Two institutions have retarded UK productivity growth in the
post-war period: industrial relations and education. The failings of
both were largely addressed in the 1980s. the productivity improvement
of the 1980s was genuine and was largely due to the reduction in union
power brought about by the trade union legislation of the 1980s. The
1980s and 1990s have also seen large falls in the proportion of the
labour force which is unqualified and rises in enrolment rates in
further and higher education, changes which tend to increase long-run
growth. But two factors have obscured the extent of the improvement.
First, the whole climate for economic growth is less favourable than it
was in the so-called Golden Age prior to the first oil shock in 1973.
Second, UK macroeconomic policy compares poorly with other OECD
countries: booms have been shorter and recessions longer, so that
microeconomic success has been masked by macroeconomic failure.
1. Introduction and summary
The great puzzle of British economic history in the years since
the Second World War is, how did Britain manage to throw away such large
advantages and finish up as a follower instead of as a leader? Britain
started the period with a substantial advantage in income per head over
most of the continental economies and she had suffered comparatively low
losses of human and physical capital during the war. This article will
not consider this question in its full generality, for which a whole
book would (and has) been required. But some attention to the reasons
for Britain's comparative failure is necessary, in order to assess
to what extent the supply side reforms of the 1980s and 1990s have been
successful in overcoming them.
Following recovery from the recession which began in 1979,
optimism began to rise. To some it seemed that the supply side reforms
of the 1980s had created an economic miracle in the UK. But since the
1990 recession, there is renewed pessimism about the future medium to
long-run prospects for the UK economy. The Treasury view is that GDP can
be expected to grow on average at about 2 1/4 per cent per annum,
implying a productivity growth rate of less than 2 per cent per annum
The corresponding productivity growth rate for the US economy (according
to the 1995 Economic Report of the President) is 1 1/4 per cent. So
despite the large productivity gap between the two countries -- the
UK's is about a third lower -- the UK is only expected to close
this gap at a very slow rate. Moreover France and Germany are also
expected by many commentators to grow faster than the UK, despite
already enjoying substantially higher living standards and despite the
UK having had faster productivity growth than these two countries in
So what happened to the UK's productivity miracle? How could
all those years of supply side reforms produce such an apparently
disappointing result? The purpose of this article is to assess the
evidence for an improvement in the underlying growth rate of output and
productivity.(1) To summarise the conclusion, I argue that the
improvement was genuine and moreover that it was indeed caused by the
reforms, particularly the reform of the industrial relations system or
to put it more bluntly, the reduction in union power which those reforms
brought about. However, two factors have masked the extent of the
improvement. First, the whole climate for economic growth is less
favourable than it was in the so-called Golden Age of the post-war
period which seems to have come to an end around 1973. Second, at the
macro level UK policy compares poorly with other OECD countries so that
genuine microeconomic success has been masked by macroeconomic failure.
Since we do not understand the reasons for the deterioration in global
economic performance, there is not much we can do about the first
factor. However we can hope to improve the UK's own macroeconomic
It is now generally agreed that the explanation for differences
between countries in long-run growth rates should be sought in
institutions (North and Thomas 1973; Barro and Sala-i-Martin 1995,
chapter 12). Factors like low investment or low R&D expenditure may
be the proximate source of poor performance but should be regarded as
symptoms not causes. This view is particularly persuasive when the
countries concerned all have access to the same pool of scientific and
technical knowledge, when trade and investment flows are relatively
free, and in the absence of war and political upheaval. Institutions are
much longer lived than individuals and companies, carry a baggage of
history with them, and require the concerted efforts of large numbers of
people with different interests in order to change them. Moreover, they
are frequently supported for distributional and non-economic reasons.
Hence even when an intellectual case has been made that some institution
harms growth it may be impossible to establish a coalition strong enough
to change it, especially since many powerful interests are opposed to
Two major UK institutions are candidates for retarding growth in
the post-war period. The first is the system of industrial relations and
the second is the educational system. Both these institutions differed
in significant ways from their counterparts in other comparable
countries. I argue below (section 3) that there are good theoretical
reasons for expecting the pre-1980s industrial relations system to
retard growth. I also argue that the empirical evidence supports the
view that the productivity improvement was largely the result of the
reduction of union power which the reforms helped to bring about.
The failings of the British educational and training system are by
now well known. What is not so well known is that the last 10 years have
seen a large increase in educational outputs, though this outcome seems
to have been largely unplanned. The implications of this for economic
growth are discussed in section 4.
Section 5 compares UK macroeconomic performance with that of other
OECD countries. First, the most recent recession was much longer in the
UK than in almost any other OECD country. Second, the evidence suggests
that since 1970 up till the present booms have been shorter in the UK,
while recessions have been longer than in most other OECD countries. For
13 OECD countries in the period 1970-94, the greater the tendency to
short, sharp booms and long recessions, the lower the GDP growth rate.
Also, countries like the UK which spent longer in recession over this
period tended to have lower average rates of growth. If this evidence is
accepted, and if (a big if!) UK macroeconomic policy improves, then a
better estimate of the UK's prospects can be gained by looking at
the period 1979-90 and ignoring the depressing experience of more recent
Finally, Section 6 offers some concluding comments. But before
getting on to explanations and controversies, we should recall the
reasons why people were once so optimistic and look at the UK's
growth and productivity performance in quantitative terms. This is the
task of the next section.
2. The improvement in UK productivity
There is evidence for an improvement in UK labour productivity
growth in both manufacturing and services. For manufacturing, we also
have estimates of total factor productivity and here too there is a
sharp improvement. However, the improvement is relative to the 1970s.
Relative to the UK's performance in the 1950s and 1960s, the
picture is one of recovery rather than improvement.
Table 1 compares the growth rate of labour productivity (output
per hour) in manufacturing for 12 major industrial countries.(3) In
1960-73, the UK had the lowest growth rate except for the United States
and in 1973-79, the lowest of all 12. However in the period 1979-89, UK
productivity grew more rapidly than in any other country, except for
Belgium and Japan. Over the longer period 1979-94, the UK is surpassed
only by Japan. UK productivity has also continued to grow rapidly even
through the latest recession. All 12 countries have had lower
productivity growth since 1979 than they did in 1960-73. Only 3 of the
12, the US, Sweden and the UK, have had higher growth in the most recent
period than they did in 1973-79. But the UK shows by far most striking
improvement. Table 1. Growth of output per hour in manufacturing, 12
countries: selected periods
Note: (a) 1979-93. Source: US Department of Labor, BLS News,
September 1995 (except US 1960-73, from BLS News, August 1991).
Table 2 compares UK productivity growth (output per person
employed in this case) in manufacturing, private services and the whole
economy, 1960-93. The figures for manufacturing and GDP are the only
official productivity statistics which are currently published. In
manufacturing the growth rate of output per person employed fell
drastically in 1973-79 but then recovered equally sharply in 1979-89.
Even including the recession years (1990-93) does not alter the picture
much. The growth rate in 1979-89 even exceeds slightly the rate during
the Golden Age of 1960-73. Manufacturing is nowadays only about a fifth
of the economy so Table 2 also shows estimates for productivity growth
in private services. These estimates exclude the large part of the
service sector for which productivity growth is zero by assumption of
the national income statisticians. The service industries included made
up 36 per cent of GDP in 1990. It has not yet been possible to push
these figures back before 1973. Here too we can see that labour
productivity growth increased sharply after 1979.
Note: Manufacturing: 23.8 per cent of 1990 GDP; Private services:
36.3 per cent of 1990 GDP Omitted sectors: Energy & water (10.6 per
cent of GDP); Agriculture (1.8 per cent); Construction (5.9 per cent);
Other (mainly public) services (21.5 per cent). Source: Manufacturing
and whole economy: Economic Trends Annual Supplement 199S, Table 3.5.
Private services: own estimates, weighted average of 5 service sectors
(36.3 per cent of 1990 GDP); output from 1994 Blue Book, employment from
Historical Statistics (Employment Department).
Total factor productivity growth, which allows for increased
efficiency in the use of capital and of intermediate input as well as of
labour, also rose substantially in 1979-86 compared with the 1973-79
period (Table 3). These estimates are based on the Census of Production
rather than national income statistics and stop in 1986. Using a
different concept of output (gross output rather than the more usual
value added), labour productivity growth in manufacturing is also seen
to recover sharply in 1979-86 after the very sharp fall in 1973-79.
However its growth is still lower than in 1954-73 (Table 3).
Source: Oulton and O'Mahony (1994, Table 5.5).
The improved UK performance has led to a substantial narrowing of
the gap between ourselves and other countries (Oulton 1994; see also
Eltis and Higham and O'Mahony in this issue of the Review). On the
other hand, if we look at output rather than productivity, the picture
is not so encouraging, particularly if we include the years from 1990
onwards (Table 4). Again there is a substantial improvement over the
1973-79 period. But even excluding the recent recession GDP growth was
still lower in 1979-89 than in 1960-73 by about 0.8 per cent per annum
From 1979 to 1994, GDP grew at 1.85 per cent per annum, which is better
than the rate achieved over 1973-79, but not to a startling extent.
Source: Economic Trends Annual Supplement 1995.
Long term growth rate comparisons
To get a better perspective on the UK's performance, data
from the Penn World Table (version 5.6) have been employed. A
distinguishing feature of this dataset is that all national income
magnitudes are measured in a common set of international prices (derived
from the various rounds of international price comparisons carried out
by international agencies). Two periods have been distinguished: 1960-73
and 1973-89. The beginning and end-dates are constrained by data
availability plus the desire not to contaminate the results with the
recent recession. The variable to be analysed is the growth rate of GDP
per worker. In order to further minimise endpoint effects, this is the
trend rate of growth in each period (derived by fitting a loglinear time
trend to each countries GDP per worker series).
The choice of countries to analyse was motivated by a number of
factors. The Penn World Table contains data on 152 countries, many very
small and many very poor. For present purposes it seems sufficient to
analyse countries which are currently reasonably large and rich and 2.5
including the UK were selected. Inclusion of middle income countries
might have been desirable but many of these countries had failing income
in the 1980s. Controlling for these factors would have complicated the
Trend growth rates for the UK and the other 24 currently
successful industrialised countries in the two periods are shown in
Table 5. As Chart 1 shows more clearly, in the period 1960-73 the UK
came fourth from the bottom. Chart 2 makes the same comparison for
1973-89. The UK's relative position is greatly improved. We are now
eighth from the top. However, Table 5 also shows that in 1973-89 the
performance of every single one of these countries, including high
fliers like Hong Kong and Singapore, deteriorated compared to 1960-73.
The average growth rate in these 25 countries was 4.31 per cent per
annum in 1960-73, but only 1.62 per cent per annum in 1973-89. Apart
from Norway, the UK's decline was the smallest. So in this rather
back-handed sense the UK comes nearly top of the league. Table 5. Trend
growth of GDP per worker, 25 countries: 1960-73 and 1973-89
Note: Trend growth rates calculated by loglinear regression on time.
Source: Penn World Table, mark 5.6 (November 1994) and own calculations.
The fact that every country's performance has deteriorated
indicates some adverse change has occurred in the climate for economic
progress, the reasons for which are not currently understood (Crafts and
Toniolo 1995). Discussion of this issue would take us too far afield.
However, we can still ask how well or badly the UK has done given what
could reasonably expected in the circumstances. Accordingly,
cross-country regressions of a fairly standard form have been fitted for
each period. The dependent variable is the trend growth rate of GDP per
worker ([Delta]y). The independent variables are the ratio of gross
investment to GDP averaged over each period (I/Y) and the initial
productivity gap between country i and the United States (GAPUS),
defined for 1960 as log(US GDP per worker in 1960 [?] country i's
GDP per worker in 1960) and analogously for 1973. These three variables
are measured as percentages. The investment ratio is as emphasised
earlier only a proximate determinant of growth but is included here in
default of measures of the relevant institutional variables for the 25
countries. It would have been highly desirable to include a measure of
investment in human capital as well but for reasons discussed below none
of the available measures seems satisfactory. The initial productivity
gap with the US is included because of the convergence hypothesis, the
widely-held view that the bigger the gap the greater the opportunities
for growth, which can be justified on both old and new growth theory
grounds (Dowrick and Nguyen 1989). Other variables which were tried but
found to be insignificant were the ratio of government consumption to
GDP and the growth rate of the labour force to pick up possible scale
effects). The results, with robust t statistics in brackets, were as
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
For 1960-73, these two variables explain a high proportion of the
variance of growth rates. GAPUS is highly significant and indicates a
convergence rate of some 2.7 per cent per annum. The investment ratio is
significant at the 6 per cent level. For 1973-89 the explanatory power
of the equation has clearly deteriorated. GAPUS is now only significant
at the 6 per cent level and the rate of convergence seems to have
fallen. The investment ratio is now only significant at the 14 per cent
level. The coefficient has fallen by a quarter, implying a substantial
decline in the return to investment.(4)
To some extent the UK's performance in the first period could
be excused since in 1960 the UK was the eighth richest country out of
25. But the UK's poor performance was also a result of a low
investment ratio, only 18.8 per cent of GDP compared with 28.5 per cent
for France, 31.1 per cent for Germany and 31.5 per cent for Italy. Even
so, the UK's actual growth rate fell below its predicted level by
0.31 per cent per annum And a low investment ratio is not the whole
story, since even if the UK had invested as much as did Germany the
growth rate would be predicted to rise to only 3.52 per cent, still
below Germany's 4.22 per cent.(5)
In the period 1973-89, the UK continued to have a low investment
ratio, in fact slightly lower than in the first period. The regression
equation predicts that the UK should have grown at 0.98 per cent per
annum In fact, the growth rate was 1.66 per cent, so in contrast to
1960-73 the UK exceeded expectations. Despite France and Germany
continuing to have much higher investment ratios, these two countries
managed to grow at only 1.20 per cent per annum
These results indicate that a genuine improvement in productivity
occurred though against a background of a less favourable international
environment. The period of relative economic decline has ended. It
should be noted that for one third of the second period, 1973-79, the UK
is almost universally agreed to have performed very badly, making the
overall result still more striking.
3. The effect of trade unions on productivity growth
The UK system of industrial relations
The UK industrial relations system exhibited a number of
distinctive features. Each of these features could be paralleled
elsewhere when taken individually but collectively they made the system
unique in the industrialised world.(6) First, union membership was high
and rising, particularly in the 1970s. By 1979, 53 per cent of employees
in employment were union members (Metcalf 1994, Table 4.1). Second, by
its heyday in the late 1960s the dual system described by the Donovan
Commission had arisen. Negotiations took place not only at company or
national level with the union's central staff but also at local,
plant level with shop stewards. Third, as a legacy of history, unions
were organised neither on a company basis as in Japan nor on an
industrial basis as in continental Europe or the United States, but on
the basis of crafts (supplemented by some large general unions). An
implication of this structure was multi-unionism, the presence of more
than one union in the workplace. Multi-unionism was particularly
important in large plants. In 1980, 41 per cent of establishments
employing 2,000 people or more had 3-5 manual unions present and a
further 38 per cent had 6 or more present. (Daniel and Millward 1983,
Table 11.21). This is significant because though large plants are small
in number they had a disproportionately large share of total employment
in this period. For example, plants employing 1,000 or more accounted
for 41 per cent of total manufacturing employment in 1979 and this
proportion had been increasing over the post-war period (Oulton 1987,
Tables 2 and 3). Fourth, under the British tradition of
`voluntarism', trade unions enjoyed a wide range of legal
immunities. Activities such as secondary picketing which would have
attracted civil or even criminal penalties in other countries such as
the United States were not actionable in Britain.
Though it is something of a commonplace amongst financial
journalists and economic commentators that trade unions have adversely
affected UK economic growth, this proposition is still quite
controversial in academic circles. The standard economic history of
Britain since the Second World War devotes some 16 pages out of 427 to
trade unions (Floud and McCloskey 1994). In a long chapter entitled
`Economic and commercial performance since 1950', Millward (1994)
has only a couple of pages on the role of trade unions, mostly devoted
to arguing that their role was very similar to that in the US.(7) The
argument of this article is not that unions always and everywhere reduce
productivity growth, only that they tended to do so in the UK. I turn
first to theoretical considerations before discussing the empirical
Unions and productivity growth: theory
It has long been recognised that if unions push up wages then
firms will wish to adopt more capital intensive techniques, so raising
the level of labour productivity. On the other hand, restrictive
practices would tend to lower labour productivity, so the net effect on
the level seems unclear a priori. It is not so obvious that unions
necessarily reduce the growth rate of productivity. Indeed, Freeman and
Medoff (1984) argued that trade unions give workers a voice and so could
raise the productivity growth rate as a result of more consultation and
better information flows. Their work was widely cited in the UK
literature, even though their evidence was drawn from the very different
institutional setting of the US.
There are in fact a number of theoretical reasons to expect an
adverse effect of unions, of least of the UK kind, on productivity
1. Up till the 1980s, the UK system of industrial relations required,
particularly in large plants, that several different unions had to be
able to reach agreement with management on changes in working methods.
If they could not agree, the status quo continued. The costs of change,
in terms of management time or interruptions to production, were high
and the probability of no agreement at the end of the day was not
negligible. It seems very likely that the greater the number of partners
to a negotiation, each with the power of veto, the lower is the
probability of agreement. Common sense suggests that change is likely to
occur at a slower pace than in systems not suffering from these
handicaps. However, though this situation cries out for treatment from a
game-theoretic point of view, I am not aware of any game theory results
which are directly applicable.(8)
2. In the standard model of investment with adjustment costs, the
rate of growth of the firm's capital stock is a function of the
marginal cost of adjustment (Lucas 1967). If unions raise this marginal
cost, for example by protracted negotiations on work practices, then the
growth rate of the capital stock will be lower.(9)
3. Where it is not possible to make binding contracts, union-firm
bargaining will result in a lower level of investment (Grout 1984). The
reason is that if investment is at least partially irreversible, then
any agreement today on wages and conditions risks a hold-up problem
tomorrow. The union cannot guarantee that it will not attempt to skim
off the firm's profit from its planned investment. Knowing this,
the firm invests less. Though this is technically a level rather than a
growth rate argument, it is certainly plausible that a growth rate
analogue of the argument exists and one has in fact been supplied by
Bean and Crafts (1995). Their engine of growth is expenditure on R&D
to develop new intermediate inputs. Unions cause a reduction of such
expenditure and so reduce the growth rate.
4. The arguments above suggest that investment and process innovation
can be slowed down by trade unions. It is also possible that product
innovation can be slowed down. Suppose that firms are in monopolistic
competition. They can raise demand for their existing products or
develop new products by expenditure on design, marketing, R&D etc.
These costs are distinct from ordinary production costs since they are
of an overhead nature and do not vary with the level of output. Product
improvement which raises the price which consumers are willing to pay is
just as much a productivity gain as is reducing the production cost of a
given product. If unions raise the wages of production workers or lower
their productivity through restrictive practices at a given wage, then
firms will spend less on product improvement. The reason is that with
higher production costs, the output over which the overhead expenditure
can be spread is lower.(10) Though this is also strictly speaking a
level of productivity argument, a dynamic version would run in terms of
firms in international competition failing to spend sufficiently on
product improvement and so growing more slowly while losing market
These theoretical arguments suggest the following broad
conclusion. Multi-unionism, multi-level bargaining, and the absence of
binding contracts meant that bargains which could potentially make all
parties better off, with losers compensated by gainers (potential Pareto
improvements), could frequently not be reached by negotiation. In this
situation, two kinds of reform could be envisaged. In the first, the
existing unions would be abolished and replaced by either company or
industrial unions within a firm legal framework. Unions would remain
powerful but they would become `encompassing' institutions as in
more successful systems. However, it is difficult to see how this could
have been carried out in practice and no-one seems to have seriously
suggested this. The second course was simply to reduce union power and
this is what the legislation of the last 15 years has progressively
done. Union recognition rights have been restricted. The closed shop
which involved 5 million workers in 1979 has been outlawed. The types of
industrial action which are legal have been restricted, with potentially
severe civil penalties for infringement available via the courts. The
requirement for balloting has made it more difficult to call strikes and
has therefore reduced the effectiveness of a threat to strike.(11)
The reduction in union power manifested itself in a number of
ways. First, there was a straightforward fall in union density, down
from 53 per cent in 1979 to 40 per cent in 1990, about the same as in
1960. Second, trade union recognition fell sharply, from about two
thirds of all establishments in 1984 to about half by 1990 (Metcalf
1994, Tables 4.1 and 4.2).(12) Thirdly, even in establishments which
continued to recognise unions there is ample anecdotal evidence for a
more co-operative approach.
However the reduction of union power, particularly shopfloor power
over work practices, was not solely due to the legislation, which tended
to interact with other, complementary policies. First, the policy of `no
bail outs' and of allowing lame ducks to die strengthened the hand
of management in forcing through change. Second, the credibility of no
bail outs' was raised by the early privatisations and sell-offs
(British Leyland (now Rover), British Telecom, British Airways and
British Aerospace), which made political interference less likely.
Third, contracting out struck at union monopoly power in the public
sector. Finally, it seems that a change in public opinion (which the
government had itself fostered) allowed the government to use the powers
of the state, e.g. employing the police to confront mass picketing, in
ways which governments of the 1970s felt inhibited from doing.
Unions and productivity growth: evidence
When it first became apparent, the productivity improvement was
often dismissed as an effect of the 1979-81 recession. In one version,
the recession caused a one-off reduction in overmanning. In another
version, closures killed off low productivity plants raising the average
productivity of the survivors (the `batting average' effect).
Though both these factors may have played some role, the persistence of
the improvement up to and through the most recent recession makes
exclusive reliance on such explanations increasingly implausible.
The widely cited survey article of Brown and Wadhwani (1990)
sceptical on whether the trade union legislation of the 1980s had had an
important role in the productivity improvement. They concluded: `The
impact of the legislation on Britain's trade unions is undeniable;
its impact on Britain's economic performance is far less
obvious.' However, the evidence they themselves cite on
productivity tends to the opposite conclusion. For example, Machin and
Wadhwani (1989) found that unionised establishments had experienced more
`organisational change' in the terminology of the 1984 Workplace
Industrial Relations Survey. Brown and Wadhwani also cite a panel study
of firms by Nickell, Wadhwani and Wall (1989). The latter found that
total factor productivity growth was higher in heavily unionised firms
over the period 1980-84. In fact, Brown and Wadhwani conclude this part
of their survey by saying: `These studies suggest that the early years
of the legislation were associated with an increase in productivity
which was especially marked for unionised establishments'.
Since they wrote, evidence of several kinds has continued to
accumulate. First, there is a vast mass of anecdotal material. Second,
there are studies using industry level data. Third, there are studies
using firm level data. Firm level studies are obviously very
illuminating since we can hope to catch changes in behaviour under the
influence of new incentives. But industry level studies are also
important since part of the process of change is the rise of firms which
respond to new opportunities and the decline of those who do not. This
`biological selection' effect can be missed by panel studies of
firms. For example, if panels are balanced, firms which fail may never
appear; if panels are unbalanced, firms which fail to adapt go out of
business and drop out of the panel. In either case, a policy change
could produce a productivity improvement but this would not be picked
Amongst the industry level studies, Oulton (1990) showed that the
productivity improvement in manufacturing of the 1980s was greater in
the industries which had been more heavily unionised in the 1970s. The
disadvantage associated with operating large plants also appeared to
have fallen. This study also controlled for inter alia the shock effect
of the 1979-81 recession, the rise in energy prices, and vintage type
effects on productivity arising from capital scrapping (see also Oulton
1987 and 1989b on the role of closures and vintage effects).
Bean and Crafts (1995) employ the industry dataset developed by
Oulton and O'Mahony (1994) combined with data from the Workplace
Industrial Relations Survey on multi-unionism to show that the great
change in the 1980s was the disappearance of the disadvantage previously
associated with multi-unionism. Since multi-unionism is particularly
important in large plants, this finding is consistent with the earlier
one in Oulton (1990).
Amongst studies using firm level data, Denny and Machin (1992)
showed that unionisation tended to reduce investment in the period
1973-85. Their figure for the gross effect is 28 per cent though there
are offsetting factors which make the net effect smaller. Their data
does not appear to allow an estimate of any improvement in the 1980s.
Gregg et al. (1993) find that the disadvantages of unionisation for
productivity growth have been reduced or even eliminated in the later
1980s as a result of management reasserting the `right to manage'.
Machin and Stewart (1995) find that financial performance of unionised
firms has improved in the 1980s. Only in a small minority of plants
where unions still seem to retain their traditional power do they now
have a significant effect on financial performance. If investment
depends on profitability, then this should eventually lead to an
increase in investment of all types. Indeed, on a `new growth
theory' view the higher the rate of profit the higher the rate of
4. Education and training
From an economic point of view, every educational system has two
aims. The first is to select and train an elite. The second is to
provide the whole population with an adequate level of education. Every
system puts some weight on each of these two objectives but the balance
differs between countries. In the British case, there can be no doubt
that the bulk of the effort went into selecting the elite. Until quite
recently, the overwhelming majority of British children left school at
the earliest legal opportunity, frequently with no formal qualification
other than that of having attended school for the prescribed period. Of
these early leavers, a minority of the male half entered a traditional
apprenticeship; the rest received little or no formal education or
training for the rest of their lives. Potential members of the elite on
the other hand were carefully selected in a series of stages through the
11 plus exam, O-levels (as they then were), A-levels and finally
university degrees. The main changes in educational policy of the
post-war period -- the raising of the school leaving age from 14 to 15
in 1945 and from 15 to 16 in 1972, the Robbins expansion of higher
education in the late 1960s, comprehensivisation in the 1970s -- did
little to correct this underlying bias.
As late as 1975, 70 per cent of the population aged 16-69 had
essentially no qualifications, educational or vocational (Table 6). Even
by 1987, 60 per cent of the UK's manufacturing labour force had no
vocational qualifications, compared with 29 per cent in (Western)
Germany (Table 7). By comparison with the rest of Western Europe,(13) it
would now be generally agreed that the UK's main deficiency has
been the low level of basic education and vocational training bestowed
on those in the lower half of the academic ability range. The extent of
the gap and the economic disadvantages which flow from it have been
demonstrated in numerous studies emanating from the National
Table 6. Highest qualification level attained: Great Britain
Key: Higher intermediate: Higher education below degree level
Lower intermediate: GCSE or O-level, grades A-C
Low or no qualifications: at best GCSE or O-level, grades D-E
Source: General Household Survey 1993, Table 10.1.
Table 7. Vocational qualifications of the manufacturing workforce, UK
and Germany, 1987
Note: Definitions of levels differ from those of Table 7. Source:
However, recent years have seen some dramatic changes. The
proportion of the population with essentially no qualifications has
fallen from 70 per cent in 1975 to 44 per cent in 1993 and is set to
decline further (Table 6). Partly this reflects the steady process of
demographic change: the exit from the labour force of older, less
qualified workers and their replacement by younger, more qualified ones.
But it is also starting to reflect a very sharp rise in enrolment rates
in both further and higher education. The proportion of 16 year olds
enrolled in full time, post-compulsory education was flat during the
first half of the 1980s and then started moving sharply upwards. It rose
from 47 per cent in 1986 to 73 per cent in 1993. The proportion of young
people enrolled in full time university education was 13 per cent in
1982 but had become 28 per cent by 1992, rising particularly sharply
after 1990. (Department for Education, Statistical Bulletin, 10/94 and
13/94). This expansion has been primarily demand led: young people have
voted with their feet. But it was facilitated by an important policy
change, the replacement of O-level, one of the traditional stages of
elite selection, with GCSE which has enabled a much higher proportion to
achieve an adequate level and so qualify for further education (Green
and Steedman 1995).(15)
How much of the gap between the UK and comparable countries has
been closed? International comparisons of educational attainments are
fraught with difficulties of inconsistent data collection and
inconsistent definitions (Steedman 1995) and there are so far no studies
which take account of these recent changes. Although other countries
have not stood still, it seems likely that there has been some closing
of the gap, if only because the UK was previously so far behind. But it
would certainly be a mistake to assume that all problems have been
solved, since there are continuing questions about the quality of basic
education. Although over 40 per cent of pupils gained 5 or more GCSE A
grades in 1994, far fewer did so in basic subjects. Only a third of
pupils gained a GCSE grade A-C in all three of English, Maths and a
science subject (National Foundation for Educational Research 1995).
A further and more difficult question is, how much of an effect on
economic growth can be expected from the increase in the proportion of
qualified people? Theory suggests two ways in which human capital may
influence growth. The traditional, `old growth theory' view as well
as that of the growth accounting literature is that a higher growth rate
of human capital is associated with a higher growth rate of output.
According to old growth theory, the ultimate sources of output and
productivity growth lie elsewhere but nevertheless growth in human
capital just like in physical capital is an essential accompaniment, a
necessary though not a sufficient condition (Mankiw et al. 1992).
According to `new growth theory' on the other hand, a higher
level of human capital causes a higher growth rate of output, either
because the rate of growth of new, useful knowledge depends on the level
of human capital (Lucas 1988) or because in a follower country a higher
level of human capital makes for a faster rate of absorption of
innovations generated in leading countries, i.e. faster convergence
(Nelson and Phelps 1966).
A difficulty with both old and new growth theory is that `human
capital' is an empty box. Nothing in the theory tells us whether
what matters is high level skills (Ph.D.s), vocational training or basic
literacy and numeracy. Still less does it tells us the answer to finer
questions such as whether we should be encouraging first degrees in
engineering rather than in media studies.
Empirically, human capital has been identified with various
measures of educational attainment and some support has been found for
both old and new growth theory views (Mankiw et al. 1992; Barro and
Sala-i-Martin 1995, chapter 12), though sometime the results are
embarrassing. For example, Barro and Sala-i-Martin find that the higher
the initial level of education of men, the higher the subsequent growth
rate. But the higher the initial level of women, the lower the
subsequent growth rate. They then struggle to resist the conclusion that
growth can be accelerated by educating fewer women and more men.
However, the main results come from cross section regressions of
countries at very different levels of development. Even if it is found
that poor countries can raise growth by increasing the proportion of
their populations with primary and secondary education, nothing much
seems to follow for the policies of countries like the UK which already
have 100 per cent primary and secondary education.
A further, major difficulty in the way of quantifying the effect
of education on growth is the quality of the data on educational
attainments. These may be adequate for drawing broad brush conclusions
comparing poor with rich countries, but are much less satisfactory for
comparisons amongst developed countries. The Barro-Lee dataset is the
best currently available (Barro and Lee 1993) but examination of some of
the numbers raises many doubts. According to this dataset, the
percentage of the population in 1985 reckoned to have `attained
secondary' education was 38.0 per cent in the UK but only 19.5 per
cent in (Western) Germany; by contrast neighbouring Austria had 30.2 per
cent., France 26.3 per cent and Italy 30.7 per cent. On a different
measure, average years of total schooling embodied in the population in
1985, Germany does much better than Austria (8.54 years compared to
6.64) but still just worse than the UK (8.65). However Germany and the
UK are substantially ahead of France and Italy (6.52 and 6.28 years
respectively). Even if we ignore the issue of quality of education,
uncritical reliance on these figures would be unwise.(16)
5. UK macroeconomic performance
The UK has experienced three major (and two minor) recessions
since 1973. Table 8 compares the length of the three major ones. The
length of a recession is defined as the number of quarters it takes for
output to regain its previous peak level (more on this measure below).
From this table we can see that the recent recession lasted almost as
long as the one which started in 1979 and longer than the 1973 one. It
took 4 years for output to regain its previous level. The previous two
recessions coincided with major shocks to the world economy. This was
not the case for the 1990 recession, even though there were some common
features such as preceding asset price booms in a number of countries.
The depth and severity of the 1990 recession, greater than in almost any
other OECD country, were home-grown, and must therefore have been
largely due to domestic policy errors.(17)
Table 8. Recessions compared: UK quarterly GDP
Note: Time to recover is number of quarters before previous peak
output is regained. Source: Economic Trends Annual Supplement 1995.
The most likely explanation is a double policy error: first, an
overexpansion of demand in the Lawson boom, and second, the decision to
enter the ERM at a greatly overvalued exchange rate (Wren-Lewis et al.
1991; Barrell et al. 1994). Financial liberalisation, which is often
blamed for the Lawson boom, caused difficulties in other countries but
need not have led to such severe results. For example, it took only 7
quarters for output to regain its previous peak level in the US. We do
not need to decide which was the more important error, the Lawson boom
or ERM entry, only to note that these errors may have severely affected
Macroeconomic policy: a comparison across the OECD
If the UK's record turns out to be substantially the same as
that of other countries, there would be little reason for optimism that
the record in the future would be any better. So there is an interest in
seeing whether in fact other countries have been more successful. The
best way to assess the UK's record is to compute the optimal policy
using a model of the UK economy, and calculate the deviation between the
present value of the utility stream under the optimal policy and the
present value under the policy actually followed. The same operation
must be repeated for every other country in the comparison. This is
obviously a mammoth undertaking. In any case, the results will be
dependent on the models chosen and on the form of the assumed utility
function. A less demanding but still complex approach is the computation
of output gaps which employs a production function to estimate trend
output levels (Barrell and Sefton 1995).
Instead a cruder approach will be employed, by looking at various
summary measures of the growth of GDP. in order to assess success or
failure in macroeconomic management, it seems essential to look at
quarterly data. The OECD transmits quarterly GDP data for 20 countries.
Five of these only have data from 1980Q1 or later: Mexico, New Zealand,
Norway, Sweden and Turkey, while Portugal's stops in 1993Q4. 14
countries have data from 1977Q1 to the present (1994Q4 except for the US
which is 1995Q1). Of these 14, 11 including the UK have data from 1970Q1
to the present.
There is a surprising amount of variety in the experience of these
20, mostly very similar, countries. Thus Canada and Australia largely
escaped the first oil shock, whereas in Switzerland the effect was
severe. Finland has yet to recover from the current world recession, but
this is the exception that proves the rule, since unlike Britain Finland
has suffered a severe external shock in the collapse of its major
trading partner, the Soviet Union. The UK, France, Germany, Italy and
the US all show a similar pattern over this period, with the three major
recessions all being clearly visible. Japan's progress appears much
smoother until the most recent recession since when stagnation has
The mean, variance and skewness of the quarterly growth rate of
GDP(19) for the 11 and the 14 countries are shown in Tables 9 and 10
respectively. The range of growth rates is not very great. Over the
longer period, japan comes top with 3.71 per cent per annum and
Switzerland bottom with 1.52 per cent per annum The UK comes tenth out
of the 11 and also tenth out of the 14.
Table 9. Quarterly growth rates of GDP, 1970Q1-1994Q4: 11 countries
Note: Austria, not seasonally adjusted in the source, has been
seasonally adjusted. Source: OECD Quarterly National Accounts (May 1995
Table 10. Quarterly growth rates of GDP, 1977Q2-1994Q4: 14 countries
Note: Austria, Denmark and Finland, not seasonally adjusted in the
source, have been seasonally adjusted. Source: OECD Quarterly National
Accounts (May 1995 diskette).
The most obvious measure of macroeconomic stability is the
variance of the growth rate. Here the UK has the third highest variance
out of 11 and the 6th highest out of 14, suggesting a poor to moderate
performance.(20) However, the variance suffers from two drawbacks.
First, it is symmetrical as between deviations above and below the mean.
But it is surely reasonable to score deviations below the mean more
heavily, since booms are generally seen as good times, their only
drawback being that they may require policy to be tightened in order to
prevent inflation taking off.(21) Second, the variance does not
distinguish between short period fluctuations with little social cost,
and long lasting recessions.
Asymmetry as between booms and slumps can be captured by the
skewness. A positive value means that the distribution of growth rates
is characterised by a relatively small number of values considerably
above the mean and a relatively large number a bit below the mean. In
more intuitive terms, positive skewness means that booms tend to be
short and sharp while recessions tend to be long and shallow.(22) Over
the longer period (Table 10), 7 countries out of 11 have negative
skewness and 4 positive. The UK is the most positively skewed. Over the
shorter period, skewness is negative for 12 of the 14 countries and only
for the Netherlands and the UK is it positive.
These results could be criticised as misleading since the variance
and the skewness are both measured relative to the mean which may not be
a good measure of the trend growth rate. The analysis has therefore been
repeated after detrending the data (in log level form) using the
Hodrick-Prescott (HP) filter. The HP filter, which is designed to be
applied to an original series in log form, divides the series into two
components, a trend and a cyclical component, such that the original
series is the sum of the two. By construction, the cyclical component
has mean zero over the sample period. The HP filter is a fairly flexible
and objective method of trend extraction but can also be criticised as
being, along with other, similar methods, essentially a smoothing
operator. A prolonged recession will show up as a fall in the trend rate
of growth followed by a subsequent rise.(23)
The variance and skewness of the cyclical components of GDP for
the 14 countries appear in Table 11. The UK has the second highest level
of skewness with Japan now coming top. To test whether growth is
affected by skewness, Chart 3 plots skewness against the trend rate of
growth (calculated as a loglinear trend for the same observations as the
skewness). It is clear that a strongly negative relationship exists for
13 of the 14 countries with Japan being the odd man out (r = -0.72).
Japan's growth record is most unusual since progress was very
steady right up till the 1990s. For these 13 countries, a parsimonious
explanation of their trend growth is provided by the following
regression (robust t statistics in brackets):
Trend growth = 0.0212 - 0.0101 Skewness
(3.5) N = 13; [R.sup.2] = 0.67
Here GAPUS is the gap with the US as defined in section 2. The
investment ratio was also tried but was not significant (its inclusion
did not affect the coefficient on skewness). The variance was also
insignificant whether included on its own or together with the skewness.
Table 11. Hodrick-Prescott detrended quarterly GDP: characteristics
of cyclical component
Note: Cyclical component calculated with A set to 1600 in HP filter.
Source: OECD Quarterly National Accounts, May 1995 diskette.
This result shows that a cyclical pattern like the UK's --
short, sharp booms followed by long, shallow recessions -- is associated
with a lower trend growth rate. It is possible that long recessions
damage confidence and so reduce investment and R&D expenditure.
Moreover the effect is economically significant. If the UK had had the
average skewness of this sample, its growth rate would be predicted to
have been 0.57 percentage points higher. Since macroeconomic policy is
usually thought capable of affecting output over business cycle
frequencies, this suggests that greater success in stabilising the
economy would increase the trend growth rate.
An alternative measure of asymmetry
An alternative measure capturing the intuitive notion that
recessions are socially costly is the proportion of time which the
economy spends in recession. But here we must be precise. The economy
will be defined as being in recession when output is below its previous
peak level. If output is actually below what has been previously
achieved, it suggests an obvious malfunction of the economic system. The
economy will be said to recover in the quarter when output first
surpasses its previous peak level. The length of a recession is then the
number of quarters before recovery occurs.
To implement this measure, we need also to define what is meant by
a peak. Two measures will be used. Under measure 1, a peak occurs if
output is higher than in the preceding quarter and output falls for at
least the next two quarters (i.e. y(t)>[is greater than]y(t-1),
y(t)[is greater than]y(t+1) and y(t+1)>[is greater than]y(t+2), where
y is GDP). We add the rider that each potential peak, to qualify, must
be higher than its predecessor. Measure 2 is very similar, except that
we only require output in the next two quarters to be lower than at the
potential peak (i.e. y(t)[is greater than]y(t-1), y(t)[is greater
than]y(t+1) and y(t)[is greater than]y(t+2)). Measure 1 corresponds to
the official US definition of a recession. These measures differ from
the one commonly used, which says that a recession ends when output
regains its trend level. This of course presupposes that we can define
what is meant by the trend level of output.
Table 12 shows the proportion of time that these 14 economies have
spent in recession. On measure 1, the UK comes second, with 48 out of
100 quarters spent in recession. Switzerland is top; however this is due
almost entirely to the depth of the Swiss recession following the first
oil shock. Under measure 2, the UK is pushed down into 3rd place by
Denmark and Austria. So on either of these two measures, the UK's
performance is amongst the worst. By contrast, even on the more
pessimistic measure 2, France has spent only 23 per cent of the period
in recession, Germany 35 per cent and the US 26 per cent.
[TABULAR DATA NOT REPRODUCIBLE IN ASCII]
There is also a negative correlation between the mean growth rate
and the proportion of time spent in recession for these 14 countries.
For measure 1, the correlation is [0.74 and for measure 2, it is [0.66,
suggesting that comparatively poor macroeconomic performance may have
reduced the average growth rate. A sceptic could argue that countries
which have a low average growth rate for some other reason will also be
more likely to have actual falls in output levels (rather than just
falls in growth rates) and so will spend longer in recession as defined
here. This is true but the present finding is in accordance with the
earlier, negative correlation between skewness and trend growth which is
not vulnerable to this criticism.
Macroeconomic policy errors and UK prospects
As we have seen, the UK's growth may have been damaged by
macroeconomic policies leading to short booms and long recessions. Even
if this analysis is not accepted, any method of trend fitting which
includes the recent recession is likely to produce a relatively
pessimistic forecast for the UK's medium to long term growth rate.
It is a sound statistical principle not to throw information away.
However, if poor UK performance primarily reflects macroeconomic policy
errors, then it may be correct to ignore or at least downgrade the
evidence of recessions -- provided of course that the errors are not
likely to be repeated! This point is illustrated with reference to an
extreme case, the Great Depression in the US.
Consider the problem of someone trying in 1937 to forecast US
economic growth over the long term. As it turned out, though this
hypothetical individual was not to know this, 1937 was a cyclical peak.
In fact, the US economy was not to regain the 1929 level of output till
1939 after recovering from another recession, a full ten years after the
onset of the Great Depression.
Table 13 sets out average peak-to-peak growth rates over selected
periods. The problem for the 1937 forecaster is how much weight to give
to the Depression. The average annual growth rate over 1919-29 was 3.48
per cent, but from 1919 to 1937 was only 1.80 per cent and was negative
from 1929 to 1937. In 1937 it might have seemed absurdly optimistic to
discard the whole experience of the Depression which we know caused deep
gloom about the future of the capitalist system itself. More
sophisticated methods such as the HP filter would also have suggested a
pessimistic view of US prospects. However, as things turned out, our
forecaster would have done best by simply ignoring the Great Depression
as if it had never happened. Even then he would have underestimated the
growth which actually occurred (e.g. 4.63 per cent per annum over
Table 13. Average annual growth rates of US GNP, selected periods,
Note: 1913, 1929, 1937, 1953, 1962 and 1973 were cyclical peaks. 1919
and 1948 were first -- normal' years after WW1 and WW2
respectively. 1937 output was still below 1929 (1939 was recovery year).
1941 was last year of peace before WW2. Source: 1913-28: Romer's
revision of official Commerce Department series (Romer 1988, Table 6).
1929-73: U.S. Dept. of Commerce, National Income and Product Accounts of
the United States, 1929-82, Table 1.2 (1986).
The explanation of this paradox came much later. As Friedman and
Schwartz (1963) showed, it was a gross policy error which turned what
would otherwise have been a normal recession into a major slump. The
failure of the Federal Reserve to perform the function for which it had
been set up, namely to act as a lender of last resort, caused the
catastrophic collapse of the US banking system and a massive decline in
the money stock. Though debate continues about the transmission
mechanism from money to output, the Friedman-Schwartz argument has
received wide acceptance (Eichengreen 1992). If our 1937 forecaster had
had access to their analysis, he might have reasoned that the Fed was
unlikely (to be allowed) to make the same mistake twice, hence he could
ignore the Depression experience as a one-off. And he would have been
By analogy, if we could be confident that recent macroeconomic
policy errors will not be repeated in the future, it would be better to
ignore UK performance since 1990 and base a judgement about long-run
prospects on the performance from 1979 to 1990.
The introduction identified two major institutions as inimical to
productivity growth: industrial relations and education. The failings of
both, it was argued, have been substantially alleviated though in the
case of education the payoff has perhaps yet to appear. This does not of
course mean that all problems have been solved and there is certainly
still room for many detailed policies to is certainly still room for
many detailed policies to address remaining weaknesses (see Eltis and
Higham 1995 for more on the latter). However, the assessment here is
much more optimistic than the conclusions reached by others.(25)
It must be admitted that much about the growth process remains
mysterious and few empirical regularities are robust (Levine and Renelt
1992). It may be that the UK's relative economic decline was due to
some quite different cause from the ones discussed here. If so, we must
wait for this factor X to be identified since in my judgement no other
plausible factors which are also amenable to change by policy have been
identified. For example, the UK's decline might have been due to
the loss of the entrepreneurial spirit or the absence of Asian values,
but it is hard to think of any policies to remedy these deficiencies.
Again, the UK might have fared better with German or Swedish trade
unions or with Japanese managers, but it is not clear how this
transformation could have been achieved in practice (except partially in
the latter case by foreign investment).
Finally, what of the prospects for better macroeconomic policies?
Since our ignominious exit from the ERM in September 1992, UK
macroeconomic policy has regained a measure of credibility. The Bank of
England has acquired greater de facto independence, though it is still a
long way short of the de jure independence enjoyed by the Bundesbank.
The `Wise Men' have injected an element of independent thinking
into the policy formation process. It is true that under the present
framework policy remains subject to manipulation for short term
political gains. Nonetheless, left to itself it might be hoped that the
new arrangements would continue to develop and that the lessons of the
past would be learnt.(26) Unfortunately however, the present framework
may well only be temporary. A majority of the British political class
together with their continental counterparts seems determined to embark
on (or at least not to be left out of) the great adventure of European
Monetary Union, currently scheduled to begin in 1999. The
characteristics and performance of a new, post-EMU system of
macroeconomic control are as yet impossible to predict.
Arrow, K.J. and Nerlove, M. (1962), `Optimal advertising policy under
dynamic conditions', Economica, 29, pp. 129-142.
Auerbach, A.J., Hassett, K.A., and Oliner. S.D. (1994), `Reassessing
the return to equipment investment', Quarterly Journal of
Economics, 109, pp. 789-802.
Barrell, R., Britton, A., and Pain, N. (1994), `When the time was
right? The UK experience of the ERM', In D. Cobham (ed.), European
Monetary Upheavals, Manchester University Press.
Barrell, R. and Sefton, J. (1995) `Output gaps. Some evidence from
the UK, France and Germany', National Institute Economic Review,
No. 15,1 (February), pp. 65-73.
Barro, R.J. and Lee, J-W. (1993), `International comparisons of
educational attainment', Journal of Monetary Economics, 32, pp.
Barro, R.J. and Sala-i-Martin, X. (1995), Economic Growth,
Bean, C. and Crafts, N.F.R. (1995), `British economic growth since
1945: relative economic decline ... and renaissance?', CEPR
Discussion Paper No. 1092, London.
Bean, C. and Symons, J. (1989), `Ten years of Mrs T', CEPR
Discussion Paper No. 316, London.
Blanchflower, D. and Freeman, R.B. (1994), `Did the Thatcher reforms
change British labour market performance?', In R. Barrell (ed.),
The UK Labour Market, Cambridge: Cambridge University Press.
Broadberry, S. (1994), `Employment and unemployment', In Floud
and McCloskey (1994).
Brown, W. and Wadhwani, S. (1990), `The economic effects of
industrial relations legislation since 1979', National Institute
Economic Review, No. 131 (February), pp. 57-70.
Bureau of Labor Statistics. (1993), International Comparisons of
Labor Productivity and Unit Labor Costs, US Department of Labor.
Crafts, N.F.R. and Toniolo, G. (1995), `Post-war growth: an
overview', CEPR Discussion Paper No. 1095. London.
Daniel, W.W. and Millward, N. (1983), Workplace Industrial Relations
in Britain: the DE/PSI/ESRC Survey, Heinemann.
DeLong, J.B. and Summers, L. (1991), `Equipment investment and
economic growth', Quarterly Journal of Economics, 106, pp. 445-502.
Denny, K. and Nickell, S.J. (1992), `Unions and investment in British
industry', Economic Journal, 102, pp. 874-87.
Dowrick, S. and Nguyen, D-T. (1989), `OECD comparative economic
growth 1950-85: catch-up and convergence', American Economic
Review, 79, pp. 1010-1030.
Eichengreen, B. (1992), `The origins and nature of the Great Slump
revisited', Economic History Review, 45, pp. 213-239.
Eltis, W. and Higham, D. (1995), `Closing the UK competitiveness
gap', National Institute Economic Review, No. 154 (November), pp.
Floud, I.R. and McCloskey, D. eds. (1994), The Economic History of
Britain since 1700. Volume 3: 1939-92(2nd edition), Cambridge University
Freeman, R.B. and Medoff, J.L. (1984), What Do Unions Do? New York:
Basic Books, Inc.
Friedman, M. and Schwartz, A.J. (1963), A Monetary History of the
United States, 1867-1960, Princeton University Press.
Green, A. and Steedman, H. (1995), Widening Participation In Further
Education And Training: Literature And Issues Survey, Further Education
Funding Council (forthcoming).
Gregg, P., Machin, S. and Metcalf, D. (1993), `Signals and cycles:
productivity growth and changes in union status in UK companies,
1984-89', Economic Journal, 103, pp. 894-907.
Grout, P. (1984), `Investment and wages in the absence of binding
contracts: a Nash bargaining approach', Econometrica, 52, pp.
Haskel, J. and Martin, C. (1993), `Do skill shortages reduce
productivity? Theory and evidence from the United Kingdom',
Economic Journal, 103, pp. 386-394.
Layard, R. and Nickell, S. (1989), `The Thatcher miracle?', CEPR
Discussion Paper No. 315, London.
Levine, R. and Renelt, D. (1992), `A sensitivity analysis of
cross-country growth regressions', American Economic Review, 82,
Lucas, R.E. (1967), `Adjustment costs and the theory of supply',
Journal of Political Economy, 75, pp. 321-334.
Lucas, R.E. (1987), Models of business cycles. Blackwell.
Lucas, R.E. (1988), `On the mechanics of economic development',
Journal of Monetary Economics, 22, pp. 3-42.
Lynch, L.M. (1993), `The economics of youth training in the
U.S', Economic Journal, 103, pp. 1292-1302.
Machin, S. and Stewart, M. (1995), `Unions and financial
performance', Centre for Economic Performance, Discussion Paper No.
Machin, S. and Wadhwani, S. (1989), `The effects of unions on
organisational change, investment and employment: evidence form
WIRS', Centre for Labour Economics Discussion Paper No. 355.
Mankiw, N. G., Romer, D., and Weil, D.N. (1992), `A contribution to
the empirics of economic growth', Quarterly Journal of Economics,
107, pp. 407-437.
Metcalf, D. (1994), `Transformation of British industrial relations?
Institutions, conduct and outcomes', In R. Barrell (ed.), The UK
Labour Market, Cambridge University Press.
Millward, R. (1994), `Industrial and commercial performance since
1950', In Floud and McCloskey (1994),
National Foundation for Educational Research (1994), Standards in
Literacy and Numeracy 1948-94. London.
Nickell, S., Wadhwani, S. and Wall, M. (1989), `Unions and
productivity growth in Britain 1974-86: evidence from UK company
accounts data', Centre for Labour Economics Discussion Paper No.
North, D.C. and Thomas, R.P. (1973), The Rise of the Western World: A
New Economic History. Cambridge University Press. OECD (1995), Education
At a Glance: OECD Indicators. OECD.
O'Mahony, M. (1995), `International differences in manufacturing
unit labour costs', National Institute Economic Review, No. 154
(November), pp. 00-00.
Oulton, N. (1987), `Plant closures and the productivity
`miracle' in manufacturing', National Institute Economic
Review, No. 121 (August), pp. 53-59.
Oulton (1989a), `Trade unions, product quality and the structure of
international trade', In J. Black and A. I. MacBean (eds.), Causes
of Changes in the Structure of International Trade, 1960-1985,
Oulton, N. (1989b), `Productivity growth in manufacturing: the roles
of new investment and scrapping', National Institute Economic
Review, No. 127 (February), pp. 64-75.
Oulton, N. (1990), `Labour productivity in UK manufacturing in the
1970s and the 1980s', National Institute Economic Review, No. 132
(May), pp. 71-91.
Oulton, N. (1992), `Investment and productivity growth', in A.
Britton (ed.) Industrial Investment As A Policy Objective, National
Institute of Economic and Social Research, Report Series No. 3. London.
Oulton, N. (1994), `Labour productivity and unit labour costs in
manufacturing: the UK and its competitors', National Institute
Economic Review, No. 148 (May), pp. 49-60.
Oulton, N. (1995), `Work force skills and export competitiveness: an
Anglo-German comparison', Forthcoming in A.L. Booth and D. Snower
(eds.), Acquiring Skills: Market Failures, Their Symptoms, and Policy
Responses, Cambridge University Press.
Oulton, N. and O'Mahony, M. (1994), Productivity and Growth: A
Disaggregated Study of British Industry, 1954-86, Cambridge University
Perron, P. (1989), `The Great Crash, the oil price shock, and the
unit root hypothesis', Econometrica, 57, pp. 1361-1401.
Prais, S.J. (ed.) (1995), Productivity, education and training.
Britain and other countries compared (volumes 1 and 2), National
Institute of Economic and Social Research. London.
Romer, C.D. (1988), `World War I and the post war depression. A
reinterpretation based on alternative estimates of GNP', Journal of
Monetary Economics, 22, pp. 91-116.
Sichel, D.E. (1993), `Business cycle asymmetry: a deeper look',
Economic Inquiry, XXXI, pp. 224-236.
Steedman, H. (1995), `Limitations of current systems of educational
classification: the challenge of measuring quality of educational
outputs', LSE, Centre for Economic Performance. mimeo.
Ullman, L. (1968), `Collective bargaining and industrial
efficiency', In R.E. Caves et al., Britain's Economic
Prospects, Allen and Unwin.
Wiener, M.J. (1981), English Culture and the Decline of the
Industrial Spirit 1850-1980, Cambridge University Press.
Wren-Lewis, S., Westaway, P., Soteri, S. and Barrell, R. (1991),
`Evaluating the UK's choice of entry rate into the ERM',
Manchester School, 54 (Supplement).
(1) To forestall misunderstanding, I do not attempt a full scale
`audit of Thatcherism' (or of Majorism) for which see for example
Bean and Symons (1989) and Layard and Nickell (1989). Neither the
desirability of economic growth, nor unemployment, nor the
distributional implications of the supply side reforms will be
(2) See Wiener (1981) for a comprehensive catalogue of the
anti-growth views of the British political and intellectual
establishment in the twentieth century.
(3) These figures differ slightly from similar ones in Oulton (1994)
due to the use of more recent data from the U.S. Department of Labor,
Bureau of Labor Statistics.
(4) Similar results have been reported by Crafts and Toniolo (1995).
(5) Delong and Summers (1991) have argued that it is investment in
plant and machinery not investment in general which is crucial for
growth. Some doubt has been cast on this view for the currently rich
countries by Oulton (1992); see also Auerbach et al. (1994). In any
case, the gap between the UK and other countries in plant and machinery
investment is much smaller than for total investment.
(6) See Metcalf (1994) for a recent overview and Ullman (1968) for an
evaluation of the system as it stood in the late 1960s.
(7) However, Broadberry (1994) does consider the relationship between
trade unions and productivity while discussing the improvement of the
(8) The general point being made here is widely accepted in fields
outside economics. For example, it is generally held that the unanimity
rule of the Polish Diet was an important cause of the collapse and
partition of Poland in the eighteenth century.
(9) Let the cost of adjusting the capital stock by one unit be a
homogeneous of degree one function C(I, K), where I is investment and K
is capital stock. The marginal cost of investment is [C.sub.I](I/K), a
function only of the ratio I/K, and this is assumed to be increasing,
[C.sub.II](I/K) [is greater than] 0. The optimal policy for the firm is
to equate the marginal cost of investment to the marginal return, which
depends on input and output prices. The result in the text then follows
if for example the firm takes prices as given.
(10) This argument was demonstrated rigorously by Arrow and Nerlove
(1962) in the context of advertising expenditure. Oulton (1989a) argues
that it applies more widely to any expenditure of an overhead character
which has the effect of shifting the marginal revenue curve to the
right. In an international setting, if product improving expenditure
requires skilled workers, and if unions push up the real wages of less
skilled production workers, then in equilibrium there will be a lower
ratio of skilled workers than in other countries where unions are less
(11) See Blanchflower and Freeman (1994) for an up-to-date survey of
the main measures. The legislation has not just reduced union power,
though this is the aspect relevant here. It has also increased the
accountability of unions to their members, a feature which made the
reforms more politically acceptable at the time and which may one day,
somewhat ironically, be the basis for a revival of union power.
(12) Some but not all of this decline in density and recognition
might be ascribed to the fall in employment in heavily unionised sectors
of the economy. But as Metcalf (1994) asks, why was it that unions were
unsuccessful in recruiting in expanding areas of the economy?
(13) It is sometimes claimed that the example of the United States
throws doubt on the need to achieve a high average level of educational
or vocational attainment. It is true that most of the non-college
educated receive no further formal training after leaving high school,
though there is a very large proportion of college educated people.
However currently about four fifths of young Americans complete high
school, which ends at age 18 (Lynch 1993).
(14) These studies by various authors are collected together in two
volumes edited by Prais (1995). In addition, Haskel and Martin argue
that skill shortages reduced labour productivity growth over the period
1980-86 by as much as 0.7 percentage points per annum.
(15) The propensity to enrol in further education is higher the
higher the GCSE grade. But the propensity to enrol has also risen at any
given grade, following the introduction of GCSE.
(16) One reason that Germany shows up poorly on the `attained
secondary' measure is probably that the dataset only covers
educational not vocational qualifications, so that German apprentices
are not included. Apprenticeship is now nearly universal for those not
following the academic route. In OECD (1995) apprenticeships are
included in the definition of attained upper secondary education giving
a figure of 60 per cent for Germany in 1992. But this just illustrates
the difficulties. There is no reason to think that this is the only
adjustment which needs to be made.
(17) According to the Bank of England Quarterly Bulletin (February
1995, p. 17), the recession in the UK was deeper than in any other G7
country. On a different measure of length (number of quarters from peak
to trough), the UK's recession was also (with Japan's) the
(18) Policy during the 1979 recession has been widely criticised as
being too tight. Surprisingly, the conduct of policy during the 1990
recession has attracted far less criticism. In defence of the 1979
recession, we can note that it coincided with a major oil shock at a
time when the UK had just become an important oil producer itself.
Criticism has focused on the appreciation of the exchange rate but at
the time many people argued that the high pound was due to
Britain's new status as an oil producer and not to tight monetary
policy. To be fair to the Bank and the Treasury, the twin policy errors
identified here may also have been more obvious ex post than ex ante.
(19) The quarterly growth rate of GDP is defined as
(20) Spain has the lowest variance. Examination of the Spanish series
shows that it is implausibly smooth, probably because the quarterly
series is basically an interpolation of the annual one.
(21) If the economy could be adequately described by a
representative, risk averse consumer, then reducing the variance of the
level of GDP might be a reasonable objective. In practice it seems
plausible that the people who lose in recessions are not identical to
the ones who gain in booms. The assumption of a representative consumer
produces estimates of the loss from macroeconomic instability which are
negligibly small (Lucas 1987). If this assumption is valid, one wonders
why anyone bothers about macroeconomic policy at all.
(22) Positive skewness is the opposite of what Sichel (1993) calls
(23) The HP filter is linear and therefore cannot induce asymmetry in
a series which is originally symmetric, and it makes stationary a series
which is trend or difference stationary (that is, I(1)); see Sichel
1993. The HP filter has one free parameter, denoted by [Lambda], which
is usually set to 1600 for quarterly data.
(24) His confidence in his judgement would have been increased by the
advent of deposit insurance in 1933. That the Great Depression really
was a one-off is confirmed by Perron (1989) who shows on the basis of a
time series model that it had a permanent effect on the level of US GDP
but not on its growth rate.
(25) Bean and Crafts (1995) are only cautiously optimistic while
Brown and Wadhwani (1990) and Blanchflower and Freeman (1994) are almost
unrelievedly negative. However, on a close reading the latter two do
accept the existence of a productivity improvement.
(26) One might certainly hope so, given that entry into the ERM at an
overvalued exchange rate was a rerun of the return to gold in 1925.
percent per annum
1960-73 1973-79 1979-89 1979-94
United States 3.28 1.41 2.34 2.47
Canada 4.44 2.03 1.45 1.81
Japan 9.59 5.15 4.58 4.18
Belgium 6.69 5.83 4.16 3.73(a)
Denmark 6.22 4.09 1.28 1.68
France 6.55 4.39 3.28 3.04
Germany 5.71 4.21 1.83 2.22
Italy 6.14 5.60 3.86 3.91
Netherlands 7.15 5.32 3.40 3.04
Norway 4.69 2.21 2.03 2.06
Sweden 6.25 2.65 2.53 2.87
United Kingdom 4.14 1.01 4.13 3.95
Table 2. UK output per person employed
growth rates, per cent per annum
Manu- Private Whole
facturing services economy
1960-68 3.43 NA 2.80
1968-73 3.88 NA 3.10
1973-79 0.62 1.68 1.21
1979-89 3.98 2.50 2.00
1979-93 3.66 2.29 1.85
Table 3. Productivity growth in UK manufacturing
percent per annum
Labour Total factor
1954-73 4.60 1.18
1973-86 2.23 -0.54
1973-79 1.66 -1.53
1979-86 2.73 0.30
1979-82 1.36 -0.95
1982-86 3.75 1.24
Table 4. UK output growth
per cent per annum
Period Manufacturing GDP
1960-68 3.19 3.15
1968-73 2.82 3.26
1973-79 -0.72 1.40
1979-89 0.91 2.39
1979-94 0.67 1.85
1989-94 0.20 0.79
Country 1960-73 1973-89 Change:
% per annum % per annum 1973-89 minus
Australia 2.75 1.11 -1.64
Austria 5.02 1.31 -3.71
Belgium 4.15 1.01 -3.14
Canada 2.70 1.40 -1.30
Denmark 2.75 1,18 -1.57
Finland 3.67 2.22 -1.45
France 4.48 1.20 -3.28
Germany 4.22 1.20 -3.03
Greece 7.35 1.60 -5.75
Hong Kong 6.75 5.12 -1.63
Ireland 4.35 2.10 -2.25
Israel 5.22 1.24 -3.98
Italy 4.92 2.09 -2.82
Japan 7.73 3.13 -4.61
Netherlands 3.85 0.33 -3.52
New Zealand 1.42 0.02 -1.40
Norway 2.90 2.26 -0.64
Portugal 6.08 1.36 -4.71
Singapore 6.91 3.71 -3.20
Spain 6.32 0.92 -5.40
Sweden 2.38 1.00 -1.39
Switzerland 2.64 0.84 -1.80
Turkey 4.51 1.59 -2.91
UK 2.41 1.66 -0.75
us 2.25 0.87 -1.39
persons aged 16-69, not in full time education, per cent
1975 1985 1993
Highest qualification level
Degree and above 5 8 10
Higher intermediate 11 16 23
Lower intermediate 14 19 23
Low or no qualifications 70 57 44
Highest qualification level
Degree and above 7.2 6.0
Higher intermediate 4.4 8.2
Lower intermediate 28.1 57.2
No vocational qualifications 60.3 28.6
Period Peak Trough Recovery Time to
quarter quarter quarter recover
1973-76 1973Q3 1975Q3 1976Q4 13
1979-83 1979Q2 1981Q1 1983Q2 16
1990-94 1990Q2 1992Q1 1994Q1 15
per cent per annum at annual rate, seasonally adjusted
Country Mean Variance Skewness
Rank Value Rank Value Rank Value
Australia 3.15 3 26.35 2 -0.47 7
Austria 2.83 5 43.84 1 0.06 4
Canada 3.29 2 15.89 4 0.07 3
France 2.53 8 7.78 10 -0.78 10
Germany 2.39 9 10.55 9 -0.31 6
Italy 2.57 7 13.94 6 0.09 5
Japan 3.71 1 12.41 7 -0.48 8
Spain 2.84 4 6.20 11 0.30 2
Switzerland 1.52 11 10.62 8 -1.89 11
UK 2.10 10 20.78 3 0.36 1
US 2.58 6 14.05 5 -0.55 9
per cent per annum at annual rate, seasonally adjusted
Country Mean Variance Skewness
Rank Value Rank Value Rank Value
Australia 3.01 2 24.99 5 -0.74 14
Austria 2.13 7 31.38 3 -0.11 4
Canada 2.64 3 13.29 7 -0.33 10
Denmark 1.97 11 26.11 4 -0.20 6
Finland 1.94 12 69.74 1 -0.14 5
France 2.04 9 6.86 12 -0.48 11
Germany 2.20 5 9.90 9 -0.22 8
Italy 2.13 6 9.26 10 -0.20 6
Japan 3.42 1 8.26 11 -0.49 12
Netherlands 1.91 13 40.60 2 0.35 1
Spain 2.12 8 3.84 13 -0.05 3
Switzerland 1.69 14 3.74 14 -0.24 9
UK 2.01 10 14.45 6 0.10 2
US 2.53 4 12.07 8 -0.55 13
Country N Skewness Variance
Australia 100 -0.5060 0.000259
Austria 100 -0.0643 0.000236
Canada 100 -0.5135 0.000289
Denmark 72 0.1906 0.000218
Finland 80 0.4408 0.000693
France 100 0.0431 0.000115
Germany 100 0.2923 0.000221
Italy 100 0.4585 0.000256
Japan 100 0.9680 0.000193
Netherlands 72 0.0738 0.000241
Spain 100 0.2353 0.000136
Switzerland 100 0.7037 0.000336
UK 100 0.6946 0.000336
US 101 -0.4012 0.000309
Period % per annum Period % per annum
1913-29 3.26 1937-62 3.88
1919-29 3.48 1941-53 3.88
1929-37 -0.25 1941-62 3.30
1929-41 2.09 1948-62 3.52
1919-37 1.80 1953-62 2.54
1937-53 4.63 1948-73 3.69