ECONOMIC POLICY, VOL. 12, NO. 2

POST-NEOCLASSICAL ENDOGENOUS
GROWTH THEORY: WHAT ARE ITS
POLICY IMPLICATIONS?

NICK CRAFTS
London School of Economics1

I. INTRODUCTION improvements in technology. Since government

policy can influence these decisions both directly
In a speech in the autumn of 1994, the Shadow through taxes and subsidies and indirectly via re-
Chancellor, Gordon Brown, referred to post-neo- form of institutional arrangements, intervention
classical endogenous growth theory. The press might in principle be used to raise investment and
seized upon this phrase and lampooned Mr Brown. hence the long-run growth rate.
This is unfortunate both since the new growth
economics deserves a wide audience and should be In this context, investment usually refers to a broader
a vital ingredient in the design of appropriate sup- concept than the physical capital accumulation
ply-side policies and also because it holds out the reported in the national accounts; human capital
prospect that government really can make a serious and/or research and development expenditures may
difference to the long-term growth rate. well also be included. The key to endogenous
steady-state growth is that there should be constant
The basic idea of endogenous growth is that long- returns to this broad capital accumulation. While
run growth in income per head depends on invest- this is a common property, the precise way in which
ment decisions rather than, as in traditional growth the result is obtained varies considerably across the
theory, resulting from unexplained or exogenous many models that have been proposed.

1
I am grateful to Wilfred Beckerman, Steve Bond, Bridget Rosewell, participants at a seminar at the National Institute of
Economic and Social Research, and an anonymous referee for their helpful comments on an earlier version. I am responsible for
all errors.

30 1996 OXFORD UNIVERSITY PRESS AND THE OXFORD REVIEW OF ECONOMIC POLICY LIMITED
 N. F. R. Crafts

Endogenous growth theory needs, however, to be ings in the empirical literature. Policy implications
examined both carefully and critically. In particu- follow in section V. Section VI concludes.
lar, it is important to distinguish between alterna-
tive models in this tradition which have substan-
tially different policy implications. It should also be II. THEORIES OF ENDOGENOUS
recognized that, at present, the empirical evidence GROWTH
relating to these models is incomplete and by no
means wholly supportive of the hypothesis that The traditional neoclassical growth model of Solow,
growth should be regarded as fully endogenous. and the growth accounting based upon it, has at its
This is for two reasons. First, it remains unclear that heart a production function characterized by con-
stimulating investment of whatever type (in physi- stant returns to scale and positive but diminishing
cal or human capital or research) will not eventually marginal productivity of factors of production. The
run into diminishing returns. Second, returns to usual version supposes a CobbDouglas produc-
investment and growth outcomes are influenced by tion function such that
the weather, in the sense of the availability of
technological opportunities. Y = TKaLb ( a + b = 1; 0 < a < 1 ) (1)

In this paper, I wish to highlight two different types where T, a scaling factor, reflects the level of
of endogenous growth theory, namely the broad technology and is often called total factor produc-
capital and endogenous innovation variants. The tivity (TFP), K refers to physical capital, and L to
first of these pays no particular attention to techno- labour.
logical change, whereas for the second it is central.
Both theories stress the importance of microeco- Using basic macroeconomics, let the rate of growth
nomic foundations of the investment decisions that of the capital stock, K / K = I / K = sY / K . Then it
give rise to growth, but only in the latter approach is apparent that any tendency for the capital stock to
is the focus firmly on the incentive structures relat- grow more quickly (slowly) than output will result
ing to the special nature of innovative activity. If in capital stock growth falling (rising) as Y/K falls
innovation is taken to be at the heart of the growth (rises) so that capital and output growth rates tend
process, then a wider range of policy issues arises. eventually to equalize. In the long run, the rate of
growth is
Assessment of the policy implications of endog-
T / T + bL / L
enous growth theory turns on empirical evidence. A Y / Y = . (2)
large body of econometric analysis of comparative (1 a )
economic growth has been undertaken by new Five important aspects of this formulation are the
growth economists and there is also useful work in following.
economic history and in the economics of technol-
ogy which can be consulted. Three specific ques- (i) In the long run, growth is independent of the
tions need to be addressed: investment rate; policies to promote investment
run into diminishing returns, ultimately affect
(i) Are there diminishing returns to broad capital only the level of output per head, and have only
accumulation? transitory effects on the growth rate.
(ii) What have been the proximate sources of dif- (ii) The share of profits in national income equals a,
ferences in growth? the elasticity of output with respect to capital,
(iii)To what extent, and how, can growth of output and can be used to estimate the impact of capital
per person be influenced by government policy? accumulation on the growth rate.
(iii)Growth of long-run income per person is pro-
In section II a basic outline of key ideas from portional to the growth rate of TFP and thus
endogenous theory is provided, followed in sec- requires improvements in technology, which is
tions III and IV by an overview of important find- not determined within the model and cannot be

31
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 12, NO. 2

influenced by policy-makers. Growth is exog- This leads us directly to the famous Rebelo (1991)
enous rather than endogenous. model of endogenous growth in which
(iv)If technological knowledge is universally avail-
~
able, the level and rate of change of TFP and Y = AK (5)
steady-state growth rates should be the same in all
~
countries. Equilibrium income levels would dif- where A is a constant and K represents broad
fer only if capital to labour ratios did not equalize. capital, a fixed proportions composite of K and H.
(v) Out of the steady state, countries which have
low initial income levels and capital to labour Rebelo (1991) not only proposes a model in which
ratios should grow more quickly for a given policies to raise the rate of investment raise long-
savings rate because they will have relatively run growth, but also sees investment decisions in
low capital to output ratios. terms of intertemporal optimizing behaviour. In
equilibrium the return to investment equals the
A very simple, but empirically unpersuasive, way return to consumption and the rate of interest. A
to obtain a model of endogenous growth would be reduction in the marginal rate of direct taxation on
to modify this production function to the form income from investment (or anything else raising
the net rate of return) leads to more investment in
Y = Ka + xLb (a + x = 1). (3) broad capital and thus to a higher steady-state
growth rate.
In this case x represents externalities in the form of
social returns, such as enhancements in productiv- While the explicit introduction of human capital
ity through additional experience, which do not into models of economic growth commands wide
accrue to the investors. Note also that there is no support, this can obviously be done in ways which
exogenous productivity growth (T has disappeared), do not imply endogenous growth. In another much-
profits share underestimates the importance of capi- cited paper, Mankiw et al. (1992) argued for what
tal to the growth process, and there are constant they termed an Augmented-Solow growth model.
rather than diminishing returns to investment. Here This might be written as
an increase in the growth of the capital stock com-
ing from higher investment will be sustained, will Y = TKaLbHc ((a + c) < 1). (6)
permanently increase the growth rate, and can de-
liver rising incomes per person. Whereas in equa- Here growth is not endogenous because the expo-
tion (1) the marginal productivity of capital is nents on K and H sum to less than one. Even so, the
positive but diminishing, in (3) it is constant and introduction of human capital does have important
marginal and average product are equal.2 implications, namely that diminishing returns to
broad capital will be less severe than to physical
A more appealing way to arrive at endogenous capital in the original Solow model, convergence to
growth is to introduce human capital into the pro- the steady state will be slower, and transitory ef-
duction function. If fects of increased investment rates on growth will
last for longer.
Y = KaLbHc (a + c = 1) (4)
One important feature of the Rebelo model is that it
where H is human capital, then the sum of the suppresses any explicit role for technological
exponents on the reproducible factors of produc- change. This is quite an important limitation as far
tion, K and H, is one and there will be constant as policy-makers are concerned and is certainly
returns to accumulation given that, as we might unlikely to appeal to applied economists. As Romer
expect, in equilibrium the ratio of K to H is constant. (1993, p. 562) recently put it,

2
The effect of an increase in the investment rate works through its impact on the rate of growth of the capital stock. For a given
savings rate, s, this can be written as sY/K. With diminishing marginal product of capitalreflected in an exponent on capital less
than oneincreases in s trigger offsetting declines in Y/K and do not permanently raise the growth rate of either the capital stock
or output.

32
 N. F. R. Crafts

our knowledge of economic history, of what production enhances the appropriability of returns to innova-
looked like 100 years ago, and of current events con- tion speeds growth up.
vinces us beyond any doubt that discovery, invention,
and innovation are of overwhelming importance in eco- The GrossmanHelpman model itself allows vari-
nomic growth and that the economic goods that come
ous factors to draw more resources into innova-
from these activities are different in a fundamental way
from ordinary objects. We could produce statistical
tion.3 Innovation and the growth rate depend posi-
evidence suggesting that all growth came from capital tively on the size of the market, the productivity of
accumulation with no room for anything called techno- labour in research, and the degree of market power
logical change. But we would not believe it. in selling the products resulting from innovation.
These are mostly intuitively appealing features;
Explicit consideration of technological change has thus, a wider market means larger fixed costs of
ramifications both in terms of innovation and tech- research can be covered by expected sales and more
nology transfer. The endogenous innovation vari- talented and/or better educated researchers will
ant of endogenous growth theory has examined the lower the costs of innovation. The neo-
first in detail, proposing models in which there is Schumpeterian suggestion that greater market power
endogenous TFP growth. Grossman and Helpman is conducive to innovation and growth is, however,
(1991) provided the seminal discussion in terms of highly contentious.
a production function
In models of this kind, once again, government can
a b d
Y = CK L D (a + b + d = 1) (7) affect the long-run rate of growthhere by influ-
encing the scale of resources allocated to the inno-
where C is a constant and D represents an index of vation sector. Indeed, the structure of these models
the designs of intermediate goods for use in produc- allows a much wider variety of policies to matter.
tion which embody technological knowledge and For example, growth might be influenced through
which are augmented by innovative activity. La- trade liberalization, competition policy, the relative
bour can be used either in research with constant attractiveness of research to talented people, or
returnswhich is justified by the spillover effects public procurement policies, as well as the more
of increases in the pool of knowledgeor with obvious routes of subsidizing R & D or modifying
diminishing returns, as usual, in production. the law on patents. A central aspect is the appropri-
ability of returns to innovation.
In this formulation, the relationship between the
growth of capital and output growth is just the same The assessment of policy based on an endogenous
as in the traditional neoclassical modelin the innovation and a broad capital endogenous growth
long run capital and output grow at the same rate model might differ considerably. A straightforward
and increases in the investment rate have only example is provided by the Single Market (1992)
transitory effects on the growth rate. Growth is once programme of the European Union. In an early
again proportional to TFP growth and thus to growth article using the new growth economics, Baldwin
in D, the variety of intermediate goods, which (1989) pointed out that, by ignoring the dynamic
increases when more labour is allocated to re- effects of 1992, the Cecchini report was potentially
search. hugely understating the likely gains. Baldwin con-
templated a permanent automatic increase to Eu-
The distinctive feature of the model is that the ropes growth rate using a Rebelo-type model;
allocation of resources to innovation is determined because 1992 would increase static efficiency it
within the model on a profit-maximizing basis. To would raise the output to capital ratio and hence, for
make innovation worthwhile, it must be possible to any given savings rate, capital stock growth and
appropriate returns to cover the fixed costs of thus the rate of economic growth. An analysis based
research and thus some element of imperfection in on an endogenous innovation growth model would
competition is required. In general, anything which be more ambivalent. Growth would obviously be

3
An easily accessible algebraic treatment is provided by Gould and Ruffin (1993).

33
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 12, NO. 2

enhanced through wider markets but there might automatic and that the social capability effec-
also be adverse effects. For example, the liberaliza- tively to assimilate new ideas varies greatly (Abram-
tion of public procurement could reduce the rents to ovitz, 1986).
be obtained by R & D, or the greater budgetary
spending on social cohesion to make the Single
Market acceptable might draw more talent away III. EVIDENCE ON BROAD CAPITAL
from innovation and into rent-seeking lobbying AND GROWTH
activities.
There now exists a substantial body of historical
It must be stressed that, while the endogenous evidence on comparative economic growth and
innovation framework is obviously attractive, it income levels. As is well known, in the recent past,
only implies endogenous growth under certain as- growth rates have differed substantially across coun-
sumptions. In particular, if the assumption of con- tries and between periods and Table 1 reports some
stant returns to labour in R & D is replaced by one estimates which provide the background to much
of diminishing returns, then the model would revert recent discussion of British growth performance.
to a kind of Augmented-Solow formulation, as
proposed by Jones (1995). In that case, Solows Table 2 shows some of the factors associated with
basic proposition would return and policy towards the growth outcomes of Table 1. At this stage, it is
innovation cannot affect the long-run growth rate. sufficient simply to note that there clearly is a
strong correlation between strong growth and a
In the traditional neoclassical growth model, tech- high investment rate. Such correlations have fea-
nology is assumed to be universally available. By tured prominently in the recent revival of empirical
contrast, writers in the endogenous innovation tra- growth economics and the correct interpretation of
dition argue that nations may fail to become rich, to these results is central to the assessment of the
catch up leading countries, or to grow as fast as their endogenous growth literature.
peer group because of idea gaps rather than ob-
ject gaps (Romer, 1993). At its simplest, this might Since the late 1980s, economists have devoted a
simply mean that flows of information are imper- great deal of effort to regression analysis of similar
fect such that, in terms of equations (1) or (6), T, the but, of course, much larger international cross-
TFP level, varies and so, therefore, does the steady- sectional data sets. Most of the early work was well-
state level of income per head. A corollary would be reviewed by Levine and Renelt (1992) who found
that if technological knowledge diffused more com- that there were relatively few reliable results in the
pletely, initially backward countries would tend to literature but argued that a robust regression for a
catch up the leaders not just through factor conver- broad cross-sectional sample for 196085 was
gence, as in the traditional or Augmented-Solow
models, but also via TFP convergence. GYP = 0.83 0.35*RGDP60 0.38GPO
+ 3.17*SEC + 17.5*INV
In a more sophisticated framework, it might be
argued that technology transfer involves similar where GYP is the growth rate of GDP per person,
investments and incentives to those required by RGDP60 is real income per person in 1960, GPO is
initial discovery and that, to be successful, coun- the population growth rate, SEC is the secondary
tries need indigenous firms to develop this techno- school enrolment rate, INV is the proportion of
logical capability and/or to be open to and attractive GDP spent on investment, and * indicates a statis-
to foreign direct investment (Teece, 1976). Here, tically significant coefficient. Thus growth is posi-
steady-state TFP levels and growth rates may differ tively related to investment in broad capital and
if local institutions, policies, or economic condi- there is evidence of catch-up in the sense that,
tions create differences in the rate of endogenous normalizing for capital accumulation, initially poor
technology transfer. Economic historians have in- countries grow faster than rich ones. These results
deed long stressed that catching-up is far from are not inconsistent with Tables 1 and 2.

34
 N. F. R. Crafts

Table 1
Income Levels and Growth Rates in Selected European and Asian Countries, 195092

Income/head ($1990) Growth rate, 195073

1950 1973 Actual Adjusted
France 5,221 12,940 4.0 5.0
Germany 4,281 13,152 5.0 6.0
Italy 3,425 10,409 5.0 5.8
UK 6,847 11,992 2.5 3.1
Japan 1,873 11,017 8.0 7.7
Korea 876 2,840 5.2 4.1
Taiwan 922 3,669 6.2 5.6

Income/head ($1990) Growth rate, 197392

1973 1992 Actual Adjusted
France 12,940 17,959 1.7 2.7
Germany 13,152 19,351 2.1 2.7
Italy 10,409 16,229 2.4 2.4
UK 11,992 15,738 1.4 2.2
Japan 11,017 19,425 3.0 3.1
Korea 2,840 10,010 6.9 5.2
Taiwan 3,669 11,590 6.2 5.3

Sources: All estimates based on Maddison (1995). Income per head is measured in purchasing power parity
adjusted international dollars; Germany and Korea refer to West Germany and South Korea. Growth
rates are of income per head in per cent per year. Adjusted growth rates allow for changes in hours worked
following the method proposed by Beckerman (1980)see text.
In the traditional neoclassical model, the initial Although it is tempting to interpret correlations
short-run effect of a rise in the investment rate between investment rates and growth in terms of a
translates into an impact on growth as follows. The causal relationship, this also is probably wrong.
change in the savings rate divided by the capital to When the LevineRenelt sample is broken down
output ratio gives the change in the rate of growth into sub-periods and analysed in terms of Granger-
of the capital stock which raises the growth rate of Causality, a statistical test based on leads and lags,
output in line with the exponent on capital. For a it turns out that there is a strong tendency for
capital to output ratio of 1.9, the average in the investment to follow growth but not for growth to
sample (King and Levine, 1994, p. 275), and assum- follow investment (Blomstrom et al., 1996). In
ing that if profits share of national income is a other words, it seems that investment responds to
guide, the elasticity of output with respect to capital better growth opportunities rather than creating
= 0.33, then a rise of 1 percentage point in the them. Both this and the size of the estimated invest-
investment rate should raise output growth by 0.33/ ment coefficient are, prima facie, much more con-
1.9 = 0.174 percentage points. Interestingly, this is sistent with a GrossmanHelpman than a Rebelo-
almost exactly the prediction of the coefficient of type new growth model.
17.5 estimated by Levine and Renelt. In other
words, although a casual glance might suggest Growth accounting has traditionally sought to meas-
otherwise, on close examination this empirical ure the sources of growth on the basis of an equation
growth equation is more in line with a traditional which could be derived from equation (1). TFP
Solow than a new Rebelo growth model. growth is calculated as a residual after working out

35
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 12, NO. 2

Table 2
Accounting for Growth

I/ Y K /K Growth due TFPGR Y/Y

to capital
195073
France 23.8 5.3 1.6 3.1 5.0
Germany 26.9 7.3 2.2 3.3 6.0
Italy 24.6 5.2 1.6 2.5 5.0
Japan 32.6 10.2 3.1 3.6 9.2
UK 18.3 5.5 1.6 1.2 3.0
197392
France 21.6 4.2 1.3 0.6 2.3
Germany 22.2 3.1 0.9 1.5 2.3
Italy 21.9 3.2 1.0 1.1 2.8
Japan 30.3 6.6 2.0 1.2 3.8
UK 18.1 3.1 0.9 0.6 1.6
196690
Korea 28.4 13.7 4.1 1.7 10.3
Taiwan 26.9 12.3 3.2 2.6 9.4

Sources: Growth accounting calculations for France, Germany, Japan, and UK from Maddison (1996), for
Italy from Rossi et al. (1992) and for Korea and Taiwan from Young (1995). Estimates for Italy are for
197390. Investment data are from OECD (1995) and from Rodrik (1995).

the contributions of capital and labour.4 The esti- rapid catch-up growth (Basu and Fernald, 1995;
mates in Table 2 are obtained from the expression Rossi and Toniolo, 1996).

Y/Y = a K /K + bL /L + TFPGR. (8) The growth accounting of Table 2 suggests that
neither episodes of rapid growth nor international
These estimates do make some allowance for the differences in growth outcomes should be attrib-
quality as well as the quantity of factor input growth uted primarily to investment in physical capital. On
and thus take into account (rather crudely) human the contrary, TFP growth seems to play a large part
capital inputs in the contribution of the labour and, in the Asian countries, so do rapid increases in
input. In the pure textbook neoclassical world of labour inputs based on rising participation rates and
equation (1), TFP growth would result solely from thus more hours worked per person.
technological change but, in practice, as growth
accountants themselves stress (Maddison, 1996), Since the newly industrializing countries of Asia
this component subsumes much more, including are often seen as an example of growth success
the effects of improved use of inputs and structural based on very high rates of accumulation of broad
change as well as perhaps failure adequately to capital, it is interesting to follow up the discussion
measure the contribution of human capital. Recent of Table 2 with a Levine and Renelt-style regres-
studies have indeed stressed that the contribution of sion-based analysis of success in East Asia com-
pure technological change to TFP growth is prob- pared with failure in Latin America (Table 3).
ably quite small, especially during episodes of Although investment and school enrolment have

4
The evidence summarized in this section suggests that the conventional use of profits share of national income as the estimate
of a in the calculation of TFP growth, which is used in Table 2, is an acceptable approximation.

36
 N. F. R. Crafts

Table 3
Estimated Sources of Differences in Growth Outcomes, 196085:
East Asia (EA) versus Latin America (LA) (% per year)

EA LA

Investment 0.20
Primary enrolment 0.17
Secondary enrolment 0.06
Population growth 0.01
Initial GDP 0.18

EA dummy 1.71
LA dummy 1.31

Actual 3.62

Source: World Bank (1993, Table 1.10).

been higher in the Asian countries, the regression years, but exceeds what the OECD countries
suggests that this accounts for relatively little (0.43) have achieved subsequently. It remains to be
of the 3.62 percentage points gap in growth rates. seen whether the East Asian countries can sus-
tain their TFP growth as they catch up the West,
By far the biggest contribution comes from the but the Japanese example suggests this will be
positive East Asian and negative Latin American difficult.
dummy variables. The implication of this is, in
effect, that TFP growth, which is not captured in the (iii)To put Asian growth in a proper perspective,
regression, differed markedly and that, presumably account should be taken of its reliance on in-
through differences in policies and institutions, creasing labour inputs per person, i.e. forgone
East Asian countries made better use of their invest- leisure, which contrasts with the opposite ten-
ments and/or gained better access to foreign tech- dency in Europe. Table 1 adjusts growth of
nology. This does indeed seem to be the message of income per head to allow for this by valuing
students of the role of technology in industrializa- changes in hours worked per person at the
tion (von Tunzelmann, 1995). average wage rate in the terminal year, as sug-
gested by Beckerman (1980). The effect is quite
A comparison of growth in East Asia and western marked and suggests that, on this broader meas-
Europe, using Tables 1 and 2, is also instructive. ure of performance, East Asian growth lies
Three points, in particular, deserve emphasis. somewhere between that of France and Ger-
many in the Golden Age.
(i) Taking 196690 as a whole, it is capital stock
growth rather than investment rates which have Recalling the models discussed in section II and
been exceptional in Korea and Taiwan. This noting the strong showing of the schooling variable
reflects lower capital to output ratios, an advan- and the implications of the estimated coefficient on
tage which will erode as these countries catch investment, prima facie, the regression results ob-
up and output grows less quickly than capital, as tained by Levine and Renelt (1992) seem closest to
has happened in Japan. the Augmented-Solow model. In a straightforward
endogenous growth model, the initial income level
(ii) Rapid TFP growth tends to be characteristic of should not affect the growth rate, whereas with
catch-up growth. TFP growth has been less exogenous growth convergence to the steady state
strong in Korea and Taiwan than in Europe would, of course, lead to an inverse correlation as
during its Golden Age in the early post-war found in the regression.

37
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 12, NO. 2

An interpretation of cross-section regressions to hypothesis of constant returns to broad capital

explain growth (and also productivity levels) along continues to be rejected.
these lines has been strongly advocated by Mankiw
et al. (1992). Imposing some additional structure on The Augmented-Solow interpretation of recent
their analysis they infer that supposing a production growth experience puts strong emphasis on the role
function of Y = K 0.33L 0.33H 0.33 gives a good ap- of human capital, proxied by the schooling of the
proximation to international experience. Not only labour force, in production. Islam (1995), in his
is this inconsistent with endogenous growththe reappraisal of the evidence, shows both that this
exponents on K and H sum to less than onebut the result is not obtained for the OECD countries subset
relatively low exponent on K (similar to profits and that the coefficient on the schooling variable
share of national income) suggests an absence of actually turns negative in all subsets of the Mankiw
externalities to physical capital formation. Mankiw et al. (1992) sample once the panel data approach is
(1995) has gone on to claim that the traditional adopted. There is, however, a close correlation
neoclassical assumption of universally available between schooling and T, and Islam suggests that it
technological knowledge is also tenable and thus may be better to view education as an aspect of
would see the inverse correlation between income social capability impinging on technology transfer
levels and subsequent growth simply as neoclassi- rather than as a direct input into production (1995,
cal factor convergence. pp. 11602).

Further research suggests, however, that these find- In terms of the input of human capital to production,
ings need, at the least, substantial qualification. The however, it may well be that it is training rather than
most reliable seems to be the rejection of the hy- education that really matters. International cross-
pothesis of substantial externalities to physical sectional data on training are unfortunately not
investment.5 Two studies are of particular note. In available. Detailed investigation of the role of la-
the first, Oulton and OMahony (1994), in an econo- bour-force skills in British and German manufac-
metric analysis of British manufacturing during turing has, however, been undertaken. In a cross-
195486, found that there was no support for the sectional econometric study for 1987, OMahony
hypothesis that weighting capital by profits share (1992) found strong effects of relative skills
underestimates capitals role in growth, nor any measured either by vocational qualifications or
evidence to indicate that social returns to invest- wage differentials on relative productivity per-
ment exceeded private returns. formance. Her results give weak support to the
hypothesis of positive externalities to skill for-
The second study by Islam (1995) returns to Mankiw mation and imply an exponent on broad capital
et al.s findings, noting that the ordinary least (K + H) of the order of 0.58 to 0.66very similar
squares estimation that they used requires that to the production function suggested by Mankiw
investment is independent of the error term, which et al. (1992) and consistent with an Augmented-
is implausible if countries differ in terms of social Solow model.
capability and thus technology. Islam takes a panel
data approach which permits the investigation of Following on from this, it is also necessary to
individual country effects. Two important results reconsider the implications of the finding of a
followfirst, that the equivalent of the T term in negative cross-section correlation between initial
equation (6) varies considerably across countries income levels and growth rates. Further economet-
even within the OECD and, second, that, correcting ric research suggests both that this is not in fact
the bias in Mankiw et al.s results, profits share still decisive evidence against the endogenous growth
seems to be an appropriate weight for capital. Islam hypothesis and also that TFP as well as factor
also confirms Mankiw et al.s other key result; the convergence was involved.

5
This is despite the much discussed study by De Long and Summers (1991) which claims a special role for equipment investment
in growth and a 30 per cent social rate of return to this category of investmentwhich would imply substantial externalities. Their
results appear highly sensitive to the exclusion of one outlying observation and for the subset of OECD countries there is no
evidence that social returns exceed private returns (see Auerbach et al., 1994, and Oulton and Young in this issue).

38
 N. F. R. Crafts

For example, van de Klundert and van Schaik nology is the same in all countries, even in the
(1996) find for the advanced countries with long- advanced world, and thus against the pure Aug-
run data that this correlation only applies in the mented-Solow interpretation of growth.
period 195073 and does not hold either before
1950 or after 1973. Similarly, Bernard and Durlauf (iii)The episode of catch-up in post-war growth is
(1995) in an analysis of the same long-run time- not decisive evidence against the endogenous
series using cointegration techniques, are unable to growth hypothesis.
reject the hypothesis of no convergence in the sense
that long-run forecasts of trends and levels of output (iv) The recent growth in East Asian countries,
are not the same. Finally, Milbourne (1995) devel- although very impressive, should not be re-
ops a test to exploit restrictions implied by the pure garded as unprecedented or miraculous.
Augmented-Solow model and finds that TFP con-
vergence plays a major role in the post-war OECD
countries growth experience. IV. EVIDENCE ON ENDOGENOUS
INNOVATION AND GROWTH
As Sala-i-Martin (1994) has pointed out in the
context of the Rebelo model of equation (5), if there Whereas the Rebelo model of endogenous growth
is an episode in which broad capital productivity is puts (broad) capital accumulation centre-stage, the
raised by new opportunities for technology transfer endogenous innovation models seek to endogenize
having most of an impact on the economies which rather than to abolish TFP growth. The key issues
are the least productive, some data sets may demon- are, of course, how well do these models do empiri-
strate strong catch-up without any implication of cally in explaining TFP growth and does the evi-
long-run convergence, as seems may well be the dence suggest that returns to investment in innova-
case for the OECD countries. tive activity can support endogenous growth? In
answering these questions, it is useful to draw
Economic history offers strong support for this rather more than growth theorists generally have on
view, given that the inverse correlation findings the economics of technology literature as well as
appear to result from data sets with starting points contributions from empirical growth economics.
in the Golden Age of Table 1. The catching-up of
this period involved significantly reduced obsta- Macroeconomic growth studies have provided some
cles to the assimilation of American technology in support. Three recent econometric exercises are
Europe as market size was expanded through eco- worth noting. First, Verspagen (1996) investigates
nomic integration and multinational companies grew the role of R & D expenditures in an aggregate
enormously in importance, while the codification production function for France, Germany, and the
of knowledge and availability of scientific and UK since 1960; he finds that R & D accounts for
engineering personnel were also much enhanced around a quarter of labour productivity growth in
(Nelson and Wright, 1992). At the same time, the first two countries but cannot reject the null
catch-up was promoted by structural change and hypothesis that it had no effect in the UK.
reconstruction as the damage from war and inter-
war policy errors was remedied (Dumke, 1990). Second, Coe and Helpman (1995) show that TFP
growth in 197190 in European countries has been
Thus far, this review of inferences from macroeco- highly responsive to increases in foreign R & D
nomic growth databases has suggested four main stocks, more strongly so the more open the economy.
conclusions. Third, Chou et al. (1995) explore the determinants
of industry-financed R & D expenditure in OECD
(i) Contrary to the Rebelo model, there are dimin- countries in the 1980s; they conclude that domestic
ishing returns to broad capital formation. market size, the degree of international economic
integration, and government spending on research all
(ii) There is strong evidence against the neoclassi- have significant positive effects, as would be predict-
cal assumption that the aggregate level of tech- ed by standard endogenous innovation models.

39
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 12, NO. 2

Table 4
Industry-financed R & D Expenditure (% of GDP)

1967 1991

Switzerland 1.78 2.07

Netherlands 1.12 0.91
UK 1.00 0.94
USA 0.99 1.36
Germany 0.94 1.57
Japan 0.83 2.13
Sweden 0.71 1.71
Belgium 0.66 1.16
France 0.60 0.99
Canada 0.40 0.59
Norway 0.35 0.77
Denmark 0.34 0.85
Italy 0.33 0.61
Finland 0.30 1.07
Ireland 0.19 0.58
Spain 0.08 0.38
Portugal 0.04 0.14

Source: Patel and Pavitt (1994).

Microeconomic studies in industrial and technol- spreading fixed costs (Cohen, 1995), rivalry among
ogy economics give further, somewhat qualified, big firms is helpful in promoting R & D (Patel and
backing to the endogenous innovation approach. Pavitt, 1992).
This body of research certainly stresses that
appropriability of returns is central to innovative While the notion of endogenous innovation sur-
effort and that market size and demand growth have vives this encounter with the empirical evidence,
positive effects; a good study which encapsulates the issue of scale effects raised by Jones (1995)
all these points is Jaffe (1988). remains as a strong objection to the hypothesis of
endogenous growth. Jones considers OECD (espe-
Nevertheless, it may be the spirit rather than the cially American) growth since 1900 and stresses
letter of the new growth models which should be that growth rates of output and TFP are not now
taken seriously, since some aspects of the models higher than ever before. He argues that the rejection
are less than convincing and it would be unfortunate of R & D-based growth models is particularly
if these were taken too seriously by policy-makers. strong: The models posit that the growth rates of
In particular, the emphasis in the growth literature per capita output and TFP should be increasing with
on market power and patents as key incentives for the level of resources devoted to R & D, which is
growth needs to be treated sceptically. Empirical wildly at odds with empirical evidence (1995, p.
evidence suggests that appropriability is related 519)as a comparison of Tables 1 and 4 confirms.
much more to lead times over rivals than to patents Jones finds the basic structure of endogenous inno-
(Levin et al., 1987) and that there is no relation vation models to be appealing, but not the endog-
between concentration and R & D in properly enous growth implications.
specified regression studies (Cohen, 1995). Moreo-
ver, while large business units may have advan- This critique is probably not quite so devastating as
tages in research in exploiting learning curves and might appear at first sight. Three points should be

40
 N. F. R. Crafts

considered. First, as was noted earlier, TFP growth (iii)TFP growth has been an important component
appears to have accrued historically from a wide of past growth but industry-financed R & D has,
range of sources and, given the relatively low at best, underpinned quite modest rates of en-
fractions of GDP spent on R & D, at least until dogenous growth in the past.
recent times relatively little TFP growth can have
come from this.6 Equally, the reasons for the slow-
down in TFP growth in the OECD economies may V. POLICY IMPLICATIONS
lie elsewhere.
This section builds on the evidence discussed in
Second, direct evidence on the potency of R & D is sections III and IV, together with some reflections
somewhat mixed and bedevilled with measurement on recent British economic history, to draw out
problems. Griliches (1994) reviews the evidence some policy lessons. A general implication of the
and concludes on balance that it is likely that R & new growth economics is that institutions and policy
D in American manufacturing had as strong an may have stronger effects on the growth rate than
effect on TFP growth in 197889 as it did in 1958 would have been predicted using the traditional
73. Furthermore, it may be that our ability accu- neoclassical growth model.7 Moreover, it is impor-
rately to measure TFP growth has declined as the tant to take a broad view of the role for policy rather
service sector has grown in relative importance and than restrict the perspective to investment subsidies
that recent official estimates are understated through and taxation; reform of institutions potentially be-
inadequate treatment of quality changes. comes a key aspect of policy-making.

Third, if, as some accounts of long-run technologi- The conclusions of the last two sections have sev-
cal change insist, major changes in techno-eco- eral broad policy implications. The most important
nomic paradigm, such as the switch since the are the following.
Golden Age from Fordist mass production to flex-
ible specialization and information technology, in- (i) A policy to stimulate investment may not be a
volve creative destruction and grave adjustment route to permanently faster growth but could
problems, then it could be that the initial impact have significant short-term effects. It would
would be felt in a slowing of productivity growth seem desirable to target broad capital rather
while the long-run effect is the opposite (Freeman than merely physical investment.
and Perez, 1988).
(ii) Given that routine investment has diminish-
Three further conclusions can now be drawn: ing returns, if faster endogenous growth is
feasible, it will result from success in raising
(i) innovative activity is responsive to a variety of the rate of technological change. Even if the
economic stimuli and can be influenced by fears of critics such as Jones (1995) are
economic policy; overstated, it should be recognized that, real-
istically, policy to stimulate R & D cannot be
(ii) it is unclear whether policies to promote inno- expected to have dramatic effects. For exam-
vation can raise long-run growth or whether ple, doubling the share of GDP devoted to
diminishing returns to R & D reassert the industry-financed R & D to 2 per cent would
Solovian world of transitory effects only; raise TFP growth by at most 0.4 percentage

6
Griliches (1995, p. 60) regards a social rate of return of about 40 per cent to industry-financed R & D as a reasonable guess.
Applying this to the R & D figures in Table 4 suggests that in the Golden Age a contribution as high as 0.4 per cent per year to
TFP growth would be quite unusual.
7
The severity of diminishing returns can be expected to be much weaker when a broad notion of capital is adopted since (a +
c) in equation (6) will tend to be much larger than a alone in equation (1), even if the conditions for endogenous growth are not
satisfied. Barro and Sala-i-Martin (1995, pp. 368 and 807) offer illustrative calculations that suggest for a model with a = 0.33
the rate of convergence to the steady state will be about 5.6 per cent per year with a half-life of 12.5 years, whereas for a model
with (a + c) = 0.75 the rate of convergence will be about 2 per cent per year with a half-life of 35 years.

41
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 12, NO. 2

points and thus, using the formula in equation tions and too little on facilitating the diffusion of
(2), the growth of output by about 0.6 percent- improvements and enhancing technology transfer
age points. (Ergas, 1987).

(iii)Nevertheless, given the evidence against the Policy choices were curtailed by continual attempts
convergence predictions of the Augmented- to seek the cooperation of the trade union move-
Solow model and the possibility that technol- ment to deliver low unemployment without exces-
ogy levels vary substantially, the costs of fail- sive wage inflation. In particular, this generally
ing adequately to exploit opportunities for tech- precluded attempts to reform industrial relations,
nology transfer may be considerable. This sug- with the exception of the ill-fated act of 1971. Given
gests a high priority for identifying and remedy- Britains idiosyncratic structure of decentralized
ing market failures in the area and the potential bargaining based on multiple unionism and weak
importance of policies in the areas of skills legislative control, this might be expected to have
acquisition and foreign direct investment de- had adverse consequences for investment and/or
signed to reduce the ideas gap. innovation by firms because of worries about the
possible hijacking of returns by workers unable to
(iv)The endogenous innovation literature argues commit themselves not to do so (Grout, 1984).
that attention needs to be given to enhancing the
appropriability of returns. Set in a contempo- Putting this insight into the context of an endog-
rary British context, this is likely to involve enous innovation growth model would imply fewer
dealing with various forms of short-termism resources devoted to innovative activity and, thus,
resulting in under-investment in both innova- lower growth. Empirical investigations give some
tive activity and physical investment. support to this for investment (Denny and Nickell,
1992) and for TFP growth, which it is estimated that
Seen in this light, the general thrust of British the presence during 195479 of multiple unionism
supply-side policy during the years of rapid relative in British manufacturing reduced by between 0.75
economic decline from the 1950s through to the and 1.1 per cent per year (Bean and Crafts, 1996).
1970s appears most unfortunate. Two general fea-
tures stand out. First, the poor targeting of subsidies The early post-war years also saw the emergence of
to investment and, second, the failure effectively to a new-style capital market in the UK, characterized
tackle questions of institutional reform. These weak- by hostile takeover and by short-term, low-trust
nesses did not go completely unrecognized but relations between managers and shareholders in
governments were generally too weak or too short- which under-investment could be expected for rea-
termist to address them (Bean and Crafts, 1996). sons similar to the Grout problem (Dickerson et al.,
1995). In this context, Mayer (1992) found that
There was a strong emphasis throughout on subsi- large British firms have generally felt compelled to
dizing physical investment, building up by the maintain dividends at the expense of R & D. Al-
1970s to a focus on selective assistance and further though evidence gradually mounted that the merger
nationalizations. By contrast, training was rela- and takeover boom of the 1960s did little to promote
tively neglected, especially prior to 1964. The UK productivity improvement (Singh, 1975; Meeks,
did establish heavy support for R & D but this 1977), policy was informed more by a belief in the
included a very heavy defence-related component advantages of size and a desire to establish national
which offered little to the rest of the economy in champions than by thinking about the implications
terms of externalities, together with the economi- for the appropriability of returns.
cally disastrous programmes to support aerospace
and nuclear power.8 Technology policy concen- These imbalances in supply-side policy have been
trated unduly on aiming to promote radical innova- redressed to some extent in the 1980s and early

8
Government contributions to civil aircraft and engine development from 1945 to 1974 totalled 1.5 billion at 1974 prices and
produced receipts of 0.14 billion (Gardner, 1976).

42
 N. F. R. Crafts

1990s, although it is still unclear whether there has the much greater likelihood that social returns ex-
been any substantial impact on the long-run growth ceed private returns in the latter areas than in the
rate. The most obvious success has come in the area former.
of industrial relations, where initially conduct and
eventually structures were dramatically changed Second, industry-financed R & D is relatively small
(Metcalf, 1994) in an era of industrial crisis, accept- in the UK, as Table 4 reports, and there may be a
ance of high unemployment, ending of attempts at case for changes in the tax position of R & D which
social contracts, and high-risk economic policy- is less generous than in many other countries. This
making.9 These changes seem to have led not only would potentially be justified by spillover effects
to immediate increases in productivity levels but and by the evidence that the American R & D tax
also to reduced negative effects of union recogni- credit had a strong impact on expenditure (Griffith
tion on investment (Denny and Nickell, 1992) and et al., 1995). As these authors warn, however,
the removal of the adverse impact of multiple designing tax incentives for R & D is difficult, not
unionism on TFP growth (Bean and Crafts, 1996). least in terms of persuading firms that the policy
will be sustained.
Department of Trade and Industry (DTI) subsidies
to physical capital formation were substantially Third, it should be recognized that the fundamental
reduced, while those to innovation were main- stance in the Competitiveness White Papers (DTI,
tained, although redirected during the 1980s (Crafts, 1994, 1995) is not inconsistent with many of the
1993, p. 71). As the emphasis on support for physi- recommendations that might come from the endog-
cal investment was reduced, eventually more atten- enous innovation school and is alert to the role of
tion was paid to reform of education and training; policy in reducing the ideas gap. In particular, this
between 1987 and 1994 the number of economi- applies to the emphasis on fair and open markets
cally active people with a qualification rose by which stresses the value both of trade liberalization
3.9m and by 1993, 32 per cent of 18-year-olds went and of inward investment, and which does not
into higher education compared with 12 per cent in suggest the lurch towards protectionism feared by
1981 (Robinson, 1994). In 1994, supply-side policy critics of the (admittedly unsatisfactory) notion of
was consolidated under Michael Heseltines Com- competitiveness (Krugman, 1994).
petitiveness Division at the DTI. The ultimate
effect of these changes remains highly controver- There is, however, a danger that government in-
sial. volvement with competitiveness has led to undue
emphasis on the achievement of intermediate ob-
Gordon Browns reference to post-neoclassical jectives, which tend to lose their value once tar-
endogenous growth theory suggested that supply- geted by politicians. A possible example is the high
side policy needed further substantial changes and profile given to numbers attaining qualifications of
that, with appropriate intervention, the growth rate various kinds whose quality is inadequately moni-
might be appreciably increased. What, then, does tored. In any event, independent assessment of the
the new growth economics suggest should be the new education and training policies in terms of their
scope and chief focus of such changes? impact on TFP growth is a high priority and cer-
tainly cannot be obtained from (or by?) the DTI.
First, it is important not to return to 1960s and 1970s
style policies to promote growth. In particular, Fourth, despite the Schumpeterian outlook of many
relatively less attention should be devoted than at new growth models, it would be wrong to conclude
that time to raising the rate of fixed capital forma- that faster growth requires the abandonment or
tion and relatively more to finding effective inter- even the weakening of competition policy, still less
ventions to raise human capital accumulation and the sponsorship of national champions. The evidence
innovative activity. This message is reinforced by on innovation in Britain appears strongly to endorse
9
The size of the shock was largely unanticipated, although the stubbornness with which the policies of the early 1980s were
pursued was both deliberate and surprising. The short-term macroeconomic outcome would almost certainly have led to electoral
rout had it not been for the Falklands War (Price and Sanders, 1994) and the creation of these circumstances could not readily
have been contemplated ex ante by any politician.

43
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 12, NO. 2

the view espoused by Porter (1990) that the positive First, it should be recognized that raising the growth
effects of rivalry in preventing the managerial quiet rate is difficult and that a return to the growth rates
life outweigh any impact of reduced expected returns of the Golden Age is highly unlikely. Although
to successful projects (Geroski, 1994). there is more reason than once there was to hope
that appropriate government policy can raise the
It is sensible, however, to recognize the impact of long-run growth rate, it should also be accepted that
government policies on the appropriation of returns the hypothesis of endogenous growth is controver-
to productivity improvements. An important exam- sial.
ple of this occurs in the context of the regulation of
privatized companies. Although British-style price- Second, it is important to take away the right
cap regulation may offer incentives to management messages from the Augmented-Solow approach to
to pursue cost reductions, here again there are risks growth. The first is that convergence to the steady
of under-investment in innovation if the regulator state means that in the long run, raising the rate of
cannot commit not to expropriate ex post. Rees and broad capital formation without changing TFP
Vickers (1995) highlight this point and note that the growth will have only a levels effect and not a
main problem in practice may be political interven- growth-rate effect on income. The second is that
tion motivated by populist pressures. Clearly, the historical experience emphatically does not en-
proposal of windfall taxes on privatized utilities dorse the view that different countries tend auto-
epitomizes this danger. matically to a steady state based on identical tech-
nology. It follows that technology transfer is a key
Fifth, policies to raise the growth rate need to pay area of policy concern.
further attention to institutional reform. Doubts
about the British financial system, in terms of its Third, in the long run growth comes from improve-
excessive reliance on ownership and control by ments in productivity resulting from better technol-
outside investors and consequent appropriation ogy which involves the use and effective assimila-
problems and under-investment in projects with tion of new ideas. In principle, it follows that the
long-term pay-offs, continue to be voiced. Changes chief focus of a policy to strengthen long-run growth
to this regime could result from reform of corporate performance should be to address market failures
law and regulation and are unlikely generally to which weaken social capability rather than to
happen without government intervention given the subsidize routine investment. In practice, reduc-
coordination problems involved (Jenkinson and ing the exposure of the investor to fears of
Mayer, 1992). expropriation, whether by workers, sharehold-
ers, or politicians, will probably have favourable
impacts both on broad capital formation and on
VI. CONCLUDING COMMENTS TFP growth.

The preceding sections have given a quite intricate Finally, as a postscript, it may be as well to bear in
review of recent work in growth economics and on mind that better supply-side policies tend to have
the basis of this have extracted some cautious their big pay-off in the far distant future, while
policy implications. Rather than reiterate these frequently jeopardizing the votes of powerful inter-
points, I wish to conclude by returning to a broad- est groups initially. Absent effective ways of achiev-
brush style to make more explicit and to emphasize ing pre-commitment, the short-termism of politi-
three basic lines of argument underlying all the cians is always likely to be an obstacle to faster
detail. long-run growth.

44
 N. F. R. Crafts

REFERENCES

Abramovitz, M. (1986), Catching-Up, Forging Ahead, and Falling Behind, Journal of Economic History, 36 36(2),
385.
Auerbach, A. J., Hassett, K. A., and Oliner, S. D. (1994), Reassessing the Social Returns to Equipment Investment,
Quarterly Journal of Economics, 109109(3), 789802.
Baldwin, R. E. (1989), The Growth Effects of 1992, Economic Policy, 9, 24881.
Barro, R. J., and Sala-i-Martin, X. (1995), Economic Growth, New York, McGrawHill.
Basu, S., and Fernald, J. G. (1995), Aggregate Productivity and the Productivity of Aggregates, NBER Working
Paper No. 5382.
Bean, C., and Crafts, N. F. R. (1996), British Economic Growth since 1945: Relative Economic Decline . . . and
Renaissance?, in N. F. R. Crafts and G. Toniolo (eds), Economic Growth in Europe Since 1945, Cambridge,
Cambridge University Press.
Beckerman, W. (1980), Comparative Growth Rates of Measurable Economic Welfare: Some Experimental
Calculations, in R. C. O. Matthews (ed.), Economic Growth and Resources, vol. 2, London, Macmillan.
Bernard, A. B., and Durlauf, S. N. (1995), Convergence in International Output, Journal of Applied Econometrics,
10
10(1), 97108.
Blomstrom, M., Lipsey, R. E., and Zejan, M. (1996), Is Fixed Investment the Key to Economic Growth?, Quarterly
Journal of Economics, 111
111(1), 26976.
Coe, D. T., and Helpman, E. (1995), International R & D Spillovers, European Economic Review, 39 39(3), 85987.
Cohen, W. (1995), Empirical Studies of Innovative Activity, in P. Stoneman (ed.), Handbook of the Economics of
Innovation and Technological Change, Oxford, Blackwell.
Chou, C., Kimura, F., and Talmain, G. (1995), R & D Effort, Domestic, and International Scale Effects, paper
presented to the Seventh World Congress of the Econometrics Society, Tokyo.
Crafts, N. F. R. (1993), Can De-industrialization Seriously Damage Your Wealth?, London, IEA.
DeLong, J. B., and Summers, L. H. (1991), Equipment Investment and Economic Growth, Quarterly Journal of
Economics, 106
106(2), 445502.
Denny, K., and Nickell, S. (1992), Unions and Investment in British Industry, Economic Journal, 102
102(4), 87487.
Dickerson, A. P., Gibson, H. D., and Tsakalotos, E. (1995), Short-termism and Underinvestment: The Influence of
Financial Systems, The Manchester School, 63 63(4), 35167.
DTI (1994), Competitiveness: Helping Business to Win, Cm 2563, Department of Trade and Industry, London,
HMSO.
(1995), Competitiveness: Forging Ahead, Cm 2867, Department of Trade and Industry, London, HMSO.
Dumke, R. H. (1990), Reassessing the Wirtschaftswunder and Postwar Growth in West Germany in an International
Context, Oxford Bulletin of Economics and Statistics, 5252(4), 45190.
Ergas, H. (1987), Does Technology Policy Matter?, in B. R. Guile and H. Brooks (eds), Technology and Global
Industry, Washington, DC, National Academy Press.
Freeman, C., and Perez, C. (1988), Structural Crises of Adjustment, Business Cycles and Investment Behaviour, in
G. Dosi, C. Freeman, R. Nelson, G. Silverberg, and L. Soete (eds), Technical Change and Economic Theory,
London, Pinter.
Gardner, N. (1976), The Economics of Launching Aid, in A. Whiting (ed.), The Economics of Industrial Subsidies,
London, HMSO.
Geroski, P. (1994), Market Structure, Corporate Performance and Innovative Activity, Oxford, Clarendon Press.
Gould, D. M., and Ruffin, R. J. (1993), What Determines Economic Growth?, Federal Reserve Bank of Dallas
Economic Review, 2(2), 2540.
Griffith, R., Sandler, D., and van Reenen, J. (1995), Tax Incentives for R and D, Fiscal Studies, 16
16(2), 2144.
Griliches, Z. (1994), Productivity, R and D, and the Data Constraint, American Economic Review, 84 84(1), 123.
(1995), R and D and Productivity: Econometric Results and Measurement Issues, in P. Stoneman (ed.),
Handbook of the Economics of Innovation and Technological Change, Oxford, Blackwell.
Grossman, G. M., and Helpman, E. (1991), Innovation and Growth in the Global Economy, Cambridge, MA, MIT
Press.
Grout, P. A. (1984), Investment and Wages in the Absence of Binding Contracts: A Nash Bargaining Approach,
Econometrica, 5252(2), 44960.
Islam, N. (1995), Growth Empirics: A Panel Data Approach, Quarterly Journal of Economics, 110 110(4), 112770.
Jaffe, A. B. (1988), Demand and Supply Influences in R and D Intensity and Productivity Growth, Review of
Economics and Statistics, 72
72(3), 4317.

45
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 12, NO. 2

Jenkinson, T., and Mayer, C. (1992), The Assessment: Corporate Governance and Corporate Control, Oxford
Review of Economic Policy, 8(3), 110.
Jones, C. I. (1995), Time Series Tests of Endogenous Growth Models, Quarterly Journal of Economics, 110 110(2),
495525.
King, R. G., and Levine, R. (1994), Capital Fundamentalism, Economic Development, and Economic Growth,
CarnegieRochester Conference Series on Public Policy, 40 40, 25992.
Krugman, P. (1994), Competitiveness: A Dangerous Obsession, Foreign Affairs, 73 73(2), 2844.
Levin, R. C., Klevonick, A. K., Nelson, R. R., and Winter, S. G. (1987), Appropriating the Returns from Industrial
R and D, Brookings Papers on Economic Activity, 3, 783820.
Levine, R., and Renelt, D. (1992), A Sensitivity Analysis of Cross-Country Growth Regressions, American
Economic Review, 82 82(4), 94263.
Maddison, A. (1995), Monitoring the World Economy, 18201992, Paris, Organization for Economic Cooperation
and Development.
(1996), Macroeconomic Accounts for European Countries, in B. van Ark and N. F. R. Crafts (eds),
Quantitative Aspects of Postwar European Economic Growth, Cambridge, Cambridge University Press.
Mankiw, N. G. (1995), The Growth of Nations, Brookings Papers on Economic Activity, 1, 275310.
Romer, D., and Weil, D. (1992), A Contribution to the Empirics of Economic Growth, Quarterly Journal
of Economics, 107
107(2), 40737.
Mayer, C. (1992), The Financing of Innovation, in A. Bowen and M. Ricketts (eds), Stimulating Innovation in
Industry, London, Kogan Page.
Meeks, G. (1977), Disappointing Marriage, Cambridge, Cambridge University Press.
Metcalf, D. (1994), Transformation of British Industrial Relations? Institutions, Conduct and Outcomes 1980
1990, in R. Barrell (ed.), The UK Labour Market, Cambridge, Cambridge University Press.
Milbourne, R. (1995), Factor Convergence versus Productivity Convergence, paper presented to the Seventh World
Congress of the Econometrics Society, Tokyo.
Nelson, R. R., and Wright, G. (1992), The Rise and Fall of American Technological Leadership, Journal of
Economic Literature, 30 30(4), 193164.
OMahony, M. (1992), Productivity and Human Capital Formation in UK and German Manufacturing, NIESR
Discussion Paper No. 28.
OECD (1995), Historical Statistics, 19601993, Paris, Organization for Economic Cooperation and Development.
Oulton, N., and OMahony, M. (1994), Productivity and Growth: A Study of British Industry, 19541986,
Cambridge, Cambridge University Press.
Patel, P., and Pavitt, K. (1992), The Innovative Performance of the Worlds Largest Firms: Some New Evidence,
Economics of Innovation and New Technology, 2(1), 91102.
(1994), National Innovation Systems: Why they are Important and how they might be Measured and
Compared, Economics of Innovation and New Technology, 3(1), 7795.
Porter, M. (1990), The Competitive Advantage of Nations, London, Macmillan.
Price, S., and Sanders, D. (1994), Economic Competence, Rational Expectations and Government Popularity in
Postwar Britain, The Manchester School, 62 62(3), 296312.
Rebelo, S. (1991), Long-run Policy Analysis and Long-Run Growth, Journal of Political Economy, 99 99(3), 50021.
Rees, R., and Vickers, J. (1995), RPI X Price-Cap Regulation, in M. Bishop, J. Kay, and C. Mayer (eds), The
Regulatory Challenge, Oxford, Oxford University Press.
Robinson, P. (1994), The Comparative Performance of the British Education and Training System, LSE Centre for
Economic Performance Discussion Paper No. 644.
Rodrik, D. (1995), Getting Policy Interventions Right: How South Korea and Taiwan Grew Rich, Economic Policy,
20
20, 55107.
Romer, P. M. (1993), Idea Gaps and Object Gaps in Economic Development, Journal of Monetary Economics,
32
32(3), 54373.
Rossi, N., Sorgato, A., and Toniolo, G. (1992), Italian Historical Statistics, 18901990, University of Venice
Dipartimento di Scienze Economiche Working Paper No. 92-18.
Toniolo, G. (1996), Italy, in N. F. R. Crafts and G. Toniolo (eds), Economic Growth in Europe Since 1945,
Cambridge, Cambridge University Press.
Sala-i-Martin, X. (1994), Regional Cohesion: Evidence and Theories of Regional Growth and Convergence, CEPR
Discussion Paper No. 1075.
Singh, A. (1975), Takeovers, Natural Selection and the Theory of the Firm: Evidence from the Postwar UK
Experience, Economic Journal, 85 85(3), 497515.

46
 N. F. R. Crafts

Teece, D. J. (1976), The Multinational Corporation and the Resource Cost of International Technology Transfer,
Cambridge, MA, Ballinger.
van de Klundert, T., and van Schaik, A. (1996), On the Historical Continuity of the Process of Economic Growth,
in B. van Ark and N. F. R. Crafts (eds), Quantitative Aspects of Postwar European Economic Growth,
Cambridge, Cambridge University Press.
Verspagen, B. (1996), Technology Indicators and Economic Growth in the European Area: Some Empirical
Evidence, in B. van Ark and N. F. R. Crafts (eds), Quantitative Aspects of Postwar European Economic
Growth, Cambridge, Cambridge University Press.
von Tunzelmann, G. N. (1995), Technology and Industrial Progress, Aldershot, Edward Elgar.
World Bank (1993), The East Asian Miracle, New York, Oxford University Press.
Young, A. (1995), The Tyranny of Numbers: Confronting the Statistical Realities of the East Asian Growth
Experience, Quarterly Journal of Economics, 110110(3), 64180.

47