The growth of output and productivity in the UK: the 1980s as a phase of the post-war period.

Feinstein, Charles ; Matthews, Robin

[SOME TABLES HAVE BEEN OMITTED]

The Review is pleased to give hospitality to the deliberations of the CLARE Group but is not necessarily in agreement with the views expressed. Members of the CLARE Group are M.J. Artis, A. J. C. Britton, W A. Brown, C. H. Feinstein, C. A. E Goodhart, D.A. Hay, J.A. Kay, R.C.0. Matthews, M. H. Miller, P. M. Oppenheimer, M.V. Posner, W.B. Reddaway, J. R. Sargent, M. F-G. Scoff, Z A. Silberston, J.H.B. Tew, J. S. Vickers, S. Wadhwani. Trends in output and productivity in the UK in the 1980s need to be considered in comparison with the whole post-warperiod, not just in comparison with the abnormal 1970s; account must be taken of the whole economy, not just manufacturing; and the outcome must be viewed in the context of the experience of other OECD countries. The authors present some new statistical estimates and offer a wide-ranging interpretative survey of what the statistics reveal They avoid giving praise or blame to governments, whose influence on long-run trends, whether for good or ill, they view with some agnosticism.

1. Introduction

The completion of Mrs Thatcher's first ten years as Prime Minister stimulated many surveys of the economic policy of the government during that time in office (for instance Layard and Nickell 1989; Bean and Symons 1989). Authors writing at that time were concerned particularly to make comparisons with the performance of the predominantly Labour governments that had been in power during the previous economic cycle, running from 1973 to 1979.

Our purpose is rather different. Whether Mr Gladstone or Mr Disraeli was in occupation of No.10 at a particular date is a question that economic historians do not often ask. Certainly it is not their custom to attribute to those worthies the credit or the blame for the performance of the British economy in a particular decade of the 19th century. Nor was it the custom of contemporaries. In our own times, the government's direct and indirect participation in the economy is far greater, of course. But it is by no means the only influence, whether for good or ill. As far as some variables are concerned, it may even be a relatively unimportant influence. It is obvious that many though not all of the trends in the past 20 years have been common to OECD countries, whatever their governments. This suggests either that governments did not affect them or else that governments had rather little room to manoeuvre.

Our topic in this paper is the performance of the British economy during the 1980s, as influenced by a number of causes, of which government policy was one, but only one. We are interested in the economy's performance during that period partly as a matter of history and partly as a possible indicator of likely future trends. By performance we mean here mainly the level and growth of real national income. There exist other important variables, like inequality, but our scope will not allow us to treat them here. We shall touch briefly on such major topics as inflation only insofar as they influenced growth, while recognising that many would regard them as ingredients of performance in their own right. A central feature of the paper will be comparison of the 1980s with earlier parts of the post-war period, as well as with the 1970s. This is essential, because the 1970s were very abnormal.

2. Output and productive potential

The level of output and its trend over time depend on productive potential and on the extent to which productive potential is utilised. (a) Productive potential The simplest method of estimating the trend in productive potential is to take the annual rate of growth between cyclical peaks. 1988 is now reasonably well established as a cyclical peak in the sense in which the word has been used in the post-WWII period. An almost equally good case could probably be made for 1989, but full 1989 data are not yet available. in the present instance, therefore, a peak-to-peak growth measure means a 1979-88 growth measure. We shall refer to the period 1979-88 loosely as the 1980s. 1979 is chosen as the initial year because it was a cyclical peak, not because it was the year when Mrs Thatcher's government took office. (If instead rates of growth were to be measured over the period 1981-88, as is sometimes done in political debate, they would need to be compared with upswing years only of earlier cycles, and the comparison would in addition be affected by the relative severity of different recessions.)

Growth rates between successive cyclical peaks since WWII are shown in the first column of table 1. As may be seen, the growth rate in the 1980s was higher than in 1973-79, but it was lower than in the average of earlier post-war cycles or in any of them individually.

Before drawing any conclusions from these figures regard must be had to two complications. They affect the comparison of the 1970s and the 1980s in opposite directions.

(i) The period 1973-9 benefitted from the exceptionally rapid expansion of the output of North Sea oil and natural gas. The importance of this sector increased both with the higher rate of extraction and with the rise in the price of oil. It accounted for only 0.01 per cent of GDP in 1975, but for 4.4 per cent in 1980 and for 6.2 per cent in 1985, calculated in each case at current prices(1). A case can be made for excluding it from GDP (on the grounds that it requires little domestic input and so more nearly resembles property income from abroad, or alternatively on the grounds that it is matched by depletion of a resource which earns royalties because its price exceeds the cost of extraction). We shall not pursue the theoretical issues here, but simply give in table 1 the figures excluding NS oil in brackets. The effect of the exclusion of NS oil is to depress the growth rate in the 1970s further and so strengthen the impression of improvement in the 1980s. it does not make any difference to the growth rate for 1979-88 itself. Hence it does not make any difference to the comparison between the 1980s and periods before 1973.

(ii) The cyclical-peak method assumes that output bears the same relationship to productive potential at each peak. This is never likely to be exactly valid. But it is subject to particular objection in the present instance. There is good reason to suppose that in 1979 output was significantly lower relatively to productive potential than in either 1973 or 1988. in 1979, a much smaller percentage of firms reported to the CBI that they were working at capacity (1973, 55; 1979,43; 1988, 68); in the labour market, unemployment fell less; in the goods market, there was less spilling over of excess demand into a visible trade deficit.

While no estimate of the necessary quantitative adjustment can be offered with confidence, there is clearly a danger that if no adjustment is made at all, any evaluation of the trend in productive capacity may be seriously distorted: growth from 1973 to 1979 may be made to look too slow and growth from 1979 to 1988 too fast. Indeed, the evidence of the 1973, 1979, and 1988 levels of output barely even rules out the strong hypothesis that the rates of growth of productive potential were actually the same in 1973-79 and in 1979-88. in order for that to be true, it would be necessary for the appropriate measure of productive potential in 1979 to be about 3 per cent higher than the measure provided by actual output-by no means an incredible adjustment. The corresponding adjustment for the series without NS oil is rather larger-nearly 5 per cent-but still not enormous.

One procedure, admittedly rather arbitrary, for dealing with the unequal intensity of the rate of activity in different peaks is to measure the rate of growth, not between one peak year and another, but between the averages for the peak years and the years immediately preceding them (e.g. between the average of 1972 and 1973 and the average of 1978 and 1979, instead of between 1973 and 1979).(2) This is done in the second column of table 1. The improvement of the 1980s over the 1970s is reduced without being completely eliminated.

The adjusted figures excluding NS oil probably give the fairest indicators. There are other measurement problems but they are unlikely to affect the outcome seriously. (3) Taking the average over the five pre-1973 cycles, these adjusted figures give the following summary picture:

1950/1-72/3 3.0

1972/3-78/9 1.3

1978/9-87/8 2.0

The rate of growth in the final period lay above that of the unsuccessful 1970s. But the success of the 1980s was a distinctly qualified one, in that its growth rate lay nearer to that of the unsuccessful 1970s than it did to that of the successful 1950s and 1960s. This is the first major point to be emphasised.

(b) Utilisation of productive potential over the cycle as a whole While peak years may give an indication of the trend in productive potential, what happens between peak years matters too. Indeed all years are equally important for the people living through them. One way of measuring the experience of all years within a cycle is shown in table 2. Cycles are defined as running from the year after a peak up to and including the next peak. Output in each year within the cycle is expressed as a relative of the previous peak, taken as 100; these relatives are averaged; and the annual rate of growth of this average from the previous peak is calculated over a period half the duration of the cycle. A similar point of view is provided by Chart 1, which shows movements of GDP over three periods of equal length, 1955-64, 1964-73, and 1979-88, along with the rather shorter period 1973-79.

Both the 1980s and the 1970s show up less well relatively to pre- 1 973 cycles on the measure in table 2 than they did on the conventional peak-to-peak measures in table 1. The reason is, of course, that in addition to having lower peak-to-peak trends, they fell below those trends by greater amounts: the recessions of 1973-75 and 1979-81 were more severe than earlier recessions, as is shown in Chart 1. in table 2, the 1980s continue to appear better than the 1970s in the unadjusted figures for GDP (though not for manufacturing), but to a rather less extent than in table 1, mainly because the recovery from the trough was slower in the 1980s than in the 1970s.

A further feature of table 2 is the following. In table 1, there appears to be an uneven plateau up to 1973, then a cliff. In table 2, premonitions of the cliff are given by the downward slope from the highest growth rate, which comes as early as 1961-64. In manufacturing a rise, then fall, in the rate of growth is even more apparent, over almost the entire post-war period. In both columns of table 2, the highest growth rate comes earlier than 1973.

The foregoing measure does not overcome the problem of the differing intensity of peaks, since it involves taking the initial peak level as 100. A measure that does deal with that is shown in table 3. Here the average of the years of one cycle is compared with the average of the years of the previous cycle; and the growth rate is calculated over a period half the duration of the two cycles, i.e. as if between the mid-points of the cycles. This is a measure with a rather different meaning from those so far considered, in that each figure depends on the attributes of two cycles, and differences between phases due to changes in utilisation are not distinguished at all from changes due to trends in productive potential. Here again, as in table 2 but more clearly, there emerges a wave, with the highest growth rate in the 1960s. There is evidence for similar wave-like movements in the post-war period in many countries (Matthews and Bowen 1988). insofar as these wave-like outcomes were not merely accidental, they suggest that it may be important to see the 1980s in the context of a longer period.

The commodity composition of output

The discrepancies between the trends in manufacturing production and GDP are well known. The growth of output by sector is set out more systematically in tables 4-6. In 1973-79 growth was slower than previously in all the broadly-defined sectors shown in table 4, other than energy. The major difference between the 1970s and the 1980s was that the slowdown in growth in the 1980s by comparison with pre-1973 was largely confined to manufacturing, while services were improving. in addition, in the 1980s compared with the 1970s, energy ceased to make a special contribution to growth, and manufacturing did rather better, though still poorly by pre-1973 standards. There were also some important relative shifts within services tables 5 and 6), the first place in the data shown so far where the direct effects of political preferences might be thought discernible.

3. Labour input

(i) Persons employed

The early 1980s were notorious for an increase in unemployment, not fully reversed by 1988. This phenomenon was common to most OECD countries. For most of the decade unemployment (on a consistent measurement basis) was higher in the UK than in other OECD countries, but by the end of 1988 unemployment had become higher in France, Italy and Canada than here. Even so, unemployment in 1988 was still abnormally high here by post-war standards, notwithstanding high vacancies. (See box.) The unemployment percentage was about the same as that at the end of the 1930s-likewise a period when shortages of labour in some occupations coexisted with high unemployment.

As in other periods when unemployment has risen sharply, such as the early 1920s, trends in total labour input were quite complicated. The total number of persons in work was about the same in 1988 as it had been in 1979. There was an increase of about 5 per cent in the population of working age, but it did not find its way into work. That increase, plus the slightly faster than usual trend increase in labour-force-participation by women, was offset by the rises in unemployment and labour-related training schemes and by the decline in labour-force-participation of older men (table 7). The latter was a long-established trend, but in the 1980s it began for the first time to extend on a significant scale to men in their 50s. The accelerated withdrawal of late-middle-aged men from the labour force was no doubt caused largely by lack of demand, institutionalised in some cases by redundancy or early retirement schemes; but its persistence in the period of such high demand for labour as the late 1980s suggests that it is likely to prove irreversible, at least for these cohorts.

The net result of these and other trends was to produce very substantial changes in the nature of the labour force at work. Between 1979 and 1988 the number of full-time male employees fell by 1,890,000 (1 4 per cent). There was also a small reduction in the number of full-time female employees (130,000), despite the increase in the total number of women at work. But these were offset by an increase of 870,000 in part-time employees (530,000 female, 340,000 male) and b a rise of over a million in the number of the self-employed. Since part-timers, females, and the self-employed all tend to earn less than the average per hour, these changes jointly and severally amounted to a shift towards labour that was cheaper per hour and probably also had a lower efficiency-wage. All three trends were observed also in the inter-war period and were commonly thought at that time to be the result of low demand for labour: a man losing his job might set up on his own or do a less than whole-time job or rely on his wife doing so.

At the same time there were great changes in the allocation of the labour force between industries. Employment in manufacturing fell by nearly two million. This was matched, in about equal parts, by increases in two categories, (a) finance and business services (b) the group comprised by public, professional, and miscellaneous services. Changes in other classes were small by comparison. (ii) Hours worked

Average weekly hours worked by full-time adult manual employees in manufacturing were broadly the same at the beginning and end of the period. They fell by about two hours (to 41.8 hours) between 1979 and 1981, then recovered to 44.3 hours in 1988. A wider measure, including non-manual workers and all industries and services, but still covering only full-time adult employees, shows a slight fall over the period from 41.5 in 1979 to 40.6 in 1988. The corresponding figure for 1973 was approximately an hour higher(4) .

For a more comprehensive estimate, covering all employed persons, including part-time workers, juveniles and the self-employed, we use the estimates of working time obtained from sample surveys conducted each Spring by the European Community. These show actual hours worked, allowing for time lost through sickness, disputes, holidays, etc., and including both paid and unpaid overtime. These estimates indicate that the number of hours worked by part-time employees (both male and female) was fairly steady at 17-18 hours per week. Self-employed persons worked appreciably longer hours than employees: for males the working week was about 52 hours, and for females, a larger proportion of whom worked on a part-time basis, it was about 34 hours.

This information on weekly hours worked can be combined with the figures for the number of persons at work to get a rough estimate of total person-hours worked in 1979 and 1988. In total the weekly personhours worked have fallen by some 4 per cent between 1979 and 1988, whereas the number at work was basically unchanged. This fall reflects the decline in the number of full-time male employees, the fall in their weekly hours, and the increase in the proportion of part-time female employees, partly offset by the increase in the proportion of self-employed.

Because the Survey is conducted in the Spring it does not make adequate allowance for the time lost by holidays. Over the same period since 1979 there has also been a general increase in paid holiday entitlements, and by 1988 almost all full-time manual employees were entitled to 4 weeks or more and one-quarter to 5 weeks or more (in addition to bank holidays). In 1979 only about one-third were given 4 weeks and the remainder had between 3 and 4 weeks. This change would represent a further reduction in annual hours worked of roughly 1.5 per cent.

This reduction of 5.5 per cent over nine years amounts to a reduction at the average annual rate of 0.6 per cent. Since the change in the total number of persons at work was negligible, this same figure, 0.6 per cent, represents the rate of reduction of total labour input in person-hours between 1979 and 1988.

Similar estimates have been made for the period 1973-79. These, together with the figures for the earlier part of the post-war period, are summarised in table 8.

The result is rather surprising. Despite the much complained-of, and quite genuine, increases in unemployment and in the number of discouraged workers, the rate of change of labour input, measured in person-hours, was almost exactly the same in the 1980s as it had been in the golden age of the 1950s and 1960s! The less favourable trend in the number of persons in employment in the 1980s was offset by a less rapid than previous reduction in hours per person. Labour input contracted at a slightly more rapid rate in 1973-79, but the difference was not great.

It is difficult to pinpoint the reasons for this constancy in an environment where there was obviously a weaker demand for labour. Trends in labour input result from a large number of demographic and economic components, not all pulling in the same direction. However, the long hours worked by the substantially increased number of self-employed was a new element in the 1980s; and the rate of reduction of hours of work of employees seems also to have been rather slower than formerly (certainly slower than in 1964-73).

From one point of view, then, British economic performance in the 1980s might be held to have been better than is commonly alleged. The economy proved capable of producing enough work to do for there to be no drastic downward break in the trend of total labour input. This was achieved, however, by means of an increase in the inequality between individuals in the amount of work done: the unemployed were at one extreme and the self-employed at the other.

4. Labour productivity

Since in 1979-88 the trend in labour input was the same as it had been on average in 1951-73, and much the same as it was in 1973-79, it follows that the comparison drawn above between those periods of the trends in output (table 1), applies with little change to trends in labour productivity, as shown in the left-hand panel of table 9. Or rather, it applies at the aggregative, GDP, level. The same is not necessarily true of productivity trends in individual sectors, to which we now turn.

Output, and hence labour productivity, in the economy as a whole grew nearly one per cent a year faster in 1951-73 than in 1979-88, as table 9 shows. At the same time, it is well known that productivity in manufacturing in the 1980s did much better, as shown in the right-hand panel of table 9: it increased faster than it had done in the 1950s and at about the same rate as it had done in 1964-73, hence faster than on the average of 1951-73. Trends in manufacturing productivity have been much studied, notably in recent work at the National Institute (Darby & Wren-Lewis 1988 and 1989, O'Mahony & Oulton 1990, Oulton 1990). it is on the face of it rather surprising that productivity should have done so well in manufacturing when output there was doing so badly-most unkaldorian! We shall revert to this shortly.

Arithmetically, it is not in itself particularly paradoxical that there should be different outcomes for GDP and for manufacturing, bearing in mind that manufacturing is only about a quarter of GDP; but good productivity performance in manufacturing in the 1980s must have been offset by poorer performance in some other sector or sectors in order for the figures to add up especially as NS oil was making a contribution to the growth of total GDP per worker in the 1980s that it had not made in the 1950s and 1960s). So we need to ask where the offset came.

The main offset was in the substantial sector referred to in table 10 as 'other services'. in 1951-73 this sector had shared, more or less, in the general increase in productivity. In the 1980s, however, productivity in the sector actually fell. Productivity in each of the three sub-sectors comprising it education and health; public administration and defence; miscellaneous services) did rise a little. But there was a very large increase (3.1 per cent per annum) in employment in one of them, miscellaneous services, which includes a great deal of low-productivity work. There was thus a shift of weight towards low-productivity employment, leading to a decline in average productivity in the sector. A similar effect is implied by the significant excess shown in table 10 of intra-industry productivity growth over total productivity growth.

It is natural to interpret this as a demand-induced effect -the result of people being forced by demand changes into sectors where their productivity was lower than in the sectors where demand was relatively contracting. This can be called either disguised unemployment or the enforced obsolescence of human and physical capital. A similar phenomenon had been observed, in more extreme form, in the interwar period, when productivity declined over a large part of the service sector (BEG, p.244). In addition to the reduction in average productivity brought about by changes in weights, the level of productivity even in the low-productivity occupations themselves may have been further reduced by the increase in labour supply.

It seems likely that the fall in productivity in other' services is to be interpreted in these ways. But caution is needed.

For example, it is quite possible that other services' attracted more than their share of part-timers in the 1980s (data on that are lacking). in that case, output per person-hour in other services' rose rather more rapidly than shown in table 1 0, and output per person-hour somewhere else rose correspondingly less rapidly.

For future prognostication, an important issue that arises is this. One hypothesis is that the relatively high rate of growth of productivity in manufacturing in the 1980s was due to circumstances which were peculiar to manufacturing, or largely so. Those circumstances could be connected with the shake-out caused by the troubles at the beginning of the decade, or with reductions in trade union power, or with some inter-action of the two (see below). On that hypothesis, there is no presumption that the good productivity performance will come to extend to other sectors; on some variants of the hypothesis, the rate of growth of productivity may even relapse in manufacturing itself. The alternative hypothesis is that it is the sectors that did less well which were subject to the special circumstances, as on the disguised unemployment hypothesis. In that case there would be good prospects for recovery in productivity in the economy as a whole. The data in table 1 0 seem to give some support to the second of these hypotheses, inasmuch as the source of weakness was largely concentrated in one particular sector which bore the brunt of the effects of weakness of demand elsewhere.

5. Capital input

Labour productivity is likely to be influenced by other inputs, including the input of capital. Movements of Gross Domestic Fixed Capital Formation as a proportion of GDP since 1963 are shown in Chart 2 (IN). Previously to 1963, the ratio had risen strongly. As will be seen, the ratio fell off from the peak during the 1970s and early 1980s but then recovered quite strongly in the late 1980s. it is generally believed that capital equipment was being retired more rapidly than before. The effect of this shows up in the CSO's estimates of the gross capital stock, annual growth rates of which are also shown in Chart 2 (A K/K). By contrast with the figures for I/Y, the rate of increase of the capital stock in 1988 was still a good deal below its previous peak.

At least as prominent as changes in the rate of growth of the capital stock during the 1980s were changes in its composition. indeed to judge by table 11, there was little if any net increase at all in the gross capital stock (though not, of course, zero GFCF) in production industries between 1979 and 1988, except for the extraction of North Sea oil.(5) The decline in investment was particularly marked in transport (sea and air transport, railways and other inland transport) as well as in manufacturing. Capital outlays on extraction of oil and natural gas, on coal-mines and on the electricity supply, all reached a peak in the early 1980s and have declined since then. As with labour input, the great increase took place in the service sectors, especially in banking, finance, and business serviceS(6). Doubts may be felt about the validity of the capital stock figures; certainly, the extent to which retirements were accelerated in the 1980s cannot be estimated with any confidence. It is fairly clear, however, that the upswing of output and productivity in the production industries was not based on a boom in investment in new capital to the extent that some previous booms had been (let alone booms in such countries as Germany and Japan). Increases in productivity were procured by other means.

6. International comparisons Our first finding for the UK was that GDP growth in the 1980s was well above that in the 1970s, but only about two thirds of the rate achieved before 1973. Much the same was found to be true of labour productivity. How far were these changes comparable with what occurred elsewhere?

In tables 12-14, countries are ranked in order of their growth rates of GDP in 1979-88. The trends described in table 12 have two chief features. In the first place, all the established industrial countries of the OECD underwent major reductions in their growth rates after 1973, with not much difference between the 1970s and the 1980s in the OECD average. In the second place, although in the 1970s countries preserved a ranking in growth rates not very different from that which had prevailed when they had all been doing better (hence the UK remained at the bottom), in the 1980s the old rankings changed considerably, with the UK moving into the top half. it has to be admitted that this league table meant less than it had done before 1973, because the differences in countries' growth rates had become much smaller; but even so it was a noteworthy change from the British point of view. Within Europe there was an inverse correlation between the rankings of the 1970s and 1980s. The result was that the rates of growth of different European countries over the whole period 1973-88 were very close to each other easily within the margin of statistical error.

Broadly similar results emerge from the data on trends in labour productivity given in table 13 (despite their rather different coverage). The UK again comes near the top of the ranking in the 1980s. It may be noted that the capital stock (in other countries continued in the 1980s to increase more rapidly than here. Consequently, in international comparisons, the UK shows up more favourably in the 1980s in measures of total factor productivity than it does in labour productivity.)

Most noteworthy, from the British point of view, is that part of labour productivity that is contributed by manufacturing. These are the statistics that are most often invoked in support of the notion of a British economic miracle (table 14). The exceptionally rapid rise in productivity in manufacturing, by comparison with other countries, occurred because the decline in manufacturing employment in the UK in the 1980s was very much greater than elsewhere, while there was some rise in manufacturing output. in the years between 1973 and 1979, almost all countries experienced a reduction in manufacturing employment and a reduction in manufacturing's share of output, but the UK was the only country that showed an actual contraction in manufacturing output; that was why its productivity record showed up so badly in that period.

Many explanations have been suggested for the improvement in Britain's relative productivity performance in the 1980s, especially in manufacturing. We should like to comment on one aspect that we think particularly important, which is the potential for catch-up created by the gap in productivity levels which developed during the boom years in the world economy. In the United States, manufacturing output per worker has been well above the UK level since the late nineteenth century, and a belated response to that gap may account for some of the improvement in Britain's performance, relative to her own previous standards, which occured in the years before 1973. But in that period Germany, France, Italy, Japan and several other countries all improved their productivity much more rapidly than Britain, and by the end of the period had achieved levels of output per worker in manufacturing very much higher than Britain's. Awareness of the gap had been growing for some time, and the exceptional severity of the crisis of 1979-81 may have provided the jolt necessary to provoke some of the changes required if the gap was to be closed. In this sense the improvement in productivity may have been an important, if unintended, consequence of the deflationary policies followed by the government in these years. The depth of the recession and the dangers it created for many firm's may have served to bring home to British workers and management the need for changes in attitudes, institutions and work practices, rather as defeat in war did for other countries after 1945.

The existence of the productivity gap creates the potential for improvement, the shock to attributes creates the incentive to take advantage of this potential. The response is most evident in the changes in shop floor practices, in managerial performance, and in trade union behaviour. The government's trade union laws may have made some contribution to this process; the growing inward investment by foreign countries, particularly Japan, was also helpful both by reinforcing the threat to jobs and profits of foreign competition and by facilitating the knowledge and diffusion of better practices and attitudes.

However, there is nothing automatic about the catch-up process; the gap in productivity levels creates an opportunity to adopt better methods, it does not ensure that this will happen. Most obviously, although there have been some significant improvements in British industry, much still remains to be done. Perhaps the most important point to stress is that many of the achievements of other countries are dependent on the training and skill of the work-force. Without this the relevant equipment and production methods cannot be used. Further success in closing the gap will thus require an acceptance of the need for training on a scale still not evident in Britain. At the same time, an improvement in the rate of growth of productivity is not an end in itself. It may be a necessary condition for an expansion of GDP, but it serves only limited purposes unless the resources it releases become available for other purposes. Success in achieving that depends on issues, in the realm of stabilisation policy, that lie beyond the scope of this paper.

The catch-up hypothesis explains part of the improvement in relative productivity performance during the 1980s, and leaves the possibility that Britain may continue for a while to grow more rapidly than countries that now enjoy higher levels, of productivity. It may also turn out that the catch-up process is not a matter of once for all convergence. Countries that have come to the fore, or nearly so, may slacken their efforts. If it were not for this, it would be remarkable that West Germany should have been content with such a slow growth rate in the 1980s.

So much for the second of the two main features that are revealed by international comparisons. Comment is needed now, finally, on one other aspect-the famous world slow-down, which has provided the context for British economic experience in the past two decades.

7. The slow-down: supply or demand?

The slow-down has been the subject of a huge literature-usually focussed on growth of productivity, but also applicable to growth of GDP. It is common ground that the maintenance of a fast growth rate is more difficult than it used to be. it is this that makes it at least possible to describe as brilliant or miraculous a rate of growth of the British economy that falls well below that achieved in earlier years. However, there is no agreement about its causes. Nor is it clear whether what has happened is better described as a slow-down, which sounds permanent, or as a wave, which may be followed by another wave (which conceivably has already started). There is further scope for debate about whether such recovery as has occurred in the UK in the 1980s has been a recovery from troubles common to the world or a recovery from troubles that were once peculiar to the UK. Hypotheses about causes of the world slow-down may be classified as follows.

(i) Real, macro hypotheses, based on ideas about many countries having exhausted the scope for easy catching-up either with arrears created by lags behind the leading country or by WWII; or based on ideas about running into diminishing returns from investment. The steady decline in the rate of growth of GDP in West Germany in each decade since the 1950s can be cited in support of this class of hypothesis. It is not, however, readily applicable to the slowdown in the US.

(ii) Real, micro hypotheses, based particularly on the substitution effects caused by the rises in the oil price. This view has never enjoyed much favour as applied to the UK. it has fallen further in esteem in the 1980s, when the fall in the oil price should have put things into reverse in the world as a whole but appeared not to.

(iii) Monetary hypotheses, or hypotheses about other causes independent of production, related, for example, to changes in inflationary expectations, unwise monetary or fiscal policies, the abandonment of fixed exchange rates, possibly the monetary effects of OPEC, and other such causes that belong within the sphere of stabilisation policy. The apparently catastrophic nature of the turnaround in 1973 gives support to this approach, but as noted above, premonitions can be found in the 1960s in most countries.

In addition, trends of one or other of these sorts may have elicited changes in attitudes, by affecting firms' viability-for example leading them to have a greater concern than in the past with profitability as compared with output; and these changes in attitudes may explain some of the features observed, such as the combination already referred to, in this country, of contraction in manufacturing output and rapid increase in manufacturing productivity.

It is easy to enumerate these hypotheses, but even large research programmes have not so far succeeded in discriminating decisively between them. Yet their relative importance affects considerably our view of the 1980s and of what they imply for the future. Hypothesis (i) suggests that most of the advanced world should not hope for too much further improvement and must even be prepared to slide back in its rate of growth, but it suggests that the best prospects for improvement in this country come from the elimination of the more stubborn elements in the combination of circumstances that have caused the UK still to have a level of labour productivity well below the European average. Hypothesis (iii) suggests that there will be a major need to provide inducements to create the additional physical and human capital required to give high-productivity employment to all the labour force. At the same time a great deal will depend on the avoidance of such severe recessions as the last two that have been experienced, and on success in the control of inflation.

8. Summary and conclusions

1. GDP in the 1980s grew more rapidly than in the 1970s, but at only about two thirds of the rate that it had done in the 1950s and 1960s. Thus although the 1980s were an improvement over the 1970s, there is no question of the 1980s having established new standards of performance.

2. in addition, the average level of output in the 1980s, over all the years in the cycle, was lower than indicated by the peak-to-peak growth rates just quoted, since the recession was more severe than earlier post-war recessions, even that of the 1970s.

3. There was a slight fall in the 1980s in the number of persons employed, whereas in earlier decades there had been a rise. However in the 1980s average hours of work fell less than before. The result was that the rate of change of labour input, measured in person hours, was much the same in the 1980s as formerly.

4. Hence the statements made about trends under

1. above apply to GDP per person-hour as well as to GDP.

5. The rate of growth of capital inputs in the 1980s was reduced by there being an above average rate of retirements, on a scale larger than the comparable enforced unemployment of labour. Hence the 1980s show up rather better in total factor productivity measures than they do in labour productivity measures.

6. Compared with the past, the 1980s scored better in manufacturing productivity than in GDP productivity. While it is not surprising that individual sectors should show some variation about the mean of all the industries comprising GDP, the case of manufacturing deserves special attention because it is the sector where the measurement of output is probably the most accurate. It is also paradoxical that manufacturing productivity should have risen so rapidly, since the output in that sector was rising very slowly. 7. The reconciliation of the trend in manufacturing productivity and total productivity is not so much that manufacturing was exceptional, but that there was one area of very poor productivity performance, other services'. This may reflect the indirect results of the reduction of demand in other sectors.

8. All the above has to be seen in the context of the slow-down in economic growth that has affected OECD countries generally since at least 1973, if not a bit earlier.

9. This slow-down has not left unaffected the previously established ranking of countries in growth rates. Over the period 1973-88 as a whole, the growth rate of GDP has been closely similar in the main European OECD countries. Some OECD countries have done better in the 1970s, others (including the UK) in the 1980s.

10. This offset between the experience of the 1970s and the 1980s can be taken to point to a catching-up process. But catch-up is not a process that can be relied on to occur whenever countries diverge in their income levels. The 1980-81 recession in the UK, despite its cost, may well have stimulated catch-up of arrears that had accumulated over a long period.

11. Prognostication is made more difficult by our lack of understanding of the causes of the world slow-down. In particular, it is unclear whether it was mainly due to the exhaustion of real forces that had previously made for exceptionally fast growth, and if so whether the exhaustion is temporary or permanent; or whether it was due mainly to matters that fall within the sphere of stabilisation policy. if the latter, it should in principle be susceptible more easily to improvement through policy, though the British lack of success in reducing inflation so far is discouraging in the latter regard.